Policy, Research, and External Affairs WORKING PAPERS | - Agricultural Policies Agriculture and Rural Development Department The World Bank August 1990 WPS 471 The Determinants of Farm Investment and Residential Construction in Post-Reform China Gershon Feder, Lawrence J. Lau, Justin Lin, and Xiaopeng Luo Forced consolidation of small Chinese farms into larger farms is unwarranted but China needs mechanisms to facilitate free- market land transactions, better supplies of such inputs as fertilizer, and - when those are available - a reoriented rural credit system. Early extensions of farmers' leases on state- owned land would reassure farmers about the government's ; commitment to the present system. The Pohcy, Research, and Extemal Affatrs Complex distributes PRE Working Papers todisseminate the findings of woik in progress and to encourage the exchange of ideas among Bank staff and all othes interested in development issues. These papers carry the names of the authors, reflect only their views, and should be used and cited accordingly. The findings, interpretations, and conclusions are the authors' own They should not be attributed to the World Bank, its Board of Diectors, its managenent, or any of its member countries. Policy, Research, and Externt4 Affairs Agricultural Policies WPS 471 This paper -a product of the Agricultural Policies Division, Agriculture and Rural Development Department - is part of an cftort to investigate rural credit markets, f.arm investmecnt, and agricultural productivity in C'hinia. That research is part of a larger effort in PRE to deternine how financial intermediation affects economic activities. Copies are available free from the World Bank, 1818 li Street NW, Washington DC 20433. Please contact Cicely Spooner, room N8-039, extension 3(064 (35 pages). After 20 years oi colleclivization, China's tion with issues of farm size and consolidation agricultural sector was reformned in the last are unwarranted. The production gains from decade. Individual fami/household units re- consolidation would be limited and the costs placed collective production. Households were substantial. given individual leases on forner commune land first for 3-5 years, but now for 15 years, and Where farms are tiny, farm size is a problem even longer for tree crops. -- but coercing consolidation or recollectiviza- tion would be harmful. It would be preferable to Household data on1 four areas in China in introduce institutional mechanisms and proce- 1987-88 revealed pattems of spending on1 dures to facilitate market-induced land transac- productive assets, durable consumer goods, and tions. More mobility of labor would also help. housing. Concems about the inadequacy of invest- Using a model of household production and ment finance for agricultural households are not investment decisions, Feder, Lau, Lin, and Luo yet justified in areas where the supply of such analyzed data on several factors that had been production inputs as fertilizer is unsatisfactory. thoughit to inhibit investment in farm capital and But once the input supply system improves, encourage residential or other nonfarm invest- limited credit will become a constraint - and ments: the typically small size of farms together the rural credit system, which is geared to rural with increasing retunis !o scale in production; industry and commerce, will have to be inadequate credit; and farmers' perceptions of reoriented. insecurity because of possible policy shifts during the life of their leases on state-owned land Radical revision of the land tenure system is or the likelihood of being assigned other lands not called for as the land Ieasing system seems when the contract matures. not to be hampering investment. But likely crosion of investment incentives will be averted What were the policy implications of the if leases are extended before they mature, study results? reassuring farmers about the government's long- term commitment to the present system. If the four study sites reflect the situation elsewhere in China, policymakers' preoccupa- The PRE Working lPaper Scries disseminates the findings of work inder way in the Bank's Policv, Rescarch, and Extemal AffairsComplex. An objective of the scries is to getthcsc firdings out quickly, even if prcsentations arc less than fully polished. The findingc, intcrpretations, and conclusions in these papers do not necessarily represent ofricial Bank policy. Produced by thc PRE Disseminationi Center The Determinants of Farm Investment and Residential Construction in Post-Reform China* by Gershon Feder, Lawrence J. Lau, Justin Lin, and Luo Xiaopeng Table of Contents I. Introduction 1 II. Description of Study Areas and Surveys 3 III. Factors Potentially hibiting Farm Investment 6 A. Farm size 6 B. Tenure security 6 C. Finance inadequacy 9 IV. Patterms of Investment and Expenditure on Housing 13 and Durables in the Study Areas V. A Model of Household Consumption and Investment 20 VI. Econometric Results 23 VIl. Summary and Conclusions 32 References 35 * The paper benefitted from helpful comments by Richard Burcroff. 1. INTRODUCTION After twenty years of collectivized agriculture, China's agricultural sector underwent a major reform during the last decade. A key element of the reform was the introduction of the "household responsibility system", which replaced the collective production mode by Individual farm household units. Households were given control over former commune land through Individual contracts (Initially for 3-5 years, but presently for 15 years with even longer periods for tree crops). Asido from a commitment to provide certain prespecifled grain quotas at government-set prices, and the obligation to pay taxes, farmers now have the freedom to make cropping and Input decislons, and are allowed to retain any profits which they earn. The Improved Incentives brought about by the Institutional reform, combined with higher prices for key agricultural products, Induced an unprecedented acceleration of agricultural growth In China. Between 1979 and 1984 the value of agricultural output (In constant prices) grew at an average of 7.5 percent per year, mostly due to Improved incentives (Lin, 1988; McMIllan et al., 1989). After 1984, the pace of growth has slackened considerably, essentially because the potential gains due to Improved Incentives under the reformed structure were exhausted. Further growth thus depentJs on the traditlonal sources of agricultural development, I.e., further Investment In physical capital, expanded material Input supplies, and technological change. In this context, the analysis of factors affecting farm Investment In China Is of much policy relevance. National data indicate that along with the reform In production organization, public sector Investment In agriculture declined, In both absolute and relative terms. Agriculture's share In the government's capital -2- construction exr'nditure fell from 12.5 percent In 1979 to 4.6 percent In 1986. However, the reintroduction of compulsory labor services for maintenance of irrigation Infrastructure was equlvalent to more than half the state's budget for capital construction In agriculture during 1987. Information on farmers' farm Investment Is scant, but there are some indications of a response to the stimull provided by the reform: In 1986 about 90 percent of the purchases of walking tractors, and a rapidly expanding share of truck purchases were by Individual farmers. On-farm investment by farm households has likely been substantial, but It Is not clear to what extent It has been overtaken by households' expenditure on non-productive assets (consumer durables and housing) and non-farm enterprises. Whilo the household responsibility system stimulated production Incentives, It can be argued that concerns regarding the stability of the land tenure system introduced by the reforms, extremely small farm sizes, and credit Inadequacies hinder farm Investment, and may have caused a preference for Investing In non- productive assets (e.g., housing) and In non-agricultural activities. Such arguments have been raised quite frequently by observers of Chinese agriculture, but there has been a paucity of empirical research to assess their validity and importan^e The c.,_Jective of this paper Is to clarify, on the basis of detailed farm level data derived from recent surveys, the importance of factors relatid to tenure security, farm size and credit availability In constralning farmers' agricultural Investment. In particular, a direct measure of farmers' perceptions regarding tenure security will be utilized, as well as Information on transactions In the credit market. The next section provides a description of the study areas. -3- It Is followed by a discussion of factors affecting farm Investment and a description of Investment patterns in the study areas. A formal model of farmers' consumption and Investment uecisions, and an econometric analysis are then presented and results are Interpreted. The last section summarizes the paper. I1 DESCRPTION OF STUDY AREAS AND SURVEYS The data underlying this study were obtained through recent farm surveys organized by the authors In four countries of China: Gongzhullng county In Jilin Province (December 1987). Tal and . urong counties In Jiangsu Province (March 1988) and Xlajlang county In Jlar -,Al Province (November 1988). The samples consist of agricultural households, Ie., households operating farms as a major activity, as distinct from other rural households which engage In commerce or are mainly employed In rural Industry. Gongzhullng county Is located in Jilin province of northeastern China. Agro-clknatic conditlons are such that only one crop season Is feasible annually, and corn Is the major crop grown. Farm sizes In the study area In Gongzhuling are large relative to typlcal farms sizes In China, and the significant surplus of output over consumption requirements makes the county a leading corn supplier In China. Tal and Jurong counties of Jiangsu province are within the Central East China region where a two-season wheat-rice cycle Is practiced annually. They are characterized by high population density and consequently have very small farm sizes. Many farmers In these counties supplement their Income through off-farm -4- employment (mostly In township or village enterprises) or non-farm business activitles. X1ljlang county Is located In Jiangxi province of Southern China where double-cropping of rice Is predominant. The county Is a major rice producer, witi typical farm sizes double that of Jiangsu province. Farm fragmentatlon In the study areas In Jilin and Jiangsu province Is less severe than the average for China (which Is 9 tracts per farm). However, In Xlajiang county fragmentatlor, Is a serlous problem, (an average of 16 plots per household) and In Jiangsu tracts are extremely small. Table 1 summarizes some key characteristics of the samples from the study areas. The fow counties covered by the study were purposefully selected from the national sample of 846 countles which are surveyed annually by the State Statistical Bureau. Each of the counties were deemed typical of the agro-clinatic reglon In which It Is located. The samples consisted of severa randomly selected townships within each county, and within each township a number of ranooly selected villages, with a total sample size of about 200 farmers per county. Both Gongzhuling and Xlajlang are surplus grain producing counties, while Tal and Jurong, where population density Is very high, have less grain surplus as farm sizes area very small. Farmers were asked In the course of Interviews to provide information on their farm operations, assets (at present and In the past), credit markot transactions, and perceptions regarding tenure security. Table 1: Character'tics of Sample Farm Households County Gongzhullng Tal Jurong Xlajlang Item (N-200) (N-200) (N1-199) (N-200) 1 Mean farm size 20.75 4.63 6.90 11.31 (Mu) a/ 2. Share of Income 18 47 47 15 from non-farm sources 3. Per capita Income 952 737 832 472 b/ In 1987/88 Yuan 4. Mean No. of land 3.7 7.05 4.90 16 parcels per HH 5. Main crop corn wheat (winter) wheat Irvter) Rice (double (single season) Rice (Summer) Rice (o. ier) cropping) a/ 15 Mu - 1 hectare _/ One season hi 1988 only -6 111. FACTORS POTENTIALLY INHBITING FAM INVES1IENT A. Farm size. If productlon exhlbits Increasing returns to scale (at least within the relevant range), then larger farms will tend to have larger capital/land ratios, provided that supplies of other Inputs or credit are not constrained.1 Put differently, with Increasing returns to scale the consolidation of farms would load to higher Investment and land productivity. Farm sizes In China are typically small even when compared to other Asian countries (.55 hectare per household Is the national average), and their effective size Is further diminished by fragmentation (the national average Is about 9 parcels per farm). Some policy makers In China argue that the small farm sizes brought about by the shift Into household-based production have hindered Investment, and have called for partlal recollectivizatlon and consolidation. The perception of Increasing returns to scale relies In part on the observation that some capital goods are not divisible (e.g. tractors and draft animals). However, various arrangements have evolved In China's agriculture to circumvent this proWem. In some areas, communal equipment and draft animals have been sold or assigned to selected households who committed to provide draft services to other farmers at agreed rates. In other areas, shared ownership of animals and equipment among several households Is practiced. B. Tenure securlty. Under the household responsibility system, farmers do not own the land, but have a lease protecting their use rights for the duration of the 1 A sufficient condition for this propositlon to hold when the productlon function is homogenous Is that Inputs are complementary In productijn (i.e., positive cross second derivatives). -7- contract. For most types of Investments, a duration of 15 years as practiced at present may be viewed as sufficiently long to amortize animals and equipment. Uncertainty may however prevall In farmers' mind regarding the possibility that a change In government pollcy (whether at the central o.. ocal level) would lead into a reallocation, before the contract expires, of some or all of the land over which they hold a contract. If some or all of the land Is taken away, farmers with excess capital may suffer losses even If the capital Is mobile (i.e., not tled to the land). This Is because of the transaction cost of liquidating capital. In addition to the risk of reallocation of land, farmers may be concerned about a possibility of a reversal of the individualizdtlon policy, leading to some form of consolidation and collective operation. This possibility entails uncertainties regarding the treatment of privately owned production capital. The risks outlined above can diminish the Incentive to Invest In farming capital (e.g., equipment, machinery and draft animals), but are less likely to affect Investment in livestock (pigs, poultry, etc.) as these have a short paycoff period and will most likely remain in private ownership regardless of policy changes. Insights regarding farmers' tenure security perceptions can be derived from responses to specific questions pertalning to the po3sibillty of land reallocatlon. These questions were Included In the f,lrvey underlying the present study. As Indicated In Table 2, In three of the counties studled, only a minority of the farmers perceive a hlgh likelihood of lanJ reallocation before the current contract expires. In contrast, in Xlajlang county three-quarters of the farmers expect land reallocation before the contract expires, I.e., a majority of the farmers In the courty do not have much faith In authorities' commitment to the present allocation. This result Is compatible with the fact that the survey In Xlajlang county took Table 2: Farmers Perceptions Regarding Security of Tenure County Item Gongzhullng Tal Jurong Xlajiang (N-200) (N-200) (N-199) (N-200) X of farmers 17 17 24 75 perceiving high likelihood of contract disruptlon before expiration date X of farmers 85 76 78 59 percelving low likelihood of being reassigned the same farms after contract expiration date place much later than In the other counties (November 1988). By the time of the survey, there have been several well publicized incidents In several areas of China durlng the second half of 1988 where local authorities forced consolidation of small farms. The changed Insecurity perceptions In Xlajlang are not likely to have had an impact on the Investment data underlying the present study, as these pertain to the poriod 1984-88. A longer-term perspective underlies the responses to a question regarding the likelihood of being assigned the same traccs of land upon contract expIration. The majority of farmers In all counties perceive low likelihood of regalning use rights to the same tracts of land. While the risk of not being asslgned the same parcels of land would clearly have a negative effect on land-embodied Investment, -9- It may also hamper Investment In mobile capital (e.g., machines), because uncertainty regarding re..location of plots can Imply also uncertalnty regarding the future size of the farm and thus uncertainty regarding tPe marginal productivity of capital once the contract explres. However, when there Is a relatively long- term land lease, concerns regarding future farm size would not have a strong kmpact on Investment during the first years, as It will be fully or almost fully depreclated by contract expiration date. Investment In housing Is viewed by Chinese farmers as completely risk free, because hcuses have been privately owned even In the years of collective agriculture, and are likely to remain privately owned. It sihould be further noted that housing Is a consumption Item with high Incc... o elasticily. The income effect Is augmented by demographic factors. In China's rural areas, a new house has become a precondition for marriage eligibility for young males. With the coming of age of Individuals born In the baby boom of the ear!y 60's, the demand for housing has Increased. These facts, and the substantlal pent-up demand due to the absence of any significant housing Investment In the pre-reform years could produce a maJor adJustment In the housing stocK once Incomes have rlsen and private construction activities have become feasible. C. Finance Inadequacy Credit constralnts can be another factor Inhibiting investment. Most Investment outlays (and housing Investment as well) ten, to be lumpy, requirlng a substantlal amount of liquld resources. Because of this lumpiness, the necessary funds cannot typically be saved from one year's Income, and long or medlum-torm credit, or accumulated savings, are needed to finance Investment. Aggregate statistics Indicate that the share of medium and long-term loans In the portfolio of the rural credit cooperatives (the main lending institutlon - 10 - deallng with agricultiral households) is a mere 3 percent (World Bank, 1988a, P. 258). In the areas covered by the present study, the share of short term loans out of all Institutlonal loans varles between 90 percent and 100 percent in the four counties (Feder et al., 1989). It Is possible, of course, to finance investment by repeatedly rolling over short-term loans. In fact, the structure of Interest rates In China provided Incentives for such a rollover, as Interest rates on short-term production loans were lower than Interest rates on medium-term loans throughout the period under study (World Bank, 1988b, paper 7, annex 1, Table 10). However, short-term loans are typically granted In smaller amounts. Theoretically, when liquidity (and credit in particular) Is fungible, a shortage of credit would affect all types of Investment (whether productive or residential). However, deficiencies and segmentatlon (by source and purpose) of credit markets could create a situation whereby funds which are available for housing Investment cannot be used for productive Investment. An assessment of the operation of credit markets In rural Chins Is thus necessary In order to determine the role of credit constralnts In explalning the o!tserved Investment patterns. Sources and uses of credit In the study areas are presented In Table 3, which describes the composition of loans undertaken by sample farmers In the most recent season prior to the Interview. The boilk of institutlonal credit Is declared to be destined for production (input purchase) as Is Indeed Intended by government's policy. However, most Institutional credit Is de-facto fungible, regardless of the stated objectives for which It Is acquired, and It Is likely that short-term production credit could be used to finance Investment In farm capital or housing. The extent to which rolled-over short-term Institutional credit could substitute for medium and long-term funds Is probably lknited, because amounts - 11 - are typically small, and the need to repeatedly roll over the loans Introduces an element of uncertainty and Inconvenience. In rural China, non Institutional credit Is less likely to be fungible, because It Is granted mostly free of Interest by relatives and friends, who have close knowledge of the activities and financial resources of the borrowers. Unlike Institutional credit, which Is by-and-large short-term, Informal credit In the study areas Is mostly medium and long term, or has no definite maturity date (Feder et al., 1989). It Is likely that relatives and friends are more Inclined to lend money for purposes of house construction and special social events (e.g. weddings) which are perceived as a basic need deserving assistance. They are less likely to provide Interest-free loans for agricultural Investment which Is undertaken for the purpose of Increasing the borrower's Income. Table 3 Indlcates that a higher share of non-institutional credit Is devoted to purposes not related to production, and a significant proportion of It Is used to finance constructlon rather than productive Investment. Because of their ability to monitor borrowers activitles, loans acquired from relatives and friends are typically used for the specific purposes for which they were granted, and thus do not add to overall liquidity. It Is therefore expected that Informal credit Is not available as a significant source of funding for farm Investment In the study areas. Table_3: Distribution of Loan Purposes by Type of Lender (Percent) Purpose Sample Production Farm Construction Consumption Social Other Size Machinery (Wedding, Funeral, etc) County and (loans) Source Gmzning Institutional 212 93.0 4.0 2.0 0 1.0 0 NMn-Institutional 47 11.0 4.0 23.0 15.0 26.0 21.0 Tai Institutional 57 89.0 2.0 4.0 0 0 3.0 Nm-institutioral 25 36.0 0 32.0 0 20.0 12.0 Jurong Institutional 29 48.0 10.0 10.0 3.0 10.0 19.0 Nm-Institutional 31 6.0 19.0 29.0 13.0 23.0 10.0 Xiajiang Institutional 158 67.9 3.8 5.8 1.3 8.4 14.8 Nb-Institutional 85 33.8 1.5 28.0 9.2 20.0 9.5 - 13 - NI. PATTERNS OF INVESTIENT AND EXPENDITURE ON HOUSING AND DURABLE! iN THE STUDY AREAS The Implementation of reforms In China's agricultural sector coincided with other economic changes which made various consumer goods more available in both urban and rural areas. Thus the increased Incomes and savings could be used for acquisitlon of consumer durables, an expense of significant magnitude relative to typical farm Incomes. Table 4 compares ownership of several major consumer durables (T.V.,radio/tape, bicycle, sewing machine, watch) at two points In time: 1983, which In the study areas was a year when the new production mode was at Initial stages of Implementation, and 1988 (or 1987 In the case of Gongzhullng) when the reform was well In place, and farmers had already benefitted from several years of higher Incomes. All Items show significant Increases. The most remarkable change occurred In ownership of T.V.'s which Increased by hundreds of percent In all countles. Consumer durables are obviously private property which wl!l not be affected by any changes In government's policies In the agricultural sector. Investments In productive farm assets have been substantlal, compared to initlal levels of capital In the study areas. As demonstrated In Table 5, capital stocks (livestock and equipment) have more than doubled In the 4-5 year period covered by the data, Implying annual growth rates of capital exceeding 15 percent.2 This represents significant real growth, as the price Index of agricultural equlpment rose by only 15 percent over the whole pe lod covered. 2 The survey covered several types of land improvements, such as clearlng of stumps and constructing bunds, but there was llttle variatlon In the data as most of these Improvements were already undertaken before the reforms. Other types of improvements were difficult to quantify. - 14 - As one would expect, the highest volume of Investment per household took place In Gongzhullng and Xiajlang where farms are larger. Table 4: Changes In Ownership of Consumer Durables in Study Areas County, Year Gongzhullng Jurong Tal Xlajiang (N-200) (N-199) (N-200) (N-200) :em 1983 1987 1983 1988 1983 1988 1983 1988 Own 1.V. 6 42 11 38 2 26 5 48 Own radio/tape 3 14 2 17 2 12 34 44 g. no. of .62 .98 1.14 1.33 1.10 1.43 1.02 1.26 bicycle , own sewing 46 57 22 37 11 33 10 19 machine =wn watch 64 83 68 94 53 91 59 82 - 15 - Table 5: Cumulative Productive Investment In the Period 1983-1988 a/ (sample means In Yuan) County Gongzhullng Jurong Tal Xlajiang Item (N-200) (N-199) (N-200) (N-200) Crop-related Investment (i) Tractor 309 139 0 95 (11) Other equlpment 173 82 42 127 (ill) Draft anknal 200 76 12 357 Sub-total 682 297 54 579 Other Productive Investment (i) Truck/boat 120 0 123 0 (ni) Livestock 104 214 150 556 Sub-total 224 214 273 556 Total Productive Investment 906 511 327 1135 Total productive Investment 159% 127% 131% 107% as % of initlal capital a/ Figures for Gongzhullng are for the period 1983-1987, while the figures for other counties pertain to the perlod 1984-1988. The figures are undeflated. The price Index for agricultural equipment Increased by 15 percent between 1983/4 and 1987/8. - 16 - The composition of Investments differs across the countles studied. Food animals (plgs, chicken) are a major component of investments In Xlajlang and Tal county. Draf t power (mechanized or anInial) composes over one half of the Investment In Gongzhuling, where farm sizes are the largest among the study areas. In Tal county, where canals are numerous, farmers Invest In boats (which are used for transportation). The figures reported for Investment In tractors and trucks may be misleading, because tractors are expensive and they raise the sample mean slgnlficantly even though only a few households have acquired them. For this reason It Is useful to observe the changes In the frequency of ownership for different capital Items, reported In Table 6. It Is noted that even In Gongzhullng, less than 10 percent of the households own tractors. The most common form of Investment Is In pigs, which constitute In all counties except Gongzhullng about 40-50 percent of productive Investment. Investment In house construction or housing Improvement Is a major form of asset accumulation for farmers In China (Tam, 1988). Table 7 describes several types of housing knprovements undertaken by sample farmers, such as the Installation of tile or tin roofs Instead of straw, and the replacement of dirt floors by concrete. Slgnificant Improvements are observed In housing standards In all study areas, and about half of the sampled farmers Indicated that they have Invested In housing Improvement or expansion since 1983. The volume of funds Invested In housing Improvement Is substantial. On average, the outlay on house Improvements for households undertaking such Improvements was more than thelr average annual Income, exceeding productive Investment by a wide margin (last llne In Table 7). The share of housing Investment In the total outlay on productive Investment and - 17- housing knprovement ranges from 64 percent In Xlajlang to 91 percent in Tal county. Some possible reasons for the heavy Investment In housing, which are not related to any of the factors potentially inhlbiting productive investment, have already been mentioned In sectlon 11 above (e.g., demographic trends and hlgh Income elasticity). But It Is also possible that some of the factors hampering Incentives for productive Investment Induce an asset composition more heavily dominated by residential capital (e.g., limited tenure security, segmented credit markets, and small farm size). These Issues need to be Investigated empirically, but prior to the econometric analysis a formal model needs to be presented to underly the empirical work. - 18 - Table 6: Changes In Productive Assets % Who Own Gongzhullng Tal Jurong Xlajlang (N-200) (N-200) (N-199) (N-200) 1983 1987 1983 1987 1983 1987 1984 1988 Tractor 1 9.5 0 0 3.5 8.0 4.0 5.5 Truck/boat 0 0.5 4.5 26.0 0 0 0 0 Pump engine 1 2 0 1.5 6 11.5 0.5 1.5 Thresher 0 3 0 0.5 2.5 12 92.0 97.0 Buffalo/oxen 5 19 0 3.0 53 77 94.5 92.5 Pigs 81 66 86.5 66.5 85 84.5 97.5 98.5 Other anknals 15 30 .5 0 1.5 0 0.5 0.5 Average value per household (yuan) a/ 570 1476 250 577 403 914 1060 2195 a/ The values are not deflated. The price Index for agricultural equipment rose by 15 percent between 1983/4 and 1987/8. - 19 - Table 7: kmprovements In Housing County, Year Gongzhullng Jurong Tal Xiajlang (N-200) (N-199) (N-200) (N-200) :em 1983 1987 1983 1988 1983 1988 1983 1988 with non-straw roof 22 51 94 98 51 85 99 99 with non-earth floor 11 23 12 27 15 44 14 36 with non-earth wall 22 48 85 95 51 85 69 80 who Invested In housing 60 47 68 46 rovement since 1983 g. size of housing nvestment (Yuan) a/ 4, 35 3,792 5,233 4,506 Ratio of housing Inprovement 2.71 3.57 10.88 1.82 nvestment to productive nvestment b/ I Calculated for the subsample of Individuals with housing klprovement. Calculated for whole sample. - 20 - V. A Model of Houseiold Consumption and investment Consider a household maxlmizing its utility over a two-period planning horizon. Utility Is defined over a composite consumption good (C) and over housing services (H). For simplicity, we assume separability of utillty. T - UO(CO) + Vo(Ho) + U1(Cj) + V1(Hj) (1) where T Is total utility. U and V are respectively the utilitles from composito consumption and housing services, and the numerical subscripts denote tine perlods. The time discount factor Is omitted for simplicity, as It can be embodied In the definitlon of U, and V1. The household has an Initial endowment of financlal resources Wo, which Is augmented with borrowed funds L. These resources can be used In the first period for consumption (Co), Investment in productive assets (I) and Investment In housing (h). Other Initlal endowments are capital (K0), land (Ao) and housing (Ho). These are assumed liliquld and cannot therefore be used for financing consumption or Investment. The budget constraint Is given by W9 + L a I + h + CO (2) In the second pe.-od, If no change In the land endowment occurs, the augmented capital stock (that Is, Initial capltal plus first period Investment) Is combined wlth the initlal land endowment to produce output via a neoclassical production function. Consumption In the second period Is then the value of output minus the debt repayment. However, If agriculture Is recollectivized or the land Is taken away (an event with probability P), then the farmer receives only - 21 - some fixed future Income Y, all debt Is cancelled, and production capital Is taken over by the state. 3 Second period consumption Is therefore C1 - F(Ko + l, Ao) - (1+r)*L with probability 1-P (3) C1 ' Y with probability P (4) where F Is the production function and r Is the Interest rate. Maxuization of the expected value of utility subject to equations (2), (3), (4) Is equivalent to Max UO(W0+L-1-h) + VO(HO) + (1-P) * U1l(F(K0+l,Ao) - (1+r)#L)l l,h + P*U1(Y) + V,(Ho + h) (5) First order conditions for optimum require -U + (1-P) * U; * Fk ' ° (6) 0 - UO + V; -O (7) where Fk Is the marginal productivity of capital. 3 The model could be formuWated with a less extreme scenario wnereby there Is a probability of losing only a portion of the land and a portion of capital. While the mathematics would be more tedious, the results wouWd be skiilar. - 22 - Comparative statie results generated by the model are summerized In Table 8. Proofs can be provided by the authors to Interested readers. Table 8: Comparative Statlc Results Effect on Productive Housing Investment Investment I h Change In Credit (L) + + Rlsk (P) *+ initlal Productive Capital (K)- + Initlal Housing (HO) + The effect of additional credit on both types of Investn2ent Is positive, as one would expect when the liquldity constraint Is binding. An Increased risk to land rights will lead to loss farm Investment and higher residential Investment. Higher Initial stocks of productive capital and housing have a negative direct effect on Investment In these Items, but positive cross-effects. The effect of farm size Is not easy to ascertain In the context of the present model. Typically a larger farm size Is also associated with a larger allocation of credit, thus a ceteris paribus change of farm size Is not meankigful. it can be shown that If returns to scale are Increasing, or If the marginal utility of consumption Is relativoly non-elastic, then It Is more likely that farm size Is mm , . _- * - 23 - positively related to investment. If credit and variable Input suppiles are not constrained (i.e., treating L as a decislon varlable In the preceding mathematical model), then Increasing returns to scale Imply higher capital/land ratios. However, If some Inputs are fixed (e.g., rationed fertilizers) and their supply Increases less than propoetionately with farm size such that productlon is conducted under de- facto decreasing returns to scale, then optimal capital/land ratios will decline wlth farm size even If credit Is not a constrairt. VL Econometric Results The fIrst empirical Issue which needs to be clarified relates to returns to scale. Proponents of consolidation and recollectivization assume that there are Increasing returns to scale In Chinese agriculture. Investment and productivity would thus be enhanced If typical farm sizes were larger. In order to estimate the returns to scale In agricultural production within our study areas, conventional Cobb-Douglas production functions were estimated (Table 9). The output In these functions relates to crops but excludes livestock products, as livestock activities (poultry and plgs) are not directly related to farm size. Corresponding to this notion of output, the measure of capital utilized includes the value of equipment, machinery and draft animals, but excludes the value of other livestock. The estimates indlcate that In the two counties with larger farm holdings (Gongzhuling and Xlajlang), productlon can be characterized by constant returns to scale, as the sum of productin elasticities does not differ slgnificantly from t (line (g). In the two counties with small farm sizes (Tal and Jurong) returns to scale are statistically significantly larger than 1. Tal county, with the smallest - 24 - Table 9: Estkiates of Production Functlons (Cobb-Douglas Specification) County Gongzhuling Jurong Tal Xlajlang Input (N-200) (N-190) (N-200) (N-196) (a) Land .732* .651 * .661 * .508* (14.738) (10.287) (7.924) (7.872) (b) Capital .015* .020 -.005 .067* (1.816) (1.345) (.621) (3.049) (c) Labor .055 .2850 .393* .263* (1.279) (4.735) (5.216) (4.625) (d) Nitrogen .147* .086* .044 .131 * (4.074) (2.691) (1.364) (4.382) (e) Manure .029* .015 .018 .004 (2.592) (1.190) (1.056) (.326) (f) R2 .921 .903 .945 .900 (g) Returns to .977 1.056 1.116 .975 scale (h) t value of .759 2.022* 4.08* .800 test for returns to scale dlfferent from 1 * Slgnificant at a one-tailed 95 percent confidence level. Notes: General: Other variables In the equation Include a constant, village dummy variables, an Indicator for production problems such as post, and human capital (age, education). The parameters are not presented. (a) Land Is measured as a combined area of all parcels, adjusted for quality differences through an Index based on parameters from a hedonic value equation. (b) Capital Is measured as the value of cropping-related equipment, machinery and draft aninals. (c) Labor Is measured as the actual days applied, with the weights of 1.0, .75,.5 for males, females, and children respectively. (d) Nitrogen Is measured as weight of pure nutrient. (e) Welght of manure. (g) Sum of coefficients (a)-(e). (h) Test against the null hypothesis that returns to scale are constant. - 25 - mean farm size (less than 1/3 hectare) has the highest estimated returns to scale (1.116). In Jurong county the returns to s ale are 1.056. These results are thus compatible wlth the observatlon that capital indivisibillty implies Increasing returns to scale within a range of smaller farm sizes.4 The demand for cropping- related capital In the countles with very small farm sizes Is thus smaller by more than a proportionate factor as compared to a hypothetical region with Identical agro-clmatic conditlons but larger farm sizes. Furthermore, the estimated productlon elasticity of capital In the two counties (Tal and Jurong) is not significantly different from zero. This suggests a low marginal productivity of capital In crop production, apparently due to the IndivisibIllty of capital and the very small size of farms. We proceed next to analyze the determinants of Investment in productive assets. In order to allow for a better correspondence between the tenure security variables and the type of Investment considered, only Investment In cropping-related capital Is con3idered (e.g., livestock Is excluded). The estknated model corresPonds to the reduced form of the system of first-order conditions (6) - (7) as swumarized in Table 8, with modifications to allow for a somewhat richer model. One modification Is the Incluslon of a third form of Investment, namely, non-crop-related capital, In addition to housing and crop-related capital. Its effect or crop-related Investment Is expected to be In the same directlon as housng (i.e., positive). While values of housing In the beginning of the perlod (Ho In the notation of the preceding section) were not available, Indicators of housing quality In 1982 were used to construct a composite Index of Initial house quality. 4 A study employing province level time serles data for the period 1984-1988 obtained an estimate of returns to scale of 1.07. See Lin (1989). - 26 - Other modificatlons Involve the Incluslon of several household characteristics: Family size adds to housing demand and Is thus expected to negatively affect productive Investment. The number of adults (proxi for household labor endowment) reduces the demand for hired labor and thus releases more liquldity for productive Investment when credit Is a binding constraint. Two variables wvere incorporated in the analysis to represent human capital: Education and experience In agriculture. The latter variable Is significantly correlated with age, and may thus represent several factors wlth differing Impacts on Investment (e.g., older farmers may be less inclined to Invest). Farm size was adjusted for land quality differences through a hedonic quallty index. A coefficlent greater than 1 would imply Increasing capital/land ratlos. The probability of land loss Is represented by two Indlcators of confidence In the present land aliocation system constructed from farmers' perceptlons as reported In section Ill. One Indicator relates to perceived likelihood of contract disruption In the short term, while the other refers to the likelihood of retalning the same farm In the ionger-run. Both Indicators were constructed as dummy variables where the value 1 implles more tenure security and 0 impiles less tenure security, thus the expected sign Is positive If land tenure considerations significantly affect Investment. Given the segmentation of formal and Informal credit markets as discussed In section Ill, a distinction was made In the empirical analysis between the two types of credit, so as to test whether they have separate effects on different Investments. Specifically, informal credit Is not expected to significantly affect productive investment, as It Is not typically provided for such purposes and It Is mostly not fungible. Because Institutional credit Is mostly short-term, the average - 27 - annual institutlonal borrowing by the household was utilized In the analysis, while for Informal credit the cumulative amount for the period was used. A positive coefficient Is expected If credit is a binding constraint. Two sets of estimates were obtained: OLS regresslons of crop-related capital stock, and Tobit estimates of Investment In housing. The capital regresslons, reported In Table 10, Indicate that the stock of capltal increases with farm size (parameters In all countles are slgnificantly greater than zero at a 95 percent one talled confidence level). Only In the case of Gongzhuling does the capital/land ratlo increase with farm size, as the coefficient of land Is greater than 1. However, one cannot reject the hypothesis that the capital/land ratio Is fixed In Gongzhullng, ceterls paribus (i.e., that the coefficient of land Is equal to 1). Similarly, In Tal county the hypothesis that the coefficient of land Is one cannot be rejected. In Jurong and Xiajlang counties, the coefficients of land are significantly less than 1, indicating that larger farms have lower capital/land ratios. This may be due to supplies of rationed variable Inputs (e.g. fertilizers) Increasing less than proportlonately with farm size, or due to less than full utilizatlon of capital ( a consequence of indivisibillty). The latter explanation Is compatible with the non-significant production elasticity of capital In Jurong county. Informal credit does not affect Investment In crop-related capital In any of the counties (none of the estimated parameters are statistically significant at the 95 percent confidence lovel). This Is compatible with the segmentation anu lack of fungibility characterizing this source of funding. Formal credit significantly and positively affects Investment In farm capital In Gongzhuling county, - 28 - Table 10: Regresslons of Crop Related Capital Stock Variable a\County Gongzhuling Jurong Tal Xiajiang (N-200) (N-190) (N-200) (N-196) I. Farm Size 1.548 .383 .560 .452 (4.491) (2.288) (1.795) (3.264) II. Credit Formal credit .170 -.004 .001 .004 (2.068) (.950) (.024) (.138) Informal credit .035 .001 .021 .025 (.891) (.079) (.680) (1.917) III. Security Perceptions Confidence In short-term -.121 .173 -.359 .174 (.310) (1.013) (1.136) (1.546) Confidence In long-term .382 .083 -.108 .015 (.910) (.499) (.317) (.132) IV. Initial Capital Stocks Crop related capital .182 .214 .256 .230 (1.980) (3.800) (2.549) (3.492) Non-crop capital .060 -.044 .098 -.015 (.672) (.768) (1.054) (.205) Housing quality -.219 -.160 .107 -.015 (1.446) (1.298) (1.103) (.190) V. Household Characteristics Family size .143 .234 .625 .239 (.240) (.711) (1.433) (1.322) No. of adults .052 .344 .226 .202 (.112) (1.159) (.635) (1.424) Education .026 .017 -.015 -.023 (.449) (.742) (.489) (1.547) Experlence .001 .001 .002 -.009 (.725) (.725) (.307) (1.996) R2 .446 .290 .616 .523 a/ A constant trm and village dummy variables were also Included In each regression, but are not reported here. Regresslons follow a double logarithmic speciflcatlorn. Sample slzes differ due to missing observations. - 29 - Table 11: Proportion of Farmers with Unsatisfled Input Demand (Percent) County Gongzhullng Jurong Tal Xiajlang Input (N-200) (N-1 99) (N-200) (N-200) Fertilizer 10 54 14 23 Diesel 10 33 29 67 Pesticides 1 32 23 23 Herbicides 1 28 26 29 Note: The numbers show the percentages of respondents who stated tsat they were not able to obtaln the desired quantities of Inputs even though they were willing to pay higher than market prices. Source: Feder et al., 1990. where Input supply problems are negligible and farm sizes are larger. In the other three counties formal credit does not significantly affect Investment, and It would thus seem that generally credit was not a binding constralnt on crop-related Investment In these counties. In Tal and Jurong counties, demand for Investment has likely been low due to the low marginal productivity of capital and small farm sizes. In Xlajlang county farm sizes are larger, but Inadequate Input supplies are a serous problem, apparently diminishing the profitability of, and hence the demand for, farm capital. Concerns regarding land reallocation did not hinder productive Investment signlficantly (none of the coefficients are statistically significant). This Is, apparently because farmers expected (at least at the time of the survey) that the - 30 - general household responsibillty productlon mode will prevail, and may have perceived the transaction costs of capital stock adjustment which will be Incurred upon reallocatlon to be minor. The parameter of the initial stock of capital Is significantly greater than zero and smaller than 1 In all counties. Other variables In the equations, however, are not significantly greater than zero. Tobit estimates of the parameters of demand for housing Investment are presented In Table 12. In all counties the parameter of Informal credit Is slgnificantly related to housing Investment, implying that the availability of Informal credit was a binding constraint on housing Investment. In Xlajlang county formal credit was also a significant factor affecting housing investment, and In all other counties the parameters of formal credit are positive, although not statistically slgnificant. Tenure security perceptlons do not affect housing Investment, except In Xlajiang county, where a counter-intuitive sign Is observed for the long term tenure security Indicator. As expected, initial housing quality Is significantly negataively related to housing Investment (except In Xlajlang county). Education Is positively related to housing Investment In three of the counties, perhaps because It Is a proxi for political status and better access to constructlon materlals. - 31 - Table 12: Tobit Estimates of Investment In Housing Varlable a\County Gongznuling Jurong Tal Xlajlang (N-200) (N-190) (N-200) (N.195) I. Farm Slze 8.890 -67.539 226.978 52.380 (.210) (.486) (.778) (.445) II. Credit Formal credit .612 6.843 2.965 4.559 (1.037) (1.324) (.329) (2.666) Informal credit .803 1.338 1.537 3.369 (3.272) (5.200) (5.241) (7.664) IlI.Security Perceptions Conf. In -763.506 -550.699 1233.640 -1613.080 short-term (.802) (1.407) (.432) (1.256) Conf. In 46.281 -856.118 -1570.250 3584.130 long-term (.046) (1.030) (1.187) (2.886) IV.nitlal Capital Stocks Crop related capital .008 -.731 11.811 1.770 (.016) (.882) (1.517) (2.100) Non-crop capital 7.219 .981 -1.379 4.449 (5.109) (.693) (.617) (1.583) Housing quality -1890.98 -1195.130 -1140.37 -948.705 (4.710) (1.963) (2.822) (1.148) V. Household Characteristics Family size 87.323 162.489 571.999 265.222 (.242) (.390) (1.269) (.616) No. of adults 25.197 752.566 -230.057 -89.100 (.069) (1.697) (.501) (.203) Educatlon 257.485 -209.292 400.977 353.298 (1.873) (1.826) (3.019) (2.172) Experlence -18.805 -29.328 -13.975 -29.625 (.589) (.960) (.407) (.585) Log-likelihood -1198.8 -957.2 -1374.3 -942.10 a/ A constant term and village dummy variables were Included In each equation. sample sizes differ due to missing observations. - 32 - VIL SUMARY AND CONCLUSIONS The data and analysis presented In the preceding sectlons, while pertalning to only four counties within China, generate plausible conclusions with likely applicability to other areas. The data confirm the well established upsurge In acqulsition of consumer durables and residential Investment In rural areas. It Is also demonstrated that productive Investment was substantially lower than Investment In housing, ralsing the question of whether there were factors which Inhibit productive Investment, thereby encouraging other Investments. Possible constraints which were suggested by scholars of Chinese agriculture Include the small farm sizes, Insecure land tenure and Inadequacy of financial arrangements. All of these potential constraints can be neutralized through policy, but any policy change Involves direct and Indirect costs, and It Is therefore Important to Identify which of the constraints are actually significant and under what circumstances. The analysis shows that the extremely small size of farms In some areas could become a factor hindering Investment and productivity, as the Indivisibillty of capital Introduces Increasing returns to scale. The emergence of some forms of customized draft services or shared ownership of capital assets has apparently not been sufficient to overcome the problem of Indivisibility. The Impact of Increasing returns to scale has not been reflected In the Investments in the areas characterized by very small farm sizes, because other Inputs (e.g., fertilizers, dlesel, herbicides) are possibly rationed In a manner not proportionate with farm size, or because capital Is underutilized In areas where most farms are below a threshold size. If, however, the problem of supplementary Inputs supplies Is rectified (e.g. by Introducing market mechanisms to the distribution system), then - 33 - a likely outcome Is an Investment pattern where larger farms have higher capital/land ratlos, unless the availability of financing becomes a binding constralnt. Under circumstances of inadequate variable Input supplies, the depressed demand for capital Is reflected in a lower demand for credit. Consequently, In the three study areas where Input supply problems were severe, credit was not a factor inhibiting Investment In crop-related farm capital. In the one province wlth adequate Input supplies, crop-related Investment was constrained by the supply of Institutional credit. In such an area, the estimates suggest that a doubling of the volume of Institutional credit (which Is officially Intended mostly for financing of current inputs) would increase capital stocks by 17 percent. An Increase of one dollar In the availability of formal credit to an average household would lead to 40 cents of additlonal Investment. Insecurity of land tenure, stemming from the absence of private ownership and apprehension regarding disruption of the existing land allocation, does not appear to have been a significant factor affecting Investments before 1989. However, as current land contracts were awarded for 15 years, Investments In the years past the mid-point of contract maturity (I.e., towards the mid 90's) may be more sensitive to perceptions regarding land reallocatlon. Since the data show that the majority of farmers think It Is likely that they will not be allocated the same parcels of land upon contract maturity, this Issue Is potentially significant and requires remedial policy (e.g., by extending current contract maturities several years before they expiro). A reform In the Input supply system Is likely to Increase the demand for Investment, and credit may therefore become a constralnt In areas where It Is presently not Inhlbiting Investment. The institutlonal credit system Is highly - 34 - centralized and controlled (although reforms are belng gradually Introduced), and a market-Induced supply response Is not likely. The non-Institutional credit market Is hlghly segmented. For agricultural households, most Informal credit Is obtained from relatives and friends, and Is not available for farm Investment. An Institutional credit expansion will likely need to augment an Input supply policy reform to facilitate both Increased Input use and Increased farm Investment. Farm size Is a factaor limiting Investment In areas where farms are extremely small. Solutions in the form of forced consolidatlon have serious repercusslons (e.g. undermining of tenure security perceptlons). If consolidatlon also Involves some forms of recollectivization, then serious Incentive problems will re-emerge, with significant productivity losses (Lin, 1990). Similarly, publicly maintained equipment rental or custom services have a mixed record In other countries. The Increased supply of mechanized equipment designed for small scale farms may alleviate some of the farm scale problems. Market-induced farm consolidation may emerge If alternative employment In the rural sector becomes an attractive option for agricultural households and If constraints to labor mobility are removed. Such a process would require the elkninatlon of various bureaucratic obstacles to land market transactions. Thus, while rental of land Is now allowed among Individuals, the actual legal and bureaucratic mechanisms to facilitate efficlent land markets have not been Implemented, and apprehension regarding land contract cancellation upon renting-out may still be a factor. - 35 - References Feder, Gershon, Lawrence J. Lau, Justin Y. Lin and Xiaopeng Luo, "Agriculture Credit and Farm Performance In China", Journal of Comparative Economics, 13, December 1989. Lin, Justin Y., "The Household Responsibillty System in China s Agricultural Reform: A Theoretical and Empirical Study". Economic Development and Cultural Change, Vol. 36 No. 3 (supplement), April 1988. Lin, Justin Y., "Rural Reforms and Agricultural Productivity Growth In China", Working Paper, Beijing University, November 1989. Lin, Justin Y., "Collectivization and China's Agricultural Crisis in 1959-1961", Journal of Political Economy, 1990 (forthcoming). McMilan, John, J. Whalley and Li Jing Zhu, "The Impact of China's Economic Reforms on Agricultural Productivity Growth", Journal of Political Economy, 1989. Tam, On-Kit "Rural Finance in China", China Quarterly, 29, March 1988. World Bank, China: Finance and Investment, Washington, D.C., 1988a. World Bank, Rural Sector Adjustment Loan, Loan Implementatlon Volume II, 1988b. PRE Workino Papgr Series Contact iL Ahor Sk olp5 WPS455 A Formal Estimation of the Effect Junichi Goto June 1990 M. T. Sanchez of the MFA on Clothing Exports 33731 from LDCs WPS456 Improving the Supply and Use of S. 0. Foster June 1990 Z. Vania Essential Drugs in Sub-Saharan Africa 33664 WPS457 Financing Health Services in Africa: Germano Mwabu June 1990 Z. Vania An Assessment of Alternative 33664 Approaches WPS458 Does Japanese Direct Foreign Kenji Takeuchi June 1990 J. Epps Investment Promote Japanese Imports 33710 from Developing Countries? WPS459 Policies for Economic Development Stanley Fischer June 1990 WDR Office Vinod Thomas 31393 WPS460 Does Food Aid Depress Food Victor Lavy July 1990 A. Murphy Production? The Disincentive 33750 Dilemma in the African Context WPS461 Labor Market Participation, Returns Shahidur R. Khandker July 1990 B. Smith to Education, and Male-Female 35108 Wage Differences in Peru WPS462 An Alternative View of Tax Anwar Shah July 1990 A. Bhalla Incidence Analysis for Developing John Whalley 37699 Countries WPS46;. Redefining Government's Role in Odin Knudsen August 1990 K. Cabana Agriculture in the Nin .ies John Nash 37946 WPS464 Does A Woman's Education Affect Shoshana Neuman August 1990 V. Charles Her Husband's Earnings? Results Adrian Ziderman 33651 for Israel in A Dual Labor Market WPS465 How Integrated Are Tropical Timber Panos Varangis August 1990 D. Gustafson Markets? 33714 WPS466 Is There An Intra-Household Lawrence Haddad August 1990 J. Sweeney Kuznets Curva? Ravi Kanbur 31021 WPS467 Structural Adjustment and Living Nanak Kakwani August 1990 B. Rosa Conditions in Developing Countries Elene Makonnen 33751 Jacques van der Gaag WFS468 Does the Structure of Production Indermit Gill August 1990 M. Abundo Affect Demand for Schooling in Peru? 36820 PBE WorkqnPerSLS Contact T ALAhon , for gpge WPS469 Modeling Economic Behavior in J. Barry Smith August 1990 M. Abundo Peru's Informal Urban Retail Sector Morton Stelcner 36820 WPS470 What Do Alternative Measures of Alexander J. Yeats August 1990 J. Epps Comparative Advantage Reveal 33710 About the Composition of Developing Countries' Exports? WPS471 The Determinants of Farm Gershon Feder August 1990 C. Spooner Investment and Residential Lawrence J. Lau 30464 Construction in Post-Reform China Justin Lin Xiaopeng Luo WPS472 Gains in the Education of Peruvian Elizabeth M. King August 19_, C. Cristobal Women, 1940 to 1980 Rosemary Bellew 33640 WPS473 Adjustment, Investment, and the Riccardo Faini August 1990 R. Sugui Real Exchange Rate in Developing Jaime de Melo 37951 Countries WPS474 Methods for Measuring the Effect Anne Maasland August 1990 P. Dixon of Adjustment Policies on Income 39175 Distribution WPS475 Does Divestiture Matter? A Ahrned Galal August 1990 G. Orraca-Tetteh Framework for Learn:ng from Experience 37646