CAN RUSSIA COMPETE? CAN RUSSIA COMPETE? RAJ M. DESAI ITZHAK GOLDBERG editors BROOKINGS INSTITUTION PRESS Washington, D.C. Copyright © 2008 THE WORLD BANK All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means without permission in writing from the Brookings Institution Press, except in the case of brief quotations embodied in news articles, critical articles, or reviews. Can Russia Compete? may be ordered from: Brookings Institution Press, 1775 Massachusetts Avenue, N.W. Washington, D.C. 20036 Telephone: 1-800/537-5487 or 410/516-6956 E-mail: hfscustserv@press.jhu.edu; www.brookings.edu. Library of Congress Cataloging-in-Publication data Can Russia compete? / Raj M. Desai and Itzhak Goldberg, editors. p. cm. Includes bibliographical references and index. Summary: “Looks at Russian government’s push toward a knowledge-based economy, particularly in the manufacturing sector. Quantifies and benchmarks sector’s relative strengths, identifying opportunities to increase Russian productivity and competitiveness. Examines underlying firm-level determinants of knowledge absorption, competitiveness, and productivity, with an eye to improving workers’ skill levels and the investment climate�—Provided by publisher. ISBN 978-0-8157-1831-4 (pbk. : alk. paper) 1. Russia (Federation)—Economic policy—1991– 2. Competition—Russia (Federation) I. Desai, Raj M., 1966– II. Goldberg, Itzhak, 1947– III. Title. HC340.12.C35 2008 330.947—dc22 2008019821 987654321 The paper used in this publication meets minimum requirements of the American National Standard for Information Sciences—Permanence of Paper for Printed Library Materials: ANSI Z39.48-1992. Typeset in Minion Composition by Circle Graphics Printed by R. R. Donnelley Harrisonburg, Virginia Contents Foreword vii Acknowledgments xiii 1 Introduction 1 Raj M. Desai and Itzhak Goldberg 2 Productivity 12 Mark Schaffer and Boris Kuznetsov 3 Fostering Knowledge Absorption 35 Itzhak Goldberg, Enrique Blanco-Armas, John Gabriel Goddard, and Smita Kuriakose 4 Upgrading Skills 69 Hong Tan, Vladimir Gimpelson, and Yevgeniya Savchenko 5 Improving the Investment Climate 91 Raj M. Desai Appendix 1 113 Appendix 2 117 v vi C ONTENTS Appendix 3 125 Appendix 4 142 Appendix 5 163 References 168 Contributors 175 Index 177 Foreword Consider Russia’s economic transformation in the past ten years. In 1999, Rus- sia’s economy was essentially bankrupt; it was leaking foreign exchange reserves; it was heavily indebted to the International Monetary Fund and highly dependent on financial assistance from the West; and it had been shrinking at about 6 percent a year since the breakup of the USSR. Today, Russia’s stock of foreign reserves—which it has been accumulating at an average rate of almost 50 percent a year—now stands at about 25 percent of its total GDP. Russia holds one of the largest current account surpluses in the world, and it is one of the largest financiers of the U.S. current account deficit. And its economy has been growing at almost 7 percent since 1999. The average Russian citizen is almost twice as rich as he or she was in the depths of the post-Soviet recession. Russia has come a long way in the last decade. But the good times that Russia’s economy has enjoyed have been driven by a series of idiosyncratic factors that have been—and still could be—mostly fleeting. After the 1998 crisis, the collapse of the ruble led to the expansion of exports and import- competing sectors. Falling GDP created spare production capacity in Russian industry that could later be brought into production at lower cost. As these early benefits began to be offset by an increasing real exchange rate and ris- ing production, oil and gas prices started their steep climb. The Russian oil and gas sector—less than 40 percent of exports in 1998—make up about 64 percent of Russia’s exports today, earning the Russian economy more than $230 billion in revenue, or about $1,600 a year for every citizen. vii viii F OREWORD That level of resource wealth is staggering, but oil and gas exports alone will not make Russia a permanently wealthy country. Despite Russia’s efforts to stabilize oil and gas revenues through its stabilization fund, volatility due to unpredictable changes in oil and gas prices (along with the continuing appreciation of the real exchange rate) can easily undermine non-energy investment and inhibit the emergence of a vibrant private sector outside of the oil and gas sectors. Natural resource wealth often creates powerful incen- tives to capture and control the power and wealth that resource abundance provides. But more important, oil and gas wealth can divert both investment and attention away from building the economy that Russia will need to com- pete in the twenty-first century. Competition in the new global economic environment is getting increas- ingly intense—a process that has been under way for some time. As this book makes clear, only a business environment that creates incentives for firms to invest productively, create jobs, and expand can deliver competitiveness and prosperity. In the past, China, India, and other emerging economies would compete solely on the basis of their low-skill, low-wage activities, slowly working their way up the value-added ladder. In the current environment, however, nations can quickly enter international markets by absorbing tech- nologies, production processes, and management practices from around the world. This volume—the result of collaboration between the World Bank and the Brookings Institution—argues that global competition has created a number of challenges for Russia, some of which Russia is ill-suited to meet unless its economic policies change. Globalization has dramatically increased the importance of productivity as a source of sustainable prosperity. Technology diffusion, skill acquisition, and the absorption of knowledge have become vital sources of economic growth and wealth. The results of knowledge and technology creation can contribute most to growth when they spread throughout the economy and are absorbed by a variety of economic agents, such as entrepreneurs, researchers, and firms of all sizes. But Russia’s labor productivity, although slightly ahead of that of Russia’s neighbors in the Commonwealth of Independent States (CIS), ranks behind the productivity of most of Central Europe, all of the advanced industrialized countries, and countries such as Brazil and South Africa. Russia’s “total factor productivity,� the share of productivity not directly explained by the quanti- ties of labor or capital used in production, also ranks much lower than that of the advanced economies that Russia aspires to emulate. At the same time, rapid technological change and growing skill intensity in many activities has increased the value of workers’ skills and education. F OREWORD ix Innovative activity is a crucial source of productivity growth, especially for middle-income countries that have already reached moderate levels of prosperity. Before the 1990s, Russia was widely regarded as a bastion of sci- entific knowledge. Russia’s education system is well-developed, and the coun- try boasts one of the highest percentages of university-educated individuals in the world. It also has large numbers of scientists and researchers per capita, and its spending on research development ranks highly. In other words, Rus- sia’s potential to innovate continues to be greater than that of most other countries at comparable levels of income. But despite the considerable human and financial resources devoted to science and technology, Russia ranks low on most indicators of innovative output—patents, scientific papers, and so forth. Although the national research and development institutions created under its formerly centrally planned economy focused largely on the defense sector, they still remain important to Russia’s innovation system; however, they have yet to create links with the international scientific community or with the private sector. At the enterprise level, there is less utilization of advanced technologies than in Indian or Chinese firms. Unlocking this scientific and technological poten- tial and linking it with a more dynamic private sector must become a central part of Russia’s economic strategy. Russian authorities have been keenly aware of these challenges and on numerous occasions have expressed a desire to diversify the economy. The Russian Ministry of Economic Development and Trade, in particular, has enacted a series of administrative reforms since 2000, reducing red tape and barriers to business start-up and operations. But on other fronts, Russian authorities have often failed to follow through on some necessary steps. Smaller and medium-sized Russian companies that have grown under strong competitive pressures and that are needed to diversify the economy, for example, are being penalized by the current policy regime. Many of these firms are targets of bribes, unauthorized inspections, regulatory harassment, and other forms of rent extraction. Moreover, the greater discretion over the economy that has resulted from the concentration of authority in the execu- tive branch of government has often led to inconsistently enforced or inter- preted regulations and to a high degree of uncertainty about the longer-term direction of Russia’s economic policy. The Russian government’s 2006–2008 Mid-Term Program emphasized policy directions aimed at unleashing an “innovation economy.� Vladimir Putin’s “Russia 2020� speech, delivered at the State Council meeting in Febru- ary 2008, emphasized a path of “innovative development� for the economy. And Russia’s new president, Dmitry Medvedev, has also stressed the need for x F OREWORD economic diversification through innovation. Several initiatives proposed in these statements are laudable—better protection and enforcement of intellec- tual property rights, increases in funding for basic scientific research, and greater investments in human capital. But the programs also emphasize more state control over the national “innovation system� through government- owned venture capital funds, government-owned development banking, and other forms of state intervention in the high-tech and knowledge sectors. That Russia faces a choice may be a truism. It could continue along its path toward becoming a “rentier economy� whose prosperity is largely based on natural resources. Natural resource–led growth would continue as Russia pursued greater state intervention in economic life while tolerating large innovation gaps, deficiencies in worker skills, and associated weaknesses in the policy environment. Along this path, Russia would not build a competi- tive manufacturing base from which it could join global markets. Or Russia can choose the path identified by Raj Desai, Itzhak Goldberg, and their colleagues in this volume—to make competitiveness the defining goal of its economic policy. Russia would maintain its commitment to sound macroeconomic management but address the weaknesses in the investment climate that have limited the country’s productivity and innovative potential. It would improve its national research and development institutions, its firm- level incentives to absorb advanced global technologies and offer worker training, and a policy environment that has limited the growth potential of its most dynamic firms. The result would be economy-wide value creation from improved firm-level productivity due to increased competition, trade flows, worker mobility, and foreign direct investment. Can Russia Compete? complements ongoing work at both the World Bank and Brookings. For the past three years, the World Bank has sponsored a series of flagship events organized by its Europe and Central Asia Region department to promote knowledge-based economic growth in client coun- tries as well as a series of investment climate assessments identifying the microeconomic and market impediments to growth. The World Bank also has sponsored several regional studies on innovation and on building knowledge-based economies. Meanwhile, at Brookings, international com- petitiveness and the drivers of growth in Brazil, India, China, and Russia— the BRIC countries—are part of an ongoing project within the Global Econ- omy and Development Program. In addition, a study of the political and economic consequences of natural resource dependence in Russia is the sub- ject a forthcoming book from Brookings Press. We hope that the findings and proposals in this book will facilitate an exchange of ideas and foster dialogue among policymakers in Russia and in F OREWORD xi the Eastern European and CIS region as a whole on the implementation of policies to develop strong, knowledge-intensive economies that can boost competitiveness, improve the productivity of jobs, and ultimately raise the region’s standard of living. STROBE TALBOTT FERNANDO MONTES-NEGRET President Director, Private and Brookings Institution Financial Sector Development Europe and Central Asia Region World Bank Acknowledgments This volume is based on work that began at the World Bank more than three years ago to produce a report on the Russian investment climate that would assess the constraints and opportunities facing the Russian Federation in its efforts to create a more diverse and innovative economy. We are, therefore, heavily indebted to those individuals at the World Bank who were initially responsible for shaping the issues and priorities of the report: Alfred Jay Watkins and Alexandra L. Drees-Gross, under whose leadership the first phases of the project were begun and who contributed to the development of the project’s overall concept and strategy. We express our deep gratitude for the support and collaboration of the Higher School of Economics (HSE), Moscow State University, which formu- lated and conducted the enterprise surveys in Russia under the direction of Professor Evgeny Yasin and Dr. Andrei Yakovlev, with the participation of Ksenia Gonchar and Viktoria Golikova. The authors benefited immeasurably from the presentation of preliminary findings at HSE and from extensive conversations with our HSE colleagues. We also benefited from additional presentations at the Global Institute’s Annual Russia-China Conference in Moscow and from feedback from Igor Fedyukin, Sergei Guriev, Ksenia Yudaeva, and Ekaterina Zhuravskaya. Vastly strengthening the arguments, evidence, and conclusions of this vol- ume were the contributions—both substantive and collaborative—of a num- ber of colleagues: Paloma Anos-Casero, Igor Artemiev, Irina Astrakhan, Gail xiii xiv A CKNOWLEDGMENTS Buyske, Alexandru Cojocaru, Jacqueline Coolidge, Carl Dahlman, Irina Dezhina, Inderbir S. Dhingra, Olga Emelyanova, Rostislav Kapelyushnikov, Laura Lanteri, John Litwack, Anna Lukyanova, Jack Martens, Emanuel Salinas Muñoz, David Tarr, and Sergei Ulatov. At the World Bank, Smita Kuriakose, Laura Lanteri, and Jasna Vukoje deserve a special mention for their invaluable production support and their overall supervision of the report through initial editing and production. We also are thankful for the extensive comments we received from those who took the time to review earlier drafts of our manu- script: Gordon Betcherman, Mark Dutz, William Maloney, and Shahid Yusuf. Each of us would like to thank Fernando Montes-Negret, director of the Finance and Private Sector Development Department of the Europe and Central Asia Region at the World Bank, for his strong support of this endeavor, for his extensive comments on previous drafts, and for his valuable insights and ideas. Kristalina Georgieva, former country director for Russia at the World Bank, also encouraged our efforts to extend the initial findings contained in earlier drafts of the report. More recently, Lael Brainard, vice president of the Global Economy and Development Program at Brookings, has been extremely supportive of our endeavor and of the collaboration between the World Bank and Brookings. Christopher Kelaher, Janet Walker, and Susan Woolen at Brookings Press and Valentina Kalk of the World Bank Publisher’s Office deserve special thanks for overseeing the cooperation between our institutions and for their guidance on numerous matters related to publication. Our manuscript was efficiently and skillfully edited by Eileen Hughes. For readers who wish to pursue the issues raised in this book, to examine the ongoing work of our institutions on these matters, or to replicate any of the results presented, all enterprise survey data used herein are available at www.brookings.edu/global (Brookings Institution) and www.worldbank. org/eca (World Bank). one Introduction R AJ M. D ESAI AND I TZHAK G OLDBERG T he Russian economy has been growing at an average nominal rate of 6 per- cent annually for the past decade. Among the most important factors con- tributing to its expansion has been the skyrocketing cost of oil and gas. In 2000, when Vladimir Putin took of�ce, the cost of oil was approximately $20 a barrel; at the end of his term, it was five times higher. During Putin’s pres- idency, Russia earned about $1 trillion in oil and gas revenues. Meanwhile, the competitiveness of Russian enterprises has become increasingly fragile because of the appreciating ruble, climbing resource prices, and rising wages as well as the exhaustion of Russia’s excess industrial capacity. And in the past few years, the chorus of voices (both inside and outside the country) raising concerns about the sustainability of the Russian economy’s perfor- mance has become louder. Observers have called for Russian authorities to take measures to counterbalance the nation’s increasing economic depen- dence on natural resources. Economic diversi�cation can cover a wide number of issues and involve many challenges, including entrepreneurship, foreign investment, regional development, and physical infrastructure. In Russia’s case, it comes down to one thing: ensuring that the manufacturing sector can compete in the global economy. Russian competitiveness will not depend on centralized, top-down efforts to pick winners but on broader policy measures designed both to improve the investment climate—which affects firms’ incentives to invest productively 1 2 R AJ M. D ESAI AND I TZHAK G OLDBERG and create jobs—and to develop a more competitive, knowledge-based economy. Russian authorities are seeking to address many of the country’s most important developmental challenges. They are emphasizing policies aimed at unleashing an “innovation economy,� through, among other things, greater government commitment to research and development, better pro- tection and enforcement of intellectual property rights, the formation of industrial technology parks, and the establishment of venture capital funds. Theirs is an ambitious program, yet it includes some controversial areas of economic policy that have yielded mixed results in other parts of the world. In particular, a “new industrial policy� aims to stimulate diversification and “knowledge absorption�—firms’ application of current global technologies to their production process—through direct state support to and interven- tion in specific sectors and firms. The government has proposed an array of familiar mechanisms to accomplish this: state-managed technology pro- grams, state-run development banks, state-owned venture capital funds, and so on. But to exploit the opportunities generated by a good investment climate, Russian �rms do not need more state intervention and support. They need a workforce with the skills required to carry out higher-value-added tasks. They also need the organizational and managerial capacity and the technical competence to invest, innovate, and enter strategic supply chain arrange- ments with other �rms. Increasing the incentives for the private sector to offer specialized training to more workers should be a priority, and human capital measures should be accompanied by additional incentives to encour- age firms to invest in commercial research and development (R&D), to absorb knowledge, and to adapt production processes so that they can move closer to the global technology frontier. Economic diversi�cation will require reducing investment risks induced by national and regional policies and lowering barriers to entry for newer, more dynamic, and innovative �rms, speci�cally by facilitating transfer of land from municipalities and from older, loss-making firms. It also will require greater inclusiveness in government decisionmaking, more trans- parency regarding government decisionmaking, and stable legislation at all levels of government. This book quantifies and benchmarks the relative strengths of Russian manufacturing and identi�es opportunities to increase its productivity and competitiveness. Drawing on new surveys of manufacturing �rms of all sizes, the book sets out proposals to Introduction 3 —enhance the innovative potential of Russian �rms —upgrade the skills of the workforce —develop a business-friendly climate characterized by lower administra- tive costs and greater policy certainty. The pursuit of sound economic policies following the �nancial crisis of the late 1990s and the rise of international prices for key natural resource exports have become the leading engines for Russia’s economic growth. The results are better current account and fiscal balances, higher domestic demand, moderate inflation, and rising export revenues. But continuing dependence on commodity exports will leave Russia hostage to unpredictable shocks in international prices. Boosting non-oil exports, however, will not be easy: industrial production has been dominated by energy exporters and by ferrous metals since 2001. Meanwhile, Russian manufacturing �rms have lower average productivity relative to labor costs than do manufacturing �rms in comparable countries. Even though Russia’s manufacturing value added per worker is about the same as China’s and India’s, its comparative disadvantage lies in its higher labor costs, which reflect shortages of skilled labor despite high enrollment at higher levels of education. Russia’s inputs in terms of share of researchers in the population and aggregate outlays for R&D in GDP are comparable with those of Germany and South Korea and far ahead of those of Brazil, China, and India. But the high level of inputs does not translate into high value added per capita. Russia lags behind OECD and other large middle-income countries in R&D outputs; it also has a relatively low number of patents and scienti�c publications per capita. This volume focuses on the challenges now facing enterprises in Russia, highlighting sources of productivity growth and competitiveness within enterprises, including technological progress (knowledge absorption and innovation), worker skills, and the investment climate. After the 1998 crisis, as gross domestic product rebounded, investment accelerated, and foreign direct investment increased dramatically, Russia’s recovery surpassed expec- tations. The recovery was driven to a large extent by the devaluation of the real exchange rate, the availability of cheap domestic inputs, and excess capacity and labor hoarding. Thanks to those three factors, the last several years have witnessed balanced growth, with a structural shift toward the ser- vice sector (consistent with Russia’s goal of joining the club of postindustrial nations). Yet, a closer look at national accounts reveals that much of that shift has produced relative price increases in (nontradable) services and full 4 R AJ M. D ESAI AND I TZHAK G OLDBERG capacity utilization in industry—indicators more characteristic of a resource- dependent economy than of successful industrial diversi�cation. Productivity Patterns and the Sustainability of Russia’s Economic Performance In chapter 2, Schaffer and Kuznetsov examine how productivity in Russian manufacturing compares with that in other large economies, such as Brazil, China, India, and South Africa, and in other developed economies. They argue that although productivity in the Russian manufacturing sector has been rising, it has not kept pace with rising real wages in recent years, limit- ing the international competitiveness of manufacturing. Russia’s productiv- ity lags behind that of Brazil, South Africa, and new EU entrants such as Poland. When adjusted for labor costs, it also lags behind that of India and China. And because real wages are rising rapidly and the ruble is rapidly appreciating, the international competitiveness of Russian manufacturing is suffering. This book argues that diversified growth will depend on better human capital, knowledge absorption and diffusion, and a favorable policy environment for business. Some of the relative decline in manufacturing competitiveness is due to the increase in real wages in recent years. According to Shaffer and Kuznetsov, real wages in manufacturing in Russia (deflated by the producer price index) have increased by 72 percent since 1999. In 2004 the current dollar monthly wage in industry was over $250, an increase of 67 percent in just two years and a remarkable 266 percent increase over the $75 per month wage in 1999. Under these conditions, international competition from countries with cheaper labor costs may become increasingly difficult for Russian manufacturers. Russia’s manufacturing productivity is now about 40 per- cent of Brazil’s and a third of South Africa’s. Labor productivity in Russia, measured by value added per employee, is higher than that in India and China, but low labor costs in those two coun- tries put Russia at a competitive disadvantage. For each dollar of wages, a Russian worker produces about half the output of an Indian or Chinese worker. The low productivity in manufacturing would be of less concern if it were matched with lower wages. But China’s wages in manufacturing are 30 percent lower than Russia’s. Improving Knowledge Absorption Improving the capacity of �rms to tap into the world technology pool is an important way to increase productivity. Trade flows, worker mobility, licensing Introduction 5 of codified knowledge, and foreign direct investment are all conduits of knowledge absorption. But adoption also requires a favorable investment climate, a skilled workforce, and sufficient domestic R&D. Chapter 3, by Goldberg, Blanco-Armas, Goddard, and Kuriakose, explores what can be done to boost the absorptive and innovative capacity of Russian enterprises. The complementarities between firm-specific absorptive capacity and R&D and innovation are supported by extensive theoretical and empirical work. Despite the large size of Russia’s R&D effort (in both expenditures and personnel), manufacturing productivity has not bene�ted. Based on research “inputs,� Russia’s productivity should be among the world’s highest—on par with Germany’s and South Korea’s. Instead, Russia’s R&D activities fall short of their potential. Goldberg and colleagues propose three major reforms to the institutional and regulatory regime that governs research and development in the Russian Federation. First, they recommend that incentives be strengthened to encour- age researchers at public R&D institutes to engage in commercial innovation with and promote knowledge absorption in private companies—and to facil- itate the spinning off of private research groups from R&D institutes. Scienti�c research teams in the public system often sell R&D services, on an informal basis, to enterprises. Although that practice may facilitate the “spontaneous privatization� of the R&D industry, it also leads to conflicts of interest between researchers and institutes, to uncertainty over the ownership of technical results, and to political concerns that the state is not capturing the returns from its investment and the resulting intellectual property. The government should consider creating incentives, the authors argue, for spinning off research groups. The objective should be to reduce the burden of public financing for R&D institutes, foster commercial knowl- edge absorption by firms, and reallocate basic research funding toward universities. It also is important to hasten the dissolution of R&D insti- tutes and teams within R&D institutes that work on obsolete scientific and industrial problems. Second, the authors indicate that the government should provide incen- tives for private �rms to invest more in their capacity for absorption-driven productivity growth. Matching grants can encourage public-private risk sharing and orient the selection of research projects toward commercial concerns. They can support new technologies and production processes, investments in soft technology by private �rms, and access to information and communication technologies and ISO (International Organization for Standardization) certi�cation. But 6 R AJ M. D ESAI AND I TZHAK G OLDBERG matching grant programs face risks from ineffective allocation due to cor- ruption, capture, or poorly designed targeting strategies. In a successful program, the funding and allocation mechanisms are immunized from interference by public officials, politicians, or private groups and authori- ties monitor and enforce the neutrality of targeting. Finally, the authors strongly urge the Russian government to avoid estab- lishing state-owned or state-managed venture capital programs. The Russian government has proposed a state venture capital initiative—a government- owned institution that would participate in existing venture funds and contribute to the creation of funds to finance new companies. The record of state-owned venture capital funds in other countries is poor, so caution is warranted. In the most successful cases, governments typically have seeded the venture capital industry by investing in privately managed funds. In such public-private partnerships, governments mitigate some of the risk in technology-oriented start-ups, and the venture capitalist provides commercial and managerial expertise. A seed capital program aimed at pro- moting knowledge absorption is likely to work best when a matching grants program provides critical funding at the earlier stages of technological development, with later support by private VCs. Upgrading Worker Skills No incentives to encourage innovation will have their intended effects unless the Russian workforce can acquire the skills needed to meet the challenges of the global marketplace. Turning to this critical issue, Tan, Gimpelson, and Savchenko argue in chapter 4 that the Russian workforce, though highly edu- cated by international standards, lacks the modern skills that �rms need to compete globally, a de�cit that can be made up for through an effective com- bination of vocational and in-house training. Russian �rms can no longer rely on state-funded schools to provide them with workers who possess the skills and quali�cations necessary for global competition. More companies are relying on in-house training to upgrade the skills of their employees, but they tend to provide it to a small fraction of employees. The government can assist �rms in overcoming the skill shortfall by boosting the incentives for in- house training and by engaging with appropriate private sector counterparts to reform and expand vocational training. In 2001 Russia had one of the most highly educated workforces in the world. For the bulk of the population (ages twenty-�ve and older), the aver- age citizen had 10.5 years of schooling, ahead of Brazil, India, China, South Africa, and other transition countries as well as Germany, Japan, and the Introduction 7 United Kingdom. Russia also had one of the highest shares of population holding a tertiary-level degree (more than 50 percent), more than in Canada and more than twice the share in other postsocialist countries. But despite that significant educational achievement, Russia faces problems with the quality of education, the deterioration of secondary education, and the absence of effective professional training. More than a third of all managers reported deterioration in the quality of their workforce between 1996 and 2005. The low quality of newly hired workers (rather than the high quality of employees who left the firms) may have been responsible for the reported deterioration. Almost half the firms hired workers with lower-quality skills, while only 10 percent improved workforce quality by hiring workers with higher-quality skills. Tan, Gimpelson, and Savchenko encourage greater use of employer- targeted incentives for in-service training. The proportion of employees who receive in-house training in Russian firms is among the lowest for countries with data available. The authors argue that the Russian govern- ment should consider putting in place employer-targeted training policies to remedy the underinvestment in in-service training. —Payroll-levy training funds. Employers should be closely involved in the governance of levy funds. Policies should be designed to increase competition in training provision from all providers, both public and private, including employers. Levy funds should be strictly earmarked for training and not diverted to other government uses. —Matching grants. Training levies do not work especially well for small and medium enterprises, which are unlikely to be served by targeted training programs. Encouraging training in smaller enterprises may require more proactive approaches to address systemic weaknesses in training, technolog- ical capability, and access to �nance. Matching grants can help to develop a training culture but by themselves will not expand the training market. Improving the Investment Climate In chapter 5, the �nal chapter, Desai looks at the policy-induced constraints on business activity that hold Russian �rms back from becoming dynamic and internationally competitive. He notes that the Russian investment cli- mate is still characterized by signi�cant policy and regulatory instability as well as a tendency to punish its most dynamic and innovative �rms. Although progress has been made since 2001, corruption, anticompetitive practices limiting entry of new �rms, and the quality of the legal system have contin- ued to deteriorate. Well-connected �rms tend to enjoy preferential treatment, 8 R AJ M. D ESAI AND I TZHAK G OLDBERG including special privileges, tax breaks, investment credits, direct subsidies, guaranteed loans, and access to state property; moreover, special economic zones have been created on the sites of specific enterprises. Firms con- trolled by regional private owners as well as by foreign investors are most likely to receive preferential treatment, and such favoritism toward the most politically influential firms hurts less influential regional firms. Many problems in investment climates around the world—the policy- induced costs and risks that firms face as well as the formal and informal barriers to competition—are driven by weaknesses in the institutions that govern the investment climate. Those weaknesses may allow administrative corruption to go unchecked or encourage powerful private firms to “buy� legislation, government decrees, and regulatory and judicial decisions. Strengthening the capacity and credibility of institutions may require improving the system of checks and balances, the restraints on administra- tive discretion, the ability of all levels of government to make and enforce laws, and transparency in business-government relations. Adopting such reforms has proven to be costly and politically complicated across the region. But the experience from investment climate reforms around the world suggests another way: to adopt manageable and sustainable reforms that encourage openness, competition, and greater integration with global markets and at the same time complement reforms to the systems for innovation and worker training. A set of credibility-enhancing reforms that, though seemingly disparate, could prompt deeper reforms by empowering and supporting the natural constituencies for openness, rule-based regulation, and innovation in the Russian economy includes the following: —greater transparency and flexibility in the acquisition and disposition of land, empowering entrepreneurs and �rms —an improved intellectual property rights regime, empowering inventors and entrepreneurs —more openness in policymaking through consultation, empowering business associations. Desai puts forward four proposals to move the Russian economy in the direction of the suggested reforms. First, he supports greater privatization of municipally held land. While many regions and municipalities have mechanisms to privatize real estate, they are neither transparent nor fair. Regions that adopted legislation on land privatization ahead of the federal law (the Land Code) tended to be the leaders in land reform. In other words, the adoption of the Land Code may have clarified the basis for land Introduction 9 transactions, but it did not always persuade unwilling regions to initiate land reform and privatize land. About 90 percent of the �rms trying to purchase land failed to �nish the procedure in half a year. The lack of competition in real estate markets contributes to the problem. Effective land privatization will require greater use of auctions and tenders for vacant land (not discre- tionary and opaque administrative procedures) and greater transparency in the processes involved. Second, Desai outlines ways of improving the allocation and protection of intellectual property rights (IPRs). Two primary weaknesses remain in the regime governing intellectual property rights. First, the assignment of IPRs remains unclear. There is an ongoing debate on who controls IPRs—the inventor, the inventor’s employer (research institute or enterprise, either state-owned or private), or the state, which may have paid R&D costs. Those uncertainties complicate collaboration between private firms and public institutes, inhibit technology transfer, impair the ability to spin off companies into independent and growing businesses, and create potential conflicts of interest for the institutes. Second, registered IPRs are weakly pro- tected due to the inability or unwillingness of public authorities to police producers or importers of pirated goods and to prosecute violators—a par- ticular concern for foreign investors and exporters facing copyright piracy or patent infringement by domestic producers or importers. What is needed? A more detailed elaboration of the distribution of IPRs among inventors, research organizations, and the state. A current draft of the Civil Code allows research organizations to become owners of IPRs for tech- nologies developed using government funds “provided that the procurement contracts do not specify otherwise.� The research and business communities are rightly concerned that this open-ended provision would allow public authorities to continue to exercise ownership of subsequent IPRs and prevent closer cooperation between innovators and �rms. Third, Desai argues that the government should strengthen the consulta- tive basis for regulatory decisions by being more inclusive and in particular by encouraging greater participation by business associations. Although firms that face competitive pressures are subject to harsher investment climate constraints than those that do not, when the same firms are members of business associations, they find themselves more protected from investment climate obstacles than their counterparts that are not members. Informing market participants about new and forthcoming legal and regulatory changes and requesting comments during a formal consul- tation period can improve the quality and stability of regulations and 10 R AJ M. D ESAI AND I TZHAK G OLDBERG encourage sector buy-in. The participation of business associations that represent smaller firms should be actively encouraged, and the govern- ment should develop adequate mechanisms to consult the community of entrepreneurs and business people in an inclusive manner, informing market participants well in advance of new proposed measures. Regulatory trans- parency and predictability are especially important for smaller domestic investors and for prospective foreign investors. In sum, there is power in numbers: through collective action, innovative firms can mitigate invest- ment climate constraints. Finally, Desai encourages the Russian government to adopt one or more periodic “regulatory review� mechanisms to ensure that the rules and statutes under which businesses operate are not rendered obsolete by technological change or by changes in other economic conditions. Several countries around the world have used similar mechanisms—ranging from full-fledged regula- tory impact assessments for proposed regulatory changes to “guillotine-style� reviews that eliminate outdated rules. Methodology and Structure The book uses different types of available data, including data in the Doing Business Database (World Bank 2006a), the World Bank Enterprise Surveys, the Business Environment and Enterprise Performance Survey (BEEPS), and the Russian Competitiveness and Investment Climate Assessment Survey, a new survey of Russian enterprises commissioned for this book, which includes the Large and Medium Enterprise (LME) Survey and the Small Enterprise (SE) Survey (see box 1-1). On the basis of those data, the book presents international comparisons and benchmarks to illustrate the challenges to creating a competitive Russ- ian economy. Where appropriate, the information is complemented by results of econometric regression analyses that illustrate the relationships among different investments, firm-level capabilities and characteristics, and the wider investment climate. The use of a combination of international comparisons based on aggregate and microeconomic data, econometric results from firm-level data, and relevant case studies—rather than a single data source—should increase confidence in the book’s conclusions and pol- icy recommendations. Introduction 11 Box 1-1. Data Sources Enterprise Surveys: Since 2002 the World Bank has undertaken its Enterprise Surveys (formerly known as the Productivity and Investment Climate Surveys) in more than seventy-six countries, covering more than 50,000 �rms. The Enterprise Surveys capture �rms’ experience in a range of areas, including regulation, tax policy, labor relations, infrastructure services, technology, and training. The surveys attempt to identify and, whenever possible, quantify �rms’ obstacles in the investment climate. Since the basic methodology is consistent across countries, many areas of the surveys allow international comparisons. The Russian Competitiveness and Investment Climate Assessment Survey: This survey included two enterprise surveys of Russian firms carried out specifically for the study on which this book is based. The first covered large and medium enterprises (LME Survey); the second covered small enterprises (SE Survey). These surveys were designed by the World Bank and undertaken in partnership with the Higher School of Economics in Russia and the Russian government. The LME Survey covered a stratified random sample of 1,000 medium and large firms; the SE Survey covered 300 small firms. Business Environment and Enterprise Performance Survey (BEEPS): Developed jointly by the World Bank and the European Bank for Reconstruction and Development, this survey was carried out in 1999, 2002, and 2005, covering countries in Central and Eastern Europe and Central Asia. The 2005 BEEPS was also conducted in Portugal, Spain, Greece, Germany, Ireland, and South Korea. The BEEPS examines a wide range of interactions between �rms and the state that affect the business environ- ment. It is designed to generate comparable measurements of corruption, state capture, lobbying, and the quality of the business environment, which can then be related to speci�c �rm characteristics and �rm performance. Systemwide data: For the benchmarking of different segments of the overall economy in Russia, the study draws on country-level information collected and disseminated by the World Bank through its World Development Indicators; trade data from UN COMTRADE; labor productivity data from the ILO; and information collected by other international organizations, such as the OECD. two Productivity M ARK S CHAFFER AND B ORIS K UZNETSOV In the �fteen years since the collapse of the Soviet Union, the institutions and structure of the Russian economy have changed greatly. Although much can be said about the inconsistency of the transition and the incompleteness of many structural reforms, there is little doubt that Russia has moved from a centrally planned economy to a genuine market economy. All three main goals of the economic reform initiated fifteen years ago have been largely achieved. Prices are liberalized. Privatization is more or less complete. And the economy is now at least as open to international competition as many other market economies. Since the prolonged economic depression of the 1990s, culminating in the major �nancial crisis of 1998, Russia has been one of the fastest-growing of the large emerging markets. But there is a certain uneasiness about the sus- tainability of its growth, both inside the country and abroad. Why? Because Russia’s economic welfare depends on world market prices for the natural resources and primary goods that account for the bulk of its exports—and productivity in many sectors of the economy is fairly low. Productivity is crucially important to the Russian economy for at least two reasons. First, the combination of an aging work force and a declining labor supply makes labor productivity growth imperative for sustainable economic growth. Second, the increasing openness of the Russian economy, combined with the negative impact of the persistent real revaluations of the ruble, is increasing the competitive pressure on Russian producers, particularly in 12 Productivity 13 manufacturing, in both internal and external markets. The oil and gas sec- tor has limited growth potential due to the continuing depletion of existing oil and gas fields and the consequent necessary move to more expensive drilling projects in undeveloped regions and the sea shelf. In any case, the energy sector, while a locomotive of economic development, will never be large enough to provide gainful employment for a labor force of more than 70 million or prosperity for a population of 140 million. That is why having productive and internationally competitive sectors, particularly in manu- facturing, is one of the most serious challenges facing the Russian economy and government. Total factor productivity (TFP) provides a standard framework for ana- lyzing productivity and growth. In very simple terms, economic output can be generated by growth in factor inputs (capital and labor) and by increases in the productivity of those inputs. The difference between output growth and the part of it that can be attributed to measured TFP growth is growth in inputs. The gap is sometimes referred to as the “TFP residual� or “Solow residual,� after Robert Solow, the Nobel Prize winner who introduced the concept fifty-odd years ago. The same framework can be used for compar- ing the productivity of firms, sectors, or countries. Thus, the difference in manufacturing value added across countries, after differences in the amounts of labor and capital in manufacturing have been accounted for, is a difference in TFP. Abramovitz (1956) called the TFP residual “a measure of our ignorance.� And indeed, a central focus of research in this area has been to measure the contribution to TFP of other factors such as technology, the quality of inputs, competition, the economic environment, and so on. There is now a vast amount of theoretical work and empirical evidence on this subject from many countries and periods at the aggregate, firm, and establishment level. This book focuses on three contributors to TFP generally agreed to be central to growth: innovation and absorption, labor skills, and the invest- ment climate, each the subject of a chapter. This chapter presents an overview of recent macroeconomic develop- ments in the Russian economy, with a focus on productivity and exports. It benchmarks Russia’s productivity and competitiveness across a range of emerging and developed market economies and then extends the bench- marking to explore Russia’s record of R&D inputs and technology-intensive outputs, the level of training and the skills of the labor force, and the quality of the investment climate. Key stylized facts, puzzles, and policy issues are identi�ed for exploration in subsequent chapters. 14 M ARK S CHAFFER AND B ORIS K UZNETSOV Recent Trends in the Russian Economy During the transition of the last �fteen years the Russian economy has had three major sources of aggregate productivity: —intersectoral structural changes, that is, the reallocation of resources among different sectors with different levels of productivity —the dynamics of productivity within sectors due to reallocation of mar- ket shares among �rms with different productivity levels —the growth of productivity inside �rms due to restructuring and the absorption of knowledge and technology. The relative importance of the three factors differed across the period, with the contribution shifting from intersectoral reallocations in the pre-1998 cri- sis subperiod to intra�rm productivity growth in recent years.1 By 2005 real GDP reached almost 90 percent of the pre-reform level of 1990. But that fact conceals wide variations across sectors and regions. Value added in manufacturing was less than 70 percent of the Soviet max- imum. The oil and gas sector reestablished its pre-reform level of output while services—trade, transport, communications, and finance—exceeded it. Those changes in the structure of GDP were accompanied by massive reallocations of inputs across sectors, particularly from manufacturing to such newer sectors as trade and services. The reallocation of labor, though motivated mostly by necessity (massive job destruction in traditional man- ufacturing), transferred resources from the shrinking sectors to the growing sectors (figures 2-1 and 2-2). After the bottoming out of the “transformational recession� in the mid- 1990s, productivity dynamics went in opposite directions. The fastest- growing sectors (oil and gas, market services) had the smallest increases in labor productivity, and the sectors with the deepest initial declines in out- put (manufacturing) saw the biggest subsequent increases in productivity. The explanation for that apparent paradox is that the downsizing that redistributed labor resources from declining to growing sectors also increased the productivity of the former.2 In the first half of the transition, productivity growth in the Russian economy depended mostly on the com- petitive pressure on manufacturing, which had high productivity growth because of significant restructuring through downsizing and exits. Oil and gas and market services, facing less competitive pressure and more market opportunities, expanded employment along with output. So, despite structural changes and resource reallocations in the 1990s, the Russian economy started to grow after eight years of economic decline, with Productivity 15 Figure 2-1. Structure of Employment, 1990a Oil and gas 0.3 (0.4%) Otherb 13.8 (18.3%) Manufacturing 19.7 (26.1%) Agriculture 10 (13.2%) Market services 12.6 (16.7%) Public services 19.1 (25.3%) Sources: Goskomstat (1995) and ILO (2005). a. Millions of workers and percent of labor force. b. Including construction. Figure 2-2. Structure of Employment, 2004a Oil and gas 0.5 (0.8%) Otherb 10 (15.0%) Manufacturing 10.4 (15.6%) Agriculture 7.2 (10.8%) Market services 18.9 (28.4%) Public services 19.5 (29.3%) Sources: Rosstat (2006a); ILO (2005). a. Millions of workers and percent of labor force. b. Including construction. 16 M ARK S CHAFFER AND B ORIS K UZNETSOV overall labor productivity at about 70 percent of the 1990 level. The trigger for the growth that started in 1999 was the 1998 �nancial crisis and deep devaluation of the ruble. Imports became relatively expensive. Labor, energy, and some other material inputs became relatively cheap. And the competi- tiveness of Russian products, particularly in domestic markets, sharply increased as a result. In manufacturing, the growth of output led to higher capacity utilization for �xed capital and labor, which contributed to growing productivity. Subsequent growth has been supported by increases in world market prices for Russia’s key exports—oil, natural gas, and metals. The growth of export revenues, together with a conservative budget policy and higher con- tract and tax discipline, provided for macroeconomic stability and improve- ments in state �nances, the banking system, and corporate �nances. Addi- tional revenues from higher production, higher efficiency, and export activities were distributed through the economy, resulting in steady growth in incomes and eventually in consumer demand—and to a lesser extent in the demand for investment goods. Growth of internal demand became the major driver of the Russian economy, fed by energy and metal exports. The fastest- growing sectors became retail trade, construction, real estate, and market services. Manufacturing also gained from the expansion in demand, though the effect was not as great or as prolonged as for non-tradables. That internally oriented growth indicates the limited competitiveness of Russian manufacturing in world markets, despite annual growth of manu- facturing productivity of 10 percent since 1998. While overall exports more than doubled between 1999 and 2004, it was mostly a result of the increases in oil and gas prices on the world market. The share of manufacturing in total exports remained very low. Even in the internal market, Russian pro- ducers of manufactured goods are beginning to lose market share. Since 2002 the volume of imports has grown much faster than that of exports despite the increases in export prices for Russian oil and gas. The appreciation of the real exchange rate explains that result to some extent, but it also can be explained by the inability of Russian industries to provide enough competitive con- sumer and investment goods to meet growing internal demand. The situation is aggravated by the fact that, in markets for both consumer and investment goods, increases in revenues, incomes, and imported goods lead not only to higher demand but also to higher standards of quality. So although the current Russian trade balance looks extremely healthy, the trends are not favorable: export revenues depend exclusively on world market prices for natural resources, while import growth is a result of the Productivity 17 relatively low competitiveness of manufacturing and will probably continue independent of exports. The competitiveness of the Russian economy and of manufacturing in particular cannot be achieved through a protectionist trade policy that closes the domestic market to international competition. Already a part of the global economy, Russia will be even more integrated after its accession to the WTO. For more balanced growth, Russia needs more diver- si�ed international trade, which can come from a more competitive and more productive manufacturing sector. The Russian Economy and Manufacturing Productivity in the International Context It is useful to benchmark Russia—its economic structure, competitiveness, and productivity—against other countries and groups of countries com- peting directly with Russian producers in the markets or for foreign direct investment (FDI) and against leading developed market economies at the world technological frontier. GDP per capita at purchasing power parity (PPP) is a standard indicator of the aggregate productivity of a country and of its standard of living and level of development. Russia is one of the richer BRICS countries (Brazil, Russia, India, China, and South Africa), comparable to Poland, but its per capita GDP is less than half that of the developed countries, even a “new developed country� such as South Korea (table 2-1). Even with relatively high rates of growth, it will take decades for Russia to catch up with the leaders. Russia is more industrialized than most of the BRICS countries. Its share of employment in agriculture is near South Korea’s and declining quickly: since 1990, employment in agriculture has fallen by more than 20 percent. China and India, the last in the group to start the catching-up process, have the biggest shares of employment in agriculture; even China, with all its achievements in manufacturing and exports, is still a relatively poor and largely rural agricultural economy. A large, low-productivity agricultural sec- tor is the main reason behind the relatively low per capita GDP of these coun- tries. Russia, with a high share of services and industry, could be expected to have a higher level of GDP per capita. The share of exports in GDP shows Russia to be an open economy, little different from the other more open countries in the group: its share is 35 per- cent, against 34 percent in China, 39 percent in Poland, 37 percent in South Korea, and 38 percent in Germany. Where Russia is different is in the struc- ture of exports: more than half of exports are fuels, ores, and metals. Russia Table 2-1. Basic Indicators for Selected Countriesa Indicator Brazil Russia India China South Africa Poland South Korea Spain Germany Population (millions) 184 144 1,080 1,296 46 38 48 43 83 GDP (US$ billions) 604.0 581.4 691.2 1,931.7 212.8 242.3 679.7 1,039.9 2,740.6 2004 GDP per capita in PPP (2000 US$) 7,531 9,101 2,885 5,419 10,286 11,924 18,840 23,019 26,013 Annual GDP growth 2000–04 2.2 6.1 6.2 9.4 3.2 2.9 4.6 3.1 0.7 Percent of employment in agriculture 19.8 10.0 56.7 44.1 10.3 18.4 8.8 5.7 2.5 (2002) (2003) (2000) (2002) (2003) (2003) (2003) (2003) (2003) Exports of goods and services 18.0 35.0 19.1 34.0 26.6 39.0 44.1 25.7 38.0 (percent of GDP) Imports of goods and services 13.4 22.3 22.5 31.4 27.1 41.0 39.7 29.3 33.1 (percent of GDP) Merchandise exports (percent of GDP) 16.0 31.6 10.9 30.7 21.6 30.9 37.3 17.2 33.3 Agricultural raw materials 0.6 1.0 0.1 0.2 0.5 0.4 0.3 0.2 0.3 Food 4.5 0.4 1.1 1.1 1.9 2.6 0.4 2.5 1.4 Fuel 0.7 15.8 0.9 0.7 2.0 1.7 1.5 0.6 0.6 Ores and metals 1.4 2.4 0.7 0.6 4.8 1.3 0.6 0.3 0.8 Manufactures 8.6 6.6 8.0 28.1 12.5 25.0 34.4 13.3 28.0 Unclassi�edb 5.3 0.1 0.1 0.3 2.3 Sources: World Bank (2006e) for population, GDP per capita, GDP growth, and exports; World Bank (2006e) for percent of employment in agriculture, except for the �gure for India, which is from the Indian National Statistical Survey. a. The BRICS group includes Brazil, Russia, India, China, and South Africa, which are considered by world �nancial markets to be a group of large emerging economies expected to grow rapidly in the medium to long term and to compete for international investors. Poland, the largest former transition country in Eastern Europe, is now an EU member. Poland is a typical com- petitor for Russia in European markets and an example of a successful transition. South Korea is a successful Asian Tiger that has an advanced economy; it can be considered a “former emerg- ing economy.� Spain, a developed European country, is a latecomer to the EU that has made major progress in catching up with Europe’s leading economies. Germany is the largest and most advanced of the major EU economies. b. Blank cells represent a negligible amount. Productivity 19 has a lower ratio of manufacturing exports to GDP (7 percent) than all the other countries, comparable only to the ratio for the relatively closed economies of Brazil and India and far behind the ratios for China (28 per- cent), South Korea (34 percent), and even Poland (25 percent). And it has by far the lowest share of manufacturing in overall merchandise exports (about 20 percent against more than 50 percent in any other comparator). The low share of manufacturing in Russian exports is due to the low productivity and poor competitiveness of Russian manufactured goods (table 2-2). International comparisons of aggregate productivity usually rely on comparisons of output per worker, partly for reasons of data—those on fixed capital often are unavailable or unreliable—and partly because studies that compare TFP and labor productivity show the two to be closely correlated.3 Labor productivity in manufacturing in Russia is low even compared with that of most of the BRICS countries. Productivity is 50 percent higher in Brazil, two times higher in Poland, and three times higher in South Africa than in Russia. Value added per worker in Germany is ten times the �gure for Russia. Manufacturing value added per worker is about the same in Russia as in China. Note that the relatively strong ruble in 2004 raises measured man- ufacturing output per worker using current exchange rates and that the adjustment for transfer pricing does the same. So in that respect, estimates of manufacturing productivity in Russia reported in table 2-2 are generous. Benchmarking of TFP in BEEPS and Enterprise Surveys A similar picture emerges when TFP is estimated directly using a very dif- ferent data source, the Business Environment and Economic Performance Survey (BEEPS). Three waves of surveys were conducted, in 1999, 2002, and 2005, covering both manufacturing and nonmanufacturing firms. In 2004–05 the BEEPS was expanded to cover six developed OECD countries. For a number of years the World Bank also has been collecting data on manufacturing firms in various developing countries through the Produc- tivity and Investment Climate Surveys (PICS), now called the Enterprise Surveys. The survey data can be used to estimate TFP levels in manufactur- ing firms in Russia and all the comparator countries considered here. The results should be treated with some caution for a variety of reasons, however. Only data on gross sales, not on value added, are available for all the countries of interest. For the most part, the surveys cover small and medium enterprises, and they may not be representative of aggregate manufacturing. Table 2-2. Productivity Indicatorsa Russia 2004 Brazil India China South Africa Poland South Korea Spain Germany Indicator 2004 Unadjusted Adjusted 2000 2004 2004 2004 2004 2004 2004 Manufacturing value added per 11,094 6,723 7,226 1,908 6,894 21,116 15,532 40,473 50,180 68,640 employee (2004 US$) Manufacturing value added per 153.5 93.0 100.0 26.4 95.4 292.2 214.9 560.1 694.4 949.9 employee (Russia adjusted = 100) TFP estimates for manufacturing 209 100 139 102 350 208 493 369 452 �rms (Russia = 100) based on BEEPS-PICS Median employment 44 50 23 172 95 15 28 20 38 Number of �rms 1,575 185 2,631 2,447 563 562 208 114 218 Annual manufacturing productivity growth (2000–04) Output 5.0 5.9 n.a. 10.6b 1.9 5.6 6.6 1.1 0.6 Employment 7.5 −4.3 n.a. 2.4b 3.5 −1.4 0.0 1.1 −1.2 Productivity −2.3 10.6 n.a. 7.9b −1.5 7.1 6.6 0.0 1.8 Monthly wages in manufacturing $309 $142 $24 $111 $493 $469 $1,484 $2,834 $3,972 in (2002 US$) Memo items Industry value added per 11,855 9,254 12,667 2,209 6,225 19,867 17,034 39,781 50,063 64,723 employee (2004 US$) Industry value added per 93.6 73.1 100.0 17.4 49.1 156.8 134.5 314.1 395.2 511.0 employee (Russia adjusted = 100) Sources: Industry and manufacturing value added: UN national accounts data in local currency units converted at current exchange rates, with estimations for China, India, and Russia, where manufacturing is estimated from manufacturing, mining, electricity, gas, and water in 2004 and manufacturing share of manufacturing, mining, electricity, gas, and water is taken from a national source for an earlier year. India nominal value added data were converted from 2000 to 2004 rupees using the domestic inflation rate. Employment: ILO (2005). Monthly wages in US$: ILO (2005). Russian value added adjusted for transfer pricing effects (World Bank 2005d). a. The table shows industry and manufacturing value added per employee in 2004 at current exchange rates. The exception is India, with productivity for 2000 in 2004 rupees and at the 2004 dollar-rupee exchange rate. The �rst column for Russia uses of�cial Rosstat data, and the other adjusts Rosstat data for transfer pricing effects using the CEM methodol- ogy (World Bank 2005d). Transfer pricing moves pro�ts from industry to trade. Most pro�ts moved are for fuel and energy, only some for manufacturing. Industry is de�ned as manu- facturing, electricity, gas, water, and construction. b. Also includes electricity, gas, and water sectors. Productivity 21 And the sample sizes are small for most countries, including Russia. The results in table 2-2 are from a simple two-input (labor and fixed capital) Cobb-Douglas production function estimation using the TFP level of the �rms in the Russia BEEPS of 2002 and 2005 as the benchmark. Despite all the caveats, the results are very similar to those from the comparisons of aggre- gate manufacturing value added: TFP in Brazilian and Polish small and medium enterprises is about twice that in Russia, and small and medium enterprises in the advanced market economies of South Korea, Spain, and Germany have TFP levels four to five times greater than those of Russian �rms. Once again, Russia’s productivity level is closest to that of China.4 Russia was an evident leader in productivity growth only in 2000–04, but that is not a cause for optimism. The period after the �nancial crisis of 1998 was one of abnormally high productivity growth due to increases in capac- ity use. A reported decline in the share of the shadow economy may have added to measured output growth. Those factors were mostly spent by 2004, and future productivity growth cannot be based on continuation of those trends. Also to be stressed is that Russia is the only country among those considered here where labor shedding has made such a large contribution to productivity growth. Manufacturing output growth in 2000–04 was not exceptional. The low productivity in manufacturing would be less of a concern if it were matched with competitive (low) wages. Comparable cross-country data for wages in manufacturing are available only for 2002. They show that while Russia’s manufacturing sector is about as productive as China’s, it loses the competitive battle in labor costs: Chinese manufacturing wages are 30 per- cent lower than those in Russia. Per dollar of wages, one employee in man- ufacturing produced about $4.00 of value added in Russia, $5.20 in India, and $6.60 in China. But compared with wages in other countries, Russian wages in 2002 were still competitive: in Brazil, one dollar of wages added about $3.00 in value, in Poland about $3.50, and in other countries even less. But that competitive advantage no longer exists. Russia has been expe- riencing rapid wage inflation since 1999, with wages growing far faster than labor productivity. The real wage in total industry in Russia (deflated by the producer price index) has increased by 72 percent since 1999. In 2004 the current dollar monthly wage in industry was over $250, an increase of 67 percent in just two years and a remarkable 266 percent increase over the $75 per month in 1999. Trying to compete internationally with cheap labor in other countries (not only China and India but also Brazil and South Africa) is not an option for 22 M ARK S CHAFFER AND B ORIS K UZNETSOV the Russian manufacturing sector. In Russia, sales per worker are $6,800, well below sales for India ($8,600), China ($10,900), and Brazil ($12,300). But at $2,600 per worker, labor costs are fairly high compared with those in India ($1,300) and China ($1,400). While labor costs in Russia are still lower than those in Brazil ($4,900), real wages in Russian industry have been growing at a rate in excess of 15 percent a year, and that relative competitive advantage will probably disappear in a few years. Dynamics of Employment and Productivity during the Transition (1992–2004) It is well known that Russian manufacturing shrank signi�cantly during the transition, both in relative and absolute terms: its share in total production fell from about 30 percent to 20 percent and it lost about 40 percent of jobs. Did job destruction and creation lead to higher productivity? Did the most ef�cient enterprises create new jobs? Although it would be interesting to see whether the strongest survived and the weakest left the market, we do not have data to analyze enterprises that ceased to exist during the period. The analysis is limited to �rms still operating at the time of our survey and with information from the beginning of the transition. Comparability problems arise for some variables within that subset, so the focus is on a few compara- ble indicators for the whole period. The subsample of firms that can be traced throughout the transition period includes 409 firms, about 40 percent of the total sample.5 The very fact that the �rms survived could be expected to bias the subsample toward better-performing enterprises, but the data do not support that assumption. In 2004 firms traced to the pre-reform period did not differ significantly from the rest of the sample in terms of productivity, so the results for the subsample are valid for at least 80 percent of the firms in the full sample (those created before 1992). The only caveat is that existing firms not rep- resented in the sample or the subsample could be different. However, test- ing or controlling for censored data would require additional data collec- tion outside the scope of this analysis.6 Total employment trended downward through the transformation period: the subsample had 40 percent fewer employees in 2004 than in 1991. Almost three-quarters of the firms (72 percent) lost jobs, and a little more than a quarter (28 percent) increased employment or maintained the same level. In a market economy, a firm’s ability to keep or increase employment over Productivity 23 a long period may be looked on as an indicator of competitiveness, and the factors underlying such competitiveness are worth investigating. Following the terminology in some of the literature (see Blanchflower and Burgess 1996), firms with lower employment are “job-destroyers� and those with higher employment are “job-creators.� Job-creators as a group generated additional employment even in the first half of the 1990s, when the economy was in recession, and generated it faster in the 2000–04 period of fast growth: net employment growth in job-creating �rms increased from 105 percent in 1991–95 to 139 percent in 2000–04. Job destruction was more common in larger enterprises and job creation in smaller ones. In 1991, at the beginning of the transition, aver- age employment was 1,417 per firm in the job-destroying firms and 441 in the job-creating group. By 2004, average employment was 722 and 658, respectively. Job destruction also speeded up over time: net employment growth in job-destroying firms was −70 percent in 1991–95 and −80 per- cent in 2000–04. Declines and increases in employment depended strongly on the sector and industry. In the food industry, 58 percent of firms were job-creators, while in textiles only 8 percent were. In the rest of industry the share of job-creators varied less, from 14 percent in machinery to 23 per- cent in the wood industry. Initially, the smaller job-creating �rms were nevertheless more “capital- ized,� with an average ratio of �xed assets to labor that was 1.4 times higher in 1991. That may be an indication not only of more �xed capital but also of newer and better capital. The difference became smaller during the transi- tion, but it was still 1.2 in 2004. Since employment in the job-creator group grew signi�cantly, the difference would indicate that job-creating �rms also were more active in making new investments. The capital-labor ratio for job- creators in 2004 was slightly higher than that for new firms (those that reported having begun production after 1992). The data on productivity for both groups suggest that Russian manufac- turing went through “creative destruction,� in the sense that net job creation occurred in more productive enterprises. It is worth noting that the produc- tivity of job-creating �rms increased during the transition. In 1992, labor productivity (measured by value added per employee) was 20 percent higher in job-creating �rms than in job-destroying ones. By 2000 it was 56 percent higher, and in 2004 it was 57 percent higher. Redistribution of employment in favor of more productive enterprises resulted in signi�cant growth of their share in total sales. In 1992, job-creators produced 17 percent of total output, 24 M ARK S CHAFFER AND B ORIS K UZNETSOV and in 2004 their share more than doubled, to 37 percent, while their share in employment grew from 12 percent to 30 percent. The employment share of job-creating �rms grew less rapidly than their share of value added produced but more than their share in total output. The creative destruction story has another side, however. Did the destruction of jobs in job-destroyers lead them to become more produc- tive? On the whole, no. Despite job cuts, they did not manage to increase their productivity up to the average rate. In 2004, labor productivity in job- destroying firms (value added per employee) was still 25 percent lower than the sample average and almost 60 percent lower than productivity in job- creating firms. That does not mean that cutting labor cannot help to increase productivity. The labor productivity of about 27 percent of job- destroying firms exceeded the sample average, and the productivity of more than 40 percent exceeded the industrial average. But for 60 percent of firms in the group, downsizing did not help much. There also was a group of evi- dent outliers whose productivity was almost three times lower than average. That group was not small—16 percent of the sample, accounting for 11 per- cent of employment. As seen, the likelihood that a �rm will create jobs in the long run depends in large part on its sector and its endowment of capital inherited from the Soviet era. The trends for the capital-labor ratio indicate that job-creating �rms are also more active in making investments, allowing them to keep their lead in capital-labor ratios, despite increases in employment. Overall, labor resources were redistributed in the right direction during the transition— from less to more productive enterprises, a reallocation that contributed to productivity growth in manufacturing. But downsizing, a passive adjustment strategy, did not lead to productivity improvements in 60 percent of �rms. As a result, about 16 percent of the surveyed enterprises and 40 percent of �rms traced to the pre-reform period were still falling behind, and the gap between that group and the rest of the industry was widening. Industrial Heterogeneity The 2005 Country Economic Memorandum (CEM) for Russia argues that the major source of future productivity improvements must come from real- location of resources between incumbent �rms and newer, more productive �rms in leading sectors of the economy. Restructuring and the entry of com- petitive �rms would drive such intrasectoral reallocation forward.7 In order to characterize the nature and strength of the links between �rm age and Productivity 25 Figure 2-3. Firm Creation in Russian Manufacturing, 1705–2005 Density .015 .01 .005 0 2005 1905 1805 1705 Year firm established Source: LME Survey (2005). productivity, the analysis in this chapter used the LME survey to investigate the heterogeneity of newer and older firms in productivity and in other characteristics. The LME sample of medium and large manufacturing firms (see appen- dix 2) consists of two main groups of firms, new entrants during the tran- sition and incumbent firms established in the Soviet era (figure 2-3). There also is a third, smaller group, of firms that date to before the twentieth cen- tury. A statistical description of that age profile would note bimodal distri- bution that peaks in the early 1990s and in the mid-twentieth century, with a fat tail. The origins of each of those waves of industrial entry reflect marked shifts in the economic system and industrial policy. Some caution is needed on the year of establishment, as managers may have reported a more recent (but still Soviet era) date for some factories constructed in the nineteenth century. The productivity of enterprises in the sample (measured by value added per worker) has a much more regular pro�le, approximately log normal. That implies that while there is a great deal of heterogeneity in the productivity of 26 M ARK S CHAFFER AND B ORIS K UZNETSOV Russian manufacturing �rms, there are no distinct industrial clusters with different average productivities. Indeed, many other �rm characteristics also are distributed as log normal, including the size of the �rm’s workforce, the size of investments in �xed capital assets (or in machinery and equipment), the proportion of employees having attained a higher educational degree, and the amounts spent on R&D (see appendix 2). Having found that labor productivity is approximately log normal, we checked whether younger firms entering during the transition were more productive than the incumbents from the Soviet and tsarist era. To test that hypothesis, we compared the labor productivity distributions for the �rms established before and after 1990. They are not signi�cantly different.8 But one has to interpret that result with care because the sample does not include small manufacturing firms, arguably the new entrants into sectors with higher productivity. In addition, most or nearly all �rms in the sample were privatized �rms; very few were new private �rms. Because of the relationship between �rm age, size, and productivity, it is not clear that newer �rms were more productive than older �rms. The reason is that labor productivity is closely connected to �rm size—larger �rms are more capital intensive and therefore would have higher labor productivity. And as older �rms tend to be large, a new �rm may have higher labor pro- ductivity than an older �rm of the same size, but the average new �rm is likely to have lower labor productivity than the average old �rm.9 The analysis here does not support the hypothesis that the productivity pro�le is different for older and newer �rms. In other words, the labor pro- ductivity distributions of the Russian manufacturing �rms entering at dif- ferent periods are statistically indistinguishable. Because of that, we believe that it is a reasonable approximation to frame the policy discussion in terms of measures that can support an “average� manufacturing �rm as opposed to a target group with precise characteristics, though some selection problems and biases appear.10 Determinants of Productivity Growth After fifteen years of transformation, the Russian productivity gap today cannot be explained simply by reference to transition shocks or to the in- efficiencies inherited from the Soviet period. The relatively low productivity in Russia is even more puzzling in light of the country’s rich natural resources and rich human capital, evident in an educated labor force and a history of major scienti�c and technological advances. Productivity 27 Absorption and R&D Expenditure Much micro- and macroeconomic evidence now exists on the links between innovative activity, R&D expenditure, and productivity growth. Another link relevant to this study concerns the relationship between trade and absorption: openness to trade is especially important in facilitating the transfer of technology between developed, middle-income, and developing countries (chapter 3). Not surprisingly, microeconomic data for Russia indicate a clear correla- tion between innovative activity by �rms and their productivity level and growth. The Russia LME Survey of approximately 1,000 medium and large manufacturing �rms collected information on a variety of measures of inno- vative activity: introduction of products and technology, acquisition of new technology, R&D expenditures, and so on. Simple log-linear estimations show that innovative activity, however measured, tends to be associated with �rms that have high or growing levels of total factor productivity—indeed, with �rms that are employing growing numbers of workers or that are invest- ing and expanding their levels of �xed capital (see appendix 2 for details). The puzzle in Russia is not the absence of a connection between technol- ogy adoption and growth at the micro level but in the low productivity of Russia’s R&D sector. Simply put, although Russia has a large R&D sector, its output is unimpressive. Despite devoting signi�cant resources at the aggre- gate level to R&D and innovative activity, Russia’s current record of knowl- edge absorption is weak. And as shown above, it has not translated into higher levels of TFP. The Soviet R&D legacy should have been a blessing. Russia began the tran- sition with an R&D sector that was large by international standards and with a history of major technological achievements in its space program and mil- itary technology. UNESCO �gures for the USSR in 1990 show the vast inputs devoted to R&D: spending of more than 5 percent of GDP and close to 6,000 researchers per million of population. Both measures of inputs are double the number found today in the advanced and technology-intensive economies of Germany and South Korea (�gure 2-4). Despite the significant decline of the R&D sector during the transition, the scale of inputs into R&D activity in Russia is still substantial. The num- ber of researchers per million of population has halved, but that has taken Russia down only to the internationally high levels of Germany and South Korea. Spending on R&D has fallen substantially, to about 1 percent of GDP, but that amount is still above the levels observed in most of the other BRICS countries or in the EU 8. The definition of R&D inputs in these 28 M ARK S CHAFFER AND B ORIS K UZNETSOV Figure 2-4. Manufacturing Value Added per Worker and Researchers per Million Population Manufacturing value added per worker (US$) $70,000 Germany $60,000 Spain $50,000 South Korea $40,000 $30,000 South Africa $20,000 Poland Brazil $10,000 Russia China $0 India 0 500 1,000 1,500 2,000 2,500 3,000 3,500 Researchers per million population Source: World Bank (2002–2006); OECD (2004b). international comparisons tends to understate the scale of R&D activity because it cannot fully capture the inputs devoted to imitating and adopt- ing new technologies (see Keller 2004). The coexistence of a large R&D sphere and low productivity in manufac- turing indicates either low productivity in R&D institutions or weak links between R&D and the economy. Some evidence supports the �rst hypothe- sis (see table A2-2 in appendix 2). The number of scienti�c publications per thousand researchers in Russia is about the same as in China—around thirty—but significantly lower than in the other comparator countries. Researchers in Poland, India, Brazil, and South Korea are generating two to three times more scienti�c publications per person and German and Spanish researchers about six times more. A very similar picture emerges with another indicator of the productivity of R&D spending—the number of U.S. patents per thousand researchers (see Jaffe and Trajtenberg 2002). Russia is at the same level as China and Poland, but Spanish researchers generate almost ten times more U.S. patents per per- son, Korean researchers sixty times more, and German researchers one hun- dred times more. The second hypothesis—poor links between R&D expen- diture and industry—also is supported by the evidence, in this case by the Productivity 29 high share of state-�nanced R&D expenditure and very low R&D activity at the �rm level (see OECD 2005b). Skills and Human Capital A wide consensus now exists on the role and importance of human capital in the generation of economic and productivity growth, and considerable research since the late 1980s has explored the connection (see Krueger and Lindahl 2001 for a review). The connection is evident in the Russia LME Survey as well. Using the same methodology as used above for R&D activity, it shows that firms that have formal training programs—or those whose workforce includes a greater percentage of employees with higher education— have higher productivity. The differences are large and signi�cant. Either a 20 percentage point difference in the share of employees with higher educa- tion or the existence of a formal training program is associated with 25 per- cent higher TFP (see appendix table A2-1 for further details). As with R&D expenditure, the puzzle is not the connection among human capital, skills, and productivity at the microeconomic level—it is how Russia compares at the macroeconomic level. Enrollment rates in formal education in Russia are at least as high as in developed countries, not only for second- ary school but also for institutions of higher learning. A very large portion of Russian students complete some form of university education—in the set of comparator countries considered here, only South Korea sends more students to university. The same picture emerges when workforce education and skills are considered: across the comparator countries, the micro-level BEEPS- Enterprise surveys show that Russian manufacturing firms have the most employees with higher education, the most managers with a university or higher degree, and skilled labor forces comparable to those in the leading countries used for benchmarking. But the higher levels of skills and education in Russia are not being trans- formed into higher productivity. Russia is a clear outlier: other countries have either high productivity and a highly educated population or low productiv- ity and a less educated population (�gure 2-5). Russia (and to a lesser extent Poland, which is emerging from a similar economic system) combines high levels of skills and education with low levels of productivity. The lack of a correlation between education and productivity in Russia is due to several factors. First, formal education indicators do not neces- sarily reflect employees’ actual qualifications for their job or their skills. Second, many educated Russians are in some sense overquali�ed for the posi- tions that they occupy—too many resources are going into producing grad- 30 M ARK S CHAFFER AND B ORIS K UZNETSOV Figure 2-5. Enrollment in Higher Education and Value Added per Employee in Manufacturing Manufacturing value added per worker (US$, thousands) 80 70 Germany 60 Spain 50 South Korea 40 30 South Africa 20 Poland Brazil 10 China Russia India 0 0 10 20 30 40 50 60 70 80 90 Enrollment in higher education (percent of population) Source: UNESCO (various years) for enrollment figures; table 2-2 for value added per employee. uates with skills and education that are not suitable for a low-productivity emerging economy. Third, the labor force training institutions are rela- tively ineffective—more suitable for generating workers for a centrally planned economy than for imparting skills for a dynamic market econ- omy (chapter 4). Investment Climate If Russia is pursuing innovative activity on a large scale and has a highly edu- cated workforce, why is productivity not higher? One answer is obvious: the R&D and human capital capacities are ineffective because they are deployed in an environment characterized by weak investment and ineffective eco- nomic institutions. The investment climate has many aspects, and measuring its quality is challenging, especially across countries. The analysis here exploits the large- scale data collection conducted for the BEEPS-Enterprise surveys and uses a standard section in the questionnaires in which managers are asked about the constraints facing their businesses with regard to infrastructure, state reg- ulations, macroeconomic stability, legal system, corruption, and so on. The Productivity 31 results reported here refer to manufacturing �rms only. A full set of compar- isons is reported in appendix table A2-4 and includes regression analysis of the relative importance of different investment climate constraints in the selected countries. A very useful feature of the data is that they include results for three separate surveys conducted in Russia, in 1999, 2002, and 2005. That allows for assessment of changes in the investment climate from the aftermath of the financial crisis of 1998 to the current period. Since 1999 the business environment in Russia has, in the perceptions of managers, improved sig- nificantly. The changes in some of the categories are a natural consequence of economic growth in the recovery period (access to finance, macroeco- nomic environment), and changes in others are due to economic reform (tax burden, licensing) and better state regulation. But most successes were achieved between 1999 and 2002. From 2002 to 2005 the situation changed little or even deteriorated. The danger here is that favorable improvements in the macroeconomic situation and in government finances may have led the government to reduce the priority given to improving the business cli- mate further. That is a serious problem. Comparisons with other countries show that Russia still has a long way to go: the Russian business environment in many aspects is signi�cantly worse than in the most developed benchmark coun- tries (South Korea, Spain, and Germany). The most serious gaps between Russia and those countries concern the effectiveness of the legal system, cor- ruption, crime, and political stability (�gure 2-6). Compared with that of the poorer but rapidly growing BRICS countries, the Russian investment climate presents a mixed picture. Access to infra- structure—electricity, telecommunications, transport—is less of an obstacle in Russia than in India and China (see appendix table A2-3). That is expected, given that the two countries are still largely agricultural, whereas Russia has been largely industrial for many years. But Russian �rms cite pol- icy uncertainty and tax administration as bigger problems than their Indian and Chinese counterparts (chapter 5). Conclusions Russian productivity has been growing steadily since 1999—but from a very low level, especially in manufacturing. The sources of productivity and output growth lay in the sharp devaluation of the ruble in 1998, the decline of nonpayments and barter, the growth of capacity utilization, and 32 M ARK S CHAFFER AND B ORIS K UZNETSOV Figure 2-6. Ranking of Selected Investment Climate Constraints, Russia and Developed Economiesa Ranking 3.5 Russia 2005 South Korea 2004 3 Spain 2005 Germany 2004 2.5 2 1.5 1 Tax rates Licensing Policy Legal system Corruption Crime uncertainty Source: BEEPS (2006). a. Ranking on a scale of 1 (not an obstacle) to 4 (a major obstacle). the contraction of the informal shadow economy. Those factors were tran- sient, so the growth that they supported may not be sustainable. Produc- tivity in Russian manufacturing is lower than in most BRICS countries and uncompetitive because of rising real and dollar wages. The lack of competitiveness is clearly visible in the structure of Russian exports. While Russia is very much a part of the global economy, as evidenced by its high ratio of trade to GDP, the manufacturing sector produces mostly for the internal Russian market. Exports consist largely of fuel (oil and gas) and raw materials, and manufacturing exports as a share of GDP are very low by the standards of both other major emerging economies and developed market economies. The paradox is that productivity in Russia is very low despite the signi�- cant resources devoted to R&D and to education and training of the work- force. The scale of inputs devoted to those activities is as high as in the lead- ing industrial countries, but it results in productivity that is below that of most BRICS countries and far from that in the leaders. The investment cli- mate, measured directly by surveys of managers, points to major improve- ments since the surveys started in 1999. Most of the improvements, however, Productivity 33 were experienced in the immediate post-crisis period, and the rate of improvement between 2002 and the most recent survey in 2005 ranges from small to negligible or even negative. The overall quality of the investment cli- mate is much worse than in the developed countries—and overall no better than in the poorer developing countries. Notes 1. For a quantitative analysis, see the World Bank Country Economic Memoran- dum (CEM) for Russia (World Bank 2005d). 2. The structural changes between aggregate sectors of the Russian economy were to some extent distorted by the large-scale practice of transfer pricing, which dimin- ished industrial output and exaggerated the share of trade. Nevertheless, the main trends in sectoral productivity stand even after correcting output for transfer pricing. More detailed analysis is in the CEM (World Bank 2005d). 3. That correlation reflects the fact that capital-output ratios vary much less across countries than do TFP levels. 4. The outlier country in the TFP results is India: �rms in the Enterprise Survey have much higher levels of productivity than would be expected given the aggregate manufacturing output per worker. The likely explanation is that India has a large informal sector in manufacturing that has very low levels of productivity and is dif�- cult to capture in a survey. 5. That does not mean that the rest of the sample consists of newly established �rms. Actually, another 40 percent of �rms in the sample started before 1992, but they cannot be traced back to the Soviet period because they changed their name or registration code or because they were not included in the available databases. This is a classic problem in constructing this type of panel data. 6. One way to collect the data would be to construct a sample representative of large and medium manufacturing �rms in the base year, large enough to include a siz- able proportion or all of the 409 �rms in the LME Survey used in the present analy- sis. Then, by tracing the evolution of the �rms forward, it would be possible to iden- tify the characteristics of exiting firms and use that information to control for censored data. For a discussion of different approaches to collecting and constructing data sets for �rm dynamics studies that compares the LME data set with other types of �rm-level data, see Haltiwanger and Schweiger (2004). Another example of this kind of study is Scarpetta and others (2002). 7. The CEM (World Bank 2005d, p. 77) advances the following idea about the heterogeneity of productivity across Russian firms: “Given that the productivity improvements in Russia are likely to continue to be located within sectors, further restructuring within the largest sectors is crucial to unlocking growth. Russia’s development, like that of other transition economies, will probably be dominated by competition between new, initially small and medium-size, but highly productive, 34 M ARK S CHAFFER AND B ORIS K UZNETSOV enterprises and old enterprises inherited from the previous system, with lower value added per worker and often in need of downsizing.� 8. That �nding is corroborated by the Kolmogorov-Smirnov statistic, which tests for the equality between two distribution functions. 9. Appendix table A3-7 shows that the dummy variable designating �rms estab- lished after 1992 in which no share of common stock is owned by the government is not signi�cantly correlated with the likelihood of engaging in R&D. Similarly, tables A3-2 and A3-3 indicate no signi�cant correlation with the probability of introducing a new product or technology. 10. For example, the discussion of matching grants outlines general principles for program design that do not explicitly target speci�c groups of �rms, but it also pays attention to selection problems that could lead to wastage and crowding out of R&D and how they could be resolved. three Fostering Knowledge Absorption I TZHAK G OLDBERG , E NRIQUE B LANCO -A RMAS , J OHN G ABRIEL G ODDARD , AND S MITA K URIAKOSE The Russian Federation devotes considerable resources and manpower to research and development (R&D), yet the Russian economy lags behind other large OECD and middle-income economies in R&D-based outputs. Micro- economic evidence shows that firms facing stiffer competitive pressures also innovate more—and that firm-level R&D has a strong, positive, and signif- icant association with technological and organizational innovation and knowledge absorption. The Russian government’s initiatives to promote innovation do not appear to recognize the signi�cant role of competition in fostering innovation and absorption. Improving absorptive capacity—the ability to tap the knowledge in the world technology pool—can be a major driver of increased productivity growth. Yet that capacity is contingent on the effectiveness of trade flows, labor mobility, licensing agreements, and foreign direct investment. It also requires a good investment climate, education, and domestic R&D. This chapter argues that increasing the capacity and incentives of private �rms to absorb knowledge and to innovate is critical. Using an empirical study of underlying �rm characteristics, investments, and environmental conditions, it shows a positive and signi�cant correlation between a �rm’s likelihood to absorb or innovate and the competitive pressures in its product market. The distinction between knowledge absorption (introducing products and technologies new to the firm) and R&D/innovation (creating capital goods and final outputs new to the world) is important in a large economy like 35 36 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE Russia’s. Although this chapter focuses on absorption, the complementarities between absorptive capacity and R&D/innovation are extensively supported by theoretical and empirical work, so R&D/innovation also is discussed at some length. Specifically, the empirical work deals with a wide range of absorptive activities of �rms, including the introduction of new and improved products and manufacturing technology, exports of technology-intensive goods, orga- nizational change and outsourcing, and soft innovation through quality con- trols, certi�cation, and automated systems. This chapter first discusses the specific characteristics and relative strengths of domestic sources of knowledge absorption, including R&D and intellectual property, through international benchmarking. Cross- country comparisons are used to pinpoint potential obstacles to foreign sources of absorption and innovation operating through trade flows and foreign direct investment (FDI), and the determinants of R&D expendi- tures and absorption-related activities in Russian manufacturing firms are identified and also compared with international results. In conclusion, poli- cies are presented that the government could undertake to support com- mercial efforts at knowledge absorption and innovation. Domestic Sources of Knowledge Absorption and International Benchmarking By international standards, Russia spends large amounts of money—and employs many workers—in research and development. Indeed, Russia now has as many researchers per capita as the R&D-intensive economies of South Korea and Germany. Domestic spending on R&D in Russia, even after plum- meting in the early 1990s, was about 1.2 percent of the country’s GDP in 2004—equivalent to China’s but higher than that of other developing economies such as Brazil, India, and South Africa (0.9 percent, 0.8 percent, 0.7 percent respectively).1 Much of the expenditure is the legacy of Soviet socialism, under which industrial R&D grew large. The persisting effects, positive and negative, of that legacy have been discussed widely, and a recent World Bank report outlines some of the remaining inefficiencies: “Currently, many of the S&T (science and technology) resources are isolated both bureaucratically (in the sense that they are deployed in the rigid hierarchical system devised in the 1920s to mobilize resources for rapid state-planned industrial devel- opment and national defense), functionally (in the sense that there are few Fostering Knowledge Absorption 37 Figure 3-1. Productivity of Russian Researchers 4.5 1000 4 Patents per million populationb 3.5 Patents per dollar of GDPa 100 3 2.5 10 2 1.5 1 1 0.5 0 0.1 de d n G a Re ce lic a Hu n C y an a n a Ge ce y n ar an n tio in ric ai pe ad io re pa So pub la ec ree an Ch Sp Un ng Ko rm ra ro an Ja Af Po Fr h ut Fe h n ia Cz Eu ss Ru Source: World Bank (2006e) and OECD (2005b). a. At purchasing power parity (2000). b. Log scale. links between the supply of S&T output by research institutes and the demand for S&T by Russian or foreign enterprises), and geographically (in the sense that many assets are located in formerly closed cities or isolated science/atomic cities). Overcoming the inefficiencies embedded in these sunk costs incurred during the socialist period and adjusting the S&T system to the demands of a market economy will require a major program of insti- tutional and enterprise reform which, in turn, will make the task more daunting, although no less necessary.�2 Despite the size of the R&D effort in both spending and personnel, Russia’s manufacturing productivity has not benefited (figure 3-1). Based on the number of researchers, Russia’s productivity should be among the highest—on par with that of Germany and South Korea. Instead, Russia’s R&D activities fall well short of their potential. The composition of R&D spending by economic sector changed in the 1990s, but it still remains financed largely by the government. In 2003 the Russian government was still financing 58 percent of R&D expenditures (the OECD average was 30 percent in 2002) and Russian industry �nanced 38 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE 23 percent of total R&D outlays (the OECD average was 62 percent).3 Pri- vate expenditures on R&D changed very little between 1999 and 2007.4 Such imbalances have been underlined as a major problem in a recent study: “The Achilles’ heel of the Russian innovation system is the weakness of corporate R&D, despite some encouraging developments over the last two to three years. The efforts to transfer near-market research from pub- lic organizations to business firms, to promote the creation of technology- based firms, to encourage private investment in R&D, and to attract R&D intensive foreign investment have not been entirely successful. The reasons are many but the fact is that business enterprises contributed no more than 20 percent of total R&D expenditure in 2002.�5 Those expenditures did not seem to translate into R&D-related “outputs�— at least of the kind that can be measured and compared across countries, such as the number of international patents accumulated or the number of articles published in scientific journals.6 The number of patent applica- tions to the patent offices in the United States, the EU, and Japan per unit of GDP (0.1 patent per unit of GDP) was significantly lower than in OECD countries (4 patents per unit of GDP in Japan and 3.5 per unit of GDP in Germany).7 Clearly, patents are only an imperfect proxy for cutting- edge innovation, given that much innovation takes place without patenting—improvements in the manufacturing process, for example, normally fall outside the domain of patents. Patents are an even worse proxy for knowledge absorption. Nevertheless, we believe that patents are a useful indicator of innovation and that registration in overseas markets is a more useful indicator of Russian innovation than patenting within Russia, for two reasons: first, Russian entrepreneurs tend to avoid register- ing patents locally due to weak enforcement of intellectual property rights (IPR) in Russia; second, they lack confidence that a Russian patent ensures international protection. The number of “triadic� patent applications by Russian entities to U.S., EU, and Japanese patent offices is significantly lower than the number by entities in OECD countries and in several non- OECD countries.8 Royalty and licensing fees paid by Russian entities in 2004 reached US$1 billion, an amount that was roughly 0.2 percent of GDP and equiv- alent to the amount paid in China and several other middle-income coun- tries. Licensing and royalty fees received by Russian entities during the same time, however, were only US$220 million. That low figure, together with the modest patenting activity, suggests that the international com- mercial value so far realized by Russian R&D expenditures is limited.9 Fostering Knowledge Absorption 39 Figure 3-2. R&D Expenditure, by Economic Sector Percent 120 100 Industry 25 Defense and space 35 80 Basic research Agriculture Energy 60 33 Other 30 40 12 25 20 4 3 3 4 16 10 0 1994 2002 Source: HSE (2005). The composition of R&D expenditure by economic sector, though, changed in the 1990s. A breakdown shows reduced emphasis on applied R&D for industrial purposes and, to a lesser degree, for agriculture (�gure 3-2). R&D expenditure on defense and aerospace increased from 1994 to 2002, when expenditures on basic research more than doubled. But those expen- ditures did not seem to translate directly into standard indicators of research activity, such as the number of patents accumulated (per thousand researchers) or the number of articles by Russian researchers published in scientific journals. The description of the Russian R&D sector in a 1994 OECD report cap- tures fundamental characteristics that still affect the size and structure of the 40 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE sector: “Once established, R&D organizations grew inexorably, following the pattern of extensive growth typical of the whole economy. There is no doubt that, in relation to the scale of the economy and its real level of development, Russia now has an excessively large S&T sector.�10 That observation is largely consistent with the data in chapter 2: the Russian R&D sector seems to be quite developed but surprisingly inef�cient. The OECD report identi�ed as a priority the need to downsize what it termed an “oversized, ill-adapted system in rapid deterioration.� Under the Soviet S&T system, the objective of servicing large state-owned enterprises (SOEs) and the defense complex undeniably molded the Russian innovation system.11 Research was done mainly in specialized technology institutes that reported to their respective line ministries, subject to the cen- tral plan. Research institutes were not in universities, and they developed very narrow areas of specialization because lateral interaction between innovative agents was not encouraged by the central planning apparatus. Intellectual property generated anywhere was formally financed by the gov- ernment and so belonged to the state. Dissemination of new knowledge, know-how, or technological advances from the most advanced establish- ments within the Soviet system—the isolated defense-oriented “Yaschik�— was prohibited. Financing for research establishments was allocated largely on the basis of such input indicators as the number of scientists and not necessarily the research program. Historically the incentives for collaboration between most government-funded research establishments and enterprises were very weak. As a result, following the dismantling of the military complex and the priva- tization of the SOEs, the innovation system that remained could hardly meet the needs of the private sector. In sum, the dif�culties inherent in transitioning from a state-�nanced to a market-based innovation system still seem to constrain Russia’s knowledge absorption capability. And private �nancing of R&D seems trapped at pre- transition levels. Foreign Sources of Knowledge Absorption Two potential channels of knowledge absorption and innovation are impor- tant for Russian industry:12 —Trade. Enterprises absorb knowledge through imported capital goods and technological inputs. Exporters bene�t from their trade with suppliers and clients in more advanced markets through learning effects. Fostering Knowledge Absorption 41 —FDI. Technology spillovers from foreign investors to suppliers and clients are bene�cial for the destination country, and foreign entry by itself can increase knowledge absorption by competitors. Trade The diffusion of knowledge and technology through trade can be increased in two major ways: an increase in trade volume and an increase in the qual- ity of trade (trading in goods with higher capital intensity).13 Gains from international diffusion of technology amplify the overall productivity gain with greater openness. But it is not only the quantity of trade that matters— trading partners also make a difference. The standard argument is that importing new vintages of capital goods can directly increase manufacturing productivity by increasing the capital- labor ratio and by raising the average quality of the capital stock. Conse- quently, a larger share of capital good imports from more advanced countries would be positively associated with industrial productivity and economic growth. In 2004 the Russian Federation imported $33.8 billion in capital goods, or 28 percent of total imports.14 In 2005 the top �ve countries export- ing capital goods to Russia were Germany ($9.3 billion, for 29.3 percent of the total), Finland ($3.3 billion, for 8 percent), the Netherlands ($2.6 billion, for 6.3 percent), Italy ($2.5 billion, for 7.2 percent), and China ($1.8 billion, for 4.4 percent). Overall, a large majority of Russia’s imports of capital goods have come from partners that are economically and technologically more advanced than Russia. In addition, simply trading with countries that have a more advanced knowledge and technology base—whether in capital goods or other commodities—can generate positive spillover in “learning� from buyers (for exports) or sellers (for imports).15 As an approximate indicator of the potential supply and demand for learning, table 3-1 presents the volume of imports and exports for Russia’s major trading partners and table 3-2 shows the top commodities in the trade. The tables point to the substan- tial share of Russia’s exports to and imports from the EU and the United States; the concentration of exports in oil, gas, and natural resources; and the importance of capital goods. Because of Russia’s heavy reliance on exports of natural resources, the overall volume of trade is probably not a very good indicator of the potential for knowledge diffusion. It is more useful to look at the trade in parts and components with countries already well integrated into global production sharing networks, because trade in such intermediate goods can facilitate the 42 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE Table 3-1. Russian Exports and Imports, by Country of Origin and Destination, 2005 Exports Destination Trade value (US$ billions) Germany 19.9 USA 16.1 China 15.9 Turkey 12.9 Ukraine 12.8 Other reporters 138.2 Total exports 215.9 Imports Origin Trade value (US$ billions) Germany 21.0 China 13.2 Italy 7.5 Ukraine 7.5 Finland 7.1 Other reporters 63.7 Total imports 120.1 Source: UNCTAD (2006a). Table 3-2. Russian Exports and Imports, by Type of Commodity, 2005 Top exports Description Trade value (US$ billions) Fuels and lubricants, primary 98.4 Industrial supplies, processed 54.5 Fuels and lubricants, processed 30.9 Industrial supplies, primary 11.3 Food and beverages, processed 3.7 Other commodities 17.2 Top imports Description Trade value (US$ billions) Industrial supplies, processed 25.3 Capital goods (except transport equipment) 24.2 Food and beverages, processed 12.9 Consumption goods, semi-durable 12.1 Parts and accessories of capital goods (except transport equipment) 9.6 Other commodities 36.0 Source: UNCTAD (2006a). Fostering Knowledge Absorption 43 acquisition of new technology through vertical knowledge spillover between suppliers and customers operating in those networks.16 It also is useful to focus attention on that type of trade because some of the literature has shown that “countries that promote exports of more sophisticated goods grow faster.�17 Macroeconomic data show that most BRICS countries and some com- parators, like Germany and Poland, are more integrated than Russia into global production networks. The growth of China’s share of parts and com- ponents is exceptional, but all other BRICS as well as the other comparator countries also are exporting more parts and components, as a share of GDP, than Russia. The picture is similar for imports, where the evidence shows that Russia is only slightly ahead of India. Moreover, imports in parts and com- ponents are larger than exports, suggesting that Russian firms are not yet competitive enough in this area. As mentioned, it is not only the amount of trade that helps a country to become better integrated in global production networks; trading partners also matter. Although the total flow of imports to and, more important, of exports from Russia is to a large extent directed from and to high-income OECD countries, trade in parts and components tells a very different story. While electrical parts and components are mostly imported from high- income OECD countries, Russia is exporting the vast majority of its own parts and components to countries in the Commonwealth of Independent States (CIS).18 Using exports in parts and components as a litmus test for integration into global production networks, our evidence suggests that Russia’s integration into those networks is less than complete. If it is true that exports can boost the spread of new manufacturing methods though learning from buyers, the data highlight the possible weakness of this channel. Foreign Direct Investment FDI can promote diffusion of international technology if the technologi- cal advantages of multinational firms do not remain restricted to one firm or its affiliates.19 Technological spillover may take place in the recipient country through demonstration effects (imitation), labor turnover, or increased competition. The channels for spillover can be horizontal (orig- inating in the entry of the multinational investor in the same sector) or vertical (originating in backward links when local suppliers supply multi- national investors or in forward links when local customers buy from multinational investors). 44 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE Because a multinational has more incentive to promote local suppliers, backward links may be more widely observed than horizontal links, which inherently are associated with increased competition. Research by Javorcik (2004) �nds a positive effect of FDI on Lithuanian local suppliers working through backward links. Another �nding is that greater productivity gains are associated with projects partially owned by foreign entities (joint ventures), suggesting that domestic capital participation increases productivity spill- over. That could serve as an interesting model for Russia: it might be able to attract FDI to complement domestic capital ownership and thereby increase productivity without yielding domestic �rms’ market share to direct foreign competition. The evidence from Russia indicates that FDI inflows are lagging behind those of some BRICS comparators (see box 3-1),20 suggesting that the ben- efits from international technology diffusion have flowed to only a few eco- nomic sectors, with FDI heavily concentrated in oil and natural resources.21 At the same time, the large increases in FDI observed since 2002 suggest a growing balance within manufacturing and between the manufacturing and service sectors, indicating that the growing domestic demand for consumer goods is driving a consistent share of total FDI flows in Russia.22 Russian multinationals continue to dominate the outward FDI of the southeastern Europe and CIS region for FDI in joint ventures and mergers, accounting for 87 percent of the total in 2005.23 Investment includes large deals to acquire and create joint ventures with enterprises in developed economies—notably Lukoil’s purchase of Nelson Resources, a Canadian- based oil company, and the recently announced merger of the aluminum and alumina assets of RUSAL, the SUAL Group, and Glencore Interna- tional. Such partnerships are likely to gain in importance, given evidence that foreign-invested enterprises (FIEs) have been important for increasing labor productivity and export competitiveness in such countries as China.24 One of the most powerful channels for technology diffusion is the infor- mation and communications technology (ICT) channel.25 Usually, a good indicator of a country’s capacity to leverage the ICT channel is the amount of FDI in communications, which in Russia remains extremely low (0.4 percent of total annual FDI in 2004–05).26 Moreover, according to various private sources, such as WITSA (2006), the amount of ICT investment in Russia, as a percent of GDP, is substantially lower than that in Central and Eastern Europe (CEE) countries. Note that neither trade in capital goods nor FDI inflows are suf�cient for the successful diffusion of technology. A country must be ready to absorb Fostering Knowledge Absorption 45 Box 3-1. Gains from Reducing Barriers to Trade and to FDI Flows in Russia Russia stands only to gain by reducing its barriers to trade and FDI inflows and thus reaping the bene�ts of global integration, increased competitiveness, and improved access to business services. Tariff barriers: The CIS has high average tariff and nontariff barriers, which would need to be reduced in the medium term in order to gain from inter- national integration. For Russia in particular, Rutherford and Tarr (2006) shows that the average tariff increased between 2001 and 2003 from 11.5 percent to between 13 and 14.5 percent, placing its tariff rates (unweighted, or weighted averages) at a higher level than those of other middle-income countries, which average 10.6 percent. A reduction in the import tariff by 50 percent will produce gains to the economy on two counts: one, improved domestic resource allocation due to a shift in production to sectors where the value of production is higher, based on world market prices, and two, an increase in Russian productivity as a result of Russian businesses being able to import modern technologies. The second impact is more important for Russia. Trade Restrictiveness Index: Kee, Nicita, and Olarreaga (2006) computes indicators of trade restrictiveness that include measures of tariff and non- tariff barriers for ninety-one developing and industrial countries. Of the indi- cators for the manufacturing sector, one focuses on the trade distortions imposed by each country on imports and another focuses on market access for exports in the rest of the world. It is interesting to note that the trade restrictiveness index (TRI) for imports for Russia (.19) is lower than that for Brazil (.22) and India (.20) but higher than that for South Africa (.06), China (.12), and the European Union (.08). Russia’s TRI is the highest in the Europe and Central Asia region, a reflection of the high tariff and nontariff barriers that it imposes on its imports. On the other hand, Russia faces less trade distortion on its exports from the rest of the world, China being the only country facing a lower level of restrictiveness. Barriers to FDI: Russia fares worse than other countries in the region, attracting one of the lowest levels of FDI inflow. Among the key restrictions on foreign service providers in Russia are the monopoly of Rostelecom on �xed-line telephone services, the prohibition of af�liate branches of foreign banks, and the restricted quota on the share of multinationals in the insurance sector. The reduction of barriers to FDI in services alone would result in a (continued) 46 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE Box 3-1. Gains from Reducing Barriers to Trade and to FDI Flows in Russia (Continued ) gain of the order of 3.7 percent of GDP, accounting for about three-quarters of the total gains to Russia from WTO accession. The reduction in barriers to FDI in the service sector would allow multinationals to obtain greater post- tax bene�ts on their investments, encouraging them to increase FDI to supply the Russian market. That in turn would lead to an increase in total service providers in Russia, giving Russian users improved access to telecommuni- cation, banking, insurance, and other business services; lowering the cost of doing business and increasing the productivity of Russian �rms using those services; and providing a growth impetus to the economy. Sources: Rutherford and Tarr (2006); World Bank (2005b); Kee, Nicita, and Olarreaga (2006). foreign knowledge and manufacturing and sales methods. In Russia vertical knowledge spillover could be hampered by the central planning legacy of large industrial plants, which were more vertically integrated than Western plants.27 Moreover, local R&D, domestic research laboratories, and workers with the right skills are key aspects of the process. Trade competition and R&D expenditures are closely interrelated: unless a country is also well endowed in R&D and invests in R&D, spillover is not likely.28 Knowledge Absorption Capacity: Evidence from Firm-Level Surveys The intensity and effectiveness of private investments in knowledge and tech- nology absorption can be investigated through the survey data available for Russia and for BRICS and EU comparator countries.29 Russia has a very large R&D sector (employing roughly the same number of researchers per capita as Germany), but its output is comparable only to that of China, reflecting a strong historical orientation toward military research; the obsolescence afflicting R&D in certain �elds now that the economy is more open to knowl- edge and technology exchanges; and the lack of effective links between pub- lic and private R&D. Here we use the LME Survey data and BEEPS data (see the box in chapter 1 for a description of the surveys) to characterize the driv- ers of absorption in Russia, exploring both outputs and inputs of absorption and comparing the results with those of comparator countries. Fostering Knowledge Absorption 47 According to the LME Survey, knowledge absorption activities in Russia are associated with medium-low-technology and medium-high-technology manufacturing sectors.30 The electrical equipment sector is a leader in intro- ducing new or improved products, and the chemical industry introduces manufacturing technologies significantly more frequently than others. Firms in electrical equipment and chemicals and also in machinery are more likely to export technology-intensive manufactures. Those results suggest that capital-intensive sectors are adopting products and technologies more than labor-intensive industries. For example, only 29 percent of �rms in the textile industry financed a new or improved product, while 47 percent of �rms in other sectors did so. Results are similar for a new or improved pro- duction technology and technology-intensive exports. The food and wood sectors also exhibit weaker absorption. Estimating the Knowledge Absorption Production Function: Methodology and Results The econometric analysis presented here is meant to address three policy questions. —First, what is the impact of investments by Russian �rms in absorption and innovation inputs (for example, R&D expenditures and absorption- enabling investments in information technology) on key absorption outputs (for example, the introduction of new and improved products and manufac- turing processes)? —Second, how is the decision to allocate resources toward absorption and the capacity to adopt hard and soft technology related to a �rm’s characteris- tics and the wider economic environment? For example, it is generally argued that innovative activities increase with the size of the �rm. —Third, how does competitive pressure affect the frequency of invest- ments in absorption, such as �nancing for the introduction of products, or the potential for exporting advanced technology products? Specifically, is competition increasing the propensity to innovate? Our analysis begins to tackle those questions by characterizing the rela- tionship between a �rm’s probability of introducing a new or signi�cantly improved product or production technology and the �rm’s exports of high- technology goods, it characteristics, and environmental characteristics. In order to shed light on the underlying determinants of absorption outcomes, it also estimates a technology absorption production function, Absorption = f ( K , L , ICT , R & D , COMPRESS , IC ) , 48 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE where K is access to �nance; L represents workers’ education, skills, and train- ing; ICT captures various variables as proxies for technological capability, such as the availability of broadband Internet, a �rm website, and an infor- mation technology department; R&D captures various measures of the expenditure on research and development and related metrics, such as pur- chase of machinery and equipment and purchase of patents and know-how; COMPRESS refers to competitive pressures in the market; and IC stands for investment climate. See appendix 3 for a further description of the variables and econometric methods used in the regressions. Technology absorption outcomes are the left-hand variable in the equa- tion. They include the introduction of new or improved products and pro- duction technologies, which in general can be expected to be new to the �rm or the country but not new to the world. Different types of investment— among which are two key outlays, for research and development and for pur- chases of new machinery and equipment—are complements in the outcomes of absorption. The absorption of products and knowledge often occurs along with the adoption of advanced production methods. Our rationale for using these variables as indicators of absorptive and innovative capacity in Russian manufacturing �rms is as follows. ICT vari- ables are included because the most notable increase in productivity in man- ufacturing since the 1990s has been linked to the adoption of information and communication devices. The Internet allows businesses, their suppliers, and their clients to share and exchange vast amounts of information and knowledge. Indeed, like electricity and steam power, ICT is a “general purpose technology� that has the potential to spur growth as it spreads across differ- ent sectors of the economy, prompting a transformation in the organization of labor and production.31 ISO certi�cation is a relevant and much-used management standard that attempts to raise quality to the international level directly. Because reaching ISO standards requires a minimum level of technological capability, ISO norms can be seen as a possible proxy for technology adoption. Moreover, because standardization reduces transaction costs, it enables economies of scale and eliminates duplication, both of which are good for productivity gains. ICT use and ISO certi�cation therefore are not only indicators but also variables of interest themselves. The relationship between firm-level R&D and innovative outcomes has been investigated thoroughly. The literature, particularly for OECD coun- tries, generally finds a positive and significant association, yet the complex- ity and uncertainty surrounding knowledge-based activities means that Fostering Knowledge Absorption 49 Table 3-3. Summary of One-by-One Regressions for Product and Process Innovationa Firm introducing Firm introducing a Firm exporting a new or new or improved technology-intensive Independent variable improved product technology products Firm has a workforce that (+)** has tertiary education Firm exports its products (+)** Firm has a website (+)*** (+)*** (+)*** Firm acquired technological (+)*** (+)*** (+)*** innovations through purchase of patents, licenses, and general know-how from Russia or abroad Firm spends more than (+)*** (+)*** (+)*** 1 million rubles on R&D expenditure Firm acquired technological (+)*** (+)*** innovations as a result of third-party R&D Firm has acquired (+)*** (+)*** (+)*** ISO certi�cation Firm faces �nancial (−)* (−)** constraints such as unavailable collateral Firm is able to get loans (+)** (+)** from banks Firm faces noticeable or (+)*** considerable competitive pressure from domestic �rms Firm faces noticeable or (+)*** (+)* (+)*** considerable competitive pressure from imported products Source: LME Survey (2005). a. Only significant results are shown. All regressions include a constant, explanatory variables (size, new pri- vate firm, firm holding, foreign owner), and sectoral dummies and use a random effects model. See appendix tables A3-2–A3-4 for detailed results. Signi�cance is given by robust standard errors clustered by regions. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. estimating a firm-level or aggregate relationship between absorption inputs and outputs is not a trivial task.32 To investigate the relationship between private R&D and absorptive outcomes, we used a probit model for the sam- ple of Russian firms (tables 3-3 and 3-4).33 The results confirm a positive and significant relationship between investments in R&D and the three absorption outcomes: introducing new or improved products, introducing new or improved production technologies, and exporting technology- intensive goods. 50 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE The fact that Russian firms exhibit a positive association between R&D spending and absorption is de�nitely a positive sign of the capacity to modernize, especially for large and medium enterprises (LMEs) in manu- facturing. But that does not imply that there are no more obstacles to mak- ing effective investments in R&D for Russian �rms. Additional econometric estimations compare the probability of achieving an innovative outcome between firms in Russia and comparator countries (see table A3-6 in appendix 3, which describes the data and model used for the estimations). They show that the ef�cacy of absorption activities is higher in Russian �rms than Chinese �rms but lower in Russian �rms than in �rms in South Africa and Brazil. Therefore room exists for improving the innovation and absorp- tion capabilities of �rms and for correcting some of the external obstacles that reduce incentives to improve R&D expenditure outcomes. As expected, there are some significant and positive correlations between absorption outcomes and absorption-enabling inputs, such as ISO certifi- cation and IT use. ISO certification is significant only for advanced tech- nology exports in the full model (table 3-4). But when taken alone—with firm characteristics as controls—ISO certification turns out to be signifi- cant for all three absorption outcomes. IT use, in this case measured by the existence of a corporate website, is strongly associated with a higher likeli- hood of innovating. Finally, the purchase of intellectual property in Russia or abroad (patent rights, licenses for using inventions, production proto- types, and utility models) is strongly and positively associated with the probability of introducing product and process improvements at the firm level and positively associated with high-technology exports. That finding implies that technical knowledge codified in intellectual property rights contributes to a learning process that improves absorptive capability. The coefficients related to other firm characteristics show mixed signif- icance. Firm size is not related to introducing new or improved products or processes, but larger firms are significantly more likely to export high- technology products. That suggests that scale is a factor in export-oriented policies that would increase knowledge absorption. But there is a positive relationship between exporting and introducing new technologies (table 3-3), yet no evidence for a relationship with the other dimensions of absorption. Tertiary education level of employees has a positive impact on introducing new products (table 3-3), but that effect is lost in the full model (table 3-4), perhaps a reflection of the relatively weak impact of skills after adding other controls (chapter 4). Fostering Knowledge Absorption 51 Table 3-4. Regression Results for the Full Model of Product and Process Innovationa Firm introducing Firm introducing a Firm exporting a new or new or improved technology-intensive Independent variable improved product technology products Firms has less than (−)*** 250 employees Private �rm that was (+)** established after 1992 Firm has a website (+)*** (+)** (+)** Firm acquired technological (+)*** (+)*** (+)* innovations through purchase of patents, licenses, and general know-how from Russia or abroad Firm spending more than (+)*** (+)*** (+)*** 1 million rubles on R&D expenditure Firm acquired technological (+)*** (+)*** innovations as a result of third-party R&D Firm has acquired (+)*** ISO certi�cation Firm is able to get loans (+)* from banks Firm cites macroeconomic (−)* instability as an impediment for enterprise’s activity and development Firm faces noticeable or (+)** considerable competitive pressure from domestic �rms Firm faces noticeable or (+)** considerable competitive pressure from imported products Observations 729 729 729 Source: LME Survey (2005). a. Only signi�cant results are shown. See appendix table A3-5 for full regression results. All regressions include a constant, explanatory variables (size, new private �rm, �rm holding, foreign owner), and sectoral dummies and use a random effects model. Signi�cance is given by robust standard errors clustered by regions. P values are given in brackets. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. Lack of financing is often cited as a major constraint on the innovative activity of firms. We used two variables to gauge the effect of finance, one measuring firms’ financial constraint and the other gauging access to financing and financial institutions. The estimation indicates a negative and significant relationship (table 3-3) between financing shortages and product or process improvements as well as export of technology-intensive 52 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE products. Similarly, there is a positive and significant relationship between those two absorption outcomes and the access-to-financing variable, as expected (table 3-3). Because the finance variables and the size of the firm are highly correlated, if the model is estimated without the size variable, the finance variables become significant for all the absorption outcomes. With regard to investment climate variables, macroeconomic instability has a negative and significant effect on the probability that a firm will introduce a new product, suggesting that firms tend to absorb knowledge when the macroeconomic environment is stable (table 3-4). Empirical Analysis of Soft Innovation and Organizational Change The econometric analysis of the determinants of �rm-level R&D expendi- tures (table A3-7 in appendix 3) reiterates many of the �ndings just reported. Again, �rms that are larger and that have adopted information and commu- nications technology and attained ISO certi�cation are likely to spend more on R&D. The signi�cance of ISO certi�cation points to possible complemen- tarities between the organizational capability of the �rm and its capacity to conduct in-house R&D. In this case the educational level of the workforce also has a positive relation with R&D expenditure. Exporters have signi�- cantly greater R&D expenditures. Even though public R&D spending in Russia is much larger than private R&D spending, only 23 percent of firms surveyed outsourced part of their R&D activities to third parties. The positive relation between third- party R&D and in-house R&D by �rms underlines the fact that the two are probably complementary. Given the imbalances between public and private research, there is surely a great deal of room for additional collaboration. To complete the analysis of innovativeness in Russian �rms, we turned to the determinants of investment activities that aim to improve organizational capabilities. Such activities are referred to as “soft innovation� because they are tied to changes in processes and systems that seek to increase productiv- ity of the existing capital stock and workforce. The econometric analysis examines �ve dimensions of soft innovation that reflect how �rms absorb and implement production practices: restructuring the organization, hiring external management consultants, outsourcing functions and business processes to specialized third-party contractors, introducing input quality control for materials and production, and introducing an automated inven- tory management system. The reason for enlarging the examination of knowledge absorption is that improving �rms’ organization can be critical for absorption and innovation, Fostering Knowledge Absorption 53 in a sense capturing how ready �rms are to adopt and enhance new products and technologies. In the light of international differences in “innovative ef�- ciency,� readiness seems highly relevant for the ability of Russian �rms to catch up (table A3-6 in appendix 3). The results of the regressions show that organizational absorption is an investment activity that also is more likely in larger �rms (table 3-5). Firm size apart, the signi�cance of the other determinants really depends on the form of organizational change. The �rms that are more likely to restructure are those that —were established before 1992 (probably because they have an un�nished restructuring agenda) —are less likely to face a �nancial constraint —are more likely to face intense competitive pressure —are more likely to be exporters of high-technology products —are more likely to outsource R&D to third parties —perceive themselves to be in environments where regulatory problems exist. With regard to the direction of causality for some of the foregoing associ- ations, one might expect that restructuring leads to an easing of the �nancial constraints and to greater R&D spending, but the data do not allow for test- ing of that question. For firms to hire external management consultants, the significant determinants are having foreign equity participation, introducing a new or improved production technology (which points to the interrelatedness of hard and soft technology), outsourcing R&D, and being affected by macro- economic instability. Business process outsourcing is more likely with �rms that are part of a holding with foreign ownership. Input quality control is strongly associated with introducing a new product, an expected result. Last, the introduction of an automated inventory management system is positively associated with firms that have a more educated workforce, that have obtained ISO certification, and that have introduced new production technologies. The overall pattern suggests a very natural correspondence between the capabilities of the firms, particularly in their innovativeness, and decisions to implement costly management improvement measures to raise quality and optimize production systems. Competition as an Incentive for Absorption One of the most robust results drawn from the estimation of the absorp- tion production function is that competitive pressures have a positive and Table 3-5. Regression Results of the Full Model for Soft Innovation and Organizational Changesa Firm Firm Firm Firm automated reorganized Firm hired outsourced introduced system of organizational external to third-party input quality inventory Independent variable structure consultants contractor control management Firms has less than 250 employees (−)*** (−)** (−)* (−)** Private �rm that was established after 1992 (−)* Firm belongs to a group of small companies (holding) (+)* (+)*** Firm with more than 51 percent of common stock owned by (+)** (+)*** foreign individuals or company Firm has a workforce that has a tertiary education (+)*** Firm exports its products Firm has acquired ISO certi�cation (+)** Firm faces �nancial constraints such as unavailable collateral Firm is able to get loans from banks (+)*** (+)* Firm faces noticeable or considerable competitive pressure (+)** from imported products Firm introduced a new or improved product (+)*** Firm introduced a new or improved technology (+)* (+)* Firm exports technology-intensive products (+)** Firm acquired technological innovations as a result of (+)** (+)** third-party R&D Firm cites macroeconomic instability as an impediment for enterprise’s activity and development (+)** Firm cites regulatory constraints as an impediment for enterprise’s activity and development (+)** Observations 731 731 731 731 731 Source: LME Survey (2005). a. Only signi�cant results are shown. See appendix table A3-8 for detailed results. All regressions include a constant and sectoral dummies and use a random effects model. Signi�cance is given by robust standard errors clustered by regions. P values are given in brackets. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. Fostering Knowledge Absorption 55 Figure 3-3. Competitive Pressure and Absorption Activities Percent likelihood that a firm will introduce a new product 55 53 50 47 45 42 40 35 33 30 No competitive Low competitive Moderate competitive Strong competitive pressure pressure pressure pressure Source: Based on simulations using the LME survey; for details, see appendix 5. significant effect on a variety of innovative and absorptive outcomes. To investigate that relation further in a Russia-speci�c context, we analyzed the evidence on competitive pressures from the �rm-level data available. We tested the hypothesis that competitive pressures have a positive impact on innovative activities of �rms, using a variety of samples to test the robust- ness of the results (see appendix table A3-9 for the full range of speci�cations tested). In the LME sample, as expected, �rms subject to competitive pressure were more likely to engage in innovative activities. In particular, competitive pressures coming from domestic �rms were positively and signi�cantly cor- related with the introduction of new or improved products. Unexpectedly, competitive pressures from domestic �rms were negatively correlated with exports of advanced technology goods.34 Firms that faced signi�cant com- petitive pressures were about 20 percent more likely to introduce new and improved products, controlling for sector, �rm characteristics, and the like (�gure 3-3). In contrast, �rms that enjoyed monopolist positions or other forms of protection were less likely to change their products and production 56 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE methods than firms facing greater competitive pressure. They also con- tributed to an unpredictable and uncertain regulatory environment. In sum, a monopolistic position can seriously hurt investments in technology absorp- tion and innovation. Competitive pressures arising from imports, by contrast, seems to bear most heavily on exports of advanced technology products, as captured by the econometric results in appendix table A3-9. Because high-tech exports often are considered a key element and a crucial contribution to sustained long-term growth, that finding seems especially important, par- ticularly in light of the recent debate on diversifying the structure of the Russian economy.35 The robustness of the association between competition and knowledge absorption under different specifications is important given the method- ological difficulties of the simultaneous determination of the two vari- ables: industry structure shapes the incentives to adopt technology and innovate and innovative activities modify the competitive environment. The most recent theoretical and empirical thinking about that relation- ship points to an inverted U-shaped curve between competitive pressure and innovation.36 The upward sloping segment of the curve reflects the fact that product differentiation and cost advantages are more important when industries are more competitive—and that innovation and absorption are effective instruments for improving product quality and reducing manufacturing costs. In neck-and-neck industries, firms compete by lowering prices because their product lines and manufacturing processes are identical. Their absorptive and innovative activities therefore become a means of escaping price competition and allowing them to compete on quality and cost. The downward sloping segment indicates that very intense competi- tion erodes firm profits and thereby reduces the resources available to invest in R&D. The impact of competitive pressures on knowledge absorption, innova- tion, and productivity has been studied quite extensively. Carlin, Schaffer, and Seabright, using a data set from a BEEPS of nearly 4,000 firms in twenty-four transition countries, “find evidence of the importance of a minimum of rivalry in both absorption and growth: the presence of at least a few competitors is effective both directly and through improving the efficiency with which the rents from market power in product markets are utilized to undertake innovation.� They conclude that “our findings Fostering Knowledge Absorption 57 strongly reinforce the message that unchallenged monopoly is a drain on dynamism.�37 Moreover, the World Development Report 2005 on the invest- ment climate focused on the importance of a vibrant, competitive business environment for firms’ investments in skills, absorption, and innovation: “If firms are not subject to competitive pressures that stimulate technical progress and the demand for more skilled workers, then effective demand for education will be weak.�38 The Role of Government in Supporting Absorption The government of the Russian Federation has proposed strengthening existing programs and creating several new initiatives to promote absorp- tion and innovation in smaller firms and larger industrial firms.39 For smaller �rms, the measures include more effective infrastructure (technol- ogy parks, incubators), and �nancial support for small �rms and start-ups. Support for larger firms and the development of absorption networks and clusters include indirect measures to stimulate exports, quality certification assistance, co�nancing for ventures into global markets, support for patent- ing abroad, and support for the development of joint R&D projects. For the proposed measures to have the expected impact, it will be very important to translate them into specific interventions. The strategy does not appear to recognize the crucial role of competitive pressures in modernizing the Russian economy or the emerging trend toward increasing exposure to for- eign R&D (primarily through FDI, imports, and licensing but also by encouraging mobility and exchanges abroad). There have been many official efforts to support firms but information on the scope of those measures is scarce, limiting the possibilities of evaluat- ing their impact. One of the most successful support instruments is the Fund for Assistance to Small Innovative Enterprises (Bortnik Fund), which has a budget of about $30 million a year. Although the fund has an important demonstration role, it has not had a major impact on the R&D sector in Russia, and it has yet to be replicated on a large scale (box 3-2). Yakovlev (2006) addresses the extent of government support in Russia using information collected through an Enterprise Survey. Even though the perception exists that government support to enterprises has been limited, a quarter of �rms surveyed received direct support for investment, absorption, or exports, although the amounts were low.40 Among those types of support, support for exports appears least pervasive, with only 3 percent of firms 58 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE Box 3-2. Fund for Assistance to Small Innovative Enterprises The Fund for Assistance to Small Innovative Enterprises (Bortnik Fund) was created in 1994 by the Russian government.1 The budget, from the government, is 1.5 percent of total federal expenditures for civilian science. The fund sup- ports small innovative companies through open competition. Proposals pre- pared by the competing companies are submitted to peer review by external reviewers from the science and business sectors (including representatives of banks and venture and investment funds). The �nancial share of the fund in the winning projects cannot exceed 50 percent, and the companies have the rights to the intellectual property created in the projects. The fund pro- vides support only to projects included in a list of government-approved critical technologies. In the late 1990s, the fund began to increase support to companies at the seed and start-up stages, obviously a riskier strategy. The change was prompted not only by the stabilization of the general economic situation but also by the fact that the �nancial resources of the fund were becoming larger. In addition, the return to the fund from earlier loans was stable and provided additional money. In 2003 the fund initiated a program called START to support start-up companies. About half of the fund’s budget—or $10 million when the fund was started in 2003 and approximately $15 million in 2006—has since been devoted to the START program. In its composition and instruments, the program resembles the U.S. Small Business Innovation Research (SBIR) program. The �rst results show that about 20 percent of supported �rms managed to �nd investors and thus get support from the fund for the second year. That is an encouraging result showing that small �rms create quite a few technology- intensive goods and services with commercial potential. With a budget of about $30 million a year, the fund has not had a large impact on the R&D sector in Russia; it can only demonstrate the success of this or that instrument or approach. So far, it seems that the experience of the fund has not been replicated by other government agencies. Saltykov and Kuznetsov did a cost-bene�t analysis of a proposed bank loan for innovation infrastructure. The results for the Bortnik Fund show that the scale is small, that the number of participants is limited, and that the Bortnik Fund �nances only �rms with an impeccable credit history. It was concluded, therefore, that the fund does not allow for �nancing start-ups. Source: Bortnik (2004); unpublished analysis by Boris Saltykov and Boris Kuznetsov. 1. Resolution no. 65 of February 3, 1994. Fostering Knowledge Absorption 59 receiving it. At the same time, 28 percent of �rms surveyed received indirect support through procurement contracts with the government. Almost half the companies surveyed (44 percent) received some sort of support, direct or indirect. Direct support was provided more often to companies that had had signi�cant innovative activity or that exported a considerable proportion of their production.41 To enhance the role of government in supporting innovative firms in Russia, we recommend the following: Proposal 1: Reform the IPR regime to encourage researchers at public R&D institutes and universities to engage in private innovation while protecting the public IPRs—and support patenting activity of firms, especially patenting abroad and patenting by smaller firms with less capacity. The legacy of the Soviet science and technology system is still having a negative effect on the orientation and overall level of appropriation and exploitation activities in the National Innovation System (NIS).42 The incentive conflicts in the cur- rent arrangement have resulted in the de facto privatization of R&D activ- ities, which can be observed in multiple scientific research teams based in the public system that provide R&D services on an informal but commer- cial basis to LMEs and SMEs. Some may see that as the encouraging begin- ning of an organic R&D industry. But there are negative consequences, including conflicts of interest between researchers and institutes, uncer- tainty about the ownership of technical results, and the government’s loss of rents from the intellectual property generated through quasi-private research. Consider, on the other hand, India’s Council of Scientific and Industrial Research: founded in 1942, it has evolved with the changing eco- nomic structure of the country to become an internationally competitive institute (box 3-3). The government should consider additional policies to create incentives for spinning off research groups. The objective would be to lower the pub- lic financing burden for R&D institutes, foster commercial innovation and knowledge absorption by firms, and reallocate basic research funding in favor of universities. Although downsizing of R&D institutes has occurred, the most promising and active researchers have also been lost. The average age of those employed in R&D has risen along with the reduction in the number of personnel. University-based research would create a nurturing ground for young researchers, following the experience worldwide. In order to move to a more efficient and effective national innovation sys- tem, it will be important to hasten the dissolution of R&D institutes and 60 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE Box 3-3. From Autarkic Self-Reliance to Internationally Competitive Research and Development in India India’s Council of Scienti�c and Industrial Research (CSIR), founded in 1942, was modeled after the United Kingdom’s Department of Scienti�c and Industrial Research. Predating most other specialized research and develop- ment institutes in India, it took on a wide range of functions, from promotion of scienti�c research to setting up research and development institutions and collecting and disseminating data on research and industry. In the �rst two decades after India’s independence, it focused on building up an extensive research and development infrastructure, from metrology to research and development over a wide range of industries with a strong focus on supporting emerging industry, especially small and medium enterprises. When India changed from an inward-oriented development strategy to a more external, market-driven strategy as a result of the 1991 crisis, the focus of the CSIR also changed, from technological self-reliance to “research and development� as a business. There was more emphasis on output and per- formance in a more competitive market and on doing research that was rele- vant for the productive sectors and that could earn income. Each laboratory was considered a subsidiary corporate entity. Incentives and rewards for meeting targets were introduced, and if laboratories did a good job of meet- ing their commitments on output and deliverables, they were given greater autonomy in operations. In addition, there have been continuous efforts to streamline the CSIR further to improve its effectiveness and ef�ciency. Although the CSIR is going through further restructuring, the results have been impressive. Between 1997 and 2002 the CSIR reduced its laboratories from forty to thirty-eight and staf�ng from 24,000 to 20,000. At the same time, output increased noticeably. Technical and scienti�c publications in the inter- nationally recognized journals increased from 1,576 in 1995 to 2,900 in 2005, and their average impact factor increased from 1.5 to 2.2. Patent �lings in India increased from 264 in 1997–98 to 418 in 2004–05, and patent �lings abroad increased from 94 in 1997–98 to 500 in 2004–05. Moreover, the CSIR accounted for around 50 to 60 percent of all U.S. patents granted to resident Indian inven- tors. The CSIR also increased its earnings from outside income from 1.8 billion rupees in 1995–96 to 3.1 billion rupees in 2005–06 (about $65 million). Today, it has 4,700 active scientists and technologists in thirty-seven research laboratories, supported by 8,500 scientific and technical personnel. Source: Bhojwani (2006). Fostering Knowledge Absorption 61 the teams within them that work on obsolete scientific and industrial problems. Proposal 2: To encourage productivity growth, provide incentives for firms to invest more in their knowledge absorption capacity. Specifically, more use should be made of instruments such as matching grants for supporting the absorption of new technologies and processes, access to ICT and ISO certi�ca- tion, and investments in soft technology. Since the 1980s there has been increas- ing awareness in OECD countries of the benefits of matching grants in encouraging firms to share and manage risk.43 We recommend matching grants because they encourage public-private risk sharing and orient the selection process toward the R&D programs most likely to generate outcomes with a high commercial impact.44 The justi�cation for supporting ICT tech- nologies and international certi�cation is that such investments have strong and positive effects on productivity and innovative behavior. Clearly, there are major risks in allocating matching grants effectively— including program corruption, capture, and sectoral targeting. Therefore one prerequisite for success is an institutional design that would immunize the funding allocation from corruption and interference by political actors and other state or specific interests. A second prerequisite is neutrality: the program should not try to steer the grants in any predetermined direction (“picking winners�). To avoid government capture and failure, a matching grant program should be as neutral and transparent as possible. To ensure sufficient checks and balances, the decisionmaking process for funding allocations needs to include representatives from the private sector, academia, and civil society and foreign experts. An optimal design would include the fol- lowing key elements: —An independent institution with a clear mandate and control mecha- nism would be responsible for administration and funding decisions, keeping its mandate separate from other public policy goals. —The funding decision would be made by an independent investment committee. To enhance transparency, the investment committee should be staffed with technical experts and foreign experts, who are less likely to be subject to political influence. A potential problem is the question of con�- dentiality and fear of industrial espionage. —The investment policy and decisionmaking processes would be insti- tuted and supervised by a board consisting of representatives of different government institutions and international advisers. 62 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE —Technical assessments of project proposals would be based on external peer reviews involving international experts when possible. —All project proposals and decisions would be recorded, tracked, and made publicly available. E-government procurement technologies should be considered to aid the process. A traditional approach to innovation, absorption, and R&D support for firms has been through tax incentives (tax credits or lower tax rates), widely used in Europe to encourage investors and companies to invest in R&D. But tax incentives have weaknesses that make them less applicable to postsocialist economies, and their weaknesses seem significant enough to preclude their use. First, tax benefits are meant to help existing enterprises that can use profits from related products to take advantage of the credits or offsets. But they do not help start-ups that have not yet accumulated sufficient profits and therefore cannot offset tax liabilities. Second, in countries with a weak tax enforcement system, tax incentives may promote distorting tax avoidance behavior rather than productive investment. Third, tax incentives cannot be used (as grants can) to promote networks and links between the private sector and universities and research insti- tutes, which lie at the heart of the type of instrument that we recommend. The Institute of Economies in Transition (IET) study unsurprisingly finds that most respondents are keen about tax benefits.45 Proposal 3: Private seed capital funds should be encouraged, but state-owned and state-managed venture capital (VC) funds proposed by the government should be avoided. The government’s program on “generating incentives for development of the manufacturing sector� envisages the establishment of an open joint stock company, a Russian venture company that would be a government-owned institution supporting innovation by participating in existing venture funds and by contributing to the creation of new funds. That would augment the existing provision of state �nancing for the creation and especially the expansion of new high-growth �rms through venture funds, building on the experience and organization of the regional venture funds created in the 1990s with EBRD (Economic Bank for Reconstruction and Development) support. An open question is whether the funding is meant to be a transitory support leading to the creation of a private venture market. If so, the government would require an exit mechanism if and when a critical mass of venture projects was in place and incentives for private investment using venture funds were satis�ed. Fostering Knowledge Absorption 63 International best practices suggest that the government could seed the venture capital industry by investing in privately managed funds. In such public-private partnerships, government participation mitigates some of the risk in technology-oriented start-ups, and the seed capitalist provides commercial and managerial expertise. It still would be advisable for the seed capital program to be flexible in targeting small start-up high-technology companies by allowing public-private risk sharing with flexible contracting between investors and managers, free from excessive regulation, and the use of limited partnerships. Internal �nancing by enterprises, government funding, and angel investors are all important in early-stage technological development (ESTD). But most important is the virtual absence of more mainstream intermediaries such as banks, private equity, and other institutional investors. Even in advanced and innovative economies, early-stage finance of innovative projects is under- taken directly by �rms, if they have the resources, or by very specialized insti- tutions, with a signi�cant role for the government. Not surprisingly, internal funds account for the biggest share of ESTD �nancing in the United States because that is the most straightforward way of overcoming information asymmetries. Established enterprises know the track record of their own inventors and employees—and they typically have a better understanding of the market and the commercial potential of internally proposed innovations than outside agents do. Enterprises use the cash flows generated by estab- lished operations to �nance innovation or to tap external funds on the basis of the strength of their balance sheet. Angel investors are another important source of ESTD funding in the United States and to some extent in Europe. The term refers to successful entrepreneurs who look for new opportunities to invest private funds earned from their previous economic activities and are willing to invest in ESTD projects in technological fields that they understand well (having “been there and done that�). Studies of the behavior of angel investors frequently find that they are often heavily involved in commercial decisionmaking and that their “business support� function can be as important as the �nancing that they offer. Giving managerial advice and maintaining commercial con- trol over the ESTD entrepreneur are typical functions of the angel investor and venture capital funding models—as well as internal funding models. Given the short history of capital accumulation and profit-generating enterprises in Russia, internal financing by enterprises and angel investors is rare in the region and does not provide a viable basis for promoting 64 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE innovation. The absence of angel investors is problematic not only from a funding perspective but also because they function as sources of managerial expertise, as information brokers, and as contacts to broader formal and informal networks of entrepreneurs and innovators. The role of government in Russia is therefore different from its role in OECD countries. The lack of angels and internal �nancing is acute, and the capacity of government agen- cies to take their place is extremely limited. In that respect, some caution is warranted regarding the Russian govern- ment’s initiative to establish venture capital companies with 100 percent state participation. International experience shows that publicly financed and managed venture capital funds may be ineffective in fostering the commer- cialization of new and improved goods and services and the expansion of innovative �rms. The role of state-owned and state-managed venture capital funds in financing innovative projects in India, for example, was limited. Most of the funds set up in India during the recent relatively high-growth period (1995–2000) tended to be private funds, �nanced heavily from over- seas by communities of nonresident Indians. It can be argued that the single most important factor for the VC and IT industries in India was the combi- nation of returnees from the diaspora and a generous supply of engineering and IT talent. The successful businesses created from that pool of talent sparked spin-off and investment activities, which, in turn, influenced the gov- ernment to improve business conditions further. The lesson is that directly intervening to manage the allocation of VC funds is unlikely to lead to the expected results. When venture capital funds managed by government entities achieve commercial success, often it is because the funds are invested not in innova- tive ventures but in small, more mature companies with less risky product lines. Moreover, capture and rent seeking are prevalent and problematic because these types of funds are dominated by political interests. One advantage of implementing proposals 2 and 3 simultaneously is that the synergies between the two can establish a more integrated frame- work for supporting entrepreneurial companies. A seed capital program aimed at promoting knowledge absorption is likely to work best when a grants program provides critical funding at the earlier stages of technolog- ical development, followed by private seed capital funds. The reason is that commercial venture capital funds typically avoid the risk connected with early-stage companies and target more established projects. Matching grants can fill the gap in the support framework for early-stage technology development. Fostering Knowledge Absorption 65 In the more mature stages of technological development, seed capital pro- vides funds to expand production and the customer base and supports the later (and most visible) stages of commercialization. To minimize the poten- tial negative distortions of both proposals 2 and 3, the instrument design should ensure neutrality, whereby the government does not decide which sec- tor or technology to support but responds to market demand.46 When applied to matching grants or to the funding of seed capital funds, that trans- lates into having the government establish universal criteria for submission and eligibility but allowing an independent entity to select projects for sup- port through a transparent process. Notes 1. These �gures are from the OECD Science, Technology, and Industry Scoreboard 2005 (OECD 2005). Not all �gures were available for 2004, the most recent year, so �gures for the most recent year available were used instead. 2. Watkins (2003). 3. OECD (2005). Not all �gures were available for the most recent year (2004), so �gures for the most recent year available were used instead. According to of�cial Russian data for 2004, government expenditures on R&D were 59.6 percent of total expenditures (CSRS 2005, p. 75). 4. According to CSRS data (2005, p. 75) the share of funding from the “business enterprise sector� was 17.3 percent in 1998 and 21.4 percent in 2004. This indicator is quite dif�cult to decipher because it includes expenditures of state-owned enterprises and thus is not “private,� but it gives an idea of the evolution over time. 5. OECD (2005b, p. 35). 6. The number of patents issued is highly influenced by export activity in a country. Given the fact that the application process implies costs to the owner of the knowledge, a patent may be pursued only when that knowledge has significant commercial value in the target market (such as the United States or European Union). 7. OECD (2005a). 8. The use of patenting in the United States as an indicator of knowledge was pioneered by Griliches (1994, 1998), followed by Jaffe and Trajtenberg (2002) and Lederman and Maloney (2003b). 9. Eurostat (2004). 10. OECD (1994). See also Rosstat (2004, p. 75) and OECD (2004b, p.192). 11. According to Boris Saltykov (Saltykov 2001), 75 percent of Soviet R&D was defense oriented. Basic science was usually emphasized because in certain areas the Soviet Union was the world leader. 66 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE 12. Hoekmann and Javorcik (2006). A third channel is direct trade in knowledge through purchasing patents and licensing. 13. Coe, Helpman, and Hoffmaister (1995); Keller (2002). 14. Data from COMTRADE on “capital goods (except transport equipment) and parts and accessories thereof,� classi�ed as BEC-4. See UNCTAD (2006a). 15. Kraay (1999). 16. Global production sharing refers to the process whereby multinational corporations—through FDI—develop international production and distribution networks around the globe. This process facilitates production and distribution in different countries at different phases of the production process, enabling “true internationalization of the manufacturing process.� Intra-industry trade, by con- trast, refers to vertically integrated international networks that usually leverage a highly skilled labor force and specialized products. 17. Rodrik (2006a). 18. Most CIS countries have not yet made the transition from “buyer-driven� production networks (wherein countries can make use of cheap, unskilled labor to leverage labor-intensive activities, typically clothing and furniture manufacturing) to capital-intensive “producer-driven� networks. That transition is usually encouraged by rising wages, but economies rich in natural resources do not always face the same binding constraints (Broadman 2006). 19. Saggi (2006). 20. The FDI benchmarking analysis in the UNCTAD World Investment Report 2006 (UNCTAD 2006b) also indicates that Russia has high FDI potential but low FDI performance. 21. World Bank (2006d). 22. Hare, Schaffer, and Shabunina (2004). 23. UNCTAD (2006b). 24. Whalley and Xin (2006) indicates that foreign-invested enterprises account for more than 50 percent of exports and 60 percent of imports and are a significant driver of economic growth (the authors calculate that around 40 percent of China’s economic growth in 2003 and 2004 was due to FIEs). These enterprises have much greater average labor productivity than non-FIEs, and their export capabilities ben- efit from access to distribution systems and products tailored for export markets. Concerning absorption, the World Investment Report (UNCTAD 2006b, p. 188) argues that “the advantage of forming a joint venture from the perspective of tech- nology diffusion within the host economy is that the local partners and the affiliate, which would be vested with a certain amount of technological and managerial expertise transferred from the parent firm, are likely to have close contacts and exchanges of personnel. Forming a joint venture is therefore the most obvious— and possibly the most effective—means by which local firms can acquire knowledge from TNCs.� 25. Ark and Piatkovski (2004). Fostering Knowledge Absorption 67 26. World Bank (2006d). 27. Hare, Schaffer, and Shabunina (2004). 28. Lederman and Maloney (2003a). 29. As in other studies using a production function approach and �rm-level data, endogeneity is a possible problem in the estimations, owing to the two-way rela- tionship between technology absorption and R&D spending, absorption outcomes and organizational change, and absorption decisions and competition. As a result, the description of the results is careful not to interpret the correlations in terms of causality but simply as an correlation between different investment activities and outcomes. 30. This follows the classification of industries in the OECD STAN Indicators, based on International Standard Industrial Classification (ISIC) Revision 3. The four technology categories are low, medium-low, medium-high, and high. Exam- ples of low-technology industries include food, tobacco, wood, and paper products; beverages; and textiles. The high-technology manufactures are pharmaceuticals; office, accounting, and computing machinery; radio, television, and communica- tions equipment; medical, precision, and optical instruments; and aircraft and spacecraft (OECD 2004c). 31. Helpman and Trajtenberg (1996). 32. For example, Acs and Audretsch (1988) confirms a positive relationship between R&D expenditures and the number of innovations in �rms. The relationship is “sensitive to the total amount of innovative activity. That is, in industries in which there is little innovative activity . . . the correlation between all the measures of tech- nical change becomes considerably weaker� (p. 682). The authors also refer to results in Griliches (1986) that show that the estimates depend on the de�nition and meas- urement of R&D. 33. Random effects and clustering are used to control for regional speci�c vari- ables that may be driving innovative activities. This empirical model is meant to test for partial correlations and links between innovation outputs and inputs, but it is not an attempt to identify causal effects between those variables. 34. The evidence here is based on different indicators, mostly a �rm’s assessment of the level of competition in its main market. Even though we tested a number of dif- ferent indicators and speci�cations, these �ndings should be interpreted more as con- �rmation of a quite developed part of empirical �ndings on the same topic. 35. Litwack (2005). As indicated by Rodrik, “China’s experience indicates that it is not how much you export, but what you export that matters� (Rodrik 2006b). 36. Aghion and others (2002). 37. Carlin, Schaffer, and Seabright (2004, p. 20, 25). In this chapter we follow the example of Carlin, Schaffer, and Seabright in using measures of market power that correspond to the perceptions of individual �rms regarding the competitive pres- sures that they face. They argue that “these are an important supplement to more conventional measures, such as shares of markets based on conventional industrial 68 I. G OLDBERG , E. B LANCO -A RMAS , J. G ODDARD , AND S. K URIAKOSE classi�cations. These can help not just in illuminating the overall pressures faced by �rms but also the way in which different constraints on managerial decision-making interact.� 38. World Bank (2004, p. 157). 39. See the “Research and Development Strategy of the Russian Federation to 2015,� approved in March 2006 (Ministry of Education and Science 2006). 40. Of �rms receiving support, 5.6 percent received two or more types. 41. This is not the case for indirect support, so it cannot be af�rmed that effec- tiveness is a consideration in the granting of procurement contracts. Exporting by itself is considered as a factor signaling effectiveness of �rms. 42. OECD (2005b). 43. The following paragraphs are based on a World Bank report, “Public Financial Support for Commercial Innovation,� which analyzes the various �nancial instru- ments to encourage innovation and recommends a series of reforms before those instruments can be put to good use (World Bank 2006c). 44. The importance of matching stems from the fact that it reduces the marginal cost of research to the �rm. A �rm facing a downward-sloping marginal research return schedule will always increase total expenditure when the marginal cost falls, precluding dollar-for-dollar crowding out. 45. CEFIR and IET (2006). Respondents in the IET survey expressed a preference for the introduction of incremental tax privileges for R&D costs and the purchase of foreign technologies. 46. World Bank (2006c). four Upgrading Skills H ONG TAN , V LADIMIR G IMPELSON , AND Y EVGENIYA S AVCHENKO R ussia, even with its highly educated workforce, faces a growing shortage of skilled workers in industry. In the transition to a market economy, Russia’s workforce underwent a wrenching reallocation of labor across industries and occupations, and many specialized and technical skills that workers acquired under central planning were no longer in demand.1 Mismatches in the labor market became widespread, with sharp shortages of some types of skilled workers coexisting with excess supplies of others. The formal education sys- tem and the specialized vocational and technical training institutions in par- ticular were poorly prepared to operate under the new market conditions and to supply the new skills that the market required. Employers who once hoarded labor are now reporting skill shortages as a major production con- straint, and some are upgrading the skills of their existing workers through various training programs. To raise labor productivity in industry, improve industry’s international competitiveness, and participate more fully in the global knowledge econ- omy, Russia must analyze the issue of skill requirements and develop poli- cies to address it. Skill shortages can directly constrain production and prevent firms from meeting demand and using available inputs efficiently, with lower productivity as a consequence. And—indirectly—they can inhibit the absorption of new knowledge, a skill-intensive activity. With respect to the general economy, the mismatch between the skills that �rms require and those that education and training institutions offer can waste 69 70 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO scarce public and private resources. With respect to individuals, the mismatch leads to sunk investments in their human capital that yield low returns and unfavorable labor market outcomes. Whether a corrective policy is appropriate depends on the cause of a skill mismatch. The likely causes include —inadequate funding or governance of education and training institu- tions, which constrains them from responding to the needs of the market —inappropriate labor regulations, which inhibit hiring and �ring by �rms to meet staf�ng shortfalls —restrictive compensation policies, which prevent some employers from paying wages that are competitive enough to attract needed labor —market failures in the training market, such as high turnover of trained workers, which inhibit the willingness of employers to invest in training to meet their own skill needs. This chapter draws on the complete Russian Competitiveness and Invest- ment Climate Assessment Survey—including the Large and Medium Enter- prise Survey (LME Survey) and the Small Enterprise Survey (SE Survey)— and related research and information sources to gain insight into skill shortages and mismatches and in-service training. It examines recent trends in the level and quality of education, the effects of economic restructuring on the skill composition of the workforce, the returns to schooling, and the aggregate supply and demand for skills in the Russian labor market. Using data from firm surveys, it then characterizes the distribution and nature of staf�ng and skill shortages among different groups of manufacturing �rms, thereby contributing to the understanding of reported skill shortages and staf�ng problems, including labor turnover, compensation policies, and the inhibiting effects of labor regulations. A discussion of worker training fol- lows, presenting evidence on the distribution, intensity, and determinants of in-service training and the implications for productivity and wages. The chapter concludes with some policy implications. Evolution of Human Capital in Russia The transition from a centrally planned to a market economy has strongly affected the evolution of human capital in Russia. Before the transition, most of Russia’s workforce was concentrated in industry and the service sector was underdeveloped. Educational attainment was high, but the educational system was oriented toward imparting narrowly defined technical skills at the expense of more general knowledge and skills. Wage inequality was arti- Upgrading Skills 71 �cially compressed, and rates of return to higher education were relatively low (in the 1 to 2 percent range). That employment structure changed dramatically after 1991. In the �rst stage of the transition (1991–98), industrial restructuring was accompanied by decreases in employment and working hours and steep declines in real wages. The second stage (1999–2006)—against the backdrop of a dynamic recovery following the crisis of 1998 that positively affected all labor market indicators—led to rising returns to education and reports of skill shortages. According to Barro and Lee (2001), in 2001 Russia had one of the most highly educated workforces in the world. For the population age twenty-�ve and over, Russia ranked seventh, with an average of 10.5 years of schooling. It was ahead of other BRIC and transition countries, as well as Germany, Japan, and the United Kingdom (appendix figure A4-1). Russia also had one of the highest shares of population age twenty-five and over that had a tertiary-level degree—more than half (57 percent), 13 percentage points more than in Canada and over 26 percentage points more than in other postsocialist countries2 (see appendix figure A4-1). Russia thus appears well situated to take advantage of knowledge-based economic activities requiring a well-educated workforce and a significant pool of researchers. In 2003 Russia had twice as many researchers per million population (3,371) as the Czech Republic, Hungary, or Poland (which aver- aged about 1,500) and �ve to ten times more researchers than Brazil (344 in 2000) or China (663). On that indicator, Russia was closer to France and Ger- many (which had about 3,200 researchers per million) but behind the United States and Japan (4,500 to 5,300). It bene�ted from downsizing in the science sector during the transition, so that now a signi�cant proportion of the work- force that is not employed in the science sector has experience in research. Despite its high formal educational attainment, Russia faces serious prob- lems with the quality of education provided, including underfunding; low quality of instruction; the deterioration of secondary education, as meas- ured by international standardized tests like PISA and TIMMS; and an ori- entation to teaching narrowly de�ned skills in professional education (see appendix 4). High rates of educational attainment are not simply a legacy of the pre- transition period. While demand for higher education fell in the immediate post-reform period, enrollment rates rose again in the mid-1990s and today exceed the rates of the late 1980s. How much of that increase in educational attainment was the result of changes in the industrial and occupational composition of employment that accompanied restructuring? How much 72 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO was the result of educational upgrading within industries and occupa- tions? A decomposition of the effects of industrial and occupational changes suggests that while changes in the structure of industry and occupations contributed modestly to upgrading of the workforce in the early 1990s, most of the subsequent educational upgrading occurred independent of restructuring.3 That the upgrading took place across the board, within all industries and occupations, suggests a strong, skill-biased process of change, reflected in technological change and in the transformation of organizational and institutional arrangements in the workplace. The demand for educa- tion is likely to increase in such an environment, given the comparative advantage of educated workers in implementing new technology or in responding to disequilibria.4 The rising returns to education in Russia help explain why the demand for education was so strong during the transition. Mincer-type wage equations suggest that private returns to an extra year of schooling prior to the transi- tion were in the range of 2 to 3 percent, reflecting wage compression result- ing from the administratively set “wage grid� system. The demise of central- ized wage setting led to a rapid increase in the education premium. Returns to an extra year of education rose to about 7 to 8 percent in the �rst �ve years of transition and by an additional 2 to 3 percent in the later period, stabiliz- ing at 8 to 10 percent by 2000–02 (see appendix 4). When returns are differentiated by level of education, specialized training tends to yield lower payoffs than more general education. Vocational train- ing increased the wages of secondary school graduates by about 5 percent. Tertiary professional and technical colleges, which provide training in spe- ci�c skills, yielded wage increases of 13 percent for males and 20 percent for females. University-educated males earned 50 percent more than those who completed only secondary school, and the wage premium for university- educated females was about 70 percent. The high returns to university edu- cation explain why enrollment rates in higher education rose over the tran- sition. And the fact that schooling returns stayed high despite the increasing supply of educated workers indicates that the demand for higher education was very strong, exceeding supply. Skill Constraints and Labor Shortages Given the concerns raised by the deteriorating quality and relevance of edu- cation and training in Russia, evidence from firm surveys, including the Russia LME Survey, provides insights into how employers perceive labor Upgrading Skills 73 and skill shortages, whether their perceptions are valid, and which firms are most affected. Several factors may constrain enterprises from respond- ing to perceived skill shortages—particularly labor turnover, compensation policies, and labor legislation. Respondents to the Russia LME Survey ranked “lack of skilled and quali- �ed workforce� as the number-two investment climate constraint to enter- prise growth and development (the number-one constraint was taxation). Small enterprises with fewer than 100 employees also ranked the skill con- straint as major or severe, though not as highly as regulation or access to and cost of �nance (see appendix 4).5 The skill constraint is not new, but it has been growing with the transition from a planned to a market economy and with the rapid economic growth since the late 1990s. Time-series data from the quarterly Russia Economic Barometer sur- veys provide insights into how overstaffing or understaffing in enterprises has changed over the last two decades.6 Before the 1998 financial crisis, the proportion of firms reporting that they were overstaffed relative to expected output in the coming year was high—in 1997, 38 percent of firms noted that they had redundant personnel. The strong recovery in indus- trial output after 1998 brought the proportion of overstaffed firms down to less than 15 percent. The proportion of firms reporting that staffing was not sufficient to meet expected demand started to grow after 1998, and by 2004 almost one in four firms reported understaffing against expected output. The shift from overstaffing to labor shortage is consistent with labor use rates, which grew from around 70 percent in the mid-1990s to 90 percent, almost full use of the workforce, in 2005. The shortages also grew with output, which increased 1.5 times over 1999–2005 as employment in the corporate sector fell slightly. In the 2005 Russia LME Survey, about 60 percent of surveyed managers rated their current staffing as “optimal� relative to current output while 27 percent rated their firm as “understaffed� and 13 percent as “over- staffed.� On average, understaffed firms were short of personnel by 17 per- cent, while overstaffed firms had 15 percent more workers than they needed (see appendix 4). That means that a sizable fraction of Russian enterprises had difficulties adjusting the size of their workforce to levels dictated by their output. The probability and levels of understaffing were highest for firms oper- ating in the textile industry, where more than 50 percent of surveyed firms reported less than optimal staffing, with the staffing gap averaging 22.6 per- 74 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO Table 4-1. Staf�ng and Workforce Skills, by Skill Category Classi�cation Other Skilled Unskilled of enterprises Managers Professionals white collar workers workers Skills and quali�cations of workforce a major or severe constraint Yes 51.1 51.8 68.4 53.8 60.3 No 40.1 38.0 40.1 25.5 38.0 Overall staf�ng in the �rm Optimal 3.0 11.8 0.7 37.0 4.9 Understaffed 8.1 37.0 4.4 95.6 29.3 Overstaffed 3.9 14.8 2.3 42.2 6.3 Total 4.5 19.1 1.9 53.6 11.7 Source: LME Survey (2005). cent relative to the desired level. New firms established in or after 1992, small enterprises with fewer than 250 employees, firms involved in metal- lurgy and machine building, and government-controlled �rms (those having more than 25 percent public ownership) also were more likely to report understaf�ng. Overstaf�ng was more prevalent among large �rms (those hav- ing more than 1,000 employees) and �rms involved with chemicals. Enterprises were concerned not only about overall staffing but also about the desired mix of skills. That finding was borne out by firms report- ing understaf�ng in several occupational groups—managers, professionals, other white-collar employees, skilled workers, and unskilled workers. Firms that ranked “skills and quali�cations of the workforce� as a major or severe constraint were more likely to report understaffing in the different skill groups than were those that did not rank skill constraints highly (table 4-1). As might be expected, firms that had less than optimal staffing were also more likely than other firms to report understaffing in all skill categories, especially skilled workers (95 percent) and professionals (37 percent). Inter- estingly, firms with optimal or more than optimal staffing also reported shortages in the same two skill categories. Speci�c shortages, especially of professional and skilled workers, can coexist with overall optimal staf�ng or overstaf�ng. That fact hints that �rms face dif�culties in adjusting and real- locating their workforce. The extent to which skill shortages are a problem varies across units within firms. Most firms identified two major problems—lack of technological capacity and lack of skilled and quali�ed workers, both concentrated in oper- ating units, that is, on production lines. A much smaller fraction of �rms Upgrading Skills 75 reported those as major problems in their economic, human resource, and research and development units (see appendix 4). Firms experiencing skill shortages cited several reasons for understaf�ng. The four most commonly cited were lack of workers with needed skills in the local labor market (72 percent), lower wages than at other �rms (41 percent), high labor turnover (30 percent), and high competition for workers in the local labor market (23 percent). Those reasons are consistent with an inade- quate supply of workers with relevant job skills in the local labor market, high labor turnover rates, and noncompetitive wages and salaries. Labor Turnover During the transition to a market economy, labor turnover in Russian �rms has been higher than in �rms in other former USSR countries. In 2004 the average rate of new hires was about 29 percent, while the job separation rate was 31 percent, giving the Russian economy a gross labor turnover rate of about 60 percent. Those turnover indicators are even higher if only industry is considered, with hiring, separation, and gross turnover rates of 30, 35, and 65 percent, respectively.7 The high turnover rates were not neutral with respect to skills. Managers surveyed in the Russia Economic Barometer were asked to compare the skill mix of those who were newly hired or separated to the skill mix of those who remained. Throughout 1996–2005, more than a third of all managers reported deterioration in the quality of their workforce, about half reported no change in quality, and a tenth reported some improvement in quality due to labor turnover. The low quality of newly hired workers, not the high quality of separations, may have been responsible for the reported deterio- ration in quality. Almost half of the firms hired workers who had lower- quality skills, while only 10 percent hired workers who were more skilled. The proportion of firms that experienced an improvement in workforce quality through job separations was roughly equal to the proportion that experienced a decline. The net outcome, at least for one segment of the firms surveyed, was that the overall quality of the workforce fell. Compensation Policies Respondents to the Russia LME Survey listed noncompetitive wages as one reason for understaf�ng. If that is true, noncompetitive wages may account for the inability of �rms experiencing labor or skill shortages either to retain their skilled workers or to hire equally or more skilled workers from the 76 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO open labor market, as the Russia Economic Barometer data indicate. Firms may not offer competitive wages if they have below-average performance and profitability—that is, if they are unable to pay wages that are high enough to retain their most skilled workers or to �ll vacant positions with the skilled labor that they need. Gimpelson (2004) used data from a survey of 300 large and medium �rms in Russia to investigate whether skill shortages were driven by supply or by demand constraints and to �nd out what enterprises were doing to respond to reported skill shortfalls.8 The analysis suggested that understaffed �rms had lower labor productivity, pro�tability, and average wages than both opti- mally staffed and overstaffed �rms. And if low-ef�ciency �rms (those with low labor productivity, pro�tability, or wages) declared that they had labor or skill shortages, they were more likely to use workers with mass (generic) skills supplied by the traditional vocational education system. In contrast, ef�cient �rms were more likely to search for workers with speci�c or unique skills, who were in limited supply. A similar pattern of reported staf�ng and �rm performance emerges in the 2005 Russia LME Survey, which includes a much larger sample of industrial enterprises. Understaffed �rms fared the worst on all performance indica- tors. Though understaffed, they lost employment and showed negative net employment change over the previous year. Overstaffed �rms, by contrast, were in slightly better economic shape and showed signi�cant (and needed) downsizing over the preceding year. The best performance in labor produc- tivity and pro�tability was by �rms with optimal staf�ng. They paid wages that were comparable to those paid by overstaffed firms and significantly above those paid by understaffed and low-productivity �rms. Labor Legislation Russian enterprises may also be constrained from meeting reported skill shortages by employment protection legislation (EPL). An emerging liter- ature suggests that overly strict employment protection legislation can impair hiring and firing, stifle job creation, and lead to higher unemploy- ment. Labor legislation—regarding minimum wages, social benefits and guarantees, employment contracts, and layoff regulations—can change the labor costs that employers face and, if strictly enforced, lower their incentives to hire new workers or discharge redundant ones, even when warranted by labor demand. On the World Bank’s “Doing Business� scale, Russia got a score of 30 on rigidity of employment, comparable with that of China but significantly Upgrading Skills 77 lower than that of either Brazil or India, which had scores of 56 and 62 respec- tively.9 Russia’s index of employment rigidity was closer to the average for OECD countries and lower than that of most other transition countries except for the Czech Republic. According to the index, �ring costs—measured in weeks of wages given as compensation for discharge—also were signi�- cantly lower in Russia than in other BRIC countries. The index may understate the extent to which employment protection regulations in Russia constrain the staffing decisions of employers10 because they are poorly and selectively enforced; their impact on staffing flexibility, therefore, may differ from one firm to another.11 The actual “rule of law� is selective and varies by region, sector, age of firm, and segment of the legis- lation.12 In large and mostly unionized firms (which account for roughly two-thirds of total employment in Russia), EPL is more strictly enforced, while the same provisions are barely binding on small firms. Instead of reducing uncertainty, Russia’s EPL regime increases it (through non- enforcement) and differentiates firms by their mandatory labor costs. Firms that enjoy discretion in applying the rules may avoid paying social benefits to their workers. Firms that abide by the rules—typically large and medium firms—avoid creating new jobs and maintain a low-wage policy; many rely on small firms as flexible suppliers of labor. Not surprisingly, managers in the Russia LME Survey did not rank EPL as signi�cant a production constraint as the shortage of skilled labor.13 Even so, about 17 percent of respondents did rank it as a notable constraint. In a sep- arate question on labor regulations, only 40 percent of respondents believed that labor regulations did not create major problems for their enterprise. One-�fth reported that rules on hiring foreign labor created serious dif�- culties, 19 percent pointed to hiring and �ring rules, and 15 percent stressed regulations regarding working hours. Overstaffed �rms tended to select hir- ing and �ring rules, working hours regulations, and rules on timing of wage payments as the most constraining labor regulations. Understaffed firms tended to stress rules governing the minimum wage and the hiring of for- eign workers. Although the use of short-term contracts is restricted by labor law, 38 percent of surveyed �rms reported using them to hire about 10 per- cent of their workforce. Labor adjustment costs induced by the EPL are likely to make it more difficult for firms to search for and hire skilled workers. The regression esti- mates in appendix table A4-6 suggest that the EPL index is positively asso- ciated with the difficulties that firms experience in searching for and hiring professionals and skilled workers. The higher the sum of EPL rankings as a 78 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO constraint, the more severe the search and hiring difficulties reported by firms for both groups. Firms able to circumvent employment protection legislation were less likely to rank searching for and hiring skilled labor as a problem. The results also confirm that firms paying low, noncompetitive wages were more likely to report difficulties in searching for and hiring skilled labor. In-Service Training One possible solution to skill shortages in the local labor market is for employers to train or upgrade the skills of their existing workforce. The hiring-versus-retraining debate is discussed in Lazareva, Denisova, and Tsukhlo (2006). In the authors’ survey of 1,000 industrial enterprises in 2004, 56 percent of firms considered retraining existing workers to be the most efficient way to meet skill shortages, 35 percent preferred hiring from the external market, and 25 percent preferred agreements with education and training institutions. The BEEPS and LME Survey data show that selected comparator countries in East Asia and Latin America had a higher incidence of training than tran- sition economies in Eastern Europe and OECD countries, with South Asia especially far behind (see appendix 4).14 How did training incidence in Russia compare with that in its BRIC competitors—Brazil, India, and China? At 58 percent, Russia trailed behind China (92 percent) and Brazil (59 percent), but it was way ahead of India (17 percent).15 The Russia LME Survey elicited data about the skill groups that received in-service training and the number trained. It suggests that managers, pro- fessionals, and skilled workers are the three skill groups most likely to ben- efit from in-service training, consistent with the kinds of skill shortages that firms reported. On average, 10 to 11 percent of managers and profes- sionals and about 8 percent of skilled workers received formal training. Those �gures, which are extremely low by international standards, sug- gest that in-service training is not �rmly entrenched in Russian �rms. World Bank (1997) estimated that in Malaysia—a fast-growing East Asia country that also ranked skill shortages highly—24 percent of managers, 32 percent of professionals and technicians, and 13 to 16 percent of production workers received formal in-service training.16 So while many more Russian �rms train their employees, they provide training to a smaller number of them than do their fast-growing counterparts in East Asia. Russia stands out among the BRIC group with respect to the very small share of its workforce trained within the firm (7.7 percent of skilled and Upgrading Skills 79 Figure 4-1. Percent of Workforce Trained and Percent of Firms Providing In-Service Training in Russia and Selected Countries Percent of workers Percent of firms 80 100 70 90 80 60 70 50 60 40 50 30 40 30 20 20 10 10 0 0 a ka ria ia l ce sia y a ia ia il n ile nd a ga an az di re ai in rb an ss an Ch ee la a ay In rtu Sp Ch Ko Br rm lg Se Ru hu Ire iL Gr al Bu Po Ge h Lit Sr M ut So Skilled workers trained Unskilled workers trained Firms that provide employees with training Source: LME Survey (2005). 1.4 percent of unskilled workers) compared with that of Brazil (53 percent and 45 percent) and China (44 and 28 percent) (figure 4-1 and appendix table A4-7).17 The share of workers trained through in-service programs also was lower in Russia than in several other transition countries (12 percent and 45 percent for skilled and 6 percent and 13 percent for unskilled workers respectively) and selected developing countries (47 percent and 70 percent for skilled and 34 percent and 53 percent for unskilled workers respectively), and it was signi�cantly below the share in OECD countries as a group (typi- cally above 50 percent). It is possible, though unlikely, that the low estimates for Russia are a statistical artifact.18 The recent survey of enterprise training practices in Lazareva, Denisova, and Tsukhlo (2006) reports a somewhat higher figure for Russia—about 20 percent of workers—though it is still below �gures for OECD countries. Which firms trained, and where did workers get their training? The inci- dence of in-service training was higher among larger, export-oriented firms and firms that invested in research and development. Long-established firms were more likely to provide in-service training than newer firms. Domestically owned �rms tended to train more than foreign-owned �rms, and government ownership made no difference in the likelihood of training. 80 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO Localities rated moderate in investment risk tended to have a higher inci- dence of training than minimal-risk or high-risk regions. The probit-regression analysis in appendix table A4-8 reinforces the cor- relations reported. First, it shows that the likelihood of in-service training is higher in larger �rms (those with more than 250 employees) and in locali- ties with moderate investment risk than in either low- or high-risk regions. Second, �rms that employ a larger proportion of workers with higher edu- cation also are more likely to train. The empirical evidence from many coun- tries indicates that both types of skills—educational attainment of the work- force and post-school training—are highly correlated.19 Educated workers are not only more productive in performing given tasks but are thought to be more adept at evaluating new information and learning from it. The probit regression analysis shows that firms that engage in R&D and, to a lesser extent, export-oriented firms also are more likely to train. The technology literature suggests that much of the productivity gain from introducing a new or improved product or technology is realized through intensive learning by doing.20 To use new technology, firms have to adjust management, reorganize production lines, and upgrade worker skills. Export orientation also can have a salutary effect on training. Employers that export have greater incentives to train their workers to produce high- quality products meeting the exacting standards of foreign buyers and to increase labor productivity to meet competitive pressures.21 The analysis shows also that training by external sources tends to be more common among long-established �rms in which the government has a con- trolling interest and in export-oriented �rms with a high share of highly edu- cated workers. Their reliance on external training appears to be a carryover from the pre-transition period, when many state-owned enterprises had arrangements to hire graduates trained in speci�c areas by related vocational and technical training institutions. In contrast, in-house training is shaped less by the share of highly educated workers and more by the �rm’s export orientation, location in moderate investment risk regions, and R&D spend- ing. Employers appear to rely more on in-house training when industry or work-relevant skills are not available locally or when innovative activities require intensive on-the-job learning and training speci�c to the new tech- nologies being developed or used. Are firms that reported understaffing more likely to train to meet skill shortfalls? Surprisingly, no. Reported understaffing was not correlated with in-service training, nor were employer assessments of occupation-specific understaffing. In fact, overstaffed firms were more likely than those with Upgrading Skills 81 optimal staffing or lack of staff to provide in-service training. One expla- nation, consistent with table 4-1, is that shortages of specific skills can coex- ist with overall optimal staffing or overstaffing, so even overstaffed firms train. Another explanation is that there are various sources of demand for training—training is required not just to make up numerical labor short- falls, but also to meet the need for specific skills related to exporting and adopting new technologies. So why do Russian firms not train in house to meet skill shortfalls? Although information on why employers might not train or might train very little was not elicited in the Russia LME Survey, it is available in the World Business Environment Survey (WBES). The WBES asked �rms to rank state- ments about what factors influenced their decisions on how much to invest in training workers.22 Firms that did not train gave the following key reasons for not training: —use of “mature� technologies that did not require training or skills upgrading —inability to afford training, which might suggest a weakness in �nancial markets —high labor turnover of trained staff, an externality that prevents �rms from recouping the cost of training. —adequate informal on-the-job training or the ready availability of skilled workers in the labor market. Productivity and Wage Outcomes of Training In-service training makes sense only if an employer’s investment in train- ing employees and upgrading their skills yields positive returns in higher productivity and profits. If formal training is associated with higher firm- level productivity, as suggested by the preponderance of evidence from both industrial and developing countries, which source of training (in-house company programs or training by external providers) has the largest impact on productivity?23 If training increases productivity, employers also need to determine whether to share productivity gains from training with workers through higher wages and if so, how much. That decision depends on how easily skills gained from training can be transferred to other potential employers.24 A production function approach was used to estimate the productivity impact of training (see appendix 4).25 It is worthwhile to note some param- eters estimated by the models before turning to the training results. First, 82 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO consistent with results reported by other studies of the Russian economy, the estimated production function parameters of capital and labor coef�cients are positive and statistically significant. Second, consistent with the belief that education raises productivity, the production function results indicate that increased educational attainment of one year in a �rm’s workforce is associated with higher �rm-level productivity of about 4 to 5 percent. Third, the productivity of regions with moderate or high investment risk is 27 to 33 percent lower, respectively, than productivity in regions with low invest- ment risk. It appears that firms in moderate- to high-risk regions have greater incentives to train in house to compensate for skill shortfalls in the local markets and for their lower overall productivity. The production function results support the hypothesis that training is associated with higher productivity. The measure for any formal in-service training is positive and statistically signi�cant at the 1 percent level, suggest- ing that training is associated with a 22 percent increase in �rm productivity. When training is disaggregated by source, only external training is signi�- cant. But when �rms are distinguished by whether they rely only on in-house training, only on external training, or on both in-house and external training, the results suggest that using both sources of training is most productive, leading to a 28 percent increase, while using only external training sources is associated with a 17 percent increase in productivity. Similar positive effects of training are found on average monthly wages in firms (see appendix 4). In general, Russian employers pay higher wages when enterprises are large (having more than 250 employees) and export oriented, when they engage in R&D activities, and when they employ a more highly educated workforce. Consistent with the earlier training-productivity finding, enterprises that train also pay monthly wages that are 16 percent higher than nontraining firms, a difference that is statistically significant at the 1 percent level. The wage effects of training differ by training source, depending on whether firm-level or occupation-level wages are studied. The firm-level results suggest that external training is associated with the largest wage gains (18 percent), while the wage effects of in-house training programs are not statistically signi�cant. Occupation-level wages are most strongly affected by in-house training (16 percent), a result that is statistically signi�cant, but they are not affected by training from external sources. When wages are averaged across occupations, �rm-level mean wages may conceal considerable within- �rm variation in wages by skill or occupation. That dispersion of skill-wage differentials within and across �rms is better explained by in-house training Upgrading Skills 83 Table 4-2. Innovation and Training Variables Percent of �rms that Sample size provided training Independent variable Yes No Yes No Firm spent more than 1 million rubles on R&D 211 779 0.79 0.67 Firm introduced new process or technology 293 697 0.80 0.65 Firm purchased patents, licenses, or know-how 161 829 0.74 0.69 Composite innovation indicator 463 527 0.77 0.63 Source: LME Survey (2005); authors’ calculations. than by external training, possibly reflecting skill-wage premiums associated with innovating �rms that rely on in-house training. Training and Knowledge Absorption Capacity Training and knowledge absorption are complementary, in the sense that a firm’s capacity to innovate or to absorb new knowledge—and benefit from absorption—depends on the skills and training of its workforce. The pre- vious analyses already have shown in-service training and R&D spending (a crude measure of knowledge absorption) to be highly correlated. Train- ing is also highly correlated with other indicators of innovativeness—such as third-party R&D or licensing of patents and know-how, introduction of new production technologies, and export of high-tech products. The more pertinent issue is not just whether training and absorption are correlated but whether absorption is possible without a highly skilled and trained workforce. A composite indicator variable was used as the measure of absorption.26 Table 4-2 shows the distribution of �rms for each of the variables that make up the composite absorption indicator: 463 firms are defined as being “absorptive� according to their R&D spending, and those so de�ned are more likely to train (77 percent) than those that are not (63 percent). Each of the absorption variables that goes into the composite indicator is similarly cor- related with training. Bivariate probit analysis confirms that training and absorption decisions are made jointly (appendix table A4-11). Moreover, the regression analysis shows that when a firm both trains its workers and innovates, the impact on productivity is greater than it is for no training or innovating, only training, or only innovating (and it is the only significant effect in the endogenous model). The wage regression using 84 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO predicted values suggests that all three states—just training, just innovating, and both activities together—are associated with wage gains. Curiously, inno- vating but not training has the largest coef�cient. The alternative “exogenous� wage model also yields different results: “just training� and “investing in both� show wage gains but “just innovating� does not. Policy Implications Together, the analyses of the Russia LME Survey data and comparisons of their findings with those from other developed and developing countries suggest that employer perceptions of shortages of skilled and qualified workers are accurate. They tell a broadly consistent story about the nature of skill shortages: —Demand for educated and skilled workers is high and rising. —The educational and training system is underfunded beneath the terti- ary level and faces numerous challenges, including deteriorating quality and lack of response to industry’s need for skilled workers. —The industrial sector faces high labor turnover (which inhibits training) and constraints on its ability to adjust its workforce and workers’ mix of skills. —Some noncompetitive enterprises are unable to pay wages that are competitive enough to attract and retain workers who have the skills that they need. The analyses suggest that most enterprises have not responded to skill shortages by training their employees in house or by training more of them, despite the productivity and wage gains that might come from such investments. These results have implications for training policy in Russia. It is clear that Russian industrial enterprises underinvest in training their employees. While the incidence of training is high, the proportion of employees pro- vided in-house training in different skill categories is one of the lowest among the countries with data available, both high-income and developing countries. If in-service training is critical to the effective use of new tech- nologies and to productivity growth, as the literature and the estimates reported in this chapter suggest, then Russia’s underinvestment in work- force skills places it at a relative disadvantage to its OECD, BRIC, and East Asia competitors. Improving the investment climate in Russia should have a salutary effect on business operations and growth and should create incentives for the private sector to invest in both physical and human cap- ital. Policies to foster greater technological change also should induce more Upgrading Skills 85 in-service training, given the evidence of a strong training-knowledge absorption nexus. Market failures diminish employer incentives to train, and whether a policy response is appropriate depends on the nature of the market failure. While not Russia-specific, available WBES data suggest that three market failures—the high cost of training, training externalities from the turnover of skilled workers, and information problems—are key constraints for training. When costs are high, financial sector reforms to improve access to fund- ing for all kinds of investments, including training, are likely to be most important for smaller enterprises. When employer incentives to train are low because of turnover or “poaching� of skilled workers by other employ- ers, mandates or collective action to get all firms to train can help inter- nalize some of the externalities. The Human Resource Development Fund of Malaysia has, since its introduction in the mid-1990s, increased training among firms.27 When poor information is the constraint, the appropriate policy response is to disseminate widely the evidence on the productivity benefits of training, best practices in training, and information about the services offered by different public and private sector training providers and the cost of services. The Russian government should consider putting in place employer- targeted training policies to remedy the underinvestment in in-service train- ing. It can draw on the experiences of many other countries, both industrial- ized and developing, that have used payroll-levy training funds, tax incentives for employer-sponsored training, and matching grants. Designing a training policy appropriate for Russia is beyond the scope of this chapter, but the global experience with training levies28 suggests several lessons: —Employers should be closely involved in the governance of levy funds, as in Argentina, Brazil, and Chile, which have vested supervision of levies in industrial bodies. —Policies should be designed to increase competition in training provi- sion from all providers, both public and private, including employers. —Levy funds should be earmarked strictly for training and not diverted to other government uses, as happened with training levies in several Latin American and African countries. Although international evidence shows that levy schemes can increase training by enterprises, such schemes also have problems. Levies have been inequitable in the sense that they tend to benefit large employers more than small or medium employers. Employer reactions to the schemes also have 86 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO been mixed, with many firms, especially smaller businesses, feeling that a levy is simply another tax that offers them very little benefit. That may also be the response of many Russian firms that already face relatively high pay- roll taxes, unless they are reassured that funds earmarked for training will not be diverted elsewhere and that training funds will be used effectively and transparently. Training levies do not work especially well for small and medium enter- prises (SMEs), the group of enterprises in Russia that exhibited the lowest incidence and intensity of in-service training. The experiences of China, Malaysia, Brazil, and Chile suggest that SMEs are not likely to adopt training policies and that targeted training programs are required to reach such �rms. Encouraging training in small and medium enterprises may require more active approaches to address systemic weaknesses both in training and tech- nological capabilities and in access to �nance. Mexico’s experience with train- ing programs for SMEs offers some lessons. An alternative is to use matching grants, which can help to develop a cul- ture of training, although by themselves grants will not expand the training market. The most successful schemes are demand driven, implemented by the private sector and intended to sustain training markets. Chile and Mau- ritius report good results from using private agents to administer their schemes. Higher investments in training have been matched by fewer enter- prise failures. A side bene�t has been a new network of industry manage- ment training consultants available to enterprises that want to invest in enterprise-based training. Singapore has a program to build up its stock of industry trainers, and Japan’s Industrial and Vocational Training Association has trained more than 30,000 industry trainers in the past thirty years. It is important to gen- erate training capacity in enterprises and increase the propensity of workers to undertake training. Grants should not be restricted to state-run training institutions. Funds should strengthen and diversify the supply of training and stimulate demand. Strong training cultures have been established in much of Europe, in Japan, Korea, and Singapore, and (judging by the levels of in-company training) in Brazil and Chile. Whatever training policy is eventually adopted in Russia, it is imperative that enterprises and employer associations have meaningful input into the design of the training system so that it is responsive to their needs and to those of other key stakeholders. Where warranted, industry could take joint responsibility with government for the management and delivery of training, as in Brazil. Upgrading Skills 87 At the institutional level, involving Russian employers in the management of individual vocational and technical institutes should help ensure a steady flow of information about the skills needed by local industry—as well as opportunities for instructors to upgrade their knowledge and for students to be placed with employers. The Indian government is taking a similar approach in its efforts to reform the country’s moribund public sector industrial training institutes. It introduced Institutional Management Com- mittees (IMCs) in 1998 to involve employers in overseeing the operations of the industrial training institutes. There are now 350 IMCs in eighteen states, with more in the pipeline. IMCs are supported by the Confederation of Indian Industries and the Federation of Indian Chambers of Commerce and Industry, with each IMC chaired by a local industry representative. As cur- rently structured, however, IMCs have limited decisionmaking powers because most states in India do not allow industrial training institutes to exercise signi�cant �nancial authority, nor do states give IMCs incentives to revamp the training that they offer. Moreover, IMCs may not retain student fees and other nontraditional sources of revenues from, for example, deliv- ering tailored training courses to employers. Employers also could form public-private partnerships to deliver demand- driven, low-cost training that is largely self-�nancing. Malaysia’s Penang Skills Development Centre suggests how the private sector in different Russian regions can partner with state governments in the reform and management of tertiary-level professional and technical institutes. Notes 1. According to Sabirianova Peter (2001), more than 40 percent of all employed workers in Russia changed occupation during 1991–98, two-thirds of them during 1991–95. The author termed this mass occupational change the “Great Human Cap- ital Reallocation.� 2. This result was due in part to the very high proportion of the population that attended professional and technical colleges (or SSUZ, in Russian). Considering only attendance at university-level institutions (or VUZ, in Russian), Russia, at a rate of 21 percent, still ranked in the top ten countries, sharing ninth and tenth place with Japan. 3. A shift-share approach is used to decompose changes over time in educational attainment attributable to different components: one that measures the results of shifts in the industry and occupational composition of employment, holding edu- cation constant; another that measures the contribution of rising education, hold- ing industry and occupation mix constant; and a third interaction term. The 88 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO 1992–97 decomposition uses six education groups, fifty occupation groups, and fif- teen industry groups, while the 1997–2002 decomposition relies on seven educa- tion, thirty-two occupation, and nineteen industry groups. 4. See appendix 4; Schultz (1975); Bartel and Lichtenberg (1987); and Tan (2005). 5. In addition to ranking each constraint on a scale of 1 to 5, with 5 being a severe constraint, enterprises in the LME and SE surveys also were asked to identify the most severe constraint from the previous list. This alternative ranking yielded broadly sim- ilar �ndings, with lack of a quali�ed workforce being ranked third by medium and large enterprises and second by small enterprises. 6. IMEMO, various years. 7. Rosstat (2006). 8. The survey in Gimpelson, conducted jointly by the Higher School of Eco- nomics (HSE) and the Russian Public Opinion Research Center (ARPORC), surveyed 304 industrial enterprises located in thirty regions of Russia in 2003. The respondents were personnel managers. 9. World Bank (2006a). 10. Until 2002, employment in Russia was regulated by the Code of Laws on Labor (KZOT); reforms to the code in 2001 eliminated many contradictory and obsolete requirements but left the employment protection section of the code almost unchanged. The major positive change was to abolish trade unions’ veto power over mass layoffs. The new code required employers to hire employees on standard open- ended contracts with a full-time work week and restricted the use of fixed-term employment contracts to speci�c cases (prompting employers to use more tempo- rary contracts on the pretense that they met the code’s conditions). In spring 2004, Russia’s supreme court ruled against the more liberal interpretation of that section of the legislation and directed that �xed-term contracts signed illegitimately had to be treated as open ended. 11. In a recent World Bank study, Rutkowski and Scarpetta (2005), the authors argue that despite strict employment protection regulations, flexible enforcement gives CIS countries considerable labor market flexibility. 12. Regulations pertaining to layoffs are enforced better than regulations on over- time work. 13. There are simple and quasi-legal ways to deal with EPL constraints and turn labor-management relations into something close to employment at will. First, employers can pressure workers to quit voluntarily. Second, there is an informal prac- tice of pressuring workers to submit an application to quit voluntarily at the same time that the worker applies for a job. That allows managers to date the application and initiate a “voluntary� quit at any moment and at no cost. These and other infor- mal practices can result in high labor turnover driven by quits with almost no layoffs. 14. The seventeen countries clustered by region include Germany, Greece, Ireland, Portugal, South Korea and Spain (OECD); China, Malaysia, and Thailand (East Asia); Upgrading Skills 89 India and Sri Lanka (South Asia); Brazil and Chile (Latin America); and Bulgaria, Lithuania, Serbia, and Russia (Eastern Europe). 15. We adopted a common weighting scheme based (arbitrarily) on the size dis- tribution of firms in the India Enterprise Survey. In India the size distribution of micro (15 or fewer workers), small (16–100 workers), medium (101–250 workers), and large enterprises (more than 250 workers) was 40, 44, 7, and 8 percent respec- tively; the corresponding size distribution for the pooled LME and SE surveys in Russia was 12, 16, 29, and 43 percent respectively. 16. In the Malaysian ICA survey, employers ranked the skills and education of the workforce as the number one “severe� or “very severe� investment climate constraint (World Bank 2005c). 17. The sample is restricted to countries included in BEEPS (2006) or in Enter- prise Surveys (World Bank 2002–2006) that asked about in-service training and the shares of skilled and unskilled workers that received formal training. Skilled workers are de�ned to include managers, professionals, and skilled production workers, while unskilled workers include unskilled production workers and other nonproduction or other white-collar employees. 18. In the LME Survey, �rms reported shares of workers trained in several ranges, including an open-ended “35 percent or more.� In calculating shares trained, the mean for the last interval was generously assumed to be 45 percent, possibly biasing training estimates for Russia upward. Tabulations of the small sample of Russian �rms in the BEEPS also revealed low shares of workers trained relative to shares in other countries in the BEEPS sample. 19. See Tan and Batra (1995) for estimates on the education-training relationship in �ve developing countries in East Asia and Latin America and Tan (2000) and World Bank (1997, 2005c) for related training analyses for Malaysia. 20. Enos (1962); Bell and Pavitt (1992). 21. Tan and Batra (1995); Batra, Kaufman, and Stone (2003). 22. The World Business Environment Survey (WBES) was an Enterprise Survey sent to more than 10,000 �rms in eighty countries between late 1998 and mid-2000. The analyses reported in Batra, Kaufman, and Stone (2003) are based on a special sur- vey module that focused on competition, trade, �rms’ technological capabilities, and worker training in twenty-eight of the WBES countries. 23. Cross-sectional studies have found a strong positive association between in- service training and �rms’ productivity and wage levels (Tan and Batra 1995; Batra, Kaufman, and Stone 2003). 24. Becker (2002); Tan (1980); and Acemoglu and Pischke (1998). 25. Production functions are economic models used to measure the average relationships between output and inputs, such as capital equipment, labor, inter- mediate inputs, raw materials, and energy. Production functions are estimated in logarithmic form so that the estimated parameters can be interpreted as elasticities. 90 H. TAN , V. G IMPELSON , AND Y. S AVCHENKO Some studies use a gross output measure; others (including this one) rely on a value-added specification. 26. The composite indicator equals 1 if a firm has substantial R&D spending (more than a million rubles); has purchased technology licenses, patents, and know- how; or has introduced new production processes. If not, the indicator is 0. 27. Taiwan and Singapore are two other East Asia economies that have used direct reimbursement of approved training expenses, funded through payroll levies, to encourage �rms to train their employees. A training grant scheme in Taiwan led to dramatic increases in training, which continued after the program ended in the 1970s. Singapore used a levy on the wages of unskilled workers to �nance training grants to employers to upgrade worker skills. The Skills Development Fund’s aggressive efforts to raise awareness of and increase direct training among �rms led to a steady rise in training, especially among smaller �rms. See Tan (2000) for an extended discussion of the Malaysian experience and an impact evaluation of the Human Resource Devel- opment Fund (HRDF) policy. 28. Middleton, Ziderman, and van Adams (1993); Gill, Fluitman, and Dar (2000). �ve Improving the Investment Climate R AJ M. D ESAI In addition to the skill base of the workforce and the capacity of firms to absorb technology, the investment climate—government policies and actions that shape the opportunities and incentives for firms to invest pro- ductively, create jobs, and expand—also affects the productivity and growth of Russia’s economy. Following the financial crisis of 1998–99, a combination of favorable macroeconomic conditions and major regulatory reforms to the business environment in 2000–02 propelled Russia’s eco- nomic expansion. However, the Russian investment climate, although much improved in recent years, is still characterized by significant administrative costs, policy-induced risks, and formal and informal barriers to competition— all limiting the innovative and productive potential of the private sector. It also is characterized by unevenness in the treatment of different kinds of firms. Some firms benefit from relatively minor investment climate obsta- cles, while for others the obstacles are far more severe. Preferential treat- ment of some influential firms—long recognized as a problem in the Commonwealth of Independent States and Eastern Europe—continues to affect the Russian economy and can deter investment, innovation, and growth.1 Since 2003 the investment climate has also suffered from growing uncertainty about regulatory and policy changes and about the character of government-business relations. 91 92 R AJ M. D ESAI Investment climate policies provide numerous opportunities to redis- tribute resources toward favored �rms. Therefore, policies that would bene- �t the economy as a whole may not be implemented because they cannot be used to reward loyalty or to strengthen personal ties between public of�cials and particular privileged �rms. Governments may suppress competition— for example, by conferring monopolies, devising market restrictions, or tol- erating cartels. Tax systems may become riddled with special exemptions, or they may be enforced selectively. And because the investment climate encom- passes a wide range of policies—from tax rules to labor regulations to �nance—reforms to the investment climate in any country often require pro- found changes in the process of government, in the way the public sector does business, in the relationships between public of�cials and �rms, and in the mechanisms of accountability, transparency, and restraint: —Competitive legislatures would permit disenfranchised groups to chal- lenge the authority of incumbents, making it more difficult for executive branch policymakers to make implicit transfers to �rms without legislative approval. —The tendency of governments to grant preferential treatment to some �rms could be checked by increasing the transparency of decisionmaking by public of�cials and by creating vehicles through which broad constituencies can express their collective demands. —A free and independent media could make the public aware of the costs of corrupt practices. —Embodying formal rules and processes in national constitutions that establish effective veto points in the decisionmaking process—such as checks and balances on different branches or levels of government—could constrain arbitrariness and business clientelism. But countries do not have to implement proposed reforms all at once. They can reap benefits by addressing important investment climate con- straints in a way that gives �rms the con�dence to invest.2 Recent reviews of case studies around the world suggest that investment climate reforms can be initiated by seizing opportunities to place reform on the policy agenda— in much the same way that administrative reforms were implemented in the early part of this decade by a new Russian government.3 To that end, the proposals offered in this chapter are modest. They are intended to ensure that the proposals in previous chapters will be supported by com- plementary reforms of the investment climate—and that the incentives of key actors to support reforms will be strengthened. In particular, the chapter argues for a clear government commitment to establishing stable Improving the Investment Climate 93 conditions for private business, including a competition policy that allows tangible and intangible assets to be acquired and exchanged by small and medium firms. Along with that commitment, the government must give a greater voice to those who demand more market openness, it must encour- age greater investment in the nonresource sector, and it must promote fur- ther economic diversification.4 Around the world, the most innovative economies are characterized by high levels of competition. Russia is no exception. In large part, competitive pressures develop not from centralized, top-down efforts to create them from scratch or solely from the efforts of public agencies to police the common market against monopolistic behavior. They develop from a policy environ- ment that encourages businesses to start up and that enables fledgling com- panies to find customers for products without significant policy-induced costs, risks, or barriers to competition—formal or informal. As shown in chapter 3, �rms facing the weakest pressures from domestic and foreign com- petitors innovate the least. This chapter explores in further detail the obsta- cles that Russian �rms face in doing business, the consequences, and how the obstacles might be ameliorated. Cross-National Comparisons In comparing the severity of investment climate constraints across countries, it is important to evaluate the choices facing an investor planning to �nance an enterprise in different countries.5 Figure 5-1 compares the severity of con- straints in Russia with those in China and India, controlling for sector, size, year, whether �rms were exporters, and whether they were foreign-owned.6 To control for the overall propensity to complain, responses of firms are included for a question on macroeconomic instability—a catch-all to capture systemic attitudes toward the investment climate. Figure 5-1 identi�es three major problems for Russian �rms relative to problems for �rms in China and India: policy uncertainty (relative to India), tax administration (relative to India and China), and access to �nance (also relative to India and China). Russian firms are generally more likely than �rms in either of the other countries to consider those constraints binding on their activities. The opposite is true for all other constraints, indicating that Russian enterprises are less likely to complain about them than are other �rms. But in the past two years, reforms have slowed on some fronts, regressed on others. Surveys con�rm the worsening of corruption, informal practices, 94 R AJ M. D ESAI Figure 5-1. The “Russia� Effect on Investment Climate Constraintsa Policy uncertainty Tax administration Relative to India Relative to China Access to finance Tax rates Worker skills Informality Permits Cost of finance Access to land Crime Telecommunications Customs and trade Transport Corruption Labor regulations –2.5 –2 –1.5 –1 –0.5 0 0.5 1 Coefficient Source: LME Survey (2005). a. Figure shows coefficients of the “Russia� country dummy variable generated from two separate ordered- logit estimations of each of the individual constraints, with +/– 95 percent confidence intervals. In addition to the Russia dummy variable, additional control variables include size, sector, exporter, and foreign-owned dummy variables, as well as a binary variable coded 1 if the firm considered macroeconomic problems to be an obstacle (moderate, major, or severe) or 0 otherwise. and the quality of the legal system, among other areas. As figure 5-2 shows, while fewer firms considered infrastructure, finance (cost and access), land, and taxes to be major obstacles in 2005 than they did in 2002, several areas show deterioration. The percentage of firms that viewed labor (skills and regulations) as a major obstacle increased by 7 percentage points. Higher increases were registered for crime and the quality of the legal system (10 percent), policy unpredictability (10 percent), and corruption (24 percent). Many costs to firms are a normal function of commercial activity, but government policies and behavior can impose additional costs. For example, Improving the Investment Climate 95 Figure 5-2. Change in Russian Firms’ Perceptions of Investment Climate Constraints, 2002–05 Infrastructure Finance Land Taxes Regulations Labor Legal Policy uncertainty Corruption –10 –5 0 5 10 15 20 25 30 Percent change in firms considering each obstacle to be major, 2002–2005 Source: BEEPS (2006). it takes less time and money (as a share of per capita income) and requires a smaller number of procedures to start a business in Russia than in Brazil, China, or India (table 5-1). Even so, as figure 5-3 indicates, the losses that Russian firms suffer because of delivery delays, infrastructure problems (mainly utility service disruptions), crime, and bribes are approximately Table 5-1. Business Start-up Costs, Selected Countries Cost of Procedures to Time to start starting a business start a business a business (percent of per Country Year (number) (days) capita income) Brazil 2005 17 152 10.1 China 2005 13 48 13.6 India 2005 11 71 61.7 2003 12 29 8.4 Russian Federation 2004 9 36 6.7 2005 8 33 5 Source: World Bank (2006a). 96 R AJ M. D ESAI Figure 5-3. Administrative and Transaction Costs as a Percent of Sales in Selected Countries Percent of sales 25 Bribes Infrastructure problems 20 Security, protection, and theft Supply delays 15 10 5 0 Indonesia Ukraine Russia China South Africa Source: World Bank (2002–2006). 12 percent of total sales—about 50 percent larger than those suffered by Chinese firms and twice as large as those for South African firms. Corruption imposes its own costs. While there has been some improve- ment over the past few years, the average bribe paid by �rms in acquiring a business license and in their dealings with inspectors, tax collectors, and the courts increased between 2002 and 2005, in sharp contrast to the declining trend for those kinds of bribes in the former Soviet states, Central Europe, and the Baltic region. According to the BEEPS, the percentage of �rms stat- ing that bribery is frequent in �re, building, environmental, and health and safety inspections was higher in 2005 than in 2002, while the CIS region over- all exhibited a decrease. Differences among Small, Medium, and Large Enterprises Many of the foregoing findings on the investment climate were confirmed in separate surveys of large, medium, and small enterprises undertaken for this book. Firms of all sizes put taxes, uncertainty (both macroeconomic Improving the Investment Climate 97 Figure 5-4. Firms’ Most Severe Investment Climate Constraints, by Firm Size Large and medium enterprises Small enterprises Uncertainty Taxes 20% Uncertainty Taxes 31% 22% 21% Labor Infrastructure Business 15% 2% regulations 5% Business Labor Finance regulations 18% 12% 2% Land Infrastructure 6% Land Corruption and 4% Finance Corruption and 7% legal system 9% legal system 14% 12% Source: LME Survey (2005); SE Survey (2005). and policy induced), and labor problems at the top of their list of com- plaints. Figure 5-4 shows the percentages of large, medium, and small firms that ranked different categories of investment climate constraints as “the most important impediment� to their activities and development. Taxes, uncertainty, and labor top the list of major obstacles for Russian enterprises of all types. Smaller firms, however, tend to rank financial issues and land issues as their most severe obstacles. Regulatory and Policy Uncertainty Investment is deterred by uncertainty. Firms’ confidence in the future— which is based, to a great extent, on expectations of the stability and pre- dictability of government policies—is critical. A major concern of �rms of all sizes is the unpredictability of economic and regulatory policy changes and inconsistency in public of�cials’ interpretation and enforcement of regula- tions. Note that bribes and policy uncertainty are separate concerns. They can be related, in that excessive bribe taking can create costs, exemptions, and inconsistencies in regulatory enforcement for the �rms from which bribes are extracted. But bribes are more properly thought of as a “cost� of doing business, while regulatory uncertainty is more appropriately considered a “policy-induced risk.� Unlike market-based risks, risks created by policy and regulatory instability are fully under government control and—except for the risk of expropriation—are generally more dif�cult to insure against. So the 98 R AJ M. D ESAI latter risks can unnecessarily thwart investment, knowledge absorption, and competitiveness because entrepreneurs cannot predict their potential losses due to shifts in government policy. In enterprise surveys the predictability of state regulations improved little between 2002 and 2005, with about 60 percent of firms indicating that regulatory unpredictability constituted a moderate or major concern to business. Russian firms still face a very uncertain policy environment characterized by unpredictable changes in regulations and by inconsistent application of existing rules. While informal practices and crime are less of a concern than uncertainty, they also have worsened for Russian firms. According to the World Economic Forum, Russia ranks 75 among 117 coun- tries in terms of regulatory unpredictability (China ranks 48, India 50, and Brazil 65). The BEEPS confirm that policy uncertainty can generate uncertainty in other areas. Russian firms that perceive a high level of policy-induced uncertainty also have less confidence that their contractual and property rights will be enforced. The BEEPS also suggest that the views of surveyed firms on the predictability of state regulations improved little between 2002 and 2005, indicating the need for continuing the effort to reform the regulatory regime. In both of the surveys (BEEPS and LME), around 60 per- cent of firms in the manufacturing sector—and firms overall—suggested that regulatory unpredictability constituted a moderate or major business concern. Administrative, Regulatory, and Tax Issues After 2001 Russian federal authorities undertook reforms to streamline the regulatory regime and reduce administrative constraints. New laws were introduced for inspections (in 2001), licensing (2002), registration (2002), simpli�ed taxation (2003), and regulations (2003). Reforms of tax and cus- toms administration took place in the same period, with the enactment of a new land code (in 2001) and customs code (2003). These reforms reduced the administrative barriers facing the private sector and improved the busi- ness community’s perception of the regulatory regime, although some evi- dence of recent reversals was detected in joint monitoring of the regulatory regime by the Center for Economic and Financial Research (CEFIR) in 2001–05 in collaboration with the World Bank and the U.S. Agency for International Development.7 Improving the Investment Climate 99 Tax Administration The 2003 Law on the Simpli�ed System of Taxation for Small Enterprises offered a reduced number of taxes and levies to small �rms that adopt the simplified system. According to the CEFIR monitoring study, firms that adopted the simpli�ed tax system paid almost six different taxes and levies in 2005, compared with more than nine types of taxes in 2002. Those �rms also gave much more optimistic subjective assessments of taxation.8 However, the recent reforms appear to have had a small effect on busi- ness perceptions of tax rates and tax administration. More than 50 per- cent of firms still considered them a problem, and tax administration was more problematic in Russia than in the CIS as a whole or in Europe and Central Asia.9 Medium and large enterprises in particular devoted considerable employee time to interactions with officials of the Tax Inspectorate (forty- five person-days a year for a median firm), much more than with those of other control bodies such as the police or the sanitary-epidemiological agency. A median small enterprise also spent two times more person-days dealing with the Tax Inspectorate than with the other control bodies (five person-days and two person-days, respectively), although the numbers were much lower than those for medium and large enterprises. Of the eleven different ministries and federal services, the Federal Tax Service was assessed by medium and large enterprises as having much more of a negative impact on their business activities than any other ministry or federal service. Forty-six percent of respondents stated that the activity of the Federal Tax Service affected their business negatively, while no other min- istry or service was given such an assessment by more than a quarter of sur- veyed �rms. Uncertainty in Asset Ownership The landmark mass privatization of 1992–94 passed to private owners all physical assets (buildings, equipment, inventories, transport, and utilities) in 17,000 medium to large enterprises and in hundreds of thousands of small businesses.10 It did not, however, fully transfer control rights over two key assets: land and intellectual property. Both remain largely in the public domain. Although buying or leasing land and premises is less problematic for larger firms, smaller firms continue to face obstacles. Surveys suggest that while 100 R AJ M. D ESAI some �rms manage to go through the procedure relatively quickly (two to three months), for others it can easily take well over a year. As reported in the fourth round of CEFIR monitoring, in 2003 about 90 percent of the �rms trying to purchase land failed to �nish the procedure in half a year. In the �fth round, the length of the procedures came down only slightly. CEFIR’s monitoring surveys indicate that one of the chief sources of the problem is the lack of competition in real estate markets. Businesses contin- ually complain that there has been very little privatization of land and that the limited privatization that has occurred has been characterized by severe inconsistencies, nontransparency, and favoritism. Other surveys reinforce these findings with complaints about the “need to rely on connections,� excessive discretion, and a higher degree of corruption associated with real estate transactions than with most other administrative procedures; these surveys also find that potential foreign investors, in particular, were con- cerned about access to land.11 Municipal Landlords The new Land Code of the Russian Federation explicitly calls for the land beneath privatized buildings to be privatized as well, thus allowing land and buildings to be consolidated in a single property registry. But most usable land remains in the hands of municipal governments. Although new federal legislation constrains municipalities’ legal discretion in setting rents and lease terms, officials still abuse their dominant position through selective use of administrative barriers—more often to favor some firms over others or to exercise undue influence over local business development. Many regions and municipalities have instituted mechanisms to privatize real estate, but few are transparent or fair, and the authorities have little incentive to improve their procedures and thereby advance land reform. There is some evidence that regions that adopted legislation on land privat- ization ahead of the federal law (the Land Code) were also the ones that con- tinued to take the lead in land reform after the law’s enactment. While the Land Code did clarify the overall principles governing land transactions, it did not necessarily persuade unwilling regions to initiate land reform and land privatization.12 As a result, few �rms have managed to gain access to land. Among small and medium enterprises (SMEs) surveyed by CEFIR, only 6 percent owned land. From surveys conducted in �fteen regions of �rms that carried out land or real estate transactions in 2004 (mostly medium �rms), the proportion of respondents reporting ownership rights to land was about 18 percent.13 Improving the Investment Climate 101 Administrative barriers in land acquisition procedures deter many �rms from applying for land ownership. For example, even the relatively straight- forward procedure of applying for ownership rights to land under a building already owned by a business takes an average of eleven procedures involving eight agencies, seventeen documents, about 220 days, and 70,000 rubles. Land transactions also are magnets for corruption. On average, according to “busi- ness intermediaries� who carry out such transactions on a regular basis, more than half of privatization transactions involved “unof�cial payments.� Firms that carried out such procedures on their own behalf reported a somewhat lower level of bribes (about 36 percent), but about 25 percent of them were required to pay “sponsorships� to various foundations suggested by govern- ment of�cials and more than 20 percent had to pay for other burdens, such as paving a nearby road. Delineating and Protecting Property Rights Key policymakers—including representatives of the Duma, the Ministry of Economic Development and Trade, the Federal Antimonopoly Service, regional authorities, the Russian Urban Institute, and the private sector— have agreed on the need to create a competitive environment and a func- tioning land market, the lack of which has already begun to constrain eco- nomic growth. Three major issues hinder market relations in the land and real estate sectors: —Legal deficiencies. There continue to be major deficiencies in the con- ceptual approaches of the law and inconsistency among various statutes governing land and construction (use of land and forest, town planning) and specific procedural rules on registration of the associated rights. —Lack of interest on the part of municipalities and of�cials in developing competition in the land and real estate market. The real estate market has been monopolized, inflating prices, blunting equal access to the facilities for sale, and nurturing corruption. —Lack of an appropriate informational and legal basis for land market development. Most land plots have not been registered in the cadastre, and rights to real estate objects (in reality uni�ed objects, such as adjacent land, buildings, and infrastructure) are fragmented and governed by separate and sometimes conflicting laws and regulations. Intellectual Property Rights Firms also face uncertainties about the ownership of intangible assets, partic- ularly intellectual property rights (IPRs), which complicate the collaboration 102 R AJ M. D ESAI of research institutes with private �rms, inhibit technology transfer, impair the development of spin-off companies into independent and growing businesses, create conflicts of interest for the institutes, and may even give rise to conflicts between the goals of individual researchers and their organiza- tions. Russia’s IPR regime has two main weaknesses: —First, the assignment of IPRs remains unclear. In the 1990s Russia introduced patent and other IPR laws in which inventions registered through “author’s certificates� (the prevalent form of registration in the Soviet economy) and new technology developed with state funding remained de jure state property. However, there is ongoing debate about who owns the IPRs: the inventor, the inventor’s employer (research institute or enterprise, either state-owned or private), or the Russian government. —Second, registered IPRs are weakly protected due to the inability or unwillingness of public authorities to police producers and importers of pirated goods and to prosecute violators—a particular concern for foreign investors and exporters facing copyright piracy or patent infringement by domestic producers or importers. Different countries have different rules for allocating IPRs developed in private settings with public funds. In the United States the Bayh-Dole Act (1980) gave universities the right to patent all discoveries resulting from fed- erally funded research in order to make their technology-transfer activities more effective and to facilitate commercialization through patent assign- ments (sales), licenses, and spin-offs. The act reduced IPR disputes (previous U.S. law did not provide a clear answer to the question of who held the right to patent federally funded research). The rise in university patenting in the United States during the ensuing period and the reports of university discoveries generating significant licensing revenues are persuading European policymakers to consider sim- ilar legislation. Although funding of private research by the European Commission does not come with conditions on the ownership of resulting IPRs, national rules governing the ownership of results from publicly sup- ported research differ. Some countries give exclusive ownership to aca- demic inventors (Finland, Iceland, Italy, and Sweden). Others give univer- sities the right to own patents on their research (Austria, Belgium, Denmark, France, Germany, Ireland, Netherlands, Norway, Poland, Spain, and the United Kingdom). Allocating intellectual property rights is one of the most discussed problems in Russian science and technology policy, especially for intellec- tual property created with budget expenditures, because of contradictions Improving the Investment Climate 103 between the laws on patents and on public procurement. A draft new law that would resolve the contradictions by transferring intellectual property ownership to research institutions has been approved by the Council of Ministers but not officially enacted. The use and disposal of tangible assets theoretically owned by the state is poorly regulated in practice, since it is difficult to exercise control over the more than 10,000 state-owned enter- prises (the “unitary enterprises�) that currently exist. Scarcity of corporate R&D is the Achilles’ heel of the Russian innova- tion system. Business enterprises contribute less than 20 percent of national R&D expenditures, while the state finances more than 60 percent. Most government-funded research and development is carried out by research institutes, design bureaus, and business enterprises. As a result, 90 percent of all IPRs registered in Russia are created with state support. Russian inventors often file patent applications merely to register a “patent pend- ing� for their inventions. Lacking funds, many Russian firms fail to com- mercialize their inventions, something that differentiates Russian firms from firms in the developed market economies, who view patenting as simply a first step to commercialization. Since it is costly to keep patents in effect for the seventeen years allowed by law, Russian patent holders tend to stop maintaining their patents after four to five years. As a result, only 35 percent of registered patents in Russia are active, and the rest cannot be enforced if competitors copy the patented invention. Patenting abroad is prohibitively expensive for most Russian technology-intensive firms. Hence it is difficult to put the intellectual property of Russian innovators to sufficient productive use; moreover, the ability of intellectual property to strengthen innovators’ position in the markets for technology-intensive goods and services is inadequate, perpetuating Russia’s specialization in raw material industries. Despite the controversies and the incompleteness of Russia’s IPR laws, most observers agree that the content and quality of the laws correspond to international standards. The institutions and procedures to enforce the laws, however, remain weak. They need to be strengthened to comply with the WTO’s Agreement on Trade-Related Aspects of Intellectual Property Rights (TRIPs), one of the stumbling blocks in Russia’s accession to the WTO. The market share of pirated goods is high, rising to 97 percent of computer soft- ware, 85 percent of optical disks, 75 percent of audio records, and 45 percent of printed works. Russian arbitrage (commercial) courts hear up to 200 cases a year related to violations of IPRs, half of which are initiated by foreign plaintiffs. Often 104 R AJ M. D ESAI cases are resolved before a verdict is reached. In only a few cases are defen- dants found guilty. Barriers to Competition Around the world, the most innovative economies are characterized by high levels of competition. In Russia, too, firms facing the strongest pressures from domestic and foreign competitors innovate the most. Table A5-1 sug- gests that the investment climate in Russia can be improved signi�cantly by enhancing competition and strengthening consultations between the public and the private sector. Competition Policy The Russian antitrust regime, with the Ministry of Antimonopoly Policy and Entrepreneurship Support (MAPES) at its center, was put in place early in Russia’s transition and has gradually increased the severity of punishment for violations of antitrust laws. The Law on Competition and the Restriction of Monopolistic Activity in Goods Markets, passed in 1991, consisted mainly of cease-and-desist orders for violators. Subsequent amendments and sup- plementary legislation expanded the responsibilities of regulators and sanc- tions on offenders. In 1999, as part of a response to anticompetitive actions by financial institutions, the antimonopoly regime underwent further reform, establishing MAPES as a cabinet-level ministry. In 2004 the Russian Federation was the first non-OECD economy to participate in the OECD’s Regulatory Reform Programme, which reviewed Russia’s efforts to foster competition, innovation, and economic growth. The review identified several flaws in the Russian regulatory framework— structural and legal problems, excessive caseloads and limited capacity, and an enforcement mechanism characterized by weak sanctions for some transgressions and limited investigative powers. In March 2004, MAPES was replaced by presidential decree with the Federal Antimonopoly Service; some of the functions of MAPES, including consumer protection, com- modity exchange supervision, small and medium business support, and telecommunications tariff regulation, were passed to other ministries. Despite limitations, authorities in Russia have helped create more com- petitive conditions, reducing barriers to the movement of goods and ser- vices and establishing consumer protection legislation. Addressing competition law and policy in Russia requires amendments to multiple laws and decrees and to the organization and operations of the Improving the Investment Climate 105 new Federal Antimonopoly Service and the local, regional, and federal governments. More important, expectations for centralized, top-down approaches to competition policy often are based on a presumption that the reforms will yield results and change the incentives of the main actors in a somewhat predictable fashion. Experience suggests that this is far from the case in a country as large and regionally diverse as Russia—and in a government as multilayered. That does not imply that reforms of the kind recommended by the OECD review are not important. There is, however, an imperative to strengthen the demand for competition from the bottom up—to empower and support the natural constituencies for competition and market openness. Competitive Pressures Monopolistic firms tend to obtain more favorable treatment than other firms from government authorities. They tend to suffer less from any given investment constraint, to pay less in bribes to secure government contracts, to suffer less from nonpayment by customers, and to spend less on pro- tection payments. Their differential treatment is most likely due to loop- holes, exemptions, or special exceptions in existing legislation, as well as the discretionary interpretation and application of regulations by public authorities (table 5-2). Previous analyses of enterprise performance in Russia revealed that firms that received preferential treatment from federal and regional gov- ernmental authorities tended to receive tax breaks, investment credits, direct subsidies, guaranteed loans, access to state property, and the cre- ation of special economic zones on the sites of specific enterprises.14 The World Bank’s Country Economic Memorandum for the Russian Federa- tion noted that firms controlled by regional private owners as well as for- eign investors were most likely to receive preferential treatment. It also found that selective, preferential treatment toward the most politically influential firms had adverse effects on the performance of less influential regional firms.15 In a similar vein, the business surveys conducted for this book suggest that certain categories of firms face a more favorable investment climate. Firms facing weak competitive pressures from domestic and foreign manu- facturers (the least innovative �rms) tend to encounter lesser constraints in conducting business, while �rms facing greater competitive pressures also face a more severe investment climate. Because �rms facing the greatest com- petitive pressures are also the most innovative, investment climate constraints 106 R AJ M. D ESAI Table 5-2. Costs of Business and Competitive Pressure for Russian Large and Medium Enterprisesa Firms facing the least Firms facing the most Constraint competitive pressure competitive pressure Percent likelihood of paying bribes 29.5 49.2 to get things done Percent of contract value paid in bribes 0.9 2.6 for government procurement Overemployment (percent of workforce) 5.1 12.2 Percent likelihood of nonpayment by customers 29.2 71.7 Protection payments as a percent of sales 0.2 0.7 Source: LME Survey (2005). a. Estimates are based on stochastic simulations of regressions in table A5-3 in appendix 5. All estimates control for size, age, former SOE status, and location of �rms. do the most harm to precisely the companies that could serve as engines of diversi�cation of the Russian economy. (Tables A5-2 and A5-3 in appendix 5 show the relationship between the competitive pressures that �rms face and the severity of various investment climate constraints.) Figure 5-5 represents these findings. After controlling for firms’ systemic favorable biases toward the investment climate in general as well as size, age, sector, and regional location, the analysis shows that firms facing com- petitive pressures are more likely to face harsher governance constraints (such as a poorly functioning legal system, policy uncertainty, crime, and unfair competition), less access to and higher costs of finance, higher taxes and a weak tax administration, more problems with labor (skill shortages and regulations), and steeper administrative barriers (customs and licens- ing barriers).16 What accounts for the apparent punishment of the most innovative firms? State capture. Powerful private interests come to control legislative and regulatory systems, guaranteeing themselves a stream of uninterrupted benefits, protections, and rents. State capture does not, however, benefit private firms alone. Local and national public officials also benefit in numerous ways—whether through direct payments from captor firms or through political financing (campaign contributions) by firms. Meanwhile, firms that are not captors do not benefit, and they may be targeted by pub- lic officials—inspectors, tax authorities, and politicians—who extract ben- efits from them.17 Firms facing the greatest competitive pressures also pay more in bribes to obtain contracts with the government, tend to have bloated payrolls (presumably because they are restricted from laying off workers), Improving the Investment Climate 107 Figure 5-5. Investment Climate Gap between Competitive and Noncompetitive Firmsa Percent likelihood that the obstacle is major 70 Firms facing strongest 60 competitive pressures 50 40 30 Firms facing weakest competitive pressures 20 10 0 Governance Administrative Taxation Labor Finance Land Infrastructure Source: LME Survey (2005). a. Estimates are generated from stochastic simulations of regressions in table A5-2 in appendix 5, which control for firm size, membership in financial-industrial conglomerates, sector, region, and systemic bias toward the investment climate. and tend to encounter more nonpayment by customers (table 5-2). How can the most vulnerable firms protect themselves from such adverse effects of state capture? Business Associations Business associations can serve as a “leveler� among �rms. Although �rms that belonged to business associations reported more severe investment cli- mate constraints (in both objective and subjective terms) than those that did not, association members also were more likely to introduce new products and new technologies (appendix table A5-4). Moreover, �rms that did not belong to business associations and that faced strong competitive pressures were more likely to have excess labor and to pay more in bribes for govern- ment contracts; they also were less likely to believe that regulations were pre- dictable. Among �rms belonging to business associations, the gap in percep- tions between �rms facing strong and weak competitive pressures disappears. The combination of competitive pressure and business association mem- bership has the most favorable effect on investment climate costs and 108 R AJ M. D ESAI uncertainties (appendix table A5-5). The problems of excess employment, regulatory burdens, regulatory unpredictability, and severity of bribe pay- ments for government contracts are less constraining for competitive firms belonging to business associations than for either members of business associations or for firms facing competitive pressures alone. There are two possible explanations for this “beneficial� effect. First, Russian firms belonging to business associations report that one of the most valuable services that associations provide is access to information on regulations, pending legislation, the economic policymaking process, stan- dards, accrediting, and markets—information that may better equip busi- ness association members to anticipate regulatory and policy shifts, to pre- pare for changes in laws, and to identify “loopholes� or ways around often complex regulatory regimes. Second, it is possible that firms—especially smaller and medium firms—that band together are better protected from rent-seeking or predatory public officials. Consequently, they may pay less in bribes for commonplace tasks and procedures than firms outside busi- ness associations. Russia’s SME associations have achieved considerable success in forging strong working relationships with their government counterparts, and Russian business associations as a whole have been characterized more by positive market-supporting behavior than by the negative rent-seeking behavior typically assumed of them. Ineffective representation at the national level had been a concern for small business leaders for some time, so they approached the government to gauge its receptivity to working with a uni�ed association of entrepreneurs. The response was very positive. The administration’s support prompted the smaller associations to unite, leading to the creation of the Association of Organizations of Small and Medium Enterprises (OPORA) in 2001. OPORA’s mission is “to create positive conditions for the development of entrepreneurship in Russia, to represent and defend the interests of entre- preneurs in dialogue with all levels of government, and to participate in the creation of a national ‘middle class’ on the basis of the best traditions of national entrepreneurship and Russian business culture.�18 One of OPORA’s �rst major projects, participating in the government’s program to reduce administrative barriers, focused on improving conditions for Russian entrepreneurs by reducing the role of government rather than by lobbying the government for concessions. To foster innovation, OPORA focuses on changing the environment for Russian small businesses. As large consumer goods companies enter the market and consumers become more Improving the Investment Climate 109 discerning, the small consumer goods trading operations that helped so many entrepreneurs get started have much less potential. Business associations can, however, quickly become instruments of cronyism. Consultative mechanisms, therefore, should be inclusive, wel- coming input from their core membership. Broadening policy dialogues to include representatives of a wider range of interests—including consumers, taxpayers, and owners and employees of smaller businesses—can enfran- chise previously excluded groups. Recommendations The enterprise surveys conducted for this report—and comparisons of their findings with those from other developed and developing countries— suggest that employer warnings of policy constraints are real and should be heeded. They tell a broadly consistent story about the investment climate hurdles encountered by potentially innovative, dynamic firms. What should be done to improve the competitiveness of Russian firms and unlock their full innovative potential? Four things stand out: —Privatize municipally held land. —Improve the allocation of intellectual property rights. —Strengthen the consultative basis for regulatory decisions. —Adopt one of several review mechanisms to deal with existing regulations. Privatize municipally held land. Improving the process for privatizing land requires greater use of auctions and tenders for vacant land and greater transparency with respect to procedures. It also requires lower prices, espe- cially for land beneath already privatized buildings. Many bureaucrats involved in land and real estate privatization and registration appear to cre- ate or threaten delays to encourage payments for “speedy processing.� One way to deter such behavior would be for the Ministry of Economic Devel- opment and Trade to draft new legislation to impose time limits on the pro- cessing of applications; a nonresponse from an agency would be taken as approval of an application, and of�cials would be required to submit of�cial written explanations for any refusals, citing objective reasons. The Ministry of Economic Development and Trade should also require annual statistical reporting by regions and municipalities regarding land and real estate own- ership, lease rights, privatization transactions, modes of privatization, and prices. The Federal Antimonopoly Service might use of�cial statistical data and survey data to identify possible abuses of market power by public sector 110 R AJ M. D ESAI bodies (such as municipal administrations) and to prosecute such cases as anticompetitive practices. Improve the allocation of intellectual property rights. A current draft of the Civil Code allows research organizations to become owners of intel- lectual property rights for technologies developed using government funds “provided that the procurement contracts do not specify otherwise.� Research and business communities are rightly concerned that this open- ended provision would allow public authorities to continue to exercise ownership for subsequent IPRs and prevent closer cooperation between innovators and firms. At the end of 2005 an important step was taken toward improving the legal regulation of intellectual property created with public funding. Accord- ing to Decree 685, “on the procedure for disposing of the rights to the results of activity in the sphere of science and technology,� the rights to the results of public-funded activity may be transferred to the inventor or researcher (with some notable limitations). This decree has not yet granted the direct authors of inventions the right to dispose of their intellectual property fully and at their own discretion. And several aspects of the decree need to be further specified, including the conditions for transferring the rights to the inventor or researcher, the obligations of the state in the early stages of tech- nological development, and the procedures for determining the amount of the compensation. Strengthen the consultative basis for regulatory decisions. Difficulties in informing market participants about new and forthcoming legal and regula- tory changes stem in part from Russia’s un�nished transition. A degree of regulatory instability is thus inevitable and generally well understood by mar- ket participants. But the government should intensify its efforts to develop adequate mechanisms to consult the business community and inform market participants in advance of new measures. Regulatory transparency and pre- dictability are especially important for small domestic investors and for for- eign investors, particularly potential investors, because they are less familiar than incumbent �rms with the national legal and regulatory environment. Small and medium �rms, moreover, are a natural constituency to demand openness and competition in markets. Government consultations with business groups can produce higher- quality regulations that achieve legitimate aims at the least possible cost. They also permit basing regulatory decisions on objective analysis and trans- parency. But they should be inclusive, obtaining input from the groups’ core membership. Broadening policy dialogues to include representatives of a Improving the Investment Climate 111 wider range of interests, including consumers, taxpayers, business owners, and employees, also can enfranchise groups previously excluded from poli- cymaking. Russia now has a fairly well-developed network of business asso- ciations representing large and small �rms alike; in practice, however, the associations rely heavily on personal ties with regional and federal govern- ment authorities for access and information, and lack of connections often is considered a strong barrier to access.19 Public disclosure of regulatory agreements as part of a broader framework for better information sharing by the government can promote consumer rights, encourage rule-based enforcement of obligations, and reduce incen- tives for corruption. Several countries have standards for public disclosure— say, by posting the full content of infrastructure contracts and regulatory and administrative procedures on government websites. Adopt one of several review mechanisms to deal with existing regulations. Although licensing and regulatory reforms in 2001–03 streamlined admin- istrative barriers, business regulation in Russia often is complicated by older rules that are still on the books. The Russian Federation should con- sider implementing one of the approaches used in Mexico, Hungary, and other countries to eliminate obsolete regulations. Requiring governing authorities to document and justify all business regulations to the national governing body by a given deadline, for example, has yielded positive results in several countries. Notes 1. Yakovlev and Zhuravskaya (2005). 2. World Bank (2004e, p. 57). 3. Kikeri, Kenyon, and Palmade (2006). 4. World Bank (2005a). 5. It is dif�cult, however, to do so by looking at mean responses of �rms to ques- tions on Enterprise Surveys regarding their perceptions of the investment climate, for three reasons. First, the totals do not control for different distributions of �rm types across countries and thus do not accurately show the “country� effect of �rms’ locations. Second, the scoring differs between Russian firms (four categories of severity) and �rms from Brazil, China, and India (�ve categories). Third, the totals do not account for the propensities of �rms to “complain� about investment climate constraints in general. 6. The benefit of the ordered-logit is that it enables analysis of the effects of various explanators on changes in categories of severity, rather than on the actual likelihood that a firm selects a particular category—and thus avoids difficulties in 112 R AJ M. D ESAI comparison that may arise due to the fact that the number of allowable categories differs across countries. 7. CEFIR (2005). 8. CEFIR (2005). 9. BEEPS (2006). 10. Ownership was transferred primarily through a system of vouchers distrib- uted to Russian citizens, with certain reservations of ownership made for enterprise insiders. 11. FIAS (2004a, chapters 1 and 4). 12. A notable exception to this is Tomsk Oblast, which made use of the opportu- nity to promote reform. 13. FIAS (2004a). 14. Slinko, Yakovlev, and Zhuravskaya (2003). 15. World Bank (2005d). 16. Investment climate constraints were generated using principal component weightings of individual impediments, generating normally distributed, continuous variables with mean 0. The results also control for �rm size, whether or not the �rm is a member of a holding company or �nancial-industrial group, and a �rm-speci�c, systemic “optimism� or favorable bias toward the investment climate. Sectoral dum- mies also are included. 17. Frye (2002). 18. Yermakov and Kaganov (2000). 19. OECD (2006); Aidis and Adachi (2007). Appendix 1 The Large and Medium Enterprise (LME) Survey sampled approximately 1,000 medium- and large-scale manufacturing enterprises. The survey strat- i�ed the sample according to the following characteristics: sectoral af�liation, size (based on the number of permanent employees), and location. Sectoral Classi�cation Table A1-1 lists sectoral activity and approximate number of firms in the population based on the 2003 Russian census of enterprises. The selected sectors of medium and large enterprises included around 75 percent of all enterprises in manufacturing, which were responsible for 87 percent of total output.1 Size In addition, the sample was strati�ed according to enterprise size, in order to ensure adequate representation of both medium and larger enterprises. A small number of very large enterprises (more than 10,000 employees), how- ever, were excluded from the survey (heads of very large enterprises being harder to interview, the time and effort required to gather all necessary infor- mation from the largest �rms being more dif�cult, and so forth). Location It is likely that regional factors could affect enterprise competitiveness (due to regional differences in the investment climate as well as regional wealth). Thus, the sample was further stratified by region. The LME Sur- vey, for logistical reasons, did not sample all eighty-nine oblasts and autonomous regions of the Russian Federation. Rather, firms from forty- nine regions were sampled from both urban and rural locations. One hun- dred forty-two enterprises were in Moscow and the Moscow Oblast, and seventy-four were in Saint Petersburg and its oblast. The two capitals included more than 20 percent of the surveyed enterprises, which corre- sponds to the share of those regions in the “universe� of enterprises. About half of the enterprises in the sample were located in regional capitals, about 40 percent in noncapital cities, and about 10 percent in the country- side (villages and settlements). 113 114 Appendix 1 Table A1-1. Sectoral Structure of Russian Manufacturing Firms Number Output of medium Percent (fourth Percent and large of the quarter of total Code Activity enterprises total 2004)a output D Processing industries 18,969 100.0 5,018.6 100.0 DA15 Food products 5,129 27.0 969.8 19.3 DB Textile and sewing industry 1,733 9.1 76.5 1.5 DD Timber and woodworking industry 780 4.1 82.6 1.6 DG Chemical production 628 3.3 394.0 7.9 DJ Metallurgy and metalworking 1,299 6.8 1,480.3 29.5 DK Machinery and equipment 2,412 12.7 337.9 6.7 DL Electrical, electronic and optical equipment 1,602 8.4 319.1 6.4 DM Transport vehicles and equipment 834 4.4 697.2 13.9 Total number of enterprises 14,417 76.0 4,357.4 86.8 Source: Rosstat (2006). a. Output in billions of rubles. Table A1-2 compares the �nal sample to the universe of �rms in the enter- prise census. The table shows some oversampling of �rms in sectors with a rather small number of enterprises (chemical and timber sectors). In gen- eral, however, the sampling structure was close enough to the universe of �rms. Similarly, the distribution of employees shows some oversampling in the timber and chemical sectors. Table A1-3 shows the distribution of enterprises by legal form of organi- zation. The majority of surveyed enterprises were joint-stock companies (opened or closed). About 9 percent of the enterprises in the sample were state-owned enterprises. About 80 percent of �rms were established during the Soviet period or earlier, while 20 percent began operations after 1991. Forty-�ve percent of the enterprises exported their products directly or through intermediaries (but the share of revenue attributable to exports was more than 10 percent for only about 20 percent of them). About one-third of surveyed enterprises were a part of larger structures (holding companies). An extensive analysis of the conditions facing small enterprises (less than 100 employees) was beyond the scope of this book. Nevertheless, a second survey of limited scope (from four regions and fourteen industrial sectors) of 304 small enterprises (SE Survey) also was completed. Table A1-2. Estimation of Sample Parameters, by Sector in the Selected Regions Electrical, Metallurgy Machinery electronic, Transport Food Textiles and Timber and and metal and and optical vehicles and production garments woodworking Chemicals working equipment equipment equipment Parameter (DA 15) (DB 17–18) (DD 20) (DG 24) (DJ 27–28) (DK 29) (DL 30–33) (DM 34–35) Total Number of enterprises in the database of 1,886 655 346 390 766 1,096 1,031 556 6,776 the State Statistics Committee (>100 and <10,000 employees across �fty regions) Ideal proportional number in the sample 281 98 51 58 112 162 154 83 1,000 based on the enterprise census Adjusted number in the sample 253 90 90 90 101 146 139 90 1,000 (minimum ninety; the rest are represented proportionally) Actual number of enterprises in the sample 248 92 84 88 103 155 142 90 1,002 Distribution (all �rms) 28.00 9.87 5.14 5.79 11.37 16.27 15.31 8.25 100.00 Distribution (sample �rms) 24.75 9.18 8.38 8.78 10.28 15.47 14.17 8.98 100.00 Employees (all �rms) 651,023 268,227 157,240 422,081 621,225 640,307 688,839 655,099 4,104,041 Employees (sample �rms) 92,514 38,854 43,212 84,830 79,795 91,555 90,138 100,514 621,412 Distribution of employment (all �rms) 15.86 6.54 3.83 10.28 15.14 15.60 16.78 15.96 100.00 Distribution of employment (sample �rms) 14.89 6.25 6.95 13.65 12.84 14.73 14.51 16.18 100.00 Source: LME Survey (2005). 116 Appendix 1 Table A1-3. Classi�cation of Enterprises in the Final Sample, by Organizational and Legal Status Number of Percent of Enterprise status observations the sample Limited liability company 229 22.85 Public corporation (joint-stock company) 480 47.9 Closed corporation (joint-stock company) 229 22.85 State-owned enterprises and other forms of enterprises 56 5.59 Source: LME Survey (2005). Note 1. As part of its effort to harmonize with international norms, in 2004 the Russian statistical authority changed its sectoral classification nomenclature from the Soviet-era classi�cation based on “industries� (known by its Russian acronym OKONH) to one based on “economic activities� (OKVED). The newer system largely conforms to the EU’s Classi�cation of Economic Activities (NACE) and the Interna- tional Standard Industrial Classi�cation (ISIC). Appendix 2 The estimation methodology uses simple log-linear equations to test for the correlation between measures of innovative activity and measures of productivity and growth. The correlations reported include several key characteristics of firms known to be related to productivity and growth from other studies. They include size, ownership (new firm established after 1991—the dataset includes virtually no state-owned enterprises), export activity, and regional characteristics. The regression results are shown in table A2-1. The �ndings can be summarized as follows. First, absorptive activity, how- ever measured, is basically uncorrelated with total factor productivity (TFP) (as shown in column 1, when TFP is estimated using the log of value added, and column 3, when TFP is estimated using the log of sales), where TFP is productivity after accounting for capital and labor inputs. The exception is R&D spending, which is positively though not robustly correlated with higher TFP. This result is supported by work done on other countries. Wiesner (2005) shows that R&D activities are an important factor in explain- ing productivity growth for OECD countries. He further suggests that the elasticities of total factor productivity to R&D are lower in Europe and Japan than in the United States, suggesting more ef�cient use of R&D capital in the United States, the existence of greater enabling institutions, or both. The fail- ure to �nd a correlation between innovative activity and TFP is essentially a multicollinearity problem. Innovative activity is strongly and positively correlated with both the capital-labor ratio (column 5) and labor productivity (columns 2 and 4). In essence, �rms that are absorbing and creating products and technology are �rms that also have lots of �xed capital, and after the endowment of �xed capital is accounted for, there is little for absorption to explain. That may be either a measurement problem (measuring the flow of capital services is notoriously difficult, measuring R&D activity even harder) or a kind of double-counting problem (many of the R&D activity measures in the LME Survey imply fixed investment, which is also captured in the fixed-capital measure). The findings for training measures are similar but stronger; the differ- ence is that the multicollinearity problem is lower or absent. Training measures are positively correlated with both TFP and labor productivity— and positively correlated with the capital-labor ratio. The explanation may lie in how labor and capital inputs are measured. Labor is measured in 117 Table A2-1. Innovation, Skills, and Firm Characteristics versus Productivity and Growth Measures in Russiaa (7) (1) (2) (3) (4) (5) (6) Log (8) (9) TFP Log L TFP Log Log TFP sales Log L Log K Item (VA) VA/L (sales) sales/L K/L growth growth growth growth Innovation Financed introduction of new product 0.101 0.196** 0.040 0.264** 0.263** 0.064* 0.094** 0.038* 0.044 Financed introduction of new technology 0.046 0.185* 0.025 0.224** 0.138 0.038 0.068* 0.040* 0.098* Exported advanced technology �nished products 0.074 0.249** 0.010 0.187** 0.339** 0.011 0.011 0.021 −0.022 Acquired tech innovation Through own R&D 0.148* 0.275** 0.068 0.175** 0.282** 0.045 0.045 0.034 −0.024 Through third-party R&D 0.029 0.108 0.030 0.141* −0.002 0.072* 0.079** 0.022 0.056 Through purchase of machinery or equipment 0.087 0.179* 0.034 0.133 0.098 0.039 0.050 0.033 0.062 Through purchase of patent, licenses, or know-how 0.034 0.150* −0.012 0.246** 0.219* 0.047 0.087** 0.053** 0.088* R&D/sales 0.165 0.291** 0.058 0.269** 0.283* 0.037 0.058 0.025 0.117 R&D expenditures > 0 −0.005 0.004 −0.002 −0.004 0.004 −0.001 0.003 0.005* 0.008 R&D expenditures > 1 million rubles 0.198* 0.388** 0.076 0.329** 0.504** 0.075* 0.071 0.030 0.030 Skills Percent of workforce with higher education 0.011** 0.016** 0.003* 0.017** 0.009* 0.002 0.001 0.001 −0.004* Percent of workforce with any formal training 0.215** 0.370** 0.077 0.346** 0.362** 0.043 0.061 0.058** 0.040 Other characteristics SME −0.144 −0.291** 0.057 −0.277** −0.374** −0.070* −0.117** −0.070** −0.139** New private �rm 0.286** 0.111 0.090 0.156 −0.425** 0.020 0.097* 0.052 0.098 Exporter 0.133 0.296** 0.050 0.258** 0.405** 0.021 0.045 0.035* 0.043 To CIS countries 0.011 0.378** 0.042 0.332** 0.436** 0.010 0.034 0.035 0.022 To developed countries −0.037 0.118 0.038 0.113 0.453** 0.007 0.018 0.019 0.107 To developing countries −0.064 0.246** 0.062 0.209** 0.433** 0.067* 0.089** 0.043 0.097 National/regional capital 0.262** 0.302** 0.121** 0.302** 0.224* −0.016 −0.026 −0.017 −0.092* Population > 500k 0.301** 0.353** 0.165** 0.380** 0.320* 0.033 0.025 −0.019 −0.022 Regional IC risk −0.163** −0.178** −0.082** −0.177** −0.149* −0.011 −0.01 0.003 −0.041 Observations 657–784 708–847 675–815 776–989 896–917 696–890 765–975 765–988 712–911 a. TFP (VA): Log (value added in 2004). Value added in 2004 = (sales in 2004 − purchase of materials in 2004) /(number of employees in 2004). TFP (sales) = Log (sales in 2004). TFP growth: Log (value added in 2004) − Log (value added in 2003). *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. Appendix 2 119 numbers of people, so there is no correction for “labor quality.� Fixed cap- ital, by contrast, is measured by value, so high-quality capital gets a larger value than low-quality capital. Since quality is not captured by our labor input measure, there is an observed relationship between labor quality (skills) and TFP. The picture is similar for growth of TFP, inputs, and output (columns 6–9). Most knowledge absorption measures are uncorrelated with TFP growth. Only a few are positively correlated, and even then the statistical sig- ni�cance is not great. But absorption is positively correlated with the growth of both inputs and output: employment growth, growth of fixed capital (investment), and growth of sales. That can again be interpreted as a kind of multicollinearity problem, because absorption activities are correlated with measures of growth of inputs and output but not with the residual (TFP). The results for training measures are complementary: they are correlated with growth in sales and especially employment. The �ndings for size (SME dummy) and new private ownership comple- ment this picture. Smaller �rms have lower labor productivity levels, but that is driven by their lower capital intensity. Thus, when a TFP equation includes a measure of �xed capital, the TFP levels of SMEs are not signi�cantly lower than those for larger �rms. New private �rms also are less capital-intensive than privatized �rms, but they do not have lower labor productivity, and their TFP, after accounting for labor inputs are accounted for, is higher. SMEs are shrinking rapidly in both inputs and output, so their TFP growth is only modestly lower than that of larger �rms. New private �rms are growing somewhat faster than privatized �rms in terms of outputs but not inputs, but the difference is not great, so their TFP growth is no faster. Export activity is not associated with higher TFP; exporting �rms are capi- tal intensive, and that explains their higher labor productivity. Interestingly, �rms growing in both TFP and sales are the ones exporting to developing countries rather than to the developed West or the Commonwealth of Inde- pendent States. Finally, there is a very strong relationship between city type (national or oblast capital) or city size and productivity: �rms in the large cities have much higher levels of productivity. That is true of TFP, labor productivity, and cap- ital intensity. When we move to growth, however, the city type–population relationships disappear. Roughly speaking, the duality gap between big- advanced and small-backward cities is not changing: low-productivity �rms in the smaller cities and rural areas are not falling farther behind �rms in the major cities and capitals, but they are not catching up either. 120 Appendix 2 The regional investment climate (IC) risk measure, from an external source, captures a single classification (low, medium, high).1 The IC risk measure is strongly negatively correlated with productivity (a move up one category is associated with 20 percent lower productivity). The IC risk meas- ure is not (as may have been supposed) correlated with other characteristics of firms, such as industry and size, that also influence productivity. The industrial structure of the sample, as well as the average size of �rms, is very similar in groups of regions with different risks. Table A2-2 presents detailed data on innovation activity in Russia and the selected comparator countries. The �rst panel shows the level of R&D inputs, as measured by R&D spending as a percentage of GDP and by the number of researchers per million population. Where possible, �gures are shown for the most recent year available, plus a snapshot from one to two decades earlier. It is worth noting, in addition to the patterns and developments in Russia dis- cussed in the main text, how the countries that have made the most gains in catching up—China, South Korea, and Spain—have also significantly increased the scale of their R&D activity during the same period. The con- trast between East and West Germany in 1989 also is noteworthy—East Ger- many, like Russia and the rest of the USSR, also had a large but economically relatively ineffective R&D sector. Note the discrepancy between the �gures for R&D spending in 1990 in Russia (2.0 percent) and the USSR (5.7 percent). Both �gures come from the UNESCO yearbooks, various years. No explanation is offered in these sources, but it is possible that treatment of military R&D spending could account for some of the difference. The second panel presents R&D outputs, as measured by the number of scientific and technical journal articles2 and utility patents (“patents for invention�) granted by the U.S. Patent and Trademark Of�ce. The data in the third panel on R&D productivity (output per thousand researchers) were cal- culated using the data in the �rst two panels. Notes 1. Expert Rating Agency (2006). 2. World Bank (2006e). Table A2-2. Basic Indicators of R&D Input and R&D Productivity for Selected Countries Measure Brazil Russia India China South Africa Poland South Korea Spain Germany R&D inputs R&D as 0.93 (2004) 1.17 2004 0.85 (2000) 1.44 (2004) 0.76 (2001) 0.58 (2004) 2.63 (2003) 1.11 (2003) 2.49 (2004) percent 2.03 (1990) 0.56 (1982) 0.60 (1996) 0.82 (1983) 1.00 (1985) 0.57 (1980) 0.44 (1981) (1989) of GDP USSR West East (year) 5.66 (1990) 2.79 3.37 Researchers 344 (2000) 3,319 (2004) 119 (1998) 708 (2004) 307 (2001) 1,581 (2004) 3,187 (2003) 2,195 (2003) 3,261 (2004) per million 175 (1981) 6,697 (1990) 131 (1982) 410 (1991) 270 (1983) 1,533 (1985) 484 (1980) 510 (1981) (1989) population USSR West East 5,856 (1990) 2,236 7,819 R&D output Scienti�c 42 (2001) 109 (2001) 11 (2001) 16 (2001) 52 (2001) 149 (2001) 233 (2001) 382 (2001) 530 (2001) papers per million population 12 (1981) 132 (1993) 13 (1986) 1 (1981) 65 (1985) 107 (1985) 4 (1981) 63 (1981) 412 (1991) U.S. PTO 0.6 (2004) 1.2 (2004) 0.3 (2004) 0.3 (2004) 2.7 (2001) 0.4 (2004) 92.1 (2004) 6.2 (2004) 130.6 (2004) patents 0.2 (1981) 0.3 (1994) 0.0 (1980) 0.0 (1981) 2.0 (1983) 0.3 (1985) 0.2 (1980) 1.5 (1981) 96.0 (1991) granted per USSR million 1.7 (1980) population R&D productivity Scienti�c 104 (2000) 31 (2001) 86 (1998) 28 (2001) 164 (2001) 100 (2001) 81 (2001) 194 (2001) 165 (2001) papers per thousand researchers U.S. PTO 1.6 (2000) 0.4 (2004) 0.7 (1998) 0.4 (2004) 8.5 (2001) 0.3 (2004) 26.1 (2003) 3.3 (2003) 42.6 (2003) patents per thousand researchers Source: R&D inputs: UNESCO (various years) and OECD (2004b). R&D outputs: World Bank (2006e) and USPTO (various years). R&D productivity: derived from data for R&D inputs and R&D outputs. Table A2-3. Formal Education Indicators, Skills, and Productivity, Selected Countries South South Indicator Brazil Russia India China Africa Poland Korea Spain Germany Gross enrollment ratios by International Standard Classi�cation of Education (ISCED) level (2002) Secondary (ISCED level 2 and 3) 110 93 50 67 89 103 91 115 100 Higher education (ISCED level 5 and 6) 18 70 11 13 15 58 82 59 48 Manager/employee education level Skilled production workers (percent of workers) 40.5 78.9 63.5 13.0 44.7 89.9 85.3 76.5 85.8 Employees with university-plus education (percent of employees) 8.5 34.5 19.0 n.a. 11.5 13.6 36.8 15.8 9.8 Top manager with university-plus education (percent of �rms) 50.7 87.9 88.2 80.1 71.0 57.7 n.a. n.a. n.a. Source: UNESCO (2006) for secondary education �gures; Eurostat (various years) for higher education �gures; World Bank (2002–2006) for workforce education. Table A2-4. Investment Climate Constraintsa South South Brazil Russia Russia Russia India China Africa Poland Korea Spain Germany Constraint 2003 1999 2002 2005 2002 2002 2003 2005 2004 2005 2004 Telecoms 1.60 n.a. 1.46 1.36 1.63 1.92 1.49 1.50 1.37 1.73 1.17 >> >> >> > > >> << Electricity 2.17 n.a. 1.43 1.55 2.56 2.31 1.72 1.54 1.46 1.66 1.18 >> >> >> > << Transport 2.32 n.a. 1.61 1.63 2.11 2.16 1.88 1.62 1.65 1.91 1.55 >> >> >> >> > Land access 1.96 n.a. 1.67 1.67 1.75 1.90 1.26 1.55 1.44 1.74 1.63 >> > << < Land title n.a. n.a. 1.84 1.63 n.a. n.a. n.a. 1.65 1.41 1.73 1.48 < Tax rates 3.80 3.62 2.60 2.72 2.58 2.68 2.31 3.43 2.14 2.25 2.58 >> >> << >> << << Tax administration 3.45 3.27 2.72 2.80 2.50 2.41 1.86 3.00 1.82 2.01 2.42 >> >> << << << > << << << Customs 3.15 2.25 2.39 2.24 2.48 2.45 2.38 2.61 1.92 2.05 2.25 >> > > >> << Licensing 2.60 2.37 2.12 2.21 2.14 2.01 1.46 1.96 1.51 2.03 1.85 >> < << << << << Anticompetitive practices 3.29 2.62 2.30 2.35 2.22 2.22 1.98 2.49 1.97 2.13 2.05 >> > << << << Labor regulations 3.46 1.90 1.70 1.96 2.42 2.50 2.86 2.68 1.68 2.26 2.26 >> << >> >> >> >> << >> >> Labor skills 3.09 n.a. 2.19 2.48 2.13 2.68 2.77 2.40 1.73 2.37 1.95 >> < << > >> << << Financial access 3.18 n.a. 2.31 2.06 2.13 2.32 1.76 2.80 1.85 2.15 2.04 >> >> << >> Financial cost 3.70 3.49 2.37 2.53 2.33 2.32 2.11 3.21 1.94 2.15 2.22 >> >> < < << >> << << << (continued) Table A2-4. Investment Climate Constraintsa (Continued ) South South Brazil Russia Russia Russia India China Africa Poland Korea Spain Germany Constraint 2003 1999 2002 2005 2002 2002 2003 2005 2004 2005 2004 Policy uncertainty 3.76 3.54 2.88 2.91 2.42 2.62 2.21 3.22 2.77 2.11 1.77 >> >> << << << >> << << Macro environment 3.80 3.62 2.82 2.80 2.37 2.58 2.71 3.25 2.93 2.34 2.10 >> >> << < >> << << Legal system 2.80 2.22 1.97 2.15 n.a. n.a. 1.87 2.55 1.54 1.77 1.61 >> << >> << << << Corruption 3.43 2.59 2.20 2.32 2.82 2.42 2.08 2.31 1.61 1.64 1.43 >> > >> < << << << Crime 3.05 2.63 1.94 1.93 2.12 2.15 2.59 2.13 1.24 1.60 1.24 >> >> > > >> > << << << Organized crime n.a. 2.57 1.75 1.73 n.a. n.a. n.a. 1.91 1.15 1.49 1.26 >> > << < << a. Investment climate evaluated by �rms responding to World Bank Enterprise Surveys. Rows correspond to answers to the question “How much of an obstacle is X to the operation of your business?� (1 = not a problem, 4 = very big problem). Results use manufacturing firms only. Entries are country means, controlling for firm size; statistical signi�cance is vis-à-vis Russia in 2005. Signi�cance and sign at the 5 percent and 1 percent levels are indicated by >>, >, <, and <<. Appendix 3 Table A3-1. Descriptive Statisticsa Standard Independent variable Observations Mean deviation Minimum Maximum Size (< 250) 1002 0.4241517 0.4944604 0 1 New private �rm 1002 0.1996008 0.3998999 0 1 Workforce tertiary education 979 19.25506 13.48214 0 100 Website 996 0.7439759 0.4366545 0 1 Firm holding company 1001 0.3226773 0.4677341 0 1 Exporter 1002 0.4530938 0.4980435 0 1 Foreign owner 808 0.0371287 0.1891943 0 1 New/improved product 1002 0.4560878 0.4983167 0 1 New/improved technology 1002 0.2954092 0.4564543 0 1 Export technology- 1002 0.2315369 0.4220253 0 1 intensive products R&D expenditure 961 0.2237253 0.4169571 0 1 (> 1 million rubles) Third-party R&D 1002 0.2215569 0.4155018 0 1 Machinery and equipment 1002 0.4491018 0.497651 0 1 Intellectual property 1002 0.1007984 0.3012119 0 1 ISO certi�cation 981 0.3792049 0.4854367 0 1 Financial constraints 1002 0.1896208 0.3921967 0 1 Macroeconomic instability 1002 0.5698603 0.4953427 0 1 Access to �nance 1002 0.8313373 0.3746407 0 1 Domestic competitive pressure 1002 0.6706587 0.4702088 0 1 Imports competitive pressure 1002 0.4510978 0.4978513 0 1 Foreign producer competitive 1002 0.2844311 0.4513684 0 1 pressure Reorganized organizational 1002 0.5000000 0.5002497 0 1 structure Hired external consultants 1002 0.1506986 0.3579335 0 1 Outsourced to third party 1002 0.0868263 0.281721 0 1 Automated inventory 1002 0.3103792 0.4628799 0 1 management system Introduced input quality control 1002 0.6007984 0.4899789 0 1 Source: LME Survey (2005). a. All independent variables are de�ned in table A3-10. 125 Table A3-2. Results of One-by-One Regressions for Introduction of New or Improved Producta Independent variable (1) (2) (3) (4) (5) (6) Size (< 250) −0.41*** −0.38*** −0.35*** −0.40*** −0.28*** −0.39*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] New private �rm 0.13 0.17 0.19 0.16 0.16 0.17 [0.28] [0.18] [0.14] [0.20] [0.21] [0.16] Food −0.02 0 0.03 −0.05 0.04 0 [0.87] [0.97] [0.84] [0.72] [0.80] [0.98] Textile −0.58*** −0.62*** −0.62*** −0.63*** −0.54*** −0.54*** [0.00] [0.00] [0.00] [0.00] [0.01] [0.01] Wood −0.57*** −0.64*** −0.60*** −0.63*** −0.57*** −0.54*** [0.01] [0.00] [0.00] [0.00] [0.01] [0.01] Chemicals −0.09 −0.11 −0.11 −0.14 −0.19 −0.1 [0.64] [0.56] [0.57] [0.46] [0.34] [0.60] Metallurgy −0.35* −0.32* −0.29 −0.30* −0.33* −0.28 [0.06] [0.07] [0.11] [0.09] [0.07] [0.13] Electric −0.05 −0.01 −0.03 −0.02 −0.06 −0.01 [0.75] [0.97] [0.86] [0.90] [0.72] [0.94] Transport −0.30* −0.34* −0.34* −0.33* −0.40** −0.33* [0.10] [0.06] [0.07] [0.07] [0.04] [0.07] Firm holding company −0.08 −0.07 −0.05 −0.06 −0.11 −0.09 [0.45] [0.46] [0.60] [0.51] [0.30] [0.36] Foreign owner −0.24 −0.22 −0.17 −0.29 −0.18 −0.15 [0.34] [0.37] [0.50] [0.25] [0.49] [0.56] Workforce tertiary 0.01** education [0.05] Exporter 0.14 [0.18] Website 0.58*** [0.00] Intellectual property 0.65*** [0.00] R&D expenditure 0.66*** (> 1 million rubles) [0.00] Third-party R&D 0.66*** [0.00] ISO certi�cation Financial constraints Access to �nance Macroeconomic instability Regulatory constraints Infrastructure constraints Domestic competitive pressure Imports competitive pressure Foreign producer competitive pressure Constant 0.11 0.18 −0.24 0.2 0.07 0.08 [0.46] [0.21] [0.14] [0.10] [0.62] [0.53] Observations 787 808 802 808 770 808 Number of region code 49 49 48 49 48 49 Source: LME Survey (2005). a. All regressions include a constant and sectoral dummies and use a random effects model. Signi�cance is given by robust standard errors clustered by regions. P values are in brackets. *Significant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. (7) (8) (9) (10) (11) (12) (13) (14) (15) −0.35*** −0.38*** −0.39*** −0.42*** −0.42*** −0.43*** −0.42*** −0.40*** −0.42*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] 0.17 0.17 0.19 0.17 0.17 0.17 0.18 0.18 0.17 [0.16] [0.15] [0.12] [0.16] [0.17] [0.17] [0.14] [0.13] [0.15] −0.04 −0.07 −0.09 −0.07 −0.06 −0.06 −0.12 −0.02 −0.07 [0.79] [0.64] [0.56] [0.62] [0.67] [0.66] [0.43] [0.89] [0.63] −0.64*** −0.65*** −0.66*** −0.64*** −0.65*** −0.65*** −0.66*** −0.70*** −0.66*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] −0.57*** −0.66*** −0.68*** −0.65*** −0.64*** −0.64*** −0.66*** −0.63*** −0.66*** [0.01] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] −0.12 −0.11 −0.11 −0.1 −0.11 −0.1 −0.13 −0.13 −0.11 [0.52] [0.57] [0.56] [0.59] [0.56] [0.60] [0.50] [0.49] [0.56] −0.31* −0.30* −0.35* −0.31* −0.34* −0.33* −0.33* −0.29 −0.31* [0.09] [0.09] [0.05] [0.09] [0.06] [0.07] [0.07] [0.11] [0.08] −0.06 −0.02 0.01 −0.03 −0.01 −0.02 −0.01 −0.01 −0.02 [0.70] [0.90] [0.97] [0.87] [0.95] [0.90] [0.94] [0.93] [0.89] −0.38** −0.34* −0.35* −0.34* −0.33* −0.34* −0.33* −0.31* −0.33* [0.04] [0.06] [0.06] [0.06] [0.07] [0.06] [0.07] [0.10] [0.07] −0.07 −0.07 −0.05 −0.07 −0.07 −0.07 −0.08 −0.07 −0.07 [0.50] [0.47] [0.62] [0.48] [0.45] [0.46] [0.44] [0.48] [0.45] −0.23 −0.2 −0.18 −0.19 −0.22 −0.2 −0.15 −0.21 −0.18 [0.37] [0.41] [0.48] [0.45] [0.39] [0.44] [0.54] [0.40] [0.46] 0.26*** [0.01] −0.23* [0.07] 0.27** [0.01] −0.11 [0.21] 0.11 [0.16] 0 [0.95] 0.27*** [0.01] 0.26*** [0.01] 0.07 [0.49] 0.14 0.30** 0.05 0.34** 0.28** 0.28** 0.11 0.14 0.26** [0.30] [0.02] [0.73] [0.01] [0.03] [0.02] [0.44] [0.31] [0.04] 790 808 808 808 802 803 808 808 808 49 49 49 49 49 49 49 49 49 Table A3-3. Results of One-by-One Regressions for Introduction of New or Improved Technologya Independent variable (1) (2) (3) (4) (5) (6) Size (< 250) −0.32*** −0.24** −0.25** −0.29*** −0.19* −0.29*** [0.00] [0.02] [0.01] [0.00] [0.06] [0.00] New private �rm −0.18 −0.15 −0.13 −0.15 −0.14 −0.13 [0.19] [0.26] [0.32] [0.26] [0.30] [0.31] Food −0.28* −0.19 −0.25 −0.29* −0.16 −0.24 [0.07] [0.24] [0.11] [0.06] [0.33] [0.13] Textile −0.42** −0.37* −0.42** −0.40* −0.28 −0.33 [0.05] [0.08] [0.05] [0.06] [0.18] [0.12] Wood 0.07 0.08 0.09 0.1 0.13 0.17 [0.74] [0.70] [0.67] [0.63] [0.55] [0.43] Chemicals 0.16 0.12 0.12 0.11 0.04 0.15 [0.42] [0.52] [0.54] [0.57] [0.83] [0.43] Metallurgy −0.17 −0.17 −0.14 −0.14 −0.12 −0.12 [0.38] [0.37] [0.44] [0.45] [0.52] [0.51] Electric −0.12 −0.11 −0.16 −0.14 −0.2 −0.13 [0.50] [0.51] [0.36] [0.42] [0.26] [0.45] Transport −0.3 −0.33* −0.34* −0.31 −0.33 −0.32 [0.13] [0.10] [0.08] [0.11] [0.11] [0.11] Firm holding company 0.11 0.11 0.12 0.12 0.11 0.09 [0.28] [0.29] [0.24] [0.26] [0.30] [0.41] Foreign owner −0.22 −0.31 −0.24 −0.35 −0.28 −0.24 [0.42] [0.27] [0.38] [0.21] [0.33] [0.39] Workforce tertiary 0 education [0.55] Exporter 0.24** [0.03] Website 0.39*** [0.00] Intellectual property 0.60*** [0.00] R&D expenditure 0.67*** [0.00] Third-party R&D 0.47*** [0.00] ISO certi�cation Financial constraints Access to �nance Macroeconomic instability Regulatory constraints Infrastructure constraints Domestic competitive pressure Imports competitive pressure Foreign producer competitive pressure Constant −0.26* −0.47*** −0.63*** −0.38*** −0.53*** −0.45*** [0.10] [0.00] [0.00] [0.00] [0.00] [0.00] Observations 787 808 802 808 770 808 Number of region code 49 49 48 49 48 49 Source: LME Survey (2005). a. All regressions include a constant and sectoral dummies and use a random effects model. Signi�cance is given by robust standard errors clustered by regions. P values are in brackets. *Significant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. (7) (8) (9) (10) (11) (12) (13) (14) (15) −0.23** −0.28*** −0.28*** −0.31*** −0.29*** −0.30*** −0.31*** −0.30*** −0.31*** [0.03] [0.01] [0.01] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] −0.13 −0.13 −0.11 −0.13 −0.13 −0.13 −0.13 −0.13 −0.13 [0.35] [0.33] [0.41] [0.32] [0.32] [0.32] [0.32] [0.32] [0.31] −0.24 −0.30* −0.31** −0.27* −0.26* −0.26* −0.30* −0.26* −0.29* [0.12] [0.05] [0.04] [0.08] [0.10] [0.09] [0.05] [0.09] [0.06] −0.42** −0.42** −0.43** −0.43** −0.39* −0.39* −0.41** −0.46** −0.42** [0.05] [0.05] [0.04] [0.04] [0.06] [0.06] [0.05] [0.03] [0.05] 0.16 0.06 0.04 0.08 0.11 0.11 0.07 0.08 0.07 [0.46] [0.78] [0.84] [0.72] [0.61] [0.61] [0.74] [0.70] [0.75] 0.14 0.14 0.14 0.16 0.17 0.18 0.14 0.12 0.15 [0.46] [0.47] [0.47] [0.41] [0.37] [0.36] [0.47] [0.53] [0.45] −0.18 −0.15 −0.18 −0.16 −0.15 −0.15 −0.16 −0.14 −0.16 [0.34] [0.44] [0.33] [0.39] [0.43] [0.44] [0.40] [0.46] [0.41] −0.2 −0.13 −0.11 −0.12 −0.1 −0.1 −0.13 −0.13 −0.13 [0.25] [0.43] [0.53] [0.48] [0.56] [0.55] [0.46] [0.44] [0.45] −0.39* −0.33* −0.33* −0.32 −0.29 −0.3 −0.32 −0.3 −0.32 [0.05] [0.09] [0.09] [0.11] [0.14] [0.14] [0.11] [0.13] [0.11] 0.11 0.11 0.13 0.1 0.09 0.09 0.1 0.11 0.1 [0.30] [0.30] [0.20] [0.34] [0.38] [0.38] [0.31] [0.30] [0.32] −0.3 −0.26 −0.24 −0.26 −0.26 −0.26 −0.24 −0.26 −0.25 [0.27] [0.34] [0.38] [0.35] [0.34] [0.35] [0.37] [0.34] [0.36] 0.30*** [0.00] −0.21 [0.12] 0.31** [0.01] 0.09 [0.37] 0.03 [0.70] 0.01 [0.94] 0.09 [0.38] 0.17* [0.08] 0.02 [0.83] −0.46*** −0.28** −0.56*** −0.36** −0.33** −0.33** −0.36** −0.39*** −0.31** [0.00] [0.03] [0.00] [0.01] [0.01] [0.01] [0.01] [0.00] [0.02] 790 808 808 808 802 803 808 808 808 49 49 49 49 49 49 49 49 49 130 Appendix 3 Table A3-4. Results of One-by-One Regressions for Export of Technology-Intensive Productsa Independent variable (1) (2) (3) (4) (5) (6) Size (< 250) −0.73*** −0.65*** −0.71*** −0.58*** −0.72*** −0.55*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] New private �rm 0.32** 0.27* 0.24* 0.28* 0.24* 0.23 [0.03] [0.06] [0.09] [0.06] [0.09] [0.11] Food −1.26*** −1.24*** −1.28*** −1.16*** −1.28*** −1.20*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] Textile −1.42*** −1.38*** −1.39*** −1.27*** −1.42*** −1.37*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] Wood −0.70*** −0.65*** −0.65*** −0.74*** −0.68*** −0.55** [0.00] [0.01] [0.01] [0.00] [0.00] [0.02] Chemicals −0.06 −0.08 −0.1 −0.1 −0.06 −0.1 [0.76] [0.68] [0.62] [0.64] [0.76] [0.61] Metallurgy −0.43** −0.45** −0.43** −0.46** −0.45** −0.50** [0.03] [0.02] [0.03] [0.02] [0.02] [0.01] Electric 0.05 0.02 0.04 0.02 0.04 −0.08 [0.79] [0.92] [0.80] [0.89] [0.80] [0.65] Transport −0.28 −0.31 −0.29 −0.27 −0.29 −0.42** [0.16] [0.12] [0.14] [0.18] [0.13] [0.04] Firm holding company −0.17 −0.13 −0.15 −0.15 −0.16 −0.18 [0.16] [0.26] [0.21] [0.22] [0.18] [0.12] Foreign owner 0.22 0.29 0.16 0.44 0.26 0.22 [0.41] [0.28] [0.56] [0.11] [0.34] [0.40] Workforce 0 tertiary education [0.81] Website 0.57*** [0.00] Intellectual property 0.54*** [0.00] R&D expenditure 0.57*** [0.00] Third-party R&D −0.05 [0.72] ISO certi�cation 0.54*** [0.00] Financial constraints Access to �nance Macroeconomic instability Regulatory constraints Infrastructure constraints Domestic competitive pressure Imports competitive pressure Foreign producer competitive pressure Constant −0.05 −0.52*** −0.08 −0.27* −0.01 −0.28* [0.77] [0.01] [0.54] [0.06] [0.96] [0.06] Observations 787 802 808 770 808 790 Number of region code 49 48 49 48 49 49 Source: LME Survey (2005). a. All regressions include a constant and sectoral dummies and use a random effects model. Significance is given by robust standard errors clustered by regions. P values are in brackets. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. Appendix 3 131 (7) (8) (9) (10) (11) (12) (13) (14) −0.65*** −0.71*** −0.72*** −0.73*** −0.67*** −0.71*** −0.67*** −0.71*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] 0.25* 0.24* 0.25* 0.24* 0.26* 0.24* 0.25* 0.25* [0.08] [0.08] [0.08] [0.08] [0.07] [0.09] [0.07] [0.08] −1.27*** −1.26*** −1.27*** −1.27*** −1.29*** −1.25*** −1.21*** −1.29*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] −1.38*** −1.43*** −1.41*** −1.42*** −1.39*** −1.41*** −1.47*** −1.43*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] −0.68*** −0.66*** −0.66*** −0.67*** −0.70*** −0.66*** −0.63*** −0.68*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.01] [0.00] −0.07 −0.05 −0.06 −0.06 −0.07 −0.05 −0.11 −0.1 [0.72] [0.79] [0.77] [0.76] [0.73] [0.81] [0.59] [0.63] −0.43** −0.46** −0.47** −0.48** −0.48** −0.44** −0.41** −0.44** [0.03] [0.02] [0.02] [0.02] [0.01] [0.02] [0.04] [0.02] 0.04 0.05 0.05 0.05 0.07 0.04 0.05 0.03 [0.84] [0.76] [0.78] [0.78] [0.67] [0.79] [0.77] [0.84] −0.3 −0.29 −0.29 −0.29 −0.3 −0.29 −0.24 −0.28 [0.12] [0.14] [0.14] [0.13] [0.12] [0.13] [0.22] [0.15] −0.15 −0.16 −0.16 −0.16 −0.12 −0.16 −0.14 −0.16 [0.19] [0.16] [0.18] [0.18] [0.30] [0.17] [0.22] [0.18] 0.24 0.25 0.26 0.23 0.28 0.24 0.25 0.27 [0.37] [0.35] [0.33] [0.40] [0.30] [0.36] [0.34] [0.30] −0.35** [0.03] 0.32** [0.02] 0.13 [0.22] 0.01 [0.90] 0.07 [0.48] −0.11 [0.33] 0.35*** [0.00] 0.15 [0.20] 0.01 −0.1 −0.02 −0.01 −0.29* 0.05 −0.22 −0.06 [0.92] [0.50] [0.88] [0.92] [0.09] [0.76] [0.13] [0.65] 808 808 802 803 808 808 808 808 49 49 49 49 49 49 49 49 Table A3-5. Regression Results for Full Model of Product and Process Innovationa Export of Introduction of a new Introduction of a new or technology-intensive Independent variable or improved product improved technology products Size (< 250) −0.16 0.02 −0.37*** [0.14] [0.90] [0.01] New private �rm 0.16 −0.18 0.36** [0.25] [0.22] [0.02] Food 0.11 0.03 −1.05*** [0.52] [0.85] [0.00] Textile −0.37* −0.24 −1.35*** [0.08] [0.30] [0.00] Wood −0.40* 0.27 −0.68** [0.09] [0.24] [0.01] Chemicals −0.3 −0.01 −0.21 [0.16] [0.98] [0.35] Metallurgy −0.3 −0.13 −0.46** [0.13] [0.53] [0.04] Electric −0.05 −0.15 −0.05 [0.79] [0.43] [0.81] Transport −0.37* −0.33 −0.35 [0.07] [0.13] [0.10] Workforce tertiary education 0.01 0 0 [0.19] [0.40] [0.49] Firm holding company −0.06 0.15 −0.13 [0.58] [0.19] [0.31] Foreign owner −0.27 −0.39 0.29 [0.36] [0.23] [0.34] Exporter 0.02 0.16 [0.86] [0.19] Website 0.34*** 0.29** 0.44** [0.01] [0.04] [0.01] Intellectual property 0.47*** 0.54*** 0.31* [0.01] [0.00] [0.09] R&D expenditure 0.41*** 0.45*** 0.40*** (> 1 million rubles) [0.00] [0.00] [0.00] Third-party R&D 0.62*** 0.43*** −0.11 [0.00] [0.00] [0.44] ISO certi�cation 0.15 0.19 0.49*** [0.18] [0.10] [0.00] Financial constraints −0.03 0.02 −0.32 [0.86] [0.91] [0.12] Access to �nance 0.15 0.28* 0.04 [0.30] [0.06] [0.79] Macroeconomic instability −0.20* 0.15 0.15 [0.07] [0.20] [0.24] Regulatory constraints 0.1 −0.07 −0.03 [0.30] [0.49] [0.79] Infrastructure constraints −0.02 −0.03 0.08 [0.79] [0.76] [0.49] Domestic competitive pressure 0.25** 0.01 −0.13 [0.02] [0.91] [0.30] Imports competitive pressure 0.13 0.03 0.30** [0.29] [0.84] [0.03] Foreign producer competitive −0.02 −0.06 0.07 pressure [0.88] [0.67] [0.63] Constant −0.82*** −1.41*** −0.96*** [0.00] [0.00] [0.00] Observations 729 729 729 Number of region code 48 48 48 Source: LME Survey (2005). a. All regressions include a constant and sectoral dummies and use a random effects model. Signi�cance is given by robust standard errors clustered by regions. P values are given in brackets. *Signi�cant at 10 percent; **signi�cant at 5 per- cent; ***signi�cant at 1 percent. Appendix 3 133 Table A3-6. Introduction of a New or Improved Product, Cross-Country Comparisona Independent variable (1) (2) (3) (4) Size (> 250) 0.147*** 0.086* 0.04 0.012 0.003 0.089 0.454 0.829 Age of �rm 0 0 0 0 0.596 0.542 0.709 0.814 Exporter 0.139*** 0.066 0.098* 0.037 0.002 0.166 0.055 0.484 China −0.420*** −0.507*** −0.414*** −0.454*** 0 0 0 0 Brazil 0.701*** 0.711*** 0.769*** 0.719*** 0 0 0 0 South Africa 0.704*** 0.674*** 0.753*** 0.704*** 0 0 0 0 ISO certi�cation 0.377*** 0.384*** 0.328*** 0 0 0 R&D intensity 0 0 0.783 0.121 R&D non-zero expenditure 0.495*** 0 Constant −0.268*** −0.350*** −0.431*** −0.589*** 0 0 0 0 Observations 3,724 3,615 3,249 3,248 Source: World Bank (2002–2006) for the comparator countries and BEEPS (2006) for Russia. a. All regressions include a constant. Signi�cance is given by robust standard errors clustered by regions. P values are in brackets. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. 134 Appendix 3 Table A3-7. Results of One-by-One Regressions for R&D Expenditurea Independent variable (1) (2) (3) (4) (5) (6) Size (< 250) −0.68*** −0.58*** −0.61*** −0.69*** −0.67*** −0.58*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] New private �rm −0.09 −0.1 −0.05 −0.12 −0.08 −0.09 [0.55] [0.51] [0.73] [0.45] [0.60] [0.59] Food −0.84*** −0.70*** −0.83*** −0.89*** −0.84*** −0.76*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] Textile −0.84*** −0.79*** −0.89*** −0.85*** −0.81*** −0.89*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] Wood −0.3 −0.35 −0.35 −0.33 −0.27 −0.23 [0.24] [0.15] [0.16] [0.18] [0.27] [0.36] Chemicals 0.3 0.23 0.24 0.24 0.27 0.27 [0.17] [0.28] [0.26] [0.26] [0.20] [0.21] Metallurgy −0.12 −0.06 −0.04 −0.01 −0.02 0.01 [0.55] [0.75] [0.83] [0.98] [0.93] [0.95] Electric 0.27 0.35* 0.30* 0.32* 0.31* 0.25 [0.13] [0.05] [0.10] [0.07] [0.08] [0.17] Transport 0 0.02 0.03 0.05 0.03 −0.05 [0.99] [0.94] [0.90] [0.80] [0.89] [0.82] Firm holding company 0.11 0.08 0.13 0.1 0.07 0.07 [0.39] [0.49] [0.30] [0.40] [0.58] [0.55] Foreign owner −0.25 −0.28 −0.21 −0.37 −0.19 −0.31 [0.43] [0.37] [0.51] [0.27] [0.56] [0.34] Workforce tertiary 0.01 education [0.25] Exporter 0.38*** [0.00] Website 0.84*** [0.00] Intellectual property 0.67*** [0.00] Third-party R&D 0.41*** [0.00] ISO certi�cation 0.49*** [0.00] Financial constraints Access to �nance Macroeconomic instability Regulatory constraints Infrastructure constraints Domestic competitive pressure Imports competitive pressure Foreign producer competitive pressure Constant −0.52*** −0.68*** −1.15*** −0.49*** −0.54*** −0.69*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] Observations 752 770 765 770 770 754 Number of region code 48 48 48 48 48 48 Source: LME Survey (2005). a. Research and development expenditure > 1 million rubles. All regressions include a constant and sectoral dummies and use a random effects model. Signi�cance is given by robust standard errors clustered by regions. P values are in brackets. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. Appendix 3 135 (7) (8) (9) (10) (11) (12) (13) (14) −0.64*** −0.64*** −0.69*** −0.68*** −0.69*** −0.69*** −0.67*** −0.69*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] −0.1 −0.1 −0.09 −0.09 −0.1 −0.09 −0.08 −0.09 [0.52] [0.51] [0.54] [0.55] [0.52] [0.56] [0.62] [0.55] −0.89*** −0.91*** −0.86*** −0.87*** −0.87*** −0.88*** −0.82*** −0.86*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] −0.84*** −0.87*** −0.87*** −0.88*** −0.88*** −0.87*** −0.92*** −0.87*** [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] [0.00] −0.38 −0.4 −0.35 −0.35 −0.35 −0.36 −0.32 −0.36 [0.12] [0.10] [0.15] [0.15] [0.15] [0.14] [0.20] [0.14] 0.25 0.25 0.27 0.25 0.27 0.26 0.24 0.26 [0.24] [0.22] [0.20] [0.23] [0.20] [0.22] [0.26] [0.22] −0.03 −0.05 −0.04 −0.04 −0.04 −0.04 0 −0.04 [0.90] [0.80] [0.83] [0.83] [0.85] [0.83] [0.99] [0.85] 0.31* 0.34* 0.32* 0.33* 0.32* 0.31* 0.33* 0.31* [0.09] [0.06] [0.08] [0.06] [0.08] [0.08] [0.07] [0.08] 0.02 0.04 0.03 0.04 0.02 0.03 0.07 0.03 [0.94] [0.83] [0.87] [0.85] [0.91] [0.88] [0.75] [0.87] 0.09 0.1 0.08 0.06 0.07 0.08 0.08 0.08 [0.46] [0.39] [0.52] [0.61] [0.57] [0.50] [0.50] [0.51] −0.2 −0.18 −0.2 −0.24 −0.27 −0.19 −0.22 −0.19 [0.52] [0.56] [0.53] [0.45] [0.41] [0.55] [0.48] [0.55] −0.35** [0.04] 0.52*** [0.00] 0.04 [0.75] 0.17* [0.07] 0.13 [0.19] 0.07 [0.54] 0.28** [0.01] 0.04 [0.75] −0.37*** −0.87*** −0.43*** −0.41*** −0.40*** −0.46*** −0.57*** −0.42*** [0.01] [0.00] [0.01] [0.00] [0.01] [0.00] [0.00] [0.00] 770 770 770 764 765 770 770 770 48 48 48 48 48 48 48 48 Table A3-8. Regression Results of Full Model for Soft Innovation and Organizational Changesa Reorganized Automated organizational Hired external Outsourced Introduced input system of inventory Independent variable structure consultants to third party quality control management Size (< 250) −0.33*** −0.36** −0.35* −0.13 −0.25** [0.00] [0.01] [0.06] [0.24] [0.03] New private �rm −0.25* −0.02 0.13 0.06 0.06 [0.07] [0.91] [0.51] [0.65] [0.69] Food −0.39** −0.03 −0.16 0.13 0.17 [0.03] [0.88] [0.52] [0.44] [0.34] Textile −0.02 −0.1 −0.76* 0.1 −0.12 [0.92] [0.72] [0.09] [0.63] [0.60] Wood −0.18 0.04 −0.6 −0.03 −0.13 [0.42] [0.89] [0.11] [0.91] [0.61] Chemicals −0.1 0.02 −0.12 0.21 0.1 [0.63] [0.94] [0.68] [0.32] [0.64] Metallurgy 0 0.35 −0.12 0.41** 0.02 [0.99] [0.14] [0.66] [0.05] [0.93] Electric −0.08 −0.33 −0.32 0.12 0.25 [0.64] [0.18] [0.23] [0.51] [0.18] Transport −0.09 0.21 −0.43 0 0.45** [0.67] [0.38] [0.16] [0.99] [0.03] Firm holding company 0.20* 0.08 0.44*** 0.04 0.14 [0.06] [0.55] [0.00] [0.73] [0.20] Foreign owner 0.36 0.67** 0.91*** 0.26 0.39 [0.20] [0.03] [0.00] [0.34] [0.15] Workforce tertiary education 0 0 0 0 0.01*** [0.75] [0.75] [0.86] [0.95] [0.01] Exporter −0.16 −0.13 0.25 −0.06 −0.12 [0.22] [0.44] [0.22] [0.64] [0.41] Intellectual property 0.06 0.26 0.36 0.1 0.27 [0.74] [0.18] [0.11] [0.57] [0.13] R&D expenditure (> 1 million rubles) 0.18 0.15 0.25 0.12 0.14 [0.18] [0.35] [0.17] [0.38] [0.32] ISO certi�cation −0.16 0.07 0.16 0.17 0.26** [0.15] [0.61] [0.35] [0.13] [0.02] Financial constraints −0.21 −0.04 0.07 0.04 −0.19 [0.18] [0.84] [0.79] [0.79] [0.27] Access to �nance 0.16 0.81*** 0.62* 0.15 0.06 [0.34] [0.00] [0.06] [0.34] [0.75] Domestic competitive pressure 0.1 0.1 0.25 0.11 −0.13 [0.38] [0.48] [0.15] [0.33] [0.26] Imports competitive pressure 0.29** −0.17 −0.07 −0.05 −0.01 [0.02] [0.25] [0.68] [0.66] [0.96] Foreign producer competitive pressure 0.05 0.22 −0.06 0.06 0.02 [0.69] [0.15] [0.74] [0.65] [0.87] Financed new product 0 0.14 −0.06 0.28*** 0.13 [0.97] [0.30] [0.73] [0.01] [0.24] Financed new technology 0.01 0.27* −0.26 0.18 0.22* [0.94] [0.05] [0.14] [0.12] [0.06] Exported technology-intensive products 0.33** −0.08 0.02 0.11 −0.05 [0.04] [0.68] [0.94] [0.50] [0.74] Third-party R&D 0.27** 0.36** 0.25 0.08 0.06 [0.03] [0.01] [0.15] [0.51] [0.62] Machinery and equipment 0.12 0.06 0.13 0.17 0.08 [0.24] [0.67] [0.42] [0.11] [0.47] Macroeconomic instability −0.14 0.32** −0.12 0 −0.04 [0.17] [0.02] [0.44] [0.98] [0.70] Regulatory constraints 0.21** −0.09 0.07 −0.09 −0.01 [0.03] [0.44] [0.61] [0.31] [0.95] Infrastructure constraints 0.1 0.13 −0.18 0.07 0.02 [0.26] [0.27] [0.27] [0.43] [0.81] Constant −0.04 −2.28*** −2.28*** −0.38 −1.01*** [0.86] [0.00] [0.00] [0.13] [0.00] Observations 731 731 731 731 731 Number of region code 48 48 48 48 48 Source: LME Survey (2005). a. All regressions include a constant and sectoral dummies and use a random effects model. Signi�cance is given by robust standard errors clustered by regions. P values are given in brackets. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. Table A3-9. Competitive Pressure and Firm Innovation and Absorptiona Introduction of a new or improved product Introduction of a new or improved technology Export of technology-intensive products Independent variable (1) (2) (3) (4) (1) (2) (3) (4) (1) (2) (3) (4) Domestic competitive 0.24** 0.17* 0.16* 0.19** 0.07 0.04 0.04 0.06 −0.19 −0.26** −0.18 −0.23** pressure [0.02] [0.07] [0.09] [0.03] [0.52] [0.67] [0.72] [0.53] [0.10] [0.01] [0.11] [0.02] Imports competitive 0.27** 0.18* 0.14 0.33*** 0.20* 0.07 0 0.16 0.39*** 0.32*** 0.29** 0.40*** pressure [0.01] [0.08] [0.18] [0.00] [0.08] [0.52] [0.97] [0.12] [0.00] [0.01] [0.01] [0.00] Foreign producer −0.11 −0.03 −0.03 −0.13 −0.09 0.08 0.09 −0.02 −0.01 −0.01 −0.03 −0.07 competitive pressure [0.35] [0.76] [0.76] [0.22] [0.47] [0.50] [0.44] [0.85] [0.94] [0.93] [0.79] [0.55] Food −0.06 0.11 0.16 −0.08 −0.26 −0.15 −0.07 −0.28** −1.16*** −1.13*** −1.08*** −1.19*** [0.71] [0.44] [0.28] [0.53] [0.10] [0.31] [0.64] [0.05] [0.00] [0.00] [0.00] [0.00] Textile −0.70*** −0.47** −0.36* −0.64*** −0.46** −0.3 −0.16 −0.41** −1.48*** −1.35*** −1.20*** −1.37*** [0.00] [0.01] [0.05] [0.00] [0.03] [0.11] [0.41] [0.02] [0.00] [0.00] [0.00] [0.00] Wood −0.63*** −0.34* −0.28 −0.57*** 0.09 0.05 0.09 −0.09 −0.62*** −0.46** −0.50** −0.59*** [0.00] [0.07] [0.13] [0.00] [0.65] [0.80] [0.65] [0.61] [0.01] [0.03] [0.02] [0.00] Chemicals −0.14 −0.09 −0.12 −0.04 0.13 0.11 0.06 0.2 −0.08 −0.09 −0.08 −0.01 [0.47] [0.61] [0.51] [0.81] [0.50] [0.55] [0.74] [0.25] [0.67] [0.60] [0.67] [0.97] Metallurgy −0.30* −0.18 −0.18 −0.23 −0.14 −0.04 −0.01 −0.06 −0.41** −0.54*** −0.54*** −0.53*** [0.09] [0.28] [0.28] [0.16] [0.45] [0.82] [0.97] [0.72] [0.04] [0.00] [0.00] [0.00] Electric 0 0.03 0.04 0.08 −0.13 −0.28* −0.30* −0.15 0.05 −0.16 −0.11 0.02 [0.99] [0.84] [0.78] [0.58] [0.46] [0.08] [0.07] [0.35] [0.76] [0.32] [0.50] [0.90] Transport −0.31* −0.34* −0.40** −0.29* −0.3 −0.32* −0.31 −0.23 −0.24 −0.35* −0.33* −0.2 [0.09] [0.06] [0.03] [0.09] [0.12] [0.08] [0.10] [0.20] [0.23] [0.06] [0.09] [0.25] Size (< 250) −0.41*** −0.29*** −0.67*** [0.00] [0.00] [0.00] New private �rm 0.19 −0.13 0.24* [0.13] [0.32] [0.08] Firm holding company −0.07 0.11 −0.15 [0.47] [0.30] [0.22] Foreign owner −0.19 −0.27 0.22 [0.46] [0.33] [0.41] Exporter 0.15 0.11 0.16* 0.14 [0.10] [0.27] [0.09] [0.17] Website 0.52*** 0.44*** 0.34*** 0.27** 0.52*** 0.48*** [0.00] [0.00] [0.00] [0.02] [0.00] [0.00] Intellectual property 0.52*** 0.50*** 0.44*** 0.44*** 0.46*** 0.36** [0.00] [0.00] [0.00] [0.00] [0.00] [0.02] Third-party R&D 0.55*** 0.50*** 0.43*** 0.38*** −0.09 −0.18 [0.00] [0.00] [0.00] [0.00] [0.46] [0.16] ISO certi�cation 0.21** 0.18* 0.19* 0.14 0.58*** 0.54*** [0.02] [0.06] [0.05] [0.14] [0.00] [0.00] R&D expenditure 0.47*** 0.51*** 0.45*** (> 1 million rubles) [0.00] [0.00] [0.00] Financial constraints −0.23* −0.05 −0.34** [0.06] [0.71] [0.02] Access to �nance 0.12 0.21* 0.2 [0.29] [0.07] [0.12] Macroeconomic −0.07 0.1 0.07 instability [0.44] [0.26] [0.51] Regulatory constraints 0.04 0.01 0.02 [0.62] [0.90] [0.83] Infrastructure constraints −0.01 −0.05 0.05 [0.88] [0.57] [0.58] Observations 808 978 940 996 808 978 940 996 808 978 940 996 Number of region code 49 48 48 49 49 48 48 49 49 48 48 49 Source: LME Survey (2005). a. All regressions include a constant and sectoral dummies and use a random effects model. Signi�cance is given by robust standard errors clustered by regions. P values are given in brackets. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. 140 Appendix 3 Table A3-10. De�nitions of Independent Variables Independent variable Description Introduction of new or improved product Dummy that takes a value of 1 for firms that introduced a new or significantly improved product Introduction of new or improved technology Dummy that takes a value of 1 for �rms that introduced a new or signi�cantly improved production technology Export of technology-intensive products Dummy that takes a value of 1 for �rms that export technology-intensive products (science-intensive and high-tech products) Size (< 250) Dummy that takes a value of 1 for �rms with fewer than 250 employees New private �rm Dummy that takes a value of 1 for �rms estab- lished after 1992 and with no share of common stock owned by the government Workforce tertiary education Percent of workers with graduate education (at least 3 years of university-level studies) R&D expenditure (> 1 million rubles) Dummy that takes a value of 1 for �rms with R&D expenditure greater than 1 million rubles ISO certi�cation Dummy that takes a value of 1 for �rms with ISO certi�cation Website Dummy that takes a value of 1 for �rms with their own website Competitive pressure from imports Dummy that takes a value of 1 for �rms facing “noticeable� or “considerable� competitive pressure from imported products Firm holding company Dummy that takes a value of 1 for �rms that belong to a holding company (a group of small companies) Exporter Dummy that takes a value of 1 for �rms that export directly or through intermediaries Foreign ownership Dummy that takes a value of 1 for �rms when more than 51 percent of common stock is owned by a foreign individual or company Intellectual property Dummy that takes a value of 1 for firms that acquired technological innovation through purchase of patents, licenses, production prototypes, utility models, and general know-how from either Russia or abroad Third-party R&D Dummy that takes a value of 1 for �rms that acquire technological innovations as a result of third-party research and development Reorganized organizational structure Dummy that takes a value of 1 for �rms that reorganized their organizational structure in recent years R&D intensity Ratio of �rm’s R&D expenditure to sales Machinery and equipment Dummy that takes a value of 1 for �rms that acquired technological innovation through purchase of machinery and equipment Financial constraints Dummy variable that takes a value of 1 if the �rms did not apply for credit due to �nancial constraints such as high interest rates or unavailable collateral Appendix 3 141 Table A3-10. De�nitions of Independent Variables (Continued ) Independent variable Description Macroeconomic instability Dummy that takes a value of 1 for �rms that cite macroeconomic instability as an impediment for enterprise’s activity and development Regulatory constraints Variable generated by conducting a factor analy- sis when �rms cited regulatory constraints such as unpredictability of state regulations, labor regulations, and obtaining licenses as impediments to their development Infrastructure constraints Variable generated by conducting a factor analysis when firms cited regulatory con- straints such as irregularity of power supply and communications as impediments to their development Domestic competitive pressure Dummy that takes a value of 1 for firms facing “noticeable� or “considerable� competitive pressure from domestically produced products Foreign producer competitive pressure Dummy that takes a value of 1 for �rms facing “noticeable� or “considerable� competitive pressure from products produced by foreign �rms operating in Russia. Hired external consultants Dummy that takes a value of 1 for �rms that hired external management consultants Outsourced to third-party contractor Dummy that takes a value of 1 for firms that outsourced to a specialized third-party contractor Introduced input quality control Dummy that takes a value of 1 for �rms that introduced input quality control for materials and interim quality control for technological parts and components of production Access to �nance Dummy that takes a value of 1 when �rms got loans from banks Automated system of inventory management Dummy that takes a value of 1 for �rms that introduced automated inventory system Age The age of the �rm since its establishment China Dummy that takes a value of 1 when country is China Brazil Dummy that takes a value of 1 when country is Brazil South Africa Dummy that takes a value of 1 when country is South Africa R&D non-zero expenditure Dummy that takes a value of 1 when a �rm spends a positive amount on R&D Source: LME Survey (2005). Appendix 4 According to Barro and Lee (2001), in 2001 Russia had one of the most highly educated workforces in the world. With an average of 10.5 years of schooling for the population age 25 and over, Russia ranked seventh in the Barro-Lee sample of countries. In �gure A4-1, which compares the educational attain- ment and GDP per capita of the Barro-Lee countries, Russia is an outlier. It is signi�cantly above the �tted line in the �rst panel comparing mean years of educational attainment. Russia is ahead of other BRIC and transition coun- tries as well as most OECD countries, leading Germany by 0.7 years, Japan by 0.8 years, and the United Kingdom by 1.1 years; only the United States is ahead of Russia, with an extra 1.8 years of education. More than half (57 percent) of Russia’s population age 25 and over has attained tertiary education, which is 13 percentage points more than in Canada and more than twice that �gure in other postsocialist countries. That result is due in part to the very high proportion of the population that attended profes- sional and technical colleges (SSUZ in Russian). However, if only attendance at university-level institutions (VUZ in Russian) is considered, Russia still ranks in the top ten countries, with 21 percent, sharing ninth and tenth place honors with Japan. Russia thus appears on the surface to be well situated to take advan- tage of knowledge-based economic activities requiring a well-educated work- force and a pool of researchers. Quality of Education Its educational achievements notwithstanding, Russia fares less well on international comparisons of education spending, with negative implica- tions for the quality of education. The share of total educational expendi- tures in GDP (3.7 percent) is lower than that in developed and other tran- sition countries but comparable to educational spending in the BRIC countries (table A4-1). If we look at annual expenditures per student rela- tive to per capita GDP, Russian funding for education is skewed toward ter- tiary education, with a ratio of 34.9 percent, which is comparable to that of France but behind that of Germany and of Japan. For secondary education, by contrast, the ratio is 9.3 percent, a level comparable to levels in Indone- sia, Uruguay, and Peru. The ratio of students to teachers in Russian educational and training insti- tutions is low by international standards. In primary schools, the ratio is seventeen pupils per teacher, typical for most of the developed and transition countries; the ratio in secondary schools, at 8.5 pupils per teacher, is the 142 Appendix 4 143 Figure A4-1. Educational Attainment in Russia and Comparators Panel A: Average years of schooling, total population age 25 and over, 2000a Years of schooling 13 United States 12 11 Russia 10 Japan 9 8 7 6 5 4 3 2 1 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 10.5 11 Log GDP per capita Panel B: Percentage of population age 25–64 that has attained tertiary education, 2003b Percent 60 Russia 55 50 45 40 United States 35 30 25 20 15 10 5 8 8.25 8.5 8.75 9 9.25 9.5 9.75 10 10.25 10.5 10.75 11 11.25 Log GDP per capita Source: Panel A, Barro and Lee (2001); panel B, OECD (2005a). a. Purchasing power parity (2000), in constant 2000 international dollars. b. Purchasing power parity (2003), in constant 2000 international dollars. 144 Appendix 4 Table A4-1. Expenditures on Educational Institutions, 2002 Expenditure on education per student as percent of per-capita GDP based on full-time equivalents Expenditure on education University as percent Tertiary and of GDP for All professional advanced All all levels of Primary secondary or technical research tertiary Comparator education education education education programs education Developed countries France 6.1 18.3 30.8 35.7 33.2 33.8 Germany 5.3 17.0 26.4 21.5 44.5 41.3 Japan 4.7 22.5 25.6 35.2 44.0 43.1 United Kingdom 5.9 17.8 22.5 n.a. n.a. 40.9 United States 7.2 22.2 25.1 n.a. n.a. 56.8 Transition countries Czech Republic 4.4 12.5 21.9 16.3 40.2 37.6 Hungary 5.6 21.0 22.2 60.5 57.0 57.1 Poland 6.1 23.1 n.a. n.a. n.a. 43.2 BRIC countries Brazil (2001) 4.0 10.9 12.3 n.a. n.a. 34.7 India (2001) 4.8 14.6 26.3 n.a. n.a. 91.7 Russia (2000) 3.7 9.3 16.9 12.6 34.9 26.5 Source: OECD (2005a). lowest in the world. In universities, the ratio is �fteen students per teacher, also lower than in most developed or transition economies. Low student- teacher ratios in the face of severe underfunding at lower levels of education can be explained by very low pay in the educational sector. In 2004, the aver- age monthly wage in education was only 62 percent of the average wage in the economy as a whole and 53 percent of the average wage in industry (not con- trolling for individual characteristics). The likely consequences of low relative pay are poor selection of faculty, diminished incentives, and a lower quality of instruction. Russia’s performance on internationally comparable standardized tests supports that conclusion. According to TIMSS (Trends in International Mathematics and Science Study), which is administered to fourth- and eighth-grade students, in 2003 the scores of Russian fourth-grade students in mathematics and science were well above the international average. Russia occupied tenth and eleventh place among the twenty-eight countries that participated but lagged behind the leaders—Singapore, Hong Kong, and Tai- wan. Russia continued to perform well at the eighth-grade level, scoring above the international average for the �fty participating countries but slip- ping to fourteenth and twenty-�rst place internationally (table A4-2). Appendix 4 145 Table A4-2. Russian Student Achievement Scores, 1995–2003 International International Test 1995 1999 2003 mean score ranking TIMSS (grade 8) Mathematics 524 526 508 467 14 in 50 Science 523 529 514 474 21 in 50 2000 2003 PISA (15-year-olds) Literacy ... 462 442 480 32 in 40 Mathematics ... 478 468 486 29 in 40 Science ... 460 489 488 24 in 40 Problem solving ... ... 479 486 28 in 40 Source: OECD (2005a). On PISA (Programme for International Student Assessment), which assesses the quality of education for �fteen-year-old students, in 2003 Russia had an average literacy score of 442, markedly below the international aver- age score of 480. That put Russia in thirty-second place among the forty countries participating and some 90 to 100 points behind the scores of the leaders—Finland, Korea, and Canada. Russia’s scores on student assessments in mathematics, science, and problem solving were similarly below the inter- national average. One explanation for Russia’s lower scores on PISA is the test’s focus (unlike TIMSS) on applied knowledge. The result is consistent with the observation that Russian schools emphasize the acquisition of ency- lopedic knowledge over problem solving, innovative thinking, and creativity.1 These test scores reveal that the quality of lower secondary education in Russia is poorer than that in other developed and in almost all other transi- tion countries and that many students enter the labor market poorly equipped for work. Furthermore, when one compares Russia’s test scores on TIMSS and PISA for several years (table A4-2), it is clear that quality has dete- riorated over time. How about student performance at higher levels of education? The IALS (International Adult Literacy Survey) assesses how well equipped adults of different levels of education are for the demands of the workplace, including by testing their ability to apply knowledge to real-world situations—a core competency highly valued by most employers throughout the world. Russia has not participated in IALS, so no international comparisons can be made of how well Russian schools prepare students for the workplace at the higher levels of education not covered by TIMSS or PISA—including upper second- ary schools and vocational, professional, and technical institutes below the 146 Appendix 4 university level. The quality and relevance in the workplace of the education and training provided by these institutions are likely to be low, given the fact that, by international standards, they are underfunded (table A4-1). While reforms have taken place, many vocational, professional, and technical insti- tutions continue to operate along pretransition supply-driven lines, teaching narrowly specialized skills that do not meet emerging market needs. The need for reform of the vocational education system (VET) in Russia is probably greater than the need for reform of either secondary or higher education. The inheritance of a supply-driven, tightly controlled, micro- managed system designed to �t into a planned economy has proven very dif- �cult to reshape to �t Russia’s current needs, not least because of stakehold- ers’ resistance to change. With the demise of the majority of state-owned enterprises (SOEs) and of the traditional settings in which vocational educa- tion operated in the past, the gap between labor market trends and the qual- ifications and training provided by vocational schools has widened. The growing mismatch has occurred at the very time that rapid technological development and global competition require a more flexible, learning-ready, and skilled workforce. Key aspects of the system that need to be reformed include governance, because a large number of different agencies oversee the VET system; rigid professional standards, which slow the adoption of a competency-based qual- ification system; lack of emphasis on core transferable skills; inadequate funding with which to �nance operations and upgrade VET infrastructure and instructor skills; and consolidation of the fragmented VET system.2 Restructuring and Human Capital Accumulation From 1989 to 2002, the proportion of persons with a university-level educa- tion (complete and incomplete) increased by 6 percentage points while the proportion with a tertiary-level (SSUV) professional and technical education rose 8 percentage points (table A4-3). The share of persons with primary, vocational, and general secondary education remained unchanged. The share of those with a lower secondary education decreased by 3.5 percentage points, while the share of those with a primary education fell by 4 percentage points and that of those with less than a primary education fell by 5.5 per- centage points. Those shifts are even more pronounced if just the employed workforce is considered. The 2002 general census suggests that almost 60 per- cent of workers had some tertiary education, while the share of low-educated workers (those with less than lower secondary schooling) had fallen below 7 percent. Appendix 4 147 Table A4-3. Schooling Completion Rates, 1989 and 2002 Percent Total population age Employed population age 15 years and older 15 years and older Highest level of schooling attained 1989 2002 1989 2002 Higher complete 11.3 16.2 14.6 23.3 Higher incomplete 1.7 3.1 1.3 3.0 Tertiary (SSUV) 19.2 27.5 24.3 35.7 Secondary vocational 13.0 12.8 17.8 15.3 Upper secondary general 17.9 17.7 20.8 16.2 Lower secondary general 17.5 13.9 13.5 5.6 Primary 12.9 7.8 6.7 0.9 Preprimary 6.5 1.0 1.1 0.1 Total 100 100 100 100 Source: Rosstat (2006). How much of the increase in educational attainment of the workforce was the result of changes in the industrial and occupational composition of employment that accompanied restructuring, and how much was due to upgrading of education within industries and occupations? A decomposition of the effects of industrial and occupational changes, done separately for 1992–96 and for 1997–2002, suggests the following results:3 —In both periods, the largest contribution to the rising educational attainment of the workforce came from educational upgrading within indus- tries and occupations rather than from the reallocation of educated workers across industries and occupations. —In the initial 1992–96 period, about a quarter of all improvements in educational attainment were associated with shifts between industries, with inter-industry shifts having a slightly positive effect on demand for workers with higher education and stronger effects for those with secondary educa- tion and tertiary-level professional and technical training. Inter-occupation shifts contributed 18.5 percent and favored the least-educated group of workers. —In the more recent 1997–2002 period, virtually all rising educational attainment came from educational upgrading within industries and occupa- tions. The contribution of shifts across industries decreased to 2.6 percent, while that of inter-occupational shifts nearly halved, to 10.6 percent. The decomposition highlights the fact that while changes in the struc- ture of industry and occupations contributed modestly to the educational upgrading of the workforce in the early 1990s, most of the subsequent educational upgrading was independent of restructuring. That the subse- 148 Appendix 4 Table A4-4. Returns to Schooling in Transition Countries Percent Reform Pre-reform Early reform Late reform Country starting point period period period China 1979 0.015 0.025 0.061 Czech Republic 1991 0.039 0.070 0.083 Estonia 1992 0.025 0.076 n.a. Hungary 1990 0.067 0.074 0.098 Poland 1990 0.046 0.067 0.072 Romania 1992 n.a. 0.046 0.056 Russia 1992 0.039 0.075 0.092 Slovak Republic 1991 0.038 0.061 0.097 Slovenia 1991 0.043 0.063 0.070 Ukraine 1992 0.040 n.a. 0.055 Source: Fleisher, Sabirianova Peter, and Wang (2004). quent upgrading took place within all industries and occupations suggests the presence of a strong skill-biased change process in technological change and in the transformation of organizational and institutional arrangements in the workplace. The demand for education is likely to increase in such an environment, given the comparative advantage that educated workers have in implementing new technology or more generally in responding to dis- equilibria.4 Returns to Education The rising returns to education in Russia help explain why demand for edu- cation was so strong over the transition period. Estimates of the rates of return, based on Mincer-type wage equations, suggest that private returns to an extra year of schooling prior to the transition were 2 to 3 percent, reflect- ing wage compression resulting from the government-set “wage grid� system. The demise of centralized wage setting led to a rapid increase in the educa- tion premium—returns to an extra year of education rose to about 7 to 8 per- cent in the �rst �ve years of transition and by an additional 2 to 3 percent in the later period, stabilizing at 8 to 10 percent by 2000–02. Similar patterns of post-reform rising returns to education can be observed in other former socialist countries. Table A4-4 reports the returns to education estimated by Fleisher, Sabirianova Peter, and Wang (2004) for sev- eral transition countries, for three separate periods: pre-reform, early reform, and late reform. Rates of return to education almost doubled between the pre-reform and late reform period for many CIS countries, while those in Russia more than doubled. For China, increases in the returns to schooling Appendix 4 149 were even more dramatic; it had much lower pre-reform rates of return, about 1.5 percent, and by the late reform period returns to schooling had quadrupled. The phenomenon of rising returns to schooling is not unique to transition economies. Rates of return to schooling have risen in many countries—in Brazil over 1982–98 and in India over 1993-2004.5 In these countries, as in the transition economies, economic change brought about by opening up economies to global trade or moving from a centrally planned to a market economy created strong demand for (and rising returns to) more educated and skilled workers. Ranking of Investment Climate Constraints Respondents to the Russian Competitiveness and Investment Climate Assess- ment Surveys (which include the LME Survey and the SE Survey) ranked “Lack of skilled and quali�ed workforce� as the number 2 investment climate constraint on enterprise growth and development (the number 1 constraint was taxation). Small enterprises with fewer than 100 employees (the SE sam- ple) also ranked the skills constraint as major or severe, though not as signif- icant as regulation or access and cost of �nance (�gure A4-2).6 Characteristics of Firms by Staf�ng Level Table A4-5 reports the distribution of staf�ng levels for the Russia ICS sam- ple according to several firm characteristics. The probability and level of understaf�ng were highest for �rms operating in the textile industry. In that sector, more than 50 percent of respondents reported suboptimal staf�ng, with the staf�ng gap averaging 22.6 percent relative to desired levels. New firms established in or after 1992, small enterprises with fewer than 250 employees, �rms operating in the metallurgy and machine-building sectors, and government-controlled �rms (with more than 25 percent public owner- ship) also were more likely to report understaf�ng. Overstaf�ng was more prevalent among large �rms (those with more than 1,000 employees) and �rms in the chemical sector. Staf�ng levels also are related to how �rms rank their competitiveness. Firms rating themselves as having a medium to high level of competitiveness are more likely to have an optimal staf�ng level (60–61 percent) and less likely to report either understaf�ng (25–27 percent) or overstaf�ng (12–15 per- cent). On the other hand, �rms that classi�ed themselves as having “low� competitiveness are less likely to have an optimal staf�ng level (48 percent) and more likely to be understaffed (35 percent) or overstaffed (17 percent). 150 Appendix 4 Figure A4-2. Enterprises Ranking Investment Climate Constraints as Major or Severe Medium and large firms Tax level Lack of qualified labor force State regulation uncertainty Tax administration Macroeconomic instability Unfair competition Access to finance (collateral) Corruption Access and cost of finance Judiciary and law enforcement Obtaining licenses Getting construction permits Land access and registration Safety Customs procedures Power Transport Labor regulations Communication 0 10 20 30 40 50 60 Small firms State regulation uncertainty Tax level Access and cost of finance Lack of qualified labor force Macroeconomic instability Tax administration Corruption Unfair competition Procedures renting space Land access and registration Getting construction permits Obtaining licenses Judiciary and law enforcement Safety Power Customs procedures Communication Transport Labor regulations 0 10 20 30 40 50 60 Source: LME Survey (2005) and SE Survey (2005). Appendix 4 151 Table A4-5. Characteristics of Firms, by Staf�ng Level Optimal staf�ng Understaffed Overstaffed Percent Percent Percent of Percent Percent of Firm characteristics of �rms of �rms understaf�ng of �rms overstaf�ng Industry Metallurgy 56.5 29.4 19.5 14.1 13.5 Chemicals 52.4 25 8.9 22.6 13.7 Machinery 57.1 29.6 17.9 13.3 15.3 Wood processing 60.7 25 11.6 14.3 11 Textiles 41.9 50.5 22.6 7.5 12.3 Food 74.3 15.5 13.2 10.2 16.8 Firm size Less than 250 62.9 29 22 8.1 13.6 251–500 58.4 28.2 15.2 13.3 16.8 501–1,000 61.4 22.8 8.6 15.8 15 More than 1,000 51.1 25.2 11.7 23.8 13.2 Exporter No 62.2 27.6 20 10.2 14.4 Yes 56.9 26.9 13.4 16.2 14.7 R&D spending No 63.6 26.8 16.9 9.6 14.4 Yes 56.3 27.7 17.1 16 14.7 New �rm (after 1992) No 60 26.5 16.5 13.6 14.6 Yes 59.1 30.2 18.6 10.7 14.8 Foreign ownership No 61.8 26.8 16.9 11.5 14.3 Yes 53.3 29 17.3 17.8 15.3 Government control No 61 26.7 14.9 12.3 14.5 Yes 56.7 28.7 21.8 14.5 14.8 Competitiveness High 60.6 24.8 11.6 14.7 15.6 Medium 61.2 26.9 15.6 11.9 14.7 Low 47.9 35 25 17.1 14.7 Source: LME Survey (2005). Whatever the factors that constrain understaffed �rms from employing the personnel that they need or overstaffed �rms from discharging redundant staff, nonoptimal staf�ng levels can adversely affect �rms’ perceptions of their level of competitiveness. Major Problems by Unit within the Firm The extent to which skill shortages are a problem varies across units within �rms. Figure A4-3 graphs the percent of �rms that ranked several key issues as being a major problem by unit within the �rm—operations (or produc- 152 Appendix 4 Figure A4-3. Major Problems, by Unit within the Firm Percent of firms 70 Operating 60 Economic R&D 50 Human resources 40 30 20 10 0 Lack of Lack of skills Lack of Lack of Other No problem technological and autonomy interaction capacity qualifications with other units Source: Authors’ calculations. tion), economics (marketing, strategy), research and development (R&D), and human resources (HR). Most �rms identi�ed two major problems—lack of technological capacity and lack of skilled and quali�ed workers—both of which were concentrated in operations, that is, on production lines. A much smaller fraction of �rms reported those problems as major constraints in the economics, R&D, or HR units. Employment Protection Legislation Table A4-6 reports estimates from an ordered probit model in which the index of dif�culty of searching for and hiring skilled labor (based on a scale of 1 to 5, with 5 indicating maximum dif�culty) is regressed on an index of employment protection legislation (EPL) as a constraint (determined by the sum of different EPL components that respondents indicate as problematic), wage levels in the firm (to test the Gimpelson hypothesis), and different enterprise characteristics that might shape demand for skills.7 In-Service Training Simple tabulations suggest that about 70 percent of manufacturing enterprises in the LME sample provide employees with in-service training (see �gure A4-4). Appendix 4 153 Table A4-6. Employment Protection Legislation as a Constraint on Hiring Skilled Labora Dif�culty in searching for and hiring labor Skilled workers Dependent variable Professionals (1) (2) EPL index .108 .094 ... (3.03)*** (2.66)*** ... EPL is not a constraint ... ... −.306 ... ... (−3.94)*** Log (wages) −.093 −.199 −.197 (−1.69)* (−3.66)*** (−3.63)*** Government controlled .218 .021 .023 (2.06)** (0.20) (0.21) Foreign owned −.153 −.036 −.042 (−1.33) (−0.32) (−0.37) Small �rm (< 250 employees) −.301 −.166 −.162 (−3.84)*** (−2.17)** (−2.11)** R&D spending indicator .208 .044 .033 (2.64)*** (0.56) (0.42) New �rm (after 1992) .078 −.115 −.122 (0.83) (−1.25) (−1.32) Number of observations 896 898 898 Likelihood ratio (chi 2) 123.85 181.13 189.56 Source: LME Survey (2005). a. Industry and regional control variables included; Z values in parentheses. EPL index is the sum of rankings for EPL-related dif�culties. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. That �gure is not comparable with �gures for other countries because the LME Survey, by design, focuses on medium and large �rms with more than 100 employees and larger �rms on average tend to train more than small ones. The BEEPS and the SE Survey, in contrast, include smaller �rms. To compare the incidence of training in Russia to that of OECD and selected comparator devel- oping countries, we adopted a common weighting scheme based (arbitrarily) on the size distribution of �rms in the India ICA survey. The size distribution of micro (15 or fewer workers), small (16–100 workers), medium (101–250 workers), and large �rms (more than 250 workers) in India was 40, 44, 7, and 8 percent respectively; the corresponding size distribution for the pooled LME and SE surveys in Russia was 12, 16, 29, and 43 percent respectively. When the weighted figures are used, Russia stands out among the BRIC group with respect to the very small share of its workforce trained within the �rm (7.7 per- cent of skilled and 1.4 percent of unskilled workers) compared with that of Brazil (53 percent and 45 percent) and China (44 and 28 percent) (table A4-7). 154 Appendix 4 Figure A4-4. Incidence of Formal In-Service Training, OECD, Regional, and Country Means Training by region Percent 70 60 50 40 30 20 10 0 South OECD Europe and Latin East Asia Asia Central Asia America and (including Russia) Caribbean Training by country China Ireland Chile Spain Brazil Russia Lithuania South Korea Thailand Germany Malaysia Greece Portugal Serbia Bulgaria Sri Lanka India 0 20 40 60 80 100 Source: BEEPS (2006); World Bank (2002–2006); LME Survey (2005); and SE Survey (2005). Appendix 4 155 Table A4-7. Percent of Workforce Receiving In-Service Training in Russia and Comparatorsa Percent of workforce trained Country Skilled Unskilled BRIC Russia 7.7 1.4 Brazil 52.8 45.4 China 44.5 28.5 Transition Bulgaria 24.0 13.0 Lithuania 11.9 4.0 Serbia 45.2 6.1 OECD Germany 37.3 27.2 Greece 53.6 36.8 Ireland 76.0 68.1 Portugal 75.2 50.2 Spain 76.1 56.2 South Korea 65.4 59.2 Developing Chile 48.3 36.0 Malaysia 69.6 52.7 Sri Lanka 47.4 34.1 Source: Authors’ calculations, based on BEEPS (2006) and World Bank (2002–2006). a. Country samples are restricted to manufacturing �rms, and shares of workers trained are weighted esti- mates using the �rm size distribution of India. WBES = World Bank Enterprise Survey for year indicated. Training Determinants The importance of the training correlates can be investigated within a regres- sion framework by using a probit model.8 The model estimates the probabil- ity of in-service training by regressing the “any formal training� variable on a set of explanatory variables, including measures of firm size, the share of workers with higher education, and other �rm attributes such as export ori- entation, R&D spending, foreign or government ownership, and independent assessments of the region’s investment climate risk (table A4-8). A corre- sponding set of regressions were estimated separately for the probability of in-house training and external training. Training Impact on Firm-Level Productivity The dependent variable—the logarithm of value added—was regressed on the logarithms of capital (book value of physical plant and equipment assets), alternative measure of training (any formal training, in-house or external 156 Appendix 4 Table A4-8. Determinants of In-Service Training and Training by Source Probit model speci�cation Any formal In-house External Dependent variable: training (1,0) training training training Firm size dummies Small size (101–250 workers) 0.084 0.108 0.102 (1.42) (1.41) (1.56) Medium size (251–1,000 workers) 0.219 0.22 0.257 (3.76)*** (2.90)*** (3.96)*** Large size (> 1,000) 0.273 0.381 0.334 (4.86)*** (4.58)*** (5.24)*** Percent higher-educated workers 0.003 0.000 0.003 (2.45)** (0.03) (2.37)** New �rm (after 1992) −0.034 −0.002 −0.068 (−0.93) (−0.05) (−1.72)* Exporter 0.055 0.065 0.063 (1.71) (1.82)* (1.81)* Positive R&D spending 0.081 0.074 0.06 (2.29)** (1.91)* (1.59) R&D sales ration 1.071 0.549 0.283 (0.99) (0.66) (0.32) Some foreign ownership (> 10 percent) −0.05 0.037 −0.079 (−1.06) (0.74) (−1.57) Government control (> 25 percent) 0.059 −0.015 0.115 (1.40) (−0.33) (2.58)** Moderate IC risk 0.097 0.106 0.058 (2.70)** (2.59)** (1.48) High IC risk 0.012 0.039 −0.009 (0.34) (0.95) (−0.22) Missing values Yes Yes Yes Observations 990 981 986 Source: Pooled LME Survey and SE Survey. *Signi�cant at 10 percent; **signi�cant at 5 percent; ***signi�- cant at 1 percent. training, and combinations of training sources), and a vector of control vari- ables for worker attributes (mean years of education) and for location in moderate– or high–investment risk regions (see table A4-9). The production functions, estimated by ordinary least squares, implicitly treat the different training variables as being exogenously determined. That assumption may be suspect if the �rms that train are also more productive and systematically self-select themselves into the training group on the basis of unobserved pro- ductivity traits, potentially biasing production function estimates. Qualita- tively similar productivity (and wage) results obtain when self-selection into training is taken into account (see the discussion of human capital at the beginning of chapter 4). Appendix 4 157 Table A4-9. In-Service Training and Productivitya Model speci�cations Dependent variable: Log (valued added) (1) (2) (3) Log (capital) 0.197 0.196 0.196 (6.77)*** (6.68)*** (6.58)*** Log (labor) 0.889 0.876 0.877 (15.57)*** (15.45)*** (15.28)*** Mean years of education 0.055 0.056 0.057 (2.69)** (2.74)** (2.72)** Moderate regional IC risk −0.277 −0.277 −0.272 (−3.08)*** (−3.01)*** (−3.00)*** High regional IC risk −0.332 −0.330 −0.333 (−4.41)*** (−4.39)*** (−4.42)*** Any formal training 0.225 (3.48)*** In-house training 0.092 (1.19) External training 0.22 (3.27)*** Only in-house training −0.003 (−0.03) Only external training 0.168 (1.99)* Both in-house and external training 0.281 (4.06)*** Constant 8.64 8.695 8.708 (30.41)*** (30.27)*** (30.21)*** Missing values Yes Yes Yes Regional cluster Yes Yes Yes Observations 784 784 784 R2 0.64 0.64 0.64 Source: Authors’ calculations. a. Value of t statistics in parentheses. *Significant at 10 percent; **significant at 5 percent; ***significant at 1 percent. Training Impact on Worker Wages A wage model was estimated both at the �rm level, using the logarithm of mean monthly wage and training of the �rm, and for the pooled sample of occupations within each �rm to exploit the availability of occupation-speci�c information on wages (see table A4-10). The logarithm of monthly per- worker �rm-level or occupation-speci�c wages was regressed on the training variables, a vector of �rm attributes, and average years of education of the workforce. For the occupational wage model, data on up to �ve occupational groups per �rm were pooled and indicator variables included for managers, professionals, skilled workers, and unskilled workers (the omitted category 158 Appendix 4 Table A4-10. Cross-Sectional Wage Models with Traininga Firm-level wage model Occupation wage model Dependent variable: log (monthly wage) (1) (2) (1) (2) Constant 10.224 10.205 7.993 7.903 (46.85)*** (47.58)*** (48.34)*** (49.81)*** Any formal training 0.16 0.091 (3.29)*** (1.58) In-house training 0.044 0.160 (0.75) (3.02)*** External training 0.178 0.096 (4.21)*** (1.48) Small enterprise (< 250) −0.162 −0.147 −0.199 −0.177 (−3.18)*** (−2.90)*** (−3.60)*** (−3.20)*** Some foreign ownership −0.109 −0.101 0.084 0.087 −(1.29) −(1.21) (1.17) (1.23) Government control 0.059 0.047 −0.06 −0.059 (0.72) (0.58) (−0.84) (−0.81) Exporter 0.102 0.095 0.06 0.044 (2.08)** (1.84)* (1.27) (0.94) Positive R&D spending 0.054 0.052 0.178 0.168 (0.94) (0.91) (3.07)*** (2.93)*** R&D sales ratio 1.728 1.765 0.069 0.074 (3.10)*** (3.37)*** (1.03) (1.09) New �rm (after 1992) 0.018 0.021 −0.199 −0.177 (0.28) (0.33) (−3.60)*** (−3.20)*** Mean years of education 0.061 0.061 n.a. n.a. (3.82)*** (3.87)*** Occupation dummies No No Yes Yes Observations 923 923 3,026 3,026 R2 0.08 0.09 0.241 0.253 Source: Authors’ calculations. a. Value of t statistics in parentheses. All regressions include control variables for missing values and for regions. *Signi�cant at 10 percent, **signi�cant at 5 percent; ***signi�cant at 1 percent. being “other white-collar employees�) in place of mean years of education, with which occupations are closely correlated. The pooled sample consisted of 3,026 occupations from the 923 �rms, and the regression model accounts for the common error structure for all occupations in the same �rm. Bivariate Probit Analysis of Training and Innovating A bivariate probit model was jointly estimated for the two decision vari- ables—whether or not to innovate and to train. Each equation had some explanatory variables in common but also others that were assumed to affect one decision but not the other.9 The bivariate regression results for the inno- vation and training equations are reported in table A4-11. Many of the train- ing results have been reported previously and will not be elaborated on fur- Appendix 4 159 Table A4-11. Bivariate Probit Regressions of Innovation and Training Dependent variables: innovation and training Coef�cient Z score Innovation equation Firm size (101–250) 0.184 0.97 Firm size (251–1,000) 0.492 2.56 Firm size (> 1,000) 1.031 4.65 New equipment dummya 0.264 3.04 Percent workforce with higher education 0.008 2.39 New �rm dummy (established after 1992) −0.019 −0.18 Exporter 0.361 4.05 Foreign ownership ( ≥ 10 percent) 0.111 0.88 Government control dummy (control ≥ 25 percent) −0.009 −0.08 Received any government support for R&D 0.047 0.37 Constant −0.939 −4.82 Training equation SME indicator (< 250 workers) −0.521 −5.73 Percent workforce with higher education 0.010 2.84 New �rm (established after 1992) −0.115 −1.09 Exporter 0.202 2.17 Foreign ownership ( ≥ 10 percent) −0.194 −1.45 Government control dummy (control ≥ 25 percent) 0.187 1.48 Dif�culty hiring skilled and professional workersb 0.108 1.16 Firm overstaffed indicator 0.015 0.15 Firm understaffed indicator 0.460 3.05 Constant 0.403 3.49 Observations 979 rho 0.1125 Wald test − chi 2 (24) 170.2 Log pseudo-likelihood = −1174.4488 Prob > chi 2 0 Source: LME Survey (2005). a. New equipment = 1 if less than 50 percent of equipment is fully depreciated. b. Dif�culty of hiring skilled and professional workers = 1 if �rm gave dif�culty for either skill group a 4 or 5 on a scale of 1 to 5 (5 being the most dif�cult). ther. The innovation equation, however, is new. The results suggest that inno- vating �rms tend to be larger, to export, to use relatively new machinery and equipment, and to employ a more educated workforce. More pertinent is the estimate of rho, which measures the covariance in the errors of the innova- tion and training equations. Both Wald and likelihood ratio tests reject the null hypothesis that rho equals 0; that is, they con�rm that the innovation and training decisions were made jointly. The bivariate probit model also yields estimates of the probabilities that firms choose one investment activity but not the other, both activities together, or neither. To simplify description, let Pr (ij) be the joint probabil- ity of innovation i and training j. For the LME sample as a whole, the least 160 Appendix 4 Table A4-12. Predicted Joint Probabilities of Innovation and Traininga Innovate (0) Innovate (0) Innovate (1) Innovate (1) Firm characteristics Train (0) Train (1) Train (0) Train (1) Total sample 0.19 0.34 0.11 0.36 Small enterprise (< 250) No 0.11 0.33 0.10 0.46 Yes 0.30 0.36 0.13 0.22 Exporter No 0.25 0.38 0.11 0.26 Yes 0.11 0.30 0.11 0.47 Staf�ng Optimal 0.20 0.33 0.12 0.34 Understaffed 0.20 0.35 0.11 0.34 Overstaffed 0.09 0.37 0.06 0.47 Source: LME Survey (2005). a. Predicted probabilities from bivariate probit model; see table 4A-11. likely probabilities are Pr (10)—firms innovate but do not train (10 per- cent)—and Pr (00), �rms engage in neither activity (11 percent). It is much more common for �rms to train but not innovate, Pr (01) of 36 percent, or invest in both innovation and training at the same time, Pr (11) of 22 percent. The model also yields estimates of the probability of one investment activity conditional on the other taking place. Denote these as Pr (i ⎟ i). ⎟ j) and Pr (j The conditional probability of innovating given training is not high, Pr (i ⎟ j) of 43 percent, suggesting that �rms have many reasons for training besides supporting innovation. In contrast, the conditional probability of training given innovation is much higher, Pr (j⎟ i) of 73 percent, supporting the main- tained hypothesis that skills and training are needed to complement invest- ments in the innovative activities of the �rm. Table A4-12 reports the predicted joint probabilities of innovation and training disaggregated by several firm attributes. The tabulations reported here are restricted to firm characteristics that are able to discriminate among the different predicted joint probabilities. First, they suggest that larger firms are more likely to invest in both innovation and training (46 percent) or training alone (33 percent) than small firms, which are more likely just to train (36 percent) than to invest in both innovation and training (22 percent). Similarly, exporting firms are more likely to both innovate and train (47 percent) than to invest in training alone (30 per- cent), while nonexporters are more likely just to train (38 percent). Finally, overstaffed firms are more likely to invest in both innovation and training Appendix 4 161 Table A4-13. Production Functions and Wage Models Estimated with Exogenous or Predicted Innovation and Traininga Log (value added) Log (wages) Dependent variables (1) (2) (1) (2) Predicted joint probabilities Innovate (0); train (1) −0.186 1.437 (−0.23) (2.31)* Innovate (1); train (0) 0.712 2.490 (0.50) (2.66)** Innovate (1); train (1) 1.199 1.889 (2.66)** (4.32)** Exogenous innovation and training Innovate (0); train (1) 0.242 0.153 (2.60)** (2.63)** Innovate (1); train (0) 0.139 0.100 (1.13) (1.50) Innovate (1); train (1) 0.303 0.165 (3.12)*** (2.66)** Mean years of workforce education 0.036 0.05 0.038 0.054 (1.82)* (2.55)** (2.79)*** (3.39)*** Observations 775 784 911 923 R2 0.64 0.63 0.13 0.12 Source: LME Survey (2005). a. Production functions include logarithms of capital, labor, and controls for missing values and region. Wage equations include �rm size, ownership, new �rm, and controls for missing values, region, and industry. *Signi�- cant at 10 percent; **signi�cant at 5 percent; ***signi�cant at 1 percent. (47 percent) or in training alone (37 percent) than are optimally staffed or understaffed firms, which are equally likely to do both or to invest in train- ing alone (about 33–35 percent). Table A4-13 reports estimates of production functions and wage equa- tions that include predicted joint probabilities of innovation and training, with the omitted category being the prediction of investing in neither innovation nor training. By construction, these predicted joint probabili- ties are uncorrelated with the error terms of the model, yielding unbiased (but inefficient) estimates of the innovation and training variables. For comparison, the same models are estimated using indicator variables of innovation and training, both treated as being exogenously determined outside the model, to characterize �rms as just innovating, just training, or doing both. The production function results indicate that only joint invest- ments in innovation and training improve �rm-level productivity. That may be contrasted with the alternative “exogenous� model, wherein both “just training� and “both training and innovating� are associated with productiv- ity increases, but not “just innovating.� The wage regression using predicted 162 Appendix 4 values suggests that all three states—just training, just innovating, and invest- ing in both activities—are associated with positive wage gains. Curiously, just innovating but not training has the largest coef�cient. The alternative “exoge- nous� wage model also yields different results, namely that both “just train- ing� and “investing in both� show wage gains, but “just innovating� does not. Notes 1. Fretwell and Wheeler (2001). 2. Canning and others (2004). 3. A shift-share approach is used to decompose changes over time in educational attainment attributable to different components: one that measures the results of shifts in the industry and occupational composition of employment, holding educa- tion constant; another that measures the contribution of rising education, holding industry and occupation mix constant; and a third interaction term. The 1992–96 decomposition uses six education, �fty occupation, and �fteen industry groups, while the 1997–2002 decomposition relies on seven education, thirty-two occupation, and nineteen industrial groups. 4. Schultz (1975); Bartel and Lichtenberg (1987); Tan (2005). 5. Blom, Holm-Nielsen, and Verner (2001); Riboud, Tan, and Savchenko (2006). 6. In addition to ranking each constraint on a scale of 1 to 5, with 5 being a severe constraint, enterprises in the LME and SE surveys also were asked to identify the most severe constraint on the previous list. That alternative ranking yielded broadly simi- lar �ndings, with lack of a quali�ed workforce being ranked number 3 by medium and large enterprises and number 2 by small enterprises. 7. Gimpelson (2004) hypothesized that understaf�ng may be the result of low- ef�ciency �rms being unable to pay competitive wages. 8. The advantage of regression analysis over tabular information is that the inde- pendent effects of each variable (or set of variables) can be analyzed holding constant the effects of other hypothesized correlates. 9. Each equation must be identi�ed by having one or more instruments that affect (or are highly correlated with) that choice variable but not the other. For the innova- tion equation, the instruments were indicators for having newer equipment and hav- ing received government support for R&D. For the training equation, the correspon- ding instruments were an index of dif�culty of searching for and hiring professionals and skilled workers and whether the enterprise was understaffed or overstaffed. Appendix 5 Survey data show that �rms facing more intense competitive pressures from domestic producers, from foreign producers operating domestically, or from imported goods also tend to be more innovative. Table A5-1 examines the effects of such competitive pressures, �rst on the sum total of innovation activities (from introducing new products and technologies to initiating training programs) and second on the likelihood that �rms introduce funda- mentally new product lines. Firms that face greater competitive pressures are more likely to do both. On the other hand, �rms that enjoy a monopolist position or other forms of protection not only innovate less than �rms fac- ing greater competitive pressures, but they also contribute to an unpre- dictable and uncertain regulatory environment. Monopolistic �rms tend to obtain more favorable treatment from government authorities: they tend to suffer less from any given investment climate constraint than other �rms, pay less in bribes to secure government contracts, suffer less from nonpayment by customers, and spend less on protection payments. It is likely that their dif- ferential treatment is due to loopholes, exemptions, or special exceptions in existing legislation as well as to the discretionary interpretation and applica- tion of regulations by public authorities. There is evidence, moreover, that firms facing the greatest competitive pressures (from domestic and foreign manufacturers) also face the severest investment climate impediments, yet they are the most innovative compa- nies in the Russian economy. Table A5-2 shows the relationship between the competitive pressures that firms face and the severity of various investment climate constraints. The constraints were generated by using principal component weightings of the individual impediments, generating nor- mally distributed, continuous variables with mean 0. The results also con- trol for �rm size, whether or not the �rm is a member of a holding company or financial-industrial group, and a firm-specific, systemic “optimism� or favorable bias toward the investment climate. Sectoral dummies also are included. The within-region effects shown in table A5-2 indicate that when systemic biases toward the investment climate in general were controlled for, firms facing competitive pressures were more likely to face harsher gov- ernance constraints (poorly functioning legal system, policy uncertainty, crime, and unfair competition), less access to and higher costs of finance, higher taxes and weak tax administration, more problems with labor (skills and regulations), and steeper administrative barriers (customs and licens- ing barriers). Table A5-3 outlines the impact of competitive pressures on the costs of doing business. 163 164 Appendix 5 Table A5-1. Competitive Pressures and Innovationa Independent variable Innovative activities Introduction of new product Competitive pressures 0.0206*** 0.0669*** (0.0072) (0.0255) External consultant 0.2513*** 0.4342** (0.0475) (0.1896) FIG dummy −0.0681* −0.1076 (0.0410) (0.1463) Log (size) 0.1962*** 0.2867*** (0.0204) (0.0755) Newly private 0.0246 0.2528 (0.0572) (0.2048) Age −0.0020*** 0.0002 (0.0006) (0.0021) Major city 0.0963** 0.1628 (0.0456) (0.1647) Observations 988 985 Number of regions 46 45 Log likelihood −1930.97 −559.79 chi 2 58.52 58.52 Probability (0.0000) (0.0000) Source: LME Survey (2005). a. The dependent variable in model 1 estimates the number of innovative activities in which a �rm is engaged. Estimation is by conditional Poisson with regional �xed effects. The dependent variable in model 2 is coded 1 if a firm introduced a new product line during 2003–04, and estimation is by conditional logit regression with regional �xed effects. Competitive pressures is an index based on �rms’ scoring the impact of competition with 1 (other domestic manufacturers), 2 (imported products), or 3 (foreign producers operating in Russia). FIG dummy is coded 1 if the firm is a group-member of a holding company, 0 otherwise. Size is number of perma- nent employees in 2004. Newly private is coded 1 if the �rm began operations as a private �rm after 1992, 0 oth- erwise. Age is the number of years a �rm has been in continuous operation. Major city is coded 1 if the �rm is located in a major urban area. ***p < 0.01; **p < 0.05; *p < 0.10. Table A5-2. Competitive Pressures and Investment Climate Constraintsa Administrative Independent variable Governance Finance Land Taxation Labor Infrastructure barriers Log (size) 0.0001 0.0799* −0.0611 −0.0930** 0.0749* −0.0094 0.1451*** (0.0577) (0.0459) (0.0454) (0.0427) (0.0406) (0.0468) (0.0403) Competitive pressures 0.0701*** 0.0419*** 0.0207 0.0555*** 0.0553*** 0.0158 0.0420*** (0.0194) (0.0152) (0.0152) (0.0143) (0.0136) (0.0157) (0.0136) FIG −0.2522** −0.0106 −0.1537* −0.1892** 0.2390*** 0.1311 −0.0794 (0.1111) (0.0872) (0.0873) (0.0821) (0.0782) (0.0901) (0.0775) Bias −1.6816*** −0.1315 −0.8513* −1.4076*** −1.1044** −0.8737* −1.2183*** (0.6356) (0.5007) (0.4997) (0.4703) (0.4477) (0.5158) (0.4445) Observations 985 961 982 987 988 987 985 Regions 49 49 49 49 49 49 49 R2 0.0399 0.0207 0.0276 0.0540 0.0552 0.0257 0.0660 Source: LME Survey (2005). a. Estimation is by OLS regression using within-region �xed effects. Sectoral dummies and intercepts are included in regressions but are not reported. Standard errors are in parentheses. Dependent variables are principal components groupings of various investment climate constraints. Size is the number of permanent employees in 2004. Competitive pressures is an index based on �rms’ scoring the “impact of competition� with 1 (other domestic manufacturers), 2 (imported products), or 3 (foreign producers operating in Russia). FIG dummy is coded 1 if the �rm is a group-member of a holding company, 0 otherwise. Bias is a valence variable intended to serve as a proxy for �rm-speci�c biases toward the invest- ment climate in general. Higher levels of “bias� mean that the �rm is more favorably disposed (less likely to complain about) investment climate constraints. ***p < 0.01; **p < 0.05; *p < 0.10. 166 Appendix 5 Table A5-3. Competitive Pressures and Business Costsa Bribes for Excess Sales on government employment credit Independent contracts (percent (percent variable (percent value) of staff) of sales) Nonpayment Competitive pressures 0.1397** 0.5963* 1.0108** 0.1524*** (0.0571) (0.3376) (0.4920) (0.0292) Former state-owned 0.2873 3.1864 −5.3956 −0.1708 enterprise (0.3782) (2.2147) (3.7059) (0.1900) Age 0.0037 −0.0942 −0.0260 0.0158** (0.0125) (0.0726) (0.1204) (0.0076) Medium 0.3280 −7.9645*** 2.5916 0.5545*** (0.3606) (2.1213) (3.1119) (0.1821) Large −0.1707 −10.2951*** 0.7585 0.3727 (0.5021) (2.9171) (4.4591) (0.2513) Observations 875 955 524 972 R2 0.0452 0.0478 0.0348 0.0691 Source: LME Survey (2005). a. Estimations for the �rst three columns are by OLS regression and for the fourth column by logistic regres- sion (pseudo R 2 is reported). Dummies for city type (village, town, minor city, major city) and intercepts are included in regressions but are not reported. Standard errors are in parentheses. ***p < 0.01; **p < 0.05; *p < 0.10. Table A5-4. Business Association Membership and Innovative Activitiesa Introduced Obtained Independent Introduced new Outsourced quality Agreed variable new product technology production certi�cation to new JV Business association 0.7161*** 0.5212** 1.0996*** 1.4806*** 1.8093*** member (0.2345) (0.2374) (0.3698) (0.3332) (0.6122) Former state-owned −0.1720 −0.0700 −0.3738 −0.2077 0.4645 enterprise (0.2836) (0.2752) (0.4877) (0.4348) (0.9836) Age −0.0018 0.0016 −0.0028 0.0071 0.0099 (0.0091) (0.0089) (0.0173) (0.0115) (0.0295) Medium 0.2219 0.0809 0.7382* 0.4904 0.0503 (0.2413) (0.2369) (0.3917) (0.3700) (0.7130) Large 0.6651** 0.3162 −0.7605 0.4910 0.3921 (0.3372) (0.3394) (0.7447) (0.5080) (0.8295) Percent foreign 0.0001 0.0037 −0.0056 −0.0004 −0.0217*** owned (0.0038) (0.0037) (0.0057) (0.0061) (0.0079) Percent state 0.0252 0.0195 0.0433 0.0354 −0.0315 owned (0.0289) (0.0276) (0.0279) (0.0277) (0.0723) Observations 539 539 539 539 472 R2 0.0347 0.0246 0.0910 0.1112 0.1888 chi 2 24.50 18.38 25.48 38.53 23.80 Probability < chi (0.0108) (0.0731) (0.0077) (0.0001) (0.0081) Source: BEEPS (2006). a. Estimation is by logistic regression. Dummies for city type (village, town, minor city, major city) and intercepts are included in regressions but are not reported. Standard errors are in parentheses. ***p < 0.01; **p < 0.05; *p < 0.10. Appendix 5 167 Table A5-5. Business Association Membership and Investment Climatea Regulatory Bribe burden payments for Independent Excess (percent of Predictability government variable employment management time) of regulations contracts Competitive 0.9276** 0.1873 −0.1101*** 0.2223*** pressures (0.3815) (0.2248) (0.0289) (0.0406) Business 5.7940 10.3745*** −0.7855* 1.5297*** association (5.9636) (3.5304) (0.4522) (0.5720) membership Competitive −1.3731* −0.7884* 0.1074* −0.1715** pressures × (0.7918) (0.4667) (0.0589) (0.0731) business association membership Former 2.8933 −0.0989 0.1507 0.5541*** state-owned (2.2126) (1.3062) (0.1617) (0.2146) enterprise Age −0.0908 −0.0399 −0.0131** 0.0158** (0.0725) (0.0427) (0.0057) (0.0062) Medium −7.3537*** −1.1862 0.2866* 0.3478* (2.1409) (1.2552) (0.1568) (0.1876) Large −8.8897*** −1.6304 0.5503** 0.2484 (2.9730) (1.7881) (0.2156) (0.2540) Observations 955 966 943 824 R2 0.0545 0.0287 0.0124 0.0402 Source: BEEPS (2006). a. Estimations for the �rst and second columns are by OLS regression and for the third and fourth columns by ordered-logit regression (pseudo R2 is reported). Dummies for city type (village, town, minor city, major city) and intercepts are included in regressions but are not reported. Standard errors are in parentheses. ***p < 0.01; **p < 0.05; *p < 0.10. References Abramovitz, Moses. 1956. “Resources and Output Trends in the United States since 1870.� American Economic Review 46 (2): 5–23. Acemoglu, Daron, and Jorn-Steffen Pischke. 1998. “Why Do Firms Train? Theory and Evidence.� Quarterly Journal of Economics 113 (1): 79–119. Acs, Zoltan J., and David B. Audretsch. 1988. “Innovation and Firm Size in Manufacturing.� Technovation 7 (3): 197–210. Aghion, Philippe, and others. 2002. “Competition and Innovation: An Inverted U Relationship.� Working Paper 9269. Cambridge, Mass.: National Bureau of Economic Research. Aidis, Ruta, and Yuko Adachi. 2007. “Russia: Firm Entry and Survival Barriers.� Economic Systems 31 (4): 391–411. Ark, Bart van, and Marcin Piatkovski. 2004. “Productivity, Innovation, and Investment Climate Surveys in Old and New Europe.� International Economics and Economics Policy 1 (2–3): 215–46. Barro, Robert J., and Jong-Wha Lee. 2001. “International Data on Education Attainment: Updates and Implications.� Oxford Economic Papers 53 (3): 541–63. Bartel, Anne, and Frank Lichtenberg. 1987. “The Comparative Advantage of Educated Workers in Implementing New Technology.� Review of Economics and Statistics 69 (1): 1–11. Batra, Geeta, Daniel Kaufmann, and Andrew H. W. Stone. 2003. “The Firms Speak: What the World Business Environment Survey Tells Us about Constraints on Private Sector Development.� Washington: World Bank. Becker, Gary S. “Human Capital.� 2002. In The Concise Encyclopedia of Economics, edited by David R. Henderson. Indianapolis: Liberty Fund. BEEPS (Business Environment and Enterprise Performance Surveys). 2006. European Bank for Reconstruction and Development and World Bank (http:// info.worldbank.org/governance/beeps). Bell, Martin, and Keith Pavitt. 1992. “National Capacities for Technological Accumulation: Evidence and Implications for Developing Countries.� Annual Bank Conference on Development Economics (ABCDE). Washington: World Bank. Bhojwani, Hiro. 2006. “Report on the Indian Public Research and Development System.� Washington: World Bank. Blanchflower, David G., and Simon M. Burgess. 1996. “Job Creation and Job Destruction in Great Britain in the 1980s.� Industrial and Labor Relations Review 50 (1): 17–38. Blom, Andreas, Lauritz Holm-Nielsen, and Dorte Verner. 2001. “Education, Earnings, and Inequality in Brazil: 1982–1998: Implications for Education Policy.� World Bank Policy Research Working Paper 2686. Washington: World Bank. 168 References 169 Bortnik, Ivan. 2004. “Desiat’ let razvitia malogo innovatsionnogo predprinematel’stva v Rossii� [Ten Years of the Development of Innovative Small Entrepreneurship in Russia]. Moscow: Foundation for Assistance to Small Innovative Enterprises. Broadman, Harry. 2006. “From Disintegration to Integration.� Washington: World Bank. Canning, Mary, and others. 2004. The Modernization of Education in Russia. Moscow: World Bank. Carlin, Wendy, Mark Schaffer, and Paul Seabright. 2004. “A Minimum of Rivalry: Evidence from Transition Economies on the Importance of Competition for Innovation and Growth.� William Davidson Institute Working Paper 670. University of Michigan, Stephen M. Ross Business School. Carlin, Wendy, and others. 2001. “Competition and Enterprise Performance in Transition Economies: Evidence from a Cross-Country Survey.� William Davidson Institute Working Paper 376. University of Michigan, Stephen M. Ross Business School. CEFIR (Center for Economic and Financial Research at the New Economic School). 2005. Monitoring the Administrative Barriers to Small Business Development in Russia: The Fifth Round. CEFIR Policy Paper Series. Moscow: New Economic School. CEFIR and IET. 2006. “Survey of Corporate Governance in Russian Companies in 2005 and 2006.� Unpublished. Moscow. Coe, David T., Elhanan Helpman, and Alexander W. Hoffmaister. 1995. “North-South R&D Spillovers.� Economic Journal 107 (440): 134–49. CSRS (Center for Science Research and Statistics). 2005. “Science in Russia at a Glance.� In 2005 Statistical Yearbook. Moscow. Dossani, Rafiq, and Martin Kenney. 2001. “Creating an Environment: Developing Venture Capital in India.� Stanford University, Asia/Pacific Research Center (A/PARC). Enos, John L. 1962. Petroleum Progress and Pro�ts: A History of Process Innovation. MIT Press. Eurostat (Statistical Of�ce of the European Communities). Various years. Europe in Figures. Luxembourg: Office for Official Publications of the European Communities. Expert Rating Agency. 2006. Investment Rating for Russia’s Regions, 2005–2006. Moscow (www.raexpert.org/ratings/regions/2006). FIAS (Foreign Investment Advisory Service). 2004a. Administrative Barriers to Investment. Washington: World Bank. ———. 2004b. Russia’s Runaway Investors. Washington: World Bank. Fleisher, Belton, Klara Sabirianova Peter, and Xiaojun Wang. 2004. “Returns to Skills and the Speed of Reforms: Evidence from Central and Eastern Europe, China, and Russia.� William Davidson Institute Working Paper 703. University of Michigan, Stephen M. Ross Business School. 170 References Fretwell, David, and Anthony Wheeler. 2001. “Russian Secondary Education and Training.� World Bank Secondary Education Series. Washington: World Bank. Frye, Timothy. 2002. “Capture or Exchange: Business Lobbying in Russia.� Europe- Asia Studies 54 (7): 1017–1036. Gill, Indermit, Fred Fluitman, and Amit Dar. 2000. Vocational Education and Training Reform: Matching Skills to Markets and Budgets. Oxford University Press. Gimpelson, Vladimir. 2004. “Defitsit Kvalifikatsii i Navykov na Rynke Truda: Nedostatok Predlozhenia, Ogranichenie Sprosa, ili Lozhnie Signaly Roboto- datelei?� [Quali�cations and Skill De�ciency in the Labor Market: Lack of Supply, Demand Constraints, or False Signals of Employers?]. Working Paper 3/2004/01. Moscow: Higher School of Economics. Goskomstat (State Committee for Statistics, Russian Federation). 1995. Natsional’nye Scheta Rossii v 1989–1994 Godakh [National Accounts of Russia, 1989–1994]. Moscow. Griliches, Zvi. 1986. “Productivity, R&D, and Basic Research at the Firm Level in the 1970s.� Working Paper W1547. Cambridge, Mass.: National Bureau of Economic Research. ———. 1994. “R&D, Patents, and Productivity.� In NBER Conference Proceedings. University of Chicago Press. ———. 1998. R&D and Productivity: The Econometric Evidence. University of Chicago Press. Haltiwanger, John, and Helena Schweiger. 2004. “Firm Performance and the Business Climate: Where Does ICA Fit In?� University of Maryland. Hare, Paul, Mark Schaffer, and Anna Shabunina. 2004. “The Great Transformation: Russia’s Return to the World Economy.� CERT Discussion Papers 0401. Edinburgh: Heriot-Watt University, Centre for Economic Reform and Transformation. Helpman, Elhanan, and Manuel Trajtenberg. 1996. “Diffusion of General Purpose Technologies.� Working Paper 5773. Cambridge, Mass.: National Bureau of Economic Research. Hoekman, Bernard, and Beata Smarzynska Javorcik. 2006. Global Integration and Technology Transfer. Washington: Palgrave Macmillan and World Bank. HSE (Higher School of Economics). 2005. Nauka v Rossyiskoy Federatsii: Statistichesky sbornik [Science in the Russian Federation: Statistical Handbook]. Moscow. ILO (International Labor Organization). 2005. The Key Indicators of the Labor Database. Geneva. IMEMO (Institute of World Economy and International Relations). Russian Economic Barometer (REB) Surveys: 1991 to the Present. Moscow (www.imemo.ru/eng/ barom/survey.htm). Jaffe, Adam B., and Manuel Trajtenberg. 2002. Patents, Citations, and Innovations: A Window on the Knowledge Economy. MIT Press. References 171 Javorcik, Beata Smarzynska. 2004. “Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers through Backward Linkages.� American Economic Review 94 (3): 605–27. Kee, Hiau Looi, Alessandro Nicita, and Marcelo Olarreaga. 2006. “Estimating Trade Restrictiveness Indices.� World Bank Policy Research Paper 3840. Washington: World Bank. Keller, Wolfgang. 2002. “Trade and Transmission of Technology.� Journal of Economic Growth 7 (1): 5–24. ———. 2004. “International Technology Diffusion.� Journal of Economic Literature 42 (3): 752–82. Kikeri, Sunita, Thomas Kenyon, and Vincent Palmade. 2006. Reforming the Investment Climate: Lessons for Practitioners. Washington: International Finance Corporation. Kraay, Aart. 1999. “Exportations et Performances Economique: Etude d’un Panel d’Enterprises Chinoises� [Exports and Economic Performance: Study of a Sample of Chinese Enterprises]. Revue d’Economie du Developpement [Review of Develop- ment Economics] 1 (2): 183–207. Krueger, Alan, and Mikael Lindahl. 2001. “Education for Growth: Why and for Whom?� Journal of Economic Literature 39 (4): 1101–36. Lazareva, Olga, Irina Denisova, and Sergei Tsukhlo. 2006. “Hiring or Retraining: Russian Firms’ Experience.� Working Paper 3/2006/11. Moscow: Higher School of Economics. Lederman, Daniel, and William F. Maloney. 2003a. “Innovation in Mexico: NAFTA Is Not Enough.� Washington: World Bank, Of�ce of the Chief Economist. ———. 2003b. “R&D and Development.� Policy Research Working Paper 3024. Washington: World Bank. Litwack, John. 2005. “Achieving Diversified Growth in Russia.� Washington: World Bank. LME Survey (Russian Competitiveness and Investment Climate Assessment Surveys: Large and Medium Enterprise Survey). 2005. Washington: World Bank. Middleton, John, Adrian Ziderman, and Arvil van Adams. 1993. Skills for Productivity. Oxford University Press. Ministry of Education and Science, Russian Federation. 2006. Strategiya Razvitiya Nauki i Innovatsii v Rossiiskoi Federatsii na Period do 2015 [Research and Devel- opment Strategy of the Russian Federation to 2015]. Moscow. Ministry of Industry, Science, and Technology, Russian Federation. 2001. “Role of the State in Creating a Favorable Innovation Climate in Russia: Background Report.� OECD Helsinki Seminar, March 1–2. OECD (Organization for Economic Cooperation and Development). 1994. Science, Technology, and Innovation Policies: Federation of Russia. Vol. 1. Paris: Centre for Co-Operation with the Economies in Transition. ———. 2004a. ANBERD (Analytical Business Enterprise Research and Development) Database. Paris. 172 References ———. 2004b. Science, Technology, and Industry Outlook 2004: Multinational Enterprises and Productivity Growth: Insight at the Firm Level. Paris. ———. 2004c. STructural ANalysis (STAN) Indicators Database. Paris (www.oecd. org/sti/stan/indicators). ———. 2005. Science, Technology and Industry Scoreboard 2005. Paris. ———. 2005a. Patent Database. Paris (www.oecd.org/sti/ipr-statistics). ———. 2005b. Fostering Public-Private Partnership for Innovation in Russia. Paris. ———. 2006. Investment Policy Review of the Russian Federation. Paris. Peter, Klara Sabirianova. 2001. “The Great Human Capital Reallocation: A Study of Occupational Mobility in Transitional Russia.� EERC Working Paper Series. Moscow: EERC (Economic Education and Research Consortium) Russia and CIS. Reynolds, Sarah. 2004. “Competition Law and Policy in Russia.� OECD Journal of Competition Law and Policy 6 (3): 7–83. Riboud, Michelle, Hong Tan, and Yevgeniya Savchenko. 2006. “Globalization and Education and Training in South Asia.� Washington: World Bank. Rodrik, Dani. 2006a. “Industrial Development: Stylized Facts and Policies.� Unpublished. Harvard University, John F. Kennedy School of Government. ———. 2006b. “What Is So Special about China’s Exports?� Faculty Research Working Paper Series RWP06-001. Harvard University, Kennedy School of Government. Rosstat (State Statistics Service, Russian Federation). 2004. Indicators of Science. Moscow: Rosstat Information and Publishing Center. ———. 2006. Statistical Register of Enterprises and Establishments. Moscow: Rosstat Information and Publishing Center. ———. 2006a. Statistics of Russia. Moscow. Rutherford, Thomas, and David Tarr. 2006. “Regional Impact of Russia’s Accession to the WTO.� Policy Research Working Paper 4015. Washington: World Bank. Rutkowski, Jan, and Stefano Scarpetta. 2005. Enhancing Job Opportunities in Eastern Europe and the Former Soviet Union. Washington: World Bank. Saggi, Kamal. 2006. “Foreign Direct Investment, Linkages, and Technology Spillovers.� In Global Integration and Technology Transfer, edited by Bernard Hoekman and Beata Smarzynska Javorcik. Washington: Palgrave Macmillan and World Bank. Saltykov, Boris. 2001. “Is Russian Science a Cherry Orchard?� Nezavisimaya Gazeta, February 16, 2001. Scarpetta, Stefano, and others. 2002. “The Role of Policy and Institutions for Productivity and Firm Dynamics: Evidence from Micro and Industry Data.� Economics Department Working Paper 329. Paris: OECD. Schultz, Theodore. 1975. “The Value of the Ability to Deal with Disequilibria.� Journal of Economic Literature 13 (3): 827–46. Schweitzer, Glenn E., and Rita S. Guenther. 2005. “Basic Principles of the Russian Federation Policy on the Development of Science and Technology for the Period References 173 to 2010 and Beyond.� In Innovating for Pro�t in Russia: Summary of a Workshop. Washington: National Academies Press. SE Survey (Russian Competitiveness and Investment Climate Assessment Surveys: Small Enterprise Survey). 2005. Washington: World Bank. Slinko, Irina, Evgeny Yakovlev, and Ekaterina Zhuravskaya. 2003. “Laws for Sale: Evidence from Russia.� Working Paper W0031. Moscow: CEFIR (Center for Economic and Financial Research). Tan, Hong. 1980. Human Capital and Technological Change: A Study of Japanese Wage Differentials in Manufacturing. Ph.D. dissertation, Yale University. ———. 2000. Malaysia Skill Needs Study. Washington: World Bank Institute. ———. 2005. The Skills Challenge of New Technology. Washington: World Bank. Tan, Hong, and Geeta Batra. 1995. Enterprise Training in Developing Countries: Incidence, Productivity Effects, and Policy Implications. PSD Department Monograph. Washington: World Bank. UNCTAD (United Nations Conference on Trade and Development). 2006a. Commodity Trade Statistics (COMTRADE) Database. Geneva (http://comtrade. un.org). ———. 2006b. World Investment Report: The Shift Towards Services. Geneva. UNESCO. 2006. Education Statistics. Montreal: UNESCO Institute for Statistics (http://stats.uis.unesco.org). UNESCO (United Nations Educational, Scientific, and Cultural Organization). Various years. Statistical Yearbook. Montreal: UNESCO Institute for Statistics. U.S. PTO (U.S. Patent and Trademark Of�ce). Various years. All Technologies (Utility Patents) Report. Washington (ftp://ftp.uspto.gov/pub/taf/all_tech.htm). Watkins, Al. 2003. “From Knowledge to Wealth: Transforming Russian Science and Technology for a Modern Knowledge Economy.� World Bank Policy Research Working Paper 2974. Washington: World Bank. Weisner, Robert. 2005. “Research and Development Productivity and Spillovers: Empirical Evidence at the Firm Level.� Journal of Economic Surveys 19 (4): 587–621. Whalley, John., and X. Xin. 2006. “China’s FDI and Non-FDI Economies and the Sustainability of Future Growth.� Working Paper 12249. Cambridge, Mass.: National Bureau of Economic Research. WITSA (World Information Technology and Services Alliance). Enhanced Digital Planet 2006. Arlington, Va. World Bank. 1997. Malaysia: Enterprise Training, Technology, and Productivity. Country Study. Washington. ———. 2002–2006. Enterprise Surveys, various countries. Washington. ———. 2004. World Development Report 2005: A Better Investment Climate for Everyone. Oxford University Press. ———. 2005a. “Achieving Diversi�ed Growth in Russia.� Russian Economic Report 10. Washington. 174 References ———. 2005b. “From Disintegration to Reintegration: Eastern Europe and the Former Soviet Union in International Trade.� Washington. ———. 2005c. “Malaysia: Firm Competitiveness, Investment Climate, and Growth.� Report 26841-MA. Poverty Reduction, Economic Management, and Financial Sector Unit, East Asia and Paci�c Region. Washington. ———. 2005d. “Russian Federation—From Transition to Development: A Country Economic Memorandum for the Russian Federation.� Report Number 32308. Washington. ———. 2006a. Doing Business 2006: Creating Jobs. Washington. ———. 2006b. Russian Federation: Country Partnership Strategy. Washington. ———. 2006c. “Public Financial Support for Commercial Innovation: Europe and Central Asia Knowledge Economy Study.� Part 1. Chief Economist’s Regional Working Paper Series 1 (1). Washington. ———. 2006d. “Russian Economic Report.� Working Paper 38402. Washington. ———. 2006e. World Development Indicators 2006. Washington. Yakovlev, Andrei. 2006. “State Support Measures in Industry: Coverage, Recipients, Ef�ciency.� Unpublished. Moscow: Higher School of Economics. Yakovlev, Andrei, and Ekaterina Zhuravskaya. 2005. “State Capture: From Yeltsin to Putin.� CEFIR Working Paper. Moscow: CEFIR. Yermakov, Viktor P., and Veniamin S. Kaganov. 2000. “Razvitie malogo predprini- matel’stva v Rossii: glavnye napravleniya (po materialam II Vserossiyskogo s�ezda malykh predpriyatiy)� [Strategies for Small and Medium Enterprise Development in Russia]. Proceedings of the Second All-Russian Conference on SME Development, March 19. Contributors Enrique Blanco-Armas is an economist in the Jakarta Country Of�ce of the World Bank. Raj M. Desai is associate professor of international development at the Edmund A. Walsh School of Foreign Service at Georgetown University and a visiting fellow at the Wolfensohn Center for Development at the Brookings Institution. Vladimir Gimpelson is director of the Center for Labor Market Studies at the Higher School of Economics at Moscow State University and a research fel- low at the Institute for the Study of Labor (Bonn). John Gabriel Goddard is an economist in the Private and Financial Sector Development Department for the Europe and Central Asia Region at the World Bank. Itzhak Goldberg is senior adviser for policy and strategy in the Private and Financial Sector Development Department for the Europe and Central Asia Region at the World Bank. Smita Kuriakose is a private sector development specialist in the Private and Financial Sector Development Department for the Europe and Central Asia Region at the World Bank. 175 176 C ONTRIBUTORS Boris Kuznetsov is a chief researcher in the Interdepartmental Analytical Center and senior researcher in the Institute for Industrial and Market Stud- ies at the Higher School of Economics at Moscow State University. Yevgeniya Savchenko is a Ph.D. candidate in the Department of Economics at Georgetown University and a consultant to the Human Development Divi- sion for the South Asia Region at the World Bank. Mark Schaffer is director of the Center for Economic Reform and Transfor- mation and a professor in the Department of Economics at the School of Management and Languages at Heriot-Watt University and a research fellow at the Center for Economic Policy Research (London), the William Davidson Institute at the University of Michigan, and the Institute for the Study of Labor (Bonn). Hong Tan is a consultant to the World Bank and former lead economist in the Human Development Division for the South Asia Region at the World Bank and the Investment Climate Enhancement Unit of the World Bank Institute. Index Abramovitz, Moses, 13 Brazil: education levels in, 71; export Absorption. See Knowledge absorption structure of, 19; and in-service train- Administrative issues and investment ing, 78, 79; and investment climate, climate, 97–98, 101 95; knowledge absorption in, 36–40, Aerospace sector, 27, 39 50; labor costs in, 21–22; labor pro- Agriculture sector, 17, 39 ductivity in, 19; R&D output in, 28; Angel investors, 63–64 rigidity of employment in, 76; TFP Argentina, 85 in, 21; training programs in, 85, 86 Association of Organizations of Small Bribes. See Corruption and Medium Enterprises (OPORA, Business associations, 107–9, 111, Russia), 108 163–67 Austria, 102 Business Environment and Enterprise Automated inventory management Performance Survey (BEEPS): on system, 52–53 business associations, 108; on cor- ruption, 96; on in-service training, Barro, Robert J., 71, 142 78; and investment climate, 30–31; Bayh-Dole Act (1980), 102 and knowledge absorption, 56; on Belgium, 102 labor pool education level, 29; Benchmarking: of knowledge absorp- methodology of, 10, 11; on policy tion, 36–40; of manufacturing pro- risks, 98; and TFP, 19–22 ductivity, 17–19; of TFP, 19–22 Blanco-Armas, Enrique, 5, 35 Canada, 71 Bortnik Fund (Fund for Assistance to Capacity utilization, 4 Small Innovative Enterprises), 57, 58 Capitalization, 23 177 178 I NDEX Capital-labor ratio and employment Denmark, 102 creation, 23, 117 Desai, Raj M., 1, 7, 9, 91 Carlin, Wendy, 56 Doing Business Database (World Bank), Center for Economic and Financial 10 Research (CEFIR), 98, 99, 100 Chemical industry, 47 Early-stage technological development Chile, 85, 86 (ESTD), 63 China: agriculture sector in, 17; educa- Economic Bank of Reconstruction and tion levels in, 71; export structure of, Development (EBRD), 62 19; and global production networks, Education: demand for, 72; funding of, 43; imports from, 41; infrastructure 70, 84; and knowledge absorption, access in, 31; and investment climate, 52, 53, 142; and productivity growth, 93, 95, 96; knowledge absorption in, 29, 82; and worker skills, 6–7, 48, 69. 36–40, 50; labor costs in, 21–22; labor See also Universities productivity in, 19; R&D output in, Electrical equipment sector, 47 28; rigidity of employment in, 76; TFP Employer-targeted incentives, 7 in, 21; training programs in, 78, 79, 86 Employment contracts, 76 Cobb-Douglas production function, 21 Employment protection legislation, Commodity exports, 3, 41 76–78. See also Labor legislation Commonwealth of Independent States Energy sector, 3 (CIS), 43, 91, 96 Enterprise Surveys: and investment Communication technologies. See climate, 30–31, 109; and knowledge Information and communication absorption, 57; and regulatory risks, technologies (ICT) 98; and TFP, 19–22 Compensation policies, 70, 75–76 Equipment expenditures, 48 Competition: and investment climate, European Union (EU), 41, 102. See also 92, 93, 104–9, 163–67; and knowl- speci�c countries edge absorption, 48, 53–57 Exports: commodities, 3; growth of, 16; Confederation of Indian Industries, 87 and in-service training, 80; structure Construction sector, 16 of, 17–19, 32; technology-intensive, Corruption: and business associations, 47, 50 108; and investment climate, 31, 94, External management consultants, 95–96, 97 52–53 Council of Scienti�c and Industrial Research (India), 59, 60 Federal Antimonopoly Service, 101, Country Economic Memorandum 104–5, 109 (CEM, World Bank), 24, 105 Federal Tax Service (Russia), 99 Crime and investment climate, 31, 94, 95 Federation of Indian Chambers of Czech Republic, 71, 76 Commerce and Industry, 87 FIEs (Foreign-invested enterprises), 44 Defense sector, 27, 39, 46 Financial crisis (Russia, 1998), 16, 71 Denisova, Irina, 78, 79 Finland, 41, 102 Index 179 Firm size: and in-service training, 80; structure access in, 31; and in-service and investment climate, 96–97; and training, 78; and investment climate, knowledge absorption capacity, 50; 93, 95; knowledge absorption in, and labor productivity, 26 36–40; labor costs in, 21–22; R&D Food sector, 23, 47 output in, 28; rigidity of employ- Foreign direct investment (FDI): barri- ment in, 76; training programs in, ers to, 45–46; and business process 87; venture capital in, 64 outsourcing, 53; and external man- Industrial and Vocational Training agement consultants, 53; and knowl- Association (Japan), 86 edge absorption, 5, 43–46 Industrial sector, 24–26, 39 Foreign-invested enterprises (FIEs), 44 Information and communication tech- France, 71, 102 nologies (ICT), 44, 48, 61 Fund for Assistance to Small Innovative Infrastructure access: and investment Enterprises (Bortnik Fund), 57, 58 climate, 31, 94, 95; and knowledge absorption, 57 Germany: education levels in, 71; and Innovation: �nancial constraints on, global production networks, 43; 51–52; and knowledge absorption, imports from, 41; and intellectual 35–36; policies supporting, 2; and property rights, 102; knowledge productivity growth, 27; and train- absorption in, 36–40; labor produc- ing programs, 83. See also Soft inno- tivity in, 19; and R&D, 27, 28; TFP vation in, 21 In-service training: constraints on, 85; Gimpelson, Vladimir, 6, 7, 69, 76 and productivity, 82; recommenda- Glencore International, 44 tions for, 3, 78–81; survey data, Goddard, John Gabriel, 5, 35 152–62; and wages, 82 Goldberg, Itzhak, 1, 5, 35 Institute of Economies in Transition Government and knowledge absorption, (IET), 62 57–65 Institutional Management Committees (IMCs, India), 87 Horizontal spillover, 43 Intellectual property rights (IPRs): and Human capital, 29–30, 70–72. See also investment climate, 8, 101–4; licens- Labor pool ing of, 4–5; and patent applications, Human Resource Development Fund of 3, 38; purchase of, 50; recommenda- Malaysia, 85 tions for, 9, 59, 110 Hungary, 71, 111 International Adult Literacy Survey (IALS), 145 Iceland, 102 International Organization for Imports and knowledge absorption, 41, Standardization (ISO): and knowl- 56 edge absorption, 5, 48, 50, 53, 61; India: agriculture sector in, 17; Council and organizational change, 52 of Scienti�c and Industrial Research, Internet and knowledge absorption, 48 59, 60; export structure of, 19; infra- Inventory management system, 52–53 180 I NDEX Investment climate, 91–112; and shortages in, 12, 72–81; during tran- administrative issues, 97–98; and sition period (1992–2004), 22–24 asset ownership risks, 99–104; chal- Labor productivity. See Productivity lenges in, 3; and competition barri- Labor turnover, 75, 84 ers, 104–9, 163–67; and employ- Land reform: and investment climate, ment creation, 23; and firm size, 94; and privatization, 8–9, 99–101; 96–97; improvement of, 7–10; inter- recommendations for, 109–10 national comparisons, 93–96; and Large and Medium Enterprise (LME) knowledge absorption, 48; and Survey: data from, 125–41; and labor shortages, 73, 149–50; and industrial heterogeneity, 25–26; on productivity growth, 30–31; recom- in-service training, 78; and knowl- mendations for, 109–11; and regula- edge absorption capacity, 47; and tory and policy risks, 97–98; and labor legislation, 77; and labor short- taxation, 98 ages, 70, 72–73; methodology, 10, Ireland, 102 113–16; and R&D expenditures, Italy, 41, 102 27–29; and worker skills, 29–30 Law on Competition and the Japan, 71, 86 Restriction of Monopolistic Activity Javorcik, Beata Smarzynska, 44 in Goods Markets (1991), 104 Job-creators, 23 Layoff regulations, 76 Job-destroyers, 23 Lazareva, Olga, 78, 79 Joint ventures, 6, 44 Lee, Jong-Wha, 71, 142 Legal system: and intellectual property Knowledge absorption, 35–68; capacity rights, 38, 102, 103–4; and investment for, 46–57; domestic sources of, climate, 31, 94; and labor law, 70, 36–40; and foreign direct invest- 76–78, 92, 94; and property rights, 101 ment, 43–46; foreign sources of, Levies for training, 85–86 40–46; international benchmarking Licensing of intellectual property, 38, 83 of, 36–40, 50; and new industrial Lukoil, 44 policy, 2; recommendations for, 4–6, 59–65; role of government in, 57–65; Machinery expenditures, 48 and trade, 41–43 Malaysia, 78, 86, 87 Knowledge spillover, 41, 43, 46 Manufacturing sector: creative destruc- Kuriakose, Smita, 5, 35 tion by, 23–24; international com- Kuznetsov, Boris, 4, 12 petitiveness of, 16, 17; and knowl- edge absorption, 47; and labor Labor costs: and EPL, 77–78; at manu- productivity, 26; and R&D, 37; real facturing �rms, 3; and productivity, wages in, 4 4; and TFP, 21–22 Market failures, 70, 85 Labor legislation, 70, 76–78, 92, 94 Market services sector, 16 Labor pool: aging of, 12; education of, Matching grant programs, 5–6, 7, 61 48, 52, 53; redistribution of, 14, 69; Mauritius, 86 Index 181 Media, free and independent, 92 Privatization, 8–9, 99–104 Metals sector, 3, 16, 74 Productivity, 12–34; benchmarking of Mexico, 86, 111 TFP, 19–22, 117–24; growth factors Military technology, 27, 46 in, 26–31; and industrial heterogene- Minimum wage legislation, 76 ity, 24–26; international compar- Ministry of Antimonopoly and isons, 17–19; patterns and sustain- Entrepreneurship Support (MAPES, ability of, 4–10; recent trends, 14–17; Russia), 104 and TFP, 19, 117–24; and training Ministry of Economic Development programs, 81–83; in transition period and Trade (Russia), 101, 109 (1992–2004), 22–24; and under- Monopolies, 104–5 staf�ng, 76; and worker skills, 69 Productivity and Investment Climate National Innovation System (NIS), 59 Surveys (PICS). See now Enterprise Nelson Resources, 44 Surveys Netherlands, 41, 102 Professional training. See In-service Norway, 102 training Pro�tability and staf�ng levels, 76 Oil and gas sector, 13, 14, 16, 44 Programme for International Student OPORA (Association of Organizations Assessment (PISA), 71, 145 of Small and Medium Enterprises, Property rights, 99–104 Russia), 108 Public-private partnerships, 6, 44 Organizational change and soft innova- Purchasing power parity (PPP), 17 tion, 52–53 Outsourcing, 52–53 Quality control, 52–53 Patents, 3, 38 Real estate sector, 16, 100, 101 Payroll-levy training funds, 7 Regulatory environment: and invest- Penang Skills Development Centre ment climate, 97–98; and labor legis- (Malaysia), 87 lation, 76–78, 92, 94; recommenda- PISA (Programme for International tions for, 110–11; and transparency, Student Assessment, 71, 145 9–10 Poland: education levels in, 71; export Regulatory Reform Programme structure of, 19; and global produc- (OECD), 104 tion networks, 43; and intellectual Research and development (R&D): property rights, 102; labor costs in, expenditures for, 48; foreign, 57; and 21; labor productivity in, 19; R&D in-service training, 79, 80; institu- output in, 28; TFP in, 21 tions, 28, 40; and knowledge absorp- Policymaking: and innovation, 2; and tion, 2, 5, 35, 36–37, 48; private investment climate, 8, 9, 93, 94, expenditures on, 38, 49, 103; and 97–98 productivity, 5, 27–29; public �nanc- Political stability and investment ing of, 5, 37–38; recommendations climate, 31 for, 59 182 I NDEX Resource reallocation, 14, 23, 24, 92 rights, 103; and R&D activity, 28–29, Retail trade sector, 16 40; understaf�ng of, 74; and venture Royalties for intellectual property, 38 capital, 62, 64 RUSAL, 44 SUAL Group, 44 Russia Economic Barometer, 73, 75, 76 Sweden, 102 Russian Competitiveness and Investment Climate Assessment Survey, 10, 11, 70 Tan, Hong, 6, 7, 69 Russian Urban Institute, 101 Tariffs, 45 Taxation: and investment climate, 31, Savchenko, Yevgeniya, 6, 7, 69 92, 93, 94, 96, 98; and knowledge Schaffer, Mark, 4, 12, 56 absorption, 62 Science sector, 3, 71 Technical training, 69, 72 Seabright, Paul, 56 Technology diffusion, 44 Secondary education, 7 Tertiary-level education, 7, 50, 71 Seed capital programs, 6, 62–65 Textiles sector, 23, 47, 73 Service sector, 3–4 TIMSS (Trends in International Simpli�ed System of Taxation for Small Mathematics and Science Study), 71, Enterprises, 99 144–45 Singapore, 86 Total factor productivity (TFP): bench- Skills. See Worker skills marking of, 19–22, 117–24; de�ned, Small Enterprise Survey, 10, 70 13 Social bene�ts and guarantees in labor Trade: balance of, 16–17; and competi- legislation, 76 tion, 46; and knowledge absorption, Soft innovation, 52–53 4–5, 41–43. See also Exports Solow, Robert, 13 Trade-Related Aspects of Intellectual Solow residual, 13 Property Rights (TRIPs), 103 South Africa: and investment climate, Trade restrictiveness index (TRI), 45 96; knowledge absorption in, 36–40, Training programs: constraints on, 85; 50; labor productivity in, 19 funding of, 7, 70; and knowledge South Korea: agriculture sector in, 17; absorption, 48, 83–84; productivity export structure of, 19; knowledge and wage outcomes of, 29, 81–83; absorption in, 36–40; and R&D, 27, recommendations for, 84–87; survey 28; TFP in, 21; training programs in, data, 152–62. See also In-service 86 training Soviet legacy: and educational attain- Transformational recession, 14 ment, 71; and R&D, 27, 36–37, 40 Transition period (1992–2004), 22–24, Space program, 27, 39 70–72 Spain, 21, 28, 102 Transparency and investment climate, 8, Special economic zones, 8 92 State capture, 106–7 Trends in International Mathematics State-owned and state-managed organi- and Science Study (TIMSS), 71, zations: and intellectual property 144–45 Index 183 TRIPs (Trade-Related Aspects of Wood sector, 47 Intellectual Property Rights), 103 Worker productivity. See Productivity Tsukhlo, Sergei, 78, 79 Worker shortage. See Labor pool Worker skills, 69–90; constraints on, Unionized �rms, 77 72–81; demand for higher, 84; evolu- United Kingdom, 71, 102 tion of, 70–72; and in-service train- United States: education levels in, 71; ing, 6–7, 78–81; and knowledge and intellectual property rights, 102; absorption, 4, 48; and productivity and knowledge absorption, 41; and growth, 29–30; recommendations regulatory issues, 98 for, 84–87; shortages of, 76. See also Universities: demand for, 72; and intel- Training programs lectual property rights, 102; recom- World Bank: Business Environment mendations for R&D at, 59; and Survey, 81; Country Economic research funding, 5, 40 Memorandum, 105; Development U.S. Agency for International Indicators, 11; Doing Business Development, 98 Database, 10; Enterprise Surveys, 10, 11; on in-service training, 78; on Venture capital programs, 6, 62–65 R&D resource inef�ciencies, 36; and Vertical spillover, 43, 46 regulatory issues, 98; on rigidity of Vocational training, 69, 72, 146 employment, 76–77 World Development Indicators (World Wages: and competition for skilled Bank), 11 workers, 84; wage grid system, 72; World Development Report 2005 on growth in, 4; inequality of, 70–71; investment climate, 57 inflation of, 21; and training pro- World Economic Forum, 98 grams, 81–83; and understaf�ng, 76. World Trade Organization (WTO), See also Compensation policies 103 WBES (World Bank Business Environment Survey), 81 Yakovlev, Evegeny, 57 The Brookings Institution is a private nonprofit organization devoted to research, education, and publication on important issues of domestic and foreign policy. Its principal purpose is to bring the highest quality independent research and analysis to bear on current and emerging policy problems. The Institution was founded on December 8, 1927, to merge the activi- ties of the Institute for Government Research, founded in 1916, the Institute of Economics, founded in 1922, and the Robert Brookings Graduate School of Economics and Government, founded in 1924. Interpretations or conclusions in Brookings publications should be understood to be solely those of the authors. Board of Trustees John L. Thornton Mario Draghi William A. Owens Chair Kenneth M. Duberstein Frank H. Pearl Strobe Talbott Alfred B. Engelberg John Edward Porter President Lawrence K. Fish Edgar Rios Robert J. Abernethy Cyrus F. Freidheim Jr. Haim Saban Liaquat Ahamed Bart Friedman Sheryl K. Sandberg Alan R. Batkin David Friend Victoria P. Sant Richard C. Blum Ann M. Fudge Leonard D. Schaeffer Geoffrey T. Boisi Jeffrey W. Greenberg Lawrence H. Summers Abby Joseph Cohen Brian L. Greenspun David F. Swensen Arthur B. Culvahouse Jr. Glenn Hutchins Larry D. Thompson Alan M. Dachs Joel Z. Hyatt Andrew H. Tisch Kenneth W. Dam Shirley Ann Jackson Laura D’Andrea Tyson Steven A. Denning Kenneth M. Jacobs Antoine W. van Agtmael Vishakha N. Desai Suzanne Nora Johnson Beatrice W. Welters Paul Desmarais Jr. Philip H. Knight Daniel Yergin Thomas E. Donilon Harold Hongju Koh Daniel B. Zwirn Honorary Trustees Leonard Abramson Teresa Heinz Mario M. Morino Elizabeth E. Bailey F. Warren Hellman Maconda Brown O’Connor Zoë Baird Samuel Hellman Samuel Pisar Rex J. Bates Robert A. Helman Steven Rattner Louis W. Cabot Roy M. Huf�ngton J. Woodward Redmond James W. Cicconi James A. Johnson Charles W. Robinson A. W. Clausen Ann Dibble Jordan James D. Robinson III William T. Coleman Jr. Michael H. Jordan Warren B. Rudman D. Ronald Daniel Vernon E. Jordan Jr. B. Francis Saul II Robert A. Day Breene M. Kerr Ralph S. Saul Bruce B. Dayton Marie L. Knowles Henry B. Schacht Charles W. Duncan Jr. James T. Lynn Michael P. Schulhof Walter Y. Elisha Jessica Tuchman Mathews Joan E. Spero Robert F. Erburu David O. Maxwell Vincent J. Trosino Henry Louis Gates Jr. Donald F. McHenry John C. Whitehead Robert D. Haas Robert S. McNamara Stephen M. Wolf Lee H. Hamilton Mary Patterson McPherson James D. Wolfensohn William A. Haseltine Arjay Miller Ezra K. Zilkha Can Russia Compete? Russia has joined the club of rapidly growing economies, driven by rising prices for oil and gas. Russian authorities are concerned about sustaining this economic performance, however. Thus they have sought to diversify and broaden the basis for growth beyond the profitable natural resource sector and to build an economy that will be able to compete in the twenty-first century. This book presents the case that Russia’s neglected manufacturing sector is in need of human capital and policy reforms that will unleash its innovative potential. Can Russia Compete? quantifies and benchmarks the relative strengths and weaknesses of that sector, identifying opportunities to increase Russian productivity and competitiveness. Drawing on original survey data from Russian firms of all sizes, the authors formu- late proposals that aim to • enhance the innovative potential of Russian firms, • upgrade the skills of their workforces, and • develop a business-friendly climate with lower administrative costs and greater policy certainty. The original research and analysis of Raj Desai, Itzhak Goldberg, and their col- leagues will be of use to anyone interested in the problems of building manufac- turing competitiveness in a globally integrated economy. It will also be of interest to organizations planning to do business or invest in Russia. Contributors Enrique Blanco-Armas, Vladimir Gimpelson, John Gabriel Goddard, Smita Kuriakose, Boris Kuznetsov, Yevgeniya Savchenko, Mark Schaffer, Hong Tan Raj M. Desai is a visiting fellow in the Global Economy and Development program at the Brookings Institution. He is also a professor in Georgetown University’s Edmund A. Walsh School of Foreign Service. Itzhak Goldberg is an adviser for policy and strategy with the World Bank and has been advising governments in postsocialist transition economies since 1990. He has previously served as a research fellow at the Hoover Institution. Brookings Institution Press Washington, D.C. www.brookings.edu Cover design by Kimberly Carusone based on an illustration by A. Salensky, Tours to USSR 1932.