Social mobility in rural Russia, 1995–2003

Abstract

Using panel survey data from three Russian villages, this article examines rural social mobility in post-communist Russia. The article finds that contemporary rural social mobility is different from that which existed during the Soviet period. During the Soviet period, rural social mobility was linked to changes in profession and the direction of mobility was primarily upward. In the post-Soviet period, rural mobility is linked to increasing income differentiation and inequality. In the post-communist period, both upward and downward social mobility have occurred. The article examines the characteristics of upwardly and downwardly mobile households. It then analyses the factors that affect mobility through regression analysis of human capital and behavioural models. The article concludes that household labour continues to have the single greatest causal effect on rural mobility.

Notes

1 Estimates differ as to the size of the Gini coefficient, with another analyst placing it at 0.31 in 1991 and increasing to .47 by the fall of 1999 [Rimashevskaya, 2001: 88]. There is a broad consensus, however, that inequality increased significantly during the 1990s.

2 According to official statistics, the greatest growth in inequality occurred during 1992–95.

3 The Chayanov model postulates that household differentiation is a function of available labour in the household. Increased human capital in turn leads to more intensive work, higher levels of production, and potential food sales, all of which contribute to differentiation [Chayanov, 1966.

4 We would just note in passing that the survey data do indicate both horizontal and vertical mobility. In the sample population, among households where someone did leave, children were the most likely to have left the household during the period 1991–2003, and they were most likely to go to the nearest big city, indicating vertical mobility. The second most frequent destination was a different, nearby, village, suggesting horizontal mobility.

5 Book length studies include O'Brien and Wegren 2002, and Wegren 2004; 2005.

6 Each broad category was further subdivided by branch of the economy, containing individual professions. There was also a separate category for pensioners.

7 The increase was not due solely to an increasing population. During the indicated time period the total Soviet population increased by about 42 million [Naseleniye Rossii za 100 let (1897–1997), 1998: 32–3].

8 The Gini coefficient increased to 0.30 and the 90/10 ratio increased to 3.75 if total wages (public and private sources) are used.

9 The Gini coefficient declined from 0.27 in 1973 to 0.24 in 1979, and the 90/10 ratio increased from 3.11 to 3.33 during the same period.

10 During the Soviet period, household land plots, were often referred to as ‘private plots’ because their production was not regulated by the state (although the land itself was not privately owned). These land plots differed in size by republic but seldom exceeded 0.50 hectares. Production from private plots and the contribution to the regional food supply varied greatly by region of the country, depending upon the level of infrastructure and access to machinery and equipment. For an analysis of regional differentiation and the factors that affected it, see Kalugina [1991: ch. 4]. In the late 1980s, in the Russian Republic, the most common size of land plot ranged from 0.31–0.40 hectares (used by a plurality of families). The average monthly income from the sale of household production in the Russian Republic was 71 rubles, although there was considerable variance by size of family, with a one-person family averaging 45 rubles a month and a five-person family averaging 81 rubles a month [Lichnoe podsobnoe, 1989: 68–9].

11 The term ‘shock therapy’ is used in the broadest possible context to include price liberalization, privatization, the introduction of market forces into the economy and economic calculations, and the reduction of state control over the economy. We acknowledge that some elements of shock therapy were not fully achieved or were achieved unevenly. Space constraints do not permit a detailed discussion of what went awry and why, but these are explained in Aslund 1995: ch. 3], Silverman and Yanowitch 1997: 3–14], Gaidar 1999 and Goldman 2003: ch. 5].

12 This section draws from Wegren 2005: 157–9].

13 The reference to ‘lowest two income categories’ refers to households with monetary incomes 0–49 and 50–75 per cent of the minimum subsistence level as established by the Russian government.

14 This is shown by results gleaned from a different survey conducted in 2001. From those data, it was found that household land plots were positively correlated with food production and non-monetary household income, but negatively with food sales. Rented land, however, was positively correlated with food sales, reflecting the fact that land has different uses. This is shown for example by the fact that the size of a household land plot was negatively correlated with a rental plot.

15 Less than 2 per cent considered themselves ‘above middle.’ Thus, in 1995, 96 per cent evaluated their economic condition as either poor or average. It is interesting to note that despite improvements in standards of living, possession of durable goods, and levels of income (all of which are quantifiable), evaluations of economic condition did not change much by 2003. In 2003, 29 per cent of respondents considered themselves ‘poor’ and 64 per cent said they were ‘average,’ thereby accounting for 93 per cent of responses.

16 The government's threshold for poverty is based upon monetary income and was drawn from a number of sources [Rossiya v tsifrakh, 1997: 65; and Rossiya v tsifrakh, 2004: 109]. The government's measure is adjusted for inflation and allows comparison from one time period to another.

17 For a discussion of measuring poverty in rural Russia and differences between government estimates and our own, see O'Brien et al. 2004: 461–4].

18 The total income scale and quartile divisions are available upon request.

19 We did not explore this systematically, but it is likely that household within quartiles share similar socioeconomic and demographic characteristics.

20 These categories are for our analytical purposes, and we recognize that the two categories overlap and that behaviour may affect human capital. For example, if a household member decides to move away this is a behavioural aspect that affects size of the household and household labour.

21 The calculation of household labour is based on a formula used by Chayanov 1966. A weighted labour potential was used based on the age of the members of the household: 0 for persons aged less than 8 years of age or more than 80; 0.25 for persons aged 8–11 and 75–79; 0.50 for persons aged 12–14 and 71–74; 0.75 for persons aged 15–16 and 66–70; and 1.0 for persons aged 17–65. These figures were then summed for each household. The result is a measure of the amount of household labour in each household.

22 From the survey data redundant and non-redundant ties were calculated. Redundant ties are defined as the sum of the number persons the respondent mentioned as providing help in one or more of six types of assistance to the household. The six categories of assistance include: borrowing money, trading goods and services, taking care of the household if someone is sick, assisting with operation of the household plot, assisting with household tasks, and discussing important matters [O'Brien et al., 2000: 73]. For redundant ties, a single individual was counted more than one time if they help with housework, talking about problems, etc. The sum of all of these helping ties then is called redundant since some people are counted more than once. Non-redundant ties refer to the total number of individuals who are mentioned, but each individual is only counted once even if they help in more than one type of assistance. Thus, ‘non-redundant ties’ is a more accurate measure of the size of a household's social network.

23 ‘Assistance’ includes a range of behaviours, from ‘providing moral support’, ‘providing labour assistance’, ‘providing material assistance’, or ‘lending money.’

24 It should be noted that assistance from large farms was not particularly significant for either type of household. For households that experienced downward mobility, 14 per cent reported ‘very little’ assistance and 75 per cent reported ‘no’ assistance from a large farm. For households that experienced upward mobility, 40 per cent reported ‘very little’ assistance and 49 per cent reported ‘no’ assistance from a large farm.

25 This general line of argumentation was borne out by statistical analysis of the relationship between the size of networks and household food production. It was found that the size of social networks has a positive and statistically significant effect on household food production [O'Brien et al., 2000: 143–8].

26 Non-redundant ties may also be considered social capital, but since this article does not employ a social capital model, it was decided to include this variable in the human capital model.

27 Our presentation does not include change in the size of the household land plot and change in the size of the rental land plot. After the model was run, we found that these two variables are not statistically significant and are signed in the wrong direction. We decided that variations in the distribution of land among the villages was biasing the statistical results. For example, following the 1991 decree that required large farms to allocate 10 per cent of their land to local governments for distribution, one village did so (Vengerovka). As a result, the local government was able to distribute land directly to individuals and did so early in the 1990s in a timely manner. In another village, Latonovo, residents were not able to obtain land plots for purchase or rental until 1999. This temporal difference in the distribution of land in turn affected household production and income, thereby skewing the statistical results.

28 It is interesting to note in passing that other aspects of what may be considered human capital were also considered, namely, the extent to which the household received assistance from relatives or family. In the regression model of human capital this assistance variable turned out to be statistically insignificant and therefore was discarded.

29 Furthermore, the ‘mechanization’ model as a whole had a weak effect on household income, reflected by an adjusted R-square of 0.03, although the model was statistically significant above the 99 per cent level of confidence. When the variables in this mechanization model are added to the behavioural model (see text below), the variables in the mechanization model have no effect on adjusted R-square, on the significance of the behavioural model, and change only slightly the explanatory power of the variables in the behavioural model. In short, the effects of changes in mechanization are minimal, reflecting the fact that most labour continues to be manual.

30 Previous statistical analysis found a positive correlation between the size of a household plot and food production, but a negative correlation between the size of a household plot and income from food sales. For rental plots, the relationship was the opposite: a positive correlation between income from food sales. The point is that different types of land plots have different uses and contribute to household income in different ways.