Decline of corporate enterprises in transitional agriculture: evidence from Lithuania.

Bezemer, Dirk J. ; Stanikunas, Donatas ; Zemeckis, Romualdas 等

JEL Classifications: P46, QI2, R20, 012, 018

INTRODUCTION

An important aspect of the economic transition in the formerly socialist economies is the change in the structure of enterprises within agricultural sectors. Following the agricultural reforms, such as privatisation of farm land and assets, in the early 1990s, traditional socialist farm structures (collective and state farms) have generally been transformed into corporate farms: capitalist enterprises with the legal labels of partnerships, joint-stock companies or limited liability companies. Simultaneously, in all of the transition economies millions of private individual farms--family farms--have emerged, often operating on a very small scale and satisfying much of rural household food consumption (Swinnen and Macours, 2000; Lerman, 2001). Corporate farms, in contrast, have declined in number and, it appears, also in economic viability. Evidence from various countries suggests that corporate farms are suffering from low profitability, high debts, and high liquidation or bankruptcy rates, even in those countries where policies towards the different farm structures are officially non-discriminatory (EBRD, 2003; Lerman et al., 2002). In this paper, we will examine the possible reasons for these problems besetting corporate farming in many transition economies, and explore their relevance for corporate farms in Lithuania.

This question is an interesting one for several reasons. Theoretically, the market reforms in agriculture were driven by a distinct view on the inadequacy of corporate enterprise structures in agriculture within a market economy, and the desirability of replacing them with family farms, which were deemed to be inherently more efficient (Schmitt, 1993). To enable this transformation to occur was a major aim of the land privatisation programmes in the early 1990s. The common perception of socialist-style farm structures in the early transformation years was that 'the evident weakness of this organizational form provides the argument for full scale privatisation' (IMF et al., 1991, pp. 157-158) and that 'privatisation in ... agriculture mainly concerns the breaking up of large units ...' (World Bank, 1995, p. 2).

It would now, a decade and a half later, be important to examine how the performance of corporate farms, as they presently operate in the new market economies, is connected to such innate inadequacies. Other studies (eg Mathijs and Vranken, 2001; Gorton et al., 2003) have found large variability within the group of corporate farms, indicating that factors other than goverance structures play a role. Consequently, corporate farms are not doomed--though they may be handicapped--by their governance structures, and it would then be possible to explore policies to enhance their performance. In this study, we explore and analyse how various factors both within corporate farms and in the farm environment affect their performance. (1)

From a policy point of view, the fate of the corporate farming sector is important for agriculture as a whole. Because alternative farm structures (family farms) are typically small and often market only part of their production, commercial agriculture is still to a large extent dominated by corporate farms. Corporate farms also work considerable shares of agricultural land in many transition economies. Improving the performance of these enterprises would be relevant to the contribution of the agricultural sector to the national economy (eg in terms of incomes, employment, and export potential).

In this paper several hypotheses on the factors affecting corporate farm profitability in a transition economy are developed. These will be examined based on both secondary information and survey data from Lithuania collected by the authors. The data provide an opportunity both for an explorative investigation of the structure, environment and performance of corporate farms in Lithuania and for an econometric analysis of the validity of the hypotheses. The paper is concluded with an overview and discussion of the findings.

HYPOTHESES

In the wider literature on the agricultural sector in transition economies, several reasons have been advanced to account for what has amounted to a crisis in agriculture in general, and in corporate agricultural enterprises in particular (Hobbs et al., 1997; Sarris et al., 1999; Bezemer, 2002b). Since ultimately such factors all translate into low farm profitability, they can be conveniently outlined with reference to a farm profit function of the form

[PI] = R - C + B

where [PI] is the farm profit, R is the revenues from agricultural production, C is the total costs connected to agricultural production, and B is any other income, not directly connected to agricultural production (such as net credit, subsidies, or income from non-agricultural activities). This can be specified as

R - pQ

O = Q(K, L) C = Lw + [Kp.sub.k]

where p is the output price, Q is the agricultural output level, K is the level of capital inputs--whether fixed (buildings, machinery) or variable (fertiliser, chemicals)--L is the amount of labour utilised in agricultural production, w is the agricultural wage, and [P.sub.K] is the price of capital. In the literature on (corporate) enterprises in transitional agriculture, four factors have been suggested which may depress (corporate) farm profit and endanger farm viability. These may be presented as four hypotheses on the shape of the farm profit function.

Hypothesis 1:

[partial derivative] ([PI]) / [partial derivative](Q) < 0

A well-known argument against the corporate governance structure in agricultural production (as distinct from a family farm structure) is that it leads to overly large farms with flawed incentive structures (Deininger, 1995). In most of the socialist economies, farms had a number of workers far exceeding the size of farms on which generalisations about scale and scope economies are based (Pryor, 1992). Corporate farms, as successor organisations to these structures, are still large, recent restructuring notwithstanding. One reason is that corporate farms need to be above a minimum scale in order to justify the overhead costs of management (which are absent in family farming). However, technical economies of scale and scope can be argued to be soon outweighed by diseconomies due to the costs of monitoring employees in team production (Alchian and Demsetz, 1972}. These costs arise because of the information asymmetry problem between managers and workers with regard to the state of the land, crop, and animals, and the appropriate amount of labour input.

Yet another reason may be that such farms, in the transition setting, are often open to state interference. They may be subject to direct state ownership (sometimes via share ownership by state voucher funds) or through dependence on state resources such as subsidies and credit. They may also depend on state services in the areas of information and expertise. National and local governments can then use these resources and services to allow continued operation of large farms, which provide rural employment. Thus, farm restructuring decisions (slimming) may become politicised, leading to overly large farm sizes compared to the profit-maximising optimum.

Hypothesis 2:

[partial derivative] ([PI]) / [partial derivative](K) > 0

A second reason for reduced farm profitability may be under-investment because of capital rationing. Capital rationing may occur because of malfunctioning credit markets.

In the centrally planned economies, relative large investments were made into the agricultural capital stock and plausibly over-investments occurred both in fixed capital--the level of mechanisation was high in relation to the costs of labour--and in variable capital, for example in the amount of fertiliser applied (Pryor, 1992). Presently, such directly subsidised investments are small or absent. Investments are now predicated on access to outside finance, of which bank credit is the most important option. However, agricultural credit markets are characterised by highly specific risk sources, enterprise capital structures, collateral options, intertemporal borrowing and repayment schemes, and other idiosyncratic factors (Barry et al., 1995). They require targeted institutional support to develop and function. Several recent studies on credit markets for agriculture in the transition economies suggest that these are often failing, for a variety of reasons (Pederson and Khitarishvili, 1997; Shrieder and Heidhues, 1998; Davis and Gaburici, 1999; Koford and Tschoegl, 1999; Swinnen and Gow, 1999; OECD, 2001; Bezemer, 2002a, 2003a, b).

In such situations and in the presence of state interest in the continued operation of farm companies providing rural employment (as explained above), the result may be 'soft budgets', where credit is used as subsidies. Given the aim of preserving employment, such funds will be used to prevent the weaker farms from collapsing, rather than flowing to its most profitable use, which would typically be in the more profitable farms. The result is that 'the growing share of support [by credit support funds] is to farms whose performance remains the worst among the various ... forms of farming' (Csaki et al., 1999: 36). This quote is actually from a World Bank study on agricultural credit in the Czech Republic. Indeed, a typical flaw in agricultural credit markets in transition economies is that credit goes to the weaker farms in terms of profitability, as opposed to the situation in well-functioning credit markets, where loans are allocated to their most profitable use (Pederson and Khitarishvili, 1997; Rother, 1999).

This mechanism need not be typical for credit only. Any supply of inputs, services or benefits more generally which is controlled (or susceptible to manipulation) by the state will, in this account, have a tendency to be allocated more towards those farms with weaker financial performances. This may include overt subsidies, specialist advice, and tax advantages (see Bezemer (2002a, 2003a, b) for detailed studies on such allocation biases). This tendency is connected to hypothesis (1): using credit and other inputs or services as subsidies is one way of keeping farms large and their employment potential high. Importantly, it implies that credit and other resources are directed away from those farms that should receive it in a well-functioning credit market; thus, such interference causes these farms to be credit rationed.

Alternatively, without state interference, failing credit markets may take the form of under-supply rather than misdirected supply. Too few or too small banks typically service agriculture. Those that do operate often perceive the sector as risky due to fluctuating yields and revenues, as insecure since land is no adequate collateral, and as unfamiliar and requiring costly, specialist knowledge. In order to hedge against these risks banks will tend to set relatively high interest rates and demand large collateral to compensate for low loan volume and high risks. In Lithuania, for instance, agricultural interest rates at the moment of writing are typically 2-3% above those for other enterprises, and loans extended often have ceilings of two-thirds of the value of collateral (LIAE, 2003). Such interest rates are typically above the return rates of most agricultural projects, and such conditions are difficult and costly to meet. With debt finance unattractive, farm management will prefer to meet investment needs from own farm resources, in line with pecking-order theory (Myers, 1984). Thus, it is only the weaker farms without own resources but with the need to meet at least operational investment needs that borrow, this time not because of rationing but because of a high price for credit. As with politically motivated loan allocation biases, this mechanism also has been observed in the transition context (Bezemer, 2003a, b). And in both cases, lending is concentrated among the weaker farms, resulting in the negative relation between capital investments and farm profit stated in the hypothesis.

We note that these two mechanisms leading to failing credit markets may be mutually enforcing. With a high probability of bailout because of soft budgets, management of weakly performing farm companies will have fewer reservations about borrowing against unsustainably high interest rates.

Hypothesis 3: The level of p is depressed relative to the level of (w + [p.sub.K]).

A third problem for farm companies may be low output price relative to prices of inputs--especially if prices are taken in the broader sense of including transaction costs as well as off-farm agricultural output price. The main reasons for depressed output prices are international competition and domestic market power. An increase in competition was inevitably implied in market reforms. In the early transition years this led, in most transition countries, to the operation of price scissors, that is an increase of input prices relative to output prices (OECD, 1997).

The main problem presently may well be domestic market power. Restructuring has been widespread in agriculture, with many small-scale family farms and many farm companies of smaller size than previously. But restructuring has typically been more limited in the food industry, where one or a few buyers still dominate regional, and sometimes national markets for a particular agricultural product (Gow and Swinnen, 1998; Gaisford et al., 2001).

Farm operators in transition economies are then facing powerful buyers in downstream markets. Their market power is typically based on high concentration and oligopsony power, where many farms deliver to a single processor or trader (Hobbs et al., 1997). It also derives from product perishability (which deprives sellers of the option to defer delivery while negotiating for better transaction conditions) and from the absence or underdevelopment of futures markets for products that are not perishable. Market power for the buyer may additionally be based on specific investments made by farmers, which are particularly high in, for instance, the dairy sector. Such investments are specific to the product, and, given high buyer concentration, bind the producer to one buyer or a few buyers. Because of asset specificity, farm operators cannot easily change product. Because of buyer concentration, they cannot easily change buyers.

This puts the processing industry in a strong bargaining position vis-a-vis farmers. They may exploit this position by enforcing transaction conditions more favourable to them. This may be lower price per product unit or delayed payment; it may also include introducing new quality requirements or passing packaging and transport costs on to producers. All these effectively depress output prices and farm profit.

Hypothesis 4:

[partial derivative]([PI] / [partial derivative](B) > 0

A final suggestion for explaining low farm profitability is that farms are overly dependent on income from agricultural activities only. As the transitional economies move towards the market system and develop, the share of agriculture in the economy--in terms of employment, GDP share and consumption share--can be expected to fall. This is a long-term trend in the developed capitalist economies, in line with Engel's Law (eg Laitner, 2000). In order to survive, farms would need to re-orientate towards non-agricultural activities and income. This argument has been made in the context of the crisis in agricultural profitability in the Western economies, where nonagricultural income has risen greatly over the past 40 years, and is now typically a considerable share of total farm income, for example between a third and half on average in the European Union countries (Eurostat, 1999). Research shows that such pluri-activity is now an important determinant of farm viability (Shucksmith et al., 1989; Evans and Ilbery, 1993; McNally, 2001). The suggestion is that this may be occurring very slowly in the transition economies, and in particular within farm companies (Greif, 1997; Swain, 1999). Operators of family farms in these countries often derive a considerable share of their income from non-agricultural sources such as wages and non-farm enterprise income - about 40% in Lithuania, according to recent research by Bezemer et al. (2003). This effectively cross-subsidises agricultural production. By contrast, farm companies in transition countries typically rely for the most part on agricultural activities for income generation (with the exception of Slovakia). This may be another cause for their vulnerability.

In sum, the main factors possibly relevant to depressed farm profits in Lithuania range from features of farm organisation to features of their economic environment. Some are typical for the transition economies or for corporate farms; some are generally relevant to all farms in transition economies or for the sector agriculture as a whole. We note that these are complementary rather than rival hypotheses. Their relevance to the Lithuanian case will be explored and analysed in sections Farm Structures, Performance, and Economic Environment and Analysis.

BACKGROUND AND DATA (2)

Agriculture in Lithuania

As in all transition countries, agriculture in Lithuania has been greatly affected by the market reforms. Since this was a heavily collectivised and subsidised sector in the centrally planned economy, deep reforms in ownership patterns, production structures, and price relations have occurred. Combined with the opening of Lithuania to foreign markets and competition, the changes in agriculture have been dramatic.

Starting from an untenable position of subsidisation in 1991, prices have moved against agriculture during transition. Input prices have generally increased faster than output prices in real terms, resulting in the 'price scissors' that were a common phenomenon in agriculture in the early part of the 1990s. In Lithuania, however, price scissors continued to operate in 1995 and 2001: input prices rose 19% and 13% more than output prices in 1995 and 2001. In the years in between the opposite was true: input and output price increases were equal in 1996, and output prices rose faster than input prices by 7, 1%, and 29% in the years 1997-2000. In sum, the terms of trade of agriculture have been unfavourable to Lithuanian agriculture during most of the transition years; and volatility in the terms of trade has implied a source of risk to farm incomes.

In response, output decreased dramatically: the share of agriculture in GDP decreased from 27% to 9% in 1989-1996, even though these were years of overall economic contraction in Lithuania. Due to economic hardship, employment in agriculture increased in these years, from 18% to 25 % of total employment in 1995, falling back to an official 16% in 2001.

Land reform and farm structures

Simultaneously, the number of people involved in small-scale agricultural production has increased greatly in Lithuania. During 1991-2000, 688,000 claims for land restitutions were filed, and 478,000 restitution decisions were taken. In 2000, there were 523,000 owners of agricultural land, which is 18% of Lithuania's 2.9 million adult population. Although the area of land privately owned has increased dramatically (it went up from 800,000 to 2 million hectares in 1995-2001), land reform in Lithuania is still incomplete. In 2001, only 53% of agricultural land was privately owned. The official collection of applications for land restitution ended in 2001, but the process of land restitution is still ongoing.

Family farms are small: 12.6 hectares on average, with 84% smaller than 10 hectares. The number of landowners involved in family farming was 132,000 in 2001, with 67,000 of them officially registered as such. (3) Thus, households working family farms and registered household plots (400,000 in 2001) between them totalled 532,000. With an average household size of 2.7 persons, this implies that 1.4 million, or over a third, of Lithuania's 3.7 million population directly depend, to some extent, on food production for their livelihoods--not counting those households indirectly dependent through gifts, exchange, and barter, and those dependent on income from the corporate farming sector. These figures suggest that Lithuania has become a more agrarian country during its transition from socialism to capitalism. Yet officially, the share of the working population employed in agriculture was only 16% in 2001, down from 25% in 1995. This, incidentally, is still very high compared to less than 5% in the European Union, which Lithuania joined in 2004.

Other farm structures include partnerships and remaining state farms. Both are corporate farm structures in that they have a separation between management and ownership, and between blue-collar and white-collar labour. Farm companies have greatly decreased in number during the second half of the transition. According to LIAE (2003), between 1995 and 2002 the number of farm partnerships decreased from 2,457 to 923. In the year 2000, there were 1,138 farm companies (923 partnerships and 215 other structures). With an average size of 486 hectares, they worked 553 thousand hectares. Their number appears to be rapidly declining. According to data from the Lithuanian Land Cadastre, in January 2001 there were 697 agricultural companies, working a total land area of 128 thousand hectares; on 1 January 2002 there were 537 agricultural companies working 80 thousand hectares. The best expert estimate at the moment of writing (early 2003) is that there are 485 functioning farm companies, of which about half are profitable and have a long-term future (LIAE, 2003). (4) This is an extreme example of the generally dire situation of farm companies in Central and Eastern Europe, with the exception of Hungary (Gorton et al., 2003).

With on average 400 companies going bankrupt annually during the transition years, the main cause of their decline was lack of economic viability. One reason may be that the reform policies were flawed, in various ways. In some cases land rented by corporate farms and owned by private individuals could only be restituted after the farm went bankrupt. This would give these individuals, owners of the farm's productive resources, an incentive to bring about such failure. Also, shares in corporate farms are tradable only within the population of shareowners, which depressed their price and decreased farm equity capital and the value of its debt collateral. Furthermore reserve capital, which can be defined with great flexibility, especially in the transition context, was exempted from incorporation in share values. Apart from further decreasing share prices, this also gave the management an incentive to increase reserves in order to keep it within the farm. This may not have been the most efficient way of using its capital.

Another comparative disadvantage for corporate farms is taxation. Members of the household working a family farm are not officially hired labour, and hence pay no income tax. Also tax evasion is easier on family farms; often they are not registered, and transparency of accounting is limited since the family either consumes or barters relatively large output volumes.

For these and other reasons, income per hectare in farm partnerships was between 16% and 53% lower on average in 1996-1998 than it was in family farms, depending on production conditions. Still, family farms are an important component of the agricultural sector. They produced 20% of official gross agricultural output in 2000, with concentrations in sugar beet and meat (both 40% of the respective totals). Farm companies typically rent their land from individuals or from the state. The Constitution did not allow them to buy land until early 2003; land could be owned only by natural persons. Companies occupy a third of Lithuania's agricultural land, and are mostly located in its most fertile parts, in the centre of the country.

Domestic agricultural policies

Lithuania operates policies designed to support agriculture under the umbrella of the Special Rural Support Programme (SRSP). In 2001, this programme allocated financial support to farms for the purposes of covering debts for near-bankrupt companies, paid part of the interest security on loans to farms (through a separate Rural Credit Guarantee Fund (RCGF)), subsidised fuel, and funded training and consultancy services. The SRSP financed programmes related to preparation for EU membership. They extend direct payments and purchases cereals at intervention prices, as in the EU Common Agricultural Policy.

The RCGF was established in 1998. Together with the Lithuanian Agency for Regulation of the Agricultural and Food Product Market (LARAFP), it supports both farms and, since 2000, enterprises in agribusiness through subsidised loans and provision of security. The RCGF and LARAFP commit to repay up to 70% and 100% of loans to farms and agribusiness enterprises, respectively, with the lender assuming the risk for the remainder. All lenders operating in Lithuania are privately owned, commercial banks, since the state sold its 76% stake in Agricultural Bank, Lithuania's third-largest bank in asset terms, to NordLB of Germany.

In the 1998-2000 years, the RCGF issued 583 guarantees to farmers and other enterprises, which amounted to 38 million litai (11 million euro). (5) The majority of farmers, 70%, used the guaranteed credits for investments in fixed capital, mostly tractors (70% of loans), and dairy equipment (5%). In the same period, one-third of the credit (13 million litai, or 3.8 million euro) was repaid. During its existence the LARAFP partially guaranteed loans worth 577 million litai (167.2 million euro), extended through banks to food processing businesses. Most (75%) was used for grain purchases; the rest for trade in other foodstuffs. The budgets of both the RCGF and the LARAFP are required to be approved by the state. The state is obliged to meet the liabilities of both institutions towards banks, which amounted to 15% of their budgets in 2000.

DATA

The main empirical basis of this study is a survey research among 53 farm companies in Lithuania (or 11% of the total number), implemented during the summer of 2002. The survey was based on a questionnaire developed in recognition of the main issues relevant to corporate farming in transition economies, as outlined above, and in consultation with staff of the Lithuanian Institute for Agricultural Economics (LIAE). Data were collected through face-to-face interviews by LIAE staff in different parts of Lithuania. In view of the small survey sample, the survey frame was targeted and designed to reflect regional differences rather than aimed to be representative on the regional or national level. It included both farms in areas defined in 2002 by the Lithuanian Ministry of Agriculture (following EU definitions) as being 'favourable to agricultural production' and regions less favourable to it, each comprising half the sample. This reflects conditions such as soil quality, sunshine, precipitation, relief, and physical infrastructure. A descriptive overview of survey findings is provided in the next section.

FARM STRUCTURES, PERFORMANCE, AND ECONOMIC ENVIRONMENT

Labour, land, livestock

A typical aspect of farm companies, as distinct from family farms, is the presence of a management team separately from the blue-collar labour force. The average management team in this sample consisted of four persons, three men and one woman, with an average age of 47 years. There will typically be a director, a general administrator and human resources person, a farm accountant, and a technical, veterinary or crop specialist. The typical level of education is agricultural college or university, while management have on average 14 years of managerial experience.

Table 1 provides further details on farm production structures in the sample and compares it to official statistics. With a land area of, on average, 834 hectares and a labour force of 47 workers, farm companies in the sample are large by most standards, and also larger than the Lithuanian nationwide average in terms of land and labour. They are more crop-oriented and have considerably less livestock than the national average. We note that this difference is due both to the small and non-representative sample, and to the different times of observation within a period of rapid structural change. As we have seen, even different official sources (the agricultural institute LIAE versus the general statistical bureau LSB) for the same year show considerable divergence on key farm structure statistics.

The table also shows that the means hide large heterogeneity and variations in production structures. Several products are produced by only some farms in the sample. Within the group of farms producing a product, standard deviations of herd and area sizes vary considerably.

Agricultural and non-agricultural activities

Farm companies in the sample were about equally divided over specialised and mixed farm types. A total of 22 farms in the sample derived over 75% of their total revenues from crop production; seven farm derived over three-quarters of income from livestock production. The remaining 34 farms had significant revenues (over 25% of the total) from both crops and livestock. While there is thus, in many farms, diversification within agricultural production, farm companies in the sample are relatively specialised in a broader sense. Presented with a number of non-agricultural activities--among them retail and wholesale trade, food processing, construction, manufacturing, crafts production, tourism, and recreational services--only six respondents reported involvement in such activities. These were in food processing and retail trade (a farm shop), both of which are closely connected to agricultural production itself.

Managers were also asked to report the share of income from agricultural production proper and of various other activities, many of which are more closely intertwined with agricultural production. They reported this for 1990, 1995 and at the moment of surveying in 2002. Table 2 presents the results.

The figures show that income from additional activities outside agricultural sales includes transport as well as machinery repair, sale, and rent (agricultural contracting). This accounts for 14% of income and occupies around 10% of the labour force on average. The figures also suggest that the relative importance of such activities has slightly increased during transition, but we note that given the imprecision in reporting and the small sample size this is not a statistically significant development. When asked about their expectations with respect to the future development of revenues from nonagricultural activities, managers mostly indicated they expected no changes, while a few foresaw moderate increases in revenues of up to 10%.

In sum, non-agricultural activities in Lithuanian farm companies are intertwined with agricultural production, and are marginal features of the farm structures in terms of income and labour force. No significant developments in their importance have occurred during transition or are expected in the foreseeable future.

Profit, investment, and finance

Many farm companies in the sample were in a difficult financial situation, with 20 of them reporting losses over the year 2001, 10 of which were in excess of 100,000 litai (29,000 euro). Half of the farms (26 cases) were just coping, making either a small profit or a small loss (of up to 10,000 litai, or 2,900 euro annually); seven farms made profits in excess of a 100,000 litai The average farm reportedly made a small loss in 2001. Almost all (45 out of 51 responding) managers reported that farm profitability had 'much worsened' or 'worsened' over the preceding 5 years (ie since 1996). Three respondents reported it has remained similar and eight that it had 'improved' or 'improved much'.

Most (45) farm management had invested in agricultural capital goods in the 1998-2001 years. Many had also invested in land through renting (24 cases), and some additionally in non-agricultural capital goods (10 cases). Respondents also indicated their three most important sources of finance. The most frequent was own farm resources, mentioned by 23 respondents as their primary source of finance; in addition, three mentioned subsidies as the most important or as the secondary source of finance. Agricultural capital goods were also most often (45 cases) financed by own farm resources; in nine cases this was a secondary source, and bank credit or subsidies the primary source. A frequent second source was bank credit (21 cases). Seven farms used subsidies, always as a secondary source, to finance agricultural capital goods. Non-agricultural capital goods were in all cases financed primarily by own resources. In one case bank credit, in another supplier credit were additionally utilised.

Market environment

Firms in the transitional economies typically operate in challenging market environments, and enterprises in the agricultural sector are no exception. Respondents were asked to rank a number of potential problems relating to market development, the quality of state assistance, and corruption and crime on a scale from 1 to 5, where higher scores indicate a larger problem in transacting. They assigned scores separately to transaction problems in agricultural and non-agricultural activities (where relevant). Nearly all managers (51) responded to this question. The two most important problems were late payment by the output buyer and disputes with the buyer over the price and quality of produce. Both scored an average 3.8 on the 5-point scale. Difficulty in finding a buyer or supplier scored slightly lower (3.4). These problems are linked to the large dependence on particularly output buyers: in this sample over 80 % of agricultural sales, on average, go to the processing industry, which often is a regional monopsony. The remaining sales are about equally divided by sales to retail shops and directly to customers.

A number of other issues were deemed to be of medium importance (score 2.3-2.5). These included changes in packaging and transport requirements by output buyers, time spent negotiating with input suppliers or output buyers, lack of wholesale markets in the proximity, bureaucracy, and theft. Of small importance (scores 1.4-1.8) were bribes paid to officials, fees for private protection of output and buildings, and a lack of information on government regulations.

Interestingly, these problems were reported in about the same ranked order of importance, but with considerably lower scores for the case of nonagricultural output (reported by 33 managers). Top scores were 'difficulties to find a buyer or supplier' and 'late payment by the output buyer'. Both scored 2.6 on average; all other issues scored less than 2.0. Indeed, the market concentration on the output side of non-agricultural activities as reported by the respondents was considerably lower than in the case of agricultural products. Over 90% of sales was directly to consumers, the rest equally divided between wholesale and retail traders.

An extension of this exploration of the economic environment is to examine supporting public and private institutions. Access to, and effective use of, such institutions is typically one of the key bottlenecks in enterprise development in the transitional economies. Table 3 presents qualitative, binary measures for both aspects, both with respect to specialist agricultural services and general business support.

Services such as banking, accounting and insurance services, as well as veterinary help, all of which are essential to enterprise operations, are accessed and used by virtually all farms in the sample. Many also have access to an agricultural marketing cooperative or other producer association; but few actually use them. This suggests that, while such associations could play a role in diminishing transaction problems by providing countervailing power, there are factors that hinder their effectiveness in coordinating farmer transactions. Information and advice, either from a state agency or from private consultants, is likewise available to many but used by relatively few. Services supportive of rural labour markets are available to all and used by a considerable number of farm managers.

A final aspect of interaction with the market environment is that Lithuanian farm companies were found to be linked, in various ways, to their local, rural economies. Respondents were asked about the geographical characteristics of their employment, input purchases, and output sales. They reported the share of these costs and revenues that came from, or went to, enterprises or households located within their region and outside the cities. On average, a fifth of outlays on variable inputs (fodder, fuel, pesticides, veterinary services) was spent with such local, rural enterprises. Of all wages, 70% were paid to employees from local areas. Just over half {56%) of all land rent went to private landowners in the local economy. Outlays on non-land real estate (construction, repairs, and maintenance of farm buildings), on machinery and on transport spent in the local area accounted for 37%, 23 %, and 9%, respectively, of the totals. While the extent of employment and production linkages thus varies over outlay category, the contribution of farm enterprises to local employment and turnover of local upstream industry are considerable. The local links at the output side of the enterprise were generally of smaller importance. Farm managers reported that of all revenues from crop and livestock products, only 11% and 6% were received from buyers in the local, rural economy.

ANALYSIS

The above exploration suggests some inferences on the relevance of the discussion and hypotheses outlined in the section Hypotheses. Corporate farms in the sample are indeed in many cases suffering from low profitability and losses. Their managers report various factors also suggested by the literature as relevant, notably transaction problems and costs. Diversification seems indeed to be near-absent. We will now examine the relation of such features with farm profitability.

Testing the hypotheses

As outlined in the section Hypotheses, the hypotheses suggest that the following factors may depress farm profitability: an overly large labour force; an overly large farm size in terms of land and livestock; insufficient income from non-agricultural activities; constrained access to credit; and problems with downstream transactions. These factors are captured by the following variables based on the sample data. Between brackets the sign of their hypothesised effect on farm profitability is given.

Hypothesis 1

Corporate farms are above profit-optimal farm sizes in terms of land, labour, and livestock.

LAND (-), land areas cultivated by the farms in hectares, CATTLE and PIGHERD (-), the size of the milk and beef cattle herd and the number of pigs. These measures reflect the most important type of fixed assets (land accounted for 46% of all fixed asset value in farm companies in 2000), and the two most common forms of livestock in the sample. We also included two labour force measures: LABOUR (-), the number of full-time equivalents of the total labour force, and MANAGEMENT TEAM (-), the size of the management team. Because of governance problems, corporate farms are expected to be oversized particularly in terms of management.

Other capital categories such as machinery and buildings were not included because of valuation problems. Size is here measured by input levels, in line with the hypothesis, rather than by output level, as is usual in farm statistics where standard gross margins measures are employed. Measuring farm size by the sum of standard gross margin per product, aggregated over output levels, is also inappropriate here because of its high correlation with profitability.

Hypothesis 2:

Corporate farms face incentives to perform weakly in order to access credit and assistance though soft-budget mechanisms.

CREDIT (-), a binary variable indicating if the farm has taken up credit as a primary source of finance in the 1998-2001 period; and EXTENS (-) a binary variable indicating if the farm has used state-provided extension services were included. Also CONSULT, a binary variable indicating use of business advice, was added. Such consultancy is mostly offered by private firms, in contrast to extension services. Including CONSULT thus provides an opportunity to explore the significance of the profit depressing effect of state involvement in information services.

Hypothesis 3

Corporate farm profitability is undermined through exploitation by downstream processing industry.

As a measure for dependence on product buyers, a Herfindahl index was constructed for the shares of sales through different market channels (consumers, retail, wholesale, and processor), named CHANNEL DEPENDENCE. The Herfindahl index equals the sum of squared shares sold through each channel, in per cent of total sales. With all sales going through one channel only, the index is equal to one. Its value is smaller as the number of sale channels is larger and as shares are more equal in magnitude.

In the sample, concentration of sales in one channel is virtually equivalent to concentration of sales to processor and wholesale traders, not to retailers or consumers. (6) In these channels, in turn, it is rare for a farm to sell produce to more than one buyer; typically, many farms deliver to one processor or wholesale trader. Large dependence on one channel in this sample therefore practically equals large dependence on one buyer. Channel dependence, which is directly observed, is therefore a good proxy for buyer dependence, which we did not directly observe.

Farms selling a significant proportion of their output to small-scale buyers such as retail businesses or directly to customers are likely to incur high transaction costs and thereby depress their profits - even while such diversification away from processors and traders plausibly improves their bargaining position. To control for this, a dummy variable DIRECT MARKETING was included, which takes the value 1 if more than 25% of farm sales are to retail businesses or directly to customers. (7)

Hypothesis 4

Corporate farms are insufficiently diversified.

DIVERS (+) was included, which is the share of total farm revenues, in per cent, that is derived from non-agricultural activities.

To control for natural and human capital, we included FAVOUR, a binary variable indicating if the farm is located in an agriculturally favourable area--as explained in the section Background and Data 3--as well as EXPERIENCE and AGE, the number of years of management experience and the ages of the two most senior people in the management team, respectively. (8)

Regression analysis: specification and findings

These variables were regressed on a measure for farm profitability. Because of sensitivity and response time concerns, respondents were asked to report the level of profit or loss of their farm in the previous (2001-2002) financial year in twelve ranges, rather than in point estimates. The resulting variable PROFIT takes values 1-12, with values 1 up to and including 6 indicating farm profitability in 2001-2002, and values 7-12 indicating the farm was loss making. (9) Because profit levels are ordered, an ordered probit estimation is appropriate. (10) In Table 4 estimation results are presented.

Estimation was based on 46 of the 53 farms in the sample, mainly because the data for CHANNEL DEPENDENCE were not reported by all respondents. We note that the standard errors of some coefficient estimates are quite substantial. Indicators of model fit are satisfactory in comparison to similar studies (11) and estimates for 12 of the 14 independent variables introduced above are statistically significant at P-levels of less than 10%. Still, the model obviously captures only part of variations in profitability. Reasons for this include the small sample size and the complexity of the relation under examination. Also, no information on social capital, plausibly an important determinant of profitability, was collected in the survey.

Turning to the results, we note that FAVOUR is positive, and is the most statistically significant variable in the model. Location and natural conditions are clearly, and unsurprisingly, major determinants of farm performance. Managerial expertise as measured by the age and experience of the two most senior members of the management team are also related to profit level. (12) Interestingly, it is older rather than younger management who tend to be more successful, even when controlling for the experience effect. The other findings suggest the following inferences on the four hypotheses developed in the section Hypotheses.

First, there is no evidence that farm size, as measured by land, labour, and livestock, would negatively affect profitability (Hypothesis 1). Farms with more pork production do significantly better in terms of profit. In contrast, the size of the farm management team is related to lower profit levels. This suggests that governance problems inherent in corporate farming may play a role in the profits crisis, but not that corporate farms themselves are oversized. This is so even though variations in the farm size variables in the sample are considerable and farms in the sample are on average larger than the Lithuanian nationwide average. Although the sample is too small, and the method inappropriate, to conclusively measure the existence and nature of any scale effects, it is safe to conclude that these findings at least do not support the idea that corporate farms are overly large and therefore loss-making.

Second, there is evidence that take-up of credit and use of extension services is negatively related to farm profit (Hypothesis 2). Both of these inputs are allocated by state-controlled (if not always formally publicly owned) institutions, which would be in line with the hypothesis that soft budgets in these allocation systems exist, causing the resources (credit and information, in this case) to be used as subsidies, which are allocated to the weaker farms in terms of profitability. An interesting finding in this respect is on the use of business consultancy, which is a service very similar in nature to that of extension, but typically offered by private enterprises rather than state-controlled bank and extension services. Use of business advice is positively related to profitability. This may be interpreted to underline the profit depressing effect of state interference in the allocation of (otherwise very similar) extension services.

Third, CHANNEL DEPENDENCE has a significant and negative association with farm profitability. This provides support for the hypothesis that corporate farm profitability is undermined by over-reliance on one sales channel, and thereby typically one buyer (Hypothesis 3). The qualification is that the coefficient of DIRECT MARKETING is also significantly negative. While some diversification in sales channels may improve profit through increased bargaining power, there is a trade-off with the rising transaction costs of shifting away from sales to processors and traders toward direct marketing.

Fourth, the estimation results also show that a larger share of nonagricultural income is significantly related to higher profit levels (Hypothesis 4), even though there is only a small number of positive observations on DIVERS.

SUMMARY, DISCUSSION AND CONCLUSIONS

This study aimed to provide insight into the causes of the ongoing farm restructuring process in transition economies. The typical trend in this area has been one of a declining number of large, corporate farm companies, and a sharp increase in the number of small-scale family farms and household plot producers. To explore these developments, the study utilises data from Lithuania, where both trends have been particularly strong. The paper provides an overview of trends in Lithuanian agricultural structure, performance, and policies in recent years. It identifies the main features of farm companies and of their relations with the economic environment, based on survey data. And it presents and tests for specific hypotheses on the causes for the crisis in profitability and declining numbers of farm companies. Based on the literature it is argued that these trends and hypotheses are interesting beyond the case of Lithuania only. The findings can be summarised as follows.

Companies in the sample are much larger than the Lithuanian average. About half have specialised production structures, mostly in crops. Nonagricultural activities are often connected to food production, and are only a marginal feature in terms of farm incomes. Profitability has worsened considerably recently, and most farms are reportedly loss making. Most farms invested in capital goods during transition, most often financed by internal resources, with credit as the most frequent secondary source of finance. Farm managers report transaction problems in their dealings with the processing industry. Most often these take the form of late payment and disputes over price and quality. The institutional environment supports basic farm functions, but especially services in the areas of training and advice are less often available or taken up. Farm companies are found to have important employment and production linkages to the local, rural economy, particularly with the upstream part of the agribusiness chain and with local households.

An econometric analysis produces no evidence that large farm size as such is a cause of sub-optimal profit levels. But we do find that large management teams, and the costs related to them, tend to depress profit. The analysis also suggests that state support to farms, which takes the form of subsidies on credit, extension services, and inputs, may be biased towards the weaker farm companies, plausibly in order to prevent them from failing. Still, farm companies are fast disappearing, mainly through bankruptcies. It appears that much of present domestic support resources are wasted on farms with no profit potential, at the cost of restructuring the viable companies.

Another finding is that the usually large reliance on one sales channel is related to lower profitability level. This finding may well be indicative of a profit squeeze by the processing industry. The alternative of sales diversification is beset by higher transaction costs connected to direct marketing. Diversification into non-agricultural activities is rare, but where it occurs it is found to significantly contribute towards farm profitability.

These results lead to several reflections and policy implications. The most urgent and specific point in the Lithuanian situation appears to be that much of agricultural support is going to agribusinesses rather than farms (although concrete figures were not available). Moreover, what support is allocated to agriculture may well imply the wrong incentives in terms of restructuring, as explained earlier. Both observations would suggest that domestic agricultural policies in the areas of financial and specialist support should be reconsidered. Additionally, the issue of farm profit skimming by the processing industry would also merit further research. Chronic hold-up problems and perpetually changing transaction conditions and disputes are a strong disincentive to increase profitability in farms, since it is not certain the benefits will actually accrue to the farm in the longer term. Solutions may be sought in regulation or re-organisation of the processing industry and its transaction practices towards primary producers, or in the more effective build up of countervailing power, such as though the functioning of marketing cooperatives or boards.

Three more general points may finally be suggested. First, the replacement of corporate with individual agriculture, while perhaps beneficial in several respects, is not without costs. Individual agriculture has largely taken the form of small-scale, labour intensive micro-farms or household plots. While this may be a good thing in a time of high unemployment and impoverishment (as argued by Lerman and Schreinemachers, 2002), it may be problematic in view of the longer-term future of commercial agriculture. Where larger-scale, capital-intensive agriculture is not possible in the individual farming sector because of size limitations following from the land reforms, the corporate sector may be necessary to maintain an infrastructure for commercial agriculture. Perhaps this is no longer possible in Lithuania given the rapid decline of corporate agriculture there; but the point may be worth considering in the wider transition context.

Second, the struggle or demise of corporate agriculture, characteristic of most transition countries, is not clearly or exclusively caused by inherent flaws in the governance structures of these enterprises, as is often argued. While these internal factors may play a role, as our analysis indeed suggests they do, this study has also suggested many other reasons to do with policies and the institutional environment for the crisis in profitability. Also studies on other countries (Mathijs et al., 1999; Gorton and Davidova, 2001; Mathijs and Swinnen, 2001; Mathijs and Vranken, 2001) have shown that farm efficiency or competitiveness (let alone profitability) is not systematically connected to governance structure (family farms versus corporate farms). The focus in research and policy should shift towards enhancing the institutional and policy environment, rather than banking on the superiority of one farm type and neglecting to provide viable conditions for the alternatives. This is what seems to have happened in the Lithuanian land reforms and subsequent policies.

Third, supporting corporate farms may also be worthwhile from a rural development perspective. The existence of employment and production linkages implies that there are benefits in the local economy to viable rural enterprises beyond their output levels and profitability. While this must never be used to rationalise mistaken support polices which imply perverse incentives, it does serve to rethink the potential of corporate farms in the rural economy of transition economies.

Acknowledgements

We thank Michael Ellman, Matthew Gorton and an anonymous referee for helpful comments. Financial support by the European Commission under its Marie Curie research grant programme, contract QLK5-CT-2000-51251 is gratefully acknowledged. The authors accept sole responsibility for the contents of this paper.

REFERENCES

Alchian, A and Demsetz, H. 1972: Production, information costs, and economic organization. American Economic Review 62: 777-795.

Barry, P, Ellinger, P, Baker, C and Hopkin, J. 1995: Financial management in agriculture. Interstate Publishers: Danville, IL.

Bezemer, D. 2002a: Credit markets for agriculture in the Czech Republic. Europe-Asia Studies 54(8): 1301-1317.

Bezemer, D. 2002b: De-collectivization in Czech and Slovak agriculture: an institutional explanation. Journal of Economic Issues XXXVI(3): 723-745.

Bezemer, D. 2003a: Credit allocation and farm structures in Czech agriculture, 1993-1997. Comparative Economic Studies 45(1): 25-43.

Bezemer, D. 2003b: Credit systems in bimodal agriculture: The Czech Republic. In: Macey, D, Pyle W and S Wegren (eds). Market Institutions in Eurasian Agriculture: Origins, Operation, and Prospects. Lexington Books: Lanham, MD, USA, (forthcoming).

Bezemer, D, Zemeckis, R and Rutkaukaite, J. 2003: "Household Income Diversity in Rural Lithuania' University College London School of Slavonic and Eastern European Studies, Social Science Department working paper series, no. 32 at http://www.ssees.ac.uk/economic.html.

Csaki, C, Debatisse, M and Honisch, O. 1999: Food and agriculture in the Czech Republic, from a 'velvet" transition to the challenges of EU Accession. World Bank Technical paper no. 437, World Bank: Washington.

Davis, J and Gaburici, A. 1999: Rural finance and private farming in Romania. Europe-Asia Studies 51 (5): 843-870.

Deininger, K. 1995: Collective agricultural production: a solution for transitional economies? World Development 23(8): 1317-1334.

EBRD. 2003: Transition report. European Bank for Reconstruction and Development: London.

Eurostat. 1999: Income of the agricultural household sector. Luxembourg: Eurostat.

Evans, N and Ilbery, B. 1993: The pluriactivity, part-time farming, and farm diversification debate. Environment & Planning 25(7): 945-959.

Gaisford, J, Hobbs, J and Kerr, W. 2001: Making markets work: Modeling agri-food systems in transition. Canadian Journal of Agricultural Economics 49 (2): 181-202.

Gorton, M and Davidova, S. 2001: The international competitiveness of CEEC agriculture. The World Economy 24(2): 185-200.

Gorton, M, Kovacs, B, Mizik, T, Davidova, S, Rafinger, T and Iraizoz, B. 2003: An analysis of the performance of commercially-oriented farms in hungary. Post-Communist Economies 15(3): 401-416.

Gow, H and Swinnen, J. 1998: Up- and downstream restructuring, foreign direct investment, and hold-up problems in agricultural transition. European Review of Agricultural Economics 25(3): 331-350.

Greif, F. 1997: Off-farm income sources and uses in transition economies. Federal Institute of Agricultural Economics and Rome, FAO/REU: Vienna, Unpublished mimeo.

Hobbs, J, Kerr, W and Gaisford, J. 1997: The transformation of the agrifood system in central and eastern Europe and the new independent states. CAB International: Wallingford, UK/New York.

Hughes, G. 1999: The total factor productivity of farm structures in Central and East Europe. Bulgarian Journal of Agricultural Sciences 5: 298-311.

International Monetary Fund, World Bank, Organization for Economic Co-operation and Development, and European Bank for Reconstruction and Development. 1991: A study of the Soviet economy, Vol. III. IMF: Washington, DC.

Koford, K and Tschoegl, A. 1999: Problems of bank lending in Bulgaria: Information asymmetry and institutional learning. MOCT-MOST Economic Policy in Transition Economies 9(2): 123-152.

Krisciunaite, L and Uzdavineine, V. 2000: The structure of Lithuanian farms and their development in Lithuania. In: Agricultural Policy and Rural Development in the Baltic States. Lithuanian Institute for Agrarian Economics: Vilnius.

Laitner, J. 2000: Structural change and economic growth. The Review of Economic Studies 67(3): 545-561.

LIAE. 2001: Agriculture in Lithuania 2001. Lithuanian Institute for Agrarian Economics: Vilnius.

LIAE. 2003: Communications from expert interviews with staff from the Lithuanian Institute for Agrarian Economics.

Lerman, Z. 2001: Agriculture in transition economies: from common heritage to divergence. Agricultural Economics 26(2): 95-114.

Lerman, Z, Csaki, C and Feder, G. 2002: Land policies and evolving farm structures in transition countries. World Bank Working Paper 2794, Washington, DC.

Lerman, Z and Schreinemachers, P. 2002: Individual farming as a labor sink: Evidence from Poland and Russia. Paper presented at the Annual Meeting of the American Agricultural Economics Association, Long Beach, CA, July 2002.

LSB. 2001: Statistical yearbook of Lithuania. Lithuanian Statistical Office: Vilnius.

LSI. 2002: Statistical yearbook of Lithuania. Lithuanian Statistical Office: Vilnius.

Macours, K and Swinnen, J. 2000: Causes of output decline in central and Eastern European agriculture. Journal of Comparative Economics 28:172-206.

Mathijs, E, Blaas, G and Doucha, T. 1999: Organisational form and technical efficiency of Czech and Slovak Farms. MOCTMOST Economic Policy in Transitional Economies 9: 331-344.

Mathijs, E and Swinnen, J. 1998: The economics of agricultural decollectivization in Easte Central Euroipe and the former Soviet Union. Economic Development and Cultural Change 74: 1-26.

Mathijs, E and Swinnen, J. 2001: Production organization and efficiency during transition: an empirical analysis of East German agriculture. Review of Economics & Statistics 83 (1): 100-107.

Mathijs, E and Vranken, L. 2001: Human capital, gender and organisation in transition agriculture: measuring and explaining the technical efficiency of bulgarian and hungarian farms. Post Communist Economies 13(2): 171-187.

McNally, S. 2001: Farm diversification in England and Wales--what can we learn from the farm business survey? Journal of Rural Studies 17(2): 247-257.

Myers, S. 1984: The capital structure puzzle. Journal of Finance 39(3): 575-592.

OECD. 1997: Agricultural polices in transition economies. Monitoring and evaluation 1997. Centre for Co-operation with Economies in Transition of the Organisation for Economic Co-operation and Development: Paris.

OECD. 2001:: Agricultural finance and credit infrastructure in transition economies : Proceedings of OECD Expert Meeting, Porturoz, Slovenia, 2001. Centre for Co-operation with Economies in Transition of the Organisation for Economic Co-operation and Development, Paris.

Pederson, G and Khitarishvili, T. 1997: Challenges of agricultural and rural finance in CEE, NIS and Baltic Countries. Center for International Food and Agricultural Policy working paper 97-5.

Pryor, F. 1992: The red and the green. The rise and fall of collectivized agriculture. Princeton University Press: New Jersey.

Rizov, M. 2003: Endogenous production organization during market liberalization: Farm level evidence From Romania. Economic Systems 27(2): 171-187.

Rother, P. 1999: Explaining the behaviour of financial intermediation: Evidence from transition countries. IMF Working Paper 99/36. IMF: Washington, DC.

Sarris, A, Doucha, T and Matijs, E. 1999: Agricultural restructuring in Central and Eastern Europe: Implications for competitiveness and rural development. European Review of Agricultural Economics 26(3): 305-329.

Schmitt, G. 1993: Why collectivization of Agriculture in socialist countries has failed: A transaction cost approach. In: Csaki, C and Kislev, Y (eds). Agricultural Co-operatives in Transition. Westview Press: Boulder.

Shrieder, G and Heidhues, F. 1998: Access constraints of Romanian peasants in relation to the formal financial sector. Discussion paper 97/12, Center for Economic Reform and Transformation of Heriot-Watt University: Edinburgh.

Shucksmith, D, Bryden, J, Rosenthall, P, Short, C and Winter, D. 1989: Pluriactivity, farm structures and rural change. Journal of Agricultural Economics 40(3): 345-360.

Swain, N. 1999: Rural development and social change in the post socialist central European countryside. Zemedelska Ekonomika 45 (2): 79-84.

Swinnen, J and Gow, H. 1999: Agricultural credit problems and policies during the transition to a market economy in Central and Eastern Europe. Food Policy 24(1): 21-47.

Swinnen, J and Macours, K. 2000: Patterns of Agrarian transition. Economic Development & Cultural Change 50(2): 365-394.

World Bank. 1995: Farm restructuring and land tenure in reforming socialist economies. A comparative analysis of eastern and central Europe. World Bank Discussion Papers no. 268. World Bank: Washington, DC.

(1) We do not analyse the performance of corporate farms relative to family, or individual farms, since we use corporate farm data only; for such studies, see Hughes (1999); Macours and Swinnen (2000); Mathijs and Swinnen (1998, 2001); and Mathijs and Vranken (2001).

(2) Based on OECD (1997), LIAE (2001, 2003), EBRD (2003) and Krisciunaite and Uzdavineine (2000).

(3) These figures should be taken with caution, since registered numbers are very volatile (the 2002 number was 30,000) and plausibly sensitive to policies stipulating particular obligations or benefits dependent on registration or farm size.

(4) We note that other sources give different numbers, albeit reflecting the same trend: LSB (2001) reports a decline of agricultural companies from 1,594 to 628 in 1995-2000, their acreage falling from 804 thousand hectares to 293 thousand hectares. Divergences between statistical sources are plausibly related to the rapid and large changes in these few years.

(5) At the moment of writing (and surveying) the Litas was pegged to the euro, at a rate or 3.45 litai per euro.

(6) Only one farm sold more than half the turnover via direct marketing to retailers; only six farms sold more than a quarter of turnover to either consumers or retailers. But all except six farms responding to this question sell 70% or more of turnover through either the processor or wholesale channel.

(7) We note that the more direct measure of transaction problems reported earlier (managers' perception of the seriousness of various specific problems) was not included for statistical reasons. Since this variable is restricted to the 1-5 domain, its variability is likewise limited, with for instance a standard deviation of 0.8 on an average of 5.8 for 'late payment'. Another reason for low standard deviations is that these transaction problems were reported as serious by almost all farm managers; only six out of 51 respondents selected scores below 3 for 'late payment'. This implies that statistical association can only be weak (relatively to other variables with higher standard deviations). This was true for all transaction problems values reported in section 4.4 as well as for combinations of them. Regression may then not be the best method for examining the relation of this variable with profitability.

(8) We also observed management education levels, but these were too similar (agricultural college/university) across respondents to be used in the regression.

(9) Ranges were defined based on LIAE expert estimates. They have bandwidths of 20,000 litai (5,800 euro), except the top and bottom bands, which are open-ended. The 12 ranges are: band (12): losses of more than 100,000 litai (29,000 euro); band (11): losses from 99,000 to 80,000 litai; band (10): from 79,000 to 60,000 litai; etc, through to band (7) losses from 19,000 and 0 litai; band (6) profits from 1 to 20,000 litai; band (5) profits from 21,000 to 40,000 litai, through to band (2) profits from 81,000 to 100,000 litai; band (1) profits of more than 100,000 litai.

(10) An alternative would be the Tobit specification since the dependent variable is censored at top and bottom while the underlying distribution is not. This alternative is, however, inferior to ranked values and ordered probit estimation. Noise would be produced by the necessity to assume some value within each profitability band (eg the band average) as dependent. The Tobit specification was explored, but did indeed produce a slightly poorer model fit (eg a pseudo [R.sup.2] of 0.1753) and fewer significant findings, with identical signs as in the probit case.

(11) Examples abound. Compare, for instance, a recent study by Rizov (2003) of farm structures in Romania, with a larger sample (1,394 farms), a multinomial regression model, and a pseudo [R.sup.2] value of 0.1227.

(12) The often observed non-linear age effect was explored by also including the square of AGE, but its coefficient estimate is highly insignificant.

DIRK J BEZEMER (1), DONATAS STANIKUNAS (2) & ROMUALDAS ZEMECKIS (2)

(1) Department of Economics, University of Groningen, The Netherlands. E-mail: [email protected]

(2) Lithuania Institute of Agrarian Economics, V.Kudirkos 18, LT-2600, Vilnius, Lithuania

Table 1: Agricultural production structure in the sample

Variables Sample farms (2002, n = 53)

 Only farms with nonzero values All
 sample
 farms

 n Minimum Maximum Mean s.d. Means

All land used (ha) 51 29 3,356 834 724 833

Crops
 Cereals 51 10 1,620 421 390 405
 Leguminous plants 15 5 91 33 26 9
 Oilseeds 17 18 300 93 68 30
 Root crops 32 1 210 60 65 37
 Vegetables 10 7 33 13 8 2
 Fruit 4 3 1,526 73 64 6
 Other crops, 22 2 910 190 290 79
 meadows

Livestock (head)
 Milk cattle 32 41 669 220 141 133
 Beef cattle 29 43 1,183 352 264 193
 Pigs and hogs 26 87 14,120 1,272 2,732 624

Labour force (full 53 4 168 47 40 47
time equivalents)

Variables All Lithuanian
 farm companies
 (2000, N = 628)

 Means

All land used (ha) 692

Crops
 Cereals 295
 Leguminous plants 12
 Oilseeds 17
 Root crops 53
 Vegetables 17
 Fruit 0
 Other crops, 309
 meadows

Livestock (head)
 Milk cattle 698
 Beef cattle 493
 Pigs and hogs 1,362

Labour force (full 42
time equivalents)

Source: Survey findings, LSI (2002)

Table 2: Agricultural and non-agricultural activities

 Share of income source
 in total income (%)

 1990 1995

Agricultural production
 Crop production 29 43
 Livestock production 62 46
Non-agricultural activities
 Food processing 3 1
 Transport 0 3
 Machinery repair, 6 7
 sale and rent
 Other 0 1
Total 100 100

 2002

 Share of income Incidence in Labour
 source in total sample (% of allocation
 income (%) total sample) (fte)

Agricultural production 42
 Crop production 46 81
 Livestock production 40 64
Non-agricultural
activities 5
 Food processing 2 3
 Transport 1 17
 Machinery repair, 9 50
 sale and rent
 Other 1 7
Total 100 NA 47

Source: Survey findings

Table 3: Access to and use of business services and institutions

Business services and institutions Access Use of
 to services/ services/
 institutions institutions

 Count % Count %

Agricultural services and institutions
 Agricultural extension or technical
 advice 41 77 24 45
 Veterinary services 50 94 38 72
 Agricultural marketing co-operative 23 43 5 9
 Other agricultural producer
 association services 46 87 35 66
 Farm management training 35 66 15 28

General services and institutions
 Accountancy advice 53 100 51 96
 Business advice 31 58 11 21
 Information centre on rural
 development policies 17 32 4 8
 Labour exchange/job centre 53 100 19 36
 Bank 53 100 43 81
 Insurance company 53 100 53 100

Source: Survey findings

Table 4: Ordered probit regression of profitability levels

 Coefficient Standard Z
Independent variables estimates errors statistics

FAVOUR 1.4083 (***) 0.5047 2.79
AGE 0.0549 (*) 0.0297 1.85
EXPERIENCE 0.0298 (*) 0.0154 1.93

LAND 0.0013 (**) 0.0005 2.37
LABOUR 0.0090 0.0072 1.24
MANAGEMENT -0.3185 (**) 0.1422 -2.24
CATTLE 0.0002 0.0009 0.20
PIGHERD 0.0003 (***) 0.0001 3.24

CREDIT -2.0232 (**) 0.8689 -2.33
EXTENS -1.1424 (**) 0.4651 -2.46
CONSULT 0.8170 (**) 0.4072 2.01

CHANNEL DEPENDENCE -5.3204 (**) 2.2611 -2.35
DIRECT MARKETING -1.7935 (**) 0.8434 -2.13

DIVERS 0.0353 (**) 0.0163 2.16

Regression statistics
 Dependent variable Profit level band (values 1-12)
 # observations 46
 LR [chi square] (14) 43.19
 Prob > [chi square] 0.0001
 Pseudo [R.sup.2] 0.2501
 Log likelihood -64.76918

Notes: Multicollinearity was tested for by regressing independent
variables on each other, with adjusted [R.sup.2] levels below 0.70.
Covariance and covariation matrices contained small and insignificant
values between any pair of variables. One asterisk (*) indicates that
the marginal effect is statistically significant for P<0.10; two
asterisks (**) indicates significance for P<0.05; with three asterisks
(***) significance also holds at P<0.01.

Source: Survey findings and authors' calculations