Report No. 20192 Dominican Republic Social and Structural Policy Review Volume 11 March 23, 2000 Poverty Reduction and Economic Management Unit Latin America and the Caribbean Document of the World Bank 1. Currency Equivalents Currency Unit: Dominican Peso (DR$) Exchange Rate US$1.00 = DR$16.35 (as of March 22, 2000) 2. Fiscal Year January 1 - Dec 31 3. Acronyms and Abbreviations CB Central Bank CEA State Sugar Council CDE National Electricity Corporation FEyD Fundacion Economia y Desarollo FTZ Free Trade Zones GDP Gross Domestic Product INESPRE Price Stabilization Institute LAC Latin America and the Caribbean ONAPLAN National Planning Office ONAPRE National Budget Office SPI Banking Sector Superintendency WTO World Trade Organization Vice Pesident David de Fen-anti Country Director Orsalia Kalantzopoulos Sector Director Guillermo Perry Task Manager John Panzer TABLE OF CONTENTS Section 1: Capital Accumulation, Technological Change and Growth 1 A. Background 1 B. Sources of Growth in the Dominican Republic 2 C. Sources of Growth excluding the FTZs 3 D. Sources of Growth for the FTZs 4 E. The Role of Public Investment on Growth 6 F. Conclusions 7 References 9 Annex 11 Endnotes 1 5 Section 2: Determinants of Poverty 16 A. Introduction 16 B. Poverty in the Dominican Republic 16 Examining the sample population 17 Rural versus urban households 18 Income distribution among rural households 18 Landed versus landless households 19 Farmers versus farm laborers 19 C. The Determinants of Per Capita Income 20 The determinants of individual income 21 Occupational choice 21 Earnings equations 22 D. Education, Educational Mobility, and Poverty 23 School attendance 23 Educational attainment 24 E. Main Findings 25 Tables and Figures 26 Section 3: Macroeconomic Policy 38 A. Broad Macroeconomic Assessment 38 Public Finances 38 The External Sector 40 Monetary and Exchange Policy 41 B. The Short and Medium run: The Conduct of Monetary Policy and the Fiscal Question 43 Background 43 The Conduct of Monetary Policy 44 C. Growth, The Demandfor Money and The Inflation Tax 47 D. Fiscal, Monetary Policy and Interest Rates 53 E. The Long Run: Some Considerations on the Future of the Monetary System 58 F. Some Considerations on the CB Independence and the Conflict between Fiscal and 59 Monetary Policy Endnotes 60 Section 4: Commercial Policy 62 A. Introduction 62 B. A General Equilibrium Model 62 C Current Commercial Policies 67 D. Commercial Policy Evaluation 70 E. Main Findings 76 Annex 77 StatisticalAnnex 79 References 81 Endnotes 81 Section 5: Regulatory Framework and Banking Sector Performance 83 A. The Legal and Institutional Framework 83 B. The System of Regulations and Controls 84 Regulation on Exposures and Connected and Related Parties 86 C. Public Disclosure and Transparency 88 D. Banking Sector Performance 88 Sector Growth and Risks 90 E. Addressing Sector Risks 91 Procedures for remedying problem bank situations 91 Institutional Requirements for Remedying Banking Problems 91 Deposit Insurance Scheme (DIS) 92 Annex 93 Endnotes 95 Section 6: Budget Law and Public Sector Management 97 A. Budget Law 97 B. The Presidential Fund 1401 and the Budget Management Process 98 C The Agenda for Reform 99 Tables 100 ii SECTION 1: CAPITAL ACCUMULATION, TOTAL FACTOR PRODUCTIVITY, AND GROWTH A. Background 1. Since the restoration of macroeconomic stability in 1991, the Dominican Republic has entered a period of remarkable economic growth. During the last seven years (through 1999), the annual average GDP growth has exceeded 6 percent (Table 1), with even higher growth rates in recent years. This positive performance came after more than a decade of low and volatile growth dating back to the late 1970s. The resumption of strong economic growth has already had a beneficial impact on poverty and recent analysis of the Central Bank of the Dominican Republic indicates that between 1992 and 1998, the incidencel of poverty declined from 31.7 to 25.8 percent. 2. With more than 2 million Dominicans still living in poverty, the sustainability of growth is a key issue. The duality of the production structure raises questions about the sustainability of current growth rates. On the one hand, there are areas such as tourism, telecommunications, and the industrial free-trade zones (FTZs), which operate in a highly competitive environment, are closely linked to the world economy, and are often shielded from state intervention through special administrative arrangements. On the other hand, there are the more traditional sectors of the economy, such as agriculture and some subsectors of industry (outside the free-trade zones), which continue to operate in the midst of strong state intervention including excessive protectionism, red tape, and insecure property rights. Without continued progress in implementing structural reforms (particularly in the more traditional sectors), sustained growth could be jeopardized. Table 1.1 Economic structure and growth, 1991-1998 1991 1998 1991-98 1991-98 (In percent (In percent Average Annual Contribution to of GDP) of GDP) Growth Growth (In percent) (In percent) Construction 7.5 12.1 13.8 20.7 Commerce 12.4 12.9 6.8 12.5 Hotel, Bar, & Rest. 4.2 7.0 14.0 14.3 Industry 14.9 13.1 4.3 7.5 Telecommunications 2.4 4.7 16.5 9.0 Agriculture 13.9 11.7 3.5 6.8 FTZ" 3.3 3.4 7.2 5.3 Total Economy 100.0 100.0 6.1 100.0 Source: Central Bank of the Dominican Republic. ' The national accounts methodology does not fully capture the contribution to growth of the FTZs, since it is based only on the wage bill and not on value added. 3. The remainder of this chapter is organized as follows. Section B considers the main sources of growth, including capital accumulation and productivity increases. However, as discussed above, the production structure in the Dominican Republic is dualistic. Hence, it is necessary to examine the sources of growth excluding the free-trade zone (Section C) and of the free-trade zones separately (Section D). Section E focuses on the role of public investment in promoting growth. Section F concludes. 1 Section 1 B. Sources of Growth in the Dominican Republic 4. Studying the determinants of growth has been at the core of economic research for half a century. The pioneering study of Solow (1956) on the mechanics of factor accumulation and GDP growth led to a host of applied papers designed to test the applicability of the exogenous growth model, the role of total factor productivity, and the validity of the standard neoclassical production function (for a survey see Madison, 1987). Most of these studies were based on the notion that economic growth was the outcome of capital and labor accumulation and technological improvements leading to factor productivity growth. Accordingly, it was assumed that value added in each sector could be expressed as:2 Logyt=r+alLogKt+a2LogLt+ Pt (1) 5. Because the variables in equation (1) are nonstationary (see unit root tests in the Technical Annex) parameter estimates using standard techniques (e.g., OLS methods) do not have standard asymptotically normal distributions and are prone to spurious correlations (Granger and Newbold, 1974). This model is therefore best estimated by using cointegration techniques. In addition, since lack of proper accounting of short-term fluctuations can potentially bias the estimation (Hargreaves, 1994), this basic specification was extended to capture the short-run dynamics of growth arising from the presence of transitory shocks by estimating: ALogyt =a A Log Kt + (1 - c 0)A Log Lt +fi[LogYt-j - Y - a Log Kt-j - ( - al ) Log Lt-J] + 6t (2) 6. In this specification, A is the first difference of the variables, and the expression in brackets corresponds to the last period's deviation of output from its long-term determinants. For example, from a condition of excess capacity the self-correcting mechanism immediately calls for a future expansion in growth. The speed of the adjustment is determined by ,3; the smaller the value of ,B, the slower is the adjustment process. The parameter cio captures the short-run effect of capital and labor on the growth rate, while parameter a, captures the long-run effect of capital and labor on output. The latter are then used to estimate the contribution of each factor of production to growth with the residual accounting for technology or productivity change. 7. The model from equation (2) was estimated for the 1970-1998 period using annual data, with separate estimates being carried out for the FTZs to capture some of the dual nature of the Dominican economy.3 The estimates presented in Table 2 indicate that the share of capital in the economy's production function (excluding the FTZs)-at about 63 percent-is similar to factor compensation shares in the country's national accounts of roughly 65 percent during 1992-1997.4 Surprisingly, the estimated equation for the FTZs does not reflect the conventional thought that the FTZs are more labor intensive than the rest of the economy. Finally, the parameter for the speed of adjustment in each sector, B, suggests that FTZs adjust to shocks about twice as quickly as the rest of the economy (1.5 years versus 3 years). 2 Capital Accumulation, Total Factor Productivity and Growth Table 1.2 Production function estimates: 1970-1998 Sectors ~~~~~~~~~~~~~~~R2R2 Error D Sectors CaO al fi YCointegration* Correction DW Free-trade zones 0.714 0.301 0.907 -4.09 0.980 0.817 2.07 (0.042) (0.212) (0.183) (0.184) Economy 0.632 0.824 0.337 -0.355 0.972 0.570 1.99 (0.168) (0.161) (0.122) (0.276) Note: (*) First-stage estimation, as described in the Technical Annex. Standard deviations are in parentheses. C Sources of Growth Excluding the FTZs 8. The estimates of sources of growth based on the production function estimates of Table 2 indicate that for the last twenty five years, the most important source of growth, by far, has been capital accumulation. At the same time, there has been a remarkably low contribution of total factor productivity growth to overall economic growth. This result is even stronger when considering that the estimate of labor growth does not take into account improvements in educational attainment that have undoubtedly taken place in the last three decades5 Consequently, it could be expected that the unmeasured improvements in human capital are being captured in the form of total factor productivity increases, which nevertheless average only 0.4 percent per year. Table 1.3 Sources of growth estimates (in percent) GDP Capital Labor P dTotal Factor GDP CptlLbrProductivity Growth Period Growth Growth Contri- Share in Growth Contri- Share in Contri- Share in (Per (Per bution to GDP (Per bution to GDP bution to GDP annum) annum) GDP Growth annum) GDP Growth Growth Growth Growth Growth Growth_ 1973-98 4.6 4.5 2.8 62.3 3.6 1.3 29.2 0.4 8.5 1973-82 5.6 5.3 3.4 59.6 4.1 1.5 26.7 0.8 13.7 1983-91 2.2 3.6 2.3 106.1 3.9 1.4 65.8 -1.6 -71.9 1992-98 6.1 4.4 2.8 45.8 2.6 1.0 15.8 2.3 38.3 9. Such low overall improvements in productivity should obviously be a source of concern, since they raise doubts about the sustainability of strong growth. However, separating the analysis into different periods (Table 3), provides a somewhat different picture. The strong growth since 1992 has relied both on capital accumulation and on important growth in total factor productivity. The increase in capital accumulation is attributable to a rebound in savings that has taken place since 1992. The ratio of investment to GDP, which had declined from about 23 percent in the 1970s to 20 percent during the 1982-91 period, has been consistently increasing since 1992 averaging more than 24 percent for the 1992-98 period. By the same token, total factor productivity growth since 1992 has averaged a historically high 2.3 percent per year. 10. Assessing the determinants of this increase in capital accumulation and total factor productivity growth is beyond the scope of this chapter. However, some important stylized facts are worth mentioning. First, there is a strong correlation between the successful macroeconomic stabilization of 1991 and the initiation of structural reforms in the early 1990s with the rebound in capital accumulation and total factor productivity growth. Second, the instability associated with the debt crisis of the 1980s not only affected capital accumulation, but also productivity growth. While some of the high productivity growth in the early 1990s may have been attributable to a rebound effect, toward the end of the 1 990s this would be less a factor. Nonetheless, the analysis would indicate that sustained strong productivity growth will not be 3 Section 1 possible without continued implementation of structural reforms. Third, the experience of other developing countries shows that sustaining productivity growth rates of the kind that have been experienced recently by the Dominican Republic is difficult. Some of the fastest growing economies in the world such as Chile, Ireland, Korea, and Taiwan have not been able to sustain average annual productivity growth rates in excess of 2 percent for more than five or six years. Table 1.4 Total factor productivity growth of other developing countries Korea Taiwan Singapore Chile Ireland Morocco Jordan 1960-70 0.60 1.40 0.10 1.40 2.50 4.60 -1.10 1970-80 0.80 1.10 0.40 0.00 1.70 -0.20 3.00 1980-86 2.50 1.80 -0.80 -1.90 1.20 -0.30 -2.60 1986-92 1.90 2.50 4.00 3.80 3.40 -1.20 -4.30 Source: Bosworth, Collins and Chen, 1995. D. Sources of Growth for the FTZs 11. The traditional methodology for estimating sources of growth assumes that the long-run structure of the economy is unaffected by the accumulation of capital and labor and, therefore, the underlying production function remains constant over time. In the presence of important structural reforms and a rapidly expanding sector such as the FTZs, where learning and adaptation is an important process, this may not be a reasonable assumption. Soto (1997) discusses why a model in which learning is substantial cannot be estimated using the standard, fixed parameter model. In that case, the existence of an S-shaped learning curve would lead to substantially distorted and nonrobust results stemming from the omission of the arrival of new information to economic agents. A more sensible assumption would be to expect a learning curve, as free-trade zones evolve from an infant to a mature industry. Taking this into consideration, the sources of growth analysis for the FTZs is carried out using a Kalman-filter specification: A Log yt = a0 A LogKt +(I - a ) A Log Lt +f, [ LogYt-l- Y - (al+ SVt) Log Kt-l - (I - al -SVt) Log Lt-lI +]ut where SVt = ° SVt-l + vt (3) 12. This specification allows parameters to evolve as agents learn to be more productive with given inputs. In standard microeconomic terms, the specification allows the FTZs to be initially at an interior point in the transformation curve and move sequentially towards the efficient frontier. A graphic display of the estimated time-evolving parameter is presented in the next figure. 4 Capital Accumulation, Total Factor Productivity and Growth Figure 1.1 Dominican Republic Kalman-Filter Estimates of the Share of Capital in FTZ Production Function 0.8 0.8 0.7 0.7 . . ,, 0.6 .r 0.6 0.5 0.5 * 0.4 0.4 0.3 - I l l l l l l l l l I I 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 13. An analysis of the sources of growth for the FTZs shows remarkable productivity growth. From 1973 to 1998, the annual average productivity growth of the FTZs has consistently been more than four times higher than that of the economy at large. It must be pointed out, however, that because of data constraints, in recent years the estimate of total factor productivity growth is probably being overestimated, while capital accumulation, and therefore the contribution of capital to growth, has been underestimated. This result stems from the absence of detailed data on investment in the FTZs. Consequently, the area of their physical plants and their level of utilization is used as a proxy for capital. However, in recent years as the constraints of market access to the United States have become more binding, there has been an important process of diversification that has included increased vertical integration-and value added (VA)-in the production of garments, and also the development of other more capital-intensive products. Since these shifts can take place in the absence of expansion of the physical plants of the FTZs, total factor productivity growth is likely being overestimated. Table 1.5 Sources of growth estimates in the FTZs (in percent) Value Capital Growth Labor Growth Total Factor Added Productivity Growth Added ~~~Contri- Contri- .Conitri- Saei Period Growth Growth butin- Share in Growth onto Share in Con Share m (Per (Per bution VA (Per butVoAn to VA bution VA annum) annum) Growth Growth annum) Growth Growth Growth Growth 1973-98 21.3 15.0 10.7 50.4 19.7 5.6 26.5 4.9 23.1 1973-82 18.9 11.7 8.4 44.4 25.0 7.1 37.9 3.3 17.8 1983-91 29.1 26.5 18.9 65.2 24.8 7.1 24.4 3.0 10.4 1992-98 14.7 4.9 3.5 23.8 5.6 1.6 10.9 9.6 65.4 14. Ultimately, however, whether because of different levels of investment and capital/labor ratios or because of differences in factor productivity growth, there is a clear difference in average productivity between the FTZs and the rest of the economy. This important difference indicates the output gains that could be derived were the type of regulatory environment that applies to the FTZs to be extended to the economy at large. 5 Section I Table 1.6 Average productivity in the economy and the FTZs Free-Trade Zones Economy without FTZs Period VA Growth Growth in Average VA GDP Growth Growth in Average per Worker GPprWre 73-98 21.3 3.28 4.6 1.12 73-82 18.9 -2.23 5.6 2.10 83-91 29.1 3.96 2.2 -1.56 92-98 14.7 8.70 6.1 3.44 E. The Role of Public Investment in Growth 15. The analysis of the preceding sections suggest that maintenance of strong growth in the Dominican Republic will hinge on at least three elements: (a) the maintenance of macroeconomic stability; (b) the continued implementation of structural reforms that can help sustain high factor productivity growth; and (c) increasing the rate of capital formation. In the past, the process of capital formation was reinforced by macroeconomic stability and the government's own public investment program. However, since late 1996, there has been a significant shift in the composition of government expenditures, away from investment and toward current expenditures (including the wage bill). While some of the increase in the wage bill has been directed to finance human capital formation, this evolution raises concerns regarding its likely impact on private investment and, subsequently, on long-run growth. Table 1.7 Composition of public expenditures, 1995-1998 (in percent of GDP) 1995 1996 1997 1998 Current expenditures 7.2 7.8 10.6 11.1 Current expenditures adjusted 6.0 6.5 9.1 n.a. Wages 3.5 3.6 5.1 5.2 Transfers 1.9 2.5 3.3 3.5 Capital expenditures 7.0 6.6 5.5 5.1 Investment 4.5 4.6 2.8 2.7 Capital transfers 1.7 1.3 1.6 1.3 Amortization 0.7 0.7 1.6 0.8 Source: Based on infornation provided by ONAPRE. ' Current expenditures adjusted equals total current expenditures minus wages for education. 16. To analyze the impact of the changing composition of public expenditures on growth and the ensuing policy implications, estimates were made of the net contribution of public investment to total investment.6 There is a vast literature on modeling the determinants of investment based on the relation between output and the cost of capital (Jorgenson, 1979), the role of uncertainty (Abel, 1983), the response of investment to transitory output shocks (Blejer and Khan, 1984), and the role of the effect of real exchange rate changes on the profitability of investment projects and the cost of imported intermediate or capital goods (Krugman and Taylor, 1978). More recently, the role of economic stability has also been tested as a key determinant of private investment (Serven and Solimano, 1993). There is evidence that public investment may either crowd in or crowd out private investment, depending on the nature of those investments (Balassa, 1988, Khan and Reinhardt, 1990, and Easterly, Rodriguez, and Schmidt-Hebbel, 1994). Estimates for the Dominican Republic are based on the following equation: 6 Capital Accumulation, Total Factor Productivity and Growth Priv. Invt Publ. Invtt Gov.Constt f =vGDPt, Opennesst, A Capital Costt, Wt, ;, t(4) GDPt GDPt GDPt where openness is measure by the ratio of exports to GDP,7 I is inflation, and W is the private real wage index. 17. The estimates presented in Table 8 show that in the Dominican Republic public investment has had a net crowding-out effect on private investment since the parameter for total public investment is negative and significant, suggesting that for every percentage point of increased public investment, private investment decreases by 0.2 percentage point. This crowding-out effect is significantly lower than that recently estimated by the World Bank (1998) for Mexico, where it reached a high 50 percent. Moreover, when separating the type of public investment between basic public goods and other nonpublic goods some interesting results and policy implications are obtained. Public investment in basic public goods has a crowding-in effect of about 30 percent, thus contributing to increased private investment, while public investment in nonbasic public goods has a crowding-out effect of 15 percent. This would imply that even in the context of declining public investment the reallocation of public investments towards core public goods would allow for continued strong capital formation. Table 1.8 The determinants of private investment, 1979-1996 Constant Chiange Lagged GDP Openness Lagged Public Investmnent AR(l) R2 DW Cost Change Growth Excess Capital Wages Capacity Total Lagged Lagged goods Other2 -2.353 -0.013 -0.234 1.481 0.279 0.069 -0.177 0.816 0.71 2.14 (-10.47) (-1.59) (-1.72) (3.70) (2.36) (3.09) (-2.50) (9.84) -1.258 -0.019 -0.129 1.030 0.259 0.045 0.296 -0.151 0.872 0.76 1.96 (-2.92) (-2.37) (-1.32) (4.41) (2.39) (1.14) _ (3.88) (-2.98) (8.51) Notes: T-statistic in parenthesis. N = 18 observations. " Basic public goods involve investments in transport, rural roads, water & sewerage. 2/ Other includes housing, irrigation, communications, energy, agriculture and urban developments. Source: Bank estimates based on public investment classification from ONAPLAN. F. Conclusions * The restoration of macroeconomic stability and the initiation of structural reforms in the early 1990s has led to strong economic growth and poverty reduction. Growth has been anchored both by a resurgence of capital formation and strong productivity growth, although the latter may be slightly overestimated given the imperfect measurement of human capital embodied in the measure of labor inputs. * With a ratio of investment to GDP of about 26 percent in 1998, a continuation of GDP growth rates between 7 and 8 percent per year will require maintaining high productivity growth and investment. The past performance record of the Dominican Republic, and international experience, indicates that in the absence of continued structural reforms current rates of productivity growth will be hard to maintain. This implies that: (1) the Dominican Republic should push forward with its implementation of structural reforms-recent developments are 7 Section I encouraging in this regard; and (2) strong capital formation needs to be maintained. To this end, increasing public investment and improving the focus of those investments to promote a crowding-in of private investment should be a priority. * Increases in real GDP growth rates and openness promote private investment. According to the empirical results, each extra percentage point of GDP in exports, will raise private investment by 0.3 percentage point of GDP. Thus, the results are consistent with the notion that trade liberalization not only reduces price distortions and increases efficiency, but also encourages private investment, and ultimately output growth. 8 Capital Accumulation, Total Factor Productivity and Growth References Abel, Andrew, 1983, "Optimal Investment Under Uncertainty," American Economic Review, Vol. 73, pp. 228-33. Balassa, Bela, 1988, "Public Finance and Economic Development," PPR Working Paper No. 31, The World Bank. Blejer, Mario and Mohsin Khan, 1984, "Government Policy and Private Investment in Developing Countries," IMF StaffPapers, Vol. 31. Braun, J. and M. Braun, 1998, "El Crecimiento Potencial: El Caso de Chile," mimeo, Pontificia Universidad Cat6lica de Chile. Dahuajre, A. et al., 1989, "Impacto de las Zonas Francas Industriales de Exportaci6n en la Repuiblica Dominicana," Fundaci6n Economia y Desarrollo. Easterly, W. C.A. Rodriguez, and K. Schmidt-Hebbel, 1994, The Macroeconomics of Public Sector Deficits, Oxford University Press for The World Bank. Engle, R. and C. Granger, 1987, "Co-Integration and Error-Correction: Representation, Estimation, and Testing," Econometrica, Vol. 35, pp. 251-276. Granger, C.W.J., 1969,"Investigating Causal Relations by Econometric Models and Cross-Spectral Models," Econometrica, Vol. 37, pp. 424-438. Granger, C.W.J. and Newbold, P., 1974, "Spurious Regressions in Econometrics," Journal of Econometrics, Vol. 2, No. 2, pp. 111-120. Hamilton, J.D., 1994, "Time Series Analysis;" Princeton University Press, N.J. Hargreaves, C., 1994, "A Review of Methods of Estimating Cointegration Relationships" in Non Stationary Time Series Analysis and Cointegration, Oxford University Press. Jorgenson, Dale, 1979, "Econometric Studies of Investment Behavior: A Survey," Journal of Economic Literature, Vol. 9, pp. 1111-1147. Khan, M. S. and C. M. Reinhart, 1990 "Private Investment and Economic Growth in Developing Countries," World Development, Vol. 18, No. 1. Krugman, Paul and Lance Taylor, 1978, "The Contractionary Effects of Devaluations," Journal of International Economics, 8:445-456. Pindyck, R., 1993, "Irreversibility, Uncertainty, and Investment," Striving for Growth After Adjustment: The Role of Capital Formation, L. Serven and A. Solimano (eds). The World Bank. Sarel, Michael, 1997, "Growth and Productivity in ASEAN Countries," IMF Working Paper No. 97/97. Senhadji, Abdelhak,1999, "Sources of Economic Growth: An Extensive Growth Accounting Exercise," IMFWorking Paper 99/77. 9 Section I Serven, Luis and Andres Solimano,1993, Private Investment and Macroeconomic Adjustment: A Survey;Striving for Growth after Adjustment. The Role of Capital Formation, L. Serven and A. Solimano (eds). The World Bank. Solow, Robert M.,1956, "A Contribution to the Theory of Economic Growth," Quarterly Journal of Economics, 70, 65-94. Soto, R., 1994,"A Macroeconomic Assessment of the Dominican Republic: Policies and Prospects," mimeo, The World Bank. Soto, R., 1997, "Non-Linearities in the Demand for Money: a Neural Network Approach," Working Paper Series No.107, ILADES-Georgetown University. World Bank, 1998, Mexico: Country Economic Assessment, mimeo. 10 Capital Accumulation, Total Factor Productivity and Growth ANNEX Economy wide data GDP is available in constant 1970 prices for the 1970-98 period. The capital stock is estimated using the perpetual inventory method, assuming that the ratio of capital to GDP was 2.5 in 1980, and the depreciation rate is 4 percent. Choosing a capital-output ratio in the range of 2.5 to 3 is customary in TFP models (see for example, Pindyck and Solimano, 1994; Sarel, 1997, and Braun and Braun, 1998). The selection of 1980 as a pivot for the estimation was based on the notion that excess capacity and macroeconomic imbalances were minimal at that time. Employment data are not available for all periods, with a gap from 1984 to 1991. The gap was filled by estimating: Log Employment, = 6.76 + 0.035 Trend, + 0.458 Log Employment,]l (139.2) (9.10) (2.69) R2 = 0.969 Breusch-Pagan Residual Correlation Test = 1.621 FTZ data Value added series in FTZs was obtained from the IMF for the 1983-98 period. For the 1970-82 period, total export figures were obtained from the study by Dahuajre et al., (1989) and it was assumed that value added was 30 percent of exports. This assumption is consistent with IMF data for the 1983-97 period, which estimates the ratio between 29 percent and 31 percent. Value added was deflated using the U.S. Producer Price Index to obtain real figures (valued in 1980 U.S. dollars). The capital stock was proxied using data on physical investment (area of buildings) from the National Council of Free-Trade Zones. The latter is used for estimating the stock of capital with a perpetual inventory method under three assumptions: (a) the stock of capital in 1984 was US$238 million (Dahuajre et al., 1989), (b) the average value of a square feet of construction and equipment in FTZs was US$80.3 in 1984, including machinery (Dahuajre et al., 1989), and (c) depreciation rates were 4 percent per year. Employment data are obtained directly from the National Council of Free-Trade Zones. 11 Section 1 Table A1.1 Data Sources Variables Description Source National accounts Official national accounts (base year Central Bank (BCRD) 1970) Private and public Breakdown of private and public ONAPLAN investment investment Price of capital goods Ratio of investment deflator to GDP BCRD deflator Nominal exchange rate Year average nominal exchange rate BCRD (in RD$ per U.S. dollar) CPI Consumer price index, base year 1984 BCRD Terms of trade Terms of trade index 1965-92 Soto (1994) 1993-97 World Bank Wages Real private wage index, base year 1965-95 World Bank 1984 1996-97 Secretaria de Trabajo Real exchange rate Proxied by eP*/CPI P* is the US CPI index (base year (RER) 1984) Capacity utilization Potential GDP estimated using a Own calculation nonlinear, quadratic model with intercept adjustments in 1980 along the Mizon and Hendry's (1994) methodology r* International interest rate U.S. 3-year t-bills, annualized Openness Exports in percent of GDP BCRD Value added in FTZ Net exports 1970-82 Soto (1994), 1983-97 IMF Labor in FTZ Total employment National Council of Free-Trade Zones Capital in FTZ Estimated using perpetual inventory National Council of Free-Trade Zones (see text) 12 Capital Accumulation, Total Factor Productivity and Growth Table A1.2 Data Used in Estimations and Simulations Free-Trade Zones Economy (Except FTZs) Value Capital stock GDP Capital Private Public ade(InadilliOns (In Employment (In st(oInk Employment investment investment Year (of 1984 thousands o (In millions mIllions o (In (In millions (In millions Uo.1 1984 U.S. thousands) of 1970 1970 thousands) of 1970 of 1970 dollars) dollars) pesos) pesos) pesos) pesos) 1970 4.0 8.0 0.5 1485.5 4682.8 1030.5 208.0 76.3 1971 8.0 21.0 1.0 1647.0 4779.4 1080.0 227.5 105.7 1972 12.7 32.2 2.3 1818.2 4882.7 1022.0 259.6 115.1 1973 12.4 55.3 3.2 2052.8 5046.1 1202.0 342.8 132.7 1974 19.3 68.1 4.8 2176.0 5297.8 1230.6 382.2 151.0 1975 26.5 81.1 7.0 2288.9 5598.2 1234.2 426.1 186.0 1976 33.6 94.5 8.6 2442.9 5935.3 1348.7 430.9 141.8 1977 42.9 98.7 10.9 2564.5 6242.4 1391.0 486.9 132.0 1978 49.3 100.1 13.5 2619.6 6611.6 1397.5 516.5 119.1 1979 53.6 92.5 16.1 2738.2 6982.8 1434.4 553.6 133.9 1980 52.2 83.9 18.3 2956.4 7391.0 1414.6 603.6 144.4 1981 62.0 81.1 20.5 3082.9 7843.4 1497.6 543.7 123.8 1982 64.4 83.5 19.6 3135.3 8197.1 1508.1 470.4 77.6 1983 120.5 92.7 22.3 3280.4 8417.2 1601.7 495.7 99.1 1984 99.1 118.4 27.1 3321.5 8675.3 1598.7 548.3 74.5 1985 85.2 156.4 35.7 3251.0 8951.0 1735.5 494.0 99.5 1986 174.2 270.9 51.2 3365.5 9186.5 1783.6 485.1 132.1 1987 188.2 295.7 66.0 3706.0 9436.2 1834.6 653.7 291.3 1988 239.8 360.6 83.8 3785.9 10003.8 1885.0 429.4 374.1 1989 336.2 424.3 122.9 3952.5 10407.1 1916.5 604.2 342.3 1990 332.5 586.7 130.0 3736.9 10937.4 1982.6 572.8 223.1 1991 423.0 630.8 135.5 3773.2 11295.8 2116.2 525.9 200.6 1992 514.4 706.2 141.1 4075.7 11570.4 2265.3 634.3 265.6 1993 665.1 725.6 164.3 4198.6 12005.6 2252.4 623.9 342.1 1994 704.5 733.1 176.3 4380.7 12593.3 2224.3 649.7 376.6 1995 761.9 774.2 165.6 4591.4 13162.1 2235.1 847.9 300.6 1996 793.4 762.8 164.6 4925.0 13755.0 2358.4 914.6 331.6 1997 974.7 780.0 182.2 5326.4 14430.4 2469.8 1055.5 244.8 1998 1080.8 877.0 195.2 5712.9 15310.1 2530.5 13 Section 1 Unit root tests for time series data The series were first tested for unit roots using standard augmented Dickey-Fuller and Phillips-Perron tests (see Hamilton, 1994 for a description). Since the series are rather short-27 observations-tests were supplemented with the simple first-order autocorrelation of the series to classify series as integrated or trend stationary. Table A1.3 Unit Root Tests First Order Variable in Levels ADF Autocorrelation ADF* PP* First Difference Value added in FTZ D.858 -1.55 -2.08 -5.94 Capital stock in FTZ 0.831 -0.80 2.40 -2.80 Employment in FTZ D.855 -2.11 3.12 4.24 DP rest of economy (ROE) D.887 -2.44 2.89 3.90 Capital stock ROE D.928 -2.61 -3.22 2.84 Employment in ROE 0.916 -1.04 1.32 -3.43 Private investment (in percent of GDP) D.577 -3.67 -3.69 ... ublic investment (in percent of GDP) 0.696 -2.87 2.99 Terms of trade 0.377 -3.53 -3.86 ... Real exchange rate 0.538 -2.97 -3.01 ... Real wages 0.746 -2.20 2.16 5.49 Alternative cost of capital 0.780 -1.34 1.20 4.78 Relative cost of capital goods 0.418 -4.12 3.12 ... Inflation 0.365 -3.79 3.81 ... TFP growth ROE 0.176 -4.12 4.08 ... Openness 0.675 -2.11 -2.08 5.98 Capacity utilization 0.169 -5.18 4.68 ... Public investment in infrastructure 0.707 -2.48 3.09 ... Public investment in social capital 0.772 -4.09** ... ... Public investment in housing and urba .863 -3.00** ... ... development I I Note: (*) ADF is the augmented Dickey-Fuller Test and PP is Phillips-Perron unit root test. Critical value is -2.62 at 90 percent. (**) The Perron structural break test was performed, allowing for ajump in the mean of the series in 1984. 14 Capital Accumulation, Total Factor Productivity and Growth Estimates of the production function Before estimating equations (2) and (3) (see main text), Granger causality tests were run between valued added and GDP and inputs. The results presented in Table A.3 show that for the FTZs it is unlikely that there are simultaneity biases arising from the estimation of the error-correction, cointegration models. Consequently, Engle-Granger estimation procedures can be undertaken without the disturbing presence of nuisance parameters. However, estimates for the variables representing the non-FTZ economy show that there is simultaneous causation between GDP and capital shocks. Simultaneity biases can be controlled, however, by including leads of capital shocks in the error-correction model, as discussed in Hargreaves (1994). Table A1.4 Causality Tests Test F-Test Test F-Test GDP does not Granger cause 1.5 Value added does not Granger cause 1.07 employment employment Employment does not Granger cause 2.92* Employment does not Granger cause 3.02* GDP value added GDP does not Granger cause capital 3.25* Value added does not Granger cause 0.36 capital Capital does not Granger cause GDP 5.23* Capital does not Granger cause value 0.35 added Endnotes 1 Percentage of population living in poverty conditions. 2 Equation (1) is derived under the assumption that the production function is Cobb-Douglas. Under constant returns to scale, the estimated parameters of inputs equation (1) are equal to the factor shares in production, while the estimated constant is a proxy (up to a scale) of the average productivity growth. Tests indicated that the null hypothesis of constant returns to scale in both the FTZs and the Economy could not be rejected and the estimation of (1) was carried out under the restriction that a I+ac2=1. 3 All results must be interpreted with due caution given that GDP data use a base year of 1970. 4 These shares are similar to those found by Senhadji (1999) (ranging from 0.52 to 0.72 for Latin American countries). 5 There are insufficient data on educational attainment to control for this variable. 6 Some public investment can crowd out private investment, reducing its net contribution. 7 Given past exchange controls and trade restrictions, the ratio of exports to GDP was seen as a better proxy for openness than the ratio of exports plus imports to GDP. 15 SECTION 2: DETERMINANTS OF POVERTY A. Introduction 1. The Dominican Republic enjoyed a strong economic growth during much of the last three decades, with an average annual growth of GDP of 7.7 percent in the 1970's, 3.8 percent in the 1980's, and more than 6 percent from 1992 to 1998. While there has been a significant growth in tourism, free trade industrial areas, and other service sectors, agriculture's share of GDP has declined steadily from 23 percent in 1970 to less than 13 percent in 1998. Agricultural share of employment has declined from 45 percent to less than half of that during the same period. Growth has not benefited all population groups equally, with poverty being a predominantly rural and agricultural phenomenon. If this process continues, and those migrating out of agriculture are the more educated and talented as observed in other countries, the relative position of the poor may even worsen in the coming years. 2. Policy has not been favorable to agriculture in past decades. Due to a combination of export taxation and industrial protection policies, the agricultural sector suffered rates of negative protection reaching as high as 40 percent. A mild trade liberalization process initiated in the 1990's and the elimination of export taxes may have improved the situation somewhat. However, strong economy-wide and sectoral distortions remain in effect. Agricultural products such as rice and beans, which are key ingredients in the diet of the population, are protected at rates of 60 percent to 80 percent through a combination of import quotas and input subsidies. This protection causes high costs to society as a whole and particularly to poor households. 3. This section analyzes the characteristics of the rural poor in the Dominican Republic, with an emphasis on the role of agriculture, using data from a 1992 Income and Expenditure Survey. Documenting the extent to which poor farmers and farm laborers rely on agricultural income, and the correlation between poverty incidence on one hand, and measures of human and physical capital and family size on the other hand, may provide a basis for anti-poverty government policy. 4. This analysis includes an examination of per-capita income distribution in the whole sample and among rural households comparing per-capita income and other characteristics of rural and urban households, landed and landless households, and farm and farm laborer households. It also includes various measures of poverty and extreme poverty; and a more explicit approach for evaluating the determinants of per-capita income, using a regression analysis. The results thus far indicate that two of the major variables correlated with poverty are education and type of employment. The final steps include taking a closer look at these variables. The first analyzes the determninants of earnings in different types of employment, after controlling for endogenous occupational choices. The second deals with educational mobility, i.e., the dependence of children's education on their parents' education. B. Poverty in the Dominican Republic 5. The 1992 Income and Expenditure survey includes about 1200 households. Besides detailed income and expenditure data, it includes demographic and socio-economic data such as age, education, and labor supply of all household members, detailed informiation on agricultural activities, and indices of living conditions. 16 Section 2 Examining the sample population 6. Table 2.1 compares household characteristics across per-capita income quintiles. It was first observed that family size is inversely related to per-capita income, declining from an average of 6.51 in the first quintile to 4.22 in the fifth quintile. This creates even larger differentials in per-capita income than in total household income. The largest relative difference in average per-capita income is between the fifth and the fourth quintiles, the former being more than three times larger than the latter. Seventy one percent of the first-quintile households reside in rural areas, relative to only 26 percent of the fifth- quintile households. This indicates that poverty is still very much a rural problem in the Dominican Republic. Poverty is also more of a problem among landed households, which constitute 46 percent of the first quintile and only 21 percent of the fifth quintile. The fact that the extent of land ownership varies less among the income quintiles than the extent of being rural indicates that within rural areas, land ownership improves the position of the household in terms of per-capita income. This point will be referred to again later in the text. The effect of being a female-headed household is not monotonic: these households are more abundant in the second and third quintiles than in the extremes. 7. Having indoor sanitation and running water is strongly correlated with per capita income. Indoor sanitation can be found in the homes of only 10 percent of lowest-quintile households, versus 68 percent of highest-quintile households. Only 7 percent of the households in the lowest income quintile have indoor running water, versus 57 percent of the households in the highest income quintile. Counting both indoor and outdoor running water, the trend still exists, although to a lower degree. 8. In terms of income sources, between 60 percent and 66 percent of households have income from employment, with very little variation across income quintiles. However, income from employment constitutes more than 40 percent of household income in the lowest quintile, but less than 30 percent of household income in the highest quintile. Having agricultural income is strongly negatively correlated with per capita income, with 50 percent of the lowest-quintile households having income from agriculture versus only 20 percent of highest-quintile households. The fraction of income derived from agriculture is also negatively correlated with per-capita income, but the differences among the quintiles are much smaller, indicating that farmers in higher quintiles derive larger portions of their income from agriculture. This point will be addressed again below. Higher-income households are more likely to derive income from self employment (58 percent in the highest quintile relative to 32 percent of the lowest quintile), and income from self employment constitutes a larger of their income (36 percent among highest-quintile households relative to 12 percent among lowest-quintile households). Households in the highest quintile are more likely to derive income from a family business (23 percent of them do so, versus less than 16 percent of the other households), and derive larger shares of their income from the business (11 percent, versus 3 percent in the first four quintiles). The likelihood of receiving income from pensions, insurance, or interest rises strongly with income, but the income shares of this source are not large. Lower-quintile households are more likely to receive remittances from within the country and less likely to receive remittances from abroad. The fraction of remittances in total household income is 7 percent in the sample as a whole, with larger fractions in the intermediate quintiles. 9. Looking at the characteristics of the household heads (Table 2.1), no systematic variation in the age distribution is observed among the different income quintiles. On the other hand, income seems to be correlated with schooling. In the lowest quintile, 35 percent of the household heads have no formal schooling, compared to 11 percent in the highest quintile. Only 15 percent of the household heads in the lowest quintile have more than six years of schooling, compared to 55 percent of the household heads in the highest quintile. Education, therefore, seems to be very important for income-class mobility. In 17 Determinants of Poverty terms of principal activity, household heads in lower income quintiles are more likely to be agricultural producers or farm laborers, and less likely to be self-employed. Rural versus urban households 10. Table 2.2 shows that about 47 percent of the sample households reside in rural areas. They have larger families than the urban households, and enjoy per capital income that is not much more than a half of that of urban households. Almost half of the rural households have access to land, while only 14 percent of urban households do. Landholdings per household is larger in the rural areas, but smaller fractions of the land are owned and titled. Twenty seven percent of urban households are headed by a female, while only 15 percent of rural households are. Only nine percent of rural households have indoor sanitation, compared to 62 percent of urban households. Only 10 percent of rural households have indoor running water, compared to 44 percent of urban households. 11. Urban households are more likely to have income from employment, self-employment, pensions, insurance and interest, and remittances from abroad. Rural households are more likely to have income from agriculture, family businesses, and internal remittances. The shares of the income sources vary accordingly, with urban households earning relatively more from family businesses and relatively less from external remittances. 12. The age distribution of household heads (Table 2.2) varies somewhat between urban and rural households, with 23 percent of urban household heads being over 60 years of age, relative to 18 percent of rural household heads. This could be related to different life expectancies in the two sectors. Differences in education exist as well, with 27 percent of the rural household heads having no formal schooling versus 12 percent in the urban sector, while almost half of urban household heads have more than six years of schooling, versus only 11 percent in the rural sector. As these differences are smaller than among the income quintiles in Table 2. 1, it can be concluded that schooling varies by income within sectors as well. Rural household heads are much more likely to be agricultural producers or farm laborers, while urban household heads are more likely to be employees or self employed. Income distribution among rural households 13. In Tables 2.3, the analysis of Tables 2.1 is repeated, this time for the rural households only. The variation in household size among the income quintiles is larger than in the whole sample, indicating a smaller variation among urban households. However, variation in per capita income is smaller than in the whole sample. Variations in the extent of having access to land are relatively small, but higher-income households seem to have larger fractions of their land owned and titled. The extent of female-headship does not vary monotonically among the income quintiles. While access to indoor sanitation and running water is positively correlated with per capita income, the correlation is not as strong as in the whole sample, indicating that the differences are predominantly between the sectors. 14. Only 40 percent of households in the highest income quintile have income from employment, compared to 60 percent of all other rural households. The tendency to have income from self- employment, pensions, insurance and interest, and remittances from abroad, rises with per capita income, while the opposite is true for the tendency to have income from agriculture and internal remittances. Income from employment constitutes a larger share of household income in lower income quintiles, while the opposite is true for income from self-employment. Income from agriculture constitutes a larger share of the income of households in the highest quintile than in the first four quintiles, which means that 18 Section 2 farmers in the highest quintile are much more successful in farming than all other farmers. This point will be addressed again below. 15. The age distribution varies somewhat among the heads of rural households in different income quintiles. Higher-income farmers seem to be older. The level of schooling is again positively correlated with per capita income, except for the highest quintile in which household heads have slightly less schooling than those in the second-highest quintile. Based on the principal activity of the household heads, it can be noted that self-employment becomes more common as per capita income rises. Farm laborers account for 30 percent of heads of households in the lowest quintile, compared to less than 15 percent in all other quintiles. This indicates that farm laborers tend to be poorer than the rest of the rural population. The likelihood of being an agricultural producer falls with income and then rises again in the highest quintile, and the opposite is true for the likelihood of being an employee. This strengthens the earlier conclusion that the highest quintile includes a group of very successful farm producers. Landed versus landless households 16. In Table 2.4 it can be seen that landed households have almost one family member more than landless households, in both the rural and urban sectors. This suggests that the child labor fertility motive may be an important factor in the Dominican Republic. Per capita income of landed and landless households is almost the same in the rural sector, but is larger among landless households in the urban sector. Female headship is much less common among landed households, in both sectors. Landed households are much less likely to have indoor sanitation or running water, in both sectors. 17. Landed households are much more likely to have income from agriculture and from internal remittances, and less likely to have income from employment, family business, pensions, insurance and interest. In the rural sector only, landed households are much less likely to have income from self- employment. The shares of the different sources in total household income behave similarly. 18. Table 2.4 shows that landed household heads are relatively older and less educated in both the rural and urban sectors. In terms of principal activity, landed households are much more likely to be agricultural producers, with only 4 percent agricultural producers among rural (and none among urban) landless household heads, and less likely to be not working, employees, farm laborers, or self employed. Farmers versus farm laborers 19. In Table 2.5, the characteristics of farm households and farm-laborer households are compared. Farm households are defined as households that have agricultural income. Farm-laborer households are defined as households that have no agricultural income but have farm-laborer income. Farm households have larger families and fewer female heads of households. About three-quarters of farm households have access to land. The others derive their farm income from sales of livestock and livestock products only. These patterns are not much different in rural and urban farm households. Land holdings, are however, somewhat larger in rural farm households. A little over 40 percent of the land is owned and titled in both rural and urban farm households. Several rural farm-laborer households also have access to land, but do not report any agricultural income. 20. Per capita income of rural farm households is almost 50 percent higher on average than that of rural farm-laborer households, but about 25 percent lower than that of urban farm households. Per capita income of urban farm-laborer households is higher than all the above, mainly because farm labor income constitutes only a quarter of total income in these households, while more than half of their income is 19 Determinants of Poverty derived from self employment. Note also that the sample includes only a small number of such households, so the averages are relatively inaccurate. In contrast, farm labor income constitutes 58 percent of the total income of rural farm-laborer households. Among farm households, rural households derive 32 percent of their income from the farm, compared to only 17 percent for urban households, and they derive more income from employment on other farms. Urban farm households derive much more of their income from non-agricultural employment and self-employment. Another thing to note is that per- capita income of rural farm households is similar to the average per-capita income of rural households as a whole (Table 2.3). 21. Table 2.5 compares the characteristics of the heads of households. Heads of farm households tend to be older and more educated than heads of farm-laborer households. In each group, heads of urban households are older and more educated than heads of rural households. Among the heads of farm households, agricultural production is the principal activity of 47 percent of the rural farmers and 35 percent of the urban farmers. Agricultural production is the secondary activity of 10 percent of the rural farmers and 18 percent of the urban farmers. Heads of farm households in rural areas are more likely to be farm laborers as a principal activity, while those in urban areas are more likely to be employees or self employed. This explains, at least in part, the difference in per capita income between the farm households in the two sectors. Among the heads of farm-laborer households, about 70 percent have farm labor as a major occupation. In rural heads of farm-laborer households 11 percent have farm labor as a major occupation, and 15 percent have self-employment. C. The Determinants of Per-capita Income 28. This section will examine the determinants of household income per capita using a multivariate regression analysis. The dependent variable will be expressed in log form, as well as the explanatory variable of landholdings per capita. Other explanatory variables at the household level include household size, and dummy variables for landed households, for livestock ownership, for having at least part of the land under title, and for female-headed households. Household head characteristics will be included as well, in particular age, education, and principal and secondary activities. Dummy variables for rural areas and for location (Santo Domingo region and Santiago region, as opposed to all other regions) will also be used. 29. The regression results are reported in Table 2.6. It can easily be seen that other things equal, rural households have 19 percent lower per-capita income than urban households, which confirms the results of the earlier sections. There is some regional variation as well, with households in the Santo Domingo area having higher per-capita income, but this effect is not significant in the rural sector. Land ownership increases per-capita income, and having a title to at least part of the land is associated with 21 percent higher per-capita income (18 percent in the rural sector). It seems like land titling is a good investment in the Dominican Republic, as is the case in Latin America (Lopez and Valdez 1999). Livestock ownership has a negative effect on per-capita income, but the effect is significant only among the rural population. Again, the causality here is questionable. It may be that livestock is, on average, a last resort adopted by households with poor alternatives. Household size has a negative and very significant effect on per-capita income. Each additional household member decreases per-capita income by nine percents. Female-headed households have lower per-capita income, but only among urban households. This does not mean, though that these households are worse off in all aspects (Rogers 1996). 30. Age of the head of household affects per-capita income in the urban sector only. Households whose head is over 40 years of age have higher per-capita income. Schooling of the head of household has a significant positive effect on household per-capita income, and the effect is stronger in the urban 20 Section 2 sector. Returns to higher education seem to be significant in the urban sector only. The type of activity of the household head also affects household per-capita income. Those who are self-employed, either as a primary or as a secondary activity, have a significant effect on income (the coefficient of secondary activity is not significant in the rural sample). Private employees also have higher per-capita incomes, but not in the rural sector. 31. These results indicate several policy-responsive variables that have a potential of raising per- capita household income. These include land titling, family size, schooling, and self-employment activities. The determinants of individual income 32. In this section, the determinants of earnings are explored at the individual level. The data set includes individual-level earnings from employment, self-employment, and farm labor. Naturally, it is preferable to differentiate between the different sources of income. However, the fact that individuals are free to choose between types of employment may result in selectivity bias. Hence, estimation of sector- specific earning functions must correct for selectivity. The two-stage method of Lee (1983) is used, in which an occupational choice model is estimated first, and its coefficients are used in a second stage to correct for selectivity in the estimation of earnings equations. This method has been applied to a similar problem by Verwimp (1998). Occupational choice 33. Primary activity of the head of household has been shown before to affect household income. In this section the determinants of primary activity of household members over 16 years of age are estimated. Except for using this analysis as a component of a selectivity-correction method, such an analysis i9 also relevant on its own for this study, since the exogenous factors which determine the primary activity affect household income indirectly. 34. The empirical model chosen for this purpose is the multinomial logit model, which is consistent with a utility-maximization behavior of the individuals. It is assumed that each individual chooses one out of a set of several discrete alternatives (in this case, primary activities). A measure of indirect utility is associated with each alternative, so the individual chooses the alternative that yields the highest indirect utility. Specifying each indirect utility as a linear function of explanatory variables, with coefficients that vary across alternatives (in principle the explanatory variables can vary as well), enables to estimate the coefficients subject to a normalization constraint. Typically, the coefficients of one activity are set to zero as a normalization constraint, so the interpretation of the estimated coefficients is of a difference between each actual coefficient and the corresponding normalized coefficient. The sign of each coefficient indicates the direction of the effect of the corresponding variable on the difference between the probabilities of choosing the relevant activity and the normalized activity. The coefficients can be subsequently used to calculate the size of the effect as well. 35. Preliminary estimations have shown that the coefficients of the activities of public employee and private employee are similar. Hence, these activities were aggregated. As a result, the model included four activities: agricultural producer, farm laborer, employee, and self-employed. The normalized activity included individuals that were not in the labor force or reported "other" primary activities. The only exception is the case of individuals who reported being out of the labor force as their primary activity, but reported a secondary activity in one of the above four activities. In these cases, the 21 Determinants of Poverty secondary activity was taken into account. After disposing of cases with missing values, the final sample included 3507 observations. 36. The results are reported in Table 2.7. It can be seen that heads of household are more likely to be in the labor force than all other family members. The only exception is that the tendency of spouses to work as farm laborers is not significantly different from that of household heads. In addition, females are significantly less likely to be in the labor force than males. Individuals living in rural areas are much more likely to choose a primary activity of agricultural producer or farm laborer, and less likely to choose a primary activity of employee or self-employed. Age also has an effect on occupational choice. Between 31 and 40 years of age, individuals are more likely to be in the labor force in general. Between the ages of 41 and 50, individuals are more likely to be agricultural producers or self employed. Between the ages of 51 and 60, the probability of being an agricultural producer is the highest and the probability of being an employee is the lowest. Over 61 years of age, people are more likely to be agricultural producers or out of the labor force. Schooling has a positive effect on being an employee or self employed and negative effect on being an agricultural producer or farm laborer. 37. Household income could not be used as a determinant of occupational choice since it is clearly endogenous. However, information was used on whether the household had income from pensions/insurance/interest or remittances. Having income from pensions increases the probability of being an employee and decreases the probability of being an agricultural producer. Having income from remittances from outside the Dominican Republic decreases the probability of being in the labor force. This confirms the ambiguous role of remittances in the development process (Baver 1995). 38. Family size has a positive effect on being a farm laborer and negative effect on being self- employed. Individuals in landed households are more likely to be agricultural producers, farm laborers, or self-employed (in this order). The probability of being an agricultural producer increases with the size of landholdings, while the probabilities of the other occupations decrease with landholdings size. The fraction of land owned and titled is negatively associated with being an agricultural producer or farm laborer. Having livestock is positively associated with being an agricultural producer. 39. The regional dummies included among the explanatory variables indicate that a person who resides in the Santo Domingo or Santiago regions has a significantly larger probability of being an employee and smaller probability of being a farm laborer. A person who resides in the Santiago area also has a significantly smaller probability of being self-employed. Earnings equations 40. In this section, separate earnings equations are estimated for individuals who are employees, self employed, or farm laborers. Income of agricultural producers was not reported on an individual basis. Each equation is- corrected for selectivity using the coefficients of the occupational choice model. The selectivity correction method is equivalent to the familiar Heckman approach, after an appropriate transformation to normality. 41. The results are reported in Table 2.8. Selectivity was found important only in the earnings equation of employees. The results show that earnings in rural areas are significantly lower for employees and farm laborers, while earnings of females is significantly lower for employees and self- employed. Earnings of employees peak between the ages of 51 and 60, while earnings of self-employed individuals is highest between the ages of 41 and 60. The earnings of farm laborers peak only after the age of 60. Schooling has a positive effect on earnings in all occupations, but the effect becomes 22 Section 2 statistically significant only with more than eight years of schooling. The effect of schooling is strongest for self-employed persons. Family size has a negative effect on the earnings of farm laborers. Earnings in all occupations are higher in the Santo Domingo region, and earnings of employees are also higher in the Santiago region. 42. These results can also be seen in Table 2.9, where annual income is simulated for the working population of the sample, conditional on type of employment, under various changes in the explanatory variables. The actual explanatory variables are used in the first line of the Table, and it can be seen that the income of farm laborers is less than one half of that of employees, and less than a third of that of the self employed. In the other simulations, the values of the explanatory variables were changed for the whole sample. For example, "all urban" is a simulation in which the rural residents were considered as urban as well as the urban residents. The results are not qualitatively different from those of Table 2.8, and as such will not be discussed, however they give a better idea as to the magnitudes of the differences. D. Education, Educational Mobility, and Poverty 43. The previous analysis has shown that lack of education is one of the major determinants of poverty. Education is an important component of human capital accumulation. Educational levels are determnined in early stages of life, but their benefits are realized throughout the life cycle. Therefore, as opposed to other determinants of poverty, policy cannot affect the educational attainments of the current generation to a large extent. However, it can affect educational attainments of the next generation. It is well known that education is one of the engines on intergenerational mobility across income classes. It is also known that opportunities to gain education are not independent of parents' income and wealth. As human capital cannot serve as collateral for loans, imperfect credit markets do not allow the poor to borrow in order to finance their children's education. 44. This issue will be explored in the context of the Dominican Republic. Specifically, the dependence of school attendance and educational attainment on parents' education, per-capita household income, and other household characteristics will be examined. Educational attainment is perhaps the most relevant variable of the two. However, the educational attainments of children who are still attending school are not known. For these, school attendance is an imperfect indicator of educational attainment. Hence, educational attainment will be examined for children who have clearly finished with formal schooling, while school attendance will be examined for school-age children. School attendance 45. Three quarters of children between the ages of 6 and 18 in the sample attend school. The correlation between school attendance and several household characteristics is examined in Table 2.10. It is first observed that school attendance reaches its maximum among 11-year old children, of which almost 92 percent of the children attend school. School attendance is positively correlated with father and mother's education. School attendance is higher among girls than among boys, on average. The correlation between school attendance and family size seems to be negative. The exception is the group of children from families with less than four members, among which school attendance is much below the hypothetical regression line. This may be due to an over-representation of single-parent families in that group. The correlation between school attendance and birth order is also negative. School attendance is lower in rural areas and in landed households. Having title to some land does not seem to make a difference. Regional variation is observed as well, with the highest school attendance being in the Santo Domingo area, and the second highest in the Santiago area. 23 Determinants of Poverty 46. School attendance is further examined using a probit analysis. The results appear in Table 2.11. One can think of the dependent variable in a probit model as being the probability that a certain event (in this case school attendance) occurs. Hence, the coefficients indicate the effects of the explanatory variables on this probability. However, the model is nonlinear and therefore the interpretation of the coefficients as partial derivatives does not hold. It is possible to use the estimated coefficients to calculate the partial derivatives. This has been done for the significant coefficients only, and the derivatives appear in the last column of Table 2.11. Since all the significant coefficients are related to dummy explanatory variables, the derivatives are calculated in the following way. The predicted probabilities were calculated twice with artificial changes in the explanatory variables, first making all observations belong to the relevant category (say, age 11-13) and second making all observations belong to the excluded category (in this case, age 6). The mean of these probabilities was taken across the whole sample, and the derivative is the difference between the mean probabilities. 47. It is first observed that females are 6 percent more likely to attend school than males, on average. The coefficients of the age dummies indicate that the probability of school attendance rises from age 6 to age 11 -1 3 and then declines up to age 18, where it comes down to about the same level. Both family size and birth order have negative effects on school attendance, but these effects are not statistically significant at the 5 percent level. School attendance is positively correlated with parents' education. The probability of attending school rises with mother's education up to secondary schooling, and is also higher when fathers have higher education. Note that father's and mother's education is strongly positively correlated in this sample. 48. Rural children are 12 percent less likely to attend school than urban children. Surprisingly, per- capita household income does not have any effect on school attendance, and neither does land holdings. However, in households that have title to at least some of their land, children's school attendance is 5 percent more likely. Household income sources also have insignificant effects on school attendance, except for households that receive remittances from outside the Dominican Republic. Children in these households are 5 percent more likely to attend school. School attendance probability is higher in the Santo Domingo and Santiago regions than in other regions, but only in the Santo Domingo region is it significantly higher, by 9 percent. Educational attainment 49. The point of interest is the determinants of educational attainment of children over 18 years of age. There is a focus on children who live with their parents since the dependence of their educational attainment on their parents' education is to be examined. By this those who have already formed their own separate household are excluded, and to the extent that household formation is not independent of educational attainment, our results may be subject to selectivity bias. Those who still attend school except if they already report higher schooling were excluded, since for those, educational attainment is not yet complete and they could reach higher educational levels in the future. 50. Five levels of educational attainment are identified in the survey: no schooling, primary schooling (1-6 years), intermediate schooling (7-8 years), secondary schooling (9-12 years), and higher schooling. Therefore, an ordered discrete response model was chosen for the analysis. Specifically, an ordered probit model was estimated. The thresholds between the alternatives were allowed to be estimated endogenously rather than using the school years as exogenous thresholds. The results are reported in Table 2.12. The coefficients can be interpreted as the effects of the explanatory variables on the latent endogenous variable that may be thought of as years of schooling. However, they are normalized by the unidentified standard deviation of this latent variable. 24 Section 2 51. In Table 2.13, the changes in the probabilities of the different educational levels were simulated with a change in a single explanatory variable. In the case of the continuous explanatory variables, this is done by changing the value of the variable for all sample observations (by one unit for family size and birth order and by 10 percent for income per capita). In the case of the discrete explanatory variables that enter the estimated equation as a set of binary variables, the probabilities were simulated several times, where all sample observations are assigned to the same category each time. For example, in the gender variable, the probabilities were calculated once where all observations are treated as males and once where all observations are treated as females. The simulations in Table 2.13 are performed only for the explanatory variables whose coefficients are found in Table 2.12 to be statistically significant. 52. The results of Tables 2.12 and 2.13 are qualitatively equivalent. They show that educational attainment is higher among females, as was the case for school attendance for school-age girls. Educational attainment rises with age up to the age group of 26-27 years, and then falls. While this can be due to a certain profile of cohort effects, it may also be a result of selectivity. The initial rise is consistent with the likelihood of staying in the parents' household being positively correlated with late school attendance. The subsequent fall is consistent with the likelihood of staying in the parents' household being positively correlated with low ability or earning potential. 53. Educational attainment is negatively related to family size, as opposed to the findings of Al- Samarrai and Peasgood (1998), and positively related to birth order, as opposed to the findings of Kaneda (1998). It is strongly positively related to both father's and mother's education. Education is lower in rural areas, and is not significantly related to landholdings. Although neither is significant, the coefficient of titled land is higher than that of untitled land, as in NaRanong (1998). Educational attainment is positively related to household per-capita income. Surprisingly, it is higher in households which derive income from agriculture, other things equal. It is also higher in the Santo Domingo region than in all other regions. 54. Both school attendance and educational attainment results show that education is significantly lower in rural areas and outside of the Santo Domingo region, other things equal. This should draw the attention of policy makers. As schooling is positively correlated across generations, investments in schooling have social benefits that may be larger than the private benefits due to this externality. Family planning policies that reduce family size may also contribute to higher educational attainments. E. Main Findings 55. This analysis has shown that poverty in the Dominican Republic is mainly a rural problem. Among the rural population, the most vulnerable are households that rely on agriculture for their livelihood, in particular farm laborer households. Several factors that are associated with higher household income are identified, including schooling, land titling, and self-employment. While land titling can be considered exogenous, schooling and choice of occupation can not. Hence, the determinants of occupational choice are analyzed individually for all adult members of the household, and the determinants of individual income within each type of employment are studied. The results show that rural residents and uneducated individuals enjoy lower earned incomes conditional on type of employment, and also tend to choose the lower-paying occupations such as farm labor. Females earn less than males as employees or self employed, if they choose these types of employment. 56. The determinants of schooling are studied by examining the determinants of school attendance for school-age children and the determinants of educational attainment for older children within the household. Schooling is found to be lower in rural areas but higher for females. It is also positively 25 Determninants of Poverty related to parents' education. School attendance is also higher in households having title to at least some of their land, and among households that receive remittances from outside the DR. Educational attainment is also positively related to per-capita income and birth order, and negatively related to family size. 57. From a policy perspective, these results imply that the fight against poverty should be focused on rural areas, with an emphasis on schooling, job creation, and land titling. 26 Section 2 TABLES AND FIGURES Table 2.1 Distribution of household characteristics across income quintiles All households 1st quintile 2nd quintile 3rd quintile 4th quintile 5th quintile Number of households 1182 236 237 236 237 236 Number of household members 5.15 6.51 5.47 5.06 4.51 4.22 Rural resident 0.47 0.71 0.57 0.44 0.38 0.26 Landed household 0.30 0.46 0.36 0.28 0.22 0.21 Female head of household 0.22 0.19 0.23 0.26 0.20 0.20 Sanitation and water Indoor sanitation 0.36 0.10 0.23 0.36 0.46 0.68 Indoor running water 0.28 0.07 0.19 0.25 0.31 0.57 Outdoor running water 0.30 0.26 0.35 0.32 0.37 0.21 Incidence of source of household income Has income from employment 0.62 0.62 0.60 0.66 0.62 0.60 Has income from self employment 0.45 0.32 0.39 0.48 0.47 0.58 Has income from agriculture 0.32 0.50 0.35 0.31 0.22 0.20 Has income from family business 0.16 0.16 0.15 0.16 0.12 0.23 Has income from pensions, insurance, interest 0.31 0.15 0.25 0.32 0.34 0.49 Has income from remittances from within the DR 0.21 0.33 0.26 0.18 0.20 0.10 Has income from remittances from outside the DR 0.31 0.17 0.28 0.30 0.39 0.39 Share of source of household income Income from employment 0.32 0.42 0.36 0.36 0.32 0.29 Income from self employment 0.30 0.12 0.20 0.23 0.27 0.36 Income from agriculture 0.07 0.12 0.09 0.08 0.06 0.07 Income from a family business 0.07 0.03 0.03 0.04 0.03 0.11 Income from pensions, insurance, interest 0.04 0.01 0.02 0.03 0.03 0.05 Income from remittances from within the DR 0.01 0.05 0.04 0.02 0.01 0.00 Income from remittances from outside the DR 0.06 0.02 0.06 0.05 0.10 0.05 Age of head of household Head of household under 30 years 0.15 0.09 0.16 0.18 0.17 0.15 Head of household between 31 and 40 0.23 0.28 0.22 0.20 0.27 0.21 Head of household between 41and 50 0.22 0.22 0.22 0.21 0.22 0.24 Head of household between 51 and 60 0.19 0.22 0.17 0.22 0.14 0.20 Head of household 61 and up 0.20 0.18 0.23 0.20 0.20 0.21 Schooling of head of household No schooling 0.19 0.35 0.22 0.15 0.13 0.11 1-6 years of schooling 0.47 0.50 0.56 0.50 0.45 0.35 7-8 years of schooling 0.11 0.05 0.09 0.14 0.13 0.14 9-12 years of schooling 0.14 0.08 0.09 0.17 0.18 0.16 More than 12 years of schooling 0.09 0.02 0.04 0.03 0.11 0.25 Principal activity of head of household Notworking outside the house 0.16 0.14 0.18 0.20 0.17 0.11 Principal activity is agricultural producer 0.17 0.28 0.19 0.16 0.12 0.09 Principal activity is public employee 0.08 0.07 0.10 0.11 0.09 0.06 Principal activity is private employee 0.18 0.11 0.16 0.21 0.19 0.21 Principal activity is famm laborer 0.08 0.21 0.07 0.03 0.07 0.03 Principal activity is self employment 0.27 0.13 0.22 0.23 0.31 0.44 a. Income quintiles are defined according to total household income (including value of consumption from own sources) per capita. 27 Determinants of Poverty Table 2.2 Comparing rural and urban households All households Rural Urban Number of households 1182 560 622 Number of household members 5.15 5.30 5.02 Landed household 0.30 0.49 0.14 Size of landholdings 88.06 91.69 76.86 Amount of land owned and titled 38.29 37.85 39.66 Female head of household 0.22 0.15 0.27 Sanitation and water Indoor sanitation 0.36 0.09 0.62 Indoor running water 0.28 0.10 0.44 Outdoor running water 0.30 0.26 0.34 Incidence of source of household income Has income from employment 0.62 0.56 0.67 Has income from self employment 0.45 0.38 0.51 Has income from agriculture 0.32 0.55 0.12 Has income from family business 0.16 0.20 0.13 Has income from pensions, insurance, interest 0.31 0.22 0.39 Has income from remittances from within the DR 0.21 0.26 0.17 Has income from remittances from outside the DR 0.31 0.21 0.39 Share of source of household income Income from employment 0.32 0.24 0.35 Income from self employment 0.30 0.22 0.33 Income from agriculture 0.07 0.18 0.02 Income from a family business 0.07 0.08 0.07 Income from pensions, insurance, interest 0.04 0.01 0.05 Income from remittances from within the DR 0.01 0.02 0.01 Income from remittances from outside the DR 0.06 0.05 0.06 Age of head of household Head of household under 30 years 0.15 0.15 0.15 Head of household between 31 and 40 0.23 0.23 0.24 Head of household between 41 and 50 0.22 0.22 0.22 Head of household between 51 and 60 0.19 0.22 0.16 Head of household 61 and up 0.20 0.18 0.23 Schooling of head of household No schooling 0.19 0.27 0.12 1-6 years of schooling 0.47 0.55 0.40 7-8 years of schooling 0.11 0.08 0.15 9-12 years of schooling 0.14 0.08 0.18 More than 12 years of schooling 0.09 0.03 0.15 Principal activity of head of household Not working outside the house 0.16 0.12 0.19 Principal activity is agricultural producer 0.17 0.29 0.06 Principal activity is public employee 0.08 0.05 0.11 Principal activity is private employee 0.18 0.13 0.22 Principal activity is farm laborer 0.08 0.15 0.02 Principal activity is self employment 0.27 0.21 0.32 28 Section 2 Table 2.3 Distribution of household characteristics across income quintiles, rural households All households 1st quintile 2nd quintile 3rd quintile 4th quintile 5th quintile Number of households 560 112 112 112 112 112 Number of household members 5.30 6.82 5.89 5.08 4.71 4.01 Landed household 0.49 0.54 0.55 0.44 0.41 0.48 Fraction of land owned and titled 0.41 0.31 0.45 0.46 0.26 0.52 Female head of household 0.15 0.13 0.14 0.19 0.16 0.15 Sanitation and water Indoor sanitation 0.09 0.03 0.04 0.09 0.10 0.18 Indoor running water 0.10 0.03 0.06 0.11 0.08 0.20 Outdoor running water 0.26 0.18 0.23 0.24 0.35 0.29 Incidence of source of household income Has income from employment 0.56 0.62 0.60 0.59 0.62 0.40 Has income from self employment 0.38 0.22 0.38 0.35 0.42 0.50 Has income from agriculture 0.55 0.66 0.59 0.48 0.50 0.50 Has income from family business 0.20 0.20 0.19 0.19 0.20 0.24 Has income from pensions, insurance, interest 0.22 0.12 0.19 0.19 0.25 0.36 Has income from remittances from within the DR 0.26 0.32 0.28 0.26 0.21 0.21 Has income from remittances from outside the DR 0.21 0.12 0.19 0.22 0.22 0.30 Share of source of household income Income from employment 0.24 0.44 0.35 0.35 0.32 0.11 Income from self employment 0.22 0.07 0.16 0.17 0.21 0.27 Income from agriculture 0.18 0.15 0.14 0.12 0.13 0.25 Income from a family business 0.08 0.05 0.03 0.04 0.06 0.11 Income from pensions, insurance, interest 0.01 0.01 0.01 0.01 0.02 0.01 Income from remittances from within the DR 0.02 0.04 0.06 0.02 0.02 0.00 Income from remittances from outside the DR 0.05 0.01 0.02 0.05 0.03 0.07 Age of head of household Head of household under 30 years 0.15 0.11 0.15 0.15 0.18 0.14 Head of household between 31 and 40 0.23 0.29 0.24 0.22 0.21 0.17 Head of household between 41 and 50 0.22 0.21 0.22 0.28 0.18 0.23 Head of household between 51 and 60 0.22 0.23 0.25 0.15 0.23 0.25 Head of household 61 and up 0.18 0.16 0.13 0.20 0.20 0.22 Schooling of head of household No schooling 0.27 0.41 0.30 0.25 0.17 0.23 1-6 years of schooling 0.55 0.50 0.52 0.57 0.58 0.57 7-8 years of schooling 0.08 0.05 0.05 0.09 0.08 0.11 9-12 years of schooling 0.08 0.05 0.08 0.08 0.13 0.06 More than 12 years of schooling 0.03 0.00 0.05 0.01 0.04 0.04 Principal activity of head of household Not working outside the house 0.12 0.09 0.10 0.19 0.15 0.09 Principal activity is agricultural producer 0.29 0.33 0.32 0.24 0.24 0.29 Principal activity is public employee 0.05 0.04 0.05 0.10 0.05 0.02 Principal activity is private employee 0.13 0.08 0.12 0.17 0.16 0.13 Principal activity is farm laborer 0.15 0.30 0.15 0.08 0.11 0.12 Principal activity is self employment 0.21 0.11 0.19 0.20 0.23 0.35 29 Detenninants of Poverty Table 2.4 Comparing landed and landless households All households Rural Urban Landed Landless Landed Landless Landed Landless Number of households 360 822 272 288 88 534 Number of household members 5.74 4.89 5.76 4.87 5.69 4.91 Female head of household 0.07 0.28 0.07 0.23 0.07 0.30 Sanitation and water Indoor sanitation 0.12 0.47 0.05 0.12 0.34 0.66 Indoor running water 0.12 0.35 0.06 0.13 0.28 0.47 Outdoor running water 0.26 0.32 0.22 0.30 0.40 0.33 Incidence of source of household income Has income from employment 0.51 0.66 0.51 0.61 0.53 0.69 Has income from self employment 0.35 0.49 0.29 0.45 0.54 0.51 Has income from agriculture 0.88 0.10 0.91 0.24 0.78 0.03 Has income from family business 0.20 0.15 0.20 0.20 0.18 0.13 Has income from pensions, insurance, interest 0.18 0.36 0.16 0.27 0.24 0.41 Has income from remiHtances from within the DR 0.29 0.18 0.29 0.22 0.26 0.16 Has income from remittances from outside the DR 0.26 0.33 0.21 0.21 0.40 0.39 Share of source of household income Income from employment 0.19 0.36 0.16 0.32 0.25 0.37 Income from self employment 0.21 0.32 0.13 0.30 0.38 0.33 Income from agriculture 0.28 0.01 0.35 0.02 0.12 0.01 Income from a family business 0.04 0.08 0.05 0.10 0.03 0.08 Income from pensions, insurance, interest 0.01 0.04 0.01 0.02 0.01 0.05 Income from remiHtances from within the DR 0.02 0.01 0.02 0.02 0.01 0.01 Income from remiHtances from outside the DR 0.07 0.05 0.06 0.03 0.08 0.06 Age of head of household Head of household under 30 years 0.09 0.18 0.10 0.19 0.05 0.17 Head of household between 31 and 40 0.19 0.26 0.18 0.27 0.20 0.25 Head of household between 41 and 50 0.24 0.21 0.23 0.21 0.25 0.21 Head of household between 51 and 60 0.25 0.16 0.25 0.19 0.25 0.15 Head of household 61 and up 0.24 0.19 0.23 0.14 0.26 0.22 Schooling of head of household No schooling 0.27 0.16 0.28 0.26 0.25 0.10 1-6 years of schooling 0.56 0.43 0.60 0.50 0.42 0.40 7-8 years of schooling 0.07 0.13 0.06 0.09 0.12 0.15 9-12 years of schooling 0.08 0.16 0.05 0.11 0.15 0.19 More than 12 years of schooling 0.02 0.12 0.01 0.04 0.06 0.16 Principal activity of head of household Not working outside the house 0.06 0.20 0.06 0.19 0.08 0.21 Principal activity is agricultural producer 0.52 0.02 0.54 0.04 0.42 0.00 Principal activity is public employee 0.06 0.09 0.04 0.06 0.13 0.11 Principal activity is private employee 0.06 0.23 0.06 0.20 0.06 0.24 Principal activity is farm laborer 0.10 0.08 0.13 0.18 0.01 0.02 Principal activity is self employment 0.16 0.32 0.13 0.29 0.24 0.33 30 Section 2 Table 2.5 Comparing farm and farm-laborer households Farm households Farm-laborer households All Rural Urban All Rural Urban Number of households 349 277 72 65 54 11 Number of household members 5.73 5.71 5.79 5.08 5.00 5.45 Female head of household 0.11 0.10 0.12 0.15 0.15 0.18 Land holdings Household with land 0.77 0.76 0.78 0.06 0.07 0.00 Size of landholdings 71.29 73.62 62.35 1.89 2.28 0.00 Amount of land owned and titled 29.03 29.57 26.93 0.02 0.02 0.00 Incidence of source of household income Has income from employment 0.51 0.50 0.56 1.00 1.00 1.00 Has income from self employment 0.36 0.32 0.49 0.32 0.30 0.45 Has income from family business 0.21 0.22 0.18 0.11 0.11 0.09 Has income from pensions, insurance, interest 0.17 0.14 0.25 0.20 0.19 0.27 Has income from remittances from within the DR 0.27 0.29 0.18 0.23 0.22 0.27 Has income from remittances from outside the DR 0.25 0.21 0.40 0.09 0.09 0.09 Share of source of household income Income from farm employment 0.10 0.12 0.04 0.47 0.58 0.25 Income from other employment 0.21 0.17 0.33 0.56 0.68 0.35 Income from self employment 0.20 0.17 0.27 0.25 0.11 0.53 Income from agriculture 0.28 0.32 0.17 0.00 0.00 0.00 Income from a family business 0.04 0.05 0.03 0.01 0.02 0.00 Income from pensions, insurance, interest 0.01 0.01 0.01 0.02 0.01 0.05 Income from remittances from within the DR 0.02 0.02 0.01 0.01 0.01 0.00 Income from remittances from outside the DR 0.06 0.05 0.07 0.02 0.01 0.03 Age of head of household Head of household under 30 years 0.10 0.10 0.07 0.13 0.13 0.10 Head of household between 31 and 40 0.21 0.21 0.21 0.33 0.36 0.20 Head of household between 41and 50 0.23 0.24 0.21 0.24 0.23 0.30 Head of household between 51 and 60 0.24 0.23 0.25 0.19 0.19 0.20 Head of household 61 and up 0.22 0.21 0.26 0.11 0.09 0.20 Schooling of head of household No schooling 0.27 0.29 0.21 0.39 0.40 0.36 1-6 years of schooling 0.54 0.57 0.41 0.53 0.57 0.36 7-8 years of schooling 0.08 0.07 0.16 0.05 0.02 0.18 9-12 years of schooling 0.08 0.06 0.15 0.02 0.02 0.00 More than 12 years of schooling 0.03 0.01 0.07 0.02 0.00 0.09 Principal activity of head of household Not working outside the house 0.08 0.08 0.09 0.14 0.13 0.18 Principal activity is agricultural producer 0.45 0.47 0.35 0.03 0.04 0.00 Principal activity is public employee 0.06 0.04 0.15 0.00 0.00 0.00 Principal activity is private employee 0.08 0.07 0.10 0.06 0.07 0.00 Principal activity is farm laborer 0.13 0.15 0.03 0.71 0.70 0.73 Principal activity is self employment 0.18 0.16 0.24 0.06 0.06 0.09 Secondary activity of head of household Secondary activity is household work 0.05 0.04 0.10 0.05 0.02 0.18 Secondary activity is agricultural producer 0.12 0.10 0.18 0.00 0.00 0.00 Secondary activity is public employee 0.01 0.01 0.01 0.00 0.00 0.00 Secondary activity is private employee 0.02 0.02 0.01 0.02 0.00 0.09 Secondary activity is farm laborer 0.06 0.06 0.04 0.09 0.11 0.00 Secondary activity is self employment 0.11 0.11 0.10 0.12 0.15 0.00 31 Determinants of Poverty Table 2.6 Per-capita income regression results All sample households Rural households only Explanatory variable Coeff. t-ratio Signif; Coeff. t-ratio Signif. Constant 6.69 55.40 0.00 6.70 41.15 0.00 Rural resident -0.19 -3.47 0.00 Santo Domingo area 0.30 5.09 0.00 0.13 1.22 0.22 Santiago area 0.15 1.78 0.08 -0.02 -0.17 0.87 Household with land -0.13 -1.57 0.12 -0.14 -1.51 0.13 In(landholdings per capita) 0.11 4.08 0.00 0.15 4.89 0.00 Has title to some land 0.21 2.63 0.01 0.18 2.04 0.04 Has livestock -0.06 -1.00 0.32 -0.19 -2.31 0.02 Missing livestock data 0.01 0.16 0.87 -0.04 -0.37 0.71 Number of household members -0.09 -9.60 0.00 -0.09 -7.14 0.00 Female head of household -0.13 -2.17 0.03 -0.01 -0.06 0.96 Head of household between 31 and 40 0.01 0.15 0.88 -0.04 -0.35 0.73 Head of household between 41and 50 0.22 2.98 0.00 0.13 1.18 0.24 Head of household between 51 and 60 0.23 2.89 0.00 0.09 0.85 0.40 Head of household 61 and up 0.22 2.73 0.01 0.17 1.42 0.16 1-6 years of schooling 0.17 2.87 0.00 0.18 2.48 0.01 7-8 years of schooling 0.40 4.69 0.00 0.32 2.50 0.01 9-12 years of schooling 0.40 4.75 0.00 0.18 1.37 0.17 More than 12 years of schooling 0.90 9.35 0.00 0.31 1.50 0.13 Principal activity is agricultural producer -0.09 -0.95 0.34 -0.10 -0.84 0.40 Principal activity is public employee -0.13 -1.38 0.17 -0.16 -0.94 0.35 Principal activity is private employee 0.16 2.00 0.05 0.14 1.04 0.30 Principal activity is farm laborer -0.07 -0.70 0.48 -0.09 -0.72 0.47 Principal activity is self employment 0.35 4.95 0.00 0.35 3.20 0.00 Secondary activity is agricultural producer -0.01 -0.12 0.90 0.06 0.43 0.67 Secondary activity is public employee 0.21 0.97 0.33 0.17 0.51 0.61 Secondary activity is private employee 0.21 1.58 0.11 0.32 1.33 0.18 Secondary activity is farm laborer -0.11 -0.86 0.39 -0.16 -1.16 0.25 Secondary activity is self employment 0.23 2.96 0.00 0.19 1.84 0.07 R-squared 0.34 0.25 F-statistic 20.96 6.60 Number of observations 1182 560 32 Table 2.7 Primary activity multinomial logit results Agricultural producer Farm laborer Employee Self employed Explanatory variable Coeff. t-ratio Signif. Coeff. t-ratio Signif. Coeff. t-ratio Signif. Coeff. t-ratio Signif. Constant -2.24 -4.46 0.00 -0.77 -1.76 0.04 -0.59 -2.15 0.02 1.01 3.94 0.00 Spouse of head -2.39 -4.28 0.00 -0.66 -1.08 0.14 -1.36 -7.75 0.00 -0.79 -4.64 0.00 Child of head -2.65 -7.90 0.00 -1.23 -4.17 0.00 -1.28 -7.45 0.00 -1.62 -8.82 0.00 Other relation to head -1.77 -5.51 0.00 -1.03 -3.15 0.00 -1.41 -7.77 0.00 -1.64 -8.31 0.00 Rural resident 0.71 3.02 0.00 1.36 5.15 0.00 -0.41 -3.31 0.00 -0.37 -2.91 0.00 Female -3.27 -7.94 0.00 -3.83 -7.59 0.00 -1.30 -10.99 0.00 -1.43 -10.51 0.00 Age 31-40 0.65 2.07 0.02 0.65 2.40 0.01 0.48 3.36 0.00 0.66 4.38 0.00 Age 41-50 0.62 1.78 0.04 0.09 0.26 0.40 0.20 1.06 0.15 0.68 3.86 0.00 Age 51-60 0.59 1.67 0.05 -0.13 -0.37 0.36 -0.67 -3.07 0.00 0.09 0.45 0.33 Age 61 and up -0.39 -1.09 0.14 -1.10 -2.94 0.00 -2.49 -9.12 0.00 -1.12 -5.12 0.00 1-6 years of schooling -0.04 -0.19 0.43 -0.18 -0.79 0.22 0.55 2.78 0.00 0.21 1.31 0.09 7-8 years of schooling -0.08 -0.25 0.40 -0.42 -1.33 0.09 0.67 3.01 0.00 0.51 2.60 0.00 9+ years of schooling -0.90 -2.50 0.01 -1.67 -4.23 0.00 1.22 5.94 0.00 0.48 2.64 0.00 Has income from pensions/interest -0.64 -2.34 0.01 -0.20 -0.80 0.21 0.90 8.57 0.00 -0.14 -1.13 0.13 Has income from internal remittances 0.13 0.59 0.28 0.28 1.30 0.10 0.05 0.35 0.36 -0.11 -0.78 0.22 Has income from external remittances -0.38 -1.62 0.05 -0.41 -1.70 0.04 -0.17 -1.61 0.05 -0.16 -1.35 0.09 Family size 0.34 0.95 0.17 0.81 2.35 0.01 0.12 0.60 0.28 -0.40 -1.88 0.03 Household with land 2.44 6.10 0.00 1.18 3.11 0.00 0.21 0.72 0.24 0.63 2.30 0.01 ln(land) 0.16 2.00 0.02 -0.39 -3.74 0.00 -0.17 -2.31 0.01 -0.24 -3.26 0.00 Fraction of land titled -0.43 -1.77 0.04 -0.97 -2.90 0.00 0.31 1.31 0.09 -0.25 -1.08 0.14 Has livestock 0.50 1.83 0.03 0.29 1.22 0.11 0.02 0.12 0.45 -0.02 -0.16 0.44 Missing livestock data -0.50 -0.97 0.17 -1.64 -4.01 0.00 -0.16 -1.32 0.09 -0.14 -1.06 0.14 Santo Domingo area -0.69 -1.55 0.06 -0.83 -2.31 0.01 0.47 3.70 0.00 0.12 0.93 0.18 Santiago area 0.15 0.38 0.35 -1.12 -2.71 0.00 0.32 1.81 0.03 -0.44 -2.15 0.02 33 Table 2.8 Individual income regression results Employee Self employed Farm laborer Explanatory variable Coeff. t-ratio Signif. Coeff. t-ratio Signif. Coeff. t-ratio Signif. Constant 23.20 2.46 0.01 27.76 1.85 0.07 19.60 4.53 0.00 Rural resident -5.87 -2.05 0.04 -6.15 -1.00 0.32 -5.18 -2.37 0.02 Female -6.30 -2.15 0.03 -30.33 -5.04 0.00 0.44 0.15 0.88 Age 31-40 12.72 4.54 0.00 13.89 1.89 0.06 1.73 1.06 0.29 Age 41-50 15.41 4.45 0.00 29.81 3.79 0.00 1.52 0.81 0.42 Age 51-60 18.42 3.92 0.00 30.41 3.34 0.00 3.36 1.61 0.11 Age 61 and up 13.27 2.05 0.04 22.19 2.21 0.03 4.60 1.86 0.06 1-6 years of schooling 2.81 0.53 0.60 3.70 0.46 0.64 1.67 1.10 0.27 7-8 years of schooling 7.07 1.18 0.24 14.32 1.41 0.16 2.08 0.85 0.40 9+ years of schooling 13.24 2.29 0.02 28.87 3.24 0.00 10.52 3.04 0.00 Family size -4.72 -1.04 0.30 -1.96 -0.16 0.87 -7.20 -3.03 0.00 Santo Domingo area 8.52 3.12 0.00 17.68 2.93 0.00 9.35 3.09 0.00 Santiago area 10.30 2.60 0.01 3.79 0.37 0.71 -1.43 -0.45 0.66 Selection term -8.22 -1.95 0.05 -8.45 -0.87 0.39 -1.33 -0.68 0.49 R-squared 0.15 0.14 0.17 F statistic 9.88 8.23 3.18 Number of observations 756 652 2151 Table 2.9 Simulations of individual annual income by type of employment ________________________________ | Self employed Employee Farm laborer Actual data 37635.72 26026.28 12070.43 All urban 39937.87 27725.97 16669.73 All rural 33786.21 21858.56 11492.51 Gender All females 18611.24 22116.99 12465.44 All males 48938.77 28418.54 12027.67 Age All age under 31 22518.70 18871.32 10634.21 All age 31-40 36411.56 31593.83 12363.20 All age 41-50 52332.79 34284.94 12150.33 All age 51-60 52932.37 37290.80 13996.65 All age 61 44707.35 32137.79 15238.55 Education All no schooling 24835.41 17433.25 10410.94 All primary schooling 28533.16 20247.84 12077.17 All intermediate schooling 39152.64 24498.39 12495.50 All higher schooling 53705.85 30673.25 20926.24 Region All Santo Domingo region 48445.88 29332.83 21042.78 All Santiago region 34548.59 31116.19 10267.97 All other regions 30762.51 20813.36 11695.42 33A Determinants of Poverty Table 2.10 School attendance of children 6 to 18 years of age Population group attending not attending total % attending All 1087 360 1447 75.12 Age 6 52 60 112 46.43 7 84 32 116 72.41 8 86 19 105 81.90 9 81 16 97 83.51 10 107 23 130 82.31 11 91 8 99 91.92 12 125 17 142 88.03 13 99 15 114 86.84 14 89 25 114 78.07 15 81 21 102 79.41 16 80 35 115 69.57 17 46 33 79 58.23 18 66 56 122 54.10 Father's education None 131 90 221 59.28 Primary (1-6 years) 478 180 658 72.68 Intermediate (7-8) 118 20 138 85.51 Secondary (9-12) 115 18 133 86.47 Superior 77 5 82 93.90 Other 9 1 10 90.00 Mother's education None 147 97 244 60.25 Primary (1-6 years) 498 201 699 71.24 Intermediate (7-8) 145 23 168 86.31 Secondary (9-12) 177 15 192 92.19 Superior 64 5 69 92.75 Other 11 2 13 84.92 Gender Male 515 214 729 70.64 Female 569 146 715 79.58 Family size 2-3 47 23 70 67.14 4 153 27 180 85.00 5 197 39 236 83.47 6 233 82 315 73.97 7 195 40 235 82.98 8 111 63 174 63.79 9 68 25 93 73.12 10 34 20 54 62.96 11 24 25 49 48.98 12 and up 25 16 41 60.98 Birth order 1 (oldest) 330 95 425 77.65 2 295 80 375 78.67 3 215 75 290 74.14 4 127 46 173 73.41 5 65 24 89 73.03 6 36 16 52 69.23 7 and up 19 24 43 44.19 Rurallurban Rural 538 270 808 66.58 Urban 540 89 629 85.85 Landholding Has land 405 192 597 67.84 Has no land 673 167 840 80.12 Land titling Has title to some land 175 57 232 75.43 Does not have title 903 302 1205 74.94 Geographic location Santo-Domingo area 313 45 358| 87.43 Santiago area 69 16 85 81.18 Other area 705 299 1004 70.22 34 Section 2 Table 2.11 Probit results of school attendance, children ages 6-18 Parameters Estimate t-value Signific; Derivative Intercept -0.28 -1.07 0.14 Female 0.25 2.89 0.00 0.06 Age 7 0.71 3.62 0.00 0.23 Age 8-10 1.14 6.80 0.00 0.33 Age 11-13 1.55 8.76 0.00 0.42 Age 14-15 1.06 5.78 0.00 0.31 Age 16 0.83 4.02 0.00 0.26 Age 17 0.46 2.07 0.02 0.15 Age 18 0.18 0.92 0.18 Family size -0.04 -1.51 0.07 Birth order -0.05 -1.42 0.08 Fatherwith primary schooling 0.12 1.05 0.15 Father with intermediate schooling 0.28 1.48 0.07 Father with secondary schooling 0.14 0.64 0.26 Father with higher schooling 0.99 2.57 0.01 0.19 Mother with primary schooling 0.21 1.87 0.03 0.06 Mother with intermediate schooling 0.69 3.79 0.00 0.17 Mother with secondary schooling 0.91 4.39 0.00 0.21 Mother with higher schooling 0.43 1.41 0.08 Single father family -0.16 -0.75 0.23 Single mother family -0.02 -0.15 0.44 Rural -0.51 -4.32 0.00 -0.12 Income per capita -0.03 -0.63 0.26 Has land -0.16 -1.15 0.13 Has title to some of the land 0.42 2.59 0.00 0.05 Titled land 0.03 0.33 0.37 Untitled land -0.01 -0.21 0.42 Has livestock 0.09 0.70 0.24 Missing livestock data -0.12 -0.84 0.20 Has income from self employment 0.12 1.28 0.10 Has income from agriculture 0.07 0.54 0.29 Has income from a family business 0.05 0.44 0.33 Has income from pension, insurance, or interest 0.05 0.49 0.31 Has income from remittances from within the DR 0.08 0.72 0.24 Has income from remittances from outside the DR 0.20 1.74 0.04 0.05 Santo Domingo area 0.40 3.05 0.00 0.09 Santiago area 0.25 1.21 0.11 Number of observations 1447 35 Deterninants of Poverty Table 2.12 Ordered probit results of educational attainment, children ages 19 and up Parameters Estimate t-value Signific. Intercept -0.36 -1.49 0.07 Female 0.37 3.98 0.00 Age 21-22 0.10 0.66 0.25 Age 23-25 0.36 2.66 0.00 Age 26-29 0.76 4.95 0.00 Age 30-36 0.30 1.83 0.03 Age 37+ -0.07 -0.33 0.37 Family size -0.06 -3.12 0.00 Birth order 0.16 3.80 0.00 Father with primary schooling 0.42 2.92 0.00 Father with intermediate schooling 0.67 2.99 0.00 Father with secondary schooling 0.94 2.67 0.00 Father with higher schooling 1.19 3.50 0.00 Mother with primary schooling 0.34 2.77 0.00 Mother with intermediate schooling 1.07 5.70 0.00 Mother with secondary schooling 0.94 4.33 0.00 Mother with higher schooling 1.07 2.39 0.01 Single fatherfamily 0.28 1.30 0.10 Single mother family 0.17 1.18 0.12 Rural -0.40 -3.56 0.00 Income per capita 0.18 3.78 0.00 Has land -0.26 -1.51 0.07 Has title to some of the land -0.27 -1.50 0.07 Titled land 0.07 1.00 0.16 Untitled land 0.01 0.24 0.41 Has income from self employment -0.02 -0.24 0.40 Has income from agriculture 0.27 1.84 0.03 Has income from a family business 0.20 1.43 0.08 Has income from pension, insurance, or interest 0.15 1.43 0.08 Has income from remittances from within the DR -0.12 -1.06 0.14 Has income from remittances from outside the DR 0.00 0.02 0.49 Santo Domingo area 0.25 2.14 0.02 Santiago area -0.06 -0.34 0.37 Threshold 1 0.58 Threshold 2 2.01 Number of observations 692 36 Section 2 Table 2.13 Averages of simulated probabilities using ordered probit coefficients No or primary schooling intermediate schooling secondary schooling higher schooling base probabilities: 0.20 0.25 0.42 0.12 Family size +1: 0.22 0.26 0.41 0.11 Birth order +1: 0.15 0.24 0.46 0.15 Income per capita +10%: 0.20 0.25 0.42 0.12 Gender all males: 0.25 0.27 0.39 0.08 all females: 0.12 0.23 0.51 0.15 Age all age 19-20: 0.30 0.27 0.36 0.07 all age 21-22: 0.27 0.26 0.38 0.09 all age 23-25: 0.15 0.24 0.49 0.13 all age 26-29: 0.05 0.17 0.57 0.21 all age 30-36: 0.16 0.26 0.46 0.12 all age 37+: 0.33 0.27 0.34 0.07 Father's education all with no schooling: 0.30 0.24 0.41 0.05 all with primary schooling: 0.12 0.24 0.50 0.13 all with intermediate schooling: 0.06 0.20 0.55 0.19 all with secondary schooling: 0.02 0.12 0.60 0.26 all with higher schooling: 0.00 0.08 0.57 0.36 Mother's education all with no schooling: 0.32 0.30 0.33 0.05 all with primary schooling: 0.15 0.29 0.49 0.08 all with intermediate schooling: 0.01 0.09 0.66 0.24 all with secondary schooling: 0.03 0.12 0.66 0.20 all with higher schooling: 0.01 0.09 0.66 0.24 RurallUrban all urban: 0.11 0.26 0.51 0.13 all rural: 0.28 0.28 0.38 0.06 Agricultural income all have no income from agriculture: 0.26 0.22 0.40 0.11 all have income from agriculture: 0.16 0.23 0.44 0.16 Region all Santo Domingo: 0.13 0.25 0.49 0.14 all other areas (excluding Santiago): 0.21 0.29 0.41 0.10 37 SECTION 3: MACROECONOMIC POLICY A. Broad Macroeconomic Assessment Public Finances 1. Experience and analysis show that macroeconomic vulnerabilities are, for the most part, related to problems in the components of the joint fiscal position of the public sector cum the Central Bank. Ultimately, it is the evolution of this fiscal position that will determine whether the CB is capable of implementing low rates of monetary growth compatible with price stability, and to generate the necessary market credibility. The fiscal situation since 1992 has been acceptable, with a minimum requirement of financing via money creation and a substantial decrease of the public sector foreign debt. Whether this will continue to be the case depends, of course, of the future behavior of the consolidated government budget deficit and, at least for the immediate future, on the level of net aggregate assets. 2. As of September of 1998, the total public external debt was estimated at about US$3,400 million. Deducting the stock of gross international reserves at the same date (around US$510 million), there was a net debt of 2,890 million dollars, equivalent to roughly 18.3 percent of GDP. This results in payment of interest is equivalent to 1.1 percent of GDP. At the same time, the outstanding internal debt of the CB1, in the form of "Certificados de Participacion" (CP's) as of December of 1998 RD$3,499.2 million, or approximately 1.14 of GDP. These CP's, incidentally, are largely held (46 percent) by individuals, with the rest shared by commercial banks (16 percent) and savings and loans and other financial institutions (34 percent). Assuming that all certificates pay the latest interest rate (16 percent per year), then payment of interest on this debt amounts to an equivalent of .23 percent of GDP. 3. The sum of annual interest payments (foreign debt and CP's) amounts then to around 1.33 percent of GDP. This very aggregate calculation seems to yield results that are acceptable, but that would require, if no additional borrowing is to be undertaken, either a corresponding surplus in the budget of the consolidated public sector (central government plus all public enterprises) or an expansion of the monetary base yielding an equivalent revenue (captured by the CB and eventually transferred, via various possible accounting procedures, to the public sector). In fact, this number of about 1.3 percent of GDP is not too different than the calculation of the revenue from creation of the monetary base that the CB has been capturing, in the average, during the period 1992-98 consistent with an inflation of about 7 percent per year. 38 Section 3 Table 3.1 Balance of payments, in millions of US dollars, 1993-1998 1993 1994 1995 1996 1997 1998 I.- Current account -532.9 -283.0 -182.8 -212.7 -163.0 -383.4 1.1 Trade Balance -1443.2 -1450.7 -1390.9 -1674.2 -1995.0 -2703.8 National exports 602.1 736.4 872.1 945.5 1017.3 923.0 Net exports free trade areas -2795.4 -2991.7 -3164.2 -3580.7 -4192.0 -4957.8 National imports 750.1 804.6 901.2 961.0 1179.7 1331.0 I.II Services balance 737.4 899.4 1000.4 1055.8 1275.3 1254.1 Revenues 1537.1 1787.9 1951.3 2140.0 2446.6 2607.8 Travel 1223.7 1428.8 1570.8 1780.5 2099.4 2270.3 Others 313.4 359.1 380.5 359.5 347.2 337.5 Expenditures -799.7 -888.5 -950.9 -1084.2 -1171.3 -1353.7 Freight -431.6 -485.5 -488.2 -531.5 -557.9 -663.9 Others -368.1 -403.0 -462.7 -552.7 -613.4 -689.8 Balance on Goods & Services -705.8 -551.3 -390.5 -618.4 -719.7 -1449.7 1.111 Investment income -697.0 -681.9 -769.0 -724.8 -795.4 -941.8 Revenues 103.6 101.4 128.1 130.3 140.4 156.4 CB 23.4 16.8 23.5 28.8 24.9 20.1 Others 80.2 84.6 104.6 101.5 115.5 136.3 Expenditures -800.6 -783.3 -897.1 -855.1 -935.8 -1098.2 Interest -275.0 -210.5 -226.9 -206.7 -179.6 -209.8 Dividend and others -525.6 -572.8 -670.2 -648.4 -756.2 -888.4 I.V Net current transfers 869.9 950.2 976.7 1130.5 1352.1 2008.1 Family remittances 720.6 756.7 794.5 914.0 1088.9 1390.7 Social Remittances 116.1 118.0 119.9 141.7 158.4 176.1 Others 33.2 75.5 62.3 74.8 104.8 441.3 II.- Capital account -182.4 455.4 176.0 73.8 451.8 572.5 11.1 Capital account 0.0 0.0 0.0 0.0 0.0 0.0 11.11 Financial account -182.4 455.4 176.0 73.8 451.8 572.5 Direct investment 189.3 206.8 414.3 96.5 420.6 668.7 Portfolio investment 0.0 -38.9 -2.9 -7.3 -7.5 -0.4 M & LT Debt (net change) -440.4 -78.0 27.3 -7.9 -32.4 -62.0 ST Debt (net change) 108.5 110.8 -7.3 76.2 130.5 16.6 Currency and deposits -49.8 100.8 15.8 10.8 2.4 22.3 Others 10.0 153.9 -271.2 -94.5 -61.8 -72.6 III.- Errors& omissions 217.3 -597.8 75.0 108.8 -193.6 -148.0 IV.- Global balance ( I + II + III) -498.0 -425.4 68.2 -30.1 95.2 41.2 Financing 498.0 425.4 -68.2 30.1 -95.2 -41.2 Change in Reserves (- increase) -156.0 386.6 -131.0 15.2 -39.5 -100.6 Borrowing total 652.4 37.4 61.8 13.8 -56.7 58.3 Gifts 1.6 1.4 1.0 1.1 1.0 1.1 Source: Central Bank of the Dominican Republic. 39 Macroeconomic Policy The External Sector 4. Trade. The profile is being repeated throughout the years: a persistent and increasing deficit in "national" trade (imports minus exports of commodities, not pertaining to the activities of the free trade areas). The accepted explanation for the further increase of the fall in "net national exports" in 1998 is related to hurricane Georges, which brought about both a fall in exports and a rise in imports. It can be noted, though, that the level of this deficit for 1998 simply continues the trend initiated in 1995, with a deficit rising at an increasing rate. There is also a clear trend in two other components of the current account in goods and services. The first is a continued growth of net exports generated by free trade areas. The second, net services (essentially fueled by "travel", i.e., and tourism) continue to show a sustained growth at least until 1998, in which there was a small decline. The contribution of free trade areas and tourism is then substantial, although well below the persistent deficit in the regular ("national") trade account. Both free trade areas (which are almost exclusively export labor services) and tourism correspond to a clear comparative advantage. 5. Financial flows. The two other items in the current account are financial flows that are the counterpart of interest and dividends, and net transfers. The first ("Investment income") adds to the goods and services deficit, on account of interest and dividends that correspond to holdings of real and financial assets owned by non-residents. This is a component that remains at stable levels and is expected to grow or remain stationary in the near future --its behavior in the long run depends, of course, on future direct and portfolio foreign investment. The second items in the financial flows are remittances, of which Family Remittances is the most important component. These remittances have shown a modest but steady increase over the period (and more so in 1998, most likely due to the occurrence of hurricane Georges), but it is not clear for how long the trend will continue. 6. Finally, the capital account reports, in the "Financial Account", the "above the line" or "autonomous" components, rather than those used to finance the overall deficit. In this group, direct investment has been growing steadily (except for 1996), while changes in short and long and medium term indebtedness (both public and private) show mixed results. These last two items reflect mostly the effects of changes in government debt, although the difference between this and "Total borrowing" reported "below the line" is rather tenuous. Theoretically, "autonomous" movements (i.e., not motivated by balance of payments considerations) are to be included here, but since money is fungible, it is subject to discussion which is the appropriate assignment of changes in government debt between the "Financial Account" and "Financing" below the line. In a strict fixed exchange rate regime, the distinction would be sharper. 7. Financing. The last three lines summarize the sources of financing of the global balance of payments deficit. Although it is useful from the viewpoint of accounting clarity to distinguish here between changes in reserves and changes in debt ("Total borrowing"), from the conceptual point of view the important magnitude is the sum of these two items, as they represent a change in the government's stock of assets. 8. Policy issues. Under a "mixed" system in which, as in the DR, the regime is theoretically one of a "free floating exchange rate", but in which the CB makes exchange rate stability one of its objectives of policy, external disequilibrium becomes an elusive concept, and its characterization depends on the extent to which the CB is ultimately prepared to defend the exchange rate. These considerations that could be interpreted as merely theoretical have an empirical counterpart, in that they make it difficult to assess "problem" areas in the evolution of the items in the balance of payments as far as macroeconomic management is concemed. Here, and with specific reference to some of the items in the balance of payments, it can be assumed that the CB is committed to a narrow path of the nominal exchange rate (i.e., to a path of the rate of devaluation), roughly comparable to the "target" inflation rate. Under these conditions, surpluses 40 Section 3 become "beneficial", and deficits are deemed to be "detrimental". 9. Among items of concern (or comfort) to be singled out are the following: (i) external debt has been decreasing for the last eight years, and it is now at a very manageable level; (ii) it is not clear whether the relatively high level of family remittances will be maintained in the future; (iii) continued strength of the tourism sector remains very important; (iv) an increase in world interest rates, that would affect the dollar value of payment of interest, is always a possibility. Indeed, the recent modest rise of rates in the US following their record lows during 1998 is an indication that such a rise cannot be ruled out. The level of both public and private debt seems to suggest, though, that even if a moderate increase would take place it would not have a very significant effect. A rise in world interest rates would also tend to decrease, other things equal, the flow of portfolio foreign investment, and eventually direct investment. But the former is a rather small flow to start with, and the latter would seem to be more sensitive to local conditions (such as the health of the tourism sector, for example) than to the world interest rate; and (v) the single most important item would seem to be the high (and rising) deficit in the "national" trade account (the difference between exports and imports not related to free trade areas). An increase in the level of traditional exports would seem to be linked to future decisions concerning trade liberalization and the deregulation in the agricultural sector. 10. The real exchange rate. Finally, some considerations need to be made concerning the "real" exchange rate (defined as the price of traded in terms of non-traded goods). After the unusually high levels between 1987 and 1989, associated with the crisis of those years, the rate has been gradually falling (i.e., domestic currency has been appreciating) since 1990, and it has done so by about 20 percent. The questions of what is the equilibrium level of the real exchange rate, and whether this level can be effectively influenced in the long (or even the short) run are highly controversial. First, such a fall in times of stabilization and monetization (as in the DR since 1992) is in general the rule rather than the exception; in fact, a 20 percent fall iii eight years is mild compared to the experience of other countries. Second, the level of the real exchange rate does not seem to be currently an important reason for lack of "competitiveness" in either national exports or the evolution of the international tourism industry. Finally, mention could be made of the perils of a monetary and exchange rate policy aimed at influencing the real exchange rate. As pointed out before, there is still a great deal of controversy concerning this possibility, but more and more the professional and policy-makers consensus is drifting towards the conclusion that nominal magnitudes (such as the nominal money stock, or the nominal exchange rate) cannot influence for long the real exchange rate, which is a relative price determined by real considerations. Indeed, there are doubts about the possibility of exchange rate policy inducing changes in the desired direction even in the short run without creating instability and time inconsistency problems. Monetary and Exchange Rate Policy H1. Since the recovery from the crisis of the late eighties, monetary policy has performed well, aided by a process of sustained growth. Since 1992 there has been an initial quick fall in the inflation rate, that has since been kept at an average rate of 7 percent per year, with an important real monetization and an accompanying appropriate rate of devaluation (see Figure 3.12) 12. Consider first the basic monetary variables. The monetary base grew at an average annual rate of 16 percent, and the trend points to a longer run level very similar to that average. Given the growth in real GDP, this growth of the monetary had allowed for a much smaller pace of inflation. As a result of both real GDP growth and probably still as a lagged reaction to the initial fall in the inflation rate, an important real monetization took place during the period, particularly for the case of M2. The percentage of Ml holdings over GDP remained at a level very close to 10 percent, while the same percentage for M2 kept an upward 41 Macroeconomic Policy trend and in 1998 reached 22.5 percent. Figure 3.1: Monetary and exchange rate variables .25 .8 .20 - .6 4 .15 .2 .10 .0- C,. ,w .05 't _ ________________ 86 88 90 92 94 96 98 92 93 94 95 96 97 98 - Annual Inflation - Ml over GDP M2 over GDP .35 0.8 .300. 0.4 92 93 94 95 96 97 98 . -0.2 ______ ____,;____ _____ _____ 92 93 94 95 96 97 98 ^ Lending rate, 90 days -Borrowing rate, 90 days - Ann Growth Mon Base -A-Trend 16 0.12 15 _ ;'C' 0.08 0.06 14 0.04 . 13_ 0. 02 -V O. .1. 12 , , 0.02oo . _ ___ . , i . . . l . . 92 93 94 95 96 97 98 92 93 94 95 96 97 98 - Banks Exch Rate - Growth of Banks Exch Rate ----Official Exch Rate --+- Trend 13. Nominal interest rates (of which the 90 days lending and borrowing rates are shown, as quite representative), after some fluctuations up to mid 1995, showed a welcome downward trend that was interrupted in the third quarter of 1997, most likely as a result of some loosening of the fiscal stance, the 42 Section 3 decision of the CB to increase the interest rate of their CP's ("Certificados de Participacion") and periodic discretional changes in reserve requirements (see Section D). The interbank exchange rate, in theory a variable "freely determined by the market" but closely watched and heavily influenced by the CB, grew during the period at a rate lower than the inflation rate (an average of 2.8 percent per year), which of course resulted in the gradual fall of the real exchange rate as reported above. The trend of the rate of devaluation is clearly on the rise, though, particularly after the unification of the exchange rate market. Eventually, as long run macroeconomic equilibrium is attained, both the rate of devaluation and the inflation rate could converge. 14. Two points could be made with respect to the exchange rate. The first one is that, despite the declaration that the monetary regime is one of "freely market determined exchange rates", there is still a group of commodities (some basic agricultural products, and petroleum) for which transactions must be channeled through the CB, at the "official" exchange rate. Although over time this "official rate" has been increasingly flexible, with very frequent adjustments to follow closely the market rate, there is no reason for an "official rate" in a floating exchange rate regime. The second observation is that in a floating exchange rate regime there is, in principle, no place for the declared CB objective of "maintaining exchange rate stability". Additionally, increases in the CP's interest rate aimed at inhibiting increases in the exchange rate, as it is well know, may eventually result in substantial increases in the level of CB debt and hence in the deterioration of the overall fiscal position. The instruments of monetary policy used by the CB include open market operations, through the CP's. However, there is also the frequent use of quantitative discretionary "freezes" of commercial banks reserves. This is an extremely inefficient instrument, and reveals the lack of decision rules that could be taken into account by the market. B. The Short and Medium Run: The Conduct of Monetary Policy and the Fiscal Question Background 15. This section includes somewhat detailed analysis of two areas that are of major concern in the short and medium run: the conduct of monetary policy, and the fiscal question. Both these issues (obviously the second, but also to an important extent the first) need to be discussed with direct reference to the fiscal position of aggregate govemment (consolidated public sectorcum the CB). For purposes of further reference, therefore, it is worthwhile to lay out the basic framework, that is given by the overall budget constraint. 16. The budget constraint of overall government, at the highest meaningful level of aggregation, can be written as (1) A b=6- - +br=- -hu+br 17. where b is aggregate real debt, 5 is the central government primary deficit, h and H are the real and the nominal monetary base (sometimes referred to as "high powered money"), r the real interest rate and u the rate of growth of the (nominal) monetary base. The symbol A, as usual, means "change in". The term on the l.h.s. is then the increase in indebtedness during the period (including foreign and domestic debt, whether incurred by the central government or by the CB); the term AH/P is the real value of the increment of base money issued by the CB, and the term b r is the payment of interest on outstanding debt during the period. Notice that the term AH/P can also be expressed ash hu, i.e., the product of the real monetary base and the rate of nominal base growth. 18. In turn, and after a couple of simple substitutions, the term accounting for the flow of real revenue 43 Macroeconomic Policy from money creation can be expressed as l = h p u Ah + h ;r 19. The first term on the right hand side, Ah, is the increase in the real stock of the monetary base; the second, h ;r is the product of the real monetary base and the inflation rate (7r) during the period. The latter is often referred to as the "inflation tax". In an economy which is growing, of course, the first of these terms is positive in equilibrium, as real cash balances and hence the monetary base also grow. The interpretation of (1) is straightforward: the primary deficit plus the payment of interest on outstanding debt (the "quasi fiscal deficit") needs to be financed either by net borrowing or via money creation. 20. This expression, of course, can to be written at lower levels of aggregation to take account, for example, of different debts with different interest rates, or different maturities, and so on. However, it is very useful in its most aggregate form as a reminder of various fundamental points. It implies, first, that indebtedness by the central government or the CB has ultimately the same effects. Second, that foreign or domestic debt eventually amounts to the same? It also stresses the obvious fact that the fiscal position of the government needs to be, in principle, sharply distinguished from the position of the "country" (government cum the private sector), although the two are obviously linked. It also stresses the fact that although at any given period a relatively high primary deficit can be accommodated by net borrowing, in the medium and long run the variable that will eventually need to adjust, if the deficit persists, is the rate of money growth, since net borrowing (Ab) cannot continue indefinitely beyond certain limits. It is therefore the magnitude of the primary deficit that will ultimately need to be kept low for monetary expansion to remain at the low levels compatible with low inflation. The independence of the CB, and whether a presumably inflation fearing CB or a deficit prone central government will prevail depends, of course, on the institutional framework and on political considerations. 21. Notice, finally, that expression (1) is an identify, that holds independently of the monetary regime --i.e., whether there is an exchange rate anchor, or a monetary rule, or any discretionary policy. These are of course elementary notions, and perhaps even trivial.4 Still, the framework in itself is indispensable for the discussion of the above mentioned two points of concern. The Conduct of Monetary Policy 22. Under a strict exchange rate anchor, with a pre-announced path of the nominal exchange rate, the question of monetary policy becomes to some extent a moot point. The CB simply acts passively, and changes in monetary aggregates simply reflect incipient imbalances in the supply and demand for foreign exchange. If there is little or no capital mobility, the CB is able to modify monetary aggregates via open market operations or changes in reserve requirements, but unless those changes match changes in the demand for money, the trade balance will in due time reverse them, resulting in a change in reserves. With perfect capital mobility the reversal is much quicker, sometimes almost instantaneous. 23. Under a floating exchange rate system, on the other hand, the monetary authority recovers the ability to control monetary aggregates, and the market, as any other price determines the exchange rate. The question then arises, as it does in a closed economy, of how the CB should conduct monetary policy. This has been the question to be resolved by the DR CB since the change from a fixed to a floating rate in the late 1980's. There are several areas of concern with reference to the current conduct of monetary policy in the DR particularly with respect to objectives, intermediate targets and instruments. 44 Section 3 24. Objectives and targets. In various documents and in the Monetary Program, the CB defines its objectives of monetary policy as "to achieve macroeconomic stability", subject to various quantitative targets. More specifically, for example, the 1998 Monetary Program states that "The objectives of the Monetary and Financial Program for 1998 aare the preservation of macroeconomic stability, consistent with a GDP growth of 6 percent, a rate of inflation not higher than 8 percent, and a loss of net international reserves of no more than US$10 million". In addition, although it is not explicitly stated as an objective, CB actions and frequent references in many documents indicate that "exchange rate stability" is another objective, sometimes commanding first priority.5 As a result of these objectives, quantitative targets are set for Net International Reserves, Net Internal Assets of the CB, and Net Credit of the Banking System to the Public Sector. Notice that the second of these magnitudes can be controlled by the CB, and to some extent the third, when quantitative controls are imposed on commercial banks; the first (changes in international reserves) is rather a result of all the others. 25. These objectives, however, can easily become incompatible with each other, and there is no statement concerning priorities that would serve as a guideline to the market.6 Consider, for example, the objectives of a change in the level of international reserves and a low inflation rate. Notice, from expression (1) that, during any period, the rate of monetary expansion can be increased (and, with it, the term h ,u), with the counterpart being a decrease in indebtedness (the term Ab) so that net assets increase. Furthermore, under a floating exchange rate system, the CB is free to vary its rate of monetary growth (or the level of the monetary base). In other words, increases in the monetary base can be used to purchase anything, including foreign exchange reserves. It is indeed not difficult to imagine a long run equilibrium with a high rate of monetary expansion and hence a high level of the inflation tax, with the "tax" being used exclusively to accumulate foreign assets. 26. Consider, further, the connection between the current exchange rate and a target level of higher reserves, presumably in anticipation of a possible future deterioration of the balance of payments that might require the injection of foreign exchange to avoid a rise in the exchange rate. Today, the CB expands the monetary base and purchases foreign exchange.7 However, the extra liquidity injected in the market will soon effectively increase the exchange rate, so that the effort ends up generating the same effect that it was intended to avoid. Of course, the CB can always sterilize the increase in liquidity by issuing CP's. Then, the exchange rate will not increase today, but the counterpart of the increase in reserves would be a rise in liabilities. These two magnitudes would in principle offset each other, but this would in general entail higher future outlays, since the interest earned in reserves (perhaps around 5 or 6 percent per year) will be lower than the interest paid on CB debt (currently 16 percent per year).8 This initial rise in the exchange rate in response to a policy intended to limit itsfuture rise can become particular important under conditions in which the exchange rate tends to "overshoot" as a result of lags in the adjustment of the price of non-traded commodities. 27. Discretion vs. Rules. Monetary policy is discretional, rather than being guided by rules. The term "discretional" is used here in its strict technical meaning, and has no negative implication, and has nothing to do with "capricious", "erratic" or "arbitrary" behavior. It simply means a regime under which the monetary authority uses its instruments in reaction to events in the economy, rather than setting exogenous targets for the magnitudes that can directly control, such as the monetary base --or, in the case of an exchange rate anchor, the exchange rate. Indeed, the CB has used its discretionality very wisely during the past few years, and continues to do so. 28. Of course, it is true that in principle the path of a certain objective (the price level, for example) corresponds to a unique path of some instrument, or variable that can be effectively controlled (the monetary base, for example). There is then a trivial sense in which it would appear not to matter whether the CB sets the former and manipulates the corresponding path of the latter, or whether it follows the inverse procedure, 45 Macroeconomic Policy setting the latter and generating the corresponding path of inflation. These are, indeed, "observational equivalent". The two procedures are not at all truly equivalent, though. Indeed, the first amounts to "discretionality", and the second to "rules". The difference between the two procedures may not even be noticeable at times at which the usual unavoidable shocks to various sectors are small and independent, as it has happened in the DR during the past eight years, so that most variables (things such as the inflation rate, the rate of devaluation, interest rates and monetary aggregates) all move more or less in accordance with their anticipated paths. Important shocks uncover the distinction, and the outcome of the two regimes may be very different. 29. There is a discussion of long tradition between proponents of "rules" and those favoring "discretionality", and the question is far from resolved. Indeed, the only true rule that CBs sometimes follow is that of an exchange rate anchor (in particular when it takes the strict form of a currency board). The so- called "monetary anchors" translate, in general, in CBs setting some objectives (in general, the price level) and discretionally using intermediate targets (such as interest rates) and indicators (such as the exchange rate). This is the long-standing CB tradition in Latin America, and it would be difficult to persuade the DR's CB to do otherwise. It would be far better, perhaps, for the CB to set targets for the aggregate it can really control (the monetary base), at levels compatible with the desired low inflation rate given expected GDP growth, and let the market determine the exchange rate, interest rates, and higher level monetary aggregates such as MI and M2. 30. The global recommendations that emerges from the analysis of these two different but related points are the following: * If a strict "rule" (such as a predetermined path of the monetary base) is out of the question, still an important improvement could be made with respect to a more focused set of objectives, or perhaps a single one. Price stability is the primary candidate (the so-called "inflation target"). * The CB could clearly announce this objective and make clear its "reaction function", i.e., what decision rules it is adopting for attaining its objective. In particular, the CB could state which indicators it is adopting (interest rates, or changes in the exchange rate). It could be stressed that the important thing is for the market to have a clear idea of what these decision rules are.9 * Moreover, the CB could also make clear to the market which instruments it will use, and how. * In particular, the CB could avoid having the exchange rate as an objective, and make this point clear to the market. Notice, with reference to one of the previous paragraphs, that to take the exchange rate as an indicator is different than to have it as an objective. 31. The instruments of monetary policy. The second area of concem in the conduct of monetary policy is the use of instruments. The CB uses open market operations ("Certificados de Participacion"). In the absence of a stock of government securities, this is an acceptable instrument. Indeed, open market operations are universally seen as the appropriate, most neutral instrument. Frequently, though, it resorts to another, rather inefficient instrument, which are changes in reserve requirements. Furthermore, changes in reserve requirements are used with a particular perverse twist, i.e., in the form of sudden "freezing" of excess reserves. This is a procedure that entails large inefficiencies, and its use could be avoided 32. There are three problems with this procedure. First, other things equal, it decreases commercial banks' profitability, which is translated into a higher spread between borrowing and lending rates. Second, and more importantly, it can only be imperfectly anticipated by the banking sector --indeed, its effectiveness depends on not being anticipated, and introduces therefore an additional element of discretionality. Third, 46 Section 3 and most importantly, it interferes with the efficiency of commercial banks' operations. It is natural that excess reserves may exist at times, as commercial banks seek to maximize the expected return of their lending --i.e., not only at the highest possible interest rate, but also with respect to the probability of being repaid (the "quality" of the loan). The presence of excess reserves plays therefore a role in the borrowing and lending process of commercial banks. The discretionary, largely unanticipated freezing of excess reserves increases the cost of using those reserves and introduces a distortion. Notice that this distortion (although with different intensities) is present whether the freezing, when it occurs, affects the preexisting stock of excess reserves or the future flow. C. Growth, The Demandfor Money and The Inflation Tax 33. During the last seven years (since the beginning of 1992), the DR has experienced a high rate of real GDP growth, which on average amounted to around 6 percent per year. One of the beneficial by-products of this rate of growth has been the real monetization associated with growth, and the difference between the rate of growth of nominal monetary aggregates (specifically, the monetary base) and the resulting inflation rate.10 On average, the nominal monetary base grew at an annual rate of 16 percent, and the annual inflation rate was 7 percent. 34. As a result, the real flow of total revenues captured by increases in the nominal monetary base (an average of 1.3 percent of GDP for the period) was well above the level of the "inflation tax" (an average of .58 percent for the period). None of these percentages is alarming at the present time, but a closer analysis of the figures reveals two potentially disturbing facts, that could greatly complicate the consequences of a loosening of fiscal discipline. The first is that the government's fiscal position, and the ability of the CB to maintain a low inflation rate, are extremely sensitive to the level and the rate of growth of real GDP. The second is that, for any given level and rate of growth of real GDP, relatively small increases in the need for financing via increases in the monetary base would lead to relatively large increases in the inflation rate. These two conclusions, although intimately related, are different, but both point in the direction of caution. 35. Notice that this is not a discussion of the uses of whatever resources are generated via the creation of monetary base --it could be to cover a primary fiscal deficit, to pay interest on the existing debt (whether it is CB or central government debt) or to accumulate net assets (in the form of net foreign reserves or via a decrease in debt). 36. The definition of real revenues from creation of monetary base is given by: dHlI= dH]IH(t) =dh [Al] (t) dt P =(t) dt H(t)P () u (t) (t) dt where H, h, P, a and 7t are the nominal and real monetary base (or "high powered money", or "reserve money", equal to currency in hands of the public plus commercial banks reserves), the general price level, the (proportional) rate of growth of the base and the inflation rate, respectively. 37. In turn, the level of "money", taken here to be equal toml (currency plus demand deposits) is equal to the monetary base times a multiplier, i.e., [A2] m,=yh 47 Macroeconomic Policy where y is the multiplier corresponding to ml. In what follows "money" is called mi, so its subscript is dropped. 38. The demand for money (ml) is taken to be a function of the level of real income (GDP) and the inflation rate. Specifically, assuming the semilogarithmic form [A3] In m = A + A, In y + A, r where A, A y > 0 and AFr < 0 are constants. Notice that in this formulation the coefficient Ay is the income elasticity of the demand for money. Finally, from [Al] and [A3], and assuming that d irldt = 0 (i.e., that the monetary sector is at its steady state) one can derive [A4] T= u -Ay and [A5] R=h(ir+aAy) where a is the (proportional) rate of growth of real income (GDP). Using [A2] and [A3], the last expression can be written as F(A + Ay In y + A, IC) [A6] R = exp ((7 +a Asuby } 39. Expression [A6] makes very clear how and why the flow of real revenues from base creation depends on the inflation rate, the level of real income and the rate at which real income is changing, i.e., the rate of growth of income. The reason for this is set forth in expression [A4]: when the economy is growing, the rate of inflation will be smaller than the rate of base growth, the difference being given by (i) the rate at which the economy grows (a), and (ii) how the growth of real income influences the demand for money, i.e., the income elasticity of the demand for money (Ay). A higher rate of real income growth, then, means that, at any given point in time (or, for any period) any given level of real revenue requires a lower inflation rate. Notice that the influence of income growth is different (and over time adds to) the influence of the levelof real income --of course, a higher rate of growth during some time results in a higher future level. 40. Estimates of the demand for money. Estimations are based on quarterly observations for the period January 1992 to the present. This time frame provides a sufficient number of observations, and corresponds, after the crisis of 1989-1990, to a period during which there has been no important "change of regime" (so that the estimations are not in principle subject to the "Lucas critique"). The Government (the central government cum the CB) captures real revenues from the creation of monetary base. What is therefore needed, in order to investigate the empirical counterpart of some of the expressions set forward in the previous paragraph, is an estimation of the relationship between the level of real base (currency held by the public, plus commercial banks reserves) and variables such the inflation rate and the level of real income. Bank reserves, though, are held on account of both ml (currency held by the public, plus demand deposits) and the other components of m2 (savings and time deposits). A more elaborate structural estimation would require, then, an estimation of the demand for both ml and m2, with a careful analysis of the evolution of the 48 Section 3 various reserve requirements over time. In order to circumvent this problem, it is observed that during the period 1992-1998, the relationship between ml and the monetary base (the multiplier for ml) has been remarkably stable, with the exception of short term variations, at approximately one. On the other side, it is also found that the demand for ml is by far more stable than the demand for m2. Estimations of the real monetary base are then obtained indirectly, by estimating ml and relating this aggregate to the real base via the value of the multiplier. 41. Estimations of ml are performed via a simple OLS, using quarterly annualized observations for the period January 1992- June 1998 (except where noted). The coefficients for real GDP and the inflation rate, therefore, have the dimension of "per year". Both the coefficients for real GDP and for contemporaneous inflation are highly significative, and although the Durbin-Watson statistic is a bit below the desirable level of 2, it is quite acceptable, indicating that the AR(1) term has captured most of the autocorrelation usually plaguing this kind of estimation. Notice that inflation enters both contemporaneously and with one-quarter lag. In the case of real income, the introduction of lags does not improve the estimation. Table 3.2 Estimation results: Real money stock LS H/ Dependent Variable is LN REAL Ml Sample(adjusted): 1992:1 1998:2 Included observations: 26 after adjusting endpoints Convergence achieved after 11 iterations Variable Coeffic Std.Error t-Statist Prob. LN REAL GDP SA 1.424 0.146 9.771 0.000 ANN INFLATION -1.170 0.580 -2.018 0.057 ANN INFLATION(-1) 0.029 0.583 0.049 0.961 C -7.837 1.252 -6.261 0.000 AR(1) 0.134 0.221 0.609 0.549 R-squared 0.862 Mean dependent var 4.367 Adjusted R-squared 0.835 S.D. dependent var 0.155 S.E. of regression 0.063 Akaike info criter -5.364 Sum squared resid 0.083 Schwarz criterion -5.122 Log likelihood 37.834 F-statistic 32.658 Durbin-Watson stat 1.798 Prob(F-statistic) 0.000 42. The revenues from money creation. Based on the estimates of the demand for money, alternative scenarios are used to determine the "equilibrium" inflation tax for given real GDP growth and needed revenues from money creation. The basis for this calculation is the estimated equation [A6], and the steady state values of the demand for money. 49 Macroeconomic Policy Table 3.3 Annual real revenues from money creation (as % of GDP) Inflation Real Income Growth Average Revenue 9% 6% 3% 0% 1992-1998 0.0 1.59 0.96 0.43 0.00 1.30 2.5 1.85 1.21 0.67 0.22 1.30 5.0 2.09 1.44 0.90 0.43 1.30 7.5 2.32 1.66 1.10 0.63 1.30 10.0 2.53 1.86 1.29 0.82 1.30 12.5 2.73 2.05 1.48 0.99 1.30 15.0 2.92 2.23 1.65 1.16 1.30 17.5 3.09 2.40 1.81 1.31 1.30 20.0 3.25 2.56 1.96 1.46 1.30 22.5 3.40 2.70 2.11 1.59 1.30 25.0 3.54 2.84 2.24 1.72 1.30 27.5 3.66 2.97 2.36 1.84 1.30 30.0 3.78 3.08 2.47 1.95 1.30 32.5 3.89 3.19 2.58 2.05 1.30 35.0 3.99 3.29 2.68 2.15 1.30 37.5 4.08 3.38 2.77 2.24 1.30 40.0 4.16 3.46 2.85 2.32 1.30 43. The complete form of expression [A4] provides another useful measure, i.e., the elasticity of the flow of real revenues from base creation (as percentage of GDP) with respect to the rate of growth of GDP. A simple but a bit tedious calculation reveals this elasticity to take the simple form aR a I [A8] aR c + Aya 44. The elasticity is higher, the higher the rate of growth and the income elasticity of the demand for money, and the lower the inflation rate. For the average values of the period 1992-1998 (7 percent for the inflation rate, and 5.9 percent for the rate of growth of GDP) and for the estimated value of the income elasticity of the demand for money (1.424159), the elasticity takes the value .5459. That implies that other things the same, a one percent rise in the rate of growth of income brings about an immediate rise in revenues of approximately half a percent, a relatively high value.1' 45. The level of real GDP is one of the determinants of real monetary aggregates, and hence of the monetary base, comprised of currency held by the public and commercial banks reserves. Other things the same, higher real GDP growth naturally leads to higher levels of real GDP, and hence to higher real revenues from the creation of monetary base for any rates of nominal base growth and inflation. In addition, therate of growth of real GDP influences the relationship between the rates of nominal base growth and the inflation rate. As a result, a higher real GDP growth rate allows the rate of nominal base expansion to be higher, for any resulting inflation rate. Or, what is the same, it generates a lower inflation rate for any rate of nominal base expansion. Since the (total) revenues from base creation depend on the latter, what this means is that higher rates of GDP growth allow higher revenues for the same inflation rate. 46. The relationship between the rate of nominal base expansion and the inflation rate, in tum, is given by both the rate of real GDP growth and the extent to which money holders increase their demand for real monetary aggregates in response to the level of real GDP (the so-called "income elasticity" of the demand for money. In the case of the DR, with an estimated income elasticity of the demand for money (more 50 Section 3 specifically, ml) of around 1.42, and for an average rate of growth of income during the period 1992-98 of 5.9 percent per year, the difference is around 8.5 percentage points. This means that, for that period, a given rate of nominal base expansion yields an inflation rate of about 8.5 percentage points less. In fact, for the period, the actual averages have been 16 and 7 percent, respectively. 47. These numbers reveal a high sensitivity of real revenues from the creation of monetary base to the level and the rate of growth of real GDP. This means, on one hand, that during the 1992-98 period the relatively high rate of GDP growth allowed for a relatively high rate of base growth (and hence a non- negligible level of revenues, which for the period was 1.3 of GDP on average) with very little inflation. Furthermore, higher rates of real GDP growth would still improve this situation in a non-trivial manner. But it also means, unfortunately, that a lower rate of real GDP growth would also require an important cut in nominal base expansion in order to keep inflation at the same low level of that in the 1992-98 period. 48. The inflationary consequences of base expansion financing. A related, but different question, is given by the sensitivity of real money holdings, and hence the real monetary base, to the inflation rate. The relevance of this sensitivity is that it indicates the extent to which additional revenues from base creation would change with an increase in the steady state inflation rate. Or, what is the same, it indicates the extent of the social cost to be paid, in terms of all the distortionary negative effects of inflation, for additional financing via the creation of nominal base. 49. In the case of the DR, as indicated in Appendix A, the demand for money (ml) and hence real base, is found to be very sensitive to the inflation rate. The estimation shows the coefficient measuring this sensitivity'2, is equal to - 1.17. This indicates that the maximum revenue from the inflation tax would occur at the rate of (1/- 1.17) = .85, or 85 percent per year, a very high inflation rate, but considerably lower than the equivalent rates reported for other countries. 50. Implications and conclusions. The obvious implications of the two previous points is that if fiscal discipline is not maintained, the inflationary consequences will be important, in particular if the rate of growth of income turns out to be lower in the near or medium run future, as shown in Table 3.3. More specifically, calculating the revenue from base creation, as a proportion of GDP, depending on alternative inflation rates (going from zero to a very high 40 percent per year) and for the case of a few possible rates of GDP growth (from zero to a high of 9 percent per year). The simulations calculate those levels of revenues 2.5 years from the base period 1995-98, i.e., can be interpreted as the medium-run result of a new long run equilibrium. The average revenue of 1.3 percent of GDP for the period 1992-98 is also included, for purposes of comparison. Figure 3.1 helps visualize the comparisons. Clearly, in this graph every curve corresponds to one of the columns in Table 3.2. Notice that one of the possible growth rates of real GDP (6 percent) corresponds to the average growth rate during the period 1992-98. 51. The following are among many relevant points highlighted by Table 3.2 and Figure 3.3. Consider, first, what happens if real GDP growth during the two and a half years proceeds at the same rate of 6 percent per year exhibited for the 1992-98 period. After 2.5 years, the same revenue of 1.3 percent of GDP could be attained with a lower inflation rate (of approximately 3 percent), due to the higher level of real GDP after 2.5 years (point A in Figure 3.1), or a higher revenue could be generated with the same rate of inflation (point A'). With the same growth rate, a rise of revenues to 2.5 percent of GDP would require an inflation rate of 20 percent (point C), and for a rise to 3.5 percent of GDP the inflation rate would have to rise to 40 percent per year (point D). I.e., for this last range, a rise of revenues of one percentage point of GDP brings about a doubling of the inflation rate. 52. Secondly, consider the effects of a fall in the rate of growth of real GDP to zero (a rather pessimistic outlook, but which is useful to emphasize the point). If this would happen, the maintenance of the same 51 Macroeconomic Policy revenue level of the past (1.3 percent of GDP) would require an inflation rate of 17.5 percent (point B in Figure 3.1), i.e., over two times the base inflation rate of 7 percent. Figure 3.2 Inflation and the revenues from inflation tax C:4 .0%+ -- ----- ----- - ------ ----- ------ 0 -30% dw 3, ot . t~- ---- -- - -----;>~ 1'2.0% 0 0% 5% 10% 15% 20% 25% 30% 350 400 Annual Inflation 9% -6% - 3% 0% -Average 1992-98 53. Finally, and in order to emphasize the role of the rate of growth of real GDP as opposed to just the level of real GDP, Figure 3.3 helps to show the separate influence of both. It shows, first, the level of revenue (also 2.5 years from the base period) for the case of a zero rate of real GDP growth. Next, it shows the level of revenue obtained with a 6 percent growth (the line labeled "6%, Tot Rev". These two are merely the repetition of the two corresponding curves in the previous Figure 3.2. But Figure 3.3 shows also a third line (labeled as "6%, Infl Rev", which depicts the hypothetical revenue that would be attained at the levelsof real GDP reached after 2.5 years of 6 percent growth per year, but with no growth taking place at that point in time. The vertical difference between these lines, then, shows the relative importance of one of these factors versus the other; clearly, the "growth factor" diminishes in relative importance for higher inflation rates.'3 But for a plausible level of the inflation rate of 10 percent per year, for example, the growth factor makes up 82 percent of the difference.'" Figure 3.3 Inflation and the revenues from inflation tax 3, 5%_ 3 . 0o6 -- -- - -- - ------------_,_ __ ' 4, 2. 5 M-- ,- , ,----, 0 , 2.0'6 -- __ _-- < or O .0'% 0% 5% 10% 15- 20% 25% 30% 35' 40% Annual Inflation E0% - 6%,Infl Rev --6%, Tot Rev 52 Section 3 54. The implication of the analysis in the previous paragraphs seems clear. They point, first, to the dangers of even a relatively small relaxation of fiscal discipline, and the need for additional monetary expansion), in terms of very high resulting inflation rates. Second, they show how vulnerable the fiscal stance is to both the level and the rate of real GDP growth. A fall in the rate of real GDP growth is not only in itself a negative event, but it is also one that changes (worsens) the trade-off between money creation financing and inflation. 55. What the analysis shows is that it would be wrong to take comfort from the current relatively low levels of inflation and inflationary financing, and that even a small relaxation of fiscal discipline can have important consequences. The policy conclusions take then the form of strong additional reasons for: i. Accelerating the implementation of fundamental policies that would minimize the odds of a fall in the rate of growth; ii. Strengthening fiscal discipline in anticipation of an eventual actual fall in the rate of growth, given the sensitivity of seigniorage revenues to the level and growth of real GDP; and iii. Strengthening fiscal discipline, given the high sensitivity of the inflation rate to additional financing via increases in the monetary base. 56. In concluding this section, two clarifications are in order. First, that while discussing here the influence of real GDP on revenues from seigniorage, it is referring to a very different phenomenon than the fall in government tax revenues that usually accompanies a fall in real GDP, which of course also contributes to a deterioration of the government's fiscal position. Second, "fiscal discipline" is referred to as the means to improve the government's overall fiscal position, although the primary deficit is not the only component of the overall government (central government plus the CB) budget constraint. In particular, payment of interest on outstanding debt is another component that may force inflationary financing. But the primary deficit is the fundamental variable that will have to ultimately adjust. Finally, notice that as far as the influence of the level of real GDP on financing via monetary expansion, a fall in the level of GDP tends,per se, to increase the level of the primary deficit and bring about, therefore, an increase in the ratio of the primary deficit to real GDP, as government expenditures do not usually decline following a fall in GDP. D. Fiscal, Monetary Policy and Interest Rates 57. As it clearly emerges from the graphical depictions in Figure 3.4, interest rates, which were falling since the beginning of 1995, reversed course in the second half of 1997 and, in the case of the lending rate for 90 days or less, increased from around 18 to nearly 28 percent per year."5 It is not obvious the reason or reasons for this behavior, since precisely during the second half of the same year (1997) the inflation rate started to fall, after a peak at 10 percent per year, and reached a low of around 3 percent per year a year later. Inflationary expectations do not seem therefore to be at the center of the process, since it could be expected that, if they existed, they would have also manifested on current inflation. The rate of MI monetary expansion does not show an increase during 1997 --in fact, it fell from an average growth of around 25 percent per year in 1996 to an average of 20 percent per year in 1997. The behavior of the inflation rate, given the rise in nominal rates, implied in turn a sharp rise in the ex-post contemporaneous real interest rate, that climbed from a low of around 9 percent per year to 25 percent in the period of one year. Spreads, that were slowly falling, also show an increase back to previous levels starting in the second half of 1997. 53 Macroeconomic Policy Figure 3.4 Money and interest indicators 0.32 0.25 0 .30 o . 1, 0.28 6 --------------- 0 .26 i f II I \ & ; 1 0.24 0.22 . , 9/oo 0 . 2 0 - - - -- - - - - - - - , -- -- 0 . 0 0 - ----- - -- --; ; - -- - - - - - - - - - --;;;-- - - - -- -.., , ; 0.18 20 --0. 1994 1995 1996 1997 1998 1994 1995 1996 1997 1998 - 90 Days Lending Rate - Annual % Base Growth 0.16 0.4 0.14 0.3------ -- - - -- -- 0.10 02 3 0.08 - 0.1-- --- ---------- ,.l 0,06- ------ ----------. --L--9 4 0.06~~~~~~~~~~~~~. 0 .0 4 - ------ ---------- --------0 .0 0.04 i.1 0.02 t______________ o___ -0..... _ ... .......... 1. ... 1994 1995 1996 1997 1998 1994 1995 1996 1997 1998 - Annual Inflation Rate - Annual % Ml Growth 1.2 0.30 0.25 .- -- - +1 -------- - --------- --- ------2 0.20 - --- ------- 1.09 0 .1 5 - -- - - - - - - - - - - - - - - - 0.9- - 0.10 ~ ~~~~~ --- - - - -- - - - - - - - - ---- - - - - 0.8 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 0.05 .9'94' 1 .9.95"1.9.96' 1i9.97' '19-98 . 1 ..... 994 1995 ....1.9'96' "1i9.97'. ' '"..93.. ... - M1 Multiplier - Real Ex-Post 90 Days Rate 58. The analysis of the short-run behavior of nominal interest rates is one of the toughest questions in macroeconomics. It is known that in the long run (i) the real interest rate converges to the real rate of return on capital, and (ii) the nominal interest rate converges to the sum of the real rate and the expected (and actual) inflation rate --the so-called "Fisher effect". More expansionary policies (i.e., higher rates of monetary expansion) are then associated with higher nominal interest rates. In the short run, though, the opposite may happen, i.e., a more expansionary (contractionary) monetary policy may be for some time associated with lower (higher) nominal interest rates. This so-called "liquidity effect" will in general be more pronounced when markets are more segmented, and when expectations are less than perfectly "rational". 59. There is some initial evidence that this "liquidity effect" may be at work in this case, with a particular twist. The first thing to notice is that although changes in the rate of monetary expansion (as measured by 54 Section 3 MI) did not seem to be a significant factor when the process started, during the second part of 1997, there was a significant change in the contribution of the two factors that determine the quantity of nominal money, i.e., the monetary base on the one hand, and the "money multiplier" on the other, as it is clear from the graphs in Figure 3.4. While the rate of growth of the monetary base rose from about 5 percent per year in April of 1997 to over 20 percent per year in March of 1998, the money multiplier fell from around 1.16 to around .95, i.e., a fall of almost 20 percent. 60. The money multiplier is of course the result of two basic ratios: the currency/deposits ratio and the reserve/deposits ratio."6 In our case, both of these ratios contributed to the fall of the money multiplier (i.e., both of them rose during the second part of 1997). Notice, as it is clear from the graph depicting these two ratios, that changes in both have been of comparable magnitude; in fact, during the period of the subsample (Jan 1994 to May 1998) the average monthly change in the currency/deposits ratio was -.004619, and the same measure for the reserves/deposits ratio was significantly lower (in absolute terms), at a value of .0002. The important thing to notice here, though, is that similar changes in the reserves/deposits ratio have a considerable larger impact on the multiplier than those in the currency/deposits ratio. More specifically, absolute changes in the money multiplier (denoted here as "x") can be written as a weighted average of changes in the two ratios, dx d(c/d) ax + d(r/d) ox dt dt O(c/d) dt O(r/d) The value of the two "weights" is " 1a -x a cld)Kc:rl a( rld) c+r and for the sample period January 1994 to May 1998, the average of these two weights were -.0065 and - .052, respectively. 61. The graph in Figure 3.5 shows, for the subsample period, the discrete one-month equivalent of the three terms in the above expression, i.e., the contribution that each of these two "weighted" changes made to changes in the multiplier. It is obvious even from visual observation that changes in the multiplier trace almost exactly changes generated by the reserve/deposits ratio, with the contribution of changes generated by the currency/deposits ratio being negligible --this is true not only for the period after June 1997 (shown as a shaded area in the corresponding graph) but also for the whole subsample period of the estimation. 55 Macroeconomic Policy Figure 3.5 Money and interest indicators 0.1 1. 0 | t; > .1.............. .... 1.2 .. i ..... l 6 ..... ...... 0.120 1.0 0~ ~ ~ . 1\ . 0~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~.0 . . ......f ti........... - 0 .0 0.0 0.8 0 . 0 4 1 . . . . . .. . . . 0. 6 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ 1 994 1 995 1 99 6 1 997 1998 1994 1995 1996 1997 1998 - Spread on 90 days rate -e-ld Ratio rid Ratio 0. 20- 0.15 ..... 0 .1 0 .------ ---- ---------- 0 . 05 ------- ---- ....... 0.10 199 J 194 1995 1996 1997 1998 -a- Change in Multiplier A--weighted Change in cld Ratio ----- Weighted Change r/d Ratio 62. These changes in the reserves/deposits ratio seem to be related to changes in reserve requirements enacted by the CB in its efforts to curtail "excess liquidity", i.e., to offset increases in the monetary base. In other words, the hypothesis that seems to be consistent with most figures is that during the period there was an important monetary base expansion (possibly in order to satisfy fiscal requirements), with the CB trying to minimize its expansionary effects via induced changes in the money multiplier --together with a 16 percent interest rate on CP's ("certificados de participaci6n"). This effort imposed an "squeeze" on commercial banks' liquidity and increased interest rates, as well as the spread. 56 Section 3 Table 3.4 Estimation results: Lending 90 days interest rate LS /l Dependent Variable is Lending 90 days Interest Rate Sample(adjusted): 1994:01 1998:05 Included observations: 53 after adjusting endpoints Convergence achieved after 21 iterations Variable Coeffic Std.Error t-Statist Prob. Inflation 0.031 0.154 0.205 0.839 Inflation (-1) 0.186 0.172 1.079 0.287 Inflation (-2) -0.027 0.177 -0.155 0.878 Inflation (-3) -0.093 0.153 -0.606 0.548 Multiplier -0.010 0.036 -0.264 0.793 Multiplier (-1) -0.086 0.036 -2.406 0.021 Multiplier (-2) -0.103 0.035 -2.904 0.006 Multiplier (-3) -0.135 0.036 -3.737 0.001 C 0.575 0.065 8.880 0.000 AR(1) 0.738 0.100 7.397 0.000 R-squared 0.943 Mean dependent var 0.249 Adjusted R-squared 0.930 S.D. dependent var 0.039 S.E. of regression 0.010 Akaike info criter -8.982 Sum squared resid 0.005 Schwarz criterion -8.611 Log likelihood 172.830 F-statistic 78.338 Durbin-Watson stat 1.974 Prob(F-statistic) 0.000 63. In order to verify whether this hypothesis could be dismissed off-hand, a lagged regression was run of the nominal 90 days lending rate on two variables (the inflation rate and the value of the money multiplier) using monthly observations from January 1994 to May 1998. The regression output (which is somewhat surprising) is reported in Table 3.4. The results for the multiplier, except for the contemporaneous month, are all significative and of the expected sign: a fall in the money multiplier is associated with a rise in the interest rate. In fact, the multiplier seems to have (in the short run) a higher explanatory power than the inflation rate. Figure 3.6 shows the actual and the forecast behavior of the 90 days lending rate, as well as the deviations. 57 Macroeconomic Policy Figure 3.6 Estimates of the demand for money 4.8 . 4. 6, _J 4. 4 -( .............. .... 4.2-. 4.0 92 93 94 95 96 97 a LN Real ml--v-- Regress Forecast 64. In principle, the policy conclusion that emerges is in accordance with the general tone of the monetary policy recommendations in the report, i.e., the need for adopting a rule for the management of one of the variables that the CB can control directly (the monetary base, in particular); and also that changing reserve requirements is not an advisable instrument. In this particular episode, it seems that the rise in interest rates and spreads have been the price paid for offsetting the effects of a rise in the monetary base on the quantity of money. Of course, a more fundamental question is why the monetary base had to increase in the first place. E. The Long Run: Some Considerations on the Future of the Monetary System 65. During the last seven or eight years a combination of fiscal restrain, favorable conditions and the rapid growth of specific sectors has allowed for a very satisfactory macroeconomic performance. Macroeconomic management has been, during these years, relatively easy. In the future, the role of macroeconomic policy could be to provide an environment in which the ensuing process of reform can take place without the hindrance of global instability and uncertainty. As crises arise (and they will, simply as a matter of probabilities) the more important a clear set of monetary policy rules (as opposed to discretionality) will become. 66. A review of the main possible vulnerabilities concerning macroeconomic variables reveals some specific areas of concern of which the future of fiscal discipline arises as the main one. In the medium and long run, the main point of concern is the lack of political consensus, and the presence of the "political cycle". The current political configuration indicates that for the foreseeable future political competition among various parties will have an effect, as it happens in many countries, on the level and structure of government expenditures, i.e., on the primary deficit. If this is inevitable, the question is how to devise a mechanism that can isolate macroeconomic variables, and that at the same time injects transparency in the system, with a clear and immediate reckoning of the cost of public funds (via taxation or central government borrowing). Of course, such a mechanism requires a wide political consensus. The current political schedule, in a period in which national priorities start being discussed before the upcoming elections, provides an opportunity for exploring the possibilities of such consensus, that would shift the "political battlefield" away from 58 Section 3 expansionary monetary policies and towards more transparency and/or competition for scarce public funds. 67. CB independence is .often cited as the major requirement for isolating monetary variables from decisions of the central government. Ultimately, and for a variety of reasons G CB independence is not sufficient. 68. Some countries have adopted a currency board with a parity fixed by law (Argentina being the primary example), and by doing so raised the cost of a disastrous fiscal situation, that could eventually force a change in the law or elimination of the board, with a high political cost. 69. Currency boards are often seen as a means of increasing credibility and reducing country risk, and therefore the country risk premium on world interest rates facing domestic residents. This is only one manifestation of its effects. The main purpose of the regime is to tie the hands of both the CB and the fiscal authority, so as to remedy the fact that sovereign governments cannot undertake legal commitments. The currency board accomplishes the same by raising the political cost of lack of fiscal discipline that would force the abolishment of the board. It seems that this, or an alternative form of covenant among various political parties would not be out of reach and have a very high payoff. 70. An alternative is outright "dollarization", a possibility that is being actively discussed recently in some South American countries. 18 Besides the natural political and national pride obstacles to dollarization, two objections have been raised, which on closer examination are found to be invalid. The first is that "seigniorage" (i.e., the flow of revenues from the creation of monetary base) is lost. In most cases, and in particular in the case of the DR, a simple calculation shows that the amount of seigniorage at low inflation rates is negligible."9 In fact, what is lost is the possibility of using seigniorage in any significant way, i.e., with high inflation rates. The second (and this, as opposed to the first, applies also to a strict currency board) is that the CB loses its role as a "lender of last resort". As mentioned before, assistance to the financial sector is costly, and the ability of the CB to pay for it via the inflation tax (a rather inefficient, disruptive and regressive tax) almost guaranties that this would be the outcome. The absence of this ability to create liquidity via the creation of "outside money" can indeed contribute to more efficient and transparent financing.20 71. The adoption of a currency board or outright dollarization for the DR is not a likely choice in the near future. Nevertheless, its discussion would have a high payoff in terms of bringing up all the questions associated with the need to isolate monetary variables from central government deficits. Also in this regard, an intermediate positive step is to legalize (i.e., make contracts in foreign exchange enforceable in a court of law) and eliminate obstacles to the adoption of foreign exchange as a unit of account. F. Some Considerations on CB Independence and the Conflict between Fiscal and Monetary Policy 72. Throughout this analysis the importance of the aggregate budget constraint has been stressed, encompassing both the central government (including its various agencies) and the CB. As mentioned before, this approach reflects the extreme view that monetary policy will ultimately end up accommodating fiscal deficits, i.e., that in the final analysis the fiscal authority (congress and the executive branch) will prevail over the monetary authority (the CB). If it is assumed, as is often the case, that the former shows a proclivity towards fiscal deficits, and the latter a tendency towards price stability, then this extreme view can be qualified as "pessimist", at least as far as a stable macroeconomic management is concerned. This is the view taken by the market, more often than not, even in highly developed economies. 73. An opposite polar view is to assume that the monetary authorities will at all times be able to conduct monetary policy independently of fiscal demands, perhaps after transferring to the public sector whatever 59 Macroeconomic Policy revenues from money creation results from the rate of monetary growth determined by the CB. In this case, the correct approach is to consider the budget constraints of the two entities separately. A consistent fiscal program, then, would be the result of the path of the primary deficit to conform to the stock of the public sector debt and therefore the flow of interest payments, in such a way that indebtedness will eventually converge to a finite sustainable value.2" 74. The question of which of these two views (or some mix of the two) is appropriate is a highly debatable one, and "political economy" considerations come into play. A powerful argument for the first of these two views (i.e., the view taken in this report) is the presence of "time inconsistency": it is optimal for the monetary authority to announce its intentions of not providing financing for the fiscal deficit, but once the public sector has engaged in a fiscally unsustainable course, it is optimal to "bail-out" government by generating an unanticipated increase in the money stock, and hence in inflation.22 75. Notice that, in spite of claims about the independence of monetary policies, there are instances in which the accommodation of monetary policy to fiscal needs takes place in a hidden manner. The classical example is a policy of targeting the interest rate on the part of the CB. When the treasury increases its net borrowing interest rates tend to rise, and as a response the CB increases the monetary base (i.e., injects liquidity via an increase in net lending), so that the final result is very much the same as if the CB would have purchased the additional government bonds in the first place. 76. Despite the strong arguments in favor of the "pessimist" view outlined above and used in the report, it is clear that anything that would increase CB independence and discourage financing of government deficits is a positive development from the viewpoint of price stability, if it decreases the temptation to run high primary deficits. Notice, for example, that a similar problem to the time consistency question raised before, with ultimately the same consequences, arises when, as it currently happens in the DR, the CB, rather than the public sector, is forced to do the borrowing, as a result of lack of appropriate "good" reputation of the fiscal authority. Although it would ultimately make no difference, it is true that easy borrowing on the part. of the CB (what could have less of a short-run impact on inflation) would encourage a lax fiscal policy. In this sense, it would be far better to institutionally force the central government to do its own borrowing, which in the case of the DR would require establishing a "track record" and building its reputation. Endtotes I The central government also has arrears in the payment to suppliers, on which there is no firm estimate. 2 The trend is derived using the Hodrick-Prescott filter. 3 The traditional distinction between "domestic" and "foreign" debt refers, really, to debt denomimted in either domestic or foreign currency. Although it is sometimes useful to separate them, it is noted that more and more frequently, particularly in countries with a history of high and variable inflation, all debt is either denominated in foreign currencies or "indexed" to some price index, i.e., denominated in real terms. 4 What is not trivial, for example, is the fact that more net borrowing today, in the context of forward looking markets, may well trigger the anticipation of future inflationary financing tomorrow, and therefore have very similar effects to outright monetary expansion today. 5 The same 1998 Monetary Program, for example (page 2), reports that "At the beginning of 1997 the Monetary Board (Junta Monetaria) adopted measures geared to eliminate pressures on the exchange 60 Section 3 rate". 6 Perhaps, as a slightly irreverent note, it could be pointed out that non-committal expressions such as "consistent with", or "compatible with", are favorite among CBs (both of the "south" and of the "north"). From an operational viewpoint they are pretty much devoid of meaning. 7 Or, more generally and amounting to the same, it expands the monetary base by a higher level than would have done otherwise, and purchases more foreign exchange (or sells less) than would have otherwise done. 8 This might indeed be justified as a way to change the composition of assets towards those that can be used quickly in an emergency. If in this case, the authorities need to be aware of the cost they are paying for this insurance. 9 Of course, the market would otherwise "learn", and eventually discover these decision rules. But this may be a long process that can be avoided. 10 Not all the real monetization was due to growth, though. During the first part of the period sta ting in 1992, it resulted also from the once-and-for-all increase in the demand for money due to the substantial reduction of the inflation rate. I I Notice that, in the restricted computation implied by this elasticity, a percentage change in R (absolute real revenue) is equivalent to the same percentage change in R/y (real revenue as a proportion of GDP), because the elasticity refers to a change in the rate of growth of real GDP, for its same level. 12 More specifically, the "semilogarithmic elasticity" of ml with respect to the inflation rate. 13 This is the counterpart of the result reported in expression [A8] in Appendix A. 14 More specifically, the zero growth total revenue is .816 per of GDP; with a 6 percent growth, the revenue rises to 1.86 percent of GDP, of which 1.046 percentage points are due to the growth factor. 15 The 90-day lending rate is taken as a reference. Similar patterns were followed for 180 and 360 days rates. 16 Strictly, the money multiplier relevant for Ml should consider only hose commercial banks reserves maintained in account of demand deposits. The total of reserves has been taken instead. This shortcut does not invalidate the significance of the changes. This is of course the reason why the reserves/deposits ratio is at times larger than one and the multiplier less than one. 17 One being, of course, the "time inconsistency" argument: once the central government has acquired debt that cannot sustain, it is optimal for the CB to bail the central government out. The moral hazard problem intrinsic to the financial system vis-a-vis the CB shows up also between the central government and the CB. 18 The term "dollarization" is used for short. In fact, the new Euro could be adopted with similar effects. 19 A cursory view at the numbers discussed in Section 3.2 show, for example, that at an inflation rate in line with the current rate in the US (say, around 2 or 3 percent per year), with a growth rate of GDP of 6 percent, the total revenue would amount to about one half of one percent of GDP. 20 On this point, see S.Fischer, "An International Lender of Last Resort", address to the American Economic Association, New York, January 1999. Also available as IMF document. 61 Macroeconomic Policy 21 Of course, the same would apply to the CB, if there is an initial stock of the bank's negative assets. 22 This is nothing else than a straightforward extension of the well-known time inconsistency problem of monetary policy as analyzed, for example, by Auemheimer ("The Honest Government Guide to the Revenues from the Creation of Money", Journal of Political Economy, 1974) and more explicitly by Calvo ("The Inconsistency of Monetary Policy", Econometrica, 1978).Indeed, the problem exists even if the central government does not need to be bailed-out, if taxes are distortionary. 61A SECTION 4: COMMERCIAL POLICY A. Introduction 1. The Dominican Republic constitutes an interesting case for commercial policy evaluation because an escalating import tariff rate structure and non-tariff import barriers aimed at protecting import- substitution activities coexist with successful export processing zones and tourism. Exports from these zones account for about 80 percent of total exports of goods and their annual growth rate has been about 26 percent in the last fifteen years. 2. As is well known, free trade status to exportable activities tends to reduce the anti-export bias of commercial policies, as tariffs are not applied on imported inputs. Commercial policies could still have a negative impact on these activities as they tend to appreciate the so-called real exchange rate and thus to reduce the competitiveness of the country in international markets. The main question is thus the extent to which current commercial policies constitute an implicit tax on exportable activities in presence of export incentives. 3. This paper has the objective of quantifying the impact of commercial policies on resources allocation, factor returns, relative prices, and welfare in the Dominican Republic. A factor-specific general equilibrium model is used to assess the impact of current trade policies on the economy vis-a-vis a free trade policy. In addition, a Government proposal of tariff reductions submitted to Congress for approval and an alternative policy of a 10 percent uniform import duty are also evaluated. The model is presented in Section A. Section B contains a summary of commercial policies. The evaluation of commercial policies is presented in Section C and concluding remarks are in Section D. B. A General Equilibrium Model 4. The model presented in this section is very similar in structure to those of R. Jones (1963), R. Batra (1973), R. Komiya (1967), and E. Bond (1995). In Batra's and Komiya's models, equilibrium prices of non-traded goods are determined only by supply conditions with demand playing no role in price determination. In this model, those prices are determined both by supply and demand for non-traded goods as a result of assuming sector-specific primary factors. 5. The model consists of three final goods (exportable, importable, and non-traded goods), four primary inputs (labor and three sector-specific factors), and three intermediate inputs (exportable, importable and non-traded inputs). Labor is mobile across activities while capital is a sector-specific factor although it is mobile among firms within each sectoi1. Production technology is assumed to have constant returns to scale and elasticity of substitution between primary factors and inputs equal to zero. On the other hand, elasticity of substitution between labor and capital is assumed equal to one. Consumers' demand for final goods is assumed to be generated by a utility function with unitary elasticity of substitution. 6. The corresponding cost functions are thus: (1) C i = [00, .(w4 r(4J) + Ei -j pi ]-Yi M,X,H /=6 62 Section 4 where w and r, are wages and return to capital in activityj, pi's are the prices of inputs, Y, is activityj's output, and M, X, and Hrefer to importable, exportable and non-traded activities. 7. Using Shepherd's Lemma, demand for primary factors and for inputs can be derived from (1): (2) L=°° 90j ' Y As what are to be evaluated are policy changes, it would be convenient expressing equations (2) and (3) in terms of a baseline scenario to assess changes in demands for primary factors: (2') (+.)=4 d r= (3') (1+k,) ( d (4') (i + dj ) =-__= =1± + where a variable () means the ae ale of that variable in the baseline scenario. 8. Production responses to changes in relative prices are assumed to take place along the production possibility frontier so that all resources are fully employed. This full employment condition requires that demand for labor and capital be equal to their total endowments: (5) (i +lM)('IM =1 (i~){4]+(id+OH)( -=1, and as labor is mobile across activities and capital is a specific primary factor. 63 Commercial Policy 9. A second general equilibrium condition is that of zero-profits. A relationship among prices of final products and prices of primary factors and purchased inputs (vis-a-vis the baseline scenario) can then be derived from the cost functions: Cj Wj Wj n= pi p j (7) - _: o + O'l -r = C. ( iJ C1 P PJ 10. Domestically produced goods are assumed to be perfect substitutes of their internationally traded counterparts so that their domestic prices are determined by their international prices plus import tariff or minus export tax rates. Border prices are going to be normalized to be equal to unity. Thus, domestic prices can be expressed as a function of the baseline and of the new import (or export) taxes. In the case of import-competing activities, their prices will be given by: (8.1) - = and (8.2) pi = (1 + tX ) Pi 1+t, 2.8 Prices of non-traded activities are determined domestically by the market clearing condition. Demand for non-traded goods as final consumer goods (Dd ) is derived from a Cobb-Douglas utility function (U) while their demand for intermediate consumption is derived from the cost function (5): (9) U = B(D da°(D d)a'(H d)a2 It can be easily shown that the associated expenditure function (E) is as follows: (10) E P(P-, P. Ph ).U = Bpxp PM Ph U where P is a price level transforming units of the numeraire into utility units. Applying the Shepherd's Lemma, the demand for non-traded activities as final consumer goods can be obtained: (11) dE= Dd = Ba @2 (a2-1) P aOP al U , or as a function of the baseline scenario, dp h HD = 2h x H dh ( 11') 1+dh DH __ )_ Px K Pm9 I U DH Px Ph, Pm U Likewise, the demand for non-traded goods as intermediate inputs by the different productive activities can be obtained from the cost functions (1): 64 Section 4 (12) l+dhJ=-i== +yj Prices of non-traded activities are then determined by the market clearing condition DH IDHI (13) = (1 + dh)[+ E (1 + dhj) Yh yDH)D 11. Equation (13) contains changes in welfare (that is, changes in U) affecting the final demand for non-traded goods. These changes in welfare can be calculated using the national accounts identity: (14) E+S =Y+ Etj(D _Sj)+Yti(D Id _Si j I where Y is national income or revenue function S is savings SF, SI are quantities domestically supplied of final and intermediate goods 12. The revenue function is a function of prices of final and intermediate goods: (15) Y = Y(pj, pi Using Shepherd's Lemma, (16) Y =Sj supply of commodityj (17) Y1, = Si - Di= excess supply of intermediate good i Change in utility is obtained by totally differentiating the utility function and using (16) and (17). (18) ___ - [( +[U)1c }=lt ±E{) E t( it ))- ji}tt3.(1d+Jj)) -E )-_j_ ] ((i + P( +1 where - - -+ are the shares of expenditure in intermediate and final consumption, and of output of importable goods in total expenditure. 65 Commercial Policy The numerator of (18) contains the welfare gains of increasing imports if domestic prices of imported goods exceed their international prices (t, and tj >0), the gains of reducing output of import-competing activities and the cost for consumers of any tariff revenue loss; the denominator is the impact of changes in tariffs on the marginal utility of income (or expenditure). 13. The general equilibrium model thus comprises 21+n equations (2')-(8.2), (11')-(13), and (18) and the same number of endogenous variables, namely, y,, pp pb U d,c, dg, p, Ip kp w, and rj. A unique solution of the model thus exists as a function of commercial policy, technological parameters, and sectoral shares in the use of labor and in expenditure. 14. The short-run impact of protection to import-competing activities on wages is ambiguous. Nominal wages would tend to increase because demand for labor by import-competing activities increases as their domestic prices rise. This impact on wages could be offset by declines in output in the non-traded goods sector. As a result of substitution and wage cost effects, supply of non-traded goods would tend to decline and their consumption to increase, their prices would increase but the level of production may decline or increase depending upon the magnitudes of changes in supply and demand. Thus, the net impact of commercial policies on nominal wages would be a function of the size of the sectors and of their labor intensities. The impact of protection on real wages would depend on the impact of protection on the price index used to deflate nominal wages, that is, on the structure of consumer's expenditure. 15. Rental on capital in import-competing activities increases as a result of protection as demand for capital by these activities increases and of capital specificity. On the other hand, rental on capital in exportable activities would decline because the marginal productivity of capital would decline as a result of the increase in wages. Finally, the impact of protection on the rate of return on capital in non-traded activities is ambiguous as the impact of protection on the level of output is also ambiguous. 16. The output responses of import-competing and exportable activities are similar to those of general equilibrium models with perfect mobility of primary factors of production: production of importable goods increases and that of exportable goods declines. As indicated earlier, the response of production of non-traded goods is ambiguous as demand for these goods increases and their supply declines as a result of substitution effects on consumption and production. 17. The model can be used to compute changes in value-added prices of importable and exportable goods (relative to that of non-traded goods) in response to changes in commercial policies and, hence, to obtain an estimate of the anti-export bias of commercial policies. As can be seen in fig. 1, the differential between relative prices of import-substitution and exportable activities measures the magnitude of distortions affecting the volume of trade. LetD,, Dm, be the excess demand for importable goods and S, S, the excess supply of exportable goods. As before, units of goods are defined such that their internal prices are unity in equilibrium under free trade. Undistorted (free trade) equilibrium occurs at pointA with a volume of trade equal to IM where it has been assumed for simplicity that trade is balanced. Demand for importables and supply of exportables would shift when an import tariff is imposed to D.1D,, and to S. respectively as a result of substitution effects on production and consumption. In the new (distorted) equilibrium B, the relative price of importable goods would increase to (l+to), 66 Section 4 Figure 4.1 Tariffs as a tax on exports +t,~~~~~~~~~~s X / / -IO I+so i l+sO ---_iI \ D , , ~\ Mo Ml M* Quantities the relative price of exportable goods would decline to (I+so), and the volume of trade would decline to Mo. As can be seen in figure 1, the same volume of trade Mo could be achieved with an implicit tax equal to (to - s0) part of which represents a tax on exportable activities. The magnitude of the burden of this tax on exportable activities (so) is thus a measure of the so-called anti-export bias of commercial policies. The mechanism through which an import tariff is converted into an implicit export tax is the increase in the price of the non-traded good mentioned earlier. This increase in the price of non-traded goods tends also to erode in part the protection that was intended to be granted to import-competing activities. 18. The anti-export bias of commercial policies can be reduced by granting a free-trade status to producers of exportables on their imports of inputs. This would increase the supply of exportable goods to SXSX and the volume of trade would expand to MI. As a result, the relative price of exportable activities would increase from (i + s ) to (i + sj) and the implicit tax on trade would be reduced to (t1 - si). There would still be an implicit tax on these activities (si) as a result of the overall impact of protection on prices of non-traded goods that cannot be eliminated by export incentives. C. Current Commercial Policies 19. In the early 1990s, the Dominican Republic made substantial progress towards reducing barriers to trade. In 1990, imports tariffs were reduced from a range of 0-200 percent to nine rates ranging from 0 to 35 percent. Tariff exemptions granted to special sectors under special agreements with the State have been eliminated. Import prohibitions have also been eliminated with the exception of several products competing with local production2. At present, there are non-tariff barriers on imports of rice, sugar, onions, garlic, milk, red beans, poultry meat, and tomato paste. These commodities represent around 40 percent of 1995 agricultural output and 12 percent of manufacturing production. 67 Commercial Policy 20. The granting of these licenses to importers do not follow either clear procedures or criteria and the government authorities have wide discretional powers. In general, imports are authorized or prohibited depending upon the conditions of the market and they are subject to statutory tariff rates. Imports of poultry and tomatoes are exclusively granted to domestic producers. Import licenses for garlic, onions and potatoes are granted to registered importers only when there are domestic supply shortages. 21. The Government has reached agreement with the World Trade Organization (WTO) on increasing until the year 2005, on year-to-year basis, the quantities that can be imported under the quota system3. According to this agreement, the volumes of onions, red beans, sugar, garlic, and milk to be imported will be increased by 37 percent by the year 2005 compared to 1997 levels; the quota volumes for corn, poultry and rice will be increased by about 40, 53, and 23 percent, respectively. This reduction in protection rates would be reinforced if a government's proposal for further reduction in import tariffs is implemented (see paragraph 23 below); this overall reduction in tariff rates (and in prices of import- competing activities) would tend to reduce the demand for goods subject to the quota system as a result of cross-substitution effects. 22. The tariff equivalents of these barriers have been quantified for the commodities for which data is available by comparing producers' prices with border prices (see Table 4.1). Table 4.1 Tariff equivalents of non-tariff barriers (%) Commodity Statutory Rate Tariff Equivalent Corn 5.00 85.17 Red Beans 25.00 69.55 Onions 25.00 37.87 Garlic 25.00 34.53 Poultry 25.00 74.82 Pasteurized Milk 20.00 48.36 Milk in Powder 20.00 53.36 Polished Rice 20.00 43.00 Raw Sugar 15.00 42.11 Refined Sugar 15.00 37.67 23. Regarding domestic indirect taxation, there is discrimination against imports in value added and excise taxation. The current value added tax rate is 8 percent and commodities such as rice, tomato sauces, condiments, cheese, bread, wheat, processed cereals, sausages, fresh, chilled and frozen meats, detergents, toothpaste, matches, salt, and primary sector commodities are exempt from the tax if they are locally produced. Imports of beer and alcoholic beverages, and cigarettes are subject to a 30 percent excise tax rate but local production is subject only to a 10 percent tax rate. If exempt from value added and excise taxation, producers would find it more profitable to increase their production vis-a-vis non- discriminatory domestic indirect taxation. This would result from the increase in producers' prices because consumers would have to pay the international price plus all taxes on imports (import and domestic indirect taxes) but producers would not get any reduction in their prices because of the exemption on domestic taxes. 24. A summary of output-weighted average nominal protection rates granted by current commercial policies to import-substitution activities is presented in Table 4.24 Import-competing agriculture is protected at a nominal rate of around 40 percent5 and its value added represents around 30 percent of the sectoral GDP. In the import-substitution manufacturing sector, activities such as grain mills (mostly rice milling), sugar refining, food processing, beverages, tobacco manufacturing (mostly cigarettes), wearing 68 Section 4 apparel and footwear are all protected at nominal rates ranging from around 30 to 45 percent. Not surprisingly, the value added of these activities represents about 80 percent of total value added of import competing manufacturing (about 16 percent of GDP that excludes public administration). Table 4.2 Nominal protection rates (%) Activity Protection Rate Agriculture* 39.56 Import-Substitution Manufacturing 33.02 Grain Mills 44.54 Food Processing 40.98 Beverages and Tobacco Manufacturing 45.16 Textiles and Wearing Apparel 32.00 Footwear and Leather Manufacturing 35.00 Petroleum Refineries 5.90 Chemicals 16.37 Rubber and Plastic Products 24.00 Non-Metal Products 18.88 Metal Products 20.00 Other Manufacturing 17.37 Intermediate Goods ** 15.69 Consumer Goods 28.53 Source: 1997 customs data and 1995 production matrix. Note: (*) Excludes paddy, livestock, forestry and fishing that are included in the non-traded group of commodities. (**) Includes imports of capital goods. 25. The government has submitted legislation to Congress to further reduce tariffs and to eliminate discrimination against imports in the value added and excise taxation. In the proposed legislation, the maximum tariff rate would be reduced to 15 percent in a period of two years and the number of rates reduced to four (0-5-10-15 percent). The proposal includes a transitional tariff structure by which the maximum import tariff would be reduced to 20 percent and the number of tariff rates reduced to five (0, 3, 8, 14 , 20 percent). 26. Nominal output-weighted protection rates as proposed by the government have been calculated for import-competing activities and they are presented in Table 4.3. Data in column (a) contains the proposed tariff rates per economic activity maintaining current import licensing for the aforementioned commodities; column (b) assumes the elimination of these quantitative restrictions. In both cases, there would be a substantial reduction in nominal protection rates, particularly for food processing and grain mills. 69 Commercial Policy Table 4.3 Proposed tariff reductions (%) Activities Protection Rates Protection Rates (a) (b) Other Agriculture 22.63 15.00 Grain Mills 32.61 15.00 Food Processing 25.78 15.00 Beverages and Tobacco Manufacturing 15.00 15.00 Textiles and Wearing Apparel 13.11 13.11 Footwear and Leather Manufacturing 14.91 14.91 Petroleum Refineries 0.23 0.23 Chemicals 5.48 5.48 Rubber and Plastic Products 9.24 9.24 Non-Metal Products 10.27 10.27 Metal Products 5.90 5.90 Other Manufacturing 6.23 6.23 Intermediate Goods 3.36 3.36 Consumer Goods 15.30 11.92 Notes: (a) Existing QRs are maintained. (b) QRs are eliminated. D. Commercial Policy Evaluation 27. The general equilibrium model is solved using the coefficients of the 1995 input-output matrix for the Dominican Republic. The matrix contains 37 sectors (excluding public administration) of which 13 are classified as import competing activities (other agriculture, grain mills, sugar refining, food processing, beverages and tobacco manufacturing, textiles, footwear, petroleum refineries, rubber and plastic products, chemicals, non-metal products, metal products and other manufacturing), 8 as exportable activities (exportable agriculture, mining, food processing, textiles, tobacco manufacturing, footwear, and other activities in export processing zones, and hotels and restaurants), and 16 as non-traded activities (paddy, livestock and services excluding public administration and hotels). The non-traded service sector is aggregated into one activity and sugar refining and food processing are also merged together so that the input-output matrix used in the calculations contains 23 activities. 28. As mentioned earlier, nominal protection rates include tariff rates, value added and excise differentials, and the calculated tariff equivalents of quantitative restrictions. The source of tariff rates is 1997 customs data containing CIF value of imports and tariff revenue so that actual tariff rates could be calculated. In order to calculate tariff rates on imports of inputs and capital goods, imports were classified according to the Broad Economic Categories (BECf of the United Nations by mapping the codes of the Customs Harmonized System (HS) into those of BEC; another mapping was used to link the local codes of national accounts with those of HS7. As a result, customs data could be matched with input-output data and tariff rates allocated to the different categories of inputs and capital goods used in production by the economic activities. 29. The current commercial policies summarized in Table 4.1 are used as the baseline scenario and compared with alternative commercial policies, namely, free trade, the Government's proposal of tariff reductions, the proposal cum elimination of quantitative restrictions, and a 10 percent uniform import duty8. Units of goods, inputs, and primary factors are defined so that their prices are equal to unity in the baseline scenario. 70 Section 4 30. The impact of the alternative scenarios on the economy is presented in Tables 4.4 to 4.10. As indicated in Section II, one consequence of protection is that exportable activities are subject to an implicit tax that can be reduced but not eliminated by granting free trade status to their producers. It is estimated that, under the free trade scenario, the relative value added prices of exportable activities would increase by 27 percent (see Table 4.4) and this would reflect the removal of the implicit tax. According to the calculations, exportable activities would currently pay around 55 percent of the total taxation of foreign trade (48 percent)9. 31. Nominal value added prices of non-traded agriculture (paddy and livestock) would substantially fall (46 percent for paddy, and 50 percent for livestock) as a result of trade liberalization'°. As most of the local demand for these commodities is for intermediate consumption by import competing activities, the reduction in (effective) protection granted to these activities is the main cause behind this decline. 32. As indicated in Section II, the impact of changes in commercial policies on real wages depends upon consumers' pattern of expenditure. Based on input-output data, it is calculated that real wages would increase by about 5 percent in spite of a reduction in nominal wages by around 13 percent as the result of a decline in the consumer price level of around 17 percent. On the other hand, return to capital in import-substitution activities would decline with the elimination of protection, and increase in export- oriented activities (see Table 4.5). This clearly indicates that protection redistributes income against labor, at least in the short run. As a result of free trade, the increase in consumers' welfare is calculated at around 8 percent over that of current commercial policies. Table 4.4 Free trade scenario - Impact on relative prices, wages, and welfare (%) Impact of Reforms on: % Change Relative Value Added Prices: Import-Substitution Activities -21.67 Exportable Activities: 26.68 - Commodities 21.66 - Tourism 33.47 Implicit Tax on Trade -48.35 Value Added Prices of Non-Traded Goods: -13.86 Paddy -45.90 Livestock -50.17 Services -11.08 Nominal Wages -12.80 Real Wages 4.90 Consumer Price Index -16.87 Consumers' Welfare 7.80 Notes: Value added prices are averages of those of the activities weighted by the sectoral shares in total value added. Real wages are nominal wages deflated by a consumer price index with weights given by the input-output matrix. 33. Under this scenario, the increase in output of export processing zones (see Table 4.5) would range between 16 (food processing) and 50 percent (footwear and leather manufacturing). Agriculture and mining outputs would increase by around 7 and 13 percent, respectively, reflecting the fact that these commodities are intensive in the use of natural resources" . Most of the outputs of import-competing activities would decline as a result of free trade; this decline would range between 6 (import competing agriculture) and 53 percent (footwear for the domestic market)'2. 71 Commercial Policy 34. The size of the response of outputs of exportable activities such as footwear and tobacco processing is not surprising as these activities are labor-intensive (labor-capital ratios are 2.53 in tobacco processing and 2.57 in leather manufacturing compared to an average of about 1.5 for the whole export processing sector) and their production costs would be thus reduced vis-ii-vis other exportable activities. Given the assumption of full employment of the labor force, these activities would need to expandvis-a- vis others in order to absorb the labor released by import-competing activities. Table 4.5 Free trade scenario - Impact on output, employment and capital rentals (%) Activities Changes in Changes in Changes in Output Employment Return to Capital Rice -26.40 -54.40 -60.20 Exportable Agriculture 6.80 24.40 8.50 Other Agriculture -6.10 -26.50 -35.90 Livestock, Forestry & Fishing -9.10 -48.00 -54.70 Mining 13.10 55.50 35.60 Grain Mills -28.90 -48.10 -54.70 Food Processing -17.10 -41.00 -48.60 Beverages & Tobacco Manufacturing -9.50 -42.50 -49.90 Textiles & Wearing Apparel -26.80 -49.50 -56.00 Footwear & Leather Manufacturing -52.70 -67.60 -71.70 Petroleum Refineries 3.80 20.90 5.40 Chemicals -3.80 -10.90 -22.30 Rubber & Plastic Products -7.10 -20.40 -30.60 Non-Metal Products -7.50 -12.80 -24.00 Metal Products -16.50 -36.00 -44.30 Other Manufacturing -7.10 -14.20 -25.20 Food Processing (EPZs) 15.60 34.00 16.80 Tobacco Manufacturing (EPZs) 43.90 66.10 44.80 Wearing Apparel (EPZs) 26.70 48.10 29.10 Footwear & Leather Manufacturing(EPZs) 49.50 74.90 52.40 Other Manufacturing (EPZs) 12.40 31.10 14.30 Hotels & Restaurants 0.00 0.00 31.60 Other Services 1.00 3.00 -10.20 35. If sector-specific capital were intemationally mobile, and rates of return were equalized for the same activities across countries in the baseline scenario, the increase in capital rentals in exportable activities would certainly attract foreign investment until capital rental differentials are eliminated. This would reinforce the short-run effect of trade liberalization on output of exportable activities. As a result, their demand for labor would increase further as capital and labor are complements in production. By the same token, the decline in output of import competing activities would be enhanced as a result of the reduction of the stock of sector-specific capital. The impact of changes in the stock of capital on nominal wages is thus ambiguous as it would depend upon the magnitudes of the changes in sectoral demands for labor. 36. The simulations of the impact of the Government's trade liberalization proposal on the economy are presented in Table 4.6. Relative value added prices for exportable activities would increase by about 12 percent. The welfare gains to consumers under this scenario (+1.9 percent) would be well below those of the free trade scenario. The difference that the govemment's proposal could make in the economy is given by the elimination of existing non-tariff barriers. If these quantitative restrictions are eliminated, then the anti-export bias would be further reduced by 3 percentage points and consumers' welfare would 72 Section 4 increase by about 2 percentage points (from 1.9 to 4.2 percent). This could be achieved without any decline in real wages'3. Table 4.6 Impact of trade reforms on relative prices, wages and welfare (%) Impact of Reforms on: (a) (b) Relative Value Added Prices: Import Substitution Activities -9.20 -12.31 Exportable Activities 12.20 15.03 - Commodities 8.77 11.35 - Hotels 16.82 19.99 Implicit Tax on Trade -21.31 -27.34 Value Added Prices of Non-Traded Goods: Paddy -9.12 -21.30 Livestock -14.96 -26.70 Services -5.60 -6.70 Nominal Wages -5.30 -7.40 Real Wages 1.80 2.60 Consumer Price Index -7.00 -9.75 Consumers' Welfare 1.90 4.20 Notes: (a) Existing QRs are maintained; (b) QRs are eliminated. 37. The response of supply of exportable goods would be stronger (see Table 4.7) if existing quantitative restrictions are eliminated and import tariffs in this scenario are reduced. In particular, outputs of footwear and leather manufacturing, and tobacco manufacturing would expand by 25 and 23 percent, respectively, compared with the baseline scenario. This would mostly be the result of releases of labor by paddy (-37 percent), food processing (24 percent), and rice milling (32 percent) activities caused by reductions in their value added prices. 73 Commercial Policy Table 4.7 Impact of trade reform proposals on output, employment and capital rentals (%) Changes in Output Changes in Changes in Activities Employment Return to Capital (a) (b) (a) (b) (a) (b) Rice -2.60 -16.80 -6.60 -37.40 -11.50 -42.00 Exportable Agriculture 2.90 3.90 10.20 13.70 4.30 5.30 Other Agriculture -2.20 -3.50 -10.10 -16.20 -14.90 -22.30 Livestock, Forestry and Fishing -1.80 -5.00 -11.80 -29.70 -16.50 -34.80 Mining 6.90 8.20 27.20 32.90 20.40 23.10 Grain Mills -2.80 -18.20 -5.30 -32.10 -10.30 -37.10 Food Processing -3.30 -9.20 -8.90 -23.70 -13.80 -29.30 Beverages and Tobacco Manufacturing -6.30 -5.80 -30.50 -28.40 -34.20 -33.70 Textiles and Wearing Apparel -16.50 -14.70 -32.60 -29.40 -36.30 -34.60 Footwear and Leather Manufacturing -32.60 -29.50 -44.80 -40.90 -47.70 -45.20 Petroleum Refineries 0.90 1.50 4.70 8.20 0.90 0.20 Chemicals -3.10 -1.90 -8.80 -5.60 -13.70 -12.50 Rubber and Plastic Products -4.80 -3.80 -14.20 -11.30 -18.70 -17.80 Non-Metal Products -1.60 1.60 -2.80 2.80 -8.00 -4.80 Metal Products -13.10 -11.10 -29.30 -25.20 -33.10 -30.70 Other Manufacturing -3.90 -1.60 -8.00 -3.40 -13.90 -10.50 Food Processing (EPZs) 6.10 8.50 12.60 17.70 6.60 9.10 Tobacco Manufacturing (EPZs) 15.80 22.50 22.70 32.80 16.20 23.00 Wearing Apparel (EPZs) 10.20 14.20 17.50 24.60 11.30 15.50 Footwear and Leather Manufacturing (EPZs) 17.90 25.30 25.80 36.80 19.20 26.80 Other Manufacturing (EPZs) 4.90 6.80 11.70 16.40 5.80 7.80 Hotels and Restaurants 0.00 0.00 0.00 0.00 19.10 20.50 Other Services 0.30 0.70 0.90 2.10 -4.50 -5.40 Notes: See Table 4.6. 38. If there are fiscal restrictions to trade liberalization policies requiring taxation of trade, the country would do much better in terms of consumers' welfare with a 10 percent uniform import tariff than under the government's proposals. Welfare of consumers would increase by about 5 percentage points over that of the baseline scenario. Under this alternative, relative value added prices of exportable activities would increase by about 17 percent and the remaining taxation of exports would be about 10 percentage points vis-a?-vis the free trade scenario. Table 4.8 10% Uniform import duty- relative prices, wages and welfare (%) Impact of Reforms on: % Change Relative Value Added Prices: Import-Substitution Activities -15.68 Exportable Activities: 16.83 - Commodities 13.67 - Tourism 21.09 Implicit Tax on Trade -32.51 Value Added Prices of Non-Traded Goods: Paddy -24.70 Livestock -32.70 Services -7.40 Nominal Wages -9.50 Real Wages 1.36 Consumer Price Index -10.71 Consumers' Welfare 5.10 74 Section 4 39. The supply responses of the activities to a 10 percent uniform import duty are presented in Table 4.9. It should not be surprising that the supply response of exportable activities is higher than under the two previous scenarios as a result of the further reduction of the anti-export bias. In particular, outputs of footwear and leather manufacturing, and tobacco manufacturing would increase by around 34 and 30 percent, respectively. The increase in employment in these two sectors would range between 44 and 50 percent. Activities such as paddy and livestock would release labor by around 47 and 40 percent of their current employment levels, respectively; import-competing footwear would reduce its demand for labor by about 60 percent. Table 4.9 10% Uniform import tariff - Impact on output, employment and capital rentals (%) Activities Changes in Changes in Changes in Output Employment Return to Capital Rice -22.10 -47.10 -52.10 Exportable Agriculture 4.40 15.40 4.50 Other Agriculture -4.60 -20.70 -28.30 Livestock, Forestry & Fishing -6.80 -38.50 -44.40 Mining 7.40 29.40 17.10 Grain Mills -24.00 -41.00 -46.60 Food Processing -12.60 -31.60 -38.10 Beverages & Tobacco Manufacturing -7.80 -36.50 -42.60 Textiles & Wearing Apparel -22.80 -43.30 -48.70 Footwear & Leather Manufacturing -46.10 -60.50 -64.20 Petroleum Refineries 5.80 33.30 20.60 Chemicals 0.20 0.50 -9.00 Rubber & Plastic Products -4.40 -12.80 -21.10 Non-Metal Products 1.40 2.40 -7.30 Metal Products -9.40 -21.60 -29.00 Other Manufacturing -0.40 -0.90 -10.30 Food Processing (EPZs) 11.10 23.50 11.80 Tobacco Manufacturing (EPZs) 30.10 44.40 30.70 Wearing Apparel (EPZs) 18.50 32.50 20.00 Footwear & Leather Manufacturing(EPZs) 33.50 49.40 35.20 Other Manufacturing (EPZs) 8.70 21.50 10.00 Hotels & Restaurants 0.00 0.00 19.00 Other Services 0.90 2.70 -7.00 40. The responses of exports of mining, exportable agriculture, and export processing zones are presented in Table 4.10 using 1995 data as the base year. As the model used in the simulations is static in the sense that it does not explain any long run trend of the endogenous variables, the export response should be understood as a once-and-for-all reaction to changes in commercial policies over any trend value. The projections show that under free trade, exports of these commodities would expand by about 25 percent in the short run. If the government's trade reform proposal were implemented, then exports would expand by only about 9 percent. Again, the country could do better in terms of export performance under the 10 percent uniform duty scenario as exports would expand by about 17 percent. Table 4.10 Projected export responses (%) Scenario Agriculture Mining EPZs Total Free Trade 46.79 11.65 25.74 25.44 Government's Proposal 2.46 4.10 9.74 9.06 Elimination of QRs 19.65 5.20 13.59 13.19 10% Uniform Import Duty 29.00 4.06 17.76 17.13 75 Commercial Policy E. Main Findings 41. It has long been recognized that tariffs and non-tariff barriers on imports reduce the volume of trade. As the balance of trade is largely unaffected by commercial policies, what happens to imports will also happen to exports. Simple accounting shows that if commercial policies reduce imports, then, the need for foreign exchange to purchase them is also diminished; consequently, exports will also be reduced as they are the ultimate source of foreign exchange. 42. This study only gives a quantitative evaluation of short run export supply response to changes in commercial policies. A diversity of dynamic effects has not been incorporated in this commercial policy evaluation because of the practical ways of incorporating them into the analytical framework used in the report. Following a more dynamic approach would have meant incorporating more assumptions and guesses about key economic and technological parameters. 44. With all its limitations, this study shows that although existing export incentives can compensate the impact of commercial policies on production costs of exportable commodities, the overall impact of protection policies on output of exportable activities can still be substantial. The main reason is that protection results in increases in prices of non-traded goods and in nominal wages. As a consequence, the value-added prices of exportable goods (relative to those of import-competing and non-trade activities) decline and their output is reduced vis-a-vis those of a non-distorted environment such as free trade. Protection is not a free lunch for the economy because it is a non-zero sum game as the result of the costs imposed on consumers and of the resources misallocation costs. Exportable activities share a great deal of the burden that current protection policies impose on the economy. 45. The quantitative evaluation of current commercial policies indicates that there is an implicit tax on exportable activities of about 27 percent although the country has been able to successfully implement tax rebates for exports in processing zones. The current government's proposal could be a significant move in the right direction towards reducing the burden of this tax but it is a timid approach as measured by its welfare gains and export response compared to available alternatives. For instance, a 10 percent uniform import tariff along with the elimination of non-tariff barriers can do better in terms of welfare gains and export performance than the government's proposal. A simple explanation for this is that if there are fiscal revenue constraints that would require taxation of the country's volume of international trade, then an uniform import duty could further reduce distortions in relative prices for consumers and producers while preserving fiscal revenue. 76 Section 4 ANNEX I. A GE Model with Endogenous Capital Response: An Illustration 1.1 In this Annex, the model of the main section is modified in order to simulate the impact of changes in the rates of return to sector-specific capital on the economy on capital accumulation in manufacturing, tourism and services activities14. This extension 'of the model specifies a simple relationship between the desired stock of capital and the real rate of return of the following type: (a. 1) Kj = b, .(1 + rj) where bko, and rj is the rate of return to capital in activityj. 1.11 In terms of changes of desired capital stock vis-a?-vis the baseline scenario, equation (a. 1) can be expressed as: d=(1+KJ)_ (i+rj) (a.2) (I + k )= I.III The results of incorporating equation (a.2) into the general equilibrium model of Section II are presented below for the free trade scenario. They show that relative value added prices of exportable activities could increase by about 20 percent and that the total taxation rate on trade is calculated at 41 percent compared to 22 and 48 percent, respectively, under given stocks of sector-specific capital. In particular, the increase in value added prices of commodities and tourism would be about 14 and 26 percent, respectively. Real wages, on the other hand, would increase by around 7 percent compared to the calculated short run response of around 5 percent. Table A4.1 Free Trade Scenario - Impact on Relative Prices, Wages and Welfare (%) Impact of Reforms on: % Change Relative Value Added Prices: Import-Substitution Activities -21.82 Exportable Activities: 19.16 - Commodities 14.07 - Tourism 26.02 Implicit Tax on Trade -40.98 Value Added Prices of Non-Traded Goods: Paddy -55.12 Livestock -77.56 Services -4.55 Nominal Wages -9.60 Real Wages 7.40 Consumer Price Index -15.82 Consumers' Welfare 12.90 I.IV Output effects of trade liberalization with endogenous capital are presented in Table A.2. According to the results, expansion of tobacco manufacturing and footwear in export processing zones would reach around 70 percent, and wearing apparel by about 40 percent as a result of additional investments. Tourism (hotels and restaurants) output would increase by about 30 percent. Exports of agricultural commodities and mining, of export processing zones, and tourism are projected to expand by 38 percent. 77 Commercial Policy Table A4.2 Free Trade Scenario - Impact on Output, Employment, and Capital Accumulation (%) Activities Changes in Changes in Changes in the Output Employment Stock of Capital Rice -36.20 -68.30 Exportable Agriculture 5.00 17.90 Other Agriculture -7.00 -29.90 Livestock, Forestry and Fishing -21.10 -80.40 Mining 9.40 38.30 Grain Mills -39.90 -47.20 -30.90 Food Processing -44.00 -55.30 -36.50 Beverages and Tobacco Manufacturing -56.40 -73.70 -51.30 Textiles and Wearing Apparel -69.90 -80.00 -57.50 Footwear and Leather Manufacturing -88.90 -92.70 -74.40 Petroleum Refineries 3.50 13.40 1.30 Chemicals -29.40 -37.50 -24.30 Rubber and Plastic Products -38.20 -49.10 -32.20 Non-Metal Products -39.10 -45.40 -29.80 Metal Products -62.90 -74.50 -52.00 Other Manufacturing -38.80 -46.60 -30.50 Food Processing (EPZs) 24.00 37.90 11.60 Tobacco Manufacturing (EPZs) 70.30 89.10 30.70 Wearing Apparel (EPZs) 41.40 58.00 19.50 Footwear and Leather Manufacturing (EPZs) 79.10 100.10 34.50 Other Manufacturing (EPZs) 19.20 33.10 9.70 Hotels and Restaurants 27.60 27.60 27.60 Other Services 0.80 6.40 -1.90 78 Section 4 STATISTICAL ANNEX Table B4.1 Tariff Equivalent of Quantitative Restrictions Commodity Producer's Price CIF Price Tariff Equivalent Corn 4282.87 2312.91 85.17 Green Beans 14.61 8.62 69.55 Garlic 31.00 23.04 34.53 Onions 11.42 8.28 37.87 Poultry 21.72 12.43 74.82 Milk in Powdera 94.84 61.84 53.36 Pasteurized Milkb 19.82 13.36 48.36 Sugar' 5489.01 3862.49 42.11 Sugard 6918.03 5025.21 37.67 Rice 8924.88 6240.66 43.01 Sources: Dominican Republic Customs data, US Customs data, and National Accounts. Notes: (a) Milk NIDO. (b) Milk RICA (uht). (c) Sugar Crema. (d) Sugar Refino. 79 Commercial Policy Table B4.2 Import Tariffs on Inputs and Capital Goods (%) Description Tariff Rates Cereals 0.00 Export Crops 0.00 Oilseeds 4.97 Textile Fibers 0.00 Leguminous 25.00 Root Crops 10.40 Vegetables 25.00 Fruits 35.00 Other agricultural products 11.14 Live animals 3.82 Other animal products 13.89 Forestry 11.95 Fisheries 10.00 Mineral coal 5.00 Petroleum 5.00 Metallic minerals 5.00 Stone, sand and clay 5.50 Salt 9.16 Other minerals 5.00 Slaughtering 15.04 Animal and vegetable oils and fats 11.51 Dairy products 19.84 Grain mills 6.45 Bakery 40.00 Sugar refining 23.24 Other food products 12.06 Alcoholic beverages 27.12 Non-alcoholic beverages 30.00 Tobacco manufacturing 20.00 Textiles, garments, and leather 23.37 Wood products 16.28 Paper pulp, paper packing 9.41 Printing 20.25 Petroleum refining 18.61 Chemical products 8.62 Rubber products 19.92 Plastic products 23.44 Non-metal industries 18.81 Iron and Steel basic products 13.24 Non-ferrous products 12.47 Metal products 22.16 Machinery of general use 14.26 Machinery for particular uses 8.17 Household equipment 26.02 Office equipment 8.36 Electrical machinery 20.02 Radio, TV and communications equipment 18.05 Watches, optical instruments 13.99 Transport equipment 17.34 Furniture and other manufacturing 26.88 Metal and non-metal waste 12.01 Average Tariff Rate 15.69 Notes: The weights for the average tariff rate are those of intermediate consumption. 80 References Batra, R., Studies on the Pure Theory of International Trade, ch. 9, MacMillan London, 1973. Bond, E., "Using Tariff Indices to Evaluate Preferential Trading Arrangements: An Application to Chile", The World Bank, 1995. Clements, K. and L. Sjaastad, How Protection Taxes Exporters, Trade Policy Research Centre, London, 1984. Harberger, A., "A Vision of the Growth Process", AER, Vol 88, N° 1, March 1998. Jones, R. "The Structure of Simple General Equilibrium Models", The Journal of Political Economy, vol. LXXIII, December 1965, 6. Komiya, R., "Non-Traded Goods and the Pure Theory of International Trade", International Economic Review, VIII, June 1967. Neary, J. P., "Short-Run Capital Specificity and the Pure Theory of International Trade", The Economic Journal, vol. 88, Sept. 1978. Endnotes 1 The assessment of the impact of changes in commercial policies on the economy is thus limited to short run effects as no capital accumulation and technological change are allowed in the model. 2 Import prohibitions that have been eliminated comprise those that were enacted by presidential decrees. The remaining prohibitions are contained in laws granting the issue of import licensing to different government bodies. 3 No agreement has yet been reached with the WTO on binding tariff rates for imports outside the quota system. 4 The 1995 production matrix was used to calculate the output-weighted nominal protection rates for the activities. Within each activity, locally produced commodities were identified and the corresponding import tariff rates allocated to each of them and weighted by imports as the matrix contains data only about production at two digit level of the local classification of products. 5 Other agricultural activities such as paddy and livestock are classified in this report as non-traded goods as there are no reported imports of these commodities. In the case of livestock, imports of live animals are mostly breeding stocks. Sugar cane, coffee and cocoa are, on the other hand, exportable commodities. 6 These broad categories are consumer, intermediate and capital goods, and passenger motor cars. The latter are included in the consumer good category. BEC codes also classify imports by degree of processing (primary and manufactured products) for consumer and intermediate goods. 7 This mapping along with customs data were kindly provided by the Department of National Accounts of the Central Bank of the Dominican Republic. 8 The 10% uniform import duty is included in the simulations as the best alternative to free trade under fiscal revenue constraints where the order of n is determined by maximizing consumers' welfare through the simulations. 81 Commercial Policy 9 L. Sjaastad and K. Clements (1984) report similar values (.60) of the incidence of trade taxation on exportable activities for an average of several countries although the methodology is different than the one used in this study. 10 In the case of livestock, forestry and fishing, the sectoral output is dominated by livestock, of which production of poultry is one of the most important components. 11 It is assumed that technology in tourism (hotels and restaurants) is such that the elasticity of substitution between labor and capital is zero; as capital is specific to the activity, the activity's output response is zero. 12 If unitary elasticity of substitution between labor and capital in tourism is assumed, then tourism output would expand by about 14 percent, and the increase in tobacco manufacturing and footwear in export processing zones would be around 34 and 39 percent, respectively. 13 Actually, real wages would increase by about 2 percentage points if quantitative restrictions are eliminated. 14 It is assumed that return to capital in mining and agriculture is mostly land rental and as such assumed given by the force of nature. 82 SECTION 5: REGULATORY FRAMEWORK AND BANKING SECTOR PERFORMANCE A. The Legal and Institutional Framework 1. The juridical body of the DR's financial sector comprehends several Laws and the regulations established by the Monetary Board of the Central Bank (JM). While the main juridical body provides the basic institutional framework, the regulations approved by the JM specifically deal with a supervisory framework of credit controls, and has been developed since the early 1990s. 2. Since 1992 the Parliament has been discussing the establishment of a comprehensive and modem regulatory framework for the financial sector, the new Monetary and Financial Code "CodigoMonetario- Financiero" (CMF), that would expand on some important financial legislation that cannot be presently covered by the JM's Regulations. A good CMF could broaden the present supervisory regulations, provide stability to a regulatory system of controls, expanding it into some important market-oriented legislation, and, ideally, could allow public prosecution of delinquent individuals and governing boards, that could help prevent corruption and make the banking system more transparent and stable. 3. Supervision Authorities. Articles 111 and 112 of the DR's 1966 Constitution, give the power to regulate the monetary and banking system to the Junta Monetaria (JM). The JM is therefore the superior body of the DR's financial and monetary system. The Law on the Central Bank' and the General Law on Banks2 also establishes responsibilities of the Central Bank (CB) and the Bank Superintendencia (SPI). While the JM sets and determines the financial and monetary policy, the CB is empowered to put into practice and manage the financial and monetary policy. The JM is formed by ten members and headed by a chairman who is, at the same time, the CB's governor. The Supervisory body of the financial and banking sector is the SPI, which needs the JM's authorisation for certain decisions3. 4. Independence. The executive power appoints the governing bodies of the DR's financial system including the JM, the CB and the SPI. Among the ten members of JM's governing body are two civil servants from the ministries of Industry and Finance. Their period of service is two or three years (specifically mentioned in art No. 10 of the Law on the CB) but can be re-appointed. The SPI's director (Superintendente) is appointed by the Government in accordance with art. No 2 of the General Law on Banks published on April 14th, 1965. The mentioned Law does not specify the duration of the appointment. The CB's governor is also appointed by the government for a period of three years4 initially, but can be re-elected5 as well. However, all civil servants designated by the government are appointed by an "undetermined period", in accordance with the Law No 277 published on June 29th, 1966. In practice, the government can appoint anybody without a specific time for its service period. Once appointed only the Supreme Court can remove the JM's members and the CB's governor before the appointment-period of service finishes. 5. There is some degree of concern over the lack of autonomy and independence of the above mentioned governing bodies of the financial system of the Dominican Republic. Once approved by the Congress, the CMF could clarify these issues and place the independence of the financial authorities as one of the main principles to provide stability controlling the financial sector and managing financial upheavals. Only an independent authority can exercise appropriate actions to supervise, monitor, control and anticipate financial problems. In this context it is recommended that the new CMF includespecif/c wording to preserve the independence of the CB and the JM and specify as well, the minimum period of service of the CB's governor and JM's members. This period has to be long enough to provide stability to the system. 83 Section 5 B. The System of Regulations and Controls 6. Supervision and Controls. A properly structured supervisory system plays an important role to prevent crises by helping to identify and anticipate problems before they become serious. Despite the absence of the Monetary and Financial Code, the DR's authorities have made important efforts to establish prudential norms and to strengthen the supervision of the financial system. The main measures undertaken can be divided into two areas: (a) risk evaluation, classification and grading of bank's loans and debtors, and (b) banking institutions minimum capital requirements. 7. Risk Evaluation, Grading, and Provisions. The JM approved a risk rating method, which is being applied by the SPI during on-site inspections. Loans are classified according the following criteria: (i) debt service record: classifying loans as good, average or deficient; (ii) financial performance: understood as debtor cash flow, or debtor's payment capacity. Classifying debtors as: very good, adequate or insufficient; and (iii) collateral, also classified as good, average or deficient. The JM's Banking Norms also introduce only four different risks or debtors for risk classification and provisioning purposes: (i) commercial operations, (ii) consumer credit, (iii) mortgage loans, and (iv) investments. In principle the Banking Norms apply the same classification criteria to all debtors and risks. However, for consumer credit and mortgages loans the analysis could stress the debt service record, and investments, could be classified in accordance to market prices. 8. The provisioning and written-off policy applied by the authorities can be summarised in four steps. The first, provides a calendar to be applied perfunctorily by financial institutions semi-annually. When the on-site credit controls carried out by the SPI differ less than 10 percent on the total amounts provisioned by the banks' in-house audit departments, these can provision the balance during the following six months without being penalised. Regarding this perfunctory calendar and the corresponding provisioning amounts, it is necessary to call on how these norms clearly favour possible abuses of large companies and banks (buying and selling rights over the counter and openly in the market6 and on their own account), against ordinary credit to small and medium size companies and individuals, despite the lack of established capital markets, that would heavily penalise listed non- performing rights. Besides, the Authorities have not provided maximum risk-exposure limits on financial institution's own trading accounts. 9. The second step is a compulsory 1 percent generic provision on financial institutions to be set up semi-annually in accordance to the mentioned JM's resolution dated December I 1th, 1992. The third step is called "castigos" (punishments), and is only the way that the Authorities provide for a system to write-off from the bank's balance sheets, those fully provisioned debtors after another 24 or 36 months. This is a measure to clean up the banks' balance sheets and recognise actual losses. There is a further fourth step, which is a discretionary global provisioning to be applied by the SPI's unilateral decision, on the basis of a clear lack of private financial institutions' adequate provisioning amounts and policies. 10. One major obstacle to acknowledge the total number of bad debts in the financial system and provide reasonable provisioning, is its high cost, due to the lack of adequate legislation and satisfactory accounting procedures that do not allow provisioning to be deducted as expenses from the bank's balance-sheet. In these circumstances, the banking sector is being penalised for the provisions and losses with no real advantages for the system encouraging under-provisioning and under-reporting of bad loans. 84 Regulatory Framework and Banking Sector Performance 11. As a result, the present system of debtor's classification and provision requirements are not sufficiently strict in comparison to international best practice. It is recommended that the authorities apply international best practice regarding the classification and provisioning of risks, which is stricter than that actually applied by the DR's FA. For example, in Argentina or Chile, which are generally regarded as best practice examples of banking supervision in Latin America, consumer loans with payments overdue over 90 days are classified as non-performing and require provisions of 25 percent (Argentina) or 60 percent (Chile). Similarly, Colombia and Peru also classify consumer loans as non- performing after 90 days of payment arrears requiring provisions of 50 and 60 percent respectively. By contrast, the DR only requires a 20 percent provisioning for consumer loans with overdue payments between 90 and 120 days. Table 5.1 Days with overdue payment to classify as non-performing loans Consumer Mortgage Commercial Argentina 90 90 90 Brazil 60 360 60 Chile 90 90 90 Colombia 90 120 120 Mexico 60/90 180 30 Peru 90 90 15 DR (category C2) 120 180 Subjective Notes: Mortgage: no provision, mortgage loan guarantee. Commercial: discretionary provision. Table 5.2 Minimum provision when marked as non-performing loans (%) Consumer Mortgage Commercial Argentina 25 12 25 Brazil 100 100 100 Chile 60 10 Subjective Colombia 50 20 20 Mexico 45 35 45 Peru 60 50 50 DR 40 20 Subjective 12. Minimum capital requirements. The Basle Committee (BC) on Banking Regulations and Supervisory Practices worked on the convergence of supervisory regulations governing the capital adequacy of international banks. These regulations were created for a group of leading banks from developing countries such as France, Belgium, Spain, Canada, Japan or United States. The BC proposals where applied in 1987 and provided a minimum level of capital7 for those international banks based on OECD countries. Banks working outside these boundaries are considered riskier and therefore could set higher capital levels. As such the BC's capital recommendations could be considered as a minimum. 13. The DR has promptly adopted the BC capital requirement recommendations. A minimum capital requirement of 9percent of risk-weighted assets was established in 1992, and this was raised to 1 Opercent in 1998. The JM established a single measure of capital which includes common and ordinary shares, legal and other reserves, retained and current earnings, and real state value changes. Capital is not divided into two or three tiers, but it is only a single core capital measurement based on what is called "effective capital" or "adjusted capital" which is finally measured by the "Solvency Index". The core capital 85 Section 5 (solvency index) is calculated on the basis of a formula that weights the different capital components (excluding provisions, depreciation and other factors) against different risks exposures', as indicated in the domestic accounting guidelines9 issued by the SPI for financial institutions. 14. The definition of capital is similar to the definition in Peru which includes: paid-in capital, legal reserves, discretionary reserves, generic reserves for losses in loan portfolio, equity investment in warehousing, brokerage, and mutual fund subsidiaries, as well as current and retained profits. There is, however, no distinction between Tier I and Tier II capital, as requested by the Basle rules, and as implemented in Brazil, Colombia and Mexico. The ratio is calculated on the basis of the weighting of assets described in Table 5.4. While the level of the ratio is adequate by Latin American standards, the actual weighting of assets is less stringent than in other countries in the region. For example, while treatment of capital is similar to Peru, loan assets in Peru are weighted at 150 percent as opposed to 100 percent in the Dominican Republic, making it easier for the latter to comply with the minimum solvency ratio. By the same token, Argentina which has a similar weight for loans (100 percent), has higher weights also for other types of assets, making it harder to comply with minimum solvency requirements. Table 5.3 Minimum capital as a % of risk-weighted assets Argentina 11.5 Brazil 8 Colombia 9 Mexico 8 Peru 8 DR 10 Table 5.4 Asset weighting for calculation of minimum capital Unsecured Mortgage loans Deposits in / securities of loans other fin. institutions Argentina 100 100 100 Brazil 100 n/a 50 Colombia 150 50 20 Mexico 100 50 20 Peru 150 n/a 25 Dominican Rep. 100 60(40 if guaranteed) 40 Regulations on Exposures and Connected and Related Parties 15. Connected parties are defined as shareholders who own more than 3 percent of the bank's equity capital and related parties are any physical or moral person directly or indirectly related to a connected party and managers or directors. Loans to connected parties cannot exceed 100 percent of the banks' equity capital. Large exposures are also regulated by the JM's Resolutions, which set two different figures on the banks' equity capital, regarding debtors guarantees: * A maximum of 15 percent exposure on risks without collateral, and * A maximum 30 percent exposure with collateral. 86 Regulatory Framework and Banking Sector Performance 16. Other Prudential Measures. The FA have also taken other prudential measures that can be briefly highlighted as follows: * There is a maximum limit on off-balance sheet exposures. This limit is set on three times a bank's equity capital. * The value of financial institutions' fixed assets cannot exceed 75 percent of the its equity capital. In some cases this figure can represent up to 125 percent, only if the 50 percent extra is financed by long-term loans with a maturity of no less than 10 years. * The authorities are promoting "Credit Houses" (Bureau de Credito), which provide credit reports on individuals or firms determined by previous credit experience and other factors. * Regulations have been introduced for "Asset Valuation Houses" for the appraisal of assets and collateral and the creation of an Official Registry. All financial institutions should make, by Law'°, a valuation of debtor's assets on loans over Pesos 250,000 (USD 15,000). * The JM is also studying the regulation and promotion of "Rating Agencies", that would evaluate equity and debt instruments issued by banks, corporate, municipalities and other institutions, to be listed and later on traded into secondary markets. 17. Reporting Requirements. The JM updated the mandatory reporting requirements of key statistical information, to be disclosed by banks, on its Resolution dated April 2nd, 1997. Annex No. 2 incorporates a summary of this Resolution. The Resolution also includes pecuniary sanctions applicable in case of delays and non-compliance. The JM's mandatory reporting requirements are not sufficient to provide adequate information coverage of the DR's financial sector and the frequency of reports is still low relative to other countries in the region. Income statements, for example, are reported on a quarterly basis when best practice in Latin America is to at least present this information on a monthly basis. Moreover, the information does not include the data needed to determine the value of risk-weighted assets such as a breakdown of their asset portfolio, which is required to determine the adequacy of capital. Hence, it is impossible to perform independent (by outsiders) assessments of solvency. 18. The authorities are now in a very active process to develop the procedures to gather much needed additional information through improved reporting requirements. All these statistics will contribute to identify problems in the financial system. However, only well developed long-term financial markets will support the complexity and scale of banking operations and will supply the tools to control and anticipate financial crises. As the authorities work to improve the information systems, it would be advisable that these include: * Breakdowns of assets and liabilities not only by clients but also by groups of clients/companies, etc. In order to do so, it would be first necessary to develop a "National Accounting Plan" that would include "Consolidated accounts" by groups. This is an urgent matter for effective supervision. * Breakdowns of assets and liabilities by maturity, currency, industry sector and by geographical area to enable the risk concentration to be identified, monitored and controlled. * Credit facilities showing total facilities granted, by customer and group. 87 Section 5 * List of all exposures to guarantors to aid in the assessment of counter-party credit risk. * Excess over facility reports detailing facilities draw down in excess of agreed limits. This is a particularly important report. * Margin variation reports on standard interest rates of both banking assets and liabilities. C. Public Disclosure and Transparency 19. It is widely recognised among policy makers that banking supervision does not prevent bank failures by itself. Evidence shows that modem public disclosure regulation on banks' quarterly, semi- annually and annual results and other important information, is a powerful tool for supervisors and regulators. The DR would benefit from regulations that promote market transparency through greater public disclosure requirements on banks. The CB could continue with its traditional policies of prudential supervision and capital adequacy and at the same time move to market-based regulations that would be centred upon: (a) public disclosure; (b) responsibility for directors; and (c) penalties. 20. Such regulations would encourage bank's management to adopt and maintain prudent risk positions as a result of the general public knowledge and awareness of the bank's risks and results, while also establishing that directors have the ultimate responsibility and that producing disclosure statements that are false or misleading can have serious consequences. Therefore, false or misleading information could be treated as a criminal offence and directors could face unlimited personal liability. 21. The Payment System. The DR's payment system is obsolete. The authorities are aware of the need to develop an inter-bank clearing system fully independent from the financial institutions, where all banks could maintain clearing accounts with a local cash settlement centre. The CB is seeking to know and understand the nature of the payment system between banks, at all times, through the implementation of a modernisation program to be launched in the year 2000. This will allow the authorities to know, for example, the extent of a failing institution's direct involvement in local, national and international payments and clearance and to control the situation in order to avoid any contagion risks. D. Banking Sector Performance 22. During the last years, the DR's financial system has been experiencing a strong growth. Between 1994 and 1998 the average annual rate of growth of assets and liabilities in the banking was roughly 24 percent, including a very strong growth in credit of 38 percent in 1997. During all this period, the rate of growth of deposits has been significantly lower than the rate of growth of credit, which has led banks to increasingly depend on foreign currency denominated lines of credit. The rate of growth of foreign exchange financing has been well over 100 percent every year since 1994. 23. Capital adequacy. Total industry capital to total assets has averaged about 9 percent in the last four years. Given the lack of information on banks' assets, it is impossible to calculate the ratio of capital to risk-weighted assets. Since according to the DR's regulations the maximum weight is 100 percent (see Table 5.4), however, it is clear that the risk-weighted figure will be higher than the gross number, probably very close to the 10 percent stipulated by the regulation. The closest approximation that could be made to risk-weighted capital would exclude liquid assets (weight of zero). This calculation yields a value of 15.6 percent for the industry. In the DR, however, financial institutions do not carry consolidated accounting with, for example, their off-shore activities. The experience in other countries 88 Regulatory Framework and Banking Sector Performance show that this sometimes facilitates accounting-window-dressing and double counting of capital. In the future, this issue needs to be addressed. 89 Section 5 Table 5.5 Multibancos in the Dominican Republic (in millions of RD$) 1994 1995 1996 1997 1998 Total Liabilities 33,477.76 40,095.27 50,067.05 65,880.29 79,800.59 Rate of Growth (lo) 19.77 24.87 31.58 21.13 As % of GDP 24.34 24.71 27.28 30.63 33.09 Total Deposits (short & long term) 16,723.60 19,821.87 24,976.56 31,066.41 34,491.86 Rate of Growth (%) 18.53 26.01 24.38 11.03 As%ofGDP 12.16 12.21 13.61 14.45 14.30 Foreign Exchange Deposits 207.52 445.58 1,109.62 2,779.87 5,832.55 Rate of Growth (%) 114.72 149.03 150.52 109.81 As % of GDP 0.15 0.27 0.60 1.29 2.42 Foreign Exchange Financing 235.15 596.57 1,312.89 2,579.30 5,167.10 Rate of Growth (°/O) 153.70 120.07 96.46 100.33 As % of GDP 0.17 0.37 0.72 1.20 2.14 Total Assets 37,143.99 44,251.82 54,878.68 72,104.36 87,953.80 Rate of Growth (%) 19.14 24.01 31.39 21.98 As % of GDP 27.00 27.27 29.90 33.53 36.47 Credit in Local Currency 18,181.31 22,251.93 27,034.79 37,431.01 43,466.70 Rate of Growth (%) 22.39 21.49 38.45 16.12 As % of GDP 13.22 13.71 14.73 17.40 18.02 Credit in Foreign Currency 320.24 526.70 1,508.51 3,412.29 7,250.53 Rate of Growth (%) 64.47 186.41 126.20 112.48 As % of GDP 0.23 0.32 0.82 1.59 3.01 Table 5.6 Average industry financial ratios (%) Dec-95 Dec-96 Dec-97 Dec-98 Capital/Assets 9.18 8.61 8.63 9.12 Foreign currency liabilities / 82.85 136.58 168.45 193.70 capital and reserves Non-performing loans / total 3.61 2.62 2.10 2.14 loans Provisions / Non-performing 50.74 85.75 100.34 143.46 loans Operational expenses / total 38.58 39.60 42.03 38.56 income Net financial income / 186.96 195.70 191.71 204.76 personnel expenditure Return on equity (before taxes 31.76 34.91 34.49 30.61 and reserves) Return on assets 1.76 1.85 1.97 1.13 Liquid assets / total deposits 39.70 30.77 31.29 34.63 Total interest expense / total 9.29 7.79 7.08 9.49 deposits 90 Regulatory Framework and Banking Sector Perfornance 24. Foreign exchange liabilities. The level of exposure to foreign exchange liabilities is high, averaging nearly 200 percent of outstanding equity and reserves, and has more than doubled since 1995. The level of this ratio is a cause of concern, in particular because there is little information available on the extent to which asset holdings in foreign currency net off some of the exposure, and on the activity of banks in currency hedging via the use of derivative products. 25. Non-performing loans. The industry-wide non-performing loans to total loans ratio has decreased significantly from a high 3.6 percent in 1995 to 2.1 percent in 1998. While a level below 3 percent is not usually considered alarming, in the case of the DR one must consider that the definition of non-performing loans is more lax that in other countries and, as in many other countries that have been undergoing fast sector development with net transfers of resources from depositors to banks, under- reporting may be significant as bad performing loans get rolled-over. 26. Provisions. As the relative size of non-performing loans has fallen the extent of provisioning has dramatically increased. The industry average level of provisioning was at 143 percent of total non- performing loans in December 1998. 27. Efficiency. A very rough measure of efficiency of the industry is used, the ratio of operational expenses to total income. This ratio has remained rather stable since 1995, but appears to be rather high, at 38.6 percent in December 1998. 28. Return on equity. Profitability has fallen significantly in the last year relative to the previous two. The return on equity from 34.5 percent in 1997 to 30.6 percent in 1998. 29. Return on assets. The return on assets fell from nearly 2 percent in 1997 to 1.1 percent in 1998. The decline in profitability is consistent with the increase in the share of liquid assets in the banks' balance sheets and the increase in the cost of funds (see liquidity and cost of funds indicators). 30. Liquidity. The industry-wide level of liquidity is well above that required by the regulation (20 percent). Total liquid assets to total deposits was 34.6 percent in 1998, down from nearly 40 percent in 1995, but higher than in both 1995 and 1996. 31. Cost of funds. The cost of funds increased significantly in 1998. The ratio of interest expense to total deposits has increased to 9.5 percent ion 1998 from 7 percent in 1996 and 7.8 percent in 1995. Sector Growth and Risks 32. The strong growth of the banking sector has undoubtedly brought with it enhanced and vulnerabilities. In the DR these are linked to (a) increasing portfolio risks inherent to the rapid growth of credit and the client base, and (b) the increased exposure to foreign exchange risk and contagion effects because of the enhanced role of foreign exchange liabilities. However, one would not necessarily expect that this enhanced risks would reflect in the short run through the banking sector indicators data. When the rate of growth of deposits is higher than the deposit interest rate (which has been fluctuating between 5 percent and 20 percent), as has consistently been the case in the Dominican Republic, especially until 1997, there is a net transfer of resources to the banks which are able to remain liquid even if they roll over their loans (including unpaid interests) and make no returns on assets. 33. The key test, however, can take place when the rate of growth of the bank deposits (including foreign exchange lines of credit) is lower than the deposit interest rate, and banks face the need to make a net transfer of funds to depositors. In order to do so, they must either extract a transfer of resources from 91 Section 5 their borrowers or run down their stock of liquidity. Problems occur when the required transfer to depositors is so high, that banks cannot extract the required resources from their borrowers or their stock of liquidity. In the short run, the market assessment of these risks seem to be low, as strong growth is expected to continue in the short and medium term. For example, Thompson's Bank-watch Rating rate 4 Dominican Banks short term debt in early 1999 with LC I (local currency risk 1), the lowest risk grade they provide in their range that goes from LC 1 to LC4. E. Addressing Sector Risks 34. Following the macroeconomic crisis of the late 1980s and early 1990s, strong economic growth, a monetization in the economy, and improvements in the Bank's regulatory system have allowed for strong sector development. Financial deepening has increased and the sector, including the regulatory agencies are in a process of modernization. However, several challenges lie ahead. In previous sections it has been noted that regulations need to be continually upgraded to bring prudential norms in line with what now are considered best practices. At the same time, the supervision capacity of the SPI needs to be reinforced as the institution is still away from having the professional capacity to adequately enforce and monitor the sector. But even with adequate and fully enforced regulations in place, additional measures need to be undertaken to address potential problems. Procedures for remedying problem bank situations 35. In March 1996, Banco de Comercio Dominicano suffered a liquidity short-fall despite its apparently strong capital adequacy ratio. The bank represented at that time, 14 percent of the DR's total banking assets and almost 17 percent of its deposits. The CB had to put in place mechanisms to avoid a run-on-deposits, including injecting liquidity; authorising a US$300 million sale of loans classified as "A" to the CB, granting of guarantees to retail depositors; and finally the CB had to buy the majority of the bank's equity capital to acquire the bank's total control in order to manage the situation. The total cost for the taxpayer was about 1 percent of the country's GDP. This problem highlights the need to establish and develop detailed procedures for remedying financial sector problems. 36. The CB has experience handling these situations. However, in the absence of an adequate Monetary and Financial Code, the current legislation is not sufficient to address these issues and further plans and actions could be taken. Presently the authorities have limited guidance and resources to dominate a financial problem. The authorities have only regulated the actions to be taken in case of capital adequacy shortfalls. JM's Resolution on capital adequacy deals with the problem of a deficient solvency index. When the ratio falls below 10 percent, the regulations call to avoid further lending; transfer to the CB any increase in deposits; avoid dividend payments; and stop any credit facility from the CB. When a bank's capital adequacy falls below 6 percent the SPI's Director can appoint a Deputy with sufficient powers for up to 90 days "to correct the situation". If the bank's capital falls below 4 percent and the situation is not regularised in 10 days, the JM can take over the failing bank in order to liquidate it. Institutional Requirements for Remedying Banking Problems 37. It would be very beneficial for the authorities to always have in their possession the "formally executed documents" establishing the terms and conditions under which banks will have access to the lender of last resort facility. These documents could be reviewed and updated periodically, and, most important, the CB could, to the fullest extent possible, have in its possession the collateral necessary for discount window borrowings. In addition, it would be useful to for the CB to have well thought out_ 92 Regulatory Framework and Banking Sector Performance "plans" whereby it could provide large-scale liquidity to the system as a whole using open market operations. These formally executed documents could detail, as far as possible, the following issues: (i) the name of the authority or authorities to intermediate in the financial problem; (ii) enforce certain measures upon the failing institution, including: the authority to change management, the mechanism to manage a failing institution until a solution is found, and the system to place top-officials to supervise the banks operations; (iii) documents that allow to: inject new capital, eliminate dividends, sale of assets, adjust capital, shrink the balance sheet, and sell affiliates or subsidiaries and or selective lines of business in order, either to raise capital or reduce points of vulnerability, or both. Deposit Insurance Scheme (DIS) 39. The establishment of a deposit insurance scheme to protect small depositors could be beneficial as long as it is part of an overall exit strategy. Until now, the Central Bank has bailed out depositors from failing institutions and there is very likely a market perception that the government will continue to act in that manner to protect depositors. In this context, establishment of a formal Deposit Insurance Scheme (DIS) could be advisable. The authorities could promote a DIS with a suitable limit or cap on the level of individual deposit-coverage. The most common elements of a well-endowed DIS would include: (i) a cap or limits on individual deposit-coverage; (ii) the scheme could be financed by the DR's financial sector in the same way that the industry is now financing the SPI; (iii) there could be an initial capitalisation that have always to come from a governmental source. The CB could stand behind the DIS in a visible manner, however, its corporate organisation could have a mixture of private and public ownership. 93 ANNEX Table A5.1 Mandatory Reporting Requirements Institution Daily Weekly Fortnightly Monthly Quarterly Semi-annually Annually SUPERINTENDENCIA (SPN) Balance Sheet (Analytic) Credit and Loans report Semi-annual Financial Situation report Profit and Loss report Annual Report Annual Report public disclosure Management Letter Asset evaluation report Source: Superintendency (SPI) and the Central Bank. 93 Table A5.2 Risk Grading System Grading Specific Commercial Operations Consumers Mortgage Loans Investments Provisions (Discount, Credits ...) Goods A) The JM's Regulations includes "Instruments" a) Debt service record: issued by corporations (such as bonds, classified as good, deventures ... ) and Banks. Today only short- average or deficient. term commercial paper up to 90 days, is issued. b) Financial performance: Classified B) These risks should be regarded as Market Risks debtor cash-flow, or regarding debt Classified regarding debt service. rather than Credit Risks and classification debtor's payment service (grading) varies in accordance with the "market capacity. Classified as: value" of the instrument and the interest very good, adequate or payment performance of the issuer (usually insufficient. banks and large corporations established in the c) Collateral, as above. DR) C) Own Banks' Grading on Investments have to be ratified (or changed) by the SPI. No liquidity problems I A 0% experienced by the issuer and Up to 45 days Up to 5 days overdue No liquidity problems experienced by the issuer and "Normnal Risk" duly paid interests on the assets overdue * duly paid interests on the assets issued issued 2 B 1% A loss not greater than 5% of the Up to 60 days Up to 180 days overdue A loss not greater than 5% of the value of the asset Potential Risk" value of the asset issued overdue issued Up to 15% loss due to market 3 C1 10% upheavals and/or liquidity Up to 90 days More than 180 days overdue Up to 15% loss due to market upheavals and/or "Real Risk" shortages experienced by the overdue liquidity shortages experienced by the issuer issuer Due to the existence of real estate 4 C2 20% Up to 25% Up to 120 days guarantees this regulation does not Up to 25% "Significant Risk" overdue provide for greater grading 5 Dl 40% Up to 55% Up to 180 days Up to 55% "High Risk" overdue 6 D2 60% Up to 80% Up to 270 days Up to 80% "Very High Risk" overdue 7 100% Up to 100% More than 270 Up to 100% l_________ ____________________________ days overdue Source: JuntaMonetaria. 94 Table A5.3 Reserve Requirements for Financial Institutions 1994-1998 Financial lnstitutions Liabilities - t Ret.rp. on R6er#e Commercial Banks Deposits 20% CDs 20% 2% annual interest rate on 50% of required reserves Foreign Exchange liabilities 30% when 300% above capital 10% when under 300% capital Mortgage Banks CDs 10% samne as above Term Depositis 20% Development Banks Investment Certificates 10% same as above same as above Financial Institutions Investment Certificates 10% 6% annual on reserves at Savings accounts 10% National Housing Bank 14% annual on reserves held Savings and Loans on Central Bank paper Financial certificates 10% 12% annual on reserves held on Central Bank paper. Source: Central Bank Dominican Republic. Endnotes Officially published in December 29th, 1962. 2 Officially published in April 14th, 1965. 3 For instance and in accordance to the General Law on Banks published in April 14th, 1965: a) Art. No 10: to authorise the establishment of foreign banks; b) Art No 13 :to authorise mergers and acquisitions; c) Art. No 18: to compel bank's governing bodies to level out their capital base in accordance with the minimum established by law, etc. 95 Section 5 4 Law on the Central Bank, published in December 29th 1962, Art. 28. 5 Law on the Central Bank, Art. No 16, second paragraph. 6 The DR has an incipient sort of "Bolsa de Valores" where short-term commercial paper up to 90 days is already traded. 7 The BC's regulators stressed that their recommendations covered the minimum capital regarding "credit risk" operations and therefore did not cover "market risks" operations that had to be covered with further capital measures. (However, since 1987 the BC has produced further regulations on market risks.) 8 The credit conversion factors applied on the weightings of balance sheet assets, reflect the actual development of the banking industry. As an example we can list some balance sheet assets and their corresponding weights as follows: a) 100 % on: Loans and investments b) 60 % on: Mortgage loans c) 40 % on: loans from other domestic financial institutions d) 15 % on: loans guaranteed by deposits, irrevocable letter of credits, etc. 9 The SPI established an "accountancy manual" for financial institutions based on Resolution No 19-94 dated December 9th, 1994. 10 JM's second Resolution dated June 29th, 1993 and amendment dated September 7th, 1994 and September 14th, 1995. 96 SECTION 6: BUDGET LAW AND PUBLIC SECTOR MANAGEMENT A. The Budget Law 1. The current Budget Law (BL) dates from 1969 and at the time it was promulgated it contained most of the elements of a modem Law. It established the link between planning and budgeting, incorporated the concept of budget by program, and regulated the overall budget process. It had, however, some very important loop-holes, which if misused could lead to an extreme concentration of power in the Presidency undermining the normal role a budget system has in terms of being a transparent instrument to express government policies and priorities, promote efficiency and transparency in the use of resources, and to be an adequate instrument for the conduct of macroeconomic policy. The application of the BL during the last fifteen years has proven that the most important shortcomings of the legislation have been: * Lack of consolidation for overall public sector. The BL pertains only to the Central Government and does not include the non-financial autonomous entities of the State. This has increasingly become a problem as progressive deterioration in the administration has required increased, often, non-anticipated budget transfers to these entities, mainly the National Electricity Company (CDE) and the State Sugar Conglomerate (CEA). For example, the transfers to public enterprises averaged about 5 percent of total central government expenditures from 1980 to 1996, rose to 9.4 percent in 1997 and is expected to have been equally high for 1998. * Existence of Special Budget Accounts that do not allow for ex-post classification of expenditures. These accounts include No 112 "Specific Executing Agencies", and No 114 "Reserves". One of the most important shortcomings of the existence of these accounts is their impact on the control of public employment and the wage bill. The lack of an exact ex-post classification of expenditures makes it impossible to know the level of employment and the associated long term budget commitments in these categories. * The prevalence of a Special Presidential Fund (Fund 1401). The existence of this Fund clearly distinguishes the DR's budget execution process from that of basically any other country with a modern budget system. Between 1986 and 1997 roughly 50 percent of the countrys budget was directly executed by the Office of the Presidency largely through the use of Fund 1401, notwithstanding that the budget presented to Congress generally provided the Presidency the control over roughly 10 to 13 percent of all resources during those years. Table 6.1 Expenditures by institution (as a % of total expenditures) 1980-1985 1986-1990 1991-1996 1997 1998* Finance 16.8 12.5 15.6 22.1 23.6 Presidency 15.6 47.6 52.3 37.8 21.5 Education 12.5 7.7 7.9 10.6 13.8 Public Health & Social Assistance 8.7 6.2 5.7 7.1 9.1 Agriculture 13.0 6.3 3.5 4.1, 7.6 Interior and Police 7.2 5.5 4.9 6.5 6.7 Armed Forces 11.1 6.6 5.3 6.2 6.2 Public Works and Communications 10.3 4.2 1.7 1.1 5.1 *Prehlminary. Source: Based on ONAPRE. 97 Section 6 2. The Fund 1401 has two sources of funding: (i) the equivalent of 75 percent of tax revenues in excess of those projected in the budget approved by Congress; and (ii) the equivalent of 100 percent of the "non-utilized funds" from the different executing agencies. These "non-utilized funds" have been the main source of funding for Fund 1401, accounting for roughly two-thirds of Fund 1401 funding from 1994 to 1998. The under-funding of line Ministries and other executing agencies that fuel the Fund 1401 are made possible by art. 30 of the BL which states that "the budget appropriations constitute a maximum limit to resources available to the institution", and therefore they are not an entitlement. The art. 30 goes on to specify that in practice, "resources will be allocated subject to the availability". Table 6.2 Ratio of executed to approved budget by institution and programs 1994-1998 1994 1995 1996 1997 1998 Presidency 4.38 5.08 5.24 5.49 2.04 All other 0.54 0.56 0.62 0.97 0.93 Defense 0.89 0.53 0.56 0.82 1.03 Justice & Order 0.74 0.56 0.55 0.95 0.99 Education 1.00 0.93 0.88 1.20 1.06 Health 0.91 1.76 0.78 0.96 0.88 Agriculture 0.93 0.71 0.74 0.92 1.05 Transport 3.48 2.67 2.34 1.52 1.02 Urban Works 8.53 4.64 69.70 54.62 15.56 Electricity 0.75 1.73 2.46 4.31 1.03 Source: Based on ONAPRE. B. The Presidential Fund 1401 and the Budget Management Process 3. While there has been a remarkable improvement in containing the use of Fund 1401 during 1998, the impact of many years of over-using this system has had a strong impact on the way the budget process is managed and on the quality of the management of public funds. The main outcomes could be summarized as follow: e the budget has become an instrument largely devoid of meaning as a sign of government's policies and priorities because differences between approved and executed budget can be, by both institutions and type of programs, very large; * the pervasive uncertainty over resources that will actually be available provides little incentive for the utilization of adequate planning tools at the implementing agency level. It permeates accountability for results as results-oriented programs cannot be implemented in this context of high uncertainty. * it encourages poor procurement practices and reduces the ability to adequately control expenditures. In principle, Ministries and executing agencies cannot enter into contracts until they have received corresponding spending authorization from the Budget Office. In practice, however, implementing units systematically violate these regulations by using supplier's credit to implement programs for which they presume they will get (or are entitled to) funding; and * it has concentrated in the Office of the Presidency enormous powers at the expense of developing the capacities of the line Ministries, municipalities, and other more decentralized mechanism for implementation. Moreover, as traditionally the system of checks and balances 98 Budget Law and Public Sector Management between the executive and other independent powers of the State has not functioned effectively, accountability of government use of resources is still weak. C. The Agenda for Reform 4. Reforms to the Budget Law towards improving the efficiency, transparency, and accountability of the use of public funds could be deemed a priority to support economic growth and reduce poverty. In that context the main areas in need for reform are: The need to integrate in the budget the whole non-financial public sector both for expenditures and sources of funds. Establish at the same time a uniform system of accounts that will eliminate the special treatment of special funds or accounts that make, for example, the control of the wage bill very difficult. Eliminate the incentive to under-fund activities and reduce the line Ministries' and other implementing agencies uncertainty over the availability of funds by providing the concept of budget appropriation its true meaning. This in turn, will make the budget formulation more meaningful. Make the budget formulation process a policy tool. As such, it could represent the government's broad objectives in terms of: (a) the mix of current vs. capital expenditures; (b) the policy of subsidies and transfers to private and public entities; (c) the goals for public employment and the wage bill; and (d) financing policies. Establish for the category of public investments an information system that can prioritize investment projects and make projections on future recurrent budget demands of current investments. 99 Section 6 TABLES Table A6.1 Difference in Executed vs. Allocated Budget by Ministry, 1994-1998 (millions of RD$) Ministries TOTAL 1994 1995 1996 1997 1998 National Congress 52.80 28.20 5.10 19.50 Presidency (43,186.30) (8,861.10) (9,100.30) (10,087.70) (10,686.70) (4,450.50) Interior and Police 537.90 307.60 407.00 272.70 (464.20) 14.80 Armed Forces 3,407.30 230.00 1,147.50 1,304.00 628.60 97.20 Exterior Relations 336.80 74.30 110.90 85.70 2.10 63.80 Finance 3,968.70 3,139.80 1,711.00 1,098.10 (2,432.20) 452.00 Education 2,668.90 769.50 721.40 815.80. 143.30 218.90 Public health 7,318.50 1,453.60 1,802.60 2,066.40 1,222.00 773.90 Sports 1,148.10 154.50 296.30 288.50 167.70 241.10 Labor (682.90) (99.10) (101.60) (124.10) (160.90) (197.20) Agriculture 5,626.20 1,186.10 1,623.50 1,420.10 926.10 470.40 Public Works 2,177.40 469.90 486.50 525.90 497.70 197.40 Industry and Commerce 358.20 74.50 73.50 103.80 86.50 19.90 Tourism 342.40 64.40 93.20 85.80 61.80 37.20 Attorney General 257.20 54.00 85.80 81.50 26.20 9.70 Judicial Power 248.90 24.40 69.70 154.30 (0.20) 0.70 Electoral Council 365.50 78.90 111.40 129.50 20.30 25.40 External Auditors Office 15.60 3.30 5.30 2.40 0.20 4.40 Source: Based on ONAPRE. 100 Budget Law and Public Sector Management Table A6.2 Difference in Executed vs. Allocated Budget by Activity, 1994-1998 (millions of RD$) Activities TOTAL 1994 1995 1996 1997 1998 1. General Services 5,128.40 1,950.40 2,081.70 2,498.40 (1,053.00) (349.10) General Administration 1,080.50 1,472.70 608.00 852.70 (1,491.00) (361.90) Justice & Public Order 1,565.90 271.20 548.90 655.10 68.80 21.90 National Defense 2,175.40 132.40 812.90 905.10 372.00 (47.00) International Relations 306.60 74.10 111.90 85.50 (2.80) 37.90 2. Social Services (6,775.50) (1,558.00) (544.00) (456.80) (3,306.50) (910.20) Education (413.80) 9.10 222.60 471.60 (793.70) (323.40) Sport 155.00 (54.20) 135.30 31.10 (59.40) 102.20 Health 2,102.40 183.10 609.10 691.00 119.00 500.20 Social Assistance (1,000.60) (102.40) 65.80 (136.40) (760.60) (67.00) Labor 113.80 10.40 36.80 29.20 18.30 19.10 Housing (3,276.70) (513.60) (900.20) (910.30) (641.20) (311.40) Sewerage & Potable Water (3,295.50) (1,118.60) (684.60) (382.00) (439.70) (670.60) Municipality Services (1,209.30) 18.00 (34.70) (268.10) (749.80) (174.70) Community Services 49.20 10.20 5.90 17.10 0.60 15.40 3. Economic Services (19,720.40) (3,395.40) (3,358.60) (5,226.90) (5,437.10) (2,302.40) Irrigation, Agriculture & 1,478.40 131.20 692.90 631.40 190.30 (167.40) Fishing Industry, Mining & (1,322.30) 15.70 (31.00) (294.80) (530.40) (481.80) Commerce Transportation (5,694.90) (1,336.30) (1,607.50) (1,945.60) (765.00) (40.50) Caminos Vecinales (1,108.40) (555.10) (115.00) (240.30) (195.60) (2.40) Communications (33.70) (14.60) (92.40) 122.60 110.60 (159.90) Urbanism (9,141.90) (1,893.60) (1,731.70) (2,171.00) (1,695.10) (1,650.50) Energy (2,980.40) 196.40 (498.20) (797.40) (1,807.80) (73.40) Sugar (1,146.40) (4.30) (68.50) (617.00) (791.10) 334.50 Tourism 229.20 65.20 92.80 85.20 47.00 (61.00) 4. Financial Services 6,328.60 2,155.80 1,369.80 1,427.50 (165.10) 1,540.60 Banking, Securities & (725.70) (517.90) 0.60 7.00 (39.90) (175.50) others Debt Servicing 7,054.30 2,673.70 1,369.20 1,420.50 (125.20) 1,716.10 iqTA2 __1. j.. Source: Based on ONAPRE. 101 Section 6 Table A6.3 Difference in Executed vs. Allocated by Expenditure Category, 1994-1998 (millions of RD$) Category TOTAL 1994 1995 1996 1997 1998 A. Current Expenditures 333.20 2,652.20 1,653.50 1,929.40 (6,524.90) 623.00 A. I Operational 3,060.10 944.40 1,898.50 2,839.50 (2,476.10) (146.20) Expenditures Personal Services (8,184.40) (122.00) (1,112.40) (382.10) (4,799.00) (1,768.90) Non-personal Services 728.60 251.00 374.60 424.90 329.30 (651.20) Supplies 10,515.90 815.40 2,636.30 2,796.70 1,993.60 2,273.90 A.2 Transfers (9,250.40) (0.20) (1,293.90) (2,318.40) (4,949.40) (688.50) To the Public Sector (7,022.80) (143.30) (941.10) (1,711.10) (3,744.40) (482.90) To the Private Sector (2,349.40) 120.40 (387.60) (625.20) (1,235.30) (221.70) To the External 121.80 22.70 34.80 17.90 30.30 16.10 A.3 Interest on the Debt 6,523.50 1,708.00 1,048.90 1,408.30 900.60 1,457.70 and Deferred Payment of Current Expenditures B. Capital Expenditures (15,372.10) (3,499.40) (2,104.60) (3,687.20) (3,436.80) (2,644.10) B. 1 Real Investment (16,167.00) (3,610.30) (3,277.00) (4,410.20) (2,106.00) (2,763.50) Machinery & Equipment 5,057.70 751.00 1,393.60 1,472.10 1,345.90 95.10 Construction (21,016.30) (4,105.90) (4,470.00) (5,941.90) (3,604.20) (2,894.30) Agriculture (208.40) (255.40) (200.60) 59.60 152.30 35.70 B.2 Asset Acquisition (555.80) (117.50) (157.10) (104.60) (15.30) (161.30) B.3 Capital Transfers (1,351.00) (731.10) 242.70 (50.70) (1,103.90) 292.00 To the Public Sector (1,312.60) (729.10) 231.00 (42.40) (1,057.80) 285.70 To the Private Sector (38.40) (2.00) 11.70 (8.30) (46.10) 6.30 B.4 Debt Amortization 3,401.20 1,455.50 907.80 731.80 (306.20) 612.30 B.5 Financial Investments (699.50) (496.00) 179.00 146.50 94.60 (623.60) 102