81680 CASE STUDY WORLD BANK | AGRICULTURE AND DEVELOPMENT AUGUST 15, 2013 Are Mega-Farms the Future of Global Agriculture? Exploring the Farm Size-Productivity Relationship for Large Commercial Farms in Ukraine* KLAUS DEININGER, DENYS NIZALOV, AND SUDHIR K. SINGH Ukraine provides an interesting setting to study both phenomena. With vast areas of fertile black soils, Ukraine has traditionally been a regional breadbasket. It is one of few countries that have seen rapid expansion of large farms using modern technology and management over the last decade. While tax and quota policies may favor larger farms, Ukraine lacks farm price support and subsidy policies of developed countries that arguably depress farm sizes artificially (Adamopoulos and Restuccia 2011). It is home to some of the largest farms on earth and has seen considerable land concentration; the country’s 40 largest agri-holdings are estimated to ABSTRACT control 4.5 million ha or 13.6% of cultivated —————–——————————————————————–———————— area (Lissitsa 2010), a fact interpreted by With farms cultivating tens or hundreds of thousands of hectares, Ukraine’s supporters of large scale agribusiness as example is often used to demonstrate the existence of economies of scale in modern grain production. Panel data analysis for all the country’s farms above evidence of such farms’ superior economic 200 ha in 2001-2011 suggests that higher yields and profits are due to unobserved performance. There are also vast differences in factors at rayon (district) and farm level rather than economies of scale. social structure and the nature of economic Productivity growth was driven not by farm expansion but by exit of unproductive activity across Ukraine’s regions. The land and entry of more efficient farms. Higher initial shares of area under farms above reforms of 1999-2001 allow exploring the 5,000 ha at rayon level significantly reduce subsequent exit and entry, suggesting extent to which initial differences in agrarian that excessive land concentration reduces productivity growth in the long run. structure matter. Implications for global evolution of farm structures are drawn out. BACKGROUND from new technology – could make reliance DESIGN on large commercial farms a viable Recent trends in soft commodity prices and development path for land-rich countries The basis for our empirical work is a expected increases in demand for food, (Collier and Venables 2011). depersonalized data collected annually from fiber, and fuel has led to a revival of interest 2001 to 2011 by Ukraine’s State Statistics in agricultural production and agrarian Whether an agrarian structure based on large Committee (through form 50SG) for the structure. Earlier, evidence of superior farms can be economically and socially universe of roughly 10,000 large commercial economic performance of owner-operated desirable hinges on two issues. One is farms in the country each year. We construct farms and the drawbacks of a highly whether the new technology has changed the consistent series for area cultivated and output unequal distribution of land assets provided negative farm size-productivity relationship value in 2010 US$, from crop production for 11 the basis for near-unanimous support of found in much of the traditional literature, crops and cost of key inputs. We restrict our smallholder-based strategies as the most leading to economies of scale in production. A sample to farms above 200 ha as they were effective way of expanding agricultural second concern is whether initial land not affected by any changes in reporting production. Now, influential voices argue concentration will affect subsequent economic requirements. Over the whole period, we have that increasing returns to scale in performance and what channels may be 89,736 observations for a panel of 16,191 agricultural production – partly resulting involved. farms registered outside cities. ECON.WORLDBANK.ORG/RESEARCH 2 ARE MEGA-FARMS THE FUTURE OF GLOBAL AGRICULTURE? These data allow exploring two types of farms decreased from 21 mn. ha in 2001 to Table 1: Overall Changes in Yields of Major questions: (i) links between productivity and 15 mn. ha in 2006 before recovering to Crops (kg/ha) farm size and (ii) impacts of initial farm 18 mn. ha in 2011. While average national 2001 2006 2011 structure on subsequent entry, exit, and farm size increased from 2,061 to 2,305 ha, Corn 3,305 3,298 5,892 productive performance. The wide variation the median area farmed decreased, from in initial inequality of operational farm sizes 1,625 to 1,429 ha. Inequality grew Soybean 1,081 1,060 1,914 are measured as the share of area significantly with the Gini of the operational Sunflower 942 1,424 1,948 cultivated by farms greater than 5,000 ha, farm size distribution shifting from 0.32 to Changes in Farm Structure across the 472 rural rayons (districts). 0.47. The almost 5-fold increase in area of Area cultivated Using this measure, we explore the extent farms above 10,000 ha during this period 20.59 15.19 17.61 (mn ha.) to which structural factors affect productivity was a key factor underlying this trend. Avg. farm size 2,061 2,222 2,305 indirectly by having an impact on the rate of (ha) entry and exit once other factors are Data on cost and output per ha for pre- and Median farm 1,625 1,630 1,429 accounted for. We use a farm-level post-2006 periods in Figures 1.1 and 1.2 size (ha) production function to investigate the suggest that in the first period, profitability Land Gini 0.316 0.352 0.473 presence of economies of scale and rayon- was limited and few super-large farms Area under farms 654,755 1,338,368 3,437,111 level analysis to identify the initial land existed. After 2006, profitability improved, > 10,000 ha concentration effect. with profits maximized at farm sizes of 2,000 Maximum farm 88,032 88,751 142,014 to 3,000 ha. This casts doubt on the notion size (ha) DESCRIPTIVE EVIDENCE that superior productive performance was Source: Own computation from Form 50. the main driver of the expansion of super- Detailed panel data illustrate three features large farms. RESULTS of agricultural development in Ukraine over 2001-2011. First, yields grew rapidly after Table 2 displays levels of productivity by Two sets of main analytical results were 2006, with sunflower, corn, and soybean farms in the initial sample that either stayed produced. First, once rayon- or farm-level yields almost doubling, prompting a marked or exited and new entrants. Entry and exit fixed effects are accounted for, the production shift to oilseeds (Table 1). Second, the seem closely linked to productive technology exhibits constant returns to scale. transformation of agricultural sector was performance; the incumbents who exited Second, initial land concentration affected more due to new entry than to existing farm had negative profits per ha (US$ -11.3), subsequent performance: a higher share of growth. Most entrants cultivated farms compared to US$ 35 for stayers and land cultivated by farms above 5,000 ha at 1,000-3,000 ha in size, which are large by US$ 65.5 for entrants. Entrants had higher rayon level in 2001 is associated with lower European standards but not super-large. cost, suggesting use of more input-intensive levels of exit and entry and productivity Finally, although land sales have not been technology. This is consistent with growth. allowed, there was massive concentration successful entrants employing less than of operational holdings via land rental average labor in both periods. Table 3 reports results for the nearly 90,000 agreements: area farmed by units above observations for the naïve cross section 10,000 ha expanded by more than 2 mn. ha Compared to the increase in per ha cost, (col. 1) and specifications with rayon (col. 2) (or 10% of the total) in the 2006-11 period. mean initial area operated by entrants and farm fixed effects (col. 3). While cross decreased over time from 1,950 ha to sectional estimates point towards significant Yield growth and shifts in output 1,445 ha. The fact that their size in both increasing returns to scale in the first period, composition prompted far-reaching periods was quite small casts further doubt when a doubling of inputs is predicted to be changes in agrarian structure. As Table 1 on the validity of the narrative of relentless associated with a 1.17 increase in output, the illustrates, the area cultivated by sample farm growth driven by economies of scale. corresponding figure was much lower (1.06) in Figure 1.1. Value of output and cost/ha against farm size, 2001-2006 Figure 1.2. Value of output and cost/ha against farm size, 2007-2011 3 ARE MEGA-FARMS THE FUTURE OF GLOBAL AGRICULTURE? Table 2: Descriptive Statistics by Type of Exit and Entry Period All Exits Stayers Entrants Profit/ha (US$) 2001/11 27.7 -11.3 35.3 65.5 2001/06 -2.4 -12.5 6.7 1.6 2007/11 69.8 15.8 77.5 96.1 Output/ha (US$) 2001/11 250.4 134.3 294.3 354 2001/06 179.6 126.1 203.8 239.1 2007/11 367.1 232.7 393.9 415.2 Cost/ha (US$) 2001/11 222.7 145.5 259 288.5 2001/06 182 138.6 197.1 237.4 2007/11 297.3 216.9 316.4 319.1 Area at start (ha) 2001/11 1911 1829 2627 1688 2001/06 2036 1801 2363 1950 2007/11 2017 1488 2453 1445 Workers per 100 ha 2001/11 3.36 4.28 3.61 2.29 2001/06 4.14 4.34 4.34 3.45 2007/11 2.28 2.7 2.34 1.81 Figure 2. Farm and rayon fixed effects by farm size (cities & rayon Source: Own computation from form 50SG centers excluded), 2001-2011 the second period. This is in line with the determinant of structural change as and entry by 3.2 percentage points, notion that market imperfections could only efficiency of incumbents who remained in the respectively. The higher the productivity of be overcome by very large sizes in the sample first-order dominates that of exiters farms above 5,000 ha, the lower the share early post-liberalization stages. Once farm- while itself being first-order dominated by of land vacated by exiters. fixed effects are included to account for permanent (though not all) entrants. unobserved differences in managerial Regression results for areas newly ability, the hypothesis of constant returns to As to the long-term land concentration effect, occupied by entrants are different from scale can no longer be rejected at 10% Table 4 reports regression results for rayon- those for exit in the following respect. The statistical significance level in both periods. level area subject to entry or exit. The interaction between large farms’ area share This near-disappearance of ‘economies of dependent variables are rayon area shares and their initial level of productivity is scale’ implies that large farms’ emergence exited from production permanently (col. 1) insignificant, consistent with the notion that, or prevalence may be explained by a or overall (col. 2) and brought in production if area shares are accounted for, large combination of three factors: (i) large farms’ permanently (col.3) or totally (col. 4). Initial farms’ initial productivity does not affect use of superior management skills; (ii) their levels of farm-level productivity are very entry. location in rayons better endowed with key significant and negative throughout, in line factors (e.g. soil quality, infrastructure); and with the notion that exit was less likely in To the extent that they affect entry and exit, (iii) benefits from the ability to vertically rayons where farms were more productive to pre-existing structural differences should integrate beyond the production stage, deal start with. The magnitude of the estimated also have an impact on the overall with factor market imperfections, and land concentration effects is economically productivity measured as crop output per possibly exercise market power. significant: an additional 10% of a rayon’s hectare. As seen in Table 5, holding other area under farms above 5,000 ha is factors constant, higher levels of initial land Separating out farm- and rayon-specific predicted to reduce subsequent exit by 2.7 concentration indeed reduce subsequent fixed effects allows comparing relative Table 3: Cobb Douglas Production Function Estimates magnitudes of location-specific factors vs. Log of output (in USD) farm-specific management. Figure 2 shows Pooled OLS Rayon FE Farm FE that if farm-level fixed effects proxy for Area 0.212*** 0.236*** 0.341*** managerial ability as an indivisible factor, Area*post 2006 -0.048*** -0.02 -0.050*** better managers operate larger farms. In Labor 0.222*** 0.189*** 0.168*** addition, the absolute magnitude of rayon- Labor*post 2006 -0.137*** -0.104*** -0.051*** level effects exceeds that of farm-fixed Seed 0.154*** 0.163*** 0.139*** effects for farms above roughly 1,500 ha. In Seed*post 2006 0.091*** 0.069*** 0.049*** other words, large farms locate in rayons Fertilizer 0.117*** 0.121*** 0.073*** with more favorable endowments and Fertilizer*post 2006 0.012** 0.010** 0.015*** benefits from the associated location- Fuel and energy 0.319*** 0.240*** 0.162*** specific rents exceed those from Fuel and energy*post 2006 -0.104*** -0.061*** 0.006 managerial efficiency. Other – agricultural services 0.076*** 0.048*** 0.043*** Other – agricultural services*post 2006 0.029*** 0.049*** 0.042*** Displaying farm fixed effects separately for Observations 89,736 89,736 89,736 those who entered, exited, and remained in Sum of coefficients period 1 1.171*** 1.062*** 0.991 the sample in Figure 3 illustrates the Sum of coefficients period 2 1.056*** 1.019*** 0.999 contribution of entry and exit to productivity Note: Statistical significance is indicated by the following: ***p<0.01, **p<0.05, *p<0.1. dynamics. Entry appears to be a key Deviation from 1 is tested for the sum of coefficients. 4 ARE MEGA-FARMS THE FUTURE OF GLOBAL AGRICULTURE? Table 4: Rayon-level Regressions for Aggregate Entry and Exit (2002-2011) Share of area exiting Share of area entering permanent churn permanent churn Init. share of cult. area > -0.273*** -0.360** -0.315** -0.385** 5000 ha Initial productivity -0.797*** -1.104*** -0.503** -0.596*** Init. share > 5000 ha -0.701*** -0.841** -0.145 -0.188 *productivity Rayon fixed effect 0.270*** 0.583*** 0.774*** 0.994*** Observations 472 472 472 472 R-squared 0.222 0.219 0.246 0.253 Note: Region dummies and constant included in all regressions but not reported; Statistical significance is indicated by the following: ***p<0.01, **p<0.05, *p<0.1. Figure 3. Farm fixed effect by nature of entry / exit productivity, with the size of estimated Table 5: Evidence of Convergence in Productivity at Rayon Level by Quintiles effect fairly similar across categories. Percentile Also, while the negative coefficient on initial 10th 25th 50th 75th 90th productivity points towards convergence of Init. share > 5000 ha -0.363*** -0.432*** -0.405*** -0.402*** -0.492*** productivity levels across farms, there is Initial productivity -0.414*** -0.530*** -0.569*** -0.553*** -0.785*** greater dispersion in the bottom of the Init. share > 5000 ha * 0.448 0.149 0.685 0.156 0.117 productivity distribution as compared to the top. Finally, Rayon fixed effect 0.977*** 0.847*** 0.628*** 0.314** 0.671*** levels of individual productivity are higher in Note: Statistical significance is indicated by the following: ***p<0.01, **p<0.05, *p<0.1. rayons with positive productivity shocks (e.g. better infrastructure). in agriculture may be a viable development Ukraine to base policy proposals on strategy. The large and robust negative empirical evidence rather than ideology or CONCLUSION effect of the share of land initially held by corporate interests. large farms on subsequent exit and entry Our study contribute to two strands of suggests that the initial agrarian structure REFERENCES literature. First, we extend evidence on the can affect development in the long term. farm size-productivity relationship, which Efforts to explore the underlying economic Adamopoulos, T. and D. Restuccia. 2011. hitherto has been largely restricted to labor and political channels in detail may help "The size distribution of farms and intensive technologies in relatively land- identify ways to close productivity gaps and international productivity differences." scarce environments, to a setting with foster greater convergence across Ukraine’s Department of Economics Working Paper abundant land and capital-intensive regions. Beyond Ukraine, the evidence of 441. Toronto: University of Toronto. technology. The hypothesis of economies land concentration causing negative of scale in agricultural production is externalities holds lessons for policy makers Collier, P. and A. J. Venables. 2011. "Land rejected even in this environment. Instead, in other countries seeking to promote rapid Deals in Africa: Pioneers and Speculators." large farms’ superior performance appears agricultural growth. Discussion Paper 8644. London: Centre for to be due to unobserved rayon- and farm- Economic Policy Research. specific attributes. Decomposition of fixed From a methodological perspective, the effects to assess the underlying factors in study shows that the dearth of studies to Lissitsa, A. 2010. "The emergence of large detail will be a key area for follow-up guide policy-making in Ukraine is not due to scale agricultural production in Ukraine: research. absence of data but the lack of capacity to Lessons and perspectives." Washington DC: process and link available information. Paper presented at the Annual Bank Second, concerning the literature on Addressing this gap could help support Conference on Land Policy and transition and agrarian structure, our research on a number of key topics including Administration, April 26 and 27, 2010. findings suggest that a causal interpretation an exploration of the extent to which smaller of the temporal coincidence of Ukraine’s farms' growth is hampered by factor market * * * productive recovery with the growth of imperfections (and how they can be over- *Full text of report is available at: mega-farms may be a fallacy. Instead, exit come); land valuation and the extent to http://go.worldbank.org/A5D4SB2IU0 of inefficient farms during the early years of which there is an under-investment in long- reform and the space this created for entry term agricultural assets (incl. soil fertility); This case study was prepared by Klaus of more efficient ones emerge as key and the links between agricultural and non- Deininger from the Development Economics drivers of higher agricultural productivity. It agricultural growth at rayon level. This could Research Group (DECRG), Denys Nizalov of implies that reduction in entry/exit barriers help inform the debate on the future of Kyiv School of Economics/ Kyiv Economics agricultural development and land markets in Institute, and Sudhir K. Singh of DECRG. The findings, interpretations, and conclusions expressed are entirely those of the authors. They do not necessarily reflect those of their respective institutions.