Report No. 22598-POL Poland The Functioning of the Labor, Land and Financial Markets: Opportunities and Constraints for Farming Sector Restructuring December 2001 Environmentally & Socially Sustainable Development Unit Europe and Central Asia Region Document of the World Bank CURRENCY EQUIVALENTS Currency Unit = Polish Zloty (PLN) US$ 1 = PLN 4.20 (as of July 25, 2001) ACRONYMS AND ABBREVIATIONS AMA Agency for Agricultural Markets APA Agricultural Property Agency ARMA Agency for the Restructuring and Modernization of Agriculture BGZ Bank for Food Economy CEE Central and Eastern Europe EU European Union GDP Gross Domestic Product GUS Polish Central Statistical Office IERiGZ Institute of Agricultural and Food Economics KRUS Farmer's Security Social Fund LFS Labor Force Survey MNL Multinomial Logic Model OECD Organisation for Economic Cooperation and Development PKO BP Polish State Savings Bank ZUS Nonfarrn Social Security Fund Fiscal Year January I to December 31 Vice President: Johannes Linn Country Director: Michael Carter Sector Director Kevin Cleaver Sector Leader: Laura Tuck Team Leader: Carlos Cavalcanti Contents CURRENCY EQUIVALENTS ACRONYMS AND ABBREVIATIONS ACKNOWLEDGEMENTS EXECUTIVE SUMMARY ........................................................... III Introduction ........................................................... iii Rural Households' Income and Activities ............................................................v Rural Land Structure and the Rural Land Market ........................................................... ix Determinants of Total Farm Revenue ........................................................... xiv The Policy Agenda ........................................................... xvi Outline of the Study ........................................................... xviii CHAPTER 1: RURAL LABOR MARKET ............................................................1 Introduction ............................................................I Economic Reforms and Changes in Income Sources for Rural Households ............................................2 Changes in the Institutional and Policy Framework ............................................................9 Place of Residence and Place of Work: The Question of the Rural-Urban Divide ............................... 24 CHAPTER 2: RURAL LAND MARKET ........................................................... 33 Introduction ........................................................... 33 Land Transactions after 1990 ........................................................... 39 Land and Farming in the Survey ........................................................... 46 Determinants of Farm Size ........................................................... 52 Measures to Achieve a Fully Functioning Land Market ........................................................... 61 CHAPTER 3: RURAL CREDIT MARKET ........................................................... 63 Introduction ........................................................... 63 Why Don't Rural Polish People Borrow? ........................................................... 65 Financial Institutions for Rural Polish People ........................................................... 67 Savings ........................................................... 68 Overview of Household Borrowing ........................................................... 71 Credit Rationing? ........................................................... 76 Conclusions ........................................................... 85 CHAPTER 4: ON THE DETERMINANTS OF SHORT-RUN FARM REVENUE IN POLAND..86 Introduction ......................................................... 86 REFERENCES ........................................................... 92 Tables Table 1: Sources of Household Income .vi Table 2: Sources of Farm Household Income by Farm Size .vii Table 3: Employment Status by Sector of Occupation in Rural Areas .viii Table 4: Breakdown of Nonfarm Business Activities .viii Table 5: Sources of Finance for Nonfarm Businesses .............................................................. ix Table 6: Territorial Distribution of Land Transactions, 1990-1996 .................................................... xi Table 7: Size and Structure of Farms With and Without Leased Land ....................... ...................... xii Table 8: Dispute Resolution Mechanisms (multiple answers allowed) ............................................ xiii Table 9: Frequency of Land Transactions and Estimated Prices per Hectare ............... ................... xiii Table 10: Estimated Elasticities of Net Revenue by Farm Size ......................................................... xiv Table 11: Productivity of Land and Labor by Farm size .............................................................. xvi Table 1.1: Rural Households by Exclusive or Main Source of Household Income, 1988 and 1995 .......3 Table 1.2: Households' Main Income Source (percent) ............................................................... 4 Table 1.3: Distribution of Rural Households by Number of Workers (%) ..............................................5 Table 1.4: Distribution of One-Worker and Two-Worker Households by Employment Status of Workers ......................................................5 Table 1.5: Employment by Sector in Four Rural Regions ......................................................6 Table 1.6: Employment by Sector and Region in February 1999 ...................................................... 7 Table 1.7: Rural Workers by Main Job Status and Second Job ...................................................... 8 Table 1.8: Percent of Rural Workers with a Second Job by Main Activity in First Job ..........................8 Table 1.9: New Trend in National Employment, 1992-1998: The Expansion of the Private Nonagriculture Sector (percent of total employment) .........................................................9 Table 1.10: Total Budgetary Allocations to Agriculture, 1986-1995 ...................................................... 11 Table 1. I1: The Macroeconomics of Social Security: Elderly Population (65+) as a Fraction of the Working-Age Population (15-64), Selected Countries ..................................................... 12 Table 1.12: The Difference between Net and Gross Wages for ZUS Contributors ................................. 13 Table 1.13: Old-Age and Disability Contributions and Benefits KRUS and ZUS: 1998 Annual Averages .............................................................. 14 Table 1.14: The Microeconomics of Social Security Contributions ....................................................... 14 Table 1.15: Land Ownership by Household .............................................................. 16 Table 1.16: Average Number of Income Earners per Household, by Land Ownership .......................... 16 Table 1.17: Distribution of Rural Workers according to Household's Land ........................................... 17 Table 1.18: Distribution of Rural Workers from Households with More than 1 Hectare of Land by Social Security System (percent within voivodship) ........................................................ 17 Table 1.19: Worker's Affiliation by Region (percent) .............................................................. 18 Table 1.20: What Is Your Employment Status? (by occupational sector) ............................................... 21 Table 1.2 1: Rural Workers: By Sector and Principal Source of Income ................................................ 21 Table 1.22: Agriculture Sector Workers Are Older, Particularly Those Who Consider Pensions to be their Main Income Source .............................................................. 22 Table 1.23: Answer to the Question. Did You Work During the Reference Week? (by main income source) .............................................................. 22 Table 1.24: The Distribution of Rural Wages by Schooling (deciles) ..................................................... 23 Table 1.25: Median Annual Income from Work (Rural Areas): By Schooling ...................................... 23 Table 1.26: Median Annual Income from Work (Rural Areas): By Occupation ................................... 24 Table 1.27: Median Annual Income from Work (Rural Areas): By Sector .24 Table 1.28: Daily Farm Wages in Rural Areas ............................................................... 24 Table 1.29: Percentage of Rural Workers That Work in Urban Areas by Main Activity and Region .... 27 Table 1.30: Commuting and Distance Traveled by Region: Main Job ................................................... 27 Table 1.31: Commuting and Distance Traveled by Region: Second Job ............................................... 28 Table 1.32: Recent Immigration to the Region: Fraction of Workers That Moved to their Current Location during the Last Year (by main activity - %) ......................................... 28 Table 1.33: Accumulated Immigration to the Region: Fraction of Workers That Moved to their Current Location during the Last Decade (by current main activity - %) ................. 29 Table 2.1: Structure of Land Use and Ownership, 1990-1997 (as percent of agricultural land) ........... 33 Table 2.2: Changes in Land Use and Ownership, 1990-1997 (in thousand hectares) ........................... 34 Table 2.3: Ownership and Use of Agricultural Land in Poland: 1980-1997 (in thousand hectares) ... 35 Table 2.4: Distribution of Agricultural Land among Different Types of Farms: 1996 ........................ 36 Table 2.5: Territorial Distribution of Land Transactions, 1990-1996 ................................................... 37 Table 2.6: Land Transactions: Cumulative for 1992-98 (Institute of Agricultural and Food Economics) ............................................................. 44 Table 2.7: Average Size of Land Transactions in the 1990s (hectares) ................................................. 44 Table 2.8: Average Prices per Hectare of Arable Land in Transactions between Farmers and with APA ............................................................. 45 Table 2.9: Number of Households in the Survey ............................................................. 46 Table 2.10: Characteristics of Nonfarming and Farming Households in the Survey .............................. 48 Table 2.11: Land Ownership and Land Use: Percent of Respondents by Ownership Category Table 2.12: Size and Structure of Farms With and Without Leased Land .............................................. 49 Table 2.13: Profile of Land Owners who Lease oOut Land ............................................................. 49 Table 2.14: Structure of Leased Land by Sources (in percent of leased land, n = 255) .......................... 49 Table 2.15: Dispute Resolution Mechanisms (multiple answers allowed) .............................................. 49 Table 2.16: Lease Payments as Reported by Lessors and Lessees .......................................................... 50 Table 2.17: Frequency of Land Transactions and Estimated Price per Hectare ...................................... 51 Table 2.18: Main Parties in Buy-and-Sell Transactions ............................................................. 51 Table 2.19: Main Reasons for Selling and Not Buying Land ............................................................. 51 Table 2.20: Change in Land Holdings between 1997 and 1999 ............................................................. 52 Table 2.21: Sources of Change in Land Holdings between 1997 and 1999 ............................................ 52 Table 2.22: Change of Farm Size between 1997 and 1999 for Farms in Different Size Categories ....... 52 Table 2.23: Characteristics of Farms in Different Size Categories ......................................................... 53 Table 2.24: Determinants of Farm Size ............................................................. 53 Table 2.25: Relationship between Land and Standard of Living .55 Table 2.26: Standard of Living as Determined by Use of Land: Multinomial Logistic Analysis using CATMOD .55 Table 2.27: Nonfarm Business Activities by Categories of Respondents .......................................... 56 Table 2.28: Types of Nonfarm Business Activities in the Sample .......................................... 56 Table 2.29: Profitability of Nonfarm Business Activities .......................................... 56 Table 2.30: Structure of Cash Family Income .......................................... 57 Table 2.3 1: Disposition of Farm Production among Sales, Consumption, and Other Uses .58 Table 2.32: Imputed Family Income including Cash Income and Value of Own Consumption of Farm Products .58 Table 3. 1: Number of Accounts and Distance to Institutions: Savers with Deposit Accounts ............. 69 Table 3.2: Mean Distance to Bank, PKO BP, and Local Cooperatives, by Voivodship ....................... 70 Table 3.3: Binary Probit Estimates of Households with Savings Accounts .......................................... 70 Table 3.4: Reasons for Not Requesting a Loan in the Past 12 months .................................................. 72 Table 3.5: Number of Loans Approved and Denied by Lender .......................................................... 73 Table 3.6: Selected Characteristics of Loans Granted by Major Lenders in 1999 ................................ 74 Table 3.7: "M ain" Uses of Loans by M ajor Lender .......................................................... 76 Table 3.8: Determinants of Households that Requested a Cash Loan in 1999 ...................................... 78 Table 3.9: Reasons for Not Requesting Cash Loans in Past Year ......................................................... 79 Table 3.10: Sources of Finance for Housing Improvements and Construction ....................................... 81 Table 3.11: Sources of Finance for Nonfarm Enterprises .......................................................... 82 Table 3.12: Determinants of Self-Finance for Nonfarm Enterprises ....................................................... 82 Table 3.13: The Probability that a Household Undertook any Farm Investment .................................... 84 Table 4.1: Net Farm Revenue Regressions .......................................................... 90 Table 4.2: Effects of Prices and Land over Factor Markets .......................................................... 91 Table 4.3: Estimated Elasticities of Net Revenue by Farm Size .......................................................... 92 Boxes Box 1: Transaction Costs in Land Sales ............................................ XII Box 2.1 Land Transactions ............................................. 41 Figures Figure 1: Number of Farms by Size Category - 1990 to 1998 ............................................ x Figure 1.1: Labor Allocation and a Labor Tax on One Sector ............................................ 15 Figure 1.2: School Attainment by Age: Rural Areas ............................................ 30 Figure 1.3: School Attainment by Age: Urban Areas ............................................ 30 Figure 2.1: Users of Ag ricultural Land, 1990-1997 ............................................ 34 Figure 2.2: Private Farms That Changed Their Size in 1990-96 ............................................ 36 Figure 2.3: Number of Farms by Size Category: GUS Surveys ............................................ 37 Figure 2.4: Number of Farms by Size Category: GUS Estimates ............................................ 38 Figure 2.5: Size Distribution of Individual Farms: 1998 and 1996 ............................................ 38 Figure 2.6: Organization of the Polish Cadastral System ............................................ 42 Figure 2.7: Poland: Land Transactions (Buying and Leasing) ............................................. 43 Figure 2.8: Nominal Price of Land and Consumer Price Index: 1991-98 ............................................ 45 Figure 2.9: Farm Size Distribution (farms that use land) ............................................. 47 Figure 2.10: Distribution of Land Ownership by Size .47 Figure 2.11: Farm Size Distribution .47 Figure 2.12: What the Family Budget Buys .54 Figure 2.13: Structure of Income: Farming Families .57 Figure 2.14: Structure of Imputed Income: Farming Families .59 Figure 2.15: Structure of Family Income: Book-Keeping Farms .59 Figure 2.16: Structure Family Income by Farm Size: Book-Keeping Farms .60 Figure 2.17 Family Income by Farm Size: Book-Keeping Farms .60 ACKNOWLEDGMENTS The study was carried out with the support and cooperation of the Polish Ministry of Agriculture, the Agricultural Property Agency of the Polish State Treasury, and the Polish Central Statistical Office - GUS. The assistance of many individuals in these organizations is gratefully acknowledged. This study was also made possible by the participation of three groups of Polish counterparts. First and foremost, the ZBES team, who shouldered the responsibility for the design and implementation of the rural household survey. This team included (in alphabetical order) Adam Czyzewski, Waldemar Dubla, Witold Orlowski, and Leszek Zienkowski, plus the largely anonymous group of experts at the GUS Lodz office who worked diligently and conscientiously on the survey questionnaires and the database. Second, the group of Polish experts who participated in the project's steering committee. This included Anna Szemberg, Andrej Rosner, Izaslaw Frenkel, Leszek Klank, and Wojciech Zientara. Their background papers provided important source material for the study. Finally, the work greatly benefited from access to the panel data of "book-keeping" farms regularly surveyed by the Institute of Agricultural and Food Economics (IERiGZ) provided by Lech Goraj. The team also gratefully acknowledges the guidance and support received from several World Bank colleagues. Mariusz Safin made important contributions to all four chapters of the study and provided assistance throughout the preparation of the study. Marcin Przybyla provided excellent research assistance. Jacek Wojciechowski organized the dissemination of the findings of this study. Ewelina Pusz, Malgorzata Dworzynska, and Magdalena Nowicka organized the team's visits to Poland and the conference where the findings of this study were presented. Anita Correa provided superb assistance in the processing of this study. The work on this study began under the direction of Michel Debatisse, and was completed by a team led by Carlos B. Cavalcanti. The team consisted of Alejandra Cox-Edwards, working on the rural labor market; Csaba Csaki, Zvi Lerman and Istvan Feher, with the assistance of Pepijn Schreinemachers, working on the rural land market; Timothy Guinnane working on the rural credit market; and Alberto Valdes, with the assistance of Gustavo Anriquez, working on the determinants of total farm revenue. Zvi Lerman was the overall scientific coordinator of the Bank-commissioned Rural Household Survey for Poland, referred to in this study as the Special Rural Household Survey. This study was written under the direction of Marcelo Selowsky, Chief Economist, ECAVP; Michael Carter, Country Director, ECC09; Kevin Cleaver, Sector Director, ECSSD; and Laura Tuck, Sector Manager, ECSSD. EXECUTIVE SUMMARY Introduction 1. This study identifies several factors that inhibit efficiency improvements in the farming sector, both in themselves and through the dynamics of their mutual interaction. The study observes that incentives faced in the labor market have important implications for the land structure and, in many ways, are at the heart of the problem of low labor productivity in agriculture. The study finds that, while rural households are increasingly diversifying their income sources out of farming, they have not moved away from rural areas because of the incentives to hold onto small agriculture plots and because of the high costs of formal employment outside agriculture. Instead, households have increasingly relied on so-called unearned income (pension benefits and other social transfers) and settled for informal employment in rural areas. This has done little to either advance the social condition of rural households, or to improve agricultural productivity. This incentive to hold onto small agricultural plots, as well as other factors limiting the availability of land for commercial farming, has lead to an increasingly polarized land structure, with small farms becoming smaller, and large farms consolidating, albeit slowly. 2. The study has five main findings:I * There is a heavy implicit tax on labor associated with crossing the frontier between farming and working in other sectors. The contribution to the regular pension system (ZUS) is over five times higher than the contribution to the farmer's pension system (KRUS), even though the two systems offer similar benefits. This, in turn, has important implications for the land structure: since KRUS eligibility requires rural residents to own at least one hectare of land, households hold onto these very small land plots, instead of releasing them for more efficient consolidation. This problem could be mitigated by, among other options, gradually raising the contribution rate for KRUS, while subsidizing contributions for low income participants. Another option would be simply close KRUS to new participants. The option of gradually increasing contribution rates has the advantage of not penalizing eligible participants close to retirement, while still taking steps to improve the financial situation of the farmer's pension fund. The findings of this study draws on the special rural household survey funded and designed by the Bank, and fielded in May 2000. The survey covered a sample of 2,835 households, including 1,320 nonfarrning and 1,515 farming households, in four rural Polish regions. The regions were: (i) Malopolskie, which is located in the center south and corresponds to rural Katowice and its surroundings; (ii) Mazowieckie, which is located at the center of the country, and corresponds to rural Warsaw and its surroundings; (iii) Wielkopoiskie, which is located in the center east, includes part of the frontier with Germany, and corresponds to the area that surrounds Poznan; and (iv) Zachodniopomorskie, which is located in the north west, is sparsely populated. The latter includes some of the most arable lands that were previously under the control of the State farms and that were liquidated in the early 1990s. The study also draws on three background papers commissioned by the Polish authorities on each of the factor markets: (i) Frenkel, Izaslaw and Andrej Rosner (2000), "Population and Labor Market in Rural Poland;" (ii) Szemberg, Anna (2000), "Transformation of the Agrarian Structure in the Private Farm Sector;" (iii) Zientara, Wojciech (2000), "Land Market in Poland in the Post-War Period;" and (iv) Klank, Leszek (2000), "Rural Financial Market in Poland." Executive Summary iv To achieve these two objectives, the increase in contribution rates would need to be spread over a 10 year period, and the funds for subsidizing contributions would need to be limited, ensuring the incentive to adequately target these subsidies. This would reduce the implicit tax of moving from farm to off-farn employment in rural areas, and would discourage rural households from holding on to small plots of land only to maintain eligibility for the farmer's pension system (KRUS). * The incentive to hold onto small plots of land has limited the availability of private land for the expansion of commercial farming. The study finds that only a small fraction of all private farms currently act as land suppliers, either through sale or leasing. Although the State continues to lease out land under its ownership, State land is really only available in significant amounts in western and northern Poland, where the number of private farms is relatively small (30 percent of the total). Also, most of the State land is under long-terrm leases, with relatively low lease payments and, therefore, is unlikely to return to the land market in the foreseeable future. * The problem of the low availability of land for the expansion of commercial farming is compounded by the high transaction costs for land trading. The cost of trading land is estimated at 12.5 percent of the value of the land sold, including a five percent treasury tax and several fees. This is high by the standards of other OECD countries (e.g., in the U.S., transaction costs are around 5 percent), and encourages informal transactions between individuals. The study finds that there is scope to reduce the cost of farmland transactions. Possible actions include (i) reducing the State Treasury tax; (ii) setting an upper limit on notary fees by defining a cap on the current percentage-based fee; (iii) standardizing contracts for either ownership transfer or lease of land; (iv) introducing a dispute settlement process accessible to all farmers; (v) improving and making more accessible the information system for land transactions prices and ownership; and (vi) eliminating the restrictions on using land as loan collateral. * Rural households rely very little on financial institutions either for savings, or for on- and off-farm investment. Less than 25 percent of all rural households report owning savings accounts, and most investments are funded from household savings. Supplier credits are important for large commercial farming operations, but these constitute only 12 percent of farms in Poland. The main reasons reported for relying so little on financial institutions are uncertainties about household incomes, and high direct and indirect costs of borrowing. The latter includes high interest rates, collateral, and fees, as well as the required paperwork and travel to the lender's office. This low level of involvement with financial intermediaries, either because of uncertainty or the cost of borrowing, implies that growth in rural areas will continue to be limited by the very low level of savings of rural households. * Farm consolidation should lead to higher productivity in agricultural production, with most of the increase in productivity coming from higher labor productivity. The productivity of farm labor, measured as farm value added per work day, rises Executive Summary v from a median of PLN 4 per work day in farms up to 1 hectare to over PLN 200 per day for farms larger than 30 hectare. Meanwhile, land productivity, measured as farm value added per hectare, remains relatively constant at PLN 3,000 per hectare across farms of different size, with the differences between farms sizes not being statistically significant. This indicates that farm consolidation will entail an overall reduction in agricultural employment. If, therefore, farm consolidation is to proceed in the interest of improved productivity and increased incomes for families that remain in farming, it needs to be accompanied by programs to develop non- farm employment opportunities in rural areas. In the absence of more non-farm employment opportunities, the gains from increasing farm productivity could be offset by the losses of households that relinquish their land and the associated income sources. 3. The rest of this executive summary reviews the evidence regarding the factors that constrain the functioning of rural factor markets.2 It begins by describing the recent trends in rural household incomes and activities, highlighting the increasing reliance on sources of unearned income (primarily pension benefits) and on informnal sector employment. Next, it examines the land structure that emerged after the first decade of transformation, with small farms getting increasingly smaller and large farms slowly increasing through consolidation. The following section attempts to measure the effect of these constraints on farming profitability by estimating a farm revenue function for four farm size categories - very small farms (less than one hectare); small farms (between 1 and 7 hectares); medium farms (between 7 and 15 hectares); and large farms (over 15 hectares). The second to last section presents the policy agenda that emerges from this study, while the final section describes each of the individual chapters of the study. Rural Households' Income and Activities 4. This section reviews the evidence on rural households' income and activities. It finds that rural households are increasingly diversifying income and employment away from farming, relying more on unearned income (e.g., old-age and disability pensions), as well as on employment in informal sector activities. However, since eligibility for these sources of unearned income (KRUS old-age and disability pensions) requires ownership of at least one hectare of land, there are strong incentives to remain in rural areas. The challenge for rural households is not necessarily to reduce reliance on farming income-by and large, that has already happened. The challenge is to reduce poverty levels by moving households from lower- to higher-productivity activities. Within farming, this requires moving toward larger-scale farms. Outside farming, this requires (i) moving toward activities with higher value added, usually entailing formal employment, and (ii) relying more on outside sources of financing. While sources of unearned income will continue to be an important source of income for rural households during this transition, accelerating the transition will require changing the incentives associated with pensions and other social benefits. Eligibility for these benefits should depend more on need than on land ownership, and formal employment should not be penalized by high taxes on labor income. 2 This executive summary was prepared by Carlos B. Cavalcanti. Executive Summary vi 5. Income diversification. The diversification of rural household income out of farmning matches the overall decline in farming as a source of employment. This is clearly reflected in the breakdown by source of household income for both fanning and nonfarming households (Table 1). On average, off-farn salaries account for over one-third of household income, and unearned income (primarily pension benefits) account for another one-third. These two sources of income account for almost 90 percent of the income of nonfarning households, and just over 50 percent of the income of farming households. Table 1. Sources of Household Income All sample Nonfarming households Farmning households (n=2,835) (n=1,320) (n='1,515) Salaries 7,244 8,111 6,489 KRUS 2,459 1,679 3,139 ZUS 3,551 4,220 2,968 Other personal income 992 996 987 Total personal income 14,246 15,006 13,583 Nonfarm business income 1,747 1,847 1,661 Farn income (net of cost of purchased inputs and payments to nonfamily labor) 4,634 -- 8,877 Total family income 20,627 16,853 24,121 Average family size 3.9 3.2 4.4 Source: Special Rural Household Survey (2000). 6. Income diversification out of farming has happened despite farming households reporting, on average, significantly higher incomes. Farmning households (families that cultivate some land) report, on average, total income of 24,000 zloty, compared with only 17,000 zloty for families that do not cultivate land, and the income gap between these two household groups is statistically significant. Average figures conceal, however, the significant variation in farming incomes across farm sizes. When the income of farming households is broken down by both sources of income and farm size, not surprisingly, farming income rises with farm size (Table 2). It is clear that households in smaller farms derive a significant share of their income from outside of farming. Farm households cultivating less than 7 hectares earn, on average, over 80 percent of their income from salaries and transfers from the farmer's pension system (KRUS) and the regular pension system (ZUS). Also, while households on medium-size farms (7 to 15 hectares) derive a larger share of their income from farning (44 percent), they do not appear to be better off. Only households in large farms (over 15 hectares) both generate a large share of their income from farming and report higher household income. 7. Informal employment. The increasing reliance on unearned income among rural households is closely linked with the rise of informal employment in rural areas. The main reason is that access to pension and unemployment benefits creates barriers in the transition from informal to formal employment. This is reflected in the difference in contribution rates for the Farmer's Social Security Fund (KRUS) and the regular Social Security Fund (ZUS). Eligibility for the Farmers' Security Social Fund (KRUS) is dependent on owning at least one hectare of land, but the contribution rate is around one-sixth of the contribution to the regular Social Security Fund (ZUS). In principle, as soon as a worker gets a formal job he/she has to start making contributions to ZUS. However, the significant difference in contribution rates, especially considering that the two systems offer similar benefits, becomes equivalent to a tax on the movement of workers out of farm employment. Executive Summary vii Table 2. Sources of Farm Household Income by Farm Size' Income source < Iha 1-7 ha 7-15 ha >15 ha Al Household Annual Income 20,140 18,189 18,621 53,198 28,313 Per Capita Annual Income 6,794 4,381 4,734 11,751 6,737 Net Farm Income 2,189 1,295 8,203 39,034 13,577 Gross Farm Income 4,117 6,340 21,587 77,985 29,819 Inputs Costs 1,912 4,882 13,189 37,711 15,781 Labor Costs 16 163 195 1,240 462 Net Income From Nonfarm Business 348 1,773 183 4,757 2,134 Salaries 8,609 7,839 4,116 3,057 5,706 Income From Transfers 8,300 6,949 5,622 5,776 6,428 KRUS 2,827 2,842 3,851 4,580 3,56 zus 4,937 3,418 1,372 815 2,343 Other Income 694 333 497 574 468 Sale of Assets (excluding land) 0 12 4 178 5 Sale of Land 622 152 394 20 208 Source: Special Rural Household Survey (2000). 1) Farms that did not report fann revenues were omitted. 8. The increase in 'informality' of employment in rural areas is reflected in the composition of employment status by sector (Table 3). A survey carried out by the Central Statistical Office in 1998 indicated that some of the sectors with the highest share of employment in rural areas (trade and repair, transportation, and construction) were, after agriculture, the sectors with the highest degree of informal labor contracts. According to the respondents, their decision to work without registration reflected (i) the absence of formal jobs, (ii) the inclination to avoid social security contributions, and (iii) the desire to increase their income. In addition, 10 percent of the respondents said that they decided to work unregistered so as not to lose their rights to other sources of income, namely pensions and social benefits. 9. The practice of collecting pension benefits and other social transfers while still working is reported to be prevalent in rural areas. For instance, during the week of the February 1999 Labor Force survey, 14 percent of old-age pensioners, 19 percent of disability pensioners, and 6 percent of unemployment benefit recipients reported having worked in rural areas. They report working as self-employed, part-time, or unpaid workers, since the transfers listed above are incompatible with formal work. Hence, it follows that, at least in rural areas, the full time work category corresponds roughly to the 'formal' sector, and the other work categories are in the 'informal' sector. By this definition, about one-half of the workers in rural areas (or 20 percent of all workers, roughly 3 million individuals) are in the 'informal' sector. While this includes almost all workers in agriculture and fishing, it also includes around 20 percent of the rural workers employed outside agriculture. Executive Summary viii Table 3. Employment Status b Sector of Occu ation in Rural Areas Sector Self Full time Part time Unpaid Share in total employment (%) Agriculture and fishing 71.4 7.0 0.8 20.8 45.9 Mining and quarrying 0.0 99.4 0.6 0.0 1.5 Manufacturing 6.1 88.6 4.6 0.7 1.7 Electricity, Gas and water supply 0.0 97.3 2.7 0.0 0.7 Construction 13.2 82.8 3.8 0.2 5.2 Trade and Repair 26.5 64.8 5.9 2.9 8.7 Hotels and restaurants 19.2 72.1 7.7 1.0 1.0 Transportation 14.1 82.6 2.9 0.4 4.3 Financial intermediation 5.5 89.0 5.5 0.0 1.0 Real Estate and other business activities 10.2 78.0 10.2 1.7 1.1 Public administration and defense 0.0 95.1 4.9 0.0 3.0 Education 0.2 92.1 7.7 0.0 4.6 Health 2.5 93.4 4.1 0.0 3.8 Other Community Services 17.2 72.8 10.0 0.0 1.7 Total 38.2 48.7 3.1 10.0 100.0 Source: State Statistical Office, Labor Force Survey, February 1999. 10. Nonfarm business activities. The shift toward 'infornal' activities outside agriculture is also reflected in nonfarm business activities identified in the Special Survey of Rural Households. Although nonfarm business activities are reported by only 8 percent of respondents, the survey found that most nonfarm businesses were in areas where inforrnal employment predominates, such as trade and repair, transportation, and construction. Very few respondents reported businesses in farm-related services, such as equipment leasing, field services, processing of farm inputs, marketing for other farmers, and delivery of farmn inputs. This suggests that nonfarm businesses aim primarily at diversifying away from activities related to farming (Table 4). Table 4. Breakdown of Nonfarm Business Activities Number of households Share of nonfarm businesses (%) Trade and repair 71 30.2 Construction 47 20.0 Manufacturing 29 12.3 Transportation, storage, packing 22 9.4 Repair work 10 4.3 Equipment leasing 0 0.0 Mechanical field services 4 1.7 Processing of farm products 4 1.7 Collection/sale from other farms 5 2.1 Delivering inputs 3 1.3 Restaurant, lodging 4 1.7 Training, teaching 3 1.3 Other 33 14.0 Total 235 100.0 Source: Special Rural Household Survey (2000). 11. The sources of financing for nonfarm businesses provide additional evidence that income diversification reflects uncertainty regarding income streams in rural areas. Among the nonfarm Executive Summary ix businesses that used outside financing, most of the financing came from personal savings (50 percent). Financing from personal savings was almost three times higher than what was financed from commercial and cooperative banks (Table 5). These nonfarn businesses are not, as one might expect, small businesses. The mean (median) net revenue for all nonfarm enterprises was PLN 34,900 (10,900), which was the equivalent to around US$ 8,800 (US$ 2,760) at the time. Although the mean (median) income of these nonfarm businesses that did not use outside financing was actually higher than the mean (median) income of those nonfarm businesses financed by loans from cooperative banks, their reliance on own savings likely limits their growth opportunities. Table 5. Sources of Finance for Nonfarm Businesses Source of Finance Number of persons involved in % nonfarm businesses Did not need it 40 16.5 Borrowed from commercial or State banks 20 8.3 Borrowed from cooperative banks 22 9.1 Borrowed from friends and relatives 19 7.9 Borrowed from NGOs 3 1.2 Personal savings 122 50.4 Grants or inheritance 3 1.2 Other 13 5.4 Total 242 100.0 Source: Special Rural Household Survey (2000). 12. Having diversified out of agriculture, the challenge for rural households is to move from lower- to higher-productivity activities. This requires moving toward activities with higher value added, which typically requires relying more on sources of outside financing. This also requires moving to farms with large-scale operations. We turn our attention next, therefore, to some of the obstacles to achieving this second objective: the current rural land structure and the functioning of the rural land market. Rural Land Structure and the Rural Land Market 13. This section examines the rural land structure that emerged during the first decade of the transition, and how the rural land market has developed. The section finds an increased polarization of land holdings, with a rising share of both very small farms (less than one hectare) and relatively large farms (over 15 hectares). It also finds an increased reliance on land leasing to transfer land into the hands of private commercial farms. This transfer is proceeding slowly, however. This slow transfer reflects (i) the limited availability of State land in areas with a relatively higher concentration of private commercial farming, and (ii) relatively few land transactions among private landowners. The latter in turn reflects the incentive to hold onto small plots of land and high transaction costs in land sales. 14. The starting point. Unlike other countries in the region, Poland did not introduce sweeping collectivization after World War II, and individual farms operating on privately owned land continuously dominated its agriculture. However, during the Soviet era, private farming in Poland did not experience the growth and consolidation that marked private farming in market economies. Government policies were restrictive, limiting farm sizes and the operation of farms. Executive Summary x This, de facto, constrained the efficient functioning of land markets. Polish agriculture, therefore, despite private land ownership and individual farming, remained much less efficient than agriculture in market economies, and less efficient than even the collective agriculture in certain Central European countries, such as the German Democratic Republic, Czechoslovakia, and Hungary. 15. Land distribution. Since land ownership in Poland was already highly skewed toward small-sized farms at the beginning of the transition, the incentives which households faced to hold onto small plots of land, and other aspects about the functioning of the land market, only accentuated this profile of land distribution. Surveys conducted by the State Statistical Office (GUS) indicate that between 1990 and 1998, the number of small farms with 1 to 5 hectares increased from 53 percent to 56 percent of all farms, while the number of relatively large farms with more than 10 hectares increased from 17 percent to 19 percent, and the number of mid-sized farms with 5-10 hectares declined from 30 percent to 25 percent (Figure 1). This skewed distribution of rural land ownership is a particular problem for Poland, because it began the transition period in a very different position from other Central and Eastern European Countries. With over three-quarters of Poland's agricultural land already cultivated by relatively small family units in 1990, land consolidation through privatization and land restitution was less of an option in Poland than elsewhere. Figure 1. Number of Farms by Size Category - % of Farms - 1990 to 1998 60 l ' ~~small 50 - _ _. - medium 40 - -large 30 - 20 - e e * * * v 10 0- 1990 1991 1992 1993 1994 1995 1996 1997 1998 Source: Polish Central Statistical Office (GUS). 16. Availability of State Land. The problem of low availability of State land is compounded by the fact that State land is unevenly distributed across regions within Poland. Most State land is concentrated in western and northern Poland, where the number of private farms is relatively small (about 30 percent of the total). In these regions, the State Land Property Agency (APA) accounted for 60 to 80 percent of land transactions between 1990 and 1998, compared to only 30 to 40 percent in of land transactions in the central-eastern and southern parts of Poland. Also, the average lease contract in western and northern regions was around 15 hectares, compared with less than 5 hectares throughout the rest of the country (Table 6). The availability of State land in specific regions emerges as an important factor in determining the growth pattern of private farms across the country. Executive Summary xi Table 6. Territorial Distribution of Land Transactions, 1990-1996 Regions Share of farms participating in Average transaction per farm, ha land transactions, % Purchased Leased Northeast 18.3 5.4 13.0 North 23.1 10.1 25.0 Central-west 15.5 6.3 11.3 Southwest 20.1 6.6 15.4 Central 10.0 3.8 4.8 Capital 9.2 3.1 5.9 Central-east 13.7 2.9 4.4 South 8.6 4.4 4.4 Southeast 6.9 1.8 3.0 Source: Institute for Agricultural and Food Economics. 17. Land leasing. Land leasing also emerges as a more common arrangement than land sales. This reflects two factors. First, transaction costs on land sales are high, with the main single component being a treasury tax set at 5 percent of the value of land. There are also a wide range of fees and charges for the preparation of maps, extracts, and new entries in the mortgage register (when land is mortgaged). The example reported in Box 1 provides a detailed cost calculation of these transaction costs, and indicates that total transaction costs add up to around 12.5 percent of the price of the land sold. Second, leasing dominates State-owned land transactions. During the 1990s, only about 15 percent of State land was sold, with the rest being leased out to private users and about 10 percent remaining as reserve.3 18. This reliance on land leasing to transfer State land to private users reflects, among other factors, pending issues with land restitution and privatization. According to the State Agricultural Property Agency (APA), the lack of a restitution law is a major factor blocking large-scale sale of leased State land. APA officials claim that thousands of claims have been filed by former landowners all across the country. Since there is no mechanism for dealing with these claims so far, even suspicion of a claim is sufficient to block APA's plans to sell a land plot. The absence of a restitution law is compounded by two other problems. The land area represented by the large number of restitution claims exceeds by many times APA's land reserves, and land that can be potentially claimed by former owners is already leased out to other users on 10- to 15-year leases. The excessive number of claims results, in part, from the fact that land expropriated from large landowners (with holdings of more than 50-100 hectares) during the post-war land reform was partitioned and distributed to individuals between 1944 and 1950. This land is, therefore, no longer controlled by the State. Also, some of the land that can be potentially claimed by former owners is leased out on long leases to others users, whose interests must be considered if any privatization is attempted. All these factors create tremendous difficulties for restitution of land to former owners, and explain why restitution of land has played a much less prominent role in Poland than in other Central and Eastern European countries during the 1990s. It also explains why APA has relied primarily on land leasing to transfer State land to private land users. 3 The 500,000 hectares of State land reserves are mainly poor-quality land in remote locations, and are very difficult to place. Executive Summary xii Box 1. Transaction Costs in Land Sales In 1998, two private individuals agreed on a transaction involving 3 hectares of undeveloped agricultural land at a price of 9,000 zloty. The costs of preparing the land survey documents were to be borne by the seller, and the remnaining costs were to be borne by the buyer. The calculation of transaction costs (in zloty) were the following: Cadastral map extract and mortgage register extract 80.00 Notary fee 270.00 Fee for extracts (6 extracts, 4 pages each, 5 zloty per page) 120.00 Treasury tax (5% of transaction value 450.00 Fee for setting up the mortgage register 30.00 Fee for entering the ownership rights into the mortgage register 144.00 Fee for printing the mortgage register 9.15 Total 1,103.15 This amounts to 12.5% of the value of agricultural land sold to the new owner. 19. The Polish authorities have considered several draft restitution laws. This is clearly an issue of great complexity, where it is important to reconcile as far as possible legal, political, and economic efficiency considerations. International experience argues in favor of settling existing claims to the extent possible. Otherwise, this will remain a pending issue. Whatever solution is adopted, it needs to create a clear basis for land ownership rights in the future, as this is an essential element for fostering the development of a land market that can direct resources progressively towards their most efficient use. It would be preferable to do this as soon as possible. The recent experience of other Central and Eastern European countries shows that attempts to restitute physical plots to claimants in the same location can be a costly and very time-consuming procedure; it may, therefore, be preferable to consider an approach based on a flexible, voucher-based mechanism that allows exchange of location and assets. 20. While land leasing is common with State land, it is not particularly widespread among private landowners. The special rural household survey found that leasing out of land is reported only by 8 percent of households with privately owned land. Among farmers, only 17 percent lease in land, and the great majority (83 percent) rely entirely on their own land. Farmers operating with leased land accumulate much larger holdings (Table 7). They cultivate, on average, 25.7 hectares, compared with 7.3 hectares for farmers who do not lease in land (the difference is statistically significant). Shortage of manpower is the main reason given by landowners for leasing out land (42 percent), followed by lack of profitability (26 percent). Landowners who lease out land are older, have smaller families, and control a smaller land endowment than those who do not lease out. Lessors on average lease out more than 60 percent of their land. Table 7. Size and Structure of Farms With and Without Leased Land All farms (respondents with land) Farmns with leased land Farns without leased land Number of farms 1515 (100%) 255 (17%) 1260 (83%) Farm size (ha.) 10.4 25.7 7.3 Own land 8.4 13.8 7.3 Leased land 2.0 11.9 0 Source: Special Rural Household Survey (2000). Executive Summary xiii 21. Among private landowners, only one-third of land is leased, and only about one-half of this land is covered by formal lease contracts. The limited number of formal leasing contract reflects, in part, lack of familiarity with formal dispute resolution mechanisms in connection with leased land. Among the respondents to the special survey who leased out land, 12 percent claimed that no dispute resolution mechanisms existed, 34 percent did not know what dispute resolution mechanisms were available, and 53 percent cited strictly informal dispute resolution mechanisms, such as talking the problem out between the parties or going with the problem to the chairman of the municipal (Gmina) council (Table 8). Table 8. Dispute Resolution Mechanisms (multiple answers allowed) Percent of respondents with leased land (n=255) Talk the problem over between the parties 46 Go to Gmina chairperson 7 Go to the court 3 Appoint an arbitrator 2 Other method 9 No dispute resolution mechanism available 12 Don't know 34 Source: Special Rural Household Survey (2000). 22. Buying and Selling of Land. Land purchases and sales are also not very frequent among private landowners. Only about 5 percent of respondents to the special survey reported selling or buying land in the five years between 1995 and 1999. Another 5 to7 percent have plans to buy or sell land in the near future. The great majority (82 percent of respondents), however, have no plans to acquire additional land in the immediate future. The average sale transaction between 1995 and 1999 involved 2.6 hectares at a price of 1,800 zloty per hectare (estimated by regression analysis). The average buy transaction involved 6.7 hectares at a price of 2,300 zloty per hectare (Table 9). The difference in prices is not statistically significant, and the survey suggests average land prices of around 2,000 zloty per hectare. This result is confirmed by the responses of potential buyers (5 percent of respondents) who report their willingness to pay an estimated price of 2,350 zloty per hectare. Since the average rate for land leasing is estimated at 80 zloty per hectare, the price of land in buy-and-sell transactions is equivalent to about 25 years of leasing. Table 9. Frequency of Land Transactions and Estimated Prices per Hectare Land transactions Percent of respondents Hectares (mean) Prices (regression estimate) Sell 1995-99 4.8% of all (n=2835) 2.6 1,800 (R-sq=0.05) Buy 1995-99 6.4% of all (n=2835) 6.7 2,300 (R-sq=0.47) Planning to sell 7.4% of owners (n=1709) NA NA Plan to buy' 4.9% of all (n=2835) 12 2,350 (R-sq=0.60) Lease in 17% of farmers (n=1515) 12 79 (R-sq=0.28) Lease out 8% of owners (n=1709) 3.4 80 (R-sq=0.18) 1) Of these potential buyers, 96% already have land. Source: Csaki and Lerman (2000). 23. The problems with State land transactions and the relatively small number of land transactions among private landowners indicate that there are limited opportunities for growth in private commercial farming. While more needs to be done to improve the functioning of the land market, more also needs to be done to understand how the availability of land and other factors of production affect farm revenues. We turn next, therefore, to this issue. Executive Summary xiv Determinants of Total Farm Revenue 24. What can we infer about the opportunities and constraints for farming sector restructuring beyond the link between labor market incentives and land structure that emerges from the analysis to this point? This section presents the results from the estimations of a short-run revenue function (per household) using the data made available by the Special Rural Household Survey for 2000.4 This includes data on farm output, prices, quasi-fixed factors, and household characteristics. In addition, a separate analysis by farm size was estimated, allowing the estimations to capture the relation between the returns from the various factors and farm size. Farms were grouped into four farm size categories: (i) minifundia, less than one hectare; (ii) small, between 1 and 7 hectares; (iii) medium, between 7 and 15 hectares, and (iv) large, more than 15 hectares. 25. The regression results explain almost 76 percent of the variability in the farmer's household revenues, and particularly significant among the variables are prices, inputs, and demographic characteristics. The regression coefficients were then expressed in terms of revenue elasticities-the percentage change in farm revenue associated with a given percentage change in the independent variables (Table 10). Table 10. Estimated Elasticities of Net Revenue by Farm Size All Sample Minifundia Small Medium Large Labor -0.042 0.001 -0.134 -0.035 -0.050 Education 1.716 10.723 11.127 3.220 0.615 Land 1.082 0.218 2.136 1.364 0.989 Specialized Machinery -0.345 -1.406 -2.128 -0.645 0.089 Tractors 0.593 0.228 1.626 1.056 0.630 Fragmentation -0.281 -0.285 -1.809 -0.512 -0.115 Works Outside HH -0.079 0.321 0.668 -0.051 -0.072 Investment 0.193 0.000 0.651 0.224 0.151 KRUS Enrollment 0.198 0.249 0.770 0.508 0.109 Source: Valdes and Anriquez (2001). 26. The results provide strong support to the argument that the farm-household allocation process is a joint decision and that studying each factor market in isolation fails to capture the full effect of relevant determinants. The results also confirm three main points highlighted in the earlier analysis: (i) farm revenues for small and medium-sized farms increase with land size, indicating that land consolidation in the 7 hectare range, through either land purchase or land lease, would increase overall farming income; (ii) returns on investment, including human capital investment, are high, indicating that the farming sector would benefit from greater access to outside savings; and (iii) returns on hired labor increase with land size, indicating that labor productivity rises with land size; and (iv) the returns from work outside the household are high 4 In this exercise, farm revenues are defined as total revenues accruing from crops, livestock, and live animal production, net of purchased inputs and out-of-farm services. The analysis is short-run because of the presence of quasi-fixed factors, household characteristics, and spatial location. The choice of a flexible functional form was deterrnined largely by the data characteristics and by the focus of the analysis. The specification chosen allowed the estimated elasticities of land, labor, capital, and other variables to vary across the sample according to the differences in land size, input use, capital, education, access to pension payments, and other factors. See Chapter 4 for a complete description of the estimation method. Executive Summary xv for families living on farms below 7 hectares, suggesting that households living on small farms should be encouraged to diversify sources of income. The latter should persuade policymakers to reduce barriers in the transition from informal employment on farms to formal employment in rural areas. It should also persuade policymakers to encourage land consolidation by reducing transaction costs on land sales and by streamlining and simplifying land registration procedures. 27. The main findings are summarized as follows: * The returns to land are positive and, except for very small farms, the land-to- revenue elasticities diminish with land size. The elasticity of land is high for small and medium-sized farms (2.1 and 1.3, respectively), indicating that an increase of farm size at this range by one percent increases revenues by more than one percent. However, minifundia and large farms have an elasticity of land below unity (< 1.0). The minifundia result indicates that there is a minimum farm size necessary for land to be more productive, while the large-farm result suggests that returns on land diminish beyond a certain size. * The farms that had made an investment (less than 5 percent of the farms) had revenues 19 percent higher that those who did not, ceteris paribus. These investments were more productive for the small and medium-sized farms, and were particularly more profitable in the livestock sector.5 * The returns on capital assets vary with farm size. For example, returns on tractors are higher for small and medium-sized farms, but considerably lower for very small farms (minifundia). On the other hand, specialized machinery (milking machines and other equipment) exhibits positive returns only for the largest farms. * The returns on work done outside the farm decrease with land size. For smaller farms, it is more profitable for the head of the household to work outside the farm. Farmers with more land to work find that working outside the farm leads to lower returns. This is consistent with separate calculations indicating that farm revenues raises faster than needed labor input with land size.6 These calculations of labor productivity, measured by farm value added per working day, suggest that labor productivity rises from a median of PLN 4 per work day in farrns up to 1 hectare to over PLN 200 per day for farms larger than 30 hectare. However, farm land productivity, measured as farm value added per hectare, remains constant at around PLN 3,000 per hectare across different farm sizes, with the differences between farm size categories not being statistically significant (Table I1). 5Returns on capital investment are net of depreciation allowances. 6 Lerman (2001). Executive Summary xvi Table 11. Productivity of Land and Labor by Farm Size Value added, Work days Value added Value added Value added Value added zloty per hectare per work day per hectare per work day (mean) (mean) (median) (median) Up to 1 ha 1,073 177 2,758 5 2,808 4 1-2 ha 4,875 311 3,274 26 2,731 10 2-5 ha 10,062 431 3,159 44 2,633 16 5-7 ha 16,545 550 2,780 47 2,370 26 7-20 ha 35,520 506 2,947 133 1,947 40 20-30 ha 73,915 542 2,987 209 1,743 77 Over 30 ha 151,110 493 2,945 1,009 2,023 226 Source: Lerman (2001). * The levels of schooling are very low for all farm sizes (9.66 years on average), and, not surprisingly, the observed education elasticities are very high. For example, a farmer with one year of education above the mean would (other things being equal) have 17 percent higher revenues. The Policy Agenda 28. A policy agenda aimed at encouraging the expansion of private commercial farming in Poland needs to acknowledge the link between social policies and the current land structure. It also needs to acknowledge that more could be done to improve the functioning of the land market. The latter includes lowering the transaction costs in land sales, and reducing institutional and procedural bottlenecks to make the market more efficient. This section reviews policy actions that would contribute toward reversing this vicious cycle, whereby rural households hold onto small plots of land to ensure eligibility for social transfers, and thereby slow down land consolidation. 29. Actions aimed at improving the functioning of the labor market and better targeting social benefits in rural areas include the following: * Encouraging self-employed workers in rural areas to contribute to the regular pension system (ZUS), rather than the farmer's pensions system (KRUS). One option would be to gradually increase the contribution to KRUS to levels comparable to ZUS over a 10 year period. Another option is to simply close KRUS off to new participants. The first option would reduce the incentive for self- employed workers in rural areas to hold on to one hectare plots of land to ensure eligibility for KRUS. Self-employed workers in rural areas unable to meet this requirement would have their contribution subsidized, although the funds available for this subsidy scheme would be limited to minimize the risk of abuse. The second option would take away any incentive altogether, but at the cost of disenfranchising eligible low income farmers close to retirement. * Excluding the size of land holdings from consideration when defining eligibility for unemployment benefits. The current eligibility requirements for unemployment benefits exclude rural residents with farm holdings of more than two hectares of Executive Summary xvii land. The evidence from the special survey indicates, however, that access to two hectares of land does not necessarily guarantee that a household can survive by farming, especially in poorer farming regions. Furthermore, the two-hectare limit acts as a deterrent to farm consolidation beyond that limit. The elimination of this eligibility criterion would still require that eligible unemployed workers to meet all the other criteria for receiving unemployment benefits,7 limiting the scope for abuse. Taking steps to improve education and training in rural areas. The survey results indicate that farmers with one year of education above the mean have (all other things being equal) 17 percent higher revenues. Also, previous analysis carried out by the World Bank shows that new labor market entrants with only basic vocational education or less, especially in rural areas, face greater difficulties in finding employment.8 This analysis suggest that improving access and the quality of education and training in rural areas requires further progress in moving away from vocational education programs, emphasizing instead general education that provides skills that are broader, transferable, and in greater demand. Actions to complete this transition can be aided by the introduction of a national learning assessment. This should help guide policymakers in monitoring quality and in allocating funds across educational programs and regions, targeting those in greatest need. In training programs, emphasis should be given to programs that target particular problems (e.g., skills in short supply), and individuals whose problems are only moderately severe. 30. Actions that would contribute to improving the functioning of land markets include the following: * Reducing the transaction costs on land sales. As indicated above, the cost of land transactions is fairly high, especially when compared to the advanced market economies. The total cost of a land transaction may be as high asl2.5 percent of the sales value, with the 5 percent treasury land transaction tax being the largest single component. This is a significant barrier to entry. Given the current conditions of Polish agriculture, where land consolidation would greatly contribute to improving returns on farming, reducing land transaction costs and simplifying the tax and fee structure should be a priority. Options include eliminating the 5 percent treasury tax on land transactions and setting an upper limit on notary fees. The latter could 7To be eligible for unemployment benefit (UTB), it must be the case that: (1) no job offers are available, no training or retaining is available, no intervention works or public works job are available, and no additionally created work places are available, and (2) in the 12 months before registering as unemployed the claimant worked at least 180 days covered by social insurance. However, the 180-day employment condition is not binding if the claimant falls under certain categories. These include having (1) been laid off by the employer because of economic difficulties; (2) left school in the last 12 months (recent school graduate); and (3) been re-employed after a period of collecting unemployment compensation, but the period was shorter than 180 days because of the economic difficulties faced by the employer. 8 World Bank (2001). Executive Summary xviii be achieved by setting a ceiling on the current percentage-based fee for land transactions. * Reviewing the land ownership registration system to streamline and simplify the cumbersome and time-consuming procedures. Although the land registration law is basically sufficient, the treatment of "unregistered rights" requires attention. Estimates indicate that 30 percent to 40 percent of agricultural land rights are not registered to reflect the current ownership. The registration procedures are very time-consuming, and the system is seriously under-funded, with a lack of adequate personnel resources. * Improving access to land mortgage facilities by streamlining and reducing the restrictions on using land as loan collateral by, for instance, increasing the percentage of the value of real estate that can be mortgaged from the current 60 percent to at least 70 percent or 80 percent. Establishing a system for monitoring and reporting all land sales. This would increase the transparency of land markets and improve information flows, ensuring that market forces can function as the primary engine for land consolidation. 31. It is important to note however that land consolidation and the increase in farm land productivity will result from the reduction in the number of number of self-employed farmers. A consolidation of farms in the interest of improved productivity and increased incomes for families that remain in farming needs therefore to be accompanied by programs to develop non- farm employment opportunities in rural areas. In the absence of more non-farm employment opportunities in rural areas, the gains from increasing farm productivity could be offset by the losses of households that relinquish their land and the associated income sources. Outline of the Study 32. The rest of the study develops the themes identified in this executive summary by examining the three principal rural factor markets-labor, land, and credit markets. For each of these markets, the study assesses the intra-market situation regarding its key constraining factors. Furthermore, since many of the decisions regarding factor allocations are joint decisions, the study also analyzes the interaction among these three markets by considering decisions on one of these markets that have direct implications on the operation of the other markets. The study is organized, therefore, in four, self-contained chapters covering each of the rural factor markets- labor, land, and credit-plus a chapter that aims at capturing the interaction among these three markets by estimating a farm revenue function in which each of these factors of production is an explanatory variable. 33. Chapter 1 focuses on the current functions and characteristics of Poland's rural labor markets. It begins with a brief overview of the impact of the initial reforms on rural households. It turns next to five aspects of the operation of rural labor: (i) the significant shift in sectoral employrnent in the overall economy, placing the rural labor market in that context; (ii) the institutional and policy framework for business operation in rural areas, describing the basic Executive Summary xix regulations that apply to formal employment; (iii) the composition of rural employment by sector and the evidence on wages; (iv) migration pattems and the links between employment creation and population movements; and (v) the time allocation of young individuals, focusing on school- to-work transitions in rural and urban areas. The chapter closes with critical areas for policy action. 34. Chapter 2 examines the current features of rural land ownership, rural land use, and rural land markets in Poland. It begins by comparing farm structures and rural land use patterns in developed market economies and Poland, showing that Poland has a fairly fragmented structure of rural land ownership and use. It finds that a large number of very small farms operate on a relatively large proportion of the available agricultural land. Although there is a positive correlation between the amount of land used by the household and the level of family income and well-being, consolidation is taking place very slowly, and it has not achieved the scope that would produce significant results in the foreseeable future. The chapter closes with a discussion of measures to achieve a fully functioning rural land market, focusing on actions to deal with pending land restitution, to reduce obstacles to land transactions, and to facilitate the consolidation of medium sized farms. 35. Chapter 3 reviews the operation of the rural credit market and finds that rural households report very low levels of borrowing from either formal or informal lenders. The chapter sets out to understand the main reasons for the very low levels of borrowing. It begins by reviewing the basic institutional background of the main lending institutions for rural areas. Next, it provides an overview of the main features about savings and borrowing in rural areas, and develops and tests two main hypotheses about reasons why so few rural households borrow. The first hypothesis is that the high level of uncertainty in rural areas reduces the demand for credit. The second hypothesis is that the high direct and indirect costs of borrowing discourage rural households from borrowing and banks and other potential lenders from lending. The chapter finds direct evidence supporting these two hypotheses. The chapter closes with a summary of the policy implications of these findings. 36. Chapter 4 presents the results from the estimations of a farm revenue function. The results provide strong support to the argument that the farn-household allocation process is a joint decision and that the relevant determinants in each market are properly captured by the farm revenue function. The results confirm three main points highlighted in the first three chapters: (i) farm revenues for small and medium-sized farns increase with land size, indicating that land consolidation in the 7-hectare range, through either land purchase or land lease, would increase overall farming income; (ii) returns on investment, including human capital investment, are high, indicating that the farming sector would benefit from greater access to outside savings; and (iii) returns on labor increase with land size, suggesting that households living on small farms should be encouraged to diversify sources of income. CHAPTER 1: RURAL LABOR MARKET Introduction 1.1 The labor market plays a central role in the transmission of economic incentives. However, the precise signals to which individuals pay attention vary with the economic system's organization. Until 1990, the majority of Polish enterprises were State-owned, and organized to comply with production targets and maintain employment, but rarely to minimize costs. For workers, the link between their standard of living and the survival of the enterprise in which they worked was extremely tight, and the notion of job mobility from firm to firm was alien. At the same time, the concept of supplementing income from work with other sources of income from market-driven activities was well developed. These activities, better characterized as fixed-term or task-driven employment contracts include: "praca zlecona," in the case of specific services that can be contracted out; "pracownik sezonowy," in the case of peak season tasks in agriculture; and "umowa-zlecenie" or "umowa o dzielo," in the case of consultant-type activities among professionals. 1.2 The process of market liberalization that began in the early 1990s has changed the composition of employment, reducing the number of life-time jobs and replacing them with other implicit or explicit contracts that are more compatible with competition and private enterprises. The novelty is not that private or temporary labor contracts are being used, since these have been part of the Polish reality for a long time. The key change is that these activities can become more important in terms of their contribution to individual and household budgets to the extent that they are brought out of the gray economy. However, there are serious barriers to this process, because the formal economy is heavily taxed. 1.3 This chapter focuses on the current functions and characteristics of Poland's rural labor markets.9 It begins with a brief overview of the impact of the initial reforrns on rural households. The subsequent presentation emphasizes five aspects of the operation of rural labor markets. First, it describes a significant shift in sectoral employment and places the rural labor market in that context. Second, it reviews the institutional and policy framework for business operation in rural areas, describes the basic regulations that apply to formal employment, and draws on interviews conducted in November of 1999 to describe how employers and workers deal with the current regulatory framework. Third, it uses the February 1999 Labor Force Survey to analyze the composition of rural employment by sector and the evidence on wages. In addition, it uses a special household survey carried out in May 2000 in four rural regions to examine more detailed aspects of labor market participation and labor utilization in rural Poland. Fourth, it examines the existing literature to describe migration patterns, and, in particular, the links between employment creation and population movements. Fifth, it examines the time allocation of young individuals, focusing on school-to-work transitions in rural and urban areas. The final section lays out the critical areas for policy action. 9This chapter was prepared by Alejandra Cox Edwards, CSULB. Chapter 1: Rural Labor Market 2 1.4 Poland's population stands at about 39 million, with 40 percent living in rural areas, which in Poland are defined as "all territory outside city limits." Under this definition, ninety- three percent of the territory is rural, and sixty percent of the territory is rural land. Although this administrative definition correlates with population density, there are many degrees of population density within rural areas. This explains why the population in rural areas as a whole has been very stable since the late 1 940s at about 15 million. 1.5 Poland's working-age population (age 15 to 64) is about 25 million, and 35 percent of this population lives in rural areas. However, an important fraction of rural workers work in urban areas or in nonfarm related rural jobs, a pretransition feature of the Polish economy. In the 1980s, dismissals from State enterprises, both agricultural and nonagricultural, had a particularly negative impact on employment for land-owning employees. These workers, mostly rural, were considered better candidates for dismissals, because they had alternative income sources. The decline in employment opportunities for rural workers brought about by reforms took place in the context of a constrained and increasingly distorted labor market. The development and opening of this labor market is essential to facilitate the reallocation of employment in the development process. This is the central subject of this chapter. Economic Reforms and Changes in Income Sources for Rural Households 1.6 We will use two approaches to examine the status of rural workers: (a) their household's key characteristics in terms of income sources, land-ownership, and other relevant factors; and (b) their individual employment and earnings. Households 1.7 According to the latest estimates, there are approximately 4 million rural households. Close to 60 percent of rural households own a farm or an agricultural plot (of at least lhectare [ha]), and the rest of the households are generally referred to as "rural nonfarm households." The farm/nonfarm household ratio has remained fairly stable since the beginning of the transition. However, there have been substantial changes in the main sources of living of rural farm and nonfarm households (see Table 1.1). 1.8 During the frst years of transition, the key component of farm-household income shifted from farming to unearned income. This shift was associated with three factors: (i) expanded social programs, such as accelerated and early retirement; increased number of persons receiving disability and family pensions, and the emergence of unemployment benefits; (ii) the decline in farm-related incomes; and (3) major shifts in nonfarm related incomes. The key component of nonfarm household income shifted out of hired labor-especially agriculture-related-into transfers. This shift was driven by dismissals from state enterprises, both nonagricultural and agricultural. The combined effect on rural farm and nonfarm households caused the share of rural households living mainly on wages and salaries to fall from 42 percent to 30 percent; the share of households living mainly on own-farm income to fall from 33 percent to 26 percent; and the share of households depending mainly on pensions or other state transfers to rise from 23 percent to 39 percent. Chapter 1: Rural Labor Market 3 Table 1.1. Rural Households by Exclusive or Main Source of Household Income, 1988 and 1995 Exclusive or main source of 1988 1995 1988 = 100 1988 1995 household income Thousand % TOTAL 4106 4116 100 100.0 100.0 Work on own farm' 1368 1062 78 33.3 25.8 Hired labor 1708 1252 73 41.6 30.4 in agriculture 375 148 39 9.1 3.6 outside agriculture 1333 1105 83 32.5 26.8 Self-employment2 71 148 208 1.7 3.6 Pensions 934 1501 161 22.7 36.5 Retirement 535 920 172 13.0 22.4 disability and other 399 580 146 9.7 14.1 Other unearned income 27 153 576 0.7 3.7 of which unemployment benefits (-) 108 X (-) 2.6 FARM HOUSEHOLDS3 2434 2304 95 100.0 100.0 Work on own farml 1368 1062 78 56.2 46.1 Hired labor 630 507 81 25.9 22.0 in agriculture 63 35 56 2.6 1.5 outside agriculture 567 472 83 23.3 20.5 Self-employment (b) 28 64 231 1.1 2.8 Pensions 405 644 159 16.6 27.9 Other unearned income 4 27 649 0.2 1.2 of which unemployment benefits (-) 20 x (-) 0.9 NONFARM HOUSEHOLDS 1673 1813 108 100.0 100.0 Hired labor 1078 745 69 64.4 41.1 in agriculture 312 112 36 18.7 6.2 outside agriculture 766 633 83 45.8 34.9 Self-employment2 44 84 193 2.6 4.6 Pensions 529 857 162 31.6 47.3 Other unearned income 23 127 563 1.3 7.0 of which unemployment benefits (-) 88 x (-) 4.9 1) or on agricultural plot (up to I hectare of agricultural land) 2) other than farming 3) including households of users of an agricultural plot Source: Frenkel and Rosner (2000), based on 1988 Census and 1995 Micro census. Chapter 1: Rural Labor Market 4 1.9 Frenkel and Rosner (2000) looked for more recent evidence on main sources of income using data from the Labor Force Survey. The Labor Force Survey (LFS) in Poland has been run since 1992, with the goal of tracking the evolution of labor market conditions and establishing a record of key indicators, such as labor force participation, employment, and unemployment rates. Their analysis suggests that, since the time of the 1995 Micro census up to the beginning of 1999, the share of households living mainly on unearned income did not change significantly. At the same time, the downward trend of the share of households depending mainly on hired work has been reversed. The share of self-employed households remained basically stable, and the share of households living mainly on income from farming continued to decline. The authors were able to capture similar trends from the Household Budget Surveys. Income Diversification within Households 1.10 From the point of view of income generation, existing information allows us to conclude that rural areas in Poland are currently predominantly nonagricultural. Approximately only one-fifth of rural households derive income mainly from farming. The rest of rural households are divided almost equally into (a) households deriving main income from nonagricultural occupations and (b) those with income from unearned sources (mainly pensions) (Table 1.2). Table 1.2: Households' Main Income Source ( percent) Income Source Rural Urban Total Nonfarm income households 37.0 52.7 46.6 Households of farmners 17.8 0.5 7.2 Households of self-employed 3.8 7.0 5.7 Households of retirees-pensioners 38.2 35.9 36.8 Households of unemnployed 0.9 0.8 0.9 Households maintained from other sources 2.3 3.1 2.8 Source: Polish Central Statistical Office, Labor Force Survey, February 1999. 1.11 In the 1995 Micro census the majority of rural households (67 percent) indicated that they had secondary sources of income. Among households living mainly on agricultural income, more than one-quarter depended exclusively on this source, more than one-half had additional unearned income, and almost one-fifth received income from work outside agriculture. 1.12 The picture captured by the LFS data suggests that rural households have diversified significantly out of the agriculture sector, both in terms of income sources and time allocation. We can also look at this issue using the Special Rural Household Survey 2000. In this case, individuals were asked whether they made any income from work during 1999. Subsequently, those that responded "yes" to the earnings question were asked to classify their main activity by sector and status. Using the answer to the earning question, and adding up by household, we generated Table 1.3. Chapter]: Rural Labor Market 5 Table 1.3: Distribution of Rural Households by Number of workers (%) Number of Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie earners None 15.70 17.43 15.57 31.30 One 20.37 18.61 25.25 33.33 Two 39.60 37.81 36.04 27.78 Three 12.16 14.92 13.46 5.42 Four + 12.17 11.23 9.68 2.17 Total 100.00 100.00 100.00 100.00 Source: Special Rural Household Survey 2000 1.13 As seen, the majority of rural households (except in the Zachodniopomorskie region) are composed of two or more income earners. It is therefore of interest to examine the degree to which income generation with households is diversified out of the agricultural sector. Using the answers to the "status" question, and aggregating by household, we generated Table 1.4. 1.14 We focus on the status classification of workers that belong to the households included in the second and third rows of Table 1.3.10 These two rows include households with one and two workers, and typically account for 60 percent of households in each of the rural voivodships (provinces). We organize these two groups of households according to the number of workers, and, within each of these categories, we examine the distribution of households by the working status of the labor force participants. Table 1.4: Distribution of One-Worker and Two-Worker Households According to Employment Status of Workers Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie One Earner Households Nonfarm Family Business 0.08 0.10 0.05 0.09 Family Farm 0.34 0.27 0.22 0.12 Employee, Private Sector 0.26 0.32 0.42 0.47 Employee, Public Sector 0.29 0.27 0.27 0.24 Two Earners Households Family Farm Only 0.31 0.45 0.27 0.26 Family Farm + Employee 0.25 0.18 0.14 0.10 Employees, Private or Public 0.25 0.25 0.43 0.43 Source: Special Rural Household Survey 2000 1.15 It is of interest to note the following: * The majority of one-earner households are employee households. That is, these households are already out of own-farming. * More than one-half of two-earner households have at least one working member engaged as an employee. The exception is the Mazowieckie region, where 45 percent of two-worker households are concentrated in the family farm category only. 10 The typical household in the first row has one main source of income-pension benefits. Chapter 1: Rural Labor Market 6 Individual Workers 1.16 During the period 1989-1993, rural employment fell, while the share of agriculture in rural employment rose. The rural labor market saw a reduction in the labor force-induced by the retirement of about 630,000 rural workers, who previously had jobs in state enterprises, and the additional retirement of farm workers who were eligible for retirement benefits. The Census data analyzed by Frenkel and Rosner (2000) indicate that, from 1988 to 1995, rural employment fell by about 867,000 and employment in agriculture fell by just 72,000. This reduction was composed of an increase in employment in individual farming (359,000), and a reduction in state farmns' employment (431,000). The combined effect of these changes caused the share of agriculture employment in rural employment to rise from 56 percent to 62 percent in the first period of transition.I 1.17 We do not have data to examine the latest trends in the Census framework. The available information with national representation is the Labor Force Survey (LFS). Unfortunately, the LFS survey results are not directly comparable to the Census data, because the LFS employment classification is based on activity in a given week, while the Census classification refers to a year. The difference in time frame causes a difference in the estimated fraction of agriculture employment in rural areas. In particular, in a short time frane, the share of farm or agricultural employment tends to be lower than within a longer one. We can see this difference as we examine recent estimates from two data sources, the Special Rural Households Survey 2000 and the 1999 LFS Survey. The first one has a time frame of one year for the employment questions, while the second one has a time frame of one week. Table 1.5: Employment by Sector in Four Rural Regions (one-year time frame) SECTOR Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie Agriculture 52.28 62.7 50.04 47.06 Manufacturing 9.74 7.73 13.29 8.12 Construction 8.69 4.81 5.63 7.76 Trade/Commerce 11.99 9.46 14.7 14.82 Transport, Storage, Communication 4.04 2.84 4.85 6 Other 13.26 12.46 11.49 16.24 Total 100.00 100.00 100.00 100.00 Source: Special Rural Households Survey 2000. 1.18 The share of agriculture sector employment that is estimated by the Special Survey12 varies between 47 percent and 63 percent, depending on the region. The Mazowieckie voivodship is much more specialized in agriculture than the other three. Indeed, Mazowieckie can be understood as the agricultural sector of the Warsaw region. In addition, these four estimates are larger than the corresponding figure estimated by the LFS for all rural areas combined. In fact, the February 1999 LFS survey indicates that the employment allocation of "During the same period (1989-95), the share of agriculture in national employment rose from 27 to 30 percent. 12 The Special survey sample includes 2,835 households in four regions. These households, in turn, include 10,971 individuals, of whom 8,548 are 15 years of age or older (the working age population). Chapter 1: Rural Labor Market 7 the rural labor force is as follows: 46 percent of employment is in agriculture; 17 percent in manufacturing; 9 percent in commerce; 5 percent in construction; 2 percent in mining; and the rest in services (see Table 1.6). Table 1.6: Employment by Sector and Region in February 1999 SECTOR Urban Rural Total Agriculture 2.15 45.93 21.26 Mining 2.5 1.5 2.06 Manufacturing 22.89 17.34 20.47 Utility 2.25 0.72 1.58 Construction 7.21 5.23 6.35 Commerce 17.76 8.72 13.81 Hotels 1.89 1 1.5 Transport 6.79 4.33 5.72 Finance 3.58 1.04 2.47 Estate 4.86 1.13 3.23 Public 6.1 2.95 4.72 Education 7.87 4.61 6.45 Health 9.29 3.78 6.88 Other services 4.85 1.72 3.48 Total 100.00 100.00 100.00 Source: Polish Central Statistical Office, Labor Force Survey, February 1999 1.19 The distribution of employment by sector as described in Table 1.6 is not unusual for the Polish level of per capita income (US$3,960). In fact, countries with similar income per capita for the same year, such as Chile (US$4,740); Brazil (US$4,420); and Malaysia (US$3,400) have 20 percent, 25 percent, and 30 percent of their respective labor forces in agriculture. There are exceptions, however. Venezuela, with an income per capita of US$3,670 in 1999, has only about 10 percent of its labor force in agriculture. By the same token, the share of employment in industry and services appears to be within the normal pattern. However, as Poland moves from that middle-income position toward upper-middle-income levels, its share of employment in services is expected to increase very rapidly. 1.20 An important observation that arises from the LFS data is that income diversification (a) takes place at the individual level to the extent that a worker holds a second job, and (b) is important at the household level to the extent that various members work in different activities. While the majority of rural workers have, or report having, only one job, as seen in Table 1.7, 10 percent of rural workers report having a second job. More precisely, about 16 percent of full- time and part-time employees (which represent about 50 percent of rural workers) report having a second job. It is interesting to note that the fraction of workers with a second job is the same among those that work full-time in their main occupation and those that work part-time, and that these two groups are significantly more likely to hold a second job than the self-employed. Chapter 1. Rural Labor Market 8 Table 1.7: Rural Workers by Main Job Status and Second Job Do you have a Main job type second job? Self-employed Full-time Part-time Unpaid famnily All employed employed worker Yes 4.61 15.80 16.21 1.73 10.13 No 95.39 84.20 83.79 98.27 89.27 Source: Polish Central Statistical Office, Labor Force Survey, February 1999. 1.21 The special rural survey allows us to examine the extent of diversification at the individual level, because each worker is asked if he/she holds a second job. We organized those answers and noted that the incidence of a second job is significantly higher in the Malopolskie region and much lower in Zachodniopomorskie. As indicated earlier, Malopolskie is a more densely populated rural area with close connections to Warsaw, and Zachodniopomorskie is sparsely populated and has a much lower degree of land ownership than the other regions. Table 1.8: Percent of Rural Workers with a Second Job by Main Activity in First Job Region Main Activity Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie Nonfarm business 0.349 0.057 0.187 0.067 Family Farrn 0.023 0.031 0.018 0.036 Private Employee 0.367 0.239 0.141 0.040 Public Employee 0.412 0.277 0.191 0.100 Day Laborer 0.364 0.071 0.400 0.071 Outwork 0.000 0.100 0.000 0.000 Other 0.667 0.250 0.000 0.143 All Workers 0.202 0.112 0.099 0.055 Source: Special Rural Household Survey 2000 1.22 As shown, the most typical status of a second job is "family-owned farming." If we examine in more detail how these second jobs are constituted, we find that the phenomenon has two distinct forms: (a) there are landowners who work as employees in their main job and have a second job as farmers in their second activity 90 percent of the time, and (b) there are landowners who are farmers in their main activity, and are nonfarm entrepreneurs or employees in their second activity. Chapter I. Rural Labor Market 9 Table 1.9: New Trend in National Employment, 1992-1998: The Expansion of the Private Nonagriculture Sector (percent of total employment) Quarter Total Urban fraction of Private fraction of Private-non agricultural Private agricultural total total of total holdings/rural 2 92 15,185 59.1 49.2 27.2 50.1 3 92 15,219 58.8 50.5 28.0 50.5 4 92 15,135 59.4 50.2 28.8 49.1 1 93 14,841 59.8 50.5 29.2 49.3 2 93 14,820 58.7 51.5 29.0 51.3 3 93 15,143 58.2 53.3 30.2 51.8 4 93 14,772 58.2 53.4 30.3 52.6 1 94 14,347 59.4 53.1 31.0 51.8 2 94 14,649 59.1 54.4 33.0 49.8 3 94 14,890 59.0 55.1 33.4 50.0 4 94 14,747 60.1 54.4 33.7 49.1 1 95 14,439 61.2 54.7 34.8 48.4 2 95 14,891 60.0 55.4 35.0 48.2 3 95 15,065 59.7 56.1 35.1 49.3 4 95 14,771 60.9 56.3 36.5 47.9 1 96 14,481 61.7 56.1 36.9 47.5 2 96 14,920 60.4 57.5 37.3 48.1 3 96 15,370 59.3 59.8 38.4 49.6 4 96 15,103 60.9 58.8 39.5 46.6 1 97 14,778 61.8 58.1 39.9 45.5 297 15,133 61.1 59.5 40.7 45.9 3 97 15,479 60.5 60.9 41.6 45.5 497 15,315 61.3 60.6 42.7 43.3 198 15,115 61.7 60.2 43.1 42.4 Source: Polish Central Statistical Office, Labor Force Survey, February 1998. 1.23 When we use the nationally representative LFS survey, we can establish the time trend for agricultural employment (using the narrow definition). Those data indicate that the fraction of private agricultural employment in rural employment has fallen steadily from 1992 to 1998 (see Table 1.9). Since public employment in agriculture has not been important and is likely to have fallen, we conclude that agriculture sector employment has fallen since 1992. 1.24 In short, the combined observations from the 1995 Census, the 2000 rural survey, and the 1992-98 LFS surveys allow us to state that a large fraction of rural households and a nontrivial number of rural workers are already engaged in activities outside agriculture. We conclude, therefore, that the challenge for rural areas is not necessarily to move workers out of agriculture. The challenge is to reduce poverty levels by moving workers from low levels of productivity to higher levels of productivity. In the next section, we examine some key institutional features of the Polish labor market that limit the incentives to reallocate work toward higher productivity activities. This is followed by an analysis of geographical labor mobility. Changes in the Institutional and Policy Framework 1.25 The very significant change in primary income sources for rural households described in the previous section has been accompanied by the creation of new institutions. The process of economic reforms seemed to promise a drastic reduction of the State's presence in day-to-day Chapter]: Rural Labor Market 10 decisions. However, there has been a tendency for the State presence to reappear in different forms. At times, it may seem as if the goal of providing a safety net has come into conflict with the goal of promoting a better allocation of resources and higher productivity. This section describes changes in incentives toward agricultural activities; budget transfers toward agriculture; access to old age and disability pensions, and unemployment benefits; and rules that apply to employment contracts, including payroll taxes. Incentives for Agricultural Activities 1.26 The main objective of the transition reforms in agriculture was to create a competitive environment. The key instruments of reform include the liberalization of markets and the privatization of state enterprises operating in farming, processing, and distribution. The collapse of Comecon (the Soviet-era Council for Mutual Economic Assistance), the reduction in import tariffs, and the elimination of food subsidies contributed to a significant decline in the demand for Polish agricultural production. The agricultural sector, mostly in private hands, was fully inserted into the chain of derived demand for Polish products, and was subject to the same type of demand shock that the state-owned enterprises had. Unlike the other sectors, agriculture did not react to the new market environment through real quantity adjustment. Instead, there were drastic adjustments of relative prices and income (FAO 1998). The sharp decline of relative agricultural prices in the early 1990s took place after a period of artificially high prices for agricultural products and high relative income levels for farmers, which were approximately 15 percent above the average income of employees outside agriculture. However, the tendency of falling agricultural prices was reversed in 1992, due to a Government policy implemented to avoid further deterioration of income levels. From 1992 onward, a relative decline of agricultural output has been accompanied by an increase in relative prices. 1.27 The initial policy agenda was soon supplemented by new objectives, such as farm income support and market stabilization. A number of new government agencies were established to support the implementation of policies more or less specifically targeted at the transition problems. The Agency for Agricultural Markets (AMA) was created in mid-1990, with the task of stabilizing farm markets and protecting farm incomes. The Agency's activities are mainly concentrated on grain, meat, and milk markets. It manages a system of minimum prices and buffer stocks and stimulates the domestic demand for agricultural products by a system of preferential credits and credit guarantees. The task of taking over and privatizing the assets of state-owned farms and the National Land Fund was delegated to the Agricultural Property Agency (APA), which was established in January 1992. The Agency for the Restructuring and Modernization of Agriculture (ARMA) was established at the end of 1993. The task of this Agency is to co-ordinate state financial support to the farm sector, mainly via subsidies on credit interest rates channeled through the rural banking system. 1.28 The restructuring and privatization of state farms has proceeded slowly. The share of the public sector in farmland ownership only declined marginally from 24.2 percent in 1990 to 22.6 percent in 1996. Despite slow privatization through sales, the reduction of labor in the formerly state-owned farms was as much as 40 percent between 1989 and 1992. (FAO, 1999). Chapter 1: Rural Labor Market 11 Budget Transfers to Agriculture 1.29 The reforms brought substantial cuts in total government spending in the agri-food sector, mainly ihrough a withdrawal of consumer food subsidies. Between 1993 and 1995, the average share of agri-food expenditure in the state budget amounted to 9.5 percent, significantly lower than the average of 20.9 percent between 1986 and 1989. However, as Table 1.10 shows, much of that spending has been redirected to the farm sector in the form of social spending. 1.30 The absolute and relative levels of subsidies have declined, and the specific forms of subsidies have changed. Direct subsidies to input prices (fertilizers, feed stuff, and pesticides) have been substituted by credit subsidies both for variable input purchases and investment. The effort to provide relief in the cost of credit was aimed at alleviating the drastic fall in the volume of investment experienced in the agriculture sector (recorded at about 70 percent in 1989-1992). Among general services, the focus has shifted from expenditure on research, advisory services, and training towards improvement of infrastructure. The dominant expenditure items in the agriculture sector budget in 1993-1995 were subsidies paid into the farmers' pension fund. This category exceeded 70 percent of all expenditure. Table 1.10: Total Budgetary Allocations to Agriculture', 1986-1995 As A Proportion Of All Public Expenditure On Agriculture (%) 1986-1989 1990-1992 1993-1995 Price and income support 6.6 7.1 4.0 Reduction of input costs 14.7 8.9 8.3 General services 10.8 16.1 11.7 Consumer subsidies 50.8 2.6 0.1 Social measures (pension fund) 12.8 60.9 71.6 Education, schools, art 4.3 4.0 4.1 Total agri-food expenditure 100.0 100.0 100.0 Share of agri-food expenditures2 20.9 9.1 9.5 Share of agri-food expenditures without pension fund 18.3 3.6 2.7 1) Unemployment payments to farm-family members are not considered. 2) In total budgetary expenditure. Source: FAO 1999 Report (citing OECD (1995), MAFE (1997), GUS data and author's calculations). Payroll Taxes and Access to Old-Age and Disability Pensions 1.31 Payroll taxes are applied differently in agriculture relative to the rest of the economy. There are two social security systems that operate in parallel, paying similar benefits and requiring significantly different contributions. The general program is ZUS and the special program for farmers is KRUS. 1.32 The Farmers' Security Fund (KRUS) was established in 1990 to provide pensions and insurance in case of accidents or sickness to workers in the rural sector. This program came as a modification to an earlier scheme that required farmers to relinquish their farms in order to receive a pension. These pensions, in turn, were financed from state funds. Under this earlier scheme, farming households could keep the land in their hands, but finance their elderly members throughout their retirement years, or relinquish the land in exchange for a retirement Chapter]: Rural Labor Market 12 benefit for the elderly members. Therefore, KRUS inherited participants that had joined in 1978, and who were eligible for pension benefits when they reached retirement age, as long as they own a minimum of 0.5 hectares. It also inherited participants from the period when the system was further modified in 1982, when benefit eligibility became subject to flat (quarterly) contributions. Not surprisingly, the number of contributors to the farmer's pension program has seen a remarkable swing. It surpassed 4.5 million contributors in 1988, but fell drastically to 2.3 million in 1990, after a 1989 amendment increased the minimum farm size to be eligible for the farmer's pension to 1 hectare. Currently, the number of contributors has stabilized at about 1.4 million. 1.33 Eligibility for an old-age pension is determined by age (60 for women, 65 for men) and payment of contributions for a minimum of 25 years. However, since the first year of actual contributions was 1983, all those farmers that have contributed since 1983 (approximately 15 years) and can prove to have been working on their farm another 10 years, qualify for benefits once they reach pension age. Farmers that have contributed for at least 30 years and stop working on the farm can become eligible for early retirement (5 years ahead of their pension age). A farmer that has been paying contributions for a minimum of 5 years and is unable to work on the farm can qualify for a disability pension. The pension cannot be lower than the minimum workers' pension (the same as in the employees' pension scheme). Currently, a farmer that becomes eligible for retirement benefits does not have to relinquish the land, and can continue working it. 1.34 The two parallel social security systems in Poland started with large unfunded liabilities. One should recall that, before the economic reforms, the Polish economy was transferring pension benefits to the elderly within a centralized budget allocation system. The pension reform in 1990 made the contributions partially tied to labor income, although it treated farrners and nonfarmers separately. As shown in Table 1.11, the current population above 65 years of age, which is only part of the population eligible for benefits13 is equivalent to 17.4 percent of the working-age population. Compared to the rest of the countries shown in the table, the Polish age distribution pyramid leads to a medium burden per potential worker. Table 1.11: The Macroeconomics of Social Security: Elderly Population (65+ years of age) as a Fraction of the Working-Age Population (15-64 years of age), selected countries Country Adult Population/Working Age Population Chile 11.1 Argentina 16.5 Poland 17.4 United States 18.1 Canada 18.6 Czech Republic 19.9 Germany 23.9 France 24.5 Sweden 26.9 Source: International Data Base, US Bureau of Census. 1.35 The evidence from the international experience indicates that the social security system's design is central for its long-term sustainability and its microeconomic impact-mainly its labor 13 Remember that women can retire at 60, and that there are disability benefits. Chapter]: Rural Labor Market 13 market incentives. In termns of labor market incentives, the Polish system has serious shortcomings. Farmers' households make contributions to KRUS at a rate equivalent to 4 payments per year of 30 percent of the minimum level of the regular old-age pension. As is shown in Table 1.13, this amounts to an annual contribution of PLN 715. Presumably, this is a way to secure pension benefits to a population of farmers that has low levels of cash incomes. Not surprisingly, this program has a large deficit. It is expected that, as the economy modernizes, most active workers will enroll in the regular pension system. 1.36 Indeed, as soon as a worker gets a formal (either farm or nonfarm) job, he/she has to start making contributions to the regular pension system (ZUS) through his/her employer. The problem with this policy is that the implicit tax on labor is very large. These contributions are proportional to individual wages as follows: Table 1.12: The Difference between Net and Gross Wages for ZUS contributors Taxable Wage 100.0 Zloty + Employers Contributions Old Age 9.76 Disability 6.20 Accidents 1.62 Labor Fund 2.45 Guaranteed Benefit Fund 0.0 (In the future it may be some 0.13 percent of wage, but currently the fund runs a surplus and employers are not paying) = gross wage 120.33 Taxable Wage 100 Zloty - Employee Contributions Old Age 9.76 Disability 6.50 Sickness 2.45 Net Wage 81.29 Source: Author's calculations. 1.37 The difference between the gross wage (PLN 120.33) and the net wage (81.29) is the payroll tax. The ZUS payroll tax corresponds to 48 percent of the net wage, or 32.4 percent of the gross wage. There is also a contribution towards health care, which is equivalent to 7.5 percent of the net wage, but individuals who pay the personal income tax have an opportunity to fully subtract this contribution from their income tax liability. 1.38 Table 1.13 summarizes the results reported by Chlon (2000) and provides estimates of contributions and benefits associated with KRUS and ZUS. Annual benefits for farmers and self-employed workers are relatively comparable across systems, although the annual contributions differ substantially. The numbers reported in Table 1.13 suggest that pensions are close to the minimum pension. For individuals that entered the system at a relatively old age, the system may seem very attractive, since they are likely to get more benefits than the present value of their contributions. However, key questions are: "How do young workers see this system?" and "How does it affect their choice of employment?" The two Polish programs pay comparable Chapter 1: Rural Labor Market 14 benefits for farmers and self-employed workers, while the contributions to the two programs are at a rate of 1 to 6 (see Table 1.13). As suggested, the KRUS program is the more attractive for those who own land because it is less expensive, creating a very good incentive to hold onto the qualifying requirement to join KRUS, namely-"to own at least 1 hectare of land." Table 1.13: Old-Age and Disability Contributions and Benefits, KRUS and ZUS 1998 Annual Averages. Annual amounts KRUS ZUS-self employed Old-age Benefits (PLN.) 464.43 552.97 Disability Benefits (PLN) 420.97 463.32 Average contributions (PLN) 715.47 4,421.56 Old Age Benefit/ Contribution (%) 0.650 0.125 Source: Chlon (2000) Tables 4 & 5 (citing KRUS, ZUS, author's calculations). 1.39 Abstracting from the impact of the personal income tax on wages, a formal sector job that pays PLN 100 and makes the contribution to the regular pension system (ZUS) is equivalent to an informal sector job (with no contribution to ZUS) paying PLN 67.6 PLUS the value individuals attach to payroll tax contributions to ZUS. To evaluate the impact of this tax on the labor market, we need an estimate of the value of this contribution for individual workers/employers. In general, the most important component of this "valuation" is the expected pension benefit associated to the contribution. The Microeconomics of Social Security Contributions 1.40 To answer the question posed above, we calculate the accumulated value of contributions from age 25 to 65, and the annuity that could be obtained from age 66 to 80, if the contributions were put aside at a market interest rate. In the calculation, we assume an interest rate of 3 percent, constant salaries, and a 10 percent cost of converting the accumulated fund into an annuity. We then compare that annuity with the old-age pension which the system pays today, and make a calculation of the implicit tax. We show that the benefits that an individual expects to obtain are a small fraction of the benefits that could be obtained from an actuarially fair investment of the funds. Estimates are presented in Table 1.14. Table 1.14: The Microeconomics of Social Security Contributions Contributions to Pensions Only KRUS ZUS (at minimum wage) (18% of the ZUS contribution) (20% of gross wage) Annual Contribution, from age 25 on PLN 358.00 PLN 1,560.00 Accumulated Contributions at age 65 PLN 28,141.00 PLN 122,715.00 Estimated Annuity, age 66 to 80 from an PLN 2,242.07 PLN 9,777.14 actuarially fair plan* Average Pension Benefit (Current) PLN 464.43 PLN 711.65 Estimated Tax * PLN 210.00 85.44%(over the payroll contribution) Source: Author's calculations. * An extreme scenario is assumed whereby the accumulated fund is used to finance two annuities (one for the worker and one for his/her spouse). That extreme scenario reduces the estimated tax. Chapter 1: Rural Labor Market 15 1.41 In the case of ZUS, the payroll contribution to pensions is approximately 20 percent of the gross wage. We use the calculation presented in Table 1.14 and compare the estimated pension benefit from the current system and the annuity that would be derived from an actuarially fair plan with the same contributions. The estimated annuity is divided by two, assuming that the fund has to finance two pensions, one for the contributor and one for the spouse. This leads us to compare PLN 711.65 with PLN 4,888.57. The first number is 14.56 percent of the second, suggesting that current pension benefits fall short of an actuarially fair contribution plan by a large margin. In other words, 85.44 percent of the contribution is "pure tax," and 14.56 percent is equivalent to a forced savings plan that pays an actuarially fair benefit. Thus, workers covered by ZUS are paying the equivalent of 17 percent (0.8544*20 percent) of their taxable wage as a "pure" tax when they make the contribution to pensions. Other payroll contributions may also amount to pure taxes, but we do not include them in these calculations. 1.42 The contribution to KRUS is not proportional to farmers' earnings. The estimated tax in this case is a fixed amount, which we calculated at PLN 210 per year. We can use a basic demand and supply model to show the impact of this tax on labor allocation between the farm and nonfarm sector. For convenience and simplification, we start with a fixed total labor supply that can seek jobs in the farm sector or in the nonfarm sector. Labor demand in the farm sector is shown as Ldf, and labor demand in the nonfarm sector is Ldnf. Starting from the demand for labor in each sector and assuming that, at the margin, workers are indifferent between the two, the equilibrium without taxes is represented by point A. Employment in nonfarm activities is Lnf, employment in farming is L-Lnf, and the wage is Wbt in both sectors. Figure 1.1: Labor Allocation and a Labor Tax on One Sector Ldn 17% tax Ld E3~~~~~~~~~-- ----- 0 Lnf' Lnf L nonfarm - v L farm = (L total - L nonfarm) 1.43 The graph captures the impact of a nonfarm employment tax on the allocation of labor between the farm and nonfarm sectors. We assume that the nonfarm sector is subject to a 17 percent tax (over the gross wage, for simplicity). This is particularly relevant, since the benefits that the two systems offer are similar. The introduction of a tax on nonfarm employment moves the equilibrium to B. The tax reduces nonfarm employment to Lnf, reduces labor market efficiency (since labor productivity at the margin is now higher in nonfarm activities), and results in lower net wages overall. The net wage in farm and nonfarm activities falls to Wn. The Chapter]: Rural Labor Market 16 economic impact of the tax would be reduced if there were tax evasion. A typical form of tax evasion is to declare the lowest possible salary for the calculation of contributions. 1.44 In any case, we argue that this model is representative of the current situation in Poland, because the current design of the Polish Social Security system requires a significant payroll contribution without clear links to individual benefits. The result is that the bulk of social security contributions are a tax on labor. This tax imposes a large distortion on the labor market. The economic impact of the tax is an increase in the cost of labor in the nonfarm sector, and a reduction of employment in all activities subject to this tax, relative to what would be observed in the absence of this tax. Because of this distortion, the sustainability of the system is questionable in the context of a free-market system. KRUS versus ZUS: Evidence on Registration from the Special Household Survey 1.45 The notion that the choice of social security system gets in the way of an efficient labor allocation is not new (Chlon, 1999), and it has also been suggested in this chapter. The data from the Special Household Survey gives an opportunity to examine what rural workers are actually doing. The survey asked two specific questions on this matter. Individuals (ages 15 and above) were asked (a) whether they were registered with KRUS, and whether they paid contributions regularly or not; and (b) whether they were registered with ZUS, and whether they paid contributions regularly or not. 1.46 To put things in context, we start with a brief summary of the sample characteristics by voivodship. Table 1.15 looks at the survey sample by household. As shown, land-ownership varies significantly by voivodship. In particular, 70 percent of households in the Zachodniopomorskie region are landless compared to an average 40 percent. In addition, there is a significant variation in household size (and number of workers) by land ownership. As seen in Table 1.16, households with larger land holdings also house a large number of workers. Thus, the distribution of workers by land ownership (Table 1.17) is different from the distribution of households by land ownership. To summarize, about 60 percent of households own land, while 77 percent of rural workers belong to land-owning households. Table 1.15: Land Ownership by Household Land Ownership Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie All Households % within voivodship no land 20.51 22.9 44.18 69.24 39.72 Less than 1 ha. 19.09 12.85 12.2 7.59 12.87 1 ha or more 60.4 64.25 43.62 23.17 47.41 100.00 100.00 100.00 100.00 100.00 All households 707 677 713 738 2,835 Source: Special Rural Household Survey 2000 Table 1.16: Average Number of Income Earners per Household, by Land Ownership Land Ownership Malopolskie Mazowieckie Wielkopolskie Zachodrniopomorskie no land 0.93 0.92 1.25 0.82 Less than I ha. 1.46 1.32 1.61 1.21 1 ha or more 2.35 2.32 2.39 2.12 Source: Special Rural Household Survey 2000. Chapter 1. Rural Labor Market 17 Table 1.17: Distribution of Rural Workers According to Household's Land Land Ownership Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie All Workers % within voivodship % % % % noland 10.11 11.20 30.88 49.41 23.08 Less than I ha. 14.76 9.07 10.95 8 10.99 I ha or more 75.13 79.73 58.17 42.59 65.93 100.00 100.00 100.00 100.00 100.00 ll workers 1,335 1,268 1,279 850 4,732 Source: Special Rural Household Survey 2000. 1.47 The results from the special survey show that, as expected, workers that are affiliated to KRUS belong to households that own at least Ihectare of land, a prerequisite for affiliation. What is more interesting is to determine what proportion of workers that own more than 1 hectare and that can register in KRUS, is affiliated with ZUS. This is presented in Table 1. 18. Table 1.18: Distribution of Rural Workers from Households with More than 1 Hectare of Land by Social Security System (percent within voivodship) Land Ownership Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie All Workers % within voivodship % % % % KRUS 31.11 46.79 54.3 58.01 44.84 ZUS 46.56 28.78 27.55 25.41 33.81 None 22.33 24.43 18.15 16.57 21.35 100.00 100.00 100.00 100.00 100.00 All workers 1335 1268 1279 850 4732 Source: Special Rural Household Survey 2000 1.48 The results shown in Table 1.18 indicate that close to one-third of workers that have the option to register in KRUS actually work as employees and contribute to ZUS. These workers must be better off working as employees, in spite of the large tax, suggesting that the net wage received as employees is at least equal to or higher than the equivalent earnings that they could obtain working in their own farm. The percentage of workers that make this choice is the highest in the Malopolskie region, characterized by a historical tradition of nonfarm-related jobs. This area of southeastern Poland belonged to Austria-Hungary during the nineteenth century. It is characterized by a great number of small farms, a relatively well-developed transportation infrastructure, and relatively dense village settlements. Combining work on the farm with a job outside agriculture has been popular in the region. Commuting to work and everyday pendulum migrations competed with permanent migration from villages to towns in the 1960s and 1970s. As a result, this region is the most densely populated rural area in Poland. The average farm area does not exceed 3-3.5 hectares in many gminas of the region (Frenkel and Rosner, 2000). 1.49 As indicated in Figure 1.1, the high tax associated with ZUS would imply that the marginal productivity of labor in the nonfarm sector would be higher than in the farm sector, and that net wages are the same in the two sectors. The differential in productivity accommodates the labor cost differential induced by the difference in payroll tax. Chapter]: Rural Labor Market 18 Table 1.19: Worker's Affiliation by Region (percent) Region KRUS ZUS None Total Malopolskie 24.7 53.8 21.5 100.00 Zachodniopomorskie 24.9 57.1 18.0 100.00 Wielkopolskie 32.1 54.2 13.7 100.00 Mazowieckie 38.5 38.3 23.2 100.00 Total 30.5 50.3 19.2 100.00 Source: Special Rural Household Survey 2000 1.50 To summarize, the affiliation to KRUS or ZUS is a function of employment patterns, which are influenced by land ownership and alternative wages. Table 1.19 shows the variation in affiliation patterns for all workers by region. On average, approximately 30 percent of rural workers are affiliated with KRUS, 50 percent with ZUS and 20 percent are not affiliated. These shares vary significantly by region. In particular, Malopolskie and Zachodniopomorskie have a smaller than average affiliation in KRUS. The reasons seem to be different. In Zachodniopomorskie (in the northwest), the degree of land ownership among workers is just 50 percent, compared to a 77 percent average for the four regions. Therefore, there is a smaller fraction of workers that is allowed in KRUS compared to other regions. On the other hand, Malopolskie (a southwest frontier region of high population density) shows the highest level of land ownership-almost 90 percent- but 35 percent of landowners are employees, compared to 22 percent for the four regions taken together. 1.51 Another comparison that is worth noting is that Malopolskie and Mazowieckie have large and similar shares of workers that own more than one hectare of land, but show significant differences in ZUS participation. Thus, Malopolskie appears as a region that offers more opportunities for employment in the nonfarm sector, even after controlling for land ownership. 1.52 On the other hand, the finding that Wielkopolskie has such a small fraction of workers nonaffiliated is explained by the age distribution of the population in that region. In fact, Wielkopolskie has the lowest fraction of 15-20-year old and 61-80-year old workers, the age groups where affiliation is typically the lowest. This suggests that the Wielkopolskie region may be a region of strong immigration. Unemployment Benefits 1.53 At present, individuals registered at a labor office are entitled to unemployment benefits if they can prove that they have worked for a minimum qualifying period. Beneficiaries have access to training, employment programs, start-up loans, and unemployment benefits. The entire program of services to the unemployed is the responsibility of the Ministry of Labor. 1.54 Up to 1996, unemployed workers had to have worked for at least six months within the previous twelve, but, subsequently, the qualifying minimum amount of work was raised to 365 days out of the previous 18 months. The unemployment benefit is about 50 percent of the net minimum wage per month, with a maximum of 12 months of coverage. The regions with higher unemployment rates and more serious problems of long-term unemployment are allowed to extend benefits. Chapter 1: Rural Labor Market 19 1.55 The longer period of previous employment required to qualify for benefits is an attempt to avoid abuse of the system. There was evidence of repeated drawing of benefits by unemployed jobseekers after short periods in seasonal jobs. Zarytcha (1998) suggests that, up to 1996, a special segment of the labor force had existed, consisting of persons who had combined short periods of formal employment with registration at employment offices and receipt of unemployment benefits while often continuing their work informally. 14 Minimum Wage 1.56 The minimum wage in February 1999 was PLN 650 (gross). This is the so called "Ubruttowiona placa", which means that it includes obligatory employees' contribution to social security (ZUS). This puts the net (take home) minimum wage at PLN 528. Hours and Overtime 1.57 The labor code establishes a maximum of 8 hours per day and 42 per week per every 3 months. Any hours above those limits are paid with premium, but the number of overtime hours is limited to 150 per year. As Bentolila (1997) notes, the legislation on hours is more restrictive than the one in the European Union. In contrast, in the U.S., there are no limits on overtime, and, in some states, such as California, the overtime premium applies only after topping 40 hours in a given week. Job Security Dismissals 1.58 Dismissal costs range between one and three months remuneration, plus 45 days notice. In the case of large establishments-with more than 1,000 employees-there are procedural steps that make dismissals very difficult. The employers whom we interviewed were very surprised by the question of the cost of dismissals, suggesting that, in the case of Poland, dismissals are just not part of the normal way of life. Workers are hired temporarily or are hired for life. 1.59 If potential employers believe that they are permanently obliged to workers at the time of hiring, then the probability of entry into employment-creating activities is much lower than in the case in which potential employers see a possible way out. The evidence from microeconomic data from several countries is very clear. Labor markets are characterized by a continuum of employment creation and employment destruction. This process is particularly important in economies that are going through profound transformations such as Poland. This means that, in order to encourage employment creation, it becomes imperative to replace the existing legislation with laws that recognize dismissals and job separations in general as a normal feature of the economic system. 1.60 In the case of Poland, there are legal restrictions on the renewal of temporary contracts; this means that activities that are subject to seasonal variations (mainly agriculture and services) cannot rely formally on repeated temporary contracts, or contingent employment. 14 Zarychta, H. Pasywna polityka rynku pracy w Polsce w latach 1990-1996 (Passive labour market policy in Poland in 1990-1996). The Institute of Labour and Social Studies, Warsaw 1998. Chapter 1: Rural Labor Market 20 Labor Intermediation 1.61 "Conducting a labor exchange or directing to work abroad for foreign employers for the purpose of attaining profit shall be prohibited." (Act on Employment and Counteracting Unemployment, 1995-art 37.4). 1.62 As stated in the law, labor intermediation is not allowed as a private, for-profit enterprise. Intermediation in other types of services, such as property rentals, is a growing activity in Poland, suggesting that the development of markets often needs the inputs of specialists who can become channels of valuable information. The suppression of private labor intermediation is detrimental to the operation of the labor market, because private labor intermediaries facilitate information flows. Labor contractors have proven to be essential actors in the process of transforming agriculture and increasing the volume of production, particularly in fruits and vegetables, since they depend on substantial seasonal labor. Intermediaries are also very important in an environment of increased labor rotation, if individuals are not familiar with the process of applying for jobs, and employers are not familiar with a process of selecting workers from a more ample pool of applicants. 1.63 Recent data for the United States indicate that close to 10 percent of total employment is characterized by nontraditional forms such as: temporary help agencies, labor contractors, contract workers, and day laborers. In addition, close to 5 percent of employment is "contingent," meaning that workers are aware that employment is unlikely to last beyond a year (see Hipple 1998, and Coheny 1996). These jobs are sometimes preferred by workers, and, even when they are not, the alternative of "no job" is likely to be inferior for workers and employers alike. 1.64 This section describes an institutional setting that provides a modest safety net to a large number of individuals and that finances transfers by taxes on the labor of a relatively small number of individuals. The current scheme is not sustainable and cannot be fixed at the margin. This fundamental problem for the operation of the labor market originated in attempts to balance collections and payments in a social security program that started with large unfunded liabilities. One alternative approach that has been practiced in the context of social security reforms is that of financing the unfunded liability (at least in part) with the proceeds of a privatization program, securing a real reduction in labor taxes. This alternative approach would encourage employment. Furthermore, if accompanied by other fundamental reforms oriented to promoting economic growth, such reform to the social security system will ensure its viability in the medium to long run, even if there is a large public sector deficit originating in the initial unfunded liability. Employment and Wages 1.65 We turn now to a description of labor force allocation. The Polish labor force is made up of about 15 million workers, of which nearly 40percent reside in rural areas. These workers belong to the various types of rural households, some with more active workers than others. The transition has so far resulted in a change in the distribution of jobs. Data compiled in a 1998 GUS publication describe a significant expansion of private, nonagricultural employment, which represented 27 percent of total employment in May of 1992 and reached 43 percent of total employment in February 1999 (Table 1.20). Chapter]: Rural Labor Market 21 1.66 Table 1.6 summarized the allocation of employment by sector. As shown, about 45 percent of rural workers are absorbed by the agricultural sector. Employment status varies significantly across sectors in rural areas. In particular, most agriculture employment is self- employment or unpaid family work, while most employment in other sectors is full-time (see Table 1.19a). Table 1.20: What Is Your Employment Status? (by occupational sector) Sector Self full time part time Unpaid Total Agriculture 71.4 7.02 0.83 20.75 100.00 Mining 0 99.36 0.64 0 100.00 Manufacturing 6.13 88.58 4.64 0.66 100.00 Utility 0 97.33 2.67 0 100.00 Construction 13.16 82.82 3.84 0.18 100.00 Commerce 26.45 64.76 5.93 2.85 100.00 Hotels 19.23 72.12 7.69 0.96 100.00 Transportation 14.13 82.56 2.87 0.44 100.00 Finance 5.5 88.99 5.5 0 100.00 Real Estate 10.17 77.97 10.17 1.69 100.00 Public 0 95.13 4.87 0 100.00 Education 0.21 92.12 7.68 0 100.00 Health 2.53 93.42 4.05 0 100.00 Other Services 17.22 72.78 10 0 100.00 Total 38.23 48.69 3.13 9.95 100.00 Source: Polish Central Statistical Office, Labor Force Survey, February 1999. 1.67 The February 1999 survey estimates that, in the 15 years and over age range of the rural population, 53.4 percent of men and 39 percent of women work. However, as seen in Table 1.21, many of these workers rely principally on other sources of income. According to the way in which individuals answer the question, "What is your principal source of income?" we learn that 15 percent of rural "workers" consider their pensions to be their most important source of income. When we separate workers by employment sector, we see that 20 percent of agriculture sector workers consider pensions to be their main source of income, while 10 percent of workers in other sectors see pensions as their main income source. Table 1.21: Rural Workers: By Sector and Principal Source of Income Sector Income Source Employee Farmer Self-employed Pension Other Total Agriculture 18.60 59.33 1.23 20.46 0.38 100.00 Other 75.53 5.98 7.56 10.39 0.54 100.00 Total 49.38 30.49 4.65 15.01 0.47 100.00 Source: Polish Central Statistical Office, Labor Force Survey, February 1999. Chapter 1: Rural Labor Market 22 Table 1.22: Agriculture Sector Workers Are Older, Particularly Those Who Consider Pensions to Be Their Main Income Source Worker's Mean Age by Sector Main Income Source Sector Employee Fanner Self-employed Pension Urban Other All Agriculture 43.7 42.1 40.7 52.9 38.7 45.2 44.7 Others 36.9 29.6 37.9 33.7 28.6 30.3 36.1 Source: Polish Central Statistical Office, Labor Force Survey, February 1999. 1.68 Income from social transfers is compatible with work. As shown below, 14 percent of old-age pensioners, 19 percent of disability pensioners, and 6 percent of unemployment benefit recipients report having worked during the week of the survey. However, when beneficiaries of state transfers report working, they are in the self-employed, part-time, or unpaid worker category. Since the transfers listed above are incompatible with formal work, it follows that the full-time work category corresponds roughly to the "formal" sector, and the rest to the "informal" sector. Table 1.23: Answer to the question: Did You Work during the Reference Week? (by main income source) Principal income source No Yes Total Nonagriculture Work 6.40 93.60 100.00 Work in agriculture 3.85 96.15 100.00 Old-age Pension 86.37 13.63 100.00 Disability Pension 80.73 19.27 100.00 Unemployment benefit 93.73 6.27 100.00 Other nonwork income 94.81 5.19 100.00 Supported by person working in nonagriculture 98.13 1.87 100.00 Supported by person working in agriculture 93.77 6.23 100.00 Supported by person having a nonwork income 96.57 3.43 100.00 Total 53.89 46.11 100.00 Source: Polish Central Statistical Office, Labor Force Survey, February 1999. 1.69 Using this definition of "informal" sector, most employment in the agriculture sector is "informal," while most nonagriculture employment is formal. Commerce, hotels, transportation, and construction (nonfarm activities) have the highest degree of informality after agriculture. A 1998 GUS survey examined the extent of unregistered employment and found that, in January- August 1998, the extent of unregistered employment was equivalent to 9 percent of those formally employed. People who worked with no registration stated that they had opted for that type of work in order to (a) increase their incomes, (b) compensate for the absence of formal jobs, and (c) avoid social security contributions. In addition, 10 percent of the respondents said that they decided to work unregistered because they did not want to lose their rights to other sources of income, i.e., pensions and social benefits. The GUS study also confirms that, in rural areas, unregistered employment is concentrated in construction, transportation services, and help in agriculture and gardening. Reported Earnings 1.70 Based on the LFS data from February 1999, we use about 5,000 rural observations and m 10,000 urban observations to examine the distribution of wages. Only full-time workers report Chapter 1: Rural Labor Market 23 eamings. Therefore, the sample of rural workers that report wage earnings is not representative of all rural workers. We then compared the distribution of earnings in rural and urban areas, and in Warsaw. The results showed quite clearly that the wage distribution has a discontinuity around the minimum wage. This discontinuity is more pronounced in rural than in urban areas, and significantly less marked in Warsaw. This finding suggests that the minimum wage is more of a binding constraint for employment creation in rural areas than in urban areas, particularly in Warsaw. 1.71 To examine the distribution of reported wages, Table 1.24 shows the entire distribution of rural wages by schooling level. The distributions are very similar for all schooling levels, except higher education. The variance is about the same for all levels of schooling, except for higher education, where the variance is higher. In general, wages are in a ratio of 2 to 1 when we compare the ninth decile with the first. However, for high levels of schooling, that ratio is almost 3 to 1. Table 1.24: The Distribution of Rural Wages by Schooling (deciles) Schooling 1st dec. 2nd dec 3rd dec. 4th dec. Median 6th dec. 7th dec. 8th dec. 9th dec. Level Higher 615 700 750 800 850 950 1000 1200 1500 Post Sec 540 590 600 650 700 700 800 860 1100 Sec Voc 500 550 600 650 700 785 850 1000 1200 Sec Genl 500 550 600 650 700 750 800 980 1100 Basic Voc 500 530 600 600 653 700 800 900 1000 Primary 460 500 550 600 600 660 700 800 900 None 480 480 500 500 650 800 800 900 900 Source: Polish Central Statistical Office, Labor Force Survey, February 1999. 1.72 The special Rural Survey 2000 allows another look at wages. We have four distinct regions examined, and we have answers relating to "annual income from labor" and "daily wages paid in farm activities." To estimate mean and median "annual incomes from labor activity," the sample was also restricted to nonzero values. We report median wages in PLN below. Table 1.25: Median Annual Income from Work (Rural Areas): By Schooling Voivodship School Achievement Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie Completed higher education 18,000 22,800 27,037 23,800 post secondary, uncompleted higher 17,000 16,060 19,000 18,000 Secondary, technical/vocational 14,100 16,500 18,580 16,797 Secondary, general 14,000 20,500 14,000 13,500 Completed primary 13,000 12,200 11,980 12,000 Incomplete primary 11,500 12,600 12,000 10,000 No education 8,900 7,700 4,500 6,000 Source: Special Rural Household Survey 2000 Chapter ]. Rural Labor Market 24 Table 1.26: Median Annual Income from Work (Rural Areas): By Occupation Region Main Activity Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie Nonfarn business 14,050 11,000 11,420 12,000 Family Farm 10,800 10,000 9,800 9,000 Private Employee 17,000 18,000 15,480 13,200 Public Employee 15,000 18,900 18,000 16,000 Day Laborer 7,000 8,000 7,800 11,520 Outwork 10,600 10,000 6,750 6,000 Other 18,800 6,560 12,000 14,600 Source: Special Rural Household Survey 2000 Table 1.27: Median Annual Income from Work (Rural Areas): By Sector Region Main Activity Sector Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie Agriculture 10,800 10,200 10,956 9,868 Manufacturing 15,600 17,940 14,400 13,200 Construction 15,700 19,000 12,000 14,000 Trade, Commerce 17,287 16,000 17,450 14,920 Transport, Storage, Comm. 18,000 17,750 15,049 12,500 Other 14,400 19,100 18,900 14,400 Source: Special Rural Household Survey 2000 Table 1.28: Daily Farm Wages in Rural Areas Region median Mean Malopolskie 50 46 Mazowieckie 40 40 Wielkopolskie 35 35 Zachodniopomorskie 30 31 Source: Special Rural Household Survey 2000 1.73 The reader should note that daily wages in rural areas appear directly related to the labor market density of the region in question, with a positive correlation between density and daily wages. It is also of interest to note that, according to Table 1.26, day laborers have the highest annual incomes in the region where daily farm wages are the lowest, and vice versa. This is, most likely, not an artifact of data errors, but a true feature of labor markets. Day labor activities are often carried out by temporary migrants and by local workers that participate in the labor force for a short period of time. It is possible that day laborers from regions with higher annual incomes work part of the year outside their local (low daily wage) area. Place of Residence and Place of Work: The Question of the Rural-Urban Divide 1.74 In the 1950s, 1960s and 1970s, there was a steady rural-urban migration process, with some important regional patterns. There are many reasons to expect changes in migration patterns as a result of economic reforms since 1989. Unfortunately, the main source of migration counts is official data on registered change of residence. These counts have always been biased Chapter1: Rural Labor Market 25 to the extent that not everyone registers. There are good reasons to believe that, after the reforms, the size of the bias has increased, resulting in a larger undercount of internal migration. 1.75 For example, before 1989, state companies were obliged to provide accommodations for employees. Each new staff member had to register as a new resident in the area and also advise the old office about his/her move. Thus, the state had more control over people's movements, and that was captured by statistics. Since 1989, companies are no longer obliged to provide housing, so the incentive to register or change registration is lower. People taking up employment in new places quite often rent a room or flat, and do not register in local offices. However, if they decide to stay in the new place for a long time, they have to register to be able to qualify for health care and education for their children. 1.76 Frenkel and Rosner (2000) offer a thorough review of recent trends in population growth and internal migration in Poland. They make three important observations: * Rural fertility rates have recently converged with urban rates at a very fast pace. This observation is notable because differences in fertility rates are a good proxy for differences in incentives between urban and rural areas. Similar fertility rates are an indication of a tighter integration of the rural and urban economic systems. * Net rural out-migration has shrunk considerably since the beginning of transition- from 144,000 in 1988 to only 9,000 in 1998, the lowest level in the whole post-war period (their Table 1.3). The shrinkage resulted mainly from the decline of the rural-to-urban stream (which dropped through the decade by more than 50 percent), and much less from the increase of the stream in the opposite direction (although the role of the latter tended to grow in the second half of 1990s). The decline in the rural-to-urban stream was associated mainly with various negative transition-related factors, particularly factors such as (a) the generally difficult situation in the labor market, (b) the weaker position of rural people in this market (to a great extent related to lower education and qualifications inadequate for urban demand), and (c) highly increased costs for housing and education in cities. However, not only the negative, but also some positive transition-related factors have probably played a role, particularly factors such as the accelerated development of small businesses and infrastructure in rural areas. * A permanent feature of sex selectivity in rural-to-urban migration is that its rate is usually higher among women than among men. The difference is highest in the most mobile age groups. The excess of women fluctuates over time; in 1998, it was much higher than in 1988. 1.77 In short, while official statistics report a significant reduction in migration (change of residence) flows during the 1990s relative to earlier periods, the evidence on fertility convergence and on increased female versus male count in urban areas are consistent with faster migration moves. Chapter ] Rural Labor Market 26 1.78 The available data, even if biased, can still shed light on the link between migration and economic conditions. Kupieszewski et al. (1996) examine population movements within rural and urban areas. They agree with others in that rural-to-urban migration has fallen substantially. However, they also note that this reduction has not reversed the depopulation of some rural areas, as "the number of units where both net migration and population growth are negative, has increased dramatically over the last decade." (p. 51). They observe a significant change in the role of cities and towns. While cities were major recipients of migrants in the 1970s and 1980s, they have tended to lose population in the mid-1990s. Apparently, cities are losing population to neighboring communities. They argue that a suburbanization process has been put into motion. 1.79 Several authors have tried to measure the relationship between migration and unemployment (see, for example, Bentolila 1997; Chlon 1998; Kupieszewski et al 1996). Using a cross section for 1993-94, Bentolila (1997) finds that people move to regions where the unemployment rate is lower, wages are higher, and there is less competition for jobs than elsewhere. However, he finds that the migration responses are low, and believes that the explanation lies in a poorly working housing market. The response of migration to labor market conditions can also be captured with micro data distinguishing migrants from nonmigrants. Chlon had the benefit of such data, and was able to observe the previous employment status of migrants (from a 1994 sample). She reports (in her Table 1.10) that migrants have a much larger probability of having been unemployed than nonmigrants. Furthermore, based on aggregate counts, Kupieszewski et al (1996) find an interesting pattern of migration. They report that, between 1984 and 1994, the communities and municipalities with the lowest unemployment rates (below 8 percent) attracted migration. All other municipalities had out-migration, and the rate of out-migration was higher for communities with higher unemployment rates. They conclude that there is a link between unemployment and migration, but that the link is weak and only extreme levels of unemployment generate significant responses in terms of migration. Their analysis goes further to show that most migration flows (80 percent) occur from communities with higher unemployment to communities with lower unemployment. 1.80 Commuting between rural and urban areas on a daily or weekly basis is another dimension of the migration question. "An important feature of employment in the rural population is that only a relatively small part of those working outside agriculture are employed in the same commune where they live. According to the 1988 census, about 65% of rural residents working outside agriculture commuted to work, mainly to towns and often for a long distance, because of lack of nonagricultural jobs on the spot. Between the 1988 and 1995 censuses the absolute number of commuters fell by 1/3 and their share to 55% (out of the total nonagricultural employment of the rural population). The latter figure shows that there was a relative increase of people working on the spot that may be explained by a fast development of small businesses in rural areas. It was, however, not big enough to compensate for jobs lost by those who commuted to towns. Since 1995, the absolute number and the share of commuters have probably increased again although there are no statistics to confirm this" (Frenkel and Rosner, 2000). 1.81 The Special Rural Household Survey 2000 collected information on geographical mobility through four types of questions. First, rural workers were asked if their jobs were in urban or rural areas; second, they were asked to estimate the distance (in kilometers) between Chapter 1: Rural Labor Market 27 their dwelling and place of work; third, they were asked if and how long they had to stay overnight in another location for work reasons; and fourth, they were asked about their location in 1999 and in 1990, allowing researchers to calculate the percentage of workers that had moved in the last year, and the percentage that had moved in the last decade. Table 1.29: Percentage of Rural Workers That Work in Urban Areas by Main Activity and Region (%) Region Main Activity Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie Nonfarm Business 24.8 10.0 12.1 16.0 Famnily Farm 0.2 0.3 0.7 2.2 Private Employee 63.7 58.2 50.3 54.9 Public Employee 66.0 53.3 50.7 47.8 Day Laborer 18.2 7.1 0.0 14.3 Outwork 60.0 40.0 22.2 30.8 Other 33.3 25.0 11.1 14.3 All Workers 28.8 20.5 25.6 31.1 Source: Special Rural Household Survey 2000. 1.82 The data tell us that between 20 percent and 31 percent of all rural workers work in urban areas. The commuting phenomenon is particularly prevalent among employees. If we exarnine the answers to the question "How far do you travel to work?" we find answers that are consistent with the first table, in the sense that the higher the incidence of work as an employee, the more travel is associated with the work, except that not all traveling takes rural workers to urban areas. We separated the answers into two categories, those that travel a nonzero distance (we called them commuters) and those that do not travel. The largest fraction of commuters is found in Zachodniopomorskie, followed closely by Malopolskie. The median travel distance is the same in these two regions, 5 kilometers, and double in the Mazowieckie region, which has the lowest percentage of commuters. Table 1.30: Commuting and Distance Traveled by Region: Main Job Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie Commuters % of Workers 63.82 41.48 55.20 71.06 Travel Distance Mean 14.5 18.5 11.8 13.0 Median 5.0 10.0 6.0 5.0 Nonconmnuters % of Workers 36.18 58.52 44.80 28.94 Travel Distance 0 0 0 0 Source: Special Rural Household Survey 2000. 1.83 The incidence of overnight stay for work purposes is very small (under 2 percent) and more prevalent in the Mazowieckie region. We know from Table 8 of the Survey (page 13) that, depending on the region, between 6 percent and 20 percent of rural workers hold second jobs in the four regions covered by the special survey. Table 1.31 takes workers with second jobs in each region and tabulates the proportion of second jobs that are urban and both the incidence of commuting and travel distance for workers that hold second jobs. The survey results show that Chapter 1: Rural Labor Market 28 second jobs tend to be more local than main jobs, indicating that rural workers generally travel to make a living, but use local activities to supplement their income. Table 1.31: Commuting and Distance Traveled by Region: Second Job Malopolskiel Mazowieckiel Wielkopolskiel Zachodniopomorskie Commuters % of Workers 63.82 41.48 55.20 71.06 Travel Distance Mean 14.5 18.5 11.8 13.0 Median 5.0 10.0 6.0 5.0 Nonconimuters % of Workers 36.18 58.52 44.80 28.94 Travel Distance 0 0 0 0 Source: Source: Special Rural Household Survey 2000. 1.84 We finally examine the survey results regarding migration to rural areas. Note that questions of previous location were only asked of workers living in rural areas, and there was no attempt to measure the extent of migration outside of rural areas. In addition, the answers pertain to change in location in terms of villages/voivodships and not necessarily from urban to rural areas. Table 1.32 tabulates results for the question pertaining to location in the previous year. The numbers indicate the fraction of workers in a given category who lived in a different village in 1999 compared to the year 2000. The fraction of recent immigrants, as a proportion of all workers, does not go over 2 percent. However, this fraction is significantly higher (up to 6.6 percent) among private employees in Malopolskie, and among day laborers is Mazowieckie. Table 1.32: Recent Immigration to the Region: Fraction of Workers That Moved to Their Current Location during the Last Year (by main activity - %) Region Main Activity Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie Nonfarm business 2.8 1.4 0.0 0.0 Family Farm 0.0 0.7 0.2 0.4 Private Employee 6.6 4.8 2.6 3.6 Public Employee 1.2 1.1 1.3 1.7 Day Laborer 0.0 7.1 0.0 0.0 Outwork 0.0 10.0 0.0 0.0 Other 0.0 0.0 0.0 14.3 All Workers 1.9 1.7 1.1 1.9 Nonworkers 2.9 1.8 1.1 2.3 Source: Special Rural Household Survey 2000. 1.85 Table 1.33 tabulates results for the question pertaining to location ten years before. The numbers indicate the fraction of workers in a given category who lived in a different village in 1990, compared to the year 2000. The fraction of accumulated immigrants, as a proportion of all workers, goes from 6 percent to 14 percent, depending on the region. In particular, Zachodniopomorskie appears to be a region of larger immigration flows, followed by Wielkopolskie. The evidence regarding relatively large immigration flows into Wielkopolskie was also apparent earlier, when data on the age distribution of the work force was examined. Chapter1: Rural Labor Market 29 The higher incidence of migration towards Zachodniopomorskie suggests that labor movements across regions are positively related to the size of the private "employees" sector. Table 1.33: Accumulated Immigration to the Region: Fraction of Workers That Moved to Their Current Location during the Last Decade (by current main activity - %) Region Main Activity Malopoiskie Mazowieckie Wielkopolskie Zachodniopomorskie Nonfarm business 5.5 10.0 9.9 10.7 Family Farm 2.1 3.7 4.5 8.7 Private Employee 10A 12.4 14.9 20.2 Public Employee 12.0 10.9 10.2 14.4 Day Laborer 0.0 7.1 0.0 7.1 Outwork 0.0 0.0 11.1 15.4 Other 16.7 0.0 0.0 14.3 All Workers 6.1 6.8 9.1 14.0 Nonworkers 4.5 3.9 5.0 10.0 Source: Special Rural Household Survey 2000 1.86 To sum up, although it is difficult to establish the change in migration flows, there is evidence that indicates that economic conditions, particularly in the labor market, have a predictable influence on migration. Nevertheless, the facts that housing costs are now explicit and that there are restrictions on the development of the housing market, reduce the intensity of labor market signals on migration incentives. Time Allocation in Rural Areas 1.87 Broad labor market indicators are very similar in rural and urban areas. For example, based on the February 1999 Labor Force Survey, which has national representation, labor force participation is 55 percent in rural areas and 54 percent in urban areas, and the unemployment rate is 12 percent in rural areas and 13 percent in urban areas. However, the age, sex, and schooling composition of the rural and urban populations are significantly different. In particular, rural areas have a larger proportion of elderly, lower levels of schooling, and a higher male-female ratio than urban areas. 1.88 In rural areas, individuals tend to leave school and enter the labor force at a younger age, resulting in lower levels of schooling and higher participation rates. The February 1999 micro data file available from the Labor Force Survey allows us to examine in detail the proportion of individuals that (a) attend school or (b) work in rural and urban areas in various age categories. Figure 1.2 shows the levels of school attainment by age. Those still in school are shown in gray. The remainder (in colors) corresponds to the proportion of individuals outside of the school system, grouped according to their completed schooling. While this picture is the result of a cross section of individuals of various ages, it is a good proxy for the pattern of out-of-school transitions for the current 15-21 year old cohort. We learn that by age 21, about 85 percent of the rural population has dropped out of school, limiting the proportion of post-secondary school graduates to less than 15 percent per cohort. About 20 percent of the cohorts achieve primary schooling, 40 percent basic vocational schooling, less than 5 percent secondary grammar school, and about 20 percent secondary-vocational. Chapter 1: Rural Labor Market 30 Figure 1.2 School Attainment by Age: Rural Areas 100% E bl!X, higher 80% lO E | iELIE 4 a in school 60% O2 | | | l~ E - post-sec - U~~ sec-voc NI sec-gram 0 basic-voc 14 15 16 17 18 19 20 21 Oprimary * none 1.89 In urban areas, on the other hand (see Figure 1.3), individuals stay in the school system about three additional years, since it is only by age 24 that 85 percent of the cohorts have dropped out of school. Furthermore, by age 24, it is apparent that nearly 30 percent of the cohorts will complete post-secondary and higher education. About 5 percent of the cohorts abandon the school system with primary schooling; 25 percent with basic vocational; 10 percent with secondary grammar; and 30 percent with secondary-vocational. Figure 1.3 School Attainment by Age in Urban Areas 100% s 11 in school 80% El higher - E post-sec 40% 0 U2 sec-gram 20% - | basic-voc 0% __ primary 14 15 16 17 18 19 20 21 22 23 24 | none 1.90 As young individuals leave the school system, a significant fraction enters the labor market. It would appear from the data that some rural-to-urban migration occurs among school- aged youth who continue higher education in urban areas, and stay working in urban areas after graduation. The observed difference in education levels between urban and rural areas is, therefore, the result of the combination of (a) a lower level of schooling among those that stay in rural areas and (b) a migration out of rural areas by individuals with higher levels of schooling. Conclusion 1.91 One of the main features of the Polish economic transformation has been the high concentration of the labor force in rural areas (20 percent) relative to agriculture's share of GDP (5 percent), and the reduction of rural employment opportunities outside the farming sector. This Chapter]: Rural Labor Market 31 has led to high (open or hidden) unemployment in rural areas, requiring the government to support the income of rural households by providing social transfers through KRUS, and by intervening in agricultural output markets. This situation has been made even worse over the past three years because of depressed producer prices, lower demand from Russia, and increased competition from subsidized EU imports. 1.92 The government has focused on price interventions and social transfers, and has paid less attention to the role of markets in facilitating the reallocation of work, land, and capital to more productive activities. The available evidence suggests that rural workers have moved away from farming, and that the labor market as a whole has moved towards nonagricultural private activities. However, the process of resource reallocation is saddled with distortions that slow it down. In particular: * there are strong incentives to hold on to small units of land in order to retain eligibility for fanner's pensions; * there are heavy taxes associated with crossing the frontier between farming and working in other sectors; and * there is a regulatory framework that severely limits the options that potential employers have regarding, wages, hours, and dismissals. 1.93 The government's efforts should focus on employers' incentives for the creation of jobs, and workers' incentives for working in a nonfarm activity. This shift in focus is required both to increase the efficiency of labor markets and to better target the budgetary funds available for the modernization of agriculture. One of the main challenges in this area is how to finance the unfunded liabilities of the pension programs. In spite of the high payroll tax, these pension programs are currently absorbing a large part of the public sector budget. Given that a high payroll tax is a serious distortion on labor market intermediation, an obvious question arises. Is it possible to have a reduction in the payroll tax and an increase in tax revenues at the same time? The answer is "yes." However, this requires a fundamental change in the design of the social security system. The reformed system must ensure that the savings of current contributors will pay off in the form of future pensions, virtually eliminating the payroll tax. 1.94 To sum up, the challenge of modernization calls for a deeper integration of rural and urban factor markets. In the particular case of the labor market, a more efficient allocation of labor needs (a) a reduction of the labor tax associated with ZUS contributions, which can be accomplished by stronger links between ZUS contributions and expected benefits; (b) a more flexible approach to the use of short-term contracts; and (c) an approach to public sector investments in infrastructure and social services that contributes to and takes advantage of the same factor markets integration. CHAPTER 2: RURAL LAND MARKET Introduction 2.1 Poland is among the largest agrarian economies of Central and Eastern Europe (CEE). Its agricultural land resources of 18.5 million hectares account for 60 percent of the country's total land area, and its rural population of 15 million is nearly 40 percent of the total population. According to official statistics based on labor surveys, Polish agriculture employs 4.4 million people, or 27 percent of the total number of those employed in the country. Yet, agriculture accounted for only 6 percent of GDP in 1997, which points to substantially lower sectoral productivity than in the rest of the economy. 15 2.2 The agricultural land resources in Poland show a secular declining trend due to zoning changes associated with growing municipal needs, construction of roads and railways, and also, to a certain extent, afforestation. The total agricultural land endowment decreased from 20.4 million hectares in 1950 to 18.5 million hectares in 1997, a decrease of 10 percent over five decades. Agricultural employment, on the other hand, has been increasing since 1990 at an annual average rate of nearly 2 percent. Land Ownership and Use 2.3 Poland was unique among the CEE countries in that its agriculture was never fully collectivized after World War II. During the 1980s, only 24 percent of land was cultivated by socialized farmns (20 percent by state farns and 4 percent by cooperatives or collectives), compared with about 98 percent throughout the rest of CEE. Thus, Poland entered the transition era with 76 percent of its agricultural land actually cultivated by family units (Table 2.1), and issues of privatization and restitution of land played a much less prominent role than in other CEE countries during the 1990s. Privatization efforts have primarily focused on the relatively modest land resources of state farms (about 4.5 million hectares). These were transferred in 1990 to the management of the Agricultural Property Agency (APA), a government organization (part of the State Treasury) that was charged with the task of reallocating state land to nonstate users by selling or leasing out its reserves. Table 2.1: Structure of Land Use and Ownership, 1990-1997 (as a percent of agricultural land) Owned land Used land 1990 1997 1990 1997 Private sector 76 78 80 92 Individual farms 72 76 76 83 Cooperatives 4 2 4 2 Other private -- -- 0 7 State sector 24 22 20 8 Total 100 100 100 18457 Source: 1998 Agricultural Statistical Yearbook. 15 This chapter was prepared by Csaba Csaki and Zvi Lerman, with research assistance from Pepijn Schreinesmachers. Chapter 2: Rural Land Market 34 2.4 APA's efforts have focused primarily on the leasing of state land, and its privatization activities have not been particularly vigorous. As a result, the changes in land ownership between 1990 and 1997 were insignificant (Table 2.2): the state privatized about 10 percent of its land reserves (449,000hectares out of 4.5 million), which declined from 24 percent of all agricultural land to 22 percent. In parallel, cooperatives lost ownership of about one-half of their land endowment, which went down from 736,000 hectares in 1990 to 394,000 hectares in 1997, primarily due to liquidation of cooperative and collective farms due to economic considerations. The privatized and de-collectivized land shifted to the individual sector, where land ownership went up from 13.5 million hectares in 1991 to 14.1 million hectares in 1997 (increasing from 72 percent of agricultural land in 1990 to 76 percent in 1997). Privatization of state land contributed 2 percentage points to the increase in the pool of individually owned land, and liquidation of cooperatives and collectives contributed another 2 percentage points. Table 2.2: Changes in Land Use and Ownership, 1990-1997 (in thousand hectares) Owned land Used land Intrasectoral transfers from owners to users 1990 1997 Change 1990 1997 Change 1990 1997 Change Private sector 14233 14506 273 14978 16982 2004 745 2476 1731 Individual farms 13497 14112 615 14228 15293 1065 731 1181 450 Cooperatives 736 394 -342 750 464 -286 14 70 56 Other private -- -- -- 0 1225 1225 0 1225 1225 State sector 4551 4102 -449 3742 1475 -2267 809 2627 -1818 Total 18784 18608 -176 18720 18457 -263 64 151 87 Source: 1998 Agricultural Statistical Yearbook. 2.5 The changes in land use between 1990 and 1997 were Figure 2.1: Users of Agricultural Land, 1990-1997 much more significant. The state million ha sector declined from 3.7 million hectares in 1990 to less than 1.5 million hectares in 1997, a 15 decrease of 60 percent (Table *state 2.2). Almost 2.3 million hectares E3Companies shifted from the state farms to the 10 *caops use of individual farms and new Olndividual private companies, increasing the s land resources of the private sector by nearly 15 percent 0 , relative to 1990. Cooperatives 1990 1991 1992 1993 1994 1995 1996 1997 lost about 300,000 hectares of the 750,000 hectares that they originally cultivated in 1990, but this de-collectivized area is offset by the decrease in the total amount of land used between 1990 and 1997. Figure 2.1 illustrates the sharp decline of the state sector, the increase of land used by individual farms, and the emergence of private companies as a new category of land users. This figure is based on the detailed time series of land ownership and land use data since 1980 which is reproduced in Table 2.3. Chapter 2: Rural Land Market 35 2.6 Comparison of land use and land ownership data in Table 2.2 suggests that most of the 1.3 million hectares cultivated by the new private companies is not privately owned and is probably leased from the state. The state is indeed an important source of land for private agriculture. In 1997, the state farm sector cultivated only 35 percent of state-owned land, and the remaining 65 percent, or nearly 2.3 million hectares was leased out for cultivation to the private sector (evenly divided between individual farms and new private companies). Table 2.3: Ownership and Use of Agricultural Land in Poland: 1980-1997 (in thousand hectares) Total Individual Cooperatives Private companies State Year Owned Used Owned Used Owned Used Used Owned Used 1980 18947 14119 1028 3698 1985 18844 14425 770 3531 1986 18804 14385 763 3526 1987 18791 14370 760 3520 1988 18742 14295 760 3521 1990 18784 18720 13497 14228 736 750 0 4551 3742 1991 18760 18764 13546 14211 695 725 12 4519 3726 1992 18741 18664 13577 14267 657 678 33 4507 3686 1993 18713 18642 13652 14602 610 623 367 4451 3050 1994 18690 18648 13732 14977 537 576 714 4421 2381 1995 18664 18622 13846 15205 476 535 980 4342 1902 1996 18663 18474 13999 15173 429 502 1179 4205 1620 1997 18608 18457 14112 15293 394 464 1225 4102 1475 Source: 1980-1988 from 1989 Statistical Yearbook; 1990-1997 from 1998 Agricultural Statistical Yearbook. For 1980-1988, cooperatives include "kolka rolnicze" listed as a separate category in the statistical abstract. Changes of Farm Structure 2.7 Private land users in Poland fall into three categories: corporate farms, individual private farms, and household plots (Table 2.4). A farm, in official Polish statistics, is a unit with more than 1 hectare of land. According to this classification, there are 2 million individual fanns averaging 7 hectares of agricultural land each. In addition, there are about one million so-called household plots averaging less than half a hectare each. The individual sector thus consists of 3 million units that cultivate about 85 percent of agricultural land. The corporate farms (all of them privatized state farms or successors of former cooperatives organized as private corporations) are much larger, averaging 620 hectares each. The total number of corporate farms is only 2,000, so that the corporate sector cultivates less than 10 percent of agricultural land (down from more than 20 percent before 1989). Because of the small concentration of land in large-scale farms, the farming structure in Poland has never displayed the sharp duality characteristic of other transition countries. The distribution of farm sizes in Poland is fairly close to the distribution in market economies, with a somewhat greater weight of the smallest farms, which in a certain sense is a sign of excessive fragmnentation. Chapter 2: Rural Land Market 36 Table 2.4: Distribution of Agricultural Land among Different Types of Farms: 1996 Number of units Percent of agricultural land Average size, ha Household plots 975,000 2% 0.4 Individual farms 2,100,000 85% 7.0 Corporate farms 2,000 7% 620.0 Unused agricultural land - 6% -- Total 3,077,000 100% Source: Uzytkowanie i jakosc gruntow, Powszechny Spis Rolny, str 6, GUS, 1996. 2.8 The farm structure in Poland is not static. Land markets began to emerge after 1990, allowing farmers to engage in land transactions with other farmers and with the state (see the section on Land Markets below). Land appears to be flowing from small farms of 1-5 hectares to larger farms with 10 hectares and more (Figure 2.2). According tndmorpublicuregistration A rdaa 6 Figure 2.2: Private Farms that changed their Size in 1990-96 to public registration data, 60 percent of farms that reduced their size between 1990 andl996 percent of farms had 1-5 hectares, compared with 40 only 35 percent among farms that 35 increased their size. On the other 30 hand, 40 percent of farms that 25 _ _ - increased their size were larger 20 ___ *Jtncrease than 10 hectares, compared with |Decrease| only 15 percent among farms that 15 - _- reduced their size. Thus, small 10 __*_ farms become smaller and 5 3 possibly go out of business, while 0 1 relatively large farms grow by 1-2 2-5 5-10 10-20 20-50 50+ acquiring more land. hectares 2.9 The number of fanns that reduce their size and thus supply land through the land market is much smaller than the number of farms that generate the demand for land: between 1990 and 1996, more than 14 percent of farms increased their size by acquiring additional land, while less than 6 percent of farms acted as land suppliers. Private land is thus insufficient to meet the full demand, and, as we have noted previously, APA plays an important role in maintaining a sufficient supply of land from state reserves. Unfortunately, the distribution of APA land across the country is extremely uneven: most state land is concentrated in western and northern Poland, where the number of private farms is relatively small (about 30 percent of the total). As a result, APA land accounted for 60-80 percent of land transactions in these regions, compared with 30- 40 percent in central-eastern and southern parts of the country. The average lease contract in western and northern regions was around 15 hectares, compared with less than 5 hectares throughout the rest of the country (Table 2.5). Availability of state land in specific regions thus determines the actual growth pattern of private farms across the country. Chapter 2: Rural Land Market 37 Table 2.5: Territorial Distribution of Land Transactions, 1990-1996 Regions Share of farms participating in Average transaction per farm, ha land transactions, percent Purchased Leased Northeast 18.3 5.4 13.0 North 23.1 10.1 25.0 Central--west 15.5 6.3 11.3 Southwest 20.1 6.6 15.4 Central 10.0 3.8 4.8 Capital 9.2 3.1 5.9 Central--east 13.7 2.9 4.4 South 8.6 4.4 4.4 Southeast 6.9 1.8 3.0 Source: Institute for Agncultural and Food Economics. 2.10 The apparent polarization of the farm structure in Poland is confirmed by long-term statistics on the number of farms in different size categories. Farm survey data published by GUS up to 1992 show that the number of smallest farms with land holdings of 1-3 hectares increased from 4 percent of all farms in 1958 to 11 percent of all farms in 1990, and the number of larger farms with land holdings greater than 10 hectares increased from 34 percent of all farms in 1958 to 41 percent in 1990 (Figure 2.3). The category of mid-sized farms with 3-10 hectares of land shrank in the same period from 62 percent to 48 percent of all farms. This trend is illustrated in Figure 2.3 by straight lines fitted to the actual yearly data. This trend apparently continued after 1990, although, for the last decade, GUS publishes only estimates, and not survey data. According to these estimates, the number of the smallest farms with 1-5 hectares increased from 53 percent of all farms in 1990 to 56 percent in 1997, the number of relatively large farms with more than 10 hectares increased from 17 percent to 19 percent of all farms, and the number of mid-sized farms with 5-10 hectares declined from 30 percent to 25 percent of all farms (Figure 2.4). Although the numbers and the statistical methodology are different for the two periods, the trend is clearly the same-we see an increase in the number of farms at the two extremes of the size scale, whereas the number of mid-sized farms is declining. Figure 2.3: Number of Farms by Size Category: GUS Surveys 70 percent of farms 60 50 40 -small -medium 30 -large 20 10 0 1958 1962 1966 1970 1974 1978 1982 1986 1990 Chapter 2: Rural Land Market 38 Figure 2.4: Number of Farms by Size Category: GUS Estimates 60 percent of farms 50 40 -small| 30 -medium _large 20 10 0 1990 1992 1994 1996 1998 2.11 The shift in farm structure is illustrated in Figure 2.5, which shows the Lorenz distribution curves for individual farms in 1988 and 1996. The share of arable land controlled by the 10 percent of largest individual farms (the right-hand tail of the curve) increased from about 25 percent in 1988 to almost 40 percent in 1996. This clearly implies that the larger farms grew even larger during this period. At the other extreme, the share of arable land controlled by the 50 percent of smallest individual farms (the left-hand tail of the curve) declined from 22 percent in 1988 to 17 percent in 1996, which implies that the smallest farms grew even smaller. The change in the shape of the curve between 1988 and 1996 provides evidence of ongoing consolidation among the larger individual farms in Poland. Figure 2.5: Size Distribution of Individual Farms: 1988 and 1996 100 percent of arable land 80 1988r/ 60 40 20 0 0 20 40 60 80 100 percent of farms Chapter 2: Rural Land Market 39 Land Markets Land Transactions in the Pretransition Period 2.12 During the early period after World War II, transactions in agricultural land were regulated by the post-war agricultural reform legislation and by the old legal provisions dating back to 1918-39. 6 Although legally allowed, formal transfer of ownership titles through buying and selling of land involved high legal costs, such as notary charges and taxes. As a result of high transaction costs, there were virtually no formal transactions in agricultural land in the 1950s. In addition, certain prohibitions were imposed on the sale and transfer of land received in the land reform of 1944-45. The July 1957 Act on Trade in Agricultural Estates relaxed some of the constraints, specifically lifting the restrictions on land-reform beneficiaries. The June 1963 Act on Limiting the Partition of Agricultural Farms introduced detailed land-market regulations based on the following principles: * preventing the purchase of agricultural land by individuals who did not guarantee its proper use; * limiting (from social considerations) the development of large private farms requiring permanent hired labor; and * preventing fragmentation of farms and influencing the attainment of the desired farm size. 2.13 As a result, only individuals judged to have suitable agricultural qualifications (formal education or proven practical experience) were allowed to purchase farmland; the maximum size of private farms was limited to 50 hectares of agricultural land in central and eastern Poland and to 100 hectares in western and northern parts. These regulations incorporated in the post-war Civil Code were an obvious obstacle to the development of land transactions in Poland. Land Transactions after 1990 2.14 The legal framework for land transactions was changed significantly by the amendment to the Civil Code of July 28, 1990. This amendment relaxed many of the restrictions on land transactions, including the highly subjective requirement of proper agricultural qualifications as a precondition for purchase of land and the limits on farm sizes. Now, any person may purchase agricultural land without any restriction on total holdings. Polish land law no longer contains draconian penalties for "irrational use" or "nonuse" of agricultural land. 16 Information concerning the legal and institutional framework of land markets in Poland was borrowed mainly from two sources: (1) R. Prosterman and L. Rolfes, "Review of the Legal Basis for Agricultural Land Markets in Lithuania, Poland, and Romania," in: C. Csaki and Z. Lerman, eds., Structural Change in the Farming Sectors in Central and Eastern Europe, Second World Bank/FAO Workshop, Warsaw, June 1999, World Bank Technical Paper 465, World Bank, Washington DC, 2000; (2) P. Dale and R. Baldwin, The Development of Land Markets in Central and Eastern Europe, ACE Project P2128R, November 1999 (see also P. Dale and R. Baldwin, "Emerging Land Markets in Central and Eastern Europe," in C. Csaki and Z. Lerman, ibid.). Chapter 2. Rural Land Market 40 2.15 Poland, unlike some CEE countries, allows ownership of agricultural land by legal bodies (corporate land ownership). Land ownership by foreigners is not explicitly prohibited, and land leasing to foreigners is apparently permitted without limitation. Indeed, APA has leased more than 100,000 hectares of state land to nearly 300 companies with majority foreign ownership. The issue of foreign ownership is one of considerable political and practical importance for Poland. On the political front, there is considerable resistance to possible foreign acquisition of agricultural land, especially by German citizens in the western areas that were German territory up to the end of World War II. On the practical front, land prices in Poland are currently lower than the prices for comparable agricultural land in the EU (around 3,000 zloty per hectare, compared with 15,000 zloty per hectare in eastern Germany and 30,000 zloty per hectare in western Germany). Thus, Polish farmland could be bought up relatively easily by wealthier EU nationals. It is also argued that allowing foreign ownership would drive prices up and make farmland less affordable to local farmers. Foreign ownership of land is one of the issues that will have to be dealt with as part of the EU accession strategy: EU member-states are generally required to eliminate all barriers to land ownership by EU nationals from any country, and only very specific exceptions are allowed. 2.16 The earlier government policy of intervening in land allocation with the purpose of attaining the desired farm size also has been abandoned. There are no minimum size restrictions on land holdings, although to be legally recognized as a farm the unit must have at least one hectare of land. Two legal provisions in social insurance and pension legislation, however, appear to favor farms of between one and two hectares. First, only rural residents with at least one hectare of land qualify for the heavily subsidized agricultural pension plan (KRUS). Second, rural residents with farm holdings of two hectares or more are not entitled to unemployment benefits if they lose their off-farm job. 2.17 Despite the changes in legislation, transaction costs apparently remain at a relatively high level. The main single component is the treasury tax set at 5 percent of the value of land affected by the transaction. In addition, there is a wide range of fees and charges for the preparation of maps, extracts, and new entries in the mortgage register (when land is mortgaged) (see Box 2.1). Prosterman and Rolfes quote an interview with a farmer who purchased 0.26 hectares for 500 zloty and had to pay another 500 zloty in notary fees, taxes, and registration charges. Zientara cites the following example of detailed cost calculations. 2.18 These relatively heavy transaction costs are a serious barrier to buying and selling land. They encourage informal transactions between individuals, which are not registered by cadastral organs and thus cannot be monitored. On the positive side, the revenue from the sale of farmland is exempt from personal income tax, provided that the transaction does not change the agricultural use of the land. Chapter 2: Rural Land Market 41 Box 2.1 Two private individuals agreed on a transaction involving 3 hectares of undeveloped agricultural land at a price of 9,000 zloty (in 1998). The costs of preparing the land survey documents were to be bome by the seller, and the remaining costs were to be borne by the buyer. Calculation of transaction costs (in zloty): - cadastral map extract and mortgage register extract - 80.00 - notary fee 270.00 - fee for extracts (6 extracts, 4 pages each, 5 zloty per page) - 120.00 - treasury tax (5% of transaction value) - 450.00 - fee for setting up the mortgage register - 30.00 - fee for entenng the ownership rights into the mortgage register - 144.00 - fee for printing the mortgage register - 9.15 Total 1103.15 The total transaction cost is 12.5% of the value of auricultural land sold to the new owner. Land Registration and Land Mortgage 2.19 Land registration has a long tradition in Poland. Land and mortgage registers were originally set up by the decree of October 1946. They were intended mainly for the benefit of private landowners, and there was (and still is) no universal obligation to register land, except in the event of sale. Currently, land registration and land mortgage are regulated by the 1982 Act on Perpetual Books and Mortgages, as amended in August 1997. The perpetual books are the land and mortgage registers maintained by nearly 300 district courts, and registration entries must be made by a judge. The system is cumbersome and obviously prone to serious delays. The overall organization of the Polish cadastral system is shown in Figure 2.5. Indeed, contrary to the views of Polish lawyers, Prosterman and Rolfes are not entirely convinced that registration of land rights under the current law provides full assurance and suggest the eventual introduction of private "title insurance," an effective, but admittedly very costly solution. 2.20 The impetus for land registration in most CEE and CIS countries was provided by restitution and privatization policies that changed the status of land rights and added many new beneficiaries to the circle of landowners. The situation in Poland is different, as this country did not need to implement sweeping restitution and privatization programs after 1989. As a result, an estimated 30-40 percent of land in Poland is not registered, and treatment of unregistered land is a major issue. An application to register unregistered land is made only when a transaction (or mortgage) affecting that land takes place. The notary handling the transaction is required to open a registration volume (a perpetual book) for the unregistered land. Land registration is not a simple process, as it draws on numerous sources of documents, including documents held by the right-claimant, notarial documents, and materials kept in government offices (especially the land and building register). With only such "sporadic," transaction-driven registration, decades may pass before all land is registered. Moreover, while the registration fees are quite high (see Figure 2.6), they are not retained for the operation and improvement of the registry system, which appears to be seriously under-funded and lacking adequate personnel. Chapter 2: Rural Land Market 42 Figure 2.6: Organization of the Polish Cadastral System (Dale and Baldwin) Ministry of Ministry of Interior and Agriculture Public Administration Ministry of Head Office Main Justice Cartography Documentation and Geodesy Center Voivodships Noties I|District District offices Documentation courts 1centers 1,100 286 267 373 k y t mortgage ) ~~~~~building J registers ~~registers 2.21 Land mortgages are regulated by the January 1998 Law on Mortgage Bonds and Mortgage Banks. This law introduced a significant change in the previous attitude of the Civil Code toward mortgage liens, raising them from sixth to third place in the order of priority of creditors' claims (mortgage liens are now preceded only by foreclosure costs and alimony claims). Yet banks complain that foreclosure proceedings, which must be conducted through the courts, are very slow and costly. Mortgagors also suffer from delays, because mortgages do not become effective until notarized and registered, which, as noted previously, is an inherently slow process in Poland. Mortgage loans are limited by law to 60 percent of the value of the real estate mortgaged, which requires the borrower to have access to considerable cash or nonmortgage borrowing. Despite these difficulties, significant numbers of loans using agricultural land as security are made by state, cooperative, and private banks. In 1998, the BGZ Bank-the Bank for Food Economy (in English), which is the apex bank of the Polish cooperative system-made about 3,500 mortgage loans for the purchase of agricultural land (these loans were subsidized by the government). The greatest impediment to agricultural land mortgage appears to be social, rather than legal, in nature: many rural residents feel uneasy with bank foreclosures and may be unwilling to buy their neighbor's land from the foreclosing bank. Chapter 2: Rural Land Market 43 2.22 A general assessment of land market performance carried out by Dale and Baldwin assigns a score of 30 to Poland (100 is the EU benchmark). This is lower than for Hungary and the Czech Republic (45 and 35, respectively), mainly because the land title database in Poland is judged to be less complete. The levels of annual title transfers and mortgage activity are also lower in Poland. Land Transactions Involving Private Farmers 2.23 There is no uniform global monitoring of land transactions in Poland, but partial information on land transactions of private farmers is available from two sources. Prof. L. Ostrowski at the Institute of Agricultural and Food Economics has been monitoring such transactions since 1992. His information base includes official data from GUS (both published and unpublished) and APA, Ministry of Justice data obtained from notary offices, and data collected by the Institute from tenders and real estate agencies. In parallel, public agricultural registration data were published for the period 1990-96. These two sources use different definitions and different coverage, and the volume of transactions reported by the Institute is substantially greater than that suggested by official registration data. Despite the difference in numbers, the general patterns are similar. The annual volume of land transactions has been generally increasing in the 1990s (Figure 2.7), and, according to the Institute data, a total of 5.7 million hectares changed hands in 1.5 million transactions during the seven year period 1992-98 (Table 2.6). Most of this land (cumulatively, 4 million hectares or nearly 70 percent) was transacted by leasing, not by buying and selling (Figure 2.7). It is significant that more than 60 percent of land in these transactions was privately owned, moving directly between individuals. Nevertheless, APA was a major player in the creation of land markets in Poland, with state- owned land representing nearly 40 percent of the total volume of transactions during this period. Figure 2.7: Poland: Land Transactions (Buying and Leasing) 1200 1000 X __X 800 . 600 M~~~~~~~~~~Leased i6Purchased 400 200 - 0 1992 1993 1994 1995 1996 1997 1998 2.24 While leasing dominates land-market transactions for both state-owned and private land, the tendency toward buying and selling is more apparent for private land. Only 25 percent of state-owned land affected by transactions was sold during this period, compared with 34 percent for private land (Table 2.6). The prominent peak in land leasing in 1994-95 (Figure 2.7) was in Chapter 2: Rural Land Market 44 fact attributable to a special campaign launched by APA to stimulate reallocation of state land reserves. The decline after the peak is a natural outcome of the long-term nature of the leases signed with APA: since the pool of state-owned land is limited and no new land becomes available, the leasing activity could not be maintained at the high levels of 1994-95. Table 2.6: Land Transactions: Cumulative for 1992-98 (Institute of Agricultural and Food Economics) A. Cumulative Volume of Transactions (thousand hectares) All transactions Buying and selling Leasing Total transaction volume 5730 1754 3976 Private land 3537 1205 2332 State land 2193 549 1644 B. Structure of Sources of Land in Different Types of Transactions (in percent of transaction volume) All sources of land Buying and selling Leasing Private land 62 69 59 State land 38 31 41 All land 100 (5,730 ha) 100 (1,754 ha) 100 (3,976 ha) C. Distribution of Land by Different Types of Transactions (in percent of transacted land) All transactions Buying and selling Leasing Private land 100 (3,537 ha) 34 66 State land 100 (2,193 ha) 25 75 All land 100 (5,730 ha) 31 69 2.25 The average leasing transaction is generally larger than the average land purchase transaction. Moreover, transactions in state land are, on average, substantially larger than transactions in private land. According to the Institute's estimates, the average lease of state land is close to 11 hectares, compared to 6 hectares for private land leases (Table 2.7). Buy-and-sell transactions are much smaller, averaging about 3 hectares only. Table 2.7: Average Size of Land Transactions in the 1 990s (hectares) Official registration data 1990-96 Institute data 1992-98 Buy/sell Lease Buy/sell Lease Pnvate land 3.8 4.5 2.6 3.0 State land 9.6 19.0 5.9 10.6 Source: Institute of Agricultural and Food Economics. Land Prices 2.26 The average price per hectare of arable land in 1998 was about 4,400 zloty, ranging from 2,500 zloty for land of poor quality to 6,000 zloty for best-quality land. This was the price for transactions in privately owned land between farmers. The price charged by APA for state land in that year was 3,000 zloty per hectare, about 30 percent less. APA prices for state land were typically 30-40 percent lower than the prices between farmers during the entire period (Table 2.8). Chapter2: RuralLandMarket 45 Table 2.8: Average Prices per Hectare of Arable Land in Transactions between Farmers and with APA Year Price between farmers APA price Consumer price index Zloty decitons of rye Zloty (1991=100) 1991 877 145 -- 100.0 1992 1220 120 405 144.3 1993 1810 86 1100 198.6 1994 2210 112 1370 257.1 1995 2421 90 1491 1996 3216 82 1871 374.9 1997 3946 99 2444 424.4 1998 4379 133 3048 465.5 Source: Statistical yearbooks, APA. 2.27 The nominal price of arable land increased between 1991 and 1998 at about the same rate as inflation, so that the real price per hectare remained constant over time (Figure 2.7). This is also evident from the price of land quoted in decitons of rye, which remained fairly steady over time, fluctuating around 100-120 (Table 2.8). 2.28 The average price of agricultural land sold in private transactions in Poland is lower than in the EU. Agricultural land in the western region of Germany sells on average for DM 15,000 per hectare (about 32,000 zloty per hectare), and such land in the eastern region of Germany sells for DM 7,000-9,000 per hectare (about 15,000-19,000 zloty per hectare). These land prices are substantially higher than the average of 4,400 zloty per hectare in Poland. Figure 2.8: Nominal Price of Land and Consumer Price Index: 1991-98 10000 1000 100 ...___________________________ 10 1991 1992 1993 1994 1995 1996 1997 1998 -zIotyIha -cpi91 Re-Privatization and Restitution Issues 2.29 As noted previously, APA has concentrated mainly on leasing out state land instead of selling it to private owners. Although the state land reserve has not been privatized, APA has virtually completed its task of redistributing former state land to private use. Only 700,000 Chapter 2: Rural Land Market 46 hectares of 4.5 million hectares of state land have been sold, but the rest has been leased out to private users, and the agency's reserve of about 500,000 hectares is mainly poor quality land in remote locations that is very difficult to place. The shift to private use of state-owned land is evident from the comparison of ownership and use figures in Tables 2.1 and 2.2. 2.30 According to APA, the lack of a restitution law is the major factor that blocks large-scale sale of leased state land. APA officials claim that thousands of claims have been filed by former landowners all across the country. There is no mechanism for dealing with these claims so far, but even suspicion of a claim is sufficient to block APA's plans to sell the land plot. 2.31 Claims are often filed for pre-1944 land holdings, which exceed many times the area of plots currently in the APA land stock. This is so because 6 million hectares of the pre-1944 private land expropriated from large landowners (with holdings of more than 50-100 hectares) according to the post-war land reform decree of September 6, 1944 was partitioned and distributed to individuals between 1944 and 1950. Moreover, some of the pre-1944 holdings have since been sold by the state to individual farmers, and are no longer controlled by APA. Finally, some of the land that can be potentially claimed by former owners is leased out on 10- 15 year leases to other users, whose interests must be considered if any re-privatization is attempted. All these factors create tremendous difficulties for restitution of land to former owners, as only a small part of the original holdings are still in state ownership and can be potentially re-privatized. Land and Farming in the Survey 2.32 The survey covered 2,835 respondents spread fairly uniformly in four voivodships (Table 2.9). The voivodships were chosen to represent the northern, central, and southern belts of the country. Table 2.9: Number of Households in the Survey Voivodship Region Capital Code Number of Percent of respondents respondents Malopolskie Southeast Krakow 12 707 25 Mazowieckie Central, east Warsaw 14 677 24 Wielkopolskie Central, west Poznan 30 713 25 Pomorskie Northwest Szczeci 32 738 26 Total 2,835 100.0 Source: Special Rural Household Survey 2000. 2.33 Land ownership is reported by 60 percent of respondents in the survey. Yet only 53 percent of respondents actually cultivate land, i.e., are classified as "farrmers" for the purposes of this study. The remaining 47 percent do not farm (although some of them own land), and their entire income is derived from sources outside agriculture. 2.34 Figure 2.9 compares the distribution of farm sizes in the survey to the national distribution reported by GUS. A "farm" in this survey is any unit where land is cultivated, i.e., where respondents report using land to grow crops or raise livestock. No a priori lower limit was imposed on what constitutes a farm. The sample contained a higher proportion of farms with Chapter 2: Rural Land Market 47 more than 20 hectares of land compared with the national GUS Figure 2.9: Farm Size Distribution (fanm that use land) distribution. This was intentional, as we wanted to 35 include a sufficiently large 30 number of relatively large and - entrepreneurially oriented farms 25 in the survey. 20 - 15 - - Land Ownership and Land Use 1_ 2.35 The average amount of 5 privately owned land is 8.5 0 hectares per household reporting 0-1 1-2 2-3 3-5 5-7 7-10 10-15 1520 20-30 30-50 >50 land ownership (i.e., based on 60 hectare percent of respondents). In the 10jfvey a sample, 13 percent of respondents own 20 or more hectares and 5 percent of Figure 2.10: Distribution of Land Ownership by Size respondents own between 31 hectares and 115 hectares. 25 percentoflandowners Half of the respondents own up to 3.7 hectares. The 20 _ _ distribution of land ownership in the sample is shown in 15 __ _ . Figure 2. 1 0. 10 2.36 The mean size of an active farm in the sample is 10.4 hectares and the median size is 4.3 (based on 53 percent 0-1 1-2 2-3 3-5 5-7 7-10 10-20 20-50 >50 of respondents who cultivate hectare land). By both these measures, the average amount of cultivated land is greater than the average amount of owned land, which means that farmers resort to land leasing. The overall distribution of farm sizes is very similar to the distribution of land ownership (compare Figures 2.10 and 2.11). Figure. 2.11: Farm Size Distribution Farms are split into 1-5 parcels for 50 percent of respondents (the 25 percent of farms median number of parcels is 3, the 20 _ __ _____ ___ mean is 4.5). The distance of the farthest parcel from the house is 1-4 5 __ _ kilometers for 50 percent of 10 respondents (2 kilometers median distance, 3.1 kilometers mean S distance). I 1 L-111L 0--i 1-2 2-3 3-5 5-7 7-10 10-20 20-50 >50 hectare Chapter 2: Rural Land Market 48 2.37 Landowners generally cultivate their land, but almost 15 percent of landowners do not farm. Among farmers, on the other hand, only 3 percent do not own any land and presumably farm on land leased from others. Farming households are, on average, larger than nonfarming households and are headed by a younger person with a higher educational attainment (Table 2.10). Among nonfarming households, there are no differences in these characteristics between landowners and those who do not own land. Table 2.10: Characteristics of Nonfarming and Farming Households in the Survey Nonfarming (n=1320) Farming (n=1515) Total (n=2835) Family size 3.2 4.4 3.9 Age of head of household 53.2 48.8 50.8 Off-farm income 15,000 13,600 14,200 Total family income 16,600 24,100 20,600 Education of head of household Higher and secondary 23 20 21 Uncompleted primary 16 8 11 Source: Special Rural Household Survey 2000. 2.38 Small landowners with up to 2 hectares show a relatively high tendency to cultivate less land than they own. Large landowners, on the other hand, frequently cultivate more land than they own (Table 2.11). Thus, among landowners with 0-2 hectares, 34 percent use only part of their land and 5 percent report using more than the land they own. Among landowners with more than 10 hectares, 18 percent use only part of their land and 36 percent use more than the land they own. This essentially means that land flows from small to large landowners, presumably because larger farms are more profitable. Table 2.11: Land Ownership and Land Use: Percent of Respondents by Ownership Category (out of total 2,835) Land ownership (ha.) | Use less than own Use = own Use more than own Total All sample 14.0 76.0 10.0 2835 0 -- 95.5 4.5 1126 0-2 33.9 60.9 5.2 614 2-10 16.8 76.4 6.9 656 >10 18.2 46.2 35.5 439 Source: Special Rural Household Survey 2000. Land Leasing 2.39 Land leasing is not particularly widespread in Poland. Among landowners, leasing out of land is reported only by 8 percent of households with privately owned land. Among farmers, only 17 percent lease land and the great majority (83 percent) rely entirely on their own land. Farmers operating with leased land accumulate much larger holdings (Table 2.12): they cultivate on average 25.7 hectares, compared with 7.3 hectares for farmers who do not lease in land (the difference is statistically significant). Chapter 2: Rural Land Market 49 Table 2.12: Size and Structure of Farmns with and without Leased Land All farms (respondents with Farms with leased land Farms without leased lT land) land Numberoffarms 1515(100%) 255(17%) 1260(83%) Farm size 10.4 25.7 7.3 Own land 8.4 13.8 7.3 Leased land 2.0 11.9 0 Source: Special Rural Household Survey 2000. 2.40 Shortage of manpower is the main reason given by landowners for leasing out land (42 percent of those who lease out), followed by lack of profitability (26 percent). Landowners who lease out land are older, have smaller families, and control a smaller land endowment than those who do not lease out. The lessors' profile is given in Table 2.13, which shows that lessors on average lease out more than 60 percent of their land. Table 2.13: Profile of Land Owners who Lease out Land Lessors (n = 133) Other landowners (n=1576) All landowners (n=1709) Age 51 48 49 Family size 3.1 4.4 4.3 Land owned 5.4 8.8 8.5 Land in use 2.1 9.7 9.1 Source: Special Rural Household Survey 2000. 2.41 Despite the active involvement of some individual landowners in lease markets, APA is the main source of leased land. Nearly 60 percent of land leased in by individual farmers is state land from the APA reserve. One-third is leased from private persons, and about half of this land is covered by formal lease contracts (Table 2.14). Transactions with APA always involve signing an official contract, so that, overall, three-quarters of the land leased in by individual farmers (76 percent) is subject to a formal lease contract. Nevertheless, farmers are generally not familiar with formal dispute resolution mechanisms in connection with leased land. Among the respondents who leased land, 12 percent claim that no dispute resolution mechanisms exist, 34 percent "don't know" what dispute resolution mechanisms are available, and 53 percent cite strictly informal dispute resolution mechanisms, such as talking the problem out between the parties or going with the problem to the gmina chairman (Table 2.15). Table 2.14: Structure of Leased Land by Sources (in percent of leased land, n = 255) From APA From private From private persons with From persons - total formal contract others Leasing by sources 59% 34% 17% 7% Source: Special Rural Household Survey 2000. Table 2.15: Dispute Resolution Mechanisms (multiple answers allowed) Percent of respondents with leased land (n=255) Talk the problem over between the parties 46 Go to gmina chairperson 7 Go to the court 3 Appoint an arbitrator 2 Other method 9 No dispute resolution mechanism available 12 Don't know 34 Chapter22: RuralLandMarket 50 Lease Payments 2.42 Land leasing usually involves some sort of payment. Thus, 74 percent of lessors report that they receive some form of lease payments for their land and 85 percent of lessees report that they pay for land that they lease from the state or other private landowners. In about 30-40 percent of the cases, the lessee is only required to pay the land tax instead of the owner, and in these cases the lease payments are relatively low (Table 2.16). In another 45 percent of the cases, the lessee is committed to higher payments in cash or in kind. Sharecropping arrangements that involve payment of an agreed percentage of output from leased land are not practiced. Table 2.16: Lease Payments as Reported by Lessors and Lessees Percent of respondents Median payment, zloty Form of lease payment Lessors (n=133) Lessees (n=255) Lessors Lessees Only land tax 31 37 150 190 Fixed amount in cash 19 20 300 500 Fixed amount in kind 18 6 275 550 Percentage of output 0 1 -- 1-20% Several forms of payment 6 21 930 1,000 All lease payments 74 85 228 400 No lease payments 26 15 -- -- Source: Special Rural Household Survey 2000. 2.43 The information on lease payments provided by lessors and lessees was used to estimate the average lease rate per hectare (by regression analysis). The results proved surprisingly consistent, producing estimates of about 80 zloty per hectare for each category of respondents (R-square=0.29 and p < 0.0001; the difference in estimates between lessors and lessees was not statistically significant). Buying and Selling of Land 2.44 In addition to leasing, land transactions in Poland also include buying and selling of land. About 5 percent of respondents report selling or buying land in the five years between 1995 and 1999. Another 5-7 percent of respondents have plans to buy or sell land in the near future. However, the great majority (82 percent of respondents) have no plans to acquire additional land in the immediate future. 2.45 The average sale transaction between 1995 an 1999 involved 2.6 hectares at a price of 1,800 zloty per hectare (estimated by regression analysis). The average buy transaction involved 6.7 hectares at a price of 2,300 zloty per hectare (Table 2.17). The difference in prices is not statistically significant, and the survey thus suggests average land prices of around 2,000 zloty per hectare. This result is confirmed by the responses of potential buyers (5 percent of respondents), who report their willingness to pay an estimated price of 2,350 zloty per hectare. Since the rate for land leasing is estimated at 80 zloty per hectare, the price of land in buy-and- sell transactions is thus equivalent to 25 years of leasing. Chapter 2: Rural Land Market 51 Table 2.17: Frequency of Land Transactions and Estimated Prices per Hectare Land transactions Percent of respondents Hectares (mean) Prices (regression estimate) Sell 1995-99 4.8% of all (n=2835) 2.6 1,800 (R-sq=0.05) Buy 1995-99 6.4% of all (n=2835) 6.7 2,300 (R-sq=0.47) Planning to sell 7.4% of owners (n=1709) NA NA Plan to buy' 4.9% of all (n=2835) 12 2,350 (R-sq=0.60) Lease in 17% of farmers (n=1515) 12 79 (R-sq=0.28) Lease out 8% of owners (n=1709) 3.4 80 (R-sq=g0.18) 1) Of these potential buyers, 96% already have land. Source: Special Rural Household Survey 2000 2.46 Land is primarily bought and sold among private individuals from the same district (Table 2.18). Buying and selling transactions with the state are infrequent. Table 2.18: Main Parties in Buy-and-Sell Transactions Main buyer Main seller Private individual, this district 61 82 Private individual, another district 20 3 State/APA 7 12 Other 12 3 Source: Special Rural Household Survey 2000. 2.47 There is no clear pattern in the reasons that respondents give for having sold land or for not planning to acquire land in the future. Nevertheless, one reason emerges clearly for all respondents: farming is not profitable for about one-third of respondents, both among those who have sold land and among those who do not plan to purchase land (Table 2.19). Table 2.19: Main Reasons for Selling and Not Buying Land Why did you sell land? Percent of Why are you not planning to acquire Percent or I res ondents land? respondents Farming not profitable 30 Farming not profitable 33 Shortage of labor/capital 23 Shortage labor/capital 17 Need cash for other purposes 20 No money to buy land II Other reasons 27 Other reasons 27 Have enough land 12 Those who sold, percent of all Those who do not plan to acquire, respondents 4.8 percent of respondents 82 No response 13 Source: Special Rural Household Survey 2000. Change in Land Holdings between 1997 and 1999 2.48 Most farms (82 percent) did not change their size between 1997 and 1999; 11 percent report size increases of slightly more than 7 hectares per farm and 7 percent report decreases of about 4 hectares per farm. In both cases, the change is nearly 40 percent of the 1997 size (see Table 2.20). Although size changes are generally infrequent, the tendency is clearly toward enlargement of the existing farm. Chapter 2: Rural Land Market 52 Table 2.20: Change in Land Holdin s between 1997 and 1999 No change Increase Decrease All (53% w/land) Number of farns 1,237/82% 164/11% 114/7% 1,515 1999 (sum) 10,499 ha 4,388 767 15,654 1997 (sum) 10,499 3,168 1,237 14,904 Change (sum) 0 +1,220 (39% of 97) -470 (38% of 97) +750 1999 (mean) 8.5 ha 26.7 6.7 10.3 1997 (mean) 8.5 19.3 10.8 9.8 Change(mean) 0 7.4 -4.1 0.5 Source: Special Rural Household Survey 2000. 2.49 Land market transactions play an important role in farm size changes (Table 2.21). In farms that increased their size between 1997 and 1999, over 90 percent of additional land was acquired through purchase or leasing. In farms that reduced their size in the same period, over 40 percent of land was sold or leased out. Table 2.21: Sources of Change in Land Holdin s between 1997 and 1999 Source of change Increase Sources of change Decrease Total change +1,220 ha Total change -470 ha Inherited 6% Gave away as a gift 9% Purchased 42% Sold 26% Leased in 51% Leased out 16% Acquired by other means 1% Disposed by other means 22% Began cultivating fallow 0% Let lay fallow 26% Source: Special Rural Household Survey 2000. 2.50 The change in land between 1997 and 1999 is positively correlated with the initial farm size in 1997 (R=0.35): on average, each additional hectare of initial holdings increases the size change by 0.10 hectares. The large farms thus show a tendency to grow through acquisition of more land, whereas small farms become even smaller by selling or leasing out some of their holdings. Table 2.22 shows that farms with up to 20 hectares of land (in 1999) on average lost 0.2 hectares between 1997 and 1999, whereas farms with more than 30 hectares (in 1999) on average gained 7 hectares during this period. The size increase in farms with more than 30 hectares is significantly greater than in farms with less than 30 hectares (by Bonferoni and similar tests); there are no statistically significant differences in farm size changes among the smaller size categories (up to 20 hectares). Table 2.22: Change of Farm Size between 1997 and 1999 for Farms in Different Size Categories Farm size (1999) Number of farns Mean change 97-99 Increase No change Decrease Up to 20 ha 1,276 -0.18 ha 76/6% 1,091/86% 109/9% 20-30 ha 124 1.30 ha 35/28% 86/69% 3/2% 30-50 ha 78 4.67 ha** (p<0.05) 33/42% 45/58% 0 Source: Special Rural Household Survey 2000. Determinants of Farm Size 2.51 Since land leasing is a relatively infrequent phenomenon, access to family-owned land is clearly the main determinant of farm size. There is a very high correlation (0.82) between farm size and "own land." Own land accounts for most of the land under cultivation in farms of all sizes (Table 2.23). Only for larger farms (over 10 hectares) is the average farm size significantly Chapter 2: Rural Land Market 53 larger than the own land under cultivation. Table 2.23 also presents additional characteristics of farms in different size categories. Larger farm sizes are associated with larger families. Larger farms are managed by younger and better educated farmers, while older farmers with lower educational attainments typically run smaller farms. In addition to human capital, availability of fann machinery is also positively correlated with greater farm size. Finally, larger farms have a relatively more commercial orientation, selling a higher proportion of their output than smaller farms. Product mix, i.e., the share of crops and livestock in farm production, does not appear to be related to farm size. Diversification of family income has a negative impact on farm size: families with a higher level of off-farm income (controlling for family size) have smaller farms, presumably because much of their working time is spent in off-farm occupations. This is contrary to initial expectations: off-farm income is often regarded as a cushion against farming risks, and thus greater income diversification is expected to enable families to maintain larger farms. Table 2.23: Characteristics of Farms in Different Size Categories (n=1515 respondents who farm) Farm Size Category 0-1 ha 1-5 ha 5-10 ha >10 ha Number of respondents 308 490 282 435 Average farm size, ha 0.5 2.6 7.4 28.2 Own land, ha 0.8 2.7 7.1 23.8 Farmer's education, percent of respondents Secondary or higher education 21 22 9 25 uncompleted primary or no education 12 9 8 3 Farmer's age 54.3 49.7 48.0 44.3 Family size 3.6 4.4 4.5 5.0 Off-farm income, zloty 16,300 16,500 12,100 9,400 Share of output sold, percent 8 27 44 56 Share of crops in output, percent 60 57 49 47 Household with any farm machinery, percent 20 59 86 97 Source: Special Rural Household Survey 2000. 2.52 Table 2.24 presents the results of a regression analysis intended to identify the determinants of farm size based on all 1,515 respondents with cultivated land. As shown above, family size, farmer's education, and share of output sold have a highly significant positive effect on farm sizes. Age, as expected, has a negative effect, but it is not significant, basically because age is highly correlated with education: higher educational attainment is observed for younger people, and as a result the education variable simply displaces the age variable from the model. Off-farn income has a significant negative coefficient, as noted above: higher off-farm income is observed for smaller farms in the sample. The product mix is not a determinant of farm size according to this analysis. Chapter 2: Rural Land Market 54 Table 2.24: Determinants of Farm Size (regression analysis based on n=1515 respondents who farm) Coefficient t-value Significance level Family size 1.18 4.90 <0.0001 Education 2.25 5.39 <0.0001 Age -0.01 0.35 0.73 Off-farm income -2.1x1O-4 6.28 <0.0001 Share of output sold 0.12 8.46 <0.0001 Product rnix: share of crops in output -0.02 1.23 0.22 Availability of mnachinery (yes/no) 6.36 6.08 <0.0001 Source: Special Rural Household Survey 2000. Land and Standard of Living 2.53 Respondents perceive considerable deterioration in their standard of living during the transition decade (Figure 2.12). The survey questions explicitly recognized five standard-of- living levels: below subsistence, subsistence, adequate, comfortable, and no material difficulties. In 2000, 45 percent of households characterize their existence as not better than subsistence, compared with 11 percent in 1990. At the other extreme, 21 percent of respondents report a comfortable (or better) standard of living today, so that in addition to the basic necessities they have enough money to save and can occasionally purchase durables. In 1990, fully 60 percent of respondents were in this category. Figure 2.12: What the Family Budget Buys Below subsistence Subsistence Adequate Comfortable l No difficulties B 0 10 20 30 40 50 percent of farms =lo 1990 2000 2.54 Use of agricultural land is one of the factors that explain the differences in the perceived standard of living. To simplify the analysis, the five original standard-of-living categories were aggregated into three, "lower" representing the original "below subsistence" and "subsistence" categories, "medium" representing the original "adequate" category, and "upper" representing the original "comfortable" and "no difficulties" categories. The sample is split evenly between Chapter 2: Rural Land Market 55 households that farm (possibly on leased land) and households that do not farm (even if they own land). Among nonfarming households, 50 percent report that their income is at best sufficient for subsistence, compared with only 40 percent among farming households. Fully 60 percent of farming households report that their standard of living is adequate or better, compared with 50 percent among nonfarming households. The differences in these percentages are statistically significant. A closer look at the use of agricultural land shows that the average holdings increase for households reporting a higher standard of living. Thus, farming households in the lower standard of living category average 7.6 hectares of land, whereas farming households in the upper standard of living category average 15.4 hectares. The differences in land holdings among the three standard-of-living categories are significant (Table 2.25). Table 2.25: Relationship between Land and Standard of Living' Standard of Nonfarming households Farming households Average land holding for living category (n=1314) (n=1515) farning households Low 49% 41% 7.6 ha Medium 31% 37% 10.7 ha High 20% 22% 15.4 ha 1) The difference in the distribution of farming and nonfarming households by standard-of-living categories is statistically significant by chi-square test. The difference in average land holdings between any two standard-of-living categories is statistically significant by the Bonferoni simultaneous comparison test. Source: Special Rural Household Survey 2000. 2.55 The probability of being in a particular standard of living category as a function of land holdings can be analyzed by multinomial logistic regression. The explanatory variable "land" included all farming and nonfarming households (2829 observations), with land in nonfarming households assigned the value zero. The logistic regression produced highly significant results, which show that an increase of one hectare in farned land increases by nearly 3 percent the odds of advancing from the lower to the upper standard-of-living category and by over 2 percent the odds of moving from the lower to the medium standard-of-living category (Table 2.26). Although the regression results leading to these conclusions are statistically significant, the model provides a relatively poor fit of the data. This is not surprising, given more than 2000 noisy observations, and it indicates that land use is not the only determinant of the standard of living: additional variables may be added to the multinomial logistic model to improve the fit. Table 2.26: Standard of Living as Determined by Use of Land: Multinomial Logistic Analysis Using CATMOD (n=2829) Lower vs. Medium vs. Lower vs. Medium Upper Upper Intercept 0.91 0.52 0.39 Land in ag. Use -0.0290 -0.00738 -0.0216 Odds ratio 0.9714 0.9926 0.9786 Percentage change in odds of being in lower standard- of-living category per one unit increase of land -2.9% -0.7% -2.1% Source: Special Rural Household Survey 2000. Chapter 2: Rural Land Market 56 Nonfarm Business Activities 2.56 Nonfarm business activities are reported by 8 percent of respondents (Table 2.27). The incidence of nonfarm business activities is significantly higher among nonfarming households and particularly among landowners who do not farmn. The number of respondents reporting nonfarm activities in these two nonfarming categories is 10 percent and 16 percent, respectively. This finding, however, does not necessarily suggest that nonfarm business activities are a substitute for farming. Farmners reporting nonfarrn business activities cultivate more land than farmers without other businesses (14 hectares and 10 hectares respectively; the difference is statistically significant). Logistic regression fails to detect any significant dependence of the choice to enter nonfarn businesses on the amount of cultivated land. It thus seems that the decision to engage in nonfarn business activities is more a reflection of entrepreneurial spirit than shortage of land. Table 2.27: Nonfarm Business Activities by Categories of Respondents All samnple Land owners Nonfarming land Farmers Nonfarmers (n=2796) (n=1684) owners (n=245) (n=1490) (n=1306) Households with nonfarm business activities 231 (8%) 141 (8%) 38 (16%) 106 (7%) 125 (10%) Households without nonfarm business activities 2565(92%) 1543 (92%) 207(84%) 1384(93%) 1181(90%) Source: Special Rural Household Survey 2000. 2.57 The most popular nonfarm businesses are trade, construction, manufacturing, and transportation (Table 2.28). Farm-related services, such as processing, marketing for other farners, or input supply to farmers, are reported by very few respondents. Equipment leasing and the provision of custom field work using one's own machinery are not common either. Over one-third of respondents with nonfarm business activities engage in a mix of other "nonconventional" activities. In most families, there is only one entrepreneur, and the average number of people engaged in nonfarm businesses is 1.1 per family (Table 2.20). Table 2.28: Types of Nonfarm Business Activities in the Sample Number of households Number of individuals Trade and commerce 71 85 Construction 47 47 Manufacturing 29 36 Transportation, storage, packing 22 26 Repair work 10 10 Equipment leasing 0 0 Mechanical field services 4 4 Processing of farm products 4 6 Collection/sale from other farms 5 5 Delivering inputs 3 3 Restaurant, lodging 4 6 Training, teaching 3 3 Other 33 36 Total 235 267 Source: Special Rural Household Survey 2000. Chapter 2: Rural Land Market 57 2.58 Table 2.29 presents a rough profit-and-loss statement of nonfarm business activities, based on a very small number of respondents (n=134). Nonfarm business activities appear to be profitable, producing an after-tax profit of 14 percent of sales. Table 2.29: Profitability of Nonfarm Business Activities N=134 Thousand Percent of Percent of Zloty sales costs Revenue 233 100 Total expenses 192 82 100 Labor 28 15 Materials, inputs 153 80 Other operating costs 11 5 Profit before tax 41 18 Tax 10 4 Profitaftertax 31 14 Source: Special Rural Household Survey 2000. Land and Family Income 2.59 Based on all 2,835 rural households in the sample, the average family income is PLN 21,000 (Table 2.30). Of this, (a) 70 percent is personal income in the form of off-farm salaries, pensions, and other transfers, (b) 22 percent is net cash farm income (excluding the value of farm products consumed by the family), and (c) 8 percent is net income from nonfarm business activities. Farming families, i.e., families that cultivate some land, report significantly higher total incomes: PLN 24,000 compared with PLN 17,000 for families that do not cultivate land. The difference between the two groups of families is statistically significant even before we take into account the value of own farm products consumed by the farming families. Table 2.30: Structure of Cash Family Income All sample Nonfarming Farming families (n=2835) families (n=1320) (n=1515) Salaries 7,244 8,111 6.489 KRUS 2,459 1,679 3,139 zUs 3,551 4,220 2,968 Other personal income 992 996 987 Total personal income 14,246 15,006 13,583 Nonfarm business income 1,747 1,847 1,661 Farm income (net of cost of purchased inputs and payments to nonfamily labor) 4,634 -- 8,877 Total family income 20,627 16,853 24,121 Average family size 3.9 3.2 4.4 Source: Special Rural Household Survey 2000. 2.60 The structure of family income for farming and nonfarming households is shown in Table 2.30, while a more detailed breakdown of the income structure of farming families is presented in Figure 2.13 (below). The difference in the amount of nonfarm business income is not statistically significant between the two groups, while personal income is statistically significantly higher for nonfarming families, mainly due to differences in salaries. Farming families are on average larger than nonfarming families: 4.4 persons, compared with 3.2. The differences in per-capita income, if any, require further analysis because of statistical problems. Chapter 2: Rural Land Market 58 Figure 2.13: Structure of Income: Farming Families Salaries 27% Personal 29% Business 7% Farm Average family income 24,121 zloty 37% Table 2.31: Disposition of Farm Production among Sales, Consumption, and Other Uses Percent of total Farmers who sell Farmers who don't sell production-all (n=1065/1040) (n=457/387) farmers (n=1488) Consumed by the famnily 43.5 27.4 78.4 Sold for cash 35.6 49.8 -- Bartered 0.7 0.8 0.4 Gifts, payments in kind, other uses 3.0 2.3 5.2 Remains in the farmn 17.2 19.8 15.9 Total output 100.0 100.0 100.0 Source: Special Rural Household Survey 2000. 2.61 Farm income in Table 2.31 above is cash farm income from sales. It is based on sales revenues net of all production costs. A certain part of the farm output, however, is not sold: some farm products are consumed by the family, some remain in storage on the farm. These unsold components are accounted for in the total production costs reported by the respondents, yet their value is not reflected in any way in sales revenue. To estimate the value of unsold production and its contribution to total family income, we need to analyze the distribution of farm production between sales and other uses. This analysis is presented in Table 2.31, which shows that farmers reporting sales revenue on average sell one-half of their output. This means that reported sales of 100 zloty should be grossed up to 200 zloty to account for the full value of production, including unsold product. In a more conservative vein, we may allow only for the value of products consumed by the family, which represent one-quarter of output, or one-half of sales. The value of own consumption for the "nonsellers" is calculated from their land holdings, using the coefficients of the regression model that estimates the output as a function of land for the "sellers." Chapter 2: Rural Land Market 59 Table 2.32: Imputed Family Income including Cash Income and Value of Own Consumption of Farm Products Farming families Sell Don't sell (n=1515) (n=1058+7) (n=457) Salaries 6,489 KRUS 3,139 ZUS 2,968 Other personal income 987 Total personal income 13,583 12,629 15,799 Nonfarm business income 1,661 2,251 466 Fann sales 21,513 30,641 0 Farm costs (inputs + nonfamily labor) -12,636 -17,024 -2,490 Farm cash income (net of cost of purchased inputs and payments to nonfamily labor) 8,877 13,617 -2,490 Cash family income 24,121 28,497 13,776 Value of own consumption (27-50% of sales) 7,913-14,655 8,273-15,321 7,194-13,323 Family income including own cons. 32,034-38,776 36,770-43,818 20,970-27,099 Average family size 4.4 4.6 4.0 Source: Special Rural Household Survey 2000. 2.62 Table 2.32 presents the structure of imputed family income, which Figure 2.14: Structure of Imputed Income: Farming includes the cash income from Table Families 2.30 plus an adjustment for the value of Personal own consumption calculated as a 20% percentage of sales (based on two Business Salaries alternatives). The imputed income of 5% 18% farming families is about 10,000 zloty higher than their cash income on account of the value of own production Farm consumed by the family, which 25 m represents one-third of total family income (Figure 2.14). Cash revenue from the farm accounts for 37 percent Value own products of cash family income (Figure 2.13). 32% The total contribution of the farm to Average family income: 24,121 zloty cash + 11,284 zloty family income, including cash revenue plus the value of own consumption, Figure 2.15 reaches 57 percent-a much higher Structure of Family Income: Book-Keeping Farms percentage of a substantially greater amount. Land and farming thus make a very significant contribution to the welfare Farm income of rural families in Poland. 69.0% / Farming and Family Income: Evidence from Other Sources \-* p _ Other off-farm 2.63 Existing information indicates that 8.5% Polish farmers rely on fairly diversified Pension 1 10.8% Salarnes 11.7% Chapter 2: Rural Land Mark-et 60 sources of income that supplement the traditional income from farming. Because of time constraints it has been impossible to analyze the family income data from the recent Household Budget Survey, but the mission managed to obtained some preliminary results from the Institute of Agricultural and Food Economics, which supervises an on-going research project based on a sample of 1,300 farmers who keep full accounts in accordance with a confidential private agreement. 2.64 In this sample of 1,300 "book- Figure 2.16 keeping" farmers, farm income Structure of Family Income by Farm Size: Book-Keeping Farms represents 70 percent of family income, while salaries account for 12 100% percent, pensions for 10 percent, and other off-farm income for 8 percent 80% - (Figure 2.15). The value of farmn 60 products consumed by the family is at 60% . i a level of about 10 percent of total 40% _ . family income (this value is included in farm income). 20% _ - 1l 0% Li - 11 L|||L 2.65 The proportions of farm and 1-2 2-5 5-7 7-10 10-15 15-20 20-50 50-100 >100 off-farm income change dramatically 103Farm Income cOff-Farm with farn size. Overall, the share of farm income increases in larger farms (Figure 2.16). Off-farm income accounts for more than 50 percent of family income in farms that have between 1 hectare and 10 hectare of land. The situation is reversed in households with more than 10 hectares, where farm income exceeds 60 percent of the total. In farms with more than 20 hectares, farm income accounts for more than 80 percent of family income, while off- farm sources make up less than 20 percent. These findings observed in the sample of "book- keeping" farms are supported by the results of the Household Budget Survey, where the respondents who report farming as the main source of family income have, on average, 12 hectares of land per household; whereas, respondents with a large component of off-farm income have, on average, less than 4 hectares of land. 2.66 There is a clear increase in total Figure 2.17 family income with the increase of Family Income by Farm Size: Book-Keeping Farms farm size (Figure 2.17). The average family income from all sources in thousand PZL farms with more than 50 hectares is 140 2.5 times the average income in 120 families with 1-2 hectares. This is 100 ___ entirely attributable to the increase in 80 -- - Farm the farm income component for larger 60 - -- , - _ IOff-farm farms, as the off-farm income 40 -- - -- component for farms with more than 20__I 50 hectares is only one-half of the off- 0 farm income in farms with 1-2 1-2 2-5 5-7 7-10 10-15 15-20 20-50 50-100 -100 hectares. Larger farms thus achieve a fammsize,ha Chapter 2: Rural Land Market 61 higher level of total family income while, at the same time, relying less on income from off-farm sources. Measures to Achieve a Fully Functioning Land Market 2.67 Our analysis highlights the current features of land ownership, land use, and land markets in Poland. The comparison with farm structures and land use patterns in developed market economies shows that Poland has a fairly fragmented structure of land ownership and use: a large number of very small farms operate on a relatively large proportion of the available agricultural land. Although there is a positive correlation between the amount of land used by the household and the level of family income and well-being, consolidation is taking place very slowly, and it has not achieved the scope that would produce significant results in the foreseeable future. Overall, the change in farming structure is relatively slow, and there is only limited activity in land markets. 2.68 This situation is the result of a number of factors. First, the current status of Polish agriculture reflects the heritage of the socialist period. Unlike other countries in the region, Poland did not introduce sweeping collectivization after World War II, and individual farms operating on privately owned land continuously dominated its agriculture. However, during the Soviet era, private farming in Poland did not experience the growth and consolidation that marked private farming in market economies. This was due to restrictive government policies, which limited farm sizes and the operation of farms, and constrained the proper functioning of land markets. As a result, Polish agriculture, despite private land ownership and individual farming, remained much less efficient than market agriculture, and, in fact, it was less efficient than even the collective agriculture in certain Central European countries, such as the German Democratic Republic, Czechoslovakia, or Hungary. 2.69 Due to the private nature of Polish agriculture, this sector received much less attention than the other state-dominated sectors at the beginning of the political transition in the early 1990s. In the mid-1990s, the opening of domestic agricultural markets to imports and the plummeting prices of agricultural commodities served to highlight the low efficiency and limited competitiveness of Polish agriculture. Unfortunately, because of the frequent changes of government in the 1990s and the strong political clout of the large farming population, which was primarily guided by short-term considerations, Poland was not ready or able to implement policies that would induce the far-reaching changes necessary for rationalizing and accelerating the consolidation of land and farming structures. This situation froze the fragmented structure of the private Polish farms without allowing adjustment, and land relations became a major liability for the country. 2.70 The current farming structure, characterized by a very large number of very small farms, is also a potential obstacle in the process of EU accession. The activation of land markets and the acceleration of the ongoing process of land consolidation in medium-sized farms is an overdue necessity, which should receive a high priority in the process of preparing for EU membership. Farm consolidation obviously cannot be separated from the complex nature of Polish agricultural strategy and policies, yet EU accession determines Poland's choices in this regard. It is already clear that the current structure of farming and the political interests that Chapter 2: Rural Land Market 62 support it lead to conflicts with the rational policy choices required both by the globalizing world economy and by the demands of EU accession. In our conclusions, we do not intend to cover the full complexity of contemporary agricultural policies in Poland, but rather focus more narrowly on farm consolidation and land market issues. 2.71 Poland has one of the best legal bases for private land ownership and agricultural land markets among all the transition economies. Polish citizens may own land and carry out the fuill range of market transactions with their land. Registration and mortgage laws exist and function well. Land use rules are also generally reasonable. However, certain basic problems persist. These include the delays in privatizing some of the former state farmland, the uncertainty surrounding pending restitution issues, the cumbersome land registration procedures, the high level of transaction costs for land sales, and issues related to restrictions on foreign ownership of agricultural land. The government needs to address these problems by completing the creation of a legal and institutional environment necessary for effective land markets. 2.72 Actions that would contribute to improving the functioning of land markets include the following: * Reducing the transaction costs on land sales. As indicated above, the cost of land transactions is fairly high, especially when compared to the advanced market economies. The total cost of a land transaction may be as high asl2.5 percent of the sales value, with the 5 percent treasury land transaction tax being the largest single component. This is a significant barrier to entry. Given the current conditions of Polish agriculture, where land consolidation would greatly contribute to improving returns on farming, reducing land transaction costs and simplifying the tax and fee structure should be a priority. Options include eliminating the 5 percent treasury tax on land transactions and setting an upper limit on notary fees. The latter could be achieved by setting a ceiling on the current percentage-based fee for land transactions. * Reviewing the land ownership registration system to streamline and simplify the cumbersome and time-consuming procedures. Although the land registration law is basically sufficient, the treatment of "unregistered rights" requires attention. Estimates indicate that 30 percent to 40 percent of agricultural land rights are not registered to reflect the current ownership. The registration procedures are very time-consuming, and the system is seriously under-funded, with a lack of adequate personnel resources. * Improving access to land mortgage facilities by streamlining and reducing the restrictions on using land as loan collateral by, for instance, increasing the percentage of the value of real estate mortgaged from the current 60 percent to a range of at least 70 percent to 80 percent. 2.73 In addition to ensuring that market forces can function as the primary engine for land consolidation, it would be useful to establish a system for monitoring and reporting all land sales. This would increase the transparency of land markets and improve information flows. CHAPTER 3: RURAL CREDIT MARKET Introduction 3.1 In the past decade, Poland as a whole has experienced outstanding economic growth, but this growth has not extended to the rural sector. There are several possible reasons for this relative under-performance. This chapter is part of this study's overall effort to understand one possible contributor to this problem, a poorly functioning rural credit market. Previous studies have found that Polish farmers borrow little from formal lenders, suggesting that problems in the credit market could be a serious bar to the modernization of farming and the creation of jobs in the nonagricultural rural sector. The credit market is addressed here both through a survey and through interviews in Poland. The interviews are not used in any systematic way here, but provide valuable information that appears indirectly in what follows. In many cases an infornant was able to clarify some aspect of a law or practice that affected rural financial markets. More often, informants offered explanations of patterns or practices that provided a starting point in analyzing the survey. 3.2 This chapter considers both farm and nonfarm rural activities. About 40 percent of the Polish population lives in rural communes, and about one-half of these do not engage in farming. Raising rural incomes requires increasing incomes in farming and also the strengthening of the nonfarm rural sector. For this reason the survey asks detailed questions about farm and nonfarm activities. The survey data come from four voivodships that were selected because they are different and help to illustrate heterogeneity within the Polish countryside. The analysis underlying this chapter pays careful attention to variations by voivodship, but only reports such differences when they are significant. 3.3 This chapter considers two general explanations for the low levels of borrowing among rural Polish households.'7 The first explanation is the "uncertainty hypothesis." Before households can borrow, either for consumption or for investment, they have to be convinced that the economic environment is sufficiently stable and predictable that they will be able to repay the loan and, in the case of an investment, do so from the proceeds of the investment project. Several factors are currently combining to create tremendous uncertainty in rural Poland. This uncertainty alone could lead to low levels of borrowing, even in the best of capital markets. The second explanation is the "credit market hypothesis." Imperfections in the banking system and other parts of a country's credit market can lead to low levels of borrowing, even when there are firms and households that are willing and able to borrow at reasonable rates. Poland's banking system is undergoing rapid change, and one might think that a banking system that has not yet made the full transition to a market economy might pose problems for rural Polish households. Neither the survey nor the situation in Poland today support straightforward, clean tests of either the uncertainty hypothesis or the credit market hypothesis. The two hypotheses are in any case not mutually exclusive. But they are useful organizing devices and this chapter is organized around demonstrating that the available evidence is more consistent with the uncertainty hypothesis than with the credit market hypothesis, although there is also evidence of problems in the credit market. 17 This chapter was prepared by Timothy W. Guinnane. Chapter 3: Rural Credit Market 64 3.4 The analysis reported here was constrained by the survey's nature and results. These constraints are simply functions of the survey's sample size and the actual situation in Poland; they do not reflect any weaknesses in the design or execution of the surveyper se. There are two types of constraints. First, with 2,835 households the sample size is more than adequate for most purposes, but borrowing is so rare that, in some cases, the effective sample size is fewer than 400 households. Thus, when examining the characteristics of households that took a specific kind of loan we quickly run out of observations, because so few households that did so are in the sample. To contend with this type of constraint, we use simple regression models that should be viewed as a fine-grained descriptive device that have the virtue of conserving on data. They are not intended as structural models of any market or household decision. A second constraint may be more serious, but again just reflects Polish reality. Some questions are simply too sensitive to ask. For example, some informnants in Poland told me that they are asked to pay certain fees associated with loan applications when the applications are, in theory, free. The survey rightly did not ask about this for fear of engendering fears of reprisals among informants. A second and related example concerns loans to and from friends and relatives. There is very little reported borrowing or lending of this sort. This type of informal credit activity is pervasive in most countries, and the apparently low levels of such credit relations might simply reflect the unwillingness of households to report such loans. Here, as in other potentially doubtful matters, we have no choice but to take the data at face value, and we do so throughout the remainder of the chapter. These caveats aside, the survey contains a great deal of information and supports some conclusions about the reasons for the low levels of borrowing that we observe in rural Poland. 3.5 In much of the analysis to follow, we find that in many cases the largest effect on a certain outcome variable is the voivodship in which a household is located. This is not completely surprising, as location can itself be correlated with characteristics of local institutions and market. The voivodship effects may also reflect other important variables (such as local living costs) that either are not in the survey or are only imperfectly captured by the survey. However, some brief comments on the four voivodships may explain later findings. Mazowieckie is the omitted voivodship in all the regression models. This district is centered on Warsaw and includes many of Poland's most economically developed areas. Malopolskie is centered on Krakow and shares some features of that relatively depressed region; incomes are low by Polish standards, although the workforce is quite well-educated. Wielkopolskie has unemployment levels that are among the lowest in rural Poland, while Zachodniopomorskie, the fourth voivodship, has very high rates of unemployment. 3.6 The chapter is organized as follows. It begins by outlining two kinds of reasons for the very low levels of borrowing. Next, it provides some basic institutional background on the main lending institutions for rural people. The principal purpose of this section is to identify the lenders associated with the credit discussed later in the chapter. Indeed, the main findings of the chapter are presented in the subsequent three sections. First, there is a review of the evidence on savings and savings accounts, followed by a discussion of the credit market in two steps. The first step is an overview, using the survey, of the prevalence and characteristics of debt among rural households. The second step consists of looking further into the issue of why so few households borrow. The final section summarizes the main conclusions and the policy implications. Chapter 3: Rural Credit Market 65 Why Don't Rural Polish People Borrow? 3.7 Rural Polish households report very low levels of involvement with financial institutions of any kind. Less than one-quarter of survey households report any financial savings, and, of those, only three-quarters have their savings deposited in any type of bank. The survey data for credit is equally striking. Only 30 percent of households contained a person who applied for a cash loan in 1999, and slightly more than 30 percent of all households had at least one member with outstanding cash credit during 1999. About 14 percent of households purchased something on credit during 1999, and fewer than 5 percent of survey households had both a cash loan and a purchase on credit during 1999. This is not the first time this fact has been noticed; Klank (no date, Table 3.7) reports that, in 1997, over one-half of all farms of less than 10 hectares reported no debt at all. Most reported debt seems to be short-term debt, rather than the long-term investment credit one would expect if farmers were modernizing and rationalizing their operations. The uncertainty hypothesis 3.8 There are two possible reasons why rural people in Poland borrow so little. The first explanation we will call the uncertainty hypothesis. This hypothesis is the favored explanation of most of the farmers interviewed in both Bocki and Modliszewice. They say that demand for credit is low, that at any reasonable interest rate there are few farmers or others who feel they have investment opportunities that justify taking loans. This kind of claim is often advanced when discussing rural credit markets. Often the claim is just a way of dismissing the credit- markets hypothesis discussed next-that is, someone who believes that markets always work well may assume away any explanation that is based on a market imperfection. Here, however, an underlying unwillingness to borrow should be taken seriously. Several farmers said there is too much uncertainty now for them to know what kind of investments to make. Most who made this argument mentioned the European Union in particular, noting that Poland's accession is not yet fixed and that the EU is considering significant changes to its agriculture policy.: 3.9 Farmers in both Bocki and Modliszewice gave concrete examples of this concern about the future. Some farmers in each region had been trying to expand their pig production, but others (out of their hearing) registered worry about the entire idea of raising pigs in Poland. Their fear is that pork production may not be profitable under new EU regulations. The survey supports this concern as a reason for low debt levels. When asked the main reason for not requesting a loan in the past 12 months, 21 percent pointed to "risk because of unstable income." Another 19 percent of households responded that they "prefer to work with own resources" which might also reflect concerns over risk.18 3.10 A variant on this argument implies that Polish farmers cannot profitably borrow until the state has made infrastructure improvements. Infrastructure is beyond the scope of this chapter, except to note that if infrastructure is complementary to an investment, then poor infrastructure might impede that investment. Consider, for instance, the example that comes from the region of 18 There are other sources of risk, and some farmers noted the recent collapse of demand from Russia as a reason not to trust prices. Chapter 3: Rural Credit Alaret 66 Bocki. Several farmers interviewed, as well as the head of the Gmina, noted that the area would be an excellent one for "agro-tourism."'9 The few farmers who have tried to run this kind of business have found that tounrsts are discouraged from coming to the area by poor roads and by recurrent pollution problems associated with poor sanitation facilities. A near-consensus view among the Bocki informants is that, if the State improves their infrastructure, they can build up a significant tourism business, with both Polish and foreign tourists, and that credit will not be an obstacle to developing that business. Several noted that there is significant interest in this kind of tounrsm in both Germany and the Netherlands and that poor roads and sanitation facilities impede the development of that market. 3.11 Thus, we have good reason to take seriously the notion that rural households do not borrow simply because they are not sufficiently certain of being able to repay the loans. Some point to a lack of complementary social investments, but the more common refrain was uncertainty about the future. The credit-market hypothesis 3.12 The other explanations for low levels of borrowing in rural Poland point to problems in the functioning of the credit markets. The other explanations include credit rationing and imply that rural people do want to borrow, but find either that (i) banks and other potential lenders will not lend to them, or will not lend as much as borrowers want, or (ii) that the costs of borrowing (both direct costs such as interest and fees, and the indirect costs including paperwork, travel to the lender's office, etc) are too high.20 The survey gives direct evidence that both problems have some force in Poland. About one-half of all borrowers say that they wanted a larger loan on the same terms as the loan they received,i` which is a simple definition of credit rationing. The cost of credit was a big reason why many never applied for a loan in the past year. About 23 percent of households that did not apply for a cash loan in 1999 give their main reason for not applying as "high interest rates and fees." 3.13 Several infonnants complained of high indirect costs, and this potential problem should be borne in mind when discussing the results. The most commonly-cited problem is the so- called "business plan" required by ARMA (Agency for the Restructuring and Modernization of Agriculture) and other lenders. Nobody seriously questioned the need for such plans. Several fanners complained, however, about the costs of providing these business plans. They claimed that the plans were so complicated that they had to pay someone to prepare them on their behalf. This claim is contradicted by both ARMA and the representatives of the extension services, who say that a big part of their job is preparing such plans and that the service is free. This 9 The term "agro-tourism" means that farmers run a bed-and-breakfast business on their farms. Some claimed that it ordinarily involves the tourist taking part in farm work, while others claimed that it did not. Some noted that the most popular agro-tourism involved horse-back riding and related rural pursuits. 20 The first problem is, strictly speaking. credit rationing, while the second is not. Ray (1998, Chapter 14) provides an excellent overview of the economics of credit markets. 21 This figure is for cash loans and pertains to the first loan. There are very few households with more than one cash loan, but for these the figure mentioned is lower, suggesting that one way households deal with credit rationing is to take multiple loans. Chapter 3: Rural Credit Market 67 discrepancy of claims may reflect different types of loan programs. When asked specifically about other indirect costs, few of the informants seemed bothered by the distance to the bank. 3.14 Rather than present any formal tests of either the uncertainty or the credit-markets hypothesis, this chapter simply works through the survey evidence to outline information that is more or less consistent with each view. Economists have developed formal tests of problems in credit markets, but these tests are very severe and, in any case, require an analytical structure that is itself controversial.22 Financial Institutions for Rural Polish People 3.15 This section outlines the primary savings and lending institutions for rural Polish people. The discussion here is intended to provide only a minimum of context, since much of the banking system in Poland is in flux, and this is particularly true of the cooperative banks. During the course of interviews it became apparent that some cooperative leaders are trying to chart their own course, and this potentially interesting development receives more attention below. 3.16 For our purposes we can divide the Polish banking system into four parts. The cooperative group includes the Bank for Food Economy (BGZ), which is the apex bank in that system;23 several regional cooperative banks; and the local cooperative banks. (At the end of 1999, there were 11 regional cooperative banks and 781 local cooperative banks, but both of those numbers have declined in 2000 and will decline further). The cooperative group is clearly the most important part of the entire banking system for rural people. The local banks are on average quite small, with collectively only 4.2 percent of all Polish banking assets at the end of 1999 (National Bank of Poland, 2000). The local cooperative banks' share in net loans and deposits taken from the public is very similar. Most cannot meet the minimum capitalization of 5 million Euros required of a new bank under Polish and EU banking law. Many local cooperative banks probably cannot meet the lower standard of 300,000 Euros required for local cooperative banks. The number of local cooperative banks declined by 408 during 1999, in almost all cases by merger with another local cooperative bank. 22 Section 5 reports a variant on these tests, but stresses their limnitations in general and with this survey. Hubbard (1988) surveys the relevant literature. 23 The Polish cooperative system has a structure roughly equivalent to France's Credit Agricole. 24 This estimate was provided by the Central Bank's General Directorate of Bank supervision. Chapter 3: Rural Credit Market 68 3.17 The second major institution in rural Polish lending is not a bank, but a government agency that offers loan subsidies for rural lending undertaken by commercial banks. ARMA runs several programs targeted to different clients, but they have a similar method of operation. For loans that ARMA approves, a commercial bank advances credit on its own risk, but with a hefty interest subsidy provided by ARMA. In an environment in which unsubsidized commercial loans can cost as much as 25 percent (nominal) per year, ARMA loan rates can be as low as 4.5 percent (nominal), depending on the program. Although ARMA runs several different programs, they all have the stated purpose of setting up a new farmer or of encouraging the creation of a new small enterprise or the enlargement of an existing enterprise. ARMA's lending guidelines peg the amount of credit to the number of new, permanent jobs created. ARMA has been a large lender in the relevant markets over the past few years, but the renewal of these programs is now under discussion. In the survey, ARMA does not appear as a major lender because most of its programs require the participation of another institution, such as the BGZ, a local cooperative bank, or a commercial bank. 3.18 A third important institution is the Polish State Savings Bank (PKO BP). The background materials present the PKO BP as a primarily urban institution, but as we shall see that it plays an important role in rural Poland's savings and credit markets. This penetration reflects, in part, the large number of offices and school savings agencies where the PKO BP collects deposits. 3.19 The final set of institutions providing credit are commercial banks and nonbank entities that provide credit to their suppliers or consumers. Commercial banks appear to be increasingly involved in making loans to large agricultural enterprises and rural agri-businesses. In an interview, the head of a commercial bank reported that margins on such loans have fallen in the past two years as a result of growing competition from other banks. Other credit sources are very large firms, such as Monsanto (for inputs) and supermarkets (for outputs) that provide credit as part of a larger business relationship. This appears to be a very important source of credit for the largest 12 percent of farms in Poland. Examples include one dairy farmer in Bocki that had purchased his milk-storage equipment from the French manufacturer on credit. A pig farmer in that same region had an arrangement by which he agreed to deliver his entire output to a single slaughterhouse in exchange for feed on low-cost credit. Savings 3.20 The survey suggests that very few households have any savings at all. A low level of savings is not surprising given the low incomes and recent shocks to the rural sector. What is surprising is the one-quarter of households with savings who do not have bank accounts. There are several possible reasons for this finding. Several informants claimed that older people do not trust financial institutions, and the survey supports this claim. For purposes of this chapter we define an older household head as someone who is 60 or more years of age.> Among older 25 The definition is arbitrary, but the results reported here and below are robust to small changes in the definition. Chapter 3: Rural Credit Market 69 household heads, only 60 percent of those with cash savings have a deposit account. For households with younger heads, the figure is 82 percent. The survey asks the main reason for not having savings in a financial institution. Only 90 households responded, and the most common reason is "other" (53 percent of respondents), so a detailed analysis is impossible.26 3.21 Table 3.1 lists the number of accounts at each of the main savings institutions, along with the average distance between the household and their branch of that institution. The two most popular institutions for savings are the PKO BP and the local cooperative banks. The average distance between the household and the depository institution is about 15 kilometers for all institutions except the local cooperatives, for which it is about 5 kilometers. The consolidation of cooperatives will presumably increase the average distance to depository institutions unless the remaining cooperatives take steps to provide for offices in the places where old cooperatives existed. In any case, the popularity of the relatively distant PKO BP does not suggest that distance to the institution is an important deterrent to holding bank accounts. Table 3.1: Number of Accounts and Distance to Institutions, Savers with Deposit Accounts Institution Number of savers Mean distance (km) BGZ 24 14.6 PKO BP 92 16.1 Private banks 21 14.3 Cooperative banks 75 5.0 Other 13 16.3 Source: Special Rural Household Survey 2001. 3.22 Table 3.2 breaks down the distance-to-bank infornation by voivodship for the two most important depository institutions, the PKO BP and the local cooperative banks. The table displays an important pattern that will reappear, in various forms, throughout this chapter: households that hold PKO BP accounts in the voivodship of Malopolskie, as the table shows, are farther from the PKO BP than are households in other voivodships. The table pertains to households with PKO BP accounts. The survey does not ask about distance for those who do not have accounts. It is possible that the average household in Malopolskie is just as far from a PKO BP office as in the other three voivodships, there is no way to test the assertion. It seems more plausible that Malopolskie is just under-served by at least this bank. We will see in later sections that, in general, households in this voivodship have less contact with the formal financial sector than those in other voivodships, even when we control for other factors.27 26 A hint of a different explanation comes from the data on nonfarm businesses. About one-third of people who run such businesses claim not to have a bank account of any kind. The most common reason stated for not having an account (besides "other") is that everyone wants cash, not checks. To the extent that this problem applies in other contexts, it implies that bank accounts do not provide effective payment services, reducing their usefulness. Payment services are probably more important for business accounts, m any case. 27 Households in Malopolskie are only slightly less likely to report cash savings than households in the other voivodship (10.4 percent versus 12.8 percent), but Malopolskie households are much more likely to refuse to answer the question about cash savings (11.6 percent versus 6.9 percent of households). It is possible that Malopolskie residents claim not to have bank accounts to avoid saying whether they have cash savings. Chapter 3. Rural Credit Market 70 Table 3.2: Mean Distance to Bank, PKO BP and Local Cooperatives, by Voivodship Bank Average distance to bank (number of accounts) Malopolskie Mazowieckie Wielkopolskie Zachodniopomorskie PKO BP 22.8 14.8 8.4 15.9 (26) (17) (19) (30) Local cooperatives 5.0 5.1 4.9 5.9 (13) (17) (38) (7) Source: Special Rural Household Survey 2001. 3.23 Table 3.3 reports a simple binary probit model that reveals some interesting pattems. The dependent variable equals 1 if the households reports that they have a bank account of some kind, and the estimation sub-sample is limited to households that report cash savings. The age quadratic peaks at about 45, showing that the reluctance of older people to have accounts is robust to the inclusion of other factors. Bank accounts are much less important in the voivodship of Malopolskie than in the other three voivodships.28 These voivodship effects are quite large. Evaluated at the means, a household in any of the three other voivodships is about 15 percent more likely to have a bank account than a household in Malopolskie. This large effect will appear again in much of the later analysis of borrowing. Table 3.3: Binary Probit Estimates of Households with Savings Accounts Probit estimates Numnber of observations = 402 LR chi2(9) 67.18 Prob > chi2 = 0.0000 Log likelihood = -207.95292 Pseudo R2 = 0.1391 Explanatory Standard 95% variables Coefficient Error z P>Fz Confidence Interval Age of the Head of Household .0637543 .0338017 1.886 0.059 -.0024958 .1300044 Squared Age of the Head of Household -.0007067 .0003154 -2.241 0.025 -.0013249 -.0000885 Head of Household is Male .4066468 .1746134 2.329 0.020 .0644108 .7488827 Second Head of Household -.8632833 .1751886 -4.928 0.000 -1.206647 -.5199201 Mazowieckie .5201182 .2075121 2.506 0.012 .1134019 .9268345 Wielkopolskie .4892828 .1972496 2.481 0.013 .1026807 .8758849 Zachodniopomorski .5324683 .1922592 2.770 0.006 .1556472 .9092894 Constant -.8375066 .8852622 -0.946 0.344 -2.572589 .8975754 Source: Author's calculations. 28 Throughout this report Malopolskie is the omitted voivodship. Thus inferences about its distinctiveness are drawn from the size of the other voivodship coefficients. Chapter 3: Rural Credit Market 71 3.24 The regression reported in Table 3.3 also . ~ * two ways in which access to the financial system is constrained by household charactelistics The regression results indicate that those with at least a secondary education are much more likely to have a bank account. (The variable is defined as 1 for households where the headl has less than a secondary education.) This is again a very large effect, with a 25 percent difference evaluated at the means. If we have controlled successfully for the presumably higher incoimes of such households, then the result implies that less-educated people are ignorant of banks, mistreated by banks, or simply do not trust banks with their money. The regression results also indicate that households headed by males are much more likely to have a bank account than other households with cash savings. It is not possible to say whether this reflects the attittudes of these households, treatment by banks, or some factor which we have not controlled. 3.25 The regression does not suggest that household income levels have a strong relationship to the probability of the household having a deposit account. One might think that households with greater incomes would have more need of payment services, and, if there are any fixed costs to opening and maintaining an account, those costs would be more worthwhile for higher-income households. This does not seem to be the case, however. Overview of Household Borrowing 3.26 The survey treats credit in two categories, cash loanis and purchases on credit, and shows that household borrowing is not very common. Table 3 4 details responses to the question "What is the main reason for not requesting a loan in the past 12 months?" Several different responses are consistent with the view that households are unwilling to borrow because of the lack of a stable economic environment. Many households say they prefer to work with their own resources. Whether this reflects uncertainty, loan costs, or other reasons is unclear. Loan costs and lack of collateral are two other important reasons. One big surpnrse in Table 3.3 is variation across voivodships in the percentage of households that cite loan costs as a reason for not applying for credit. Households in Malopoiskie are more than twice as likely as those in Zachodniopomorskie to cite high costs as a reason for not applying for a loan. This claim is somewhat puzzling at first glance. Programs such as ARMA's subsidized loans provide loan rates that are nearly equal across the country. The cooperative banks, for example, share resources within the cooperative system, so it is hard to believe that households in Malopolskie face higher lending rates than households elsewhere. The perception of higher costs in Malopolskie might reflect indirect costs such as distances from banks. This would square with my earlier findings on savings. Another feature of Table 3.4 is hard to square with that interpretation, however. Malopolskie respondents seem confident that they would qualify for loans if they wanted them. This is a different matter from indirect loan costs, of course, but, if the main cost is distance from banks and lack of engagement with the financial system, it is hard to understand why these households are so confident that they would qualify for credit. Chapter 3: Rural Credit Market 72 Table 3.4: Reasons for Not Requesting a Loan in the Past 12 Months Reason Percentage of households, by voivodship Malopolskie Mazoweckie Wielkopolskie Zachod. All Loan would not be approved .92 5.67 2.44 9.21 4.6 Lack of collateral 5.55 5.04 6.87 9.96 6.9 Lack of co-signer .55 .42 .22 .94 .55 Highinterestrates/fees 32.16 22.27 21.95 15.6 23.1 Procedure too time-consuming .18 .84 .67 .19 .45 Does not know how to apply for loan .18 .42 .22 .56 .35 Fear of losing collateral .92 .42 .44 .56 .6 Credit is risky because of unstable 19.04 19.33 22.17 22.93 20.85 income Prefer to work with own resources 19.41 19.75 23.28 15.98 19.45 Do not have investment opportunity .92 .84 .22 0.0 0.5 Do not need loan; work with own resources 12.75 16.18 11.09 14.85 13.75 Do not need loan; enough debt already 1.66 1.05 .22 2.63 1.45 Other 5.73 7.77 10.2 6.58 7.45 Number of households that answered 541 476 451 532 2000 the question Note: percentages add up to 100 within each colunn. Soyrce: Special Rural Household Survey 2001. o3.27 Table 3.5 details information on loan applications in 1999. The survey confirms that the local cooperative banks are the largest provider of cash loans by a wide margin, followed by the PKO BP, private banks, and the BGZ. Friends and relatives by this account stand as minor lenders. The second column does not tell us how many loan applications were turned down for each lender, but, if we assume that there was only one application denied per lender per household, we find, for example, that the local cooperatives approved about 86 percent of applications. We cannot tell, unfortunately, whether the loans denied by (for example) the credit cooperatives were loans sought in addition to those approved by another lender. Some lenders were much more likely than others to grant two loans to the same household in 1999. The leader in this regard was the credit cooperatives. Fully three-quarters of households with a second loan from a cooperative also received their first loan there. Two-thirds of households with their third loan from a cooperative also received their first and second loan there. Chapter 3: Rural Credit Market 73 Table 3.5: Number of Loans Approved and Denied by Lender Lender Number of households that were tumed Number of households that received a loan down by this lender in 1999 from this lender in 1999: Loan 1 Loan 2 Loan 3 Loan 4 BGZ 22 63 19 4 1 PKO BP 37 107 23 5 2 Private banks 17 70 14 10 4 Kasa Budowlana Not asked 1 0 0 0 Co-operative 88 414 94 18 5 banks NGOs 2 6 0 0 0 Friends and 21 40 19 10 5 relatives Moneylenders Not asked 1 0 0 0 Other lenders 23 139 53 18 3 Source: Special Rural Household Survey 2001. 3.28 Table 3.6 provides an overview of cash loan characteristics for the four largest lenders (by number of loans). In many respects, the lenders have similar characteristics, or differ in small ways that are consistent with basic differences in their operations. For example, there are so many cooperatives that it is not surprising that they are closer to their borrowers. The relatively large size of BGZ loans, and their high degree of collateralization, just reflects its lending programs. The same can be said for the interest-rate differences (including the fractions that are subsidized). One surprise in Table 3.6, given some of the comments made by farmers and cooperative leaders, is the relatively small differential in the length of time it takes to get a loan approved. While some of the people interviewed described BGZ as bureaucratic and inflexible, a 2-day mean difference in the loan approval rate does not support that view. The difference in medians between the BGZ and the credit cooperatives here is 3 days, so the comparison here is not driven by a few very large loans. 3.29 Given our earlier discussion of Malopolskie and distance from banking offices, it is worth breaking down the distance-to-lender data in Table 3.6 by voivodship. Again, this effort is limited by the small sample size, but, for the two largest lenders, the PKO BP and the local cooperatives, we have enough observations to draw some conclusions. The PKO BP's borrowers in Malopolskie are indeed a good deal farther from their lender (about 5 kilometers in each case) than its borrowers in either Mazowieckie or Wielkopolskie, but, at least by this measure, the Malopolskie PKO BP borrowers are no more inconvenienced than are those in Zachodniopomorskie. The local cooperative banks' borrowers are about the same distance from their lender, on average, in all four of the voivodships. Chapter 3: Rural Credit Market 74 Table 3.6: Selected Characteristics of Loans granted by Major Lenders in 1999 Characteristic BGZ PKO BP Private banks Local cooperatives Mean distance to lender (in km) 16.2 16.4 19.8 6.9 Mean number of trips required to get loan 3.7 2.7 2.8 2.6 Mean number of days required to get loan 13.1 10.9 11.5 10.9 Percent of loans disbursed as cash only 96.8 97.2 97.1 99.8 Mean size of loans disbursed as cash only ('000 PLN) 21.8 13.5 8.7 8.3 Percent of loans with collateral 88.9 61.7 64.3 85.5 Value of collateral ('000 PLN) 58.9 33.7 19.8 24.1 Mean loan-to-value ratio' .80 1.15 .96 1.10 Mean number of months repayment term 113.4 144.5 155.5 112.8 Percent with a fixed interest rate2 46.7 47.0 56.2 64.1 Mean interest rate when loan received 13.7 19.8 17.7 12.3 Percent of loans with subsidized interest rate2 64.8 22.6 27.3 72.6 Percent of loans "rationed" 58.1 33.6 38.6 54.0 Percent of loans with payments more than 30 days late 7.9 8.6 7.2 6.8 Percent of loans rescheduled or overdue 1.6 0 2.9 2.7 Note: This table is limited to "Loan 1" characteristics. 1) Value of loan/value of collateral 2) Limnited to respondents who knew Source: Special Rural Household Survey 2001. 3.30 Table 3.6 contains two surprises. One is the very high loan-to-value ratios for these loans. BGZ's average of 80 percent is already quite high, and the other lenders are, according to the survey, accepting as collateral property that is worth less than the loan.29 There are two possible interpretations of this finding. First, one might wonder whether the survey respondents are reporting the purchase price of an asset but not its current price. That is, the lender might correct land prices for current market prices, yielding "true" loan-to-value ratios less than those reported in Table 3.6. Again, we have no choice but to take the survey at its word. Second, to the extent that these high loan-to-value ratios are accurate, they suggest that collateral values are not a large problem for borrowers who can satisfy other lender requirements. 3.31 The second important finding in Table 3.6 is the percentage of all loans that were "rationed" in the sense that the borrower wanted a larger loan at the terms he was offered, but was denied the larger loan by the lender. The very large fractions of rationed loans reported for the BGZ and the credit cooperatives may just reflect the characteristics of the households that qualify for their lending programs. At a subsidized interest rate we expect some credit rationing; 29 This calculation is restricted to loans for which the respondent provided an estimate of the collateral's value. Chapter 3: Rural Credit Market 75 at a low interest rate many households will want to borrow, while the lender does not have unlimited funds available at this rate. So the BGZ (or any other lender) must either decline some worthy applicants or lend give smaller loans to applicants it approves. 3.32 Loan security depends, not surprisingly, on the lender and the loan's size. About one- half of all loans are granted on the security of a co-signer only. This figure is slightly higher (57 percent) for the cooperatives, lower for the other major lenders. Larger loans are less likely to have just a co-signer; of the 71 loans worth 20,000 PLN or more, only 30 percent have just a co- signer. There are some other surprises in the lending security information. About 25 percent of all loans, respondents say, have no security at all. This is more likely the case for loans from the PKO BP and private banks than from other major lenders, and most common for loans of less than 5,000 PLN. The remaining loans are secured mostly by vehicles, land, dwellings and other real estate, and "other." Lenders appear not to rely on livestock, inventory, or future harvests when making loans. 3.33 Table 3.7 reports the "main" uses of loans, again by lender. The differences here reflect in large measure the different mandates of the various institutions. PKO BP, for example, is charged with lending for household purposes, which shows up in table. Several of BGZ's programs target agriculture, which again accounts for their concentration in this area. Several additional patterns are noteworthy. First, the private banks have not yet become deeply involved in lending for either agriculture or nonagricultural enterprise. However, this appears to be changing, especially with the entry of more foreign banks into the Polish market. Second, there is very little refinancing of other loans. This may reflect lender policy, but it may also be an additional sign that rural Polish households are very conservative about taking on debt. Purchases on credit 3.34 The survey treats purchases on credit separately. Credit taken in this way is even rarer than cash loans; only 15 percent of households have outstanding purchases on credit. According to the survey, the vast majority (69 percent of these households) of these loans were to purchase a household durable. Another 17 percent were for some other household use. The size of the loans is surprisingly large. The mean value of the objects financed this way is about PLN 5,600, with an average down payment of 700 PLN. Interest rates for this type of credit are similar to those for cash loans, with a mean rate of 15.5 percent. One important difference is that 11 percent of these borrowers claim to have an interest rate of zero. A zero interest rate is plausible as part of a marketing program; the Bocki farmer who said he purchased equipment on credit paid no interest costs. 3.35 These loans, just like the cash loans, are nearly all current, and 94 percent of borrowers say they have never been as much as 30 days late with a payment. About 58 percent of these loans, respondents say, have no security whatsoever. (This low figure may reflect some confusion; the item they purchased may be security for the loan.) About 9 percent have a co- signer, and another 29 percent have "other" security of some form. Chapter 3: Rural Credit Market 76 Table 3.7: "Main" Uses of Loans by Major Lender Use BGZ PKO BP Private banks Local cooperatives All loans Agricultural machinery, tools, equipment 5 1 1 24 32 Agricultural inputs, raw materials, livestock 21 3 4 199 230 Agricultural land 3 0 1 16 25 Agriculture-other uses 7 1 0 36 44 Nonag. machinery, tools, equipment 1 2 1 1 10 Nonag. inputs, raw materials, merchandise 0 0 1 12 15 Nonag. other uses 1 1 0 12 15 Household durables 4 9 3 7 38 Household emergency 3 4 5 11 31 Household buy dwelling/home improvement 6 39 28 55 214 Other household uses 4 19 9 26 101 Repay another loan 0 1 0 3 5 Vehicles 3 14 12 8 43 Other 2 13 5 13 47 Note: Limited to "Loan 1." Figures are numbers of loans; final column includes all loans, not just those from the major lenders Source: Special Rural Household Survey 2001. Credit Rationing? 3.36 As mentioned, Polish rural households have little involvement with the formal financial system. Relatively few households report any cash savings, and many of those with savings do not have deposit accounts. Nearly two-thirds of households report taking no credit at all in 1999, neither in cash nor by financing a purchase. The fundamental question for this report is why this is the case-why, especially, do households borrow so little. Above we noted that there are two competing explanations. The first explanation says that households do not borrow because they do not think the economic future in Poland is predictable enough to justify commitments to repaying loans. The second says that problems in the lending market ration credit; households would borrower if costs were lower and other terms less restrictive. Given the survey data, we cannot report a simple test of either view. 3.37 This discussion proceeds in three parts. First we look more closely at the differences between households that did and did not request credit, and the reasons they give for not requesting credit. Second, we look at evidence on investments of various sorts to examine whether and how much households relied on external finance. Third, we examine the dependence of investments on current income. This is a rough test of credit-market imperfections that, in this case, may tell us more about the characteristics of households that have imperfect access to the credit market. Chapter 3: Rural Credit Market 77 3.38 The survey contains clear, if indirect, evidence that the banking system is not the primary problem in low levels of borrowing. If rural Poland was full of households that wanted to save and households that wanted to borrow, but was saddled with a banking system unable to perform its intermediation function very effectively, then we would expect to see much more inter- personal lending than the survey reports. The tight-knit, stable environments of these rural communities are precisely the type of environment where one would expect such lending to work well. The survey does not suggest much credit of this type, however. One possible explanation is that the survey simply does not track loans of this type. This may well be true on the savings side, but it seems much less likely on the borrowing side. Another explanation is that there are few households with enough income to accumulate savings. This is possible, but development economists have found significant savings in households in even the poorest economy. A final explanation, of course, is that offered by several informants: they reject the credit-rationing hypothesis, and instead say that they do not borrow because they do not think they have predictable investment opportunities. This explanation would account for the lack of inter- personal loans and the low levels of debt in general. Who borrows? 3.39 We begin by looking more closely at which households borrow and the stated reasons for not borrowing. The binary probit model reported in Table 3.8 has a dependent variable that has the value 1 if anyone in the household applied for a cash loan during 1999. The basic idea of the regression is to see what characteristics of a household make it likely to apply for credit. The controls are (a) the gender of the household head and the head's education, as in Table 3.3; (b) farmer status; (c) a dummy for whether the household holds any land, (d) a continuous variable that is total hectares owned; (e) province dummies; and (f) total family income. The age quadratic turns at about 39, which is similar to the turning point in the bank-account regressions. The logic of the gender and education controls is similar to their use in the regression for saving accounts. Here they have little effect. Male-headed households and households headed by more educated people are not more likely to apply for credit. The income effect has a strong positive impact; the quadratic term turns at about 350, which is less than the reported value for only a few observations. The strong dependence on income can be viewed in one of two ways. Perhaps low-income households simply feel that they cannot affoid interest payments. On the other hand, low-income households might feel they will be denied credit. 3.40 The voivodship dummies once again suggest that something is wrong in Malopolskie; households there are much less likely to apply for credit, even after we have controlled for other factors, than are households in the other three voivodships. Households in the other three voivodships are more likely to apply for loans (we cannot reject the null that the effects in all three of the other voivodships are the same). The effect is not enormous in this case. The largest difference between Malopolskie and the other areas implies an 1 1-percent difference in probabilities evaluated at the means. This result confirms our earlier impression that households in this area have less contact with the banking system. Chapter 3: Rural Credit Market 78 Table 3.8: Determinants of Households that Requested a Cash Loan in 1999 Probit estimates Number of observations = 2835 LR chi2 (9) 220.79 Prob > chi2 = 0.646 Log likelihood = -1598.3579 Pseudo R2 = 0.0646 Explanatory Standard 95% variable Coefficient Error z P > z Confidence Interval Age of the Head of Household .0610308 .0131575 4.638 0.000 .0352425 .086819 Squared Age of the Head of Household -.0007821 .0001283 -6.098 0.000 -.0010335 -.0005307 Head of Household is Male .0133883 .06548 0.204 0.838 -.1149501 .1417267 Second Head of Household -.0811048 .0644773 -1.258 0.208 -.2074781 .0452684 Mazowieckie .1861775 .0755513 2.464 0.014 .0380997 .3342552 Wielkopolskie .2943012 .0732379 4.018 0.000 .1507575 .4378449 Zachodniopomorski .1499893 .074163 2.022 0.043 .0046326 .295346 Constant -1.708482 .3319302 -5.147 0.000 -2.359053 -1.057911 Source: Author's calculations. 3.41 The direct measures of income do not affect the probability that a household requested credit in 1999. This is not surprising, because loan applications can be for varying amounts. The last three dummy variables show that income relative to perceived household needs does have a very large impact on the decision to ask for a loan. The survey asks respondents to say whether they have enough income to meet their needs and possibly save or buy luxury goods, classifying the answer under 5 categories. The dummies recode these answers into four categories and produce very large impacts on the probability that the households applied for a cash loan in 1999. Evaluated at the means, the difference between the least and most comfortable households is a 37-percent difference in probabilities. Taken by themselves these dummies account for about one-half of all the variance explained by the regression in Table 3.8. 3.42 These results strongly indicate that an important reason for the low levels of debt among survey households is the perception that they could not afford the repayment obligations. This observation is puzzling in its own way, however. For consumer purchases it is sensible; if a household cannot afford car payments it should not purchase a car on credit. But it cannot explain why households are reluctant to take on debt to finance income-generating activities. 3.43 Table 3.9 reports another econometric model intended to explore the reasons for not applying for cash loans. The dependent variable tries to aggregate the 13 different reasons given in the survey into four groups: (i) did not want credit, (ii) thought that loan would be denied, (iii) costs, and (iv) "other" reasons.30 The multinomial logic model (MNL) reported in Table 3.9 compares each group of reasons for not applying for credit to households that did apply for credit. Each branch of the model is independent of every other, which is a strong restriction implied by the MNL model and, in this case, probably sensible. The best way to view these 30 The variable is based on question E2. The base category is those who did apply for credit. Value 1=7-12; 2=1-3; 3=4-5; 4=6,7,13. As Table 3.9 shows, most answers fall into a few categories, so the more doubtful assignments here do not materially affect the regression results. Chapter 3: Rural Credit Market 79 results is to think of households deciding whether to apply for credit, and, if not, citing a single reason for that decision. The coefficients reflect the probability that a household refrains from requesting credit for that reason. Each branch shows the probability of deciding not to apply for credit for households that only consider the one factor. There are two new controls in this regression, a dummy for whether the household has land and a second, continuous variable, which is the number of hectares owned. Table 3.9: Reasons for not Requesting Cash Loans in past year (base category is "Did request a loan") Multinomial regression Number of observations = 2835 LR chi2 (48) 538.24 Prob > chi2 = 0.0000 Log likelihood 3705.9331 Pseudo R2 0.0677 Explanatory Standard 95% Variables Coefficient Error Z P > z Confidence Interval Age of the Head of Household -.0898703 .024462 -3.674 0.000 -.1378149 -.0419257 Squared Age of the Head of Household .0011901 .0002392 4.975 0.000 .0007213 .0016589 Head of Household is Male .2317868 .1221605 1.897 0.058 -.0076435 .4712171 Second Head of Household .0220422 .1199989 0.184 0.854 -.2131514 .2572358 Farmner -.2532124 .1261803 -2.007 0.045 -.5005213 -.0059035 Owns Land .1555138 .1293622 1.202 0.229 -.0980314 .4090589 Land Area -.031968 .0062981 -5.076 0.000 -.0443121 -.019624 Mazowieckie -.0425736 .1421867 -0.299 0.765 -.3212543 .2361072 Wielkopolskie -.2907427 .1414714 -2.055 0.040 -.5680216 -.0134638 Zachodniopomorski -.0398601 .1527242 -0.261 0.794 -.339194 .2594739 Constant 1.752361 .6151402 2.849 0.004 .5467081 2.958013 Age of the Head of Household -.0770285 .0368448 -2.091 0.037 -.149243 -.0048141 Squared Age of the Head of Household .0009191 .0003476 2.645 0.008 .000238 .0016003 Head of Household is Male -.2906631 .1779753 -1.633 0.102 -.6394883 .0581621 Second Head of Household .9410151 .2599449 3.620 0.000 .4315325 1.450498 Farmer -.6131628 .2331155 -2.630 0.009 -1.070061 -.1562648 Ows Land .1525072 .2043782 0.746 0.456 -2.480668 .5530811 Land Area -.0716886 .0210864 -3.400 0.001 -1130172 -.0303601 Mazowieckie .4412859 .2473102 1.784 0.074 -.0434331 .926005 Wielkopolskie .0300848 .2592998 0.116 0.908 -.4781335 .5383031 Zachodniopomorski .8062203 .2463072 3.273 0.001 .3234671 1.288974 Constant .1304342 .9657542 0.135 0.893 -1.762409 2.023278 Age of the Head of Household -.0886637 .0289888 -3.059 0.002 -.1454808 -.0318467 Squared Age of the Head of Household .0011019 .0002803 3.931 0.000 .0005526 .0016512 Head of the Household is Male .2260075 .1519257 1.488 0.137 -.0717615 .5237764 Second Head of Household .1903297 .1535551 1.239 0.215 -.1106328 .4912922 Farmer -.3856023 .1552442 -2.484 0.013 -.6898753 -.0813292 Owns Land .1225655 .1592191 0.770 0.441 -.1894983 .4346293 Land Area -.0268491 .0085851 -3.127 0.002 -.0436756 -.0100227 Mazowieckie -.4559145 .1678055 -2.717 0.007 -.7848072 -.1270218 Wielkopolskie -.7704828 .1705791 -4.517 0.000 -1.104812 -.4361539 Zachodniopomorski -.8612089 .1906994 -4.516 0.000 -1.234973 -.487445 Constant 1.348536 .7362292 1.832 0.067 -.0944465 2.791519 Chapter 3: Rural Credit Market 80 Table 3.9 (continued) Other Age of the Head of Household -.149229 .0392955 -3.798 0.000 -.2262467 -.0722113 Second Head of Household .001912 .0003575 5.349 0.000 .0012114 .0026126 Head of Household is Male .1841413 .2094169 0.879 0.379 -.2263082 .5945909 Second Head of Household .0564879 .2384919 0.237 0.813 -.4109476 .5239233 Farmer -1.099899 .2834053 -3.881 0.000 -1.655363 -5.444348 Owns Land -.2215108 .2254423 -0.983 0.326 -.6633697 .220348 Land Area -.013322 .0141221 -0.943 0.346 -.0410009 .0143569 Mazowieckie .0942612 .2630707 0.358 0.720 -421348 .6098704 Wielkopolskie .0492011 .2609434 0.189 0.850 -.4622386 .5606407 Zachodniopomorski -.2792641 .2809527 -0.994 0.320 -.8299213 .271393 Constant .9990664 1.064141 0.939 0.348 -1.086611 3.084744 (Outcome why_not=Did appl. is the comparison group). Source: Author's calculations. 3.44 Some controls have similar impacts across reasons, while a few do not. Age, for example, differs in the magnitude of effects, but the basic patterns are similar; middle-aged people do not borrow as much as those younger and older.3' Male-headed households are more likely to refrain from requesting credit because they do not want it. Households with more educated heads are less likely to assume they would be denied credit. This parallels earlier findings that more educated people have a greater familiarity with the expectations of the credit system. Farming status makes a household more likely to borrow for all reasons, but, unfortunately, the largest impact is in the "other" category. Households with larger landholdings are more likely to request credit, and there is an especially large, negative impact from land on the chance that a household will assume it cannot get credit. Since we have controlled for income, this effect presumably reflects a perceived advantage in meeting collateral requirements. Households with higher incomes are also less concerned that they will be denied credit and less likely to say they do not want credit. 3.45 Once again, and, by this point, not at all surprising, is the difference between Malopolskie and other voivodships. Table 3.8 says that households in this voivodship think credit is, in their eyes, too expensive. Our results are just as interesting for what they do not say. Being in Malopolskie does not make a household any more likely to think it will not qualify for credit. How do householdsfinance investment? 3.46 The next logical step is to look at investments and how they are financed. Here again we can gather some clues as to how rural households view the financial system and whether that system can serve their needs. We can break this into three parts: investments in dwellings, in farm activities, and in nonfarm enterprises. 3.47 Table 3.10 gives an overview of household investments in new or improved dwellings in 1999. About 35 percent of all households made investments of this type in 1999. The mean (median) value of these investments was PLN 10,500 (4,250), which makes them quite sizeable. 31 This is the same finding as in the previous regression; the signs are reversed because the dependent variable here is why households do not apply for credit. Chapter 3. Rural Credit Market 81 The table shows, quite remarkably, that a large majority of all households that make such improvements finance them out of their own resources. Bank loans help with the very largest investments, but most households undertaking serious investments do so without any formal lender assistance. Given the very low levels of external finance for dwelling investments, it is not practical to investigate the determinants of this decision in any detail. But some hypotheses can be discounted. One way to read this evidence is that many households find the "rate of return" to investing in their homes higher than what is available from bank accounts. We cannot rule this claim out, but there is almost no difference in reported savings between households that invested in their homes and those that did not, and between households that relied on their own resources for dwelling investments and those that did not. Another possible interpretation is contradicted by the data. Given that 80 percent of households finance their own home investments, binary probit models of the sort used earlier are not going to tell us a great deal about which households do and do not borrow to improve their dwellings. But a simple model (not reported) suggests that more educated household heads are more likely to rely on their own resources to improve their homes, even controlling for income, farmer status, and the other variables used in the earlier models. This does not support the view that the households most likely to rely on their own resources to improve their homes are those with least confidence in their ability to borrow. Beyond this, given the sample size and low levels of external finance for this purpose, we cannot say a great deal. Table 3.10: Sources of Finance for Housing Improvements and Construction Source Percent of households (of those who Mean Size of Investment (Median); in invested in housing) thousands of PLN Own resources 80.0 5.5 (3.0) Bank loan 14.0 20.0 (5.0) Friends and relatives 7.9 4.9 (3.0) Other sources 13.8 7.7 (3.6) Note: ARMA accounted for one loan of 30,000 PLN. Percentages add to more than 100 because households can list multiple sources. Source: Special Rural Household Survey 2001. 3.48 A second question is to ask how households finance investments in both farm and nonfarm activities. At this writing, the data on farm investment were incomplete, so we turn to investments in nonfarm enterprises. Table 3.11 lists, for the first three individuals in the household, the number of nonfarm enterprises and the main source of finance used to start them. The primary source for all of these activities was personal savings, with only about 17 percent financed by any kind of bank. The category "Did not need it" is puzzling. These are not, as one might expect, the smallest businesses. The mean (median) net revenue for all nonfarm enterprises run by the first person in the household is PLN 34,900 (10,900). Businesses where the individual claimed not to need any finance have a mean (median) net revenue of PLN 30,200 (8,900). This mean is actually larger than those financed by loans from cooperative banks. Chapter 3: Rural Credit Market 82 Table 3.11: Sources of Finance for Nonfarn Enterprises Source of Finance Person 1 Person 2 Person 3 Did not need it 27 12 1 Borrowed from commercial or state banks 13 6 1 Borrowed from cooperative banks 15 6 1 Friends and relatives 9 6 4 Borrowed from NGOs 3 0 0 Personal savings 87 24 11 Grants or inheritance 1 1 1 OTHER 8 2 3 Source: Special Rural Household Survey 2001. 3.49 The binary probit model reported in Table 3.12 attempts to distinguish nonfarm activities that were started out of own savings from others. It is important to recall that the income variable here is current income, while the business might have been started several years ago. (About one-half of all businesses run by Person 1 were started prior to 1994.) Two impacts stand out in Table 3.12. First, males are much more likely to rely on their own resources in starting the activity. The change in probabilities evaluated at the means is about 0.25, so thfis is a very large effect. Presumably, this reflects males' ability to direct household savings to enterprises they run, as we have not found any evidence of discrimination against males by banks. The other notable finding here is that individuals who live in Malopolskie are more likely to have started their businesses using their own resources, even when we have held these other factors constant. This finding, once again, points to problems in the lending market in that voivodship. Family income makes little difference in this decision. This finding may reflect the small scale of these activities. Table 3.12: Determinants of Self-Finance for Nonfarm Enterprises Probit estimates Number of observations = 231 LR chi2 (9) = 28.63 Prob > chi2 0.0007 Log likelihood = -138.69714 Pseudo R2 = 0.0936 Explanatory Standard 95% Variables Coefficient Error Z P> z Confidence Interval Age of the Head of Household .1648653 .0635594 2.594 0.009 .0402911 .2894394 Squared Age of the Head of -.0018387 .0006744 -2.726 0.006 -.0031606 -.0005169 Household Head of Household is Male .899425 .2887963 3.114 0.002 .3333947 1.465455 Second Head of Household -.0669113 .1850976 -0.361 0.718 -.4296959 .2958733 Mazowieckie -.2689962 .2542981 -1.058 0.290 -7.674113 .2294188 Wielkopolskie -.4379837 .2292447 -1.911 0.056 -.887295 .0113275 Zachodniopomorski -.4842988 .2463346 -1.966 0.049 -.9671057 -.0014918 Constant -4.203761 1.5187 -2.768 0.006 -7.180359 -1.227163 Source: Author's calculations. Chapter 3: Rural Credit Market 83 Investment and income 3.50 The literature on corporate finance has developed econometric tools to test for the presence of financial-market constraints on investment. The logic of the tests is simple, but their implementation is difficult and controversial. The central idea is that, if a firm faces no financial constraints, then its investment behavior will be unrelated to its cash-flow. In a perfect capital market, a firm with worthwhile investment projects would simply borrow whatever it needs to finance those projects, regardless of its current cash flow. Similarly, a firm with a healthy cash flow, but no worthwhile investment projects would lend to other firms, rather than undertake inferior investments. In practice, these tests amount to a regression of investment on cash flow measures. If we can reject the null hypothesis that the coefficient on cash flow is zero, then we can infer that the firm faces some kind of financing constraints. 3.51 There are several problems with implementing this kind of analysis in the case of rural Poland. First, we know the capital market is not perfect. Rejecting the hypothesis of capital- market perfection is not very informative. Some researchers have tried to use similar tests to ask whether Firm A faces worse financing constraints than does Firm B, but this extension is controversial and requires unverifiable assumptions. Second, many households in the survey report investments, but do not report the value of the investment. 3.52 We can, however, appeal to the logic of these tests to study the relationship between family income and investments in farm or nonfarm activities. This allows us to ask whether families with higher incomes are better able to make investments. A positive relationship between income and the propensity to invest reflects, presumably, both greater credit-worthiness in the eyes of a bank and a greater independence from such loans. (It may also, unfortunately, reflect a sort of reverse causation: a farmer who began an investment project in 1998 may have a higher income from that investment project, and, if he continues the investment into 1999, we may find a spurious relationship between income and the propensity to invest. Given the type of investments most at issue, this long-term investment program seems unlikely). 3.53 Table 3.13 reports a binary probit model in which the dependent variable is 1 if the household reports any farm-related investment in 1999. Because only 15 percent of households report farm investments, the dependent variable is badly skewed. As a precaution, this model was re-estimated as both a binary logic and a linear probability model. Neither model yields results substantially different from those reported in Table 3.13. The independent variables here are total family income and its square; a dummy for whether the household head is male; the household head's age and its square; a dummy for whether the household head has completed secondary school; a dummy for farmer status; and dummies for three of the four sample voivodships.32 We find that older heads are less likely to invest, but neither the household head's gender nor level of education has a significant impact on farm investment decisions. The farmer dummy was included to account for the fact that nonfarm households may find it difficult to diversify into farming, and it has the expected positive sign. 32 In this econometric model, family income has been divided by 100 to put it on the same scale as the other variables. Chapter 3: Rural Credit Market 84 Table 3.13: The Probability that a Household Undertook Any Farm Investment Probit estimates Number of observations 2835 LR chi2(10) = 435.54 Prob > chi2 = 0.0000 Log likelihood = -983.64031 Pseudo R2 = 0.1813 Explanatory Standard 95% Variables Coefficient Error z P>Izl Confidence Interval Head of Household isMale .0471261 .0851822 0.553 0.580 -.119828 .2140802 Age of the Head of Household -.0365104 .0148442 -2.460 0.014 -.0656045 -.0074162 Squared Age of the Head of Household .0001735 .0001447 1.199 0.231 -.0001102 .0004571 Second Head of Household -.1004875 .0801978 -1.253 0.210 -.2576723 .0566973 Farmer .9720407 .0652527 14.897 0.000 .8441479 1.099934 Mazowieckie -.2651493 .0861416 -3.078 0.002 -.4339837 -.0963149 Wielkopolskie -.1910309 .0846756 -2.256 0.024 -.3569919 -.0250698 Zachodniopomorski -.5572586 .0985944 -5.652 0.000 -.7505002 -.3640171 Constant .0085483 .3712806 0.023 0.982 -.7191484 .7362449 Source: Author's calculations. 3.54 The model implies, once again, that the voivodship of Malopolskie is different. Here the difference is simply a greater propensity to make farm investments, and is not necessarily reflective of any problem in the banking system. The most important effects are family income and voivodship location. The family income variable has a large and positive impact on the probability that a household invests, which suggests that self-financing is an important part of farm investment in rural Poland. Given that our model is not appealing to any structural relationship between cash flow and investment, we cannot say whether the estimated relationship is "large" in any comparative sense. We can say that the relationship between income and investment does not differ across voivodships. A different version of this model (not reported) interacts the family income variable with voivodship, and finds that the degree of dependence of investment on income does not differ in a significant way across voivodships. Nonetheless, the strong impact of family income on farm investment is a clear indication that households are being forced to use family resources to undertake farm investments. That, in turn, shows their difficulty in tapping the credit market to finance the growth or modernization of their farms. 3.55 Fewer survey households invested in nonfarm enterprises. Analogous regressions (not reported) also show that nonfarm investment have no appreciable dependence on family income. This result is somewhat surprising, given the farm investment results. The simplest interpretation of this finding is that banks and other lenders are more willing to lend for this purpose than for farm investments. This explanation seems hard to credit, except to observe that some ARMA programs are specifically intended to build up nonfarm rural economic activity. There are several other possible explanations which we cannot, unfortunately, disentangle with the data at hand. Nonfarm investment may be so inexpensive that neither income nor borrowing constraints are any kind of deterrent. Nonfarm activities may also be financed by the suppliers or consumers of these activities. Chapter 3: Rural Credit Market 85 Conclusions 3.56 The evidence shows that rural Polish households are not borrowing for both of the reasons outlined at the outset. Informants in the countryside stressed the lack of a predictable economic environment when asked why they were reluctant to borrow, and several pointed to friends and neighbors who had borrowed to invest, only to lose their investment when the economic environment changed. The survey also supports this position. Some direct answers reflect uncertainty, while in other cases investment decisions do not display the dependence on current income that we expect of household investments in an imperfect capital market. Nevertheless, there is also evidence to support the second hypothesis, which is that capital- market imperfections are discouraging borrowing. Two findings point to this conclusion. First, the strong and nearly pervasive differences among the four sample voivodships do not seem consistent with the uncertainty hypothesis. There are, to be sure, local differences in what the future might hold, but much of the uncertainty facing rural households reflects macroeconomic factors and is common across regions. Second, there is strong evidence that recent investment in farming depends on current household income. This finding can, in principle, be reconciled with the uncertainty hypothesis (perhaps high-income households are more willing to shoulder risk), but it is more clearly consistent with a view that households are undertaking these investments out of family income. The need to use family income to finance investment is a straightforward implication of a credit market that cannot match loanable funds with good investment projects. CHAPTER 4: ON THE DETERMINANTS OF SHORT-RUN FARM REVENUE IN POLAND Introduction 4.1 To analyze farm revenue determinants, we estimate a short-run revenue function using household survey data on farm output quantities, prices, quasi-fixed factors, and household characteristics.33 Farm revenues are defined as total revenues accruing from three principal output categories-agricultural crops, livestock products, and live animals.34 A general representation of this multiple-output, short-run farm revenue function is given by: (1) R(pag,pli,pla;Z) -Max{pagqag + p1iXqi + plql0 I G(qag,qli,qla;Z) = °; q where R represents household farm revenue, p are prices, q are quantities, and the three output categories (agricultural products, livestock products, and live animals) are respectively denoted by the subscripts ag, li, and la. The short-run production possibilities are embodied by G( and depend on a vector, Z, of quasi-fixed factors, household characteristics, and spatial location.35 The analysis is short-run because the elements contained in the vector Z are all quasi-fixed; they are taken as given over a one year period. 4.2 Specifically, Z includes the following quasi-fixed factors: hired-labor, M; land employed by the operation, L; an index of specialized machinery (e.g., milking machines and other farrn equipment), S; tractors in operation, T; and the number of parcels in the farm unit as a proxy for the fragmentation of the operation, F. The vector Z also includes the following household characteristics: average years of education of the head of household and spouse, E; age of the head of household, A; a dummy variable that is equal to 1 when the head of household is enrolled in the KRUS program, K; a dummy variable indicating whether the head of household works outside the household, W; and a dummy variable recording whether the household made an investment funded by a formal financial institution during the past year, D. Finally, voivodship dummies are included to test for systematic farm revenue differences with respect to geographical location. 4.3 Revenue functions are positively, linearly homogenous of degree one, increasing, continuous and convex in prices. The first-degree homogeneity property allows us to normalize the function to two relative prices. In what follows, we normalized the revenue function by the 33 This chapter was prepared by Alberto Valdes and Gustavo Anriquez. They would like to thank Johan Mistiaen for all his helpful suggestions on an earlier draft. 34 Purchased inputs and out-of-farm services are netted out from farm revenue because no quantity data was available for the latter. 35 Chambers (1988) provides an insightful description of the theoretical foundations and properties of revenue functions. For a recent application in the context of rural development and poverty, see L6pez and Romano (2001). Chapter 4: Rural Factor Market 88 price of livestock products. Hence, normalized farm revenue, R/pli, is a function of two relative prices; pla/pui and Pg/Pii. Normalization facilitates the empirical analysis by reducing the number of variables one needs to estimate. Because the exact technological possibilities embodied in G( are unknown to us, we will utilize a flexible functional form approach to approximate the latter. Following the seminal work by Diewert (1971), several flexible functional forms have been developed in the literature.36 The choice of flexible form is determined largely by a combination of data characteristics and the focus of the analysis (e.g., whether one wants to test for homogeneity or separability, rather than to impose it). In this analysis, the data imposed only one main restriction: because not all farms in the sample produce all three product categories, we could not use logarithmic-based functional forms (e.g., the translog). 4.4 We opted for a generalized quadratic specification, because this functional specification allows for zero quantities and has the advantage of allowing the effects of the explanatory variables on revenue to vary across the sample as elements embodied in Z(.) change. Thus, the estimated elasticities of land, capital, KRUS enrollment, etc., are functions of Z(), rather than fixed values-as for example in a Cobb-Douglas specification. For instance, the returns from a marginal increase in land might depend heavily on the capital and labor endowments. In other words, this approach allows us to examine potential complementarities and synergies among the various factors of production. 4.5 Price indices for these three productive sectors were created using the shares of each product or sub-sector in the total revenues of each sector. The normalized generalized quadratic revenue function is specified as follows: 2 2 2 1 1 1 11 11 (2) R? -_R(pag vp,i,p pl2;Z)1pji E ,6 pij( Pj)+ IYik(Pi ZJ +jj-kl(Zk Z,), i=0 j=0 i=O k=I k=1 1=1 where, p , pj ptla / Pli Pag / pli, and k, I index the columns of the vector Z = [M,L,S,T,F,E,A,K,W,D]. Rather than postulating a single intercept, we allow for geographical heterogeneity by specifying separate intercepts for each voivodship. Thus, the revenue function is estimated using a "fixed effects" approach to account for possible regional differences that cannot be explained by the variables included in vector Z. 4.6 The 2000 Rural Poland Survey includes observations for 1,515 farmers, of whom 1,093 reported farm revenues. The survey distinguishes between those without farm revenue and those that report a missing value despite earning some farm revenue. There are no systematic differences between these groups in either demographic characteristics or farm size distribution. 4.7 Table 4.1 presents the estimated coefficients of equation 2. The goodness of fit of the estimated revenue function is high, with a large number of statistically significant variables, including several interactive variables. The regression explains almost 76 percent of the 36 The two volumes by Fuss and McFadden (1978) provide a very thorough survey of this literature. See Chambers (1988) for some more recent applications. Chapter 4: Rural Factor Market 89 variability in the farmers' household revenues. Particularly significant among these variables are prices, inputs, and demographic characteristics. A simple likelihood ratio test between the generalized quadratic specification and a simple linear estimation without the cross products and squared terms does not reject the former specification. The fact that several of the cross-product variables are significant in the regression suggests that marginal returns are highly inter-related among assets. In addition, a separate analysis by farm size, which is allowed by the flexible function, indicates that there are marginal returns on assets such as additional land, machinery, credit, and other variables that vary substantially with farmn size. 4.8 Returns to labor, for example, significantly fluctuate with farm size. The derivative (partial) of the revenue function with respect to hired men is computed to calculate the marginal returns of labor to farmer revenues. This exercise is performed as follow: (3) R = MYM + mm M2 + y, *M *Pla +ymp-a M-Pag +6ML *M-L+ - aR/ aM =M + 26MM -M + rm P,A + rMp7 - Pag +6'ML L, where P5a and Pag naturally are the nornalized prices. However, as Table 4.1 indicates, not all coefficients are statistically significant, specifically, YM is not different from zero, and is treated as such in the calculations. Now, to calculate the elasticity of revenue with respect to labor, we need to compute percentage change of revenues, given a percentage change in labor. This exercise is done as follows: () aRIR i3R M r~SA *Mr Pi L1 MS (4) =R-* M_; =- [2J,mM * M + yMp, * Pla, + rmp_a' Pa5,g + /6ML J LR M aMIM aM R R 4.9 The elasticity presented in Table 4.3 is calculated using the coefficients '5.MM, Y- YM; , and 5ML from Table 4.1, and the sample means of the variables R, M, rag' and L. On the other hand, the elasticities for a specific farm size group can be computed using the means of the same variables, but for that group. As shown in Table 4.3, the elasticity of labor varies substantially with farm size. Table 4.3 presents the estimated elasticity of household income by farn size. The fann size categories that were used are: less than 1 hectare (minifundia, 90 farms), 1 to 7 (small, 446 farms), 7 to 15 hectares (medium, 249 farms) and over 15 hectares (large, 308 farms). 4.10 Returns to labor are slightly negative. Only minifundia shows a slightly positive elasticity with respect to labor, which indicates over-hiring of labor in farms, given the low productivity. Note that labor elasticity should be equal to the share of expenditure in labor out of total revenue. If farmers are behaving optimally, the marginal revenue of labor should be equal aR M wM to the market wage rate, Rl/ aM = w. Thus, the elasticity: a -* R should be equal to the input share of labor on revenues, if farmers behave optimally. This equality is observed for minifundia, but not for larger farms, which indicates low (negative) productivity of labor, and over-hiring. Chapter 4: Rural Factor Market 90 Table 4.1. Net Farm Revenue Regressions Variable Estimator Standard Error Voivodship Dummyj -152.809 54.744 Voivodship Durmmy2 -121.450 47.411 Voivodship Dumnny2 -105.791 52.001 Voivodship Dummy3 -168.009 54.714 Index of Price of Live Animals 61.858 75.157 Square of Price of Live Animals 7.675 5.487 Index of Price of Agricultural Crops 71.006 163.142 Square of Price of Agricultural Crops' 34.470 18.085 Price of Live Animals x Price of Agricultural Crops' -35.053 20.144 Hired Men 15.292 12.730 Square of Hired Men2 0.831 0.395 Price of Live Animals x Hired Men2 42.262 17.456 Price of Agricultural Crops x Hired Men2 -91.650 46.593 Land x Hired Men3 -0.689 0.205 Years of Education of Head of Household2 8.514 4.078 Square of Years of Education 0.087 0.219 Price of Live Animals x Years of Education -1.268 6.383 Price of Agricultural Crops x Years of Education -16.508 14.942 Age of Head of Household 0.557 0.375 Land -0.676 1.306 Square of Land2 -0.008 0.003 Price of Live Animals x Land 0.155 0.972 Price of Agricultural Crops x Land3 6.209 2.359 Special Machinery2 -22.752 10.478 Square of Special Machinery3 2.194 0.607 Price of Live Animals x Special Machinery/ -23.204 11.742 Price of Agricultural Crops x Special Machinery' 47.667 24.748 Land x Special Machinery2 0.374 0.165 Tractors in Operation 23.700 25.201 Square of Tractors3 12.592 4.700 Price of Live Animals x Tractors 16.116 25.590 Price of Agricultural Crops x Tractors -83.774 59.200 Parcels 2.474 2.620 Square of Parcels -0.057 0.053 Price of Live Animals x Parcels' 7.084 3.628 Price of Agricultural Crops x Parcels2 -18.365 8.203 Credit' -58.623 37.319 Price of Live Animals x Credit' 141.273 74.175 Price of Agricultural Crops x Credit -106.537 112.843 Head of Household Enrolled in KRUS DurnMy2 17.441 6.991 Head of Household Works Outside Dunimy 9.504 20.264 Price of Live Anirals x Works Outside Dummyt -41.677 22.238 Price of Agricultural Crops x Works Outside Dummy' 102.034 53.697 Land x Works Outside Dununy3 -1.711 0.664 Number of Household Memnbers 0.251 2.136 2 R'= 0.76, 1093 observations. I) Denotes variable significant at the 10% level; 2) Variable significant at the 5 % level; and 3) Variable significant at the 1% level. White's heteroscedasticity consistent covariance matrix is used for standard errors. Source: Special Rural Household Survey 2001. Chapter 4: Rural Factor Market 91 4.11 The returns to land, as theory indicates, are positive and diminishing, as indicated by the negative coefficient on the square of land. The elasticity of the sample is relatively high, except for minifundia, indicating that an increase of the farm size by one percent increases revenues by slightly more than one percent. The fact that the minifundia category has an elasticity of land clearly less than one shows that there is a minimum farrn size for land to be more productive. 4.12 Education elasticities are very high. It should be noted that levels of education are very low for all farm sizes, 9.66 years on average. Most of these years, on average, correspond to the compulsory primaiy education that in Poland amounts to 9 years. Perhaps the average levels of education are too low, which explains the high returns estimated. For example, a farmer with one year of education above the mean would, everything else equal, have revenues 17 percent higher. For smaller farms the estimated elasticity of labor is very high, which is explained by their relatively lower revenues. 4.13 Returns on capital assets vary significantly with farm size and asset type. Returns to tractors are high for all farm types, but are especially higher for small and medium sized farms. This result is consistent with budgetary constraints. For very small farms (minifundia), tractors are not productive, but, for bigger farms, tractors are very productive, but less productive for the largest farms, all of which indicates under-use of tractors by small and medium-sized farms, possibly because of financial constraints. On the other side, specialized machinery -indexing milking machines, specialized cropping machines, and other machinery-exhibits positive returns only for the largest farms. That is, specialized machinery is only profitable given a certain farm size. Also, the returns to specialized machinery are positively correlated with the price of agricultural crops as indicated in Table 4.2. Less than 5 percent of the households made an investment in the farm, funded at least in part by a formal financial institution. Table 4.3 indicates that farms that made an investment had, ceteris paribus, revenues 19 percent higher than those who did not. These investments are particularly more productive for the small and mid-sized farms, and are particularly more profitable in the live animals sector. Table 4.2. Effects of Prices and Land over Factor Markets RETURNS ON Land Education Labor Investment Capital in Spec. Tractors Machinery Price of Live Anirnals 0 0 + + 0 Price of Agricultural Crops + 0 - 0 0 + Price of Livestock Products - 0 - - 0 Land - N/A - N/A N/A + Source: Author's calculations. 4.14 Fragmnentation of the operation as expected affects revenues negatively. It should be noted that the average farm operates almost five different parcels. This excessive fragmentation of farns is very costly for farmers' revenues, as indicated by an elasticity of -0.281. If farms operated only 2.5 different parcels on average, revenues could be increased by 14 percent. Fragmentation is less detrimental for larger farms, probably because those larger farms exhibit economies of administration. Chapter 4: Rural Factor Market 92 Table 4.3. Estimated Elasticities of Net Revenue by Farm Size All Saple Minifundia Small Medium Large Hired Labor -0.042 0.001 -0.134 -0.035 -0.050 Education 1.716 10.723 11.127 3.220 0.615 Land 1.082 0.218 2.136 1.364 0.989 Specialized Machinery -0.345 -1.406 -2.128 -0.645 0.089 Tractors 0.593 0.228 1.626 1.056 0.630 Fragmentation -0.281 -0.285 -1.809 -0.512 -0.115 Works Outside Household -0.079 0.321 0.668 -0.051 -0.072 Investment 0.193 0.000 0.651 0.224 0.151 KRUS Enrollment 0.198 0.249 0.770 0.508 0.109 Source: Author's calculations. 4.15 The households where the head works outside the farm, on average, earn lower revenues. However, this result varies with farm size. For smaller farms it is profitable for the head to work outside the farms. The larger the farm, the costlier it is for the head to work outside the farn, with an elasticity of -0.072 for farms bigger than 15 hectares. Finally, those farms that are enrolled in the KRUS program, 55 percent of the households, earn more farm revenues. The estimations indicate that, everything else equal, farms enrolled in this public program earn revenues almost 20 percent higher. 93 REFERENCES Bedi, A. S. 1998. "Sector Choice, Multiple Job Holding and Wage Differentials: Evidence From Poland." The Journal of Development Studies, October. Bentolila, Samuel. 1997. Polish Labor Market Institutions on the Road to the EU. Madrid: Centro de Estudios Monetarios y Financieros. Working Paper No. 9712. Chlon, Agnieska. 1998. "Does Unemployment Influence Migration in Poland?" The Warsaw School of Economics. Chlon, Agnieska. 2000. "The Polish Farmers' Social Security System." Draft Discussion Note. Cohany, Sharon. 1996. "Workers in alternative employment arrangements." Monthly Labor Review 119, no. 10: 31-45. Cohany, Sharon. 1998. "Workers in Alternative Employment Arrangements: A Second Look." Monthly Labor Review 121, no. 11: 3-21. FAO. 1999. Polish Agriculture: Economic Transition and Long Run Trends. June draft. Frenkel, Izaslaw, and Andrzej Rosner. 2000. "Population and Labor Market in Rural Poland." Manuscript. GUS (Polish Statistics Office). 1994. "The Situation in the Rural Labor Market with Special Consideration of Individual Farmers." Hipple, Stephen. 1998. "Contingent Work: Results from the Second Survey." Monthly Labor Review. 121, no. 11: 22-30. Hubbard, R. Glenn.. 1998. "Capital Market Imperfections and Investment." Journal of Economic Literature, XXXVI: 193-225. Klank, Leszek,. 2000. "Rural Financial Market in Poland." Manuscript. Kupiszewski M., H. Durham, P. Rees. 1998. "Changes in the Regional Population Dynamics in Poland, 1980-1984." Studia Demograficzne, 3/133: 3-18. Kupiszewski M., Durham H., Rees P. 1998. "Internal Migration and Urban Change in Poland." European Journal of Population, 14, no. 3: 265-290. Lerman, Zvi. 2001. "Poland Rural Household 2000 Survey: Land and Labor Productivity Analysis." Manuscript. National Bank of Poland (General Inspectorate of Banking Supervision). 2000. Summary Evaluation of the Financial Situation of Polish Banks. 1999. 94 Nesporowa, Alena. 1999. "Employment and Labour Market Policies in Transition Economies." Geneva: ILO. Newell A., and M. Socha. 1998. "Wage Distribution in Poland: The Roles of Privatization and International Trade, 1992-96." Economics of Transition, 6, no. 1: 47-65. Ray, Debraj. 1998. Development Economics. Princeton: Princeton University Press. Rutkowski, J. 1996. "Changes in the Wage Structure during Economic Transition in Central and Eastern Europe." World Bank Technical Paper No. 340. Rutkowski, J. 1997. "Low-Wage Employment in Transition Economies of Central and Eastern Europe." MOCT-MOST 7:105-130. Szemberg, Anna. 2000. "Transformation of the Agrarian Structure in the Private Farm Sector." Manuscript. World Bank. 1999. "Poland-the Functioning of the Labor, Land, and Financial Markets: Opportunities and Constraints for Farming Sector Restructuring." World Bank concept paper. World Bank. 2001. "Poland Labor Market Study - The Challenge of Job Creation." World Bank report No. 22490-POL, June 2001, Washington, DC. Zientara, Wojciech. 2000. "Land Market in Poland in the Post-War Period." Manuscript.