The allocative efficiency of material input use in Russian agriculture.

During Russia's transition from a planned to a market economy, the volume of material inputs used in agriculture has declined severely. From 1990 to 2002, use of mineral fertiliser, other soil-enhancing chemicals (such as lime), and gasoline all fell by over 80 percent, while use of electricity

METHODOLOGY AND DATA

We assess the allocative efficiency (AE) of input use by comparing the price of input i ([P.sup.i]) with the value of its marginal product in producing commodity j (VM[P.sup.i.sub.j]). If [P.sup.i] > (<) VM[P.sup.i.sub.j], the farm should decrease (increase) the use of input i in order to improve allocative efficiency (and profit). We accordingly present the results of AE analysis in terms of the AE ratio, which is calculated as the ratio of VM[P.sup.i.sub.j] to [P.sup.i]: when the AE ratio is less than 1, the input's use should be reduced; when the AE ratio is greater than 1, the input's use should be increased. All marginal product estimates used in the AE tests are from econometrically estimated generalized Cobb-Douglas functions explaining the value of farm output. For each output/input combination, multiplying the estimated coefficient in the Cobb-Douglas function by the (geometrical) mean product of the input gives the input's estimated VMP (the methodology of AE analysis is briefly summarised by Grazhdaninova and Lerman, 2005).

This article provides a summary overview of the AE calculations carried out by various researchers as part of the BASIS Russia project) We only report the AE results for use of material inputs--fertiliser, seeds, animal feed, fuel, electricity, spare parts, and services; corresponding results for land and labour are reported in Grazhdaninova and Lerman (2005). All the AE calculations examined, with one exception, pertain to the behaviour of the large corporate farms that succeeded the former collective and state farms during transition. The exception is the work by Sazonov for Tarnbov Oblast, which covers the activities of private family farms (so-called peasant farms) that have arisen during transition, rather than the former state and collective farms (also see Sazonov and Sazonova, 2005).

The calculations vary by researcher with respect to (1) the regions covered; (2) the data used; (3) the specific output/input combinations examined. Uzun covers all of Russia, using a Goskomstat database of the annual reports of some 24,000 corporate farms. Liefert et al. also cover all of Russia, but they use aggregate national statistics and examine only the AE of fertiliser to produce grain. Grazhdaninova and Lerman cover the three oblasts where data were collected for 144 corporate farms as a part of the 2003 BASIS survey--Rostov, Ivanovo, and Nizhnii Novgorod. Epshtein's analysis is specific to Leningrad Oblast, and it is based on the annual reports of all 195 corporate farms in that region. Sazonov's analysis relies on a survey of 56 peasant farms in Tambov Oblast.

Some researchers examine the value of total farm output, including both crop and livestock production in essentially diversified Russian farms (Uzun, Epshtein, in part Lerman). In these analyses, the explanatory variables represent the total use of inputs, expressed in monetary values or in physical quantities. Other researchers analyze only crop or livestock production (Lerman, Sazonov (3)), or focus even more narrowly on grain, recognising that this commodity is the main consumer of fertilisers (Liefert et al., Grazhdaninova). In these cases, the use of inputs is specifically restricted to the enterprises or commodities analysed, while the output is expressed in value terms (aggregate crops or livestock) or in physical quantities (grain).

The data for input and output prices used in the AE tests pertain to the specific regions covered in each analysis. The price data used by Grazhdaninova and Lerman are from the 2003 BASIS farm survey. Epshtein and Sazonov use Leningrad and Tambov oblast prices, respectively. Uzun and Liefert et al. use national prices published by Goskomstat for the agricultural sector as a whole.

ALLOCATIVE EFFICIENCY RESULTS

Table 1 gives AE calculations for material inputs in the aggregate. In these analyses, the input variable is the total cost of all material inputs (fertiliser, seeds, animal feed, fuel, electricity, spare parts, and services) used to produce the output. The input cost is expressed in rubles, as reported by the farms in their annual reports. The AE ratio is less than 1 in all cases. The results thus strongly support a conclusion of overuse of inputs. Based on these results, it would be efficient and profit-maximising for farms to reduce their purchase and use of material inputs.

Table 2 presents AE results specifically for fertiliser use. The input variable is generally expressed in toils of fertiliser. Despite the conclusion from Table 1 that material inputs in the aggregate are overused, fertiliser is underused. Although Grazhdaninova's AE calculations are below 1, all the other calculations, including Uzun's that cover all corporate farms in Russia, are decisively greater than 1. In fact, Uzun computes a ratio of 6.3, which suggests that fertiliser use at the margin was earning farms revenue more than six times the price farms paid for the fertiliser. Lerman calculates an even higher ratio of 7.47, when fertiliser and seeds are the only two material inputs specified in the production function, and seeds are measured in rubles and fertiliser in physical units. This high number, however, sits oddly with Lerman's other calculation of 1.47 when the only difference is that both fertiliser and seeds are measured in value terms (rubles). Yet, even if Uzun's and Lerman's calculations are exaggerated outliers, the results support the general conclusion that fertiliser is underused.

TECHNICAL BIASES IN ALLOCATIVE EFFICIENCY CALCULATIONS

Because of data and measurement issues, however, the AE calculations in Tables 1 and 2 likely suffer from a number of biases, some of which push the results toward input underuse, while others bias them toward overuse. One likely measurement bias stems from the 'management bias' problem (Mundlak, 1961). Farm managers vary in ability, with the more capable generating a higher VMP from inputs. This results in them using more inputs in production. A cross-sectional regression will therefore overstate the input's VMP on an average farm. The reason is that the estimated output elasticity measures the gain in output in better managed farms compared to poorly managed ones, and incorrectly attributes the higher output to greater use of inputs rather than correctly crediting it to superior management. As a consequence the VMP estimated from the production functions will overstate the actual VMP. This will bias the AE calculations upward and thereby toward input underuse.

Another bias, again toward input underuse, results from the fact that most of the studies are specific to the year 2001, which was a good weather year for grain production, with output at 82 million metric tons (mmt), compared to average annual production over 1996-2003 of 68 mint (USDA, 2004). If the results for 2001 were used to assess performance for a broader period of time around this year, the relatively high output of grain in 2001 would generate unrepresentative estimates of inputs' VMP that are biased upward. The upward weather bias affects particularly the results covering crops, and most strongly the results specific to grain. It is therefore more relevant for crop-specific inputs, such as fertiliser.

The two issues just examined--management bias and favourable weather in the period of analysis--both bias the AE calculations toward input underuse. Correcting for management bias and the 2001 weather bias may reduce the AE ratios in Table 2 and thus make the results of fertiliser underuse less prominent (or even eliminate them altogether). The downside of this correction is that the AE ratios in Table 1 will become even smaller, thus strengthening the evidence of overuse of all material inputs in the aggregate.

The AE calculations suffer from a third bias, though in this case in the direction of input overuse. With the exception of Sazonov who covers peasant farms, all the studies cover operations by former state and collective farms. All households on these corporate farms continue to operate their own plots, which average 0.4 ha in size. While the corporate farms produce most of Russia's bulk crops (grain, oilseeds, sugar beet), the household plots account for over half of the country's output of meat, potatoes, and vegetables, and about half of milk production (Uzun, 2005). One reason for the plots' striking performance is that the plotholders receive inputs from their parent farms, either as legitimate in-kind payment or in some cases as stolen goods. Inputs commonly transferred to plot use in this way are animal feed, use of vehicles by households to truck output to farmers' markets for sale, and fuel for the vehicles. Some fertiliser and seeds are probably also transferred from the corporate farm to the use of household plots.

All the researchers who estimated production functions for corporate farms intended that the functions be based only on output produced by the farms' corporate operations, and only on those inputs actually used in farms' corporate operations (thereby excluding output and input associated with household plot production). Yet, given that much of the input transfer to plots is unofficial, separating out input use between these two channels is virtually impossible. This input transfer distorts the AE calculations for material inputs. The reason is that the bulk, if not all, of the transferred inputs are included in the estimation of farms' production functions, which exclude output on the household plots. If the inputs transferred to household plot use were also excluded in the estimation, the inputs' VMP would be higher.

As mentioned before, the household plots specialise in production of livestock products, potatoes, and vegetables. The downward bias in the estimated VMP for material inputs is therefore particularly strong for inputs used heavily in livestock production, such as animal feed, and for production functions whose output is livestock products. In crop production, which is the main specialisation of corporate farms, the bias towards input overuse from not accounting for the transfer of inputs away from farms' corporate operations to household plot use has the effect of balancing the bias toward input underuse resulting from the management bias problem and favourable weather. Taking all these biases into consideration, therefore, does not alter the general conclusion of Table 1 that material inputs in the aggregate appear to be overused.

As for the use of fertiliser specifically, the underuse biases outweigh the overuse biases, as the latter are mainly relevant for livestock-specialised household production. For these reasons, we can conclude that the AE calculations for fertiliser are biased in the direction of underuse, and actual allocative efficiency of fertiliser use is better than the calculations indicate. Accepting that the AE calculations for fertiliser are biased towards underuse also reduces any inconsistency between the finding that fertiliser is underused, while material inputs in the aggregate appear to be overused.

REASONS FOR ALLOCATIVE INEFFICIENCY

One possible reason for the allocative inefficiency of input use is poor decision-making by farm managers. During central planning, farm managers' objective was not profit maximisation, but rather achieving mandated output targets. Such a goal encouraged managers to lobby for low-output targets and high-input allocations, while being conservative with production techniques so as not to jeopardise meeting short-run output targets. Managerial incentives did not encourage initiative and flexibility in changing output/ input combinations and production methods to achieve maximum output from a given mix of inputs. These attitudes still persist, and farm managers have probably not yet fully developed the mentality and skills necessary to achieve maximum allocative efficiency in input use.

Another reason allocative inefficiency might exist, especially in the form of input underuse, is financial constraints, such that farms lack the working capital, financed either from their own revenues or credit, to purchase inputs. A well-operating commercial credit system does not yet exist for Russian agriculture and access to credit is limited (Subbotin, 2005; Yastrebova, 2005). If production cycles are short, then a situation where an input's VMP exceeds its purchase price generates more revenue than cost and the input purchase becomes self-financing. In agriculture, on the contrary, inputs must be purchased months before output is ready for sale, and the unavailability of credit or working capital can preclude obtaining inputs whose acquisition might be profitable over the full production cycle.

For tradable inputs such as fertiliser and fuel, input underuse could exist for the reason that, even though farms might be willing to pay higher prices for the inputs, suppliers are averse to selling at even greater prices. The most likely reason for this behavior is that Russian fertiliser manufacturers receive higher prices if they export their output than if they sell to domestic users. In 2000, fertiliser manufacturers exported nitrogen and potash fertiliser at average prices of 1,738 and 2,598 rubles/ton, respectively (Customs, 2001), compared to only 1,201 and 1,471 rubles/ton received from domestic sales (Goskomstat, 2002). Another export incentive is that input manufacturers can keep, and invest, their hard currency earnings abroad. Responding to these higher export prices, since the mid-1990s the Russian fertiliser industry has been exporting about 80% of its output, mostly to West European countries, and the use of fertiliser per hectare of sown land has dropped by more than 80% (Serova and Shick, 2005).

The severe drop in fertiliser use can explain why the VMP of fertiliser is now relatively high, while the disparity between domestic and world trade prices for fertiliser can explain why Russian farms have difficulty obtaining more fertiliser to close the gap between its VMP and domestic purchase price. To get fertiliser, Russian farms usually need the help of regional government, which 'sells' fertiliser to farms at attractively low prices in return for the farms' commitment to sell back their output, or at least sell within the region.

Russian domestic fertiliser markets, therefore, reflect not only a disequilibrium between input prices and the input VMP, but also the fact that domestic prices differ so strongly from world trade prices. While the AE ratio of fertiliser use calculated at domestic prices is 2.16 (see Liefert et al. in Table 2), the same calculation using the trade prices for fertilizer and grain equals 0.99. This indicates that from the point of view of world trade prices, rather than domestic prices, corporate farms were using fertiliser in the production of grain at the optimal level of allocative efficiency. In other words, if Russian farms had paid world prices for fertiliser rather than low domestic prices, in 2000 they would have purchased almost exactly the amount of fertiliser they were actually using (in their grain-producing activities). The low domestic prices farms pay for the fertiliser they receive explains the disequilibrium that exists between fertiliser's VMP and domestic purchase price. From the point of view of world prices, fertiliser is not underused. (4)

The policy options for improving the allocative efficiency of input use follow closely from the examination of why current input use might be inefficient. One option is to improve farm management, especially in the area of economic decision making. Raising the allocative efficiency of input use requires that managers flexibly alter output levels and the mix of inputs used in production in response to changes in output and input prices.

Another policy option is to decrease, or even wholly eliminate, the gap that exists between world market prices for exportable agricultural inputs such as fertiliser and fuel and the inputs' domestic purchase prices. By reducing the incentive of input producers to export rather than sell to domestic users, such a policy change would encourage the commercial and institutional development of domestic input markets. Narrowing the price gap would also improve the allocative efficiency of input use, with prices for tradable inputs now being determined by their more meaningful opportunity cost values--world trade prices. On a broader level, it would increase Russia's economic gains from trade according to comparative advantage and greater integration into world markets.

Although benefiting the economy at large, such a change would not be in the narrow interests of the agricultural sector. The jump in domestic purchase prices for such inputs would increase farm costs (without necessarily affecting fertiliser use, as demonstrated above). Since economic reform began in the early 1990s, agricultural producers' terms of trade--the prices received for output compared to prices paid for inputs--have deteriorated substantially. The reason is that the controlled price system of the Soviet period kept input prices low relative to output prices. Such price-setting behaviour means that in the pre-reform period producers were subsidised not only through direct budget subsidies, but also indirectly through the price system. When price and trade liberalisation began in the early 1990s, prices for inputs rose to levels closer to their real cost of production, and closer to world market values. The increase in real (or relative) prices of inputs is a major reason why purchase and use of inputs by farms have declined so substantially. Yet, domestic prices for tradable inputs, such as fertiliser and fuel, still remain below world market levels, which suggests that the transition-driven rise in their real prices has not yet ended. This process may be accelerated by Russia's negotiations for accession to WTO.

CONCLUSION

The results indicate that from the point of view of allocative efficiency, material inputs in the aggregate in Russian agriculture have been overused, though fertiliser has been underused. However, the efficiency calculations for fertiliser are biased toward underuse, the main reason being that the year for which most studies were done--2001--was a good weather year for Russian grain production. This overstates inputs' estimated marginal product, and consequently biases the AE calculations for fertiliser and other inputs used mainly in grain production toward underuse. This qualification reduces any inconsistency in the results that material inputs in the aggregate are overused, while fertiliser appears to be underused.

The results support the argument that the severe drop in use of material inputs in Russian agriculture during transition has been economically rational. As such, the results do not provide economic justification for the appeal by Russian farms that the government take action to correct the decline in input use. In fact, in coming years Russian domestic prices for exportable inputs, such as fertiliser and fuel, are more likely to rise rather than fall in real terms, moving closer to world market prices. Farms would respond to the price hikes by further reducing their purchase and use of these inputs.

Narrowing or eliminating the gap between domestic and world market prices for tradable inputs is therefore a policy the government could adopt to improve the economy' overall allocative efficiency of input use, as it would result in input prices being determined by their most meaningful opportunity cost values--world market prices. Yet, the government would have to face the consequences that the price increases would further hurt farms, decreasing both their output and profit (reducing input use would minimise the financial harm suffered by farms from the price hikes, but farm profit nonetheless would fall). Other policy options for improving input allocative efficiency include upgrading the quality of farm management and developing a system of farm finance, which would provide farms with working capital to buy inputs.

Table 1: Allocative efficiency calculations for material inputs in
aggregate

Researchers     Output (a)            Regions

Uzun            All agriculture       All Russia
Lerman          All agriculture       Three survey regions (c)
Lerman          Crops                 Three survey regions
Lerman          Livestock products    Three survey regions
Epshtein        All agriculture       Leningrad
Epshtein (d)    All agriculture       Leningrad
Sazonov         Crops                 Tambov

Researchers     Time period    AE ratio (b)

Uzun                2001          0.80
Lerman              2001          0.55
Lerman              2001          0.70
Lerman              2001          0.44
Epshtein            2001          0.92
Epshtein (d)        2001       0.94/0.92
Sazonov           2001-02         0.65

(a) Produced by corporate farms, with exception of Sazonov, who covers
output by peasant farms.

(b) Ratio of input VMP to input price.

(c) Rostov, Ivanovo, and Nizhnii Novgorod.

(d) Estimated with material input costs split into two complementary
components: the cost of purchased inputs and the cost of all other
material inputs (inputs from own production and input received in
kind).

Table 2: Allocative efficiency calculations for fertiliser

Researchers       Output (a)         Regions

Uzun              All agriculture    All Russia
Liefert et al.    Grain              All Russia
Lerman (c)        Crops              Three survey regions
Lerman (d)        Crops              Three survey regions
Grazhdaninova     Grain              Three survey regions
Grazhdaninova     Grain              Rostov
Epshtein          All agriculture    Leningrad

Researchers       Time period    AE ratio (b)

Uzun                 2001            6.30
Liefert et al.       2000            2.16
Lerman (c)           2001            1.47
Lerman (d)           2001            7.47
Grazhdaninova        2001            0.95
Grazhdaninova        2001            0.70
Epshtein             2001            2.87

(a) Produced by corporate farms.

(b) Ratio of input VMP to input price.

(c) Fertiliser measured in rubles.

(d) Fertiliser measured in tons.

(2) The AE calculations have been carried out by six groups of researchers working independently. The results of Uzun, Epshtein, and Sazonov are unpublished and can be obtained from the author on request (ill Russian). The results of Grazhdaninova and Lerman are reported in a separate article in this issue (Grazhdaninova and Lerman, 2005). The results of Liefert's group have been published in Liefert et al. (2003).

(3) The Tambov peasant farms in Sazonov's analysis specialise in Crops and have no livestock. Sazonov's results on the allocative efficiency of crop production therefore represent also total farm efficiency for this sample.

(4) Price differences may also explain a notable discrepancy in Table 2, where Grazhdaninova's AE ratio for the three survey oblasts points to (slight) overuse of fertiliser, contrary to the other results. All other researchers using the national average price of less than 1,500 rubles/ton obtained AE ratios greater than 1, pointing to underuse of fertiliser. Grazhdaninova calculated the AE ratio using the survey mean price of 2,600 rubles/ton (adjusted for nutrient content), which was very close to the world trade price. This gave an AE ratio of 0.95 for fertiliser use in grain production in the three survey oblasts, which is very close to the point of allocative efficiency. The low AE ratio for Rostov Oblast in Table 2 probably points to actual fertiliser overuse in this agriculturally productive region.

REFERENCES

Customs. 2001: Tamozhennaia statistika vneshnei torgovli Rossiiskoi Federatsii 2000. State Customs Committee: Moscow.

Goskomstat. 2002: Tseny v Rossii 2002. Goskomstat: Moscow.

Grazhdaninova, M and Lerman, Z. 2005: Allocative and technical efficiency of corporate farms in Russia. Comparative Economic Studies 47(1): 200-213.

Liefert, W, Gardner, B and Serova, Eu. 2003: Allocative efficiency in Russian agriculture: the case of fertilizer and grain. American Journal of Agricultural Economics 85(5): 1228-1233.

Mundlak, Y. 1961: Empirical production functions free of management bias. Journal of Farm Economics 43(1): 44-56.

Sazonov, S and Sazonova, D. 2005: Development of peasant farms in Central Russia. Comparative Economic Studies 47(1): 101-114.

Serova, E and Shick, O. 2005: Markets for purchased farm inputs in Russia. Comparative Economic Studies 47(1): 154-166.

Subbotin, A. 2005: Determinants of access to credit for corporate farms in Russia. Comparative Economic Studies 47(1): 181-187.

USDA. 2004: Online database at www.fas.usda.gov/psd/.

Uzun, V. 2005: Large and small business in Russian agriculture: adaptation to market. Comparative Economic Studies 47(1): 85-100.

Yastrebova, O. 2005: Nonpayments, bankruptcy and government support in Russian agriculture. Comparative Economic Studies 47(1): 167-180.

WILLIAM M. LIEFERT, The author thanks Zvi Lerman for helpful comments, and bears responsibility for any remaining deficiencies. The views expressed are the author's alone and do not in a any way represent official USDA views or policies.

Economic Research Service, US Department of Agriculture, Washington, DC, USA. E-mail: wliefert@ers.usda.gov