Inertia of peasant farms in European Russia in the second half of the 19th - early 20th centuries: statistical analysis of data on sown areas and marketability of grain

Rozinskaya Nataliya Anatol'evna ORCID: 0000-0002-6907-0440 PhD in Economics Associate Professor; Faculty of Economics; Lomonosov Moscow State University 1 Leninskie Gory str., Moscow, 119991, Russia natalja.rozinskaja@yandex.ru Chapligina Irina Gennad'evna ORCID: 0000-0002-3407-2462 PhD in Economics Associate Professor; Faculty of Economics; Lomonosov Moscow State University 1 Leninskie Gory str., Moscow, 119991, Russia chaplygina@econ.msu.ru

Sorokin Aleksandr Sergeevich ORCID: 0000-0002-9328-7017 PhD in Economics Associate Professor; Higher School of Cybertechnology, Mathematics and Statistics; Plekhanov Russian University of Economics 36 Stremyanny Lane, Moscow, 115054, Russia alsorokin@mail.ru

1. Introduction

The inertia of peasant farms as a phenomenon characteristic of the economies of various regions and epochs has been repeatedly studied by economists ^[1-6]. In this paper, we plan to use statistical data on Russia at the end of the 19th – beginning of the 20th century to verify the validity of this hypothesis.

But first of all, it is worth recalling what is meant by the concept of peasant farming. According to T. Shanin, in Western literature, this type of economy was allocated to a separate category in the late 1960s and early 1970s, when, as a result of post-war decolonization and the formation of programs to support independent countries, economists encountered a special economic structure characteristic of these countries, where the dominant role is played by family farms engaged in agriculture. ^[7]. From Shanin's point of view, this rediscovery of the "peasant world" is of great importance, since its special organization is the "determining factor" of the most acute social phenomena of the era: "the Vietnam War, Indian poverty, Latin American guerrillas, African stagnation and the Chinese "great leap forward" ^{[8, p.8]}.

Shanin identifies four key characteristics of peasant farming as a special economic phenomenon, in fact, a special type of economic agent or institution.: 1) the family nature of work, which does not lead to an autarkic economy that is actively involved in commodity turnover, but determines the family principle of the division of labor, as well as the priority of family rather than individual needs; 2) agriculture as the main (only) sphere of production, which leads to low specialization (peasants perform a lot of different tasks). 3) a special type of behavior characterized by traditionalism (past experience), conformism (community pressure), and normative control over each other; however, Ellis disputes this characteristic. speaking of which, peasant farms adapt perfectly to changing conditions, although perhaps not very quickly ^{[9, p. 5]}; 4) subordinate position in the general socio-political hierarchy, extreme remoteness from sources of power ^[8]. It is worth noting that, unlike Shanin, B. Galensky emphasizes precisely the autonomy of the peasants' labor ^{[10, p. 104]}, who are able to fully support themselves, unlike all other industries that cannot exist without farmers. This is an old idea that is easy to find in the works of French economists of the 18th century, such as F. Quesnay or A. Turgot, but which rather refers to the agricultural sector as a whole and does not exclude the existence of trade relations between farmers and other industries. She simply does not consider such connections as inevitable, or rather considers them necessary only for the city and industry, but not for the countryside.

In Russian and Eastern European literature, peasant farming as an independent phenomenon became the subject of economic analysis much earlier. Famous works of the organizational and production school under the leadership of Chayanov, Chelintsev, etc. Russian economists based on a detailed analysis of empirical data back in the early The 20th century has shown the specificity of peasant behavior and the dynamics of their production. A. V. Chayanov introduces the concept of a family-labor peasant farm, which implies: 1) the family nature of work, including that dictated by the lack of a developed rural labor market; 2) the lack of other types of income other than labor (while agricultural labor is combined with the practice of latrines, which Chayanov considers an important characteristic); 3) dependence on climatic conditions; 4) dependence on the demographic cycle of the family (ratio the number of consumers and workers), as well as the presence of a constant 5) problem of low land ^[11].

Chayanov emphasizes the complexity of the target function of such farms. He writes that the result of economic activity appears in the form of total labor income, which cannot be divided into the classical categories of wages, capital gains, rents, etc. ^[12]. We find a similar view in Galensky, who writes that "in many cases it is impossible to separate the production and consumer aspects of investments" ^{[10, p. 112]}. The same idea is important in the concept of "rural consumer economy" by N. P. Makarov, where the employee, organizer, and owner are united and appear as one economic entity (peasant, family) ^[13].

If we turn to the work of Ellis, who tried to consider all the existing patterns of behavior of the peasantry at the beginning of the 20th century, he identifies three key characteristics: 1) a family farm, 2) partially included in 3) an undeveloped and imperfect market ^{[9, p. XIV]}.

2. The hypothesis of inertia of peasant farms

The hypothesis about the inertia of peasant farms was put forward by Russian economists at the beginning. XX century. It was a response to the problem they identified, which was that peasant farms maintained a low marketability of their production.: they did not seek to increase the volume of bread sold in response to an improvement in market conditions, which contradicts the principles of rationality accepted in economics.

2.1. N. D. Kondratiev's version

The problem of the inertia of the grain market is posed by the Russian economist of the beginning. By N. D. Kondratiev in connection with the analysis of the problems of grain procurement during the First World War and the Revolution ^[1]. Kondratiev argues that the difficulties in achieving the growth of marketable bread are largely related to the high share of peasant farms in the market (about 85-90% before the First World War. in the total amount of acreage). By inertia, Kondratiev means the tendency of the peasant grain market to reduce the marketability of grain in favorable market conditions.

Kondratiev cites data (see Table 1) demonstrating the bread marketability rate (the ratio of the volume of transported bread to the gross harvest) of owner and peasant farms in different regions and types of bread, which shows that peasant farms, in principle, are much less likely to throw bread on the market.

Table 1.

Marketability rates in peasant and owner-owned farms by different regions and types of grain (%)

Source: Kondratiev, 1922, Ch.1. p. 5; O – the general marketability rate for all farms, K – for peasant farms, B – for owner farms.

But the problem of inertia is that in conditions of high grain prices, the marketability rate of peasant farms decreases. Kondratiev illustrates his hypothesis with the following data:

- based on peasant budgets for the Simbirsk province (grain-producing) and the Volokolamsk district (bread-consuming) Kondratiev demonstrates the growth of the revenue side of the budget of peasant farms over the expenditure side in 1914-1916.;

- provides data showing an increase in the consumption rate of peasant farms in the period from 1911 to 1915 in 5 producing provinces (14.9 poods per capita in 1915 versus 13 poods in 1911-1913);

- provides data on a sharp drop in the marketability of bread from 1909 to 1915 (from 12.4 to 7.4 for all loaves)^[1].

N. Kondratiev explains such dynamics by the peculiarities of motivation of peasant farms. He writes that as the family budget surplus grows, peasants lose the incentive to throw bread on the market and prefer to increase their internal consumption rate. This process is facilitated by the fact that in pre-revolutionary Russia, bread consumption rates were quite low compared to European countries (England, France, Belgium, Germany). Thus, Kondratiev explains the main reason for the decline in marketable bread by the fact that the peasants themselves are consumers of their products and, in conditions of increasing profitability of their farms, reduce the share of bread sold by increasing their own consumption. Kondratyev sees another reason for the low development of industrial goods markets in rural areas. The poor assortment of goods does not encourage peasants to use their bread for exchange.

2.2. A.V. Chayanov's model of avoiding peasant labor

A similar observation was made by the organizational and industrial school and described, in particular, in the works of A. V. Chayanov ^[11]. But they provide a different explanation for this phenomenon. For Chayanov, the key characteristic is the labor character of a peasant farm, which forces him to interpret the motivation of such a farm differently. Chayanov draws a graph of the balance of the economy, illustrating the behavior of such farms, similar in idea to the graph once constructed by U. Jevons for an economic agent ^{[14, p.125]}. Chayanov writes that in the labor economy, an increase in well-being is invariably associated with an increase in labor costs, "but the expenditure of physical energy for the human body is far from unlimited. After a relatively small expenditure required by the body, further expenditure of energy already requires volitional effort" ^{[2, p. 70]}. Taking into account the fact that the usefulness of the product, which is delivered by each subsequent unit of labor, tends to decrease, Chayanov concludes that such a labor economy will achieve the optimal size of production at the point where the curve of decreasing utility CD intersects the curve of increasing labor intensity AB (see Fig. 1).

Figure 1.

Figure 1.a Figure 1.b

Source: A.V. Chayanov, Essays on the theory of labor economy// A.V. Chayanov. Peasant farming. Selected works. M., Economics, 1989, pp. 71, 73.

Chayanov is interested in how such a labor economy will behave with an increase in labor productivity. It should be noted that this growth appears as an increase in the value of a product produced by one unit of labor, thus, Chayanov's reasoning is applicable to a situation when the value of manufactured products increases due to an improvement in market conditions, which formed the problem of inertia of peasant farms in N. Kondratiev's terms. Using data from the budgets of small and medium-sized peasant farms in Switzerland for 1910, Chayanov shows that in farms with a higher value of labor, income per consumer is also higher, but this difference is not as significant as it should be. Further, he mathematically proves that with an increase in labor productivity, the growth rate of income caused by the transition of the economy to a new point of equilibrium will be relatively lower. Although an increase in labor productivity leads to the fact that at the previous point of equilibrium, the severity of labor is now lower than the value of the product it brings, and this will force the peasant to increase the amount of labor applied, nevertheless, since the value of additional goods will decrease each time with an increase in these volumes, this increase in volumes will not be proportional to the increase in productivity. It will be smaller. Chayanov's conclusions are graphically presented in Figure 1.

Based on his reasoning, Chayanov argues that the inertia of peasant farms is an inevitable consequence of the labor nature of the economy, in which the described behavior corresponds to a rational strategy. Arguing that the burden of labor is extremely high in the Russian countryside due to its low productivity, Chayanov concludes that peasant farms are rapidly reaching a point at which the usefulness of further income growth does not compensate for the increasing inefficiency of additional labor efforts.

2.3. Self-employed behavior model

In modern economic theory, there is a model of the self-employed, which assumes that stationary work, the desire to give up work after reaching a certain income level, is generally characteristic of those economic agents who use their own labor, which means that they perceive an increase in production volumes as an increase in the hardships of work. In such a model, which practically repeats the already mentioned theory of According to Jevons, economic agents achieve maximum individual benefit by working only as long as the increase in income exceeds the increase in the hardships of work. In modern literature, such a behavior model is attributed to representatives of different professions, for example, taxi drivers ^[15].

2.4. Non-economic motives for the preservation of small-scale farming

Polish economist B. Galenski ^[10], recognizing that the desire to expand one's economy is traditional, draws attention to the fact that this is a motive "rarely found in today's Poland" ^{[10, p. 112]}. Explaining this phenomenon, he focuses not on economic, but rather on the social factors that shape this behavior. In his opinion, the increase in the scale of the economy is associated for the peasant (he calls them farmers) with a change in the class and professional status of the family.: 1) the peasant loses the opportunity to work in other production areas and may not like such a narrow specialization; 2) he loses time to help his neighbors (which is important given the communal rural culture); 3) he is forced to switch to labor management not only by a member of his family, but also by hired workers, which requires him to skills; 4) he himself loses the functions of a producer and becomes a pure entrepreneur; 5) he is forced to enter larger markets and become a professional trader. Thus, the expansion of a farm is not a quantitative, but a serious qualitative change, which can slow down the process of enlarging farms. In general, Galensky attributes the conservatism of peasants to the fact that a peasant's industrial life is almost inextricably linked to his personal lifestyle. Therefore, changes in production processes are perceived more painfully by them, since they lead to changes in this family structure ^{[10, p. 116]}.

3. Empirical analysis

One of the possible ways to empirically confirm the inertia of peasant farming is to analyze the reaction of peasants to changes in grain prices, that is, to analyze the dynamics of acreage in response to changes in grain prices. At the first stage of the study, we tried to find this relationship at the macro level using data from 50 provinces of European Russia. At the second stage, the same dependence was investigated, but using the example of a single province. According to N. P. Makarov, an increase in the price of bread does not mean an increase in the incomes of peasant farms. Since their structure is very complex and economic indicators are mixed, the impact of price dynamics on income levels should be the subject of a separate study. In this regard, the approach used by Kondratiev, analyzing the dynamics of peasant household budgets, seems to be more competent. But our proposed study nevertheless also seems to be correct.

Of course, the volume of plowing depends not only on price dynamics, but also on the development of transport routes, elevators, the development and nature of the trading apparatus, and climatic conditions over a long period of time. To account for these effects, we, firstly, used the data available to us to model control variables, and, secondly, we considered in parallel the dynamics of the acreage of peasant and owner farms. Since farms are located in the same temporal and spatial conditions, assuming that the detection of differences in the behavior of peasants and landlords in response to prices can serve as an argument in favor of the thesis of the inertia of the peasant economy. The reason for the differences in the reaction to prices may be the economic structure and non-market behavior of the peasants. At the same time, we realize that there is another important difference between peasant and private farms, which can influence their economic behavior – different access to information ^[16]. Unfortunately, we were unable to find data on resellers' prices, which the peasants were most likely targeting, so we assumed that their dynamics coincided with market prices, meaning that as market prices increased, resellers' prices also increased. However, we understand that this factor could also influence the formation of differences.

3.1. Data used

For empirical analysis, we used Gubernian data on acreage, labor price, level of urbanization, prices and yields of three grain crops (wheat, rye and oats) for the period from 1881 to 1913. The main sources were collections of the Central Statistical Committee of the Ministry of Internal Affairs, the Statistical Department of the Department of Agriculture and Rural Industry (since 1894, renamed the Department of Rural Economics and Agricultural Statistics of the Ministry of Agriculture and State Property, which since 1905 has been reorganized into the Main Directorate of Land Management and Agriculture), Zemstvo statistics. A database was compiled based on these sources^[17]. For more information about the data sources for each time period, see Appendix A1.

Due to the fact that there is a discussion in the scientific literature about the reliability of pre-revolutionary crop statistics ^{[18; 19]}, before proceeding to the description of the specifications of the econometric models under consideration, we would like to discuss the reliability of the statistical data used.

There were several data sources in pre-revolutionary Russia. Until 1880, information about crops and grain harvests was recorded in the governor's reports. The information was collected by the national food commissions through the county leaders of the nobility, the chambers of state property and the appanage offices. With the introduction of zemstvo institutions in 1864, the national food commissions were abolished and their duties assigned to the zemstvo bodies. The smallest unit where a survey was conducted to obtain crop data was the parish. The data collected by these authorities was much more reliable, but, nevertheless, the opinion was confirmed in the literature that the data from this period should be used with great caution ^{[20, 21]}. Therefore, we decided to use data from 1881, when the registration of harvests was entrusted to the Central Statistical Committee (CSC).

To obtain the necessary information, the Central Statistical Committee sent question forms to each volost; some of the forms were intended for peasants, separately for those who had large, medium or small allotments (6 forms per volost), some for private owners, separately for large farms, for medium and for small ones. The forms asked questions about the number of tithes sown with each type of grain, the number of seeds sown and the amount of bread harvested from this area. The average figure of sowing and harvesting of each type of grain (per tithe) for each county was derived from the data obtained. In addition, since 1870, the division into peasant and owner lands has been adopted, and headings have also been added separately for each type of grain (before that, they were divided only into "spring" and "winter"). As a result, fairly reliable figures were obtained ^[20].

Each year, the CSC received testimony from about 150 thousand farms, which were evenly distributed throughout the country; therefore, the average conclusions obtained in relation to these farms can rightfully be generalized and extended to all other farms ^[20]. Among peasant farms, the choice of typical ones for a given area was not difficult, and the choice of owner farms was often quite random, so information related to them is probably less reliable than information about harvests on peasant lands.

Simultaneously with the Central Statistical Committee, in 1880, the Statistical Department of the Department of Agriculture and Rural Industry began collecting and developing periodic information on harvests. The Department received information from volunteer correspondents from among the rural owners. The information provided by the correspondents themselves has considerable reliability, but since the correspondents of the Department themselves undoubtedly belonged to the best hosts, conclusions based on this information, when extrapolated to all farms in the country, may not be representative.

The Department usually had more correspondents from the strata of private owners than from peasants; therefore, obtaining information about owner farms is comparatively better. Due to the lack of correspondents, the Department used CSK data to develop its gubernatorial figures starting in 1884.

Some zemstvos were also engaged in crop registration. They received information, like the Department of Agriculture, from volunteer correspondents; but they had more of the latter than the Department, so their information is more reliable ^[20].

Many researchers compared the data obtained by the CSK and the Department of Agriculture, and they all came to the conclusion that the difference between them is very small ^[22]. And since conclusions about harvests were made by the CSK and the Department of Agriculture based on data they received from different sources and in different ways, the proximity of these conclusions to each other may indicate that they correctly reflect the existing reality ^[18]. In addition, Kovalchenko showed that the data of the CSK and the Department of Agriculture on the area of crops are confirmed by the materials of the agricultural census of 1916 ^{[23, p.48]}. Thus, the researchers conclude that the materials of the Central Statistical Committee meet the minimum reliability criteria and can be used for research, since "... the dynamics of these data are transmitted fairly accurately due to ... the uniformity of collection methods throughout the entire period of 33 years" ^{[18, p.210]}.

To test the hypothesis at the regional level, we used data from zemstvo statistics. As noted above, the data of the zemstvo statistics are considered to be fairly reliable and reliable. The problem is that they are difficult to use at the macro level, as different zemstvos had their own methods and programs. But these shortcomings do not affect the data in the study of a single province. That is, the zemstvo data on the Kherson province at the county level may well be used to test the hypothesis outlined above.

3.2. Descriptive statistics of the data used

For macro-level modeling, we took statistical data on 50 provinces over 33 years in panel data format. The panel was unbalanced, however, considering the nature of the omissions to be exogenous and taking into account that their number as a percentage of the number of observations varied on average from 0 to 25%, we used the same methods of parameter estimation as for balanced panel data. Table A2-1 of Appendix A2 provides descriptive statistics of the initial quantitative endogenous and exogenous variables. It should be noted that most variables have significant variation, the average value is very different from the median. This is due to the specifics of the data used – panel data for 50 provinces over 33 years. Within one province, the distribution of variables is close to normal, so we did not transform the data for the final models. However, when calculating and searching for optimal models, the authors also used logarithmic values of variables that have an exponential or logarithmically normal distribution over the entire time period.

For modeling at the regional level, we used data on 6 counties of the Kherson province for 20 years, a total of 120 observations in panel data format, data on crops and harvests for 1904 and 1908 are missing.

Table A2.2 of Appendix A2 provides descriptive statistics of the initial quantitative variables. The panel was also unbalanced, as in the macro-level simulation.

4.1. Modeling at the macro level

To test the hypothesis about the difference in the effect of grain price changes on the amount of plowing between peasant farms and landowner farms for three grain crops (rye, wheat, barley) We have built two panel data models – for peasant farms and for landowner farms. We were guided by the fact that if in these models we obtained statistically significant coefficients of the same sign for peasants and landlords with variable prices, then the hypothesis of the study was rejected, since this clearly meant that peasants and landlords had the same reaction to price changes. If the statistically significant coefficients were different, or the coefficient was statistically significant in only one of the models (for peasants or landlords), then the hypothesis of the study was confirmed, since this meant that peasants and landlords had different reactions to price changes.

The basic model for estimating the parameters was a panel data model of the form:

𝑌𝑖𝑡=𝛽0+𝛿∗𝑡+𝛽1∗pi𝑡+𝛾∗𝑍𝑖+𝜀𝑖𝑡, (1)

where

𝑌𝑖𝑡 - sown areas of the corresponding crop for peasants or landowners for the 1st province in the year t (government tithes);

i = 1, ...., 50 – the index of the province;

t = 1, ... , 33 – time index;

pi𝑡 is the price of the corresponding crop in autumn (kopecks per pood) for the 1st province in t year or t–1 year;

𝑍𝑖 is a variable that characterizes the specific features of each 1st province;

𝛽0, 𝛿, 𝛽1, 𝛾 – coefficients.

𝜀𝑖𝑡 - accidental error.

It should be noted that for rye, a winter crop, we took prices in the same year as the acreage. For spring wheat and barley, we took prices with a lag of 1, because sowing for these grains takes place in the spring, and at that time the peasants and landlords could only focus on the prices of the previous year.

To estimate the parameters of equation (1), we performed an intra-group regression, considering a fixed-effects model that explains the variation of the dependent variable around the average value for a group of observations related to a given object from the variation of the average independent variable:

, (2)

where

; ; .

The choice of a fixed effects regression panel data model versus a pooled regression model or a random effects model was based on appropriate statistical tests: the Hausmann test, the Breusch-Pagan test, and the linear constraint test. For each grain crop and for each type of economy (landowners and peasants), the choice was made in favor of a fixed-effects model, which corresponded to our logical expectation: each province had an individual specific level of reaction.

Note that in the specifications of the equations we considered the lagged values of the explanatory variable. This empirical strategy makes it possible to reflect the reactive (rather than proactive) behavior of peasant farms.

It should be noted that the main problem we face when estimating the parameter with a variable crop price in equation (2) is the problem of endogeneity. Given the discussion about the reliability of pre-revolutionary crop statistics described in detail above, we can talk about a possible bias and inconsistency in estimating the coefficient due to errors in measuring the regressor, including due to an unbalanced panel.

Another classic reason for endogeneity is the non–inclusion of regressors that significantly affect the dependent variable in the model. In our case, the task was not to accurately predict the acreage of crops using multiple regression. The aim was to test the hypothesis of inertia of peasant farms by the sign of a significant regression coefficient of only an independent variable price. However, we are aware that household decisions are influenced not only by grain prices, but also by many other factors, including, for example, weather conditions. To solve this problem, we used control variables, when they were included in the regression equation (2), we analyzed in detail the change in the sign of the coefficient and its significance for a variable price. Despite the strong limitation of the choice of variables due to the availability of data for the period under review, we built models with different combinations of control variables, the general appearance of the specifications of which is as follows:

, (3)

where

– values of the jth control variable in deviations from the provincial average:

x 1 – yield from one government-owned tithe of land of landowners or peasants of the corresponding crop (poods);

x 2 – the population in the province in villages (thousand people);

x 3 – the population in the province in cities (thousand people);

x 4 is a binary variable of whether or not there was a harvest in the province;

– coefficient for the jth control variable.

Considering equations (1)–(3), it can be assumed that the price and volume of agricultural output are endogenous variables determined as a result of the interaction of two equations: supply and demand.

To solve this problem, we used two strategies: firstly, the use of a dynamic model for panel data, where lagged variables are used as instrumental variables ^[24] and, secondly, we used a model with an instrumental variable x 5 - the labor cost of a worker on foot (on his grub) per sowing (in pennies), through which the endogenous variable of the crop price was estimated:

, (4)

where

– the price of the corresponding crop in deviations from the average for the group, obtained through the instrumental variable x 5.

We assume that the price of labor can be used as an instrumental variable for the price of grain, since the cost of labor, being part of the cost, can affect the price of the grain received. At the same time, given the fact that peasants mostly cultivate their own land, the cost of labor does not play a significant role when deciding whether to increase or decrease acreage. To test our assumptions, we constructed a correlation matrix of the variables under discussion. The selected instrumental variable has no significant correlation with the dependent variables for spring crops, a slight correlation with the dependent variable for rye, as well as an average correlation with all independent price variables.

For peasant farms, we also additionally tested the hypothesis of a change in peasant behavior after 1906. We suggested that the Stolypin reform, which was associated with changes in land ownership rights and stimulated peasants to leave the community, could help change the motivation of peasants, promote more rational behavior, bring their behavior closer to a more market-oriented one, that is, reorient peasants from maximizing income to maximizing profit. In this case, the coefficient of elasticity of the acreage of a certain crop for the price of this crop should have been positive.

To test the hypothesis of changes in peasant behavior after the Stolypin reform, we introduced a fictitious variable into equation (4) to assess the structural shift and its interaction with price to assess a possible change in the sign of the coefficient during regression:

, (5)

where

– a fictitious variable to reflect the period before and after the Stolypin reform.

4.2. Simulation results at the macro level

Of all the models constructed within the framework of the specifications outlined above to test our hypothesis of inertia of peasant farms at the macro level, a significant result was obtained for two families of models: models of the dependence of rye acreage for peasants and models of the dependence of acreage of all types of grain for landlords. To a certain extent, this can be explained by the fact that rye was the main crop of the peasants, since the main purpose of the peasant economy was to satisfy their own needs. Wheat was relatively expensive for the peasants, so they mainly consumed products made from rye flour and, accordingly, rye was the main crop for the peasants ^[25].

For peasants in the model according to equation (2), we obtained a negative regression coefficient at price, but it turned out to be statistically insignificant (see Table 2). After entering control variables and evaluating the parameters of equation (3), the coefficient remained negative and insignificant. Such results are quite expected due to possible problems with endogeneity. After entering the price obtained through the instrumental variable into equation (4), we also obtained a negative regression coefficient, significant at 0.05.

Table 2.

The results of the evaluation of coefficients of models of dependence of the sown area of rye for farmers at the macro level

* the regression coefficient is significant at the level of p<0.05

Calculated by the authors based on the data in Appendix A1.

For all models, the results of statistical tests for linear constraints, the Breusch-Pagan test and the Hausman test, confirmed the correctness of choosing a model with fixed effects compared to a model with random effects or complete regression.

From the estimates of the coefficients of the models under consideration for peasants, we can assume that farmers, with an increase in prices, were most likely inclined to reduce their rye acreage. For landlords, we observe a positive significant regression coefficient for all models (see Table.3): with the price increase, landlords reacted by increasing the acreage of wheat, i.e. they acted as capitalist agents according to the principle of profit maximization. Thus, the hypothesis of a difference in the reaction to rising grain prices between peasant farms and landowner farms was confirmed.

Table 3.

The results of the evaluation of coefficients of models of dependence of the sown area of wheat for landowners at the macro level

** the regression coefficient is significant at the level of p<0.01

*** the regression coefficient is significant at the level of p<0.001

Calculated by the authors based on the data in Appendix A1.

5.1. Formulation of a hypothesis at the regional level (Kherson province)

The Kherson province was deliberately chosen for research at the regional level as an example of one of the most export-oriented provinces, where market relations could presumably be more developed than in other regions. The discovery of "non-market" behavior of peasant farms in this province could serve as more reliable evidence that this behavior was typical for most regions of Russia. Kondratiev in his work "Bread Market ..." considered the Kherson province as an example of a region demonstrating high marketability of bread. Accordingly, according to the structure of the economy, this province should have shown a tendency to inertia to a lesser extent, and the hypothesis of a reduction in acreage in response to rising prices was the least likely to be confirmed. At the same time, since this is a grain-producing province, and it was in such provinces that Kondratiev showed a significant increase in consumption rates in response to an improvement in the market, the inertia effect may well be detected. And if it is found in such a province, it will indicate the stability and prevalence of such behavior among Russian peasants.

The following most important characteristics of this province are highlighted in the literature: underdeveloped infrastructure (roads, elevators), underdeveloped financial market, the presence of a large number of small dealers, weak opportunities for consumption growth or savings. With regard to grain trade, the Kherson province represented a wheat export region (Kondratiev gives a figure of 86.6% of exports in relation to all marketable bread), which was serviced by a number of major ports. Proximity to ports ensured the smallest gap between local and exchange prices ^[1], which is also important for the study in connection with the mentioned possible asymmetry of price information between peasant and owner farms, the majority of the region's population was engaged exclusively in agriculture. The factory industry was developed only in the south-east of the province, in the Krivoy Rog region.

The agricultural sector had an extensive grain character, which meant the desire to produce as much grain as possible and harvest it as soon as possible. This contributed to a more active use of machinery, but farming in the region was practically undeveloped. Most of the harvested crops were consumed on the farm. The remaining part and grain imported from other regions were exported to the southern coastal cities (Nikolaev, Odessa) and from there were sent abroad. The Kherson province, due to the proximity of its ports, was the largest grain collector for further shipment abroad.

Despite the export nature of grain trade, the organization of local trade, as well as in non-export regions, was poorly developed. Small-scale intermediation, small-scale purchase of peasant bread, was quite widespread, leading to the grain passing through the hands of a significant number of intermediaries. The purchase of grain from the peasants took place for the most part on the spot in villages, railway stations, marinas, etc., where the peasants brought grain. It was bought by agents of larger firms, as well as small buyers and resellers. It was difficult to sell directly by export firms, even for large landlords, because grain trading required a lot of trouble and a lot of acquaintances among exporters ^[16].

Farmers, unable to leave grain until the next harvest, were forced to sell it in early autumn, while large producers sought to sell the crop at the most favorable time of the year, sometimes in autumn, sometimes delayed the sale until spring.

The provision of advances and loans has been relatively little developed. Most buyers worked at their own expense, and most did not have large amounts of capital; some worked at the expense of firms on a commission basis. For the most part, the demand of local buyers was completely dependent on the conjuncture of foreign markets and their requirements. Prices in the Kherson province were set mainly under the influence of the southern ports, and wheat was no less influenced by flour mills, especially in the areas where mills were concentrated; for other types of grain, they are set solely by export demand ^[16].

It is worth noting that the above-described features of trade organization can be found in almost all export-oriented provinces. This gives us some reason to assume that the results obtained can be attributed to other similar regions.

5.2. Modeling at the regional level.

For the Kherson province, to test the hypothesis of our study, we selected the same models as in the previous section. The methodology of econometric modeling was similar, we considered panel data models with fixed effects of the form (2), as well as models with random effects of the form:

, (6)

where

– Individual random effects.

Due to limitations in the available source data, we were unable to use control variables, but to neutralize the effect of shifting the coefficient value due to the problem of endogeneity, we used an instrumental yield variable through which crop prices were estimated, and also considered a dynamic panel data model with the introduction of a lag value of the dependent variable.

5.3. Modeling results at the regional level.

For peasants, in the first specification of the model, the model with random effects turned out to be the best, however, the regression coefficient at the rye price turned out to be positive and not significant, we do not see the reverse effect of responding to price changes among peasants (see Table 4). After entering the lag value of the endogenous variable into the model, we see a change in the sign of the regression coefficient, However, according to the available data sample, it turned out to be statistically insignificant. And the final model to test our hypothesis with an estimate of the price value through an instrumental variable showed a negative significant coefficient value for the variable of interest.

Table 4.

The results of the evaluation of coefficients of models of dependence of the sown area of rye for peasants in the Kherson province

** the regression coefficient is significant at the level of p<0.01

*** the regression coefficient is significant at the level of p<0.001

Calculated by the authors based on the data in Appendix A1.

For landlords (see Table.5) in all calculated models of the dependence of the sown area of wheat on the price of a given crop, we obtained a positive regression coefficient (statistically significant in two models).

Table 5.

The results of the evaluation of coefficients of models of dependence of the sown area of wheat for landowners in the Kherson province

*** the regression coefficient is significant at the level of p<0.001

Calculated by the authors based on the data in Appendix A1.

Thus, based on fixed-effect models with instrumental variables, we made a final conclusion about the difference in the reaction to rising grain prices between peasant farms and landowner farms at the micro level, using the example of the Kherson province.

4. Conclusion

The aim of the work was to verify the thesis, put forward separately by N. Kondratiev and A. Chayanov in the mid-20s of the XX century, that peasant farms demonstrate non-commercial behavior, namely, they do not seek to increase the volume of bread sold in response to rising prices. Both economists proved this thesis in their works based on the analysis of data from peasant budgets, i.e. at the micro level with a very limited amount of data (observations). In this article, the hypothesis of inertia of peasant farms was tested at two levels: the macro-level and the regional level (the level of a separate province). After analyzing the dynamics of the volume of peasant plowing of bread in response to price dynamics in individual provinces of Russia, the authors found a significant negative relationship. From comparing the results of the study on peasant farms with the results of studies of similar data on private farms, which did not find such a relationship, or found a positive relationship, the authors concluded that a negative relationship can be perceived as confirmation of the specific behavior of peasant farms, that is, the inertia of peasant farms.

Considering that the results of the first stage of the analysis (at the macro level) and the second stage (at the regional level) give a similar result, as well as the fact that non-market behavior was found in the province with the most developed market relations, it can be assumed that the inertia of peasant farms in pre-revolutionary Russia was predominant. In addition, we tried to test the hypothesis of a possible change in peasant behavior after Stolypin's agrarian reform. However, the coefficients for the fictitious period variable and its product with price turned out to be insignificant, that is, in the six years that have passed since the reform, the behavior of peasants regarding the issue under study has not undergone significant changes.

applications

Appendix A1

Data sources for compiling the database: ^[17]

Appendix A2

Table A2.1 – Descriptive statistics of quantitative variables for modeling for the macro model*

Table A2.2 – Descriptive statistics of quantitative variables for modeling for the macro model*

* Note: The results are presented as: N is the number of valid observations; (min; max) is the minimum and maximum values; Mean ± SD is the mean and standard deviation; Me (Q1; Q3) is the median, 1st and 3rd quartiles.