Serfdom, Emancipation, and Economic Development in Tsarist Russia Steven Nafziger1
Department of Economics, Williams College
Paper prepared for
“Quantifying Long Run Economic Development,” Venice, Italy, March 22-24, 2011
Economic History Society’s Annual Conference, Cambridge, England, April 1-3, 2011
Extremely preliminary and incomplete – all comments welcome but please do not cite.
Serfdom has often been cited as a significant institutional constraint on Russian economic development in the late Tsarist period. Moreover, Alexander Gerschenkron and other scholars have long viewed emancipation and other peasant reforms of the 1860s as additional brakes on Russia’s economic growth. But this literature has seldom considered what the available empirical information actually says regarding these issues. This paper explores several types of empirical and legal sources to outline exactly what was meant by “serfdom,” how emancipation and land reforms changed factor endowments and the economic rights of the peasantry, and whether these changes resulted in different patterns of development across European Russia. These exercises are initial steps towards a causal evaluation of what serfdom and emancipation meant for Russia’s economic development in the century prior to 1917.
“The year 1861 begot the year 1905.”2 1 Introduction
Institutions matter for economic development. Good ones promote savings, investment, trust, and a variety of market transactions. Bad ones generate rent seeking, undermine incentives, and limit market development. Identifying precisely how a given institution does or does not help generate economic growth has become a key theme in much recent economic history. But history offers relatively few experiments for evaluating the consequences of different institutional arrangements. This paper explores the development implications of one type of institution and its evolution over time – serfdom and serf emancipation in Tsarist Russia.
One important strand of the recent literature on such “natural experiments” has focused on the development implications of slavery and slave emancipation. As an institutional arrangement, slavery varied widely throughout history. At its most basic level, slavery, or the related institution of serfdom, entailed the ownership of a laborer’s product by someone else. In practice, such regimes were often complemented by limitations on legal, political, and economic rights. In agricultural societies, much of the variation in coercive labor regimes was very local, with differences evident from estate to estate. In the case of Russia, only some of the peasants in were formally the property of private landowners prior to 1861, and serfdom itself took a wide variety of forms across space and over time. There were two dominant variants that emerged by the early 19th century: quit-rent (obrok) estates where peasants were liable for yearly payments but were granted substantial freedom to engage in a variety of economic activities, and labor service (barshchina) estates where serfs were responsible for providing labor on the owner’s land in return for access to some land of their own. Most historians argue that the both types of serfdom imposed costly restrictions on Russian economic development. In practice, this meant limitations on the mobility of labor, laws against certain types of peasant economic activity, restrictions on property ownership, and the deadweight losses implied by feudal rents. These have all been viewed as contributing factors towards the slow pace of agricultural growth and industrial expansion in Russia, which, to many contemporaries, was made especially salient with the defeat of Russia in the Crimean War (1853-1856).3 Prior to emancipation in 1861, Russia was increasingly falling behind the Western powers, and the archaic set of rules and relationships that characterized serfdom were often blamed.
As famously outlined by Engerman and Sokoloff (2002), slavery and the distribution of factor endowments upon emancipation mattered for institutional arrangements and inequality evolved over the long run in the Americas. In a similar fashion, the emancipation of the serfs occurred in a unique way that created substantial post-1861 variation in factor endowments and institutions. Former serfs were granted land but generally only as members of communal villages, and they received much worse deals than non-seigniorial peasants did. Therefore, in considering the long-run implications of serfdom as an institution, it is necessary to consider variation not only in the coercive labor system itself, but also in how emancipation differed across space, among different groups of peasants, and even between individual estates. The current essay begins a discussion in these directions.
Works by Nathan Nunn (2008a and 2008b), Melinda Miller (2009), and others have employed plausibly exogenous differences in the extent of slavery (or in the conditions of emancipation) to study the long-run implications for economic development.4 In the case of Nunn’s work, the greater penetration of slavery into an African society during the centuries of the slave trade is highly correlated with lower per capita income levels today. In Miller’s research, the greater property rights granted to former Cherokee slaves relative to otherwise comparable ex-slaves at the end of the American Civil War were associated with better outcomes in the decades that followed. In a similar vein to these works, the current study turns to geographic variation on the extent and nature of serfdom, and differences in ways that emancipation occurred, in order to investigate the long-run implications of the institution and its demise for economic development.5 However, unlike these other studies, we do not yet have a plausibly exogenous source of variation in serfdom or emancipation
Previous studies of Russian serfdom and emancipation have tended to focus only on legal changes and have considered very limited empirical evidence. Most have also ignored important complications to the story that provide possible ways to identify of the long-run economic effects of these institutional arrangements. In this paper (and future work), I aim to rectify these problems in the literature in three ways. In the first section, I quickly summarize the literature on serfdom and the emancipation reforms and present newly collected district-level data on serfdom just prior to emancipation. These allow me to characterize the variation in this institution across European Russia in a finer way than previous studies. Such data also enable an initial inquiry into whether Evsey Domar’s (1970) hypothesis about the determinants of serfdom – particularly the relatively high land-labor ratio of rural Russia – was borne out in practice.
In the second section of the paper, I explore several channels through which the emancipation and accompanying land reforms may have generated long-run differences between formerly serf and non-serf villages. The reforms of the 1860s affected all members of the peasant soslovie, or social class, and not just those who were obligated to the nobility as serfs. According to the 10th tax census of 1857-58, approximately 38-42% of peasant male souls (dushi – the primary tax unit) were serfs, while around 53% resided on state land and were administered by the Ministry of State Domains (Kabuzan, 1971, p. 176).6 In contrast to the former serfs, the 1866 reform of the state peasants endowed them with essentially the same property rights as they held before 1861 at a lower cost. Although they also resided in communal villages, several studies have argued that even before 1861, state peasant households had more secure individual property rights and were able to allocate their labor with less outside interference (Crisp, 1959; Deal, 1978; and Ivanov, 1945). Thus, the experience of these peasants offers a potential baseline upon which the long-run economic development of the former serf villages may be gauged.
After briefly outlining the structure of emancipation and later land reforms, I explore how the resulting variation in endowments and the institutional backgrounds of a sample of former serf and state peasant villages in Moscow province translated into longer-term differences in factor usage and labor market outcomes. I observe these villages twice: once in 1876-77 and again in 1899. The period of time covered by the data prevents an explicit test of how the land settlements affected the labor decisions of village members, but I focus on whether the level of involvement off-the-farm activity varied systematically with factor endowments, obligation levels, or institutional differences.7 Cross-tabulations show that former serf communities faced persistently higher tax and land obligations, worked less land, and had less productive agriculture than did former state peasant villages. However, the two groups showed only small differences in the prevalence of non-agricultural activity – a surprising findings given the large endowment differences. I then analyze the determinants of changes in off-farm work between 1876 and 1899. The results indicate that members of former serf communities were less inclined to shift away from working in the village in response to high obligation levels, improved market access, or population growth. Former serfs were more slightly more likely to respond to these factors by moving into local non-farm employment. These findings are is consistent with some persistent differences between the two types of villages that had faced divergence institutional regimes prior to the 1860s.
In the final section of the paper, which is planned for future work, I step back from the village-level analysis and consider district-level information on a number of socio-economic outcomes in the post-1861 period. I intend to explore the correlations between the extent or nature of serfdom prior to 1861 and five such outcomes: urbanization rates, literacy levels, land ownership, occupational structure, and different wage measures and income proxies. In particular, I will examine whether areas characterized as more “serf” in the late 1850s (in terms of a share of the population), areas that had relatively more serfs on large estates, or areas that had relatively more serfs under quit-rent rather than labor service obligations, showed persistent differences in outcomes over ensuing decades.
Although these are important first steps in evaluating the economic consequences of serfdom and emancipation, the three parts of this study are very preliminary in nature. Understanding where the variation in serfdom came from is critical for ascribing any causal interpretation to variation in the institution. Hopefully additional geographic and historical information will address these concerns, but for now, the village and district-level analyses should be cautiously interpreted as evidence of any causal effects of serfdom and emancipation. The village-level analysis only considers one province (Moscow), and the implications of serfdom may have been quite different in other locations. To properly identify the ways that serfdom and emancipation might have mattered in the long run in a district-level analysis, it is necessary to employ appropriate empirical methods (such as instrumental variables) that derive from an explicit model of just how variation in the institution arose, how the reforms affected or solidified the existing regime, and how these changes might have impacted economic development in the decades that followed. These issues will all be the subjects of future research.
2 Serfdom: The Basics + A Preliminary Empirical Analysis
Russian serfdom apparently emerged in the 16th and 17th centuries as a solution to the problem of scarce labor and widely available land. In return for service – military or otherwise – the Tsars granted land to favored individuals. Making these grants productive required a labor force, but the mobility of the peasantry initially made it difficult to guarantee production. Thus, a series of decrees slowly circumscribed the mobility of peasants resident on such holdings while also making other aspects of their lives increasingly subject to oversight by the class of servitors. The legal code of 1649 (the Ulozhenie) formally tied the peasants to the land, and a series of Tsars passed measures transforming the servitor population into a full-fledged noble estate with rights and privileges that extended over their peasant population. This formalized serfdom as the institution that would become synonymous with rural Russian society by the 19th century.8 The rights of serf-owners included various forms of seigniorial rents and obligations, along with control over all aspects of their peasants’ lives. Extractions either took the form of simple cash or in-kind payments, or were demanded as labor on the demesne. In the northern and central provinces around Moscow, poor soil and climate conditions led many peasants to turn to non-agricultural occupations to generate income and pay their tax and seigniorial obligations. Estate lands were often granted in their entirety to the serfs to do with as they wished, as long as they paid their quit rents to their noble landlords. Many serfs and state peasants in the region engaged in small-scale trading or the exploitation of natural resources, often under “hire” by local estate owners (Bohac, 1989; and Dennison, 2011). Serf-owners frequently allowed their peasants to travel for urban or factory-based employment (ibid.; Melton, 1987; Rudolph, 1985; and Tugan-Baranovsky, 1907 , Part 1).9 The Moscow region became widely known for textile production, with numerous serf and state peasant villages foregoing agriculture to concentrate on spinning and weaving (Vodarksii, 1972). In contrast, the provinces to the south and west, where soil and weather conditions favored grain cultivation, were characterized by the presence of substantial estate demesnes. In this region, serf obligations frequently included labor services for a particular number of days or for certain tasks. In such areas, most notably described by Steven Hoch in a series of works, the landowner was much more directly involved in managing the economic activities of his or her serfs.10 These were the estates that may have embodied Domar’s (1970) model of serfdom as a solution to the (agricultural) labor scarcity problem.
The strengthening of serfdom in the 18th and 19th centuries was matched by the emergence of the peasant commune as an important institution in rural Russian society.11 On serf estates, the commune organized agricultural production and managed land granted by the serf-owner. Such coordination occurred under the open-field system of mixed grain and livestock farming that prevailed in 19th-century Russia (Moon, 1999, pp. 122-126; and Pavlovsky, 1968). At the same time, local administrative and judicial tasks were often informally devolved to the commune and managed by elected communal elders or the assembly of household heads (the skhod). Communal control was also necessitated by substantial landlord absenteeism and the immense size of some estates (Blum, 1977). Authors such as Hoch (1986) have emphasized that even on estates with more direct management, there was a close relationship between communal and seigniorial authorities. The commune came to be responsible for the fulfillment of seigniorial duties and other state obligations that were collectively imposed on the households of the community. This collective responsibility, or krugovaia poruka, for external obligations was to be a defining feature of rural Russia well after the 1860s.12 By definition, serfdom entailed constraints on the mobility and the economic decision-making of peasants. Soviet scholars often bent over backwards trying to prove that serfdom was in crisis before 1861 (e.g. Koval’chenko, 1967), but more careful studies employing simple microeconomic theory and some limited empirical sources have found that it remained profitable until the very end (Domar and Machina, 1984). The evident mobility of peasants under serfdom suggest that this institutional structure did not prevent engagement in many forms of commerce, artisan work, or industrial labor. Moreover, evidence presented by Tracy Dennison (2011) suggests that in the absence of much state interference, large estate owners found it profitable to provide legal structures, limited public goods, and a relatively light hand when it came to internal governance. Relative to areas or villages without such institutional structures, such estates may have actually seen environments more conducive to market development and investment.
At the same time, serfs did not comprise the majority of peasants in European Russia by the middle of the 19th century. Several factors contributed to the growth in the number of peasants residing on land formally controlled by the state or the Romanov family: the tradition of granting land in return for service eased over the 18th century; in the 1760s, the state took over all lands owned by the Orthodox Church; and continued expansion of the state to the south and southwest opened up new areas for migration. The resulting “state peasantry” (gosudarstvennye krest’iane) was a diverse group that included former residents of monastic estates, state-sponsored settlers in under-populated areas, recent transfers from freed serfs and military conscripts, and other residents on state-owned land throughout the empire. In provinces such as Moscow, these communities existed alongside serf villages and managed land and obligations communally in much the same manner as the seigniorial peasants did on their allotments. However, many authors argue that in comparison to the former serfs, state peasants could more easily engage in contracts and choose occupations, had stronger traditions of individualized property rights, and faced lower levels of monetary and administrative obligations (Crisp, 1976; Deal, 1981; and Ivanov, 1945).13 2.1 Quantitative Evidence on Serfdom in the 1850s
Surprisingly, exactly where serfdom existed and what form it took are issues that are not well understood empirically. Soviet scholars were very interested in quantification, but their evidence was often limited to a few estates or a small geographic area. Depictions in the western historiography tend to be limited to broad provincial or regional comparisons (e.g. Moon, 2001). Here, I provide a brief snapshot of what serfdom looked like in the late 1850s, drawing on newly collected district-level data.
According to the data collected in the 10th tax revision of 1857-1858 (Table 1), the districts of European Russia (in 50 provinces minus Poland and Finland) contained approximately 24 million serfs out of a population of 60 million (40.8 percent – Table 2). Roughly 90 percent of districts (460 out of approximately 500) contained some serfs, although their share of the total population exceeded 80 percent in only 19 and many districts had a negligible number (Figure 1a). The serf share of the total population was somewhat lower than earlier in the 19th century, as manumissions, transfers of indebted estates to state ownership, and differential population growth (by region) all served to reduce the serf population relative to other groups of state peasants (Hoch and Augustine, 1979). The geographic distribution of serfs was concentrated in a horizontal S-shaped region between roughly Kiev and the upper Volga (Figure 1b).
According to information collected by provincial committees of the nobility in 1858-59, the share of serfs (male souls) obligated for quit-rents alone was 25.7 percent (Table 2). The other 74.3 percent were required to perform some form of labor services, which may have included working in estate-owned enterprises or several forms of agricultural work on the demesne, with or without the provision of draft power. In a separate category, the share of serfs engaged in domestic work on the estate – a group known as “household serfs,” in contrast to “field serfs” – comprised around 8.4 percent of the serf population. This category of serfs received no access to land in the emancipation reforms. Thus, over 50 percent of serfs remained obligated for some amount of labor directly on the demesne or in enterprises owned by the serf-owner. As Table 2 indicates, those serfs that paid quit-rent in some form were liable, on average and as reported by the noble committees, for over 25 rubles per tiagla, or labor unit (often a husband and wife with horse or ox).
Prior to emancipation, data on large serf estates (greater than 100 serfs, although the data do include some smaller estates) were collected for the Editorial Commission writing the reform. The same provincial committees of nobles as the above apparently supplied these data, and the same issues of self-reporting (such as underreporting land holdings) may be present. Despite this concern, these data suggest that roughly 60% of serfs resided on larger estates. The share of serfs that were on quit-rent only, or worked in the owner’s household, were slightly less than across all serfs, which is consistent with these estates perhaps involving more directly managed production under the oversight of the owner or estate administrators.
Finally, the data reported to the Editorial Commission also included information on the amount of land allotted to serfs out of estates, as well as comparative information on holdings for peasants obligated to the state or directly to the Tsar. Although 460 districts had serfdom, serfs received substantive allotments in only 389 of the 426 that provided these data. In those districts, the serf estates had approximately 11.8 desiatina per soul (and the other districts had considerably larger estates). Thus, serf allotments were slightly less than 1/3 of estate land preior to emancipation. Moreover, the average serf allotment – roughly 3.3 desiatina per male soul – were significantly less than the amount of land available to either the court or state peasants, although a large number of the latter group resided in more sparsely settled northern and eastern provinces.
2.2 Examining the Variation in Serfdom
What explains the variation in where serfdom was located and the form it took? The institution emerged during the expansion of the Muscovy state in the 16th and 17th centuries. In return for (primarily) military service and continued obedience, the Tsars allowed servicemen conditional grants of land (pomest’ia) or supported the hereditary tenure of existing nobility. But to make these holdings worth something, the state responded to the demands of its servitor class by slowly constraining the mobility of the peasants on the land, eventually leading to the Ulozhenie of 1649. This degree led to the end of any formal remnants of slavery in Russia and formalized the dependence of peasants on the owners of the land they resided upon (Hellie, 1971). The most prominent 19th century Russian historian, Vladimir Kliuchevskii, further argued that the indebtedness of the peasantry caused many of them to voluntarily enter into serfdom (cited in Domar, 1970). By the 18th century, the “second serfdom” was a defining feature of the Russian Empire.
In reflecting upon the Russian experience, Domar (1970) famously treated serfdom as the product of a high land-labor ratio and the state’s willingness to impose mobility restrictions on the peasant labor force as a way of supporting a land-based elite. Following earlier historians who had made similar arguments, Domar emphasized that in the context of a very elastic supply of land, the Russian state’s policies allowed the emerging nobility to extract the shadow (labor) scarcity rents from production on their land they owned. A similar logic forms the basis of more complicated dynamic models of slavery in land abundant societies, such as in Lagerlof (2009) or Fenske (2010).
Domar has little to say regarding the emergence of quit-rent obligations. His model focuses on the role of scarce labor in a land-abundant society. But once serfdom was consolidated as formal institutional arrangement, the tying of peasants to elite-owned land became equivalent to noble ownership of peasants themselves. This allowed those elites with relatively unproductive land to extract rents from the product of their serfs’ labor in other activities, leaving (generally) a subsistence income to the peasants. While this simple story seems to be supported by the geography of obrok and barshchina (Figure 2), it has never been formally tested.
Relying on population data from the 10th tax census (revisii) of 1857-8, Figure 3 plots the district-level relationship between population density and the portion of the population who were serfs. It shows a positive relationship, which is primarily the result of peripheral districts that had very low population densities and little serfdom. This picture is not obviously consistent with Domar’s basic point about the institution taking hold in land-abundant areas. But it seems like those districts with more productive land and, hence, higher population densities, were those where relatively more land had been granted to servitors in an earlier time period. When thinking about the causal effects of serfdom, this figure suggests that the institution was not plunked down exogenously across European Russia. Better land was scooped up by the noble class or allocated out in return for service.
But again, it is also important to keep in mind that many serfs were not engaged in agricultural production as a primary occupation by the middle of the 19th century. By 1858, population density was unlikely to have been only a function of land productivity in agriculture. Figure 4 plots the relationship between the share of serfs exclusively on quit-rents and the overall population share of serfs. While the (regression) relationship is only slightly upward sloping, there is little evidence that serfdom was only a matter of tying labor to agricultural production. Serf-owners of the mid-19th century were perfectly willing to let their serfs work in non-agricultural trades as long as their obligations were fulfilled.14 Table 3 presents some preliminary regression results that test and extend the findings in Figures 3 and 4.15 In the first three columns, the dependent variable is the share of the population that was serf in 1858; in the last four columns, it is the share of the serf population that were on obrok only. To directly evaluate the central component of Domar’s hypothesis, I include the population density in 1858. Although not significant in either basic regression specification, the coefficient is negative and strongly statistically significant in the 2nd and 3rd models with province fixed effects. Thus, once broader and fixed geographic differences are taken into account, Domar’s basic insight about the incentives for holding serfs in more land abundant areas appears to hold, at least slightly.
The particulars of serfdom evolved as the outcome of a political process in Russia, and so the role of labor scarcity, while important to its overall consolidation, may not have a large role in explaining exactly where serfdom existed or what form it took. Supporting the notion that serfdom in the Russian case was an outcome of Muscovite policy as the state expanded, columns 1 and 2 show that relatively less of the population were serfs the further one went from Moscow (even within provinces). Land was more likely to be granted to the nobility the closer it was to Moscow, with more peripheral areas being relatively undesirable or only incorporated into the Empire late in the 18th century (and relatively sparsely populated before that).16 Such a distance effect does not persist for the quit-rent share models when province fixed effects are added, suggesting that this decision had more to do with local economic conditions. To evaluate the role of geography in more depth, models 3, 6, and 7 substitute the latitude and longitude for the distance to Moscow variable. For model 3, the size and significance on the latitude variable again suggests that serfdom was more prominent in the regions closer to the old Muscovite state of the north, and less prominent in the newly settled, more agriculturally productive areas to the south. This is also supported by the similarly sized and slightly significant result for the quit rent share in model 7. As you moved west to east in a line, there were few major differences in the prevalence or nature of serfdom, especially within provinces, and as Figure 1b might suggest.
Unfortunately, other district-level information for the pre-1861 era is very scarce. Some data are available on the conditions of cities (and urbanization), the exact timing of incorporation into the Russian Empire, and land characteristics, and these will be explored in future versions of the models in Table 3. After 1861, considerably more data are available, but using them to explain (causally) the variation in serfdom is problematic for a number of reasons, including the basic timing problem But there may be interesting possibilities in this direction, as suggested by the result on the mean oat yields variable included in model 7. While this variable reflects information from the 1880s and 1890s, it likely does proxy for overall land productivity, as investments that would have dramatically improved soil quality or new agricultural practices were likely limited, especially when comparing across districts within a province. That this variable was negatively related to the share of serfs on quit-rent is consistent with the logic that serf owners were trying to take advantage of relatively better labor market opportunities off the estate. In future work, it may be possible to identify additional post-1861 variables available at the district level that would help in identifying just what drove variation in serfdom in European Russia.
3 Emancipation, Redemption, and Long-Run Labor Market Adjustments
Alexander II’s manifesto of February 19, 1861 initiated emancipation of the serfs and began a sequence of complimentary rural reforms. These measures not only granted new legal freedoms to the rural population, but they also transferred formal land rights to the peasantry in a mortgage-like process referred to as “redemption.”17 In comparison to other cases of rural reform in 19th-century Europe or to slave emancipation in much of the Americas, Russian peasants received substantial land rights, albeit in the form of communal allotments with associated collective liabilities. Rather than simply expropriating the peasants or the landlords, the state constructed the reforms as a series of steps that slowly transferred land rights to the peasants while compensating the nobility for their losses. Peasants repaid the state’s financing in a process that was only projected to conclude in the 1910s. Moreover, there was substantial heterogeneity built into this process: the price and amount of land allowed to each community was locally differentiated; leeway was left to bargaining between peasants and landowners; and different reforms were initiated for very small estates, for peasants that resided on state or Tsar-owned land, and for serfs employed as domestics. Indeed, the fact that only a minority of peasants was formally serfs in the sense of being tied to privately owned property creates an important source of heterogeneity in how the institution of slavery may have affected later development. Overall, these reforms led to hopes that the Russian economy would begin to modernize and catch up with the industrializing nations of Western Europe. Almost immediately, however, contemporaries identified a growing economic crisis in the countryside and attributed this to particular features of the emancipation reforms (Doklad, 1873).
Indeed, the unique way that peasant reform occurred in Tsarist Russia generated a long-running debate over its implications for economic development. According to Alexander Gerschenkron (1965), the collective liability of households in the commune effectively tied labor to the land and restricted the flow of resources into industry. This institutional framework had antecedents under serfdom, but the strengthening of the commune in the reforms continued these restrictions into the 20th century. Soviet scholars emphasized that the reforms fixed land endowments too low and set their “price” too high, which led directly to growing poverty and other economic difficulties in the countryside (e.g. Khromov, 1967). More recently, some scholars have begun to question whether the reforms had much of an impact at all. Hoch (2004), Gatrell (1994), Mironov (1999), and others argue that the reforms had minimal effect, either because the institutional constraints of the commune were not enforced, or because the land settlements did not significantly change the amount or the “price” of peasant land-holdings. According to this interpretation, industrial development after 1861 built upon pre-Emancipation conditions in the form of seasonal migration, relationships between factory and proto-industrial production, and the rural location of manufacturing establishments. Given these different stories, it is remarkable that relatively little empirical work has been done on precisely what serfdom and emancipation meant to long-run economic development in Russia.
3.1 A Brief Account of Serf Emancipation and Redemption
The Emancipation Statutes – the Main statutes and subsequent legislation – called for the formulation of ustavnye gramoty, or regulatory charters, between the former serf communities and their previous landlords. These documents described the population of the community involved, the seigniorial obligations, and precisely which and how much land the peasants had previously worked or had access to. These charters were to be completed by 1863, with hundreds of newly named mirovye posredniki (peace mediators) aiding in their writing and ratification. Based on rules laid out in Local Statutes, the charters translated the previous rights and obligations of the serfs into a new collective land endowment and set of labor duties or cash payments. The number of obligated souls (dushi – a tax unit roughly equivalent to one working-age male) was set on the basis of the tax census of 1857-58.18 If the amount of land per soul (a soul “allotment,” or nadel) exceeded a local maximum norm defined in the Local Statute, the excess could be “cut-off” and retained by the landlord. If soul allotments fell below one-third of this maximum norm, land was to be added to the new endowment.19 At a minimum, landlords also had the right to keep at least one-third of their arable land, and until 1870 they could reduce peasant allotments to one-third of the maximum norm at will. These rules pertained only to the arable land on the serf estate. Distinct conditions held for garden plots within the village and for other types of land. Significantly, the former serf-owner also retained all rights to forests and meadows. These were vital inputs into livestock production. Gaining access to these resources after the charters were drawn up often required additional communal payments or labor service on the landlord’s fields (e.g. Hourwich, 1892). Even in the case of arable land, the landlords had the right to pick and choose the land they retained as long as the amount available to the peasants followed the statutes.
By the end of 1864, almost all of the regulatory charters were signed by both parties and communities entered into “temporary obligations.” During this stage, the households assigned to a commune were collectively liable for the revised cash or labor obligations outlined by the charters. The local maximum allotment norm corresponded to either an amount of labor (in days per year) or a fixed payment.20 Households assigned to the commune could only exit with the unanimous approval of the rest of the members. Those wishing to leave had to give up all rights to a share of the commune’s land, and the commune had to agree to take up their outstanding debts and shares of obligations.
Temporary obligations were intended to last until the financial arrangements were made to legally transfer the land to the peasant commune. This involved the formulation of a vykupnaia sdelka, or a redemption deal, where the state agreed to finance the purchase of the land from the former serf-owner. These deals, which resembled mortgage transactions, documented the boundaries and value of the land to be formally transferred. The yearly payments (or the monetary equivalent of any labor services) to former serf-owners under temporary obligations were capitalized at a 6% interest rate to establish the aggregate redemption value of each land allotment to be transferred.21 According to the Redemption Statutes, deals could be initiated through mutual agreement between the community and the former serf-owner (requiring a 2/3 vote in the communal assembly) or, more commonly, at the demand of the former seignior.22 The process of formulating these redemption deals was drawn out. A substantial number of communities were still engaged in temporary obligations in 1880.23 Regardless of whether deals were mutually agreed upon or not, the newly reconstituted State Bank financed the purchase of land through 49-year loans made to the communes. The conditions of these loans depended on whether or not the deal was mutual, and on the amount of existing debt owed by the estate (Gerschenkron, 1965; and Zaionchkovskii, 1960).24 A key feature of the redemption program was that the commune was collectively responsible for making payments on the outstanding redemption debt. To enforce household contributions under this joint liability, the communal assembly was granted legal authority over the immovable property and labor allocation decisions of those in arrears (Burds, 1998; and Gerschenkron, 1965). The statutes stated that renewals of passports for work outside the village were only possible if arrears were paid off. If a commune failed to make one of the twice-yearly redemption payments, local officials could sell non-essential assets or punish communal officers. After the community entered redemption, households could only legally alienate their share of communal land by paying off their portion of the loan in its entirety. These legal restrictions lasted into the 20th century and have led many historians – most prominently Alexander Gerschenkron – to see the emancipation and redemption reforms as re-imposing many of the same constraints on mobility as existed under serfdom, thereby generating a “considerable obstacle” for economic growth that lasted until the Stolypin reforms of the 1900s (1962, p. 121).
The central component of Gerschenkron’s argument is that assigning households to communes and to collective responsibility for taxes, land payments, and other obligations reduced the mobility of labor out of the relatively backward agricultural sector.25 Since communal members were jointly liable for external obligations, it was in the interest of those with shares of the aggregate burden to limit household exits. The reforms allocated the commune control over the issuance of documents to work outside the village, the property of households in arrears, and the allocation of land itself. Communal authorities could call on police or local officials to return wayward migrants or punish the members that failed to contribute their share (see Nafziger, 2010). According to Gerschenkron (1965), since households were forced to retain ties to the commune and its property endowment, this effectively tied labor to the land. This lowered the supply of labor into industry, forced manufacturing to be overly capital-intensive, and slowed industrial growth.
In contrast to Gerschenkron’s emphasis on the negative implications of collective property rights and obligations, Soviet and other scholars have emphasized that the first-order effects of the reforms for later development stemmed from substantially worsened peasant endowments. As a result of the reforms, land “cut-offs” (otrezki) from peasant holdings often reached significant levels. The former serfs of Simbirsk province lost over 30% of the land they previously utilized (Kanatov, 1964). In Moscow province, they lost 14.2% of their land (Zaionchkovskii, 1958, p. 182). This change in endowments often forced the former serfs into rental contracts, whereby communes and households rented land from their former landlords for cash payments (or labor services) that typically exceeded the agricultural value of the property (Anfimov, 1980; and Filippova, 1959, p. 378). Moreover, Soviet scholars long understood that the loss in allotment land was matched by a rise in its “price” under temporary obligation or redemption. Considering data from 9 of the 13 districts of Moscow province, Litvak found that average payments per soul actually decreased from 9.36 to 8.44 rubles. However, the average per desiatina increased by 8.3%.26 Many scholars – both Soviet and Western – have also emphasized that serf obligation levels were set relatively high to begin with. Before emancipation, seigniorial obligations were not exclusively based on the productivity of the land but were extracted from the total income of serf labor in both agricultural and non-agricultural activities. As a result, post-1861 payments in provinces such as Moscow, where seasonal non-agricultural employment was prevalent before Emancipation, greatly exceeded the agricultural (rental) value of the land (Hourwich, 1891; and Ianson, 1881).27 Furthermore, Soviet scholars specifically argued that these cash demands caused previously autarkic agricultural households to look off the farm for income to pay their obligations.28 However, recent research by Sergei Kashchenko and his colleagues on regulatory charters from the northwestern provinces of Novgorod, Pskov, and Petersburg has somewhat refined the earlier Soviet findings (Degtiarev et al., 1989; and Kashchenko, 1996 and 2002). They document the entire distributions of land and obligations (for formerly quit-rent estates), both before and after the formulation of the land charters. Their research establishes that the two distributions narrowed around the norms proscribed in the Emancipation Statutes, with little change in the median size of land holdings. Hoch (2004) considers this research and asserts that the equalization of allotments benefited the rural economy by increasing the size and security of endowments for households previously undersupplied with land. He also argues that the prices peasants paid for these allotments were actually quite close to the fair market prices for the land. Hoch goes on to conclude that the settlements were not as exploitative as Soviet scholars thought, nor did they introduce new prohibitive institutional restrictions on labor mobility and economic development as Gerschenkron argued (ibid., p. 274).29 However, this recent scholarship focuses less on the final redemption deals than on the intermediate step of the emancipation charters and temporary obligations. Moreover, it does not really confront the possibility that the regime of collective property rights and obligations may have gained official state sanction and strength after 1861. The price/endowment effect of emancipation and redemption for former serfs was accompanied by numerous other changes. Therefore, to help identify whether the institution of serfdom or its reform mattered for long-run development in Russia, I identify a comparison group – the state peasants – who experienced a different institutional regime and reform process.
3.2 What about the Non-Seigniorial Peasants?
Peasants residing on privately owned land were not the only ones affected by the reforms of the 1860s. Serf emancipation was followed by similar acts for the former court (udel’nye) and state peasants.30 Under an 1866 measure, the state initiated a process to fully document the holdings of the state peasants. Communal land allotments were described in “ownership notes” (vladennye zapiski). These were compiled in a similar manner to the regulatory charters, but they were based on official cadasters of state property from the 1840s and 1850s, rather than any mediated bargaining process.31 These settlements typically granted state peasant communes all the land that they currently held.32 In return for this property, communities were made collectively liable for 20 years of quit-rent payments (obrochnye podati) at the level that they were already paying (PSZ, Ser. 2, No. 43888). Initially, these communal endowments did not entail full ownership rights, as the property was intended to be held under the “perpetual use” of the communities (Zaionchkovskii, 1960, p. 274). However, after legislation in 1886, these payments were converted into redemption obligations, and the former state peasant communities gained the same type of property rights that the former serfs held over redemption land (PSZ, Ser. 3, No. 3807).
Overall, the land and cash obligations received by the former state peasants have commonly been thought to be virtually identical to what they previously held. Furthermore, reforms enacted by P.D. Kiselev (the Minister of State Domains) in the 1840s had already formally established the communal basis for the land rights and collective obligations of state peasant villages (Adams, 1985; and Ivanov, 1945). As a result, the land settlements may have had little impact on both the institutional structure of the state peasant villages, or the land endowments and obligation levels they faced.
However, limited evidence on the ownership notes and redemption deals does suggest that the state peasants experienced some changes in their land endowments during their reform process. State peasants in Moscow province lost some arable land and access rights to a substantial amount of forested area (Druzhinin, 1978, p. 108). In Simbirsk, state peasants lost 14.8% of their land, although this was less than half of the percent lost by the former serfs (Kanatov, 1964). Even with these losses, the relationship of land to obligations was likely more favorable than that faced by the former serfs. Besides the pre-1861 differences in landholdings presented in Table 2, statistics collected by the Central Statistical Agency in 1877-78 show that on average, the former state peasants continued to hold substantially larger holdings than did the former serfs (cited in Druzhinin, 1978, p. 108). Moreover, it appears that the total obligations of the state peasant were substantially lower than the payments made by the former serfs and were very close to what they paid before the reforms (Ianson, 1881; and Ivanov, 1945, pp. 112-121).33 In provinces such as Moscow, the two types of peasant communities often lived alongside each other. The agricultural techniques of the serf and state peasants were remarkably similar, despite some efforts by the Ministry of State Domains to improve the techniques of the state peasants before 1861 (Deal, 1981; and Ivanov, 1945, p. 128). This suggests that the two groups were similar in many ways. However, economic conditions of state peasant villages did differ from those of the serfs before the reforms.34According to Mironov, serfs had 6-7% higher output per acre that state peasants in the 1850s (1996, p. 324). Deal (1981, p. 111) compares a random sample of serf estates and state peasant villages in Kharkov province and finds that serfs had higher per capita output on their allotment land in the 1850s. These numbers reflect an important pre-reform difference between serfs and state peasants: serfs were relatively specialized in agriculture and allocated more labor to grain cultivation, resulting in higher output per acre.
The relative diversification of the state peasant economy was at least partially due to more liberal property rights and policies regarding labor mobility. State peasants were granted the possibility of owning land in their own name far earlier than the seigniorial peasants (1801 versus 1848). They could enter contracts, own and inherit land, and freely engage in non-agricultural work without the approval of seigniorial officials (Blum, 1961, pp. 485-488).35 By the 1850s, over 90% of state peasant males in Moscow province were involved in some type of non-agricultural activity (Ivanov, 1945, p. 103). This was higher than the percentage of serf males, although there is less quantitative evidence on the pre-1861 allocation of labor by seigniorial peasants.36 Crisp notes that, “The better opportunities for earning and the relatively low rents of the state peasants gave them greater possibilities of accumulating capital” (1976, p. 93). According to Moon (2002, p. 355), former state peasants remained more geographically mobile in the period after 1861, in part because their greater wealth allowed them to take on the risks associated with volatile non-agricultural opportunities. He offers little empirical support to prove this hypothesis over an alternative one, where former serfs were more readily drawn into off-farm work due to the costly settlements they received.
After 1866, state peasants apparently retained most of the land they previously cultivated at a relatively low payment level. Serfs lost more land and remained responsible for larger obligations. Contemporaries and modern scholars argue that the economic and institutional conditions of the state peasants were substantially better before the reforms, and these translated into better outcomes afterwards.37 Some of the differences between the two types of villages – endowments and obligation levels – are measureable, while others – the amount of communal restrictions – are unobservable. But later measures in the 1880s and 1890s began to reduce former serf redemption payments, brining them more in line with those of state peasants by 1900.38 Moreover, Gerschenkron and others go on to argue that the two types of peasants really faced identical institutional conditions when it came to land rights and collective obligations at least after 1866.39 Statutes in 1886 and 1893 reiterated state and communal control over inheritance practices, land allocation, and the possibility of household exit from the burdens of redemption for both state peasants and former serfs.40 In this way, the rural institutional framework initiated in the 1860s came to encompass all types of peasant communities. This makes estimating the long run “effects” of serfdom a tricky endeavor.
Soviet scholars did not test their arguments regarding the economic effects of the serf reforms. Rather, they juxtaposed the immediate changes of the land settlements alongside various measures of landlessness, poverty, migration, and non-farm work. Early Western writers emphasized that the reforms reinforced communal restrictions on labor mobility. Both groups of scholars considered the reforms to have dramatically and negatively altered the path of Russian economic development. However, they took opposing sides on whether the reforms restricted or encouraged labor mobility.41 Recent research tends to view the reforms in a more positive light, emphasizing that post-1861 growth was relatively high and that the land settlement process had little effect on the rural economy. Implicit in the work of these scholars is the notion that the reforms had little additional effect on the mobility of labor out of agriculture.
To try and distinguish between these theories and better understand the long run implications of serfdom and emancipation, I compare former state peasant and serf villages in terms of the changes in non-agricultural activity that they experienced in the post-reform period. By focusing on changes over time, I am able to difference away fixed, unobservable determinants of off-farm activity that may pollute cross-sectional comparisons. This is not a true “difference-in-differences,” as I compare changes between the two types of villages where the “treatment” – serf or state peasant – was non-randomly assigned.42 Moreover, little information is available from before 1861, and so the available data restricts the analysis to changes in off-farm involvement well after 1861. I cannot tell when or exactly how differences between former serf and state peasant villages might have arisen. Despite these difficulties, evidence of any differential labor market flexibility does shed some light on which story about the post-reform rural economy contains more truth: Gerschenkron’s argument about communal restrictions (especially in former serf villages) or the Soviet view that the settlements received by the former serfs predisposed them to work more off the land than did those granted to the former state peasants.
3.3 Comparing State Peasant and Serf Villages: A Case Study from Moscow Province
In this section, I report preliminary results from an exercise that exploits variation in the institutional heritage of different villages in rural Russia to help identify how serfdom and emancipation may have generated long-run economic effects. As noted above, there are a number of reasons to think that formerly serf villages may have seen worse outcomes than formerly state peasant villages along a number of different dimensions, and that this gap may have persisted until the abrogation of redemption debt and collective responsibility in the early 1900s.In particular, higher obligation levels for worse land constituted worse endowments, and the constraints of communal property and obligations may have translated these endowments into efficiency losses and sub-optimal labor decisions. Little empirical work has been done to investigate whether these possibilities actually held. Deal’s (1978) important study is limited to simple comparisons between different types of villages in Riazan province, but it is not clear that his sample is representative and his focus is on the pre-1861 period. In contrast, I constructed a representative sample of villages from Moscow province and pulled together data on a variety of characteristics from three points in time: 1858, the late 1870s, and 1899. I use these data to directly examine whether formerly serf villages did show worse outcomes, and if so, did these persist.
These dataset employed here was primarily constructed from the output of two different zemstvo research programs carried out in Moscow province between 1870 and 1900.43 The first program was a village-level survey from 1876-77 that documented every one of the more than 5000 rural settlements in the province. This survey was enumerated at communal assembly meetings, where members jointly answered questions about conditions in the village. The output produced during this research was published in the volumes of Sbornik (1877-1882). This study collected information on each communal village’s population, land holdings, livestock, grain yields, obligations, and involvement in nonagricultural activities. The second research program was a survey of every village of Moscow province in 1899-1900 (Moskovskaia, 1903-1908). This household-level survey solicited responses from over 130000 peasant households, which were published in a summary format at the village level. The data collected in this study are similar to that from the 1876-77 survey but with more detail regarding agriculture, land, and communal practices. As a result of this effort (summarized below), village-level observations can be linked across the two surveys.44 I focus on Moscow province for a variety of reasons, not the least of which was the richness of the source materials produced by the zemstvo and published or retained in the archives.45 The substantial secondary literature on the growth of Moscow and the development of the regional economy describes an especially diverse and dynamic agrarian economy, with a large number of rural factories and significant protoindustrial employment, local labor migration, and agricultural production for industrial purposes and urban markets (Baker, 1978; Burds, 1998; and Johnson, 1979). Although Moscow resembled the other provinces in the Central Industrial Region (CIR), it may have been the case that serfdom took a somewhat different form in this especially dynamic province. The short agricultural season and poor soil productivity of Moscow province lowered agricultural returns, and the few former serf-owners that continued to farm tended to complain about their inability to hire labor away from non-farm work (see the examples in Zvenigorodskii, 1902).46 Moreover, the commune may have had less of a hold on peasants, with formerly serf villages facing a more flexible institutional environment as a result.
3.1a Constructing the Village Sample
Moscow province contained over 5000 communal villages at the time of the reforms. To study the determinants of involvement in labor activities off of the farm, this paper focuses on a sample of villages matched between the two surveys (and to some limited pre-1861 information). Both of the surveys provide information on community demographics, asset holdings, and agricultural production. Each survey gives a detailed breakdown of village allotment land and the level of involvement in non-agricultural activities (described below). Importantly, the surveys both specify whether village residents were former state peasants or serfs. But there are aspects of the two surveys that differ in important ways. The 1876 survey breaks out obligations into several different categories, including redemption and state peasant quit-rent duties, local taxes, and zemstvo payments. The later survey only provides the total obligations per acre of allotment land. Furthermore, the 1899 survey gives more detailed information on off-farm work by location and even provides data on things such as literacy and housing. Although data on all Moscow province settlements are available, this preliminary sample-based approach is flexible, offers a starting point for considering the reforms, and can be scaled up within Moscow or through other provinces with similar surveys.
The target size of the sample was set at 150 villages. I began with the 1876 survey. In the published volumes of the 1876 survey, townships are listed alphabetically by district. Each township’s data is then divided into rows corresponding to individual communities. On average, the 160 townships in the province contained roughly 32 settlements. In some cases, multiple villages were combined with only one summary line of data provided. I treat this as one “village.” Villages were not listed alphabetically but were sub-divided into categories depending on their status: temporarily obligated, in redemption, court peasants, state peasants, or full property owners with gift allotments or having fully redeemed their land. This necessitated a two-stage sampling approach. I first chose a 10% (N = 16) random sample of townships. Noting the number of observations from in each township, I assigned probability weights to villages from each township and then selected the final sample based on these.47 The resulting dataset includes villages randomly sampled from across the entire province and over the different peasant categories.
I then moved on to the 1899 survey and matched the 1876 sample to the corresponding observations. In approximately 25 cases, the 1876 “villages” were divided into more than one observation in the 1899 data. I consolidated these observations into single ones by either taking averages or summing, depending on the variable.48 Controlling for these consolidated villages in the empirical work below had no effect on the results. Of the 151 observations on villages in the 1876 survey, this procedure matched 149 in the 1899 data. These 149 villages comprise a 2.5-3% sample of the settlements in Moscow province.49 3.1b Documenting Economic Change in the Moscow Province
Table 4 documents the size of the sample villages, the communal land endowments, and the size of yearly payment obligations.50 This land is specifically denoted as nadel’naia zemlia, or allotment land, in the two surveys. This was the property that was transferred to both groups of peasants in their respective land settlements. Within the “serf” village category, I include the former seigniorial peasants and the small number of “full property owners.” This latter group had either already redeemed their land or had received their property as gift (darstvennye) allotments from their former seigniors. The category of “state peasants” includes the former court peasants. This classification was made from information in the 1876 survey. The results are not sensitive to changes in these definitions.
As Table 4 indicates, the former state peasant villages were considerably larger than the former serf villages in terms of population and land holdings. However, average household size in the two types of villages was similar in both years, falling from about 7 individuals in 1876 to just over 5.5 in 1899. For both types of villages, the average size of a household’s land allotment fell between the surveys. This was due to an increase in the number of households, rather than any change in the total amount of allotment land.51 At the same time, Table 4 also shows that there were large differences in per person and per acre obligations between the two groups of villages. Former state peasant communities had substantially lower tax and land obligations, a finding that is consistent with previous scholarship. However, this difference shrinks over time – the overall per acre burden for former serf villages dropped from 1.44 rubles to 0.85 in 1899, while for former state peasants, the drop was only 0.86 to 0.77 rubles. Land payments comprised a larger percentage (59% versus 44%) of total payments among former serf communities in 1876. The larger relative decline in obligations among the former serfs may, in part, reflect reductions in land payments that occurred in the early 1880s.52 It is useful to check whether the information on land obligations in the 1876 survey is indicative of what peasants actually paid in their land settlements. Documentation of the redemption loans made to former serf communities in Moscow province is available in the archival materials of the Main Redemption Administration, which are held in RGIA in St. Petersburg. I draw on lists of the redemption loans made in Moscow province that were compiled upon the special lowering of payments in 1881. These lists provide the total allotment acreage and loan amount for villages that financed redemption through the state.53 I was able to locate 62 former serf villages of the linked sample in these records.
To compare the samples, I took the portion of the yearly obligations attributed to redemption payments in 1876 and capitalized this amount at a 6% interest rate to arrive at the total redemption loan per allotment acre made to these villages (see Zaionchkovskii, 1958, pp. 304-305). Table 5 compares the results with the equivalent loan data provided in the archival documents. Although the correlation is not perfect, the means are quite close despite the roughness of this calculation.54 Thus, the payment information in the 1876 survey does a good job of describing the “price” of land in the former serf settlements. There is little reason to think that the same is not true for the information on former state peasant villages. Since both groups of peasants were equally affected by changes in overall tax policies over the period, the closing of the gap in obligation levels was most likely driven by the relative decline in the redemption component of total former serf payments.
State peasant communities had higher yields of the major food crop, rye. As Table 6 indicates, there was some catch-up between the two surveys, but agricultural output per acre remained lower in the former serf communities.55 State peasant villages had a higher percentage of allotment land in hay meadows, forested areas, and other non-arable property. According to the data in Tables 4 and 6, both types of community possessed approximately 12 acres of arable land per household. The extra non-arable land held by state peasant villages presumably allowed them to support more cattle and horses per household (Table 6).56 As a result, livestock inputs (manure and motive power) on cultivated land were likely higher among the former state peasants. This helps explains the higher grain yields in former state peasant villages.
The information in Tables 4 and 6 confirms that the former serf communities held smaller land endowments that were likely to have been less productive. At the same time, they were obligated for higher obligations per acre than the former state peasants. Of course, these conditions may have reflected pre-existing differences between serf and state peasant villages (in terms of the amount and productivity of land) that were then fixed by the land settlements. Rather than build on their tradition of relative mobility off the farm, former state peasants may have been less inclined to engage in non-agricultural occupations for precisely this reason.
Table 7 summarizes information on the allocation of labor to non-agricultural pursuits, both in and outside the village.57 Specific information on occupations in the villages is unavailable, but the two surveys both indicate how many individuals worked na otkhod, or outside the village. These occupations were typically temporary or seasonal and required travel documents. The 1899 survey also gives the number of individuals occupied in any type of non-agricultural trade (promysly), regardless of the location. The 1876 survey provides the total number of households not occupied in agriculture. I derive an equivalent measure for 1899 by subtracting the number of households farming their land from the total number of present and absent households.
Former serfs were slightly more likely than state peasants to be involved in non-agricultural activities in 1876.58 However, these small differences disappeared by 1899, with none of the gaps between the groups remaining statistically significant. Over 50% of males from both groups of peasants were engaged in some form of non-farm work in 1899 (0.43 or 0.44 ÷ 0.83), predominantly outside the village. This percentage includes underage males in the denominator. Females were far less likely to be involved in non-agricultural activities (only 12-15% were in 1899), but they were slightly more likely to remain in the village when they did work off the farm. There is some evidence of an increase in the number of households specializing away from agriculture: the overall percentage of landless households increased from approximately 10 to 20%. This is consistent with the Soviet view that there was a growing pool of potential recruits for non-agricultural employment. However, the increase in the share of landless households, regardless of how they are classified, does not fully account for the apparent decline in agricultural employment between the surveys.59 To summarize, former serf communities had smaller land holdings and paid a higher price for them than did the former state peasants after 1861. Previous scholarship suggests that serfs were more specialized in agriculture before 1861. However, I observe former serfs with a slightly greater level of involvement in non-agricultural activities outside the village in 1876. The disappearance of this small difference by 1899 may reflect the relative decline in the amount of the former serfs’ land payments (i.e. their need for cash). But there may be other explanations for this finding. The higher share of former serfs working outside the village in 1876 may simply indicate a different mix of non-agricultural work inside and outside the village than what the former state peasants engaged in, although, there are no apparent differences in the location of non-agricultural activities between the two groups by 1899. Alternatively, the small initial difference in off-farm activity may be evidence that former state peasants were simply continuing to take advantage of better agricultural endowments. The reduction of this gap may be evidence of the inability of former serfs to take advantage of non-farm opportunities in the presence of institutional rigidities (i.e. a stronger communal system of land control and obligations). The next section takes a stab at untangling these explanations by focusing on the determinants of changes in off-farm involvement between the two surveys.
3.1c Econometric Evidence on Peasant Labor Allocation
The former state peasants received different “treatments” in their land settlements than did the former serfs, but their initial conditions were different as well. Such pre-existing differences in endowments make it difficult to infer a causal relationship between the harsher reforms of the former serfs and their adoption of non-farm occupations. This is especially the case in the cross section, where unobservable factors such as soil quality or other characteristics of the seigniorial estates likely affected both initial conditions and the severity of the land settlements themselves. For these reasons, and given the available data, I focus on changes in the level of involvement in non-agricultural activities between 1876 and 1899. Thus, I consider the effects of changes in obligations and population (which drove the decline in mean land endowments) on the extent of off-farm work, and how these relationships differed between former serfs and state peasants. This approach helps to distinguish the role played by endowments and prices from that of any persistent institutional differences between the two types of communities.
An important empirical issue is the dearth of information on conditions before the land settlements. The evidence presented above suggests that the 1876 data do capture the essential features of the land settlements. But the land settlements themselves likely reflected pre-existing differences. Unlike Kashchenko’s work, I only observe villages for the first time 15 years after Emancipation. If any changes wrought by the reforms led to adjustments by 1876, these will not show up in my data, and any responses to the settlements that I observe will be muted.60 Given that communal land endowments remained fixed over the entire period, I focus on whether changes in obligations and the availability of labor between 1876 and 1899 were handled differently by the two groups of peasants. This tests whether involvement in nonagricultural activities was a product of population growth, demands for cash payments, or institutional differences between the two types of villages.61 Understanding how community characteristics were related to off-farm activity is an important step towards evaluating different interpretations of the reforms or the long-run effects of serfdom.
To study the determinants of changes in off-farm activity, I begin with specifications of the form:
Yi = b0 + b1 Obligationsi + b2Populationi+ b3Xi + ei (1)
Here, Yi indicates changes in measures of participation in non-agricultural activities by the residents of village i between the two surveys. The right-hand-side variables include changes in the level of obligations, village population, and other possible determinants of off-farm activity, specifically the proximity of the village to a railroad station. I denote Population by two variables – the percentage change in the total village population and the percentage change in the number of member households between the surveys. I divide demographic change in this way to capture two distinct effects.62 Larger population changes should be positively correlated with involvement off the farm if households with fixed land allotments allocated excess labor to non-agricultural activities. At the same time, more households (holding population fixed) may have affected the prevalence of off-farm work if labor markets were imperfect or returns to scale in household production were not constant. Furthermore, the literature on the Russian peasant household in this period emphasizes that there was a sharp increase in household divisions after 1861. Younger households were more likely to spilt off from the larger, patriarchal household that had long characterized rural Russia. This was possible because of growing non-agricultural opportunities for young men and women, who no longer had to wait for a share of the commune’s land to start an independent household (Moon, 1999; and Worobec, 1995). I control for this possibility by including the change in the number of households as an independent variable. The other regressors are defined as the percentage change in their values between the two surveys. Table 8 summarizes these variables.
The 1899 survey only provides information on total community obligations per acre. I use the equivalent measure from the 1876 survey to derive the percentage change in the “price” of allotments over the period. If high obligation levels were critical determinants of off-farm activity, then greater values of (percentage) Obligations – which means a smaller decline over the period – should be associated with greater increases in off-farm involvement.63 In the tables presented below, the Xi vector only includes DistanceToRRStation. This variable is intended as a proxy for any increase in the integration of a village with more distant markets. The more negative this variable, the closer a village found itself to off-farm opportunities by the end of the period. By reducing the costs of accessing non-agricultural employment, this should positively affect the decision to allocate labor off the farm. Other variables such as the change in the percentage of working-age villagers, the change in the percentage of land that was arable, and changes in the number of households were considered and found to be unrelated to off-farm activity (or suffer from serious endogeneity concerns).
I extend the basic model by interacting the explanatory variables with an indicator variable that equals one if the village was comprised of former serfs.64 I also include the indicator variable on its own to allow for different trends in the two types of village. This makes it possible to run a comparison. What drove village-level differences in off-farm employment: different institutional backgrounds (the direct effect of serfdom) or differential responses to the changes in population or land “prices” (serfdom interacted with other variables). I present estimates of Equation 1 without controls for the type of the village in Table 9 and those that control for the former status of the village in Table 10.
The measures of non-agricultural work in these two tables were determined by their comparability across surveys. I focus on three: changes in the share of males working in non-agricultural occupations outside the village, the equivalent share for females, and the share of households not engaged in agriculture. This last measure is taken with landless households defined to be non-agricultural.65 Since the 1876 data lacked information about non-farm work within the village, I only consider changes in the amount of labor allocated to non-agricultural occupations outside the village. Thus, my individual measures should be interpreted as changes in the prevalence of migrant labor among villagers. In Section 4.3, I use the 1899 survey to analyze the differences in the correlates of non-agricultural work that took place inside versus outside the village.
These three dependent variables are all defined as changes in percentage points. However, their form differs from that of the regressors, which are defined as percentage (point) changes. A cross-sectional version of Equation 1 would have the dependent variables in percentages and the right-hand-side variables as levels. Formulating a dynamic version of such a regression required slightly different transformations on each side. A one-percentage point increase in a regressor is relative to the 1876 level. An estimated coefficient then translates this increase into changes in the dependent variables that are not relative to any baseline level. It was necessary to define the dependent variables in this way since there were a substantial number of villages where men, women, or households were fully occupied in agriculture in 1876. Thus, a measure of off-farm work in percentage changes could not be constructed.66 Running these regressions with different transformations of the variables led to results very similar to those presented here.67 3.1d Econometric Results
Table 9 displays the results from estimating Equation 1 for the three different outcome measures. All the models in this and the following tables control for clustering of the standard errors at the township level. Including township-level fixed effects weakened the results slightly but had little impact on the size or signs of the coefficients. Since estimating Equation 1 should already difference away unobserved heterogeneity that was fixed (and entered linearly), I do not present the fixed-effect results here to conserve space. Table 9 provides evidence that changes in non-agricultural work outside the village were associated less with a reduction in obligations than with changes in village demographics.
The persistence of higher payment requirements was unrelated to all three measures of non-farm work. However, all three dependent variables were associated with population growth. For individual males and females, an increase in population was strongly associated with more workers leaving the village to find work. An increase in population of two percent from 1876 levels (the mean amount) was associated with 0.25 (women) to 0.5 (men) percentage points more individuals who were occupied outside the village. The effect is small and not statistically significant for households, but greater population growth was related to more household-level diversification (i.e. continued involvement in agriculture). Growth in the number of households had opposite effects. Households was associated with a reduction in the shares of males and females working outside the village and an increase in the number of households not working the land, although the latter two effects were not significant. Overall, the two demographic variables were jointly significant for males and females (p-values under 0.05) but not for households. Finally, households and individuals in those villages that were closer to markets by 1899 (more negative values of DistancetoRailroad) showed no differences in the prevalence of non-agricultural work. One explanation for this could be that long-distance market integration was less important as a source of non-farm employment than local factories, artisan work, and proto-industrial production.
Table 10 shows that much of the variation in how labor was allocated outside the village and away from agriculture stems from differences between former serf and state peasant communities. The first line of Table 10 captures the different growth trends in non-agricultural activity in the two types of villages. Changes in individual employment outside the village were smaller for the former serf villages (significant for females). The first coefficient in the third model of Table 10 indicates that former serf households were more likely to leave agriculture over time. This effect is not significant, but one explanation for the seemingly conflicting trends of individual work outside the village and household specialization away from agriculture is that the two types of villages offered different local non-agricultural opportunities. I explore this possibility in more detail below.
In Table 10, Households proved to have little relationship to measures of off-farm work.68 For males and females, the point estimates on Population and the interaction term suggest that the former state peasants were much more likely to shift labor off the land in response to increases in population than were former serfs.69 For the share of households not working the land, members of former serf villages displayed a different response to closer market access than did former state peasants. If a railroad station was 40% closer to a former state peasant village in 1899 than in 1876 (one standard deviation of DistanceToRRStation), then approximately 10% more households were likely to shift out of agriculture (than if the railroad was no closer). However, in former serf villages, households were less likely. Finally, former serf households were less likely to shift into non-agricultural occupations when payment demands remained high. The point estimates for males suggests that larger declines in obligations between 1876 and 1899 led to increases in the shares occupied outside the village (implying an easing of tighter communal obligations?). However, Obligations and its interaction term were not jointly significant for either males or females.
Conditional on these variables, individuals in former serf villages were less likely to shift to work outside the village, while households were more likely to give up farming their own land. Individuals in former serf communities showed less willingness to shift to non-agricultural work beyond the village in the face of population growth. Similarly, increased access to off-farm opportunities and the persistence of high obligation levels did not induce any further growth in the share of households who quit farming their allotment. By 1899, the slightly higher prevalence of non-agricultural work among former serfs that was evident in 1876 had disappeared. These results are surprising given the apparent comparative disadvantage of former serf villages in agriculture, as well as their higher cash obligations.
Moreover, these findings are inconsistent with the Soviet/Marxist view that conditions created by the land settlements “priced” former serfs off the land and forced them to look outside the village for employment. Rather, these results are sympathetic to interpretations of the peasant reforms that emphasize the persistence of significant institutional restrictions on labor mobility in former serf villages relative to former state peasant communities. Even though former serfs were more likely to participate in the labor market outside of their villages in 1876, something prevented these peasants from building on this over the ensuing decades. The better agricultural endowments of former state peasants may have led them to remain more involved in agriculture in 1876. But these endowment differences do not help explain the relative decline in off-farm activity by the former serfs.
There is another possible, albeit related, explanation for these findings. Tables 9 and 10 do not take into account the possibility that there were differences in the location of nonagricultural activities between former serf and state peasant villages. Former serfs may have been able to turn to a relatively larger number of local opportunities when they left agriculture. Indeed, if there were institutional constraints keeping former serfs within the village, this may have created reinforced conditions (cheap labor, entrepreneurial finances from former serf owners, etc.) that were conducive to locating production there as well. This would explain the apparent contradiction between the falling shares of individuals working outside the village and the increasing number of households not working their land in former serf communities. This possibility can be explored with the 1899 data.
3.1e Extension: Migration vs. Local Non-Farm Work in 1899
The first two columns of Tables 9 and 10 focus on an individual’s decision to work outside the village. Such occupations included factory employment, work in the transport sector, or other urban service jobs. They required extra passport documentation that was costly to obtain and renew and involved some greater amount of separation from the household. Alternatively, households and individuals could undertake non-agricultural activities in the home (such as textile weaving or handicrafts) or in local putting-out establishments that produced intermediate goods for finishing in larger factories (Tugan-Baranovsky, 1907 ). Indeed, if entrepreneurs took advantage of (possibly) cheaper labor in former serf communities by locating production there, these peasants may not have turned to opportunities outside of the village. Alternatively, former serf-owners may have provided relatively cheap finance for former serf businessmen, initiated home production of manufactured goods, or directed other local, non-agricultural activities themselves.
The 1899 data offer additional information on where non-agricultural work took place.
Specifically, this survey indicates the total number of individuals occupied in non-agricultural activities and whether these men or women worked inside or outside the village (but not their exact occupations). Women worked in non-agricultural occupations in approximately equal shares inside (10%) and outside (9%) the village, while the equivalent percentages for men were 9% and 42%. Thus, men were much more likely to migrate to find employment. I use this information to investigate the differences between participation in migratory labor and involvement in non-agricultural activities that took place in the village.
I do this by estimating cross-sectional regressions of the shares of men and women occupied in the two types of locations. (I do not estimate these regressions in terms of percentage points, but the conversion simply requires multiplying the coefficients by a factor of 100.) These specifications closely follow those of the previous section. However, with only one time period, unobservable heterogeneity may be a significant problem. The inclusion of township-level fixed effects weakens but did not fundamentally change the results (I do not report these results here to conserve space).70 As in Tables 9 and 10, I focus on the correlations between off-farm activity and factors such as total obligations per acre, population density, and proximity to markets. I chose population density rather than overall population because it more exactly captures the average land-labor ratio in the village (and avoids multi-collinearity from including separate population and land-holding measures). I define this as the total population of the village divided by the acreage of allotment land, and I also control for the portion of a community’s land that was suitable for grain cultivation.71 Obligation levels and distance to railroads are measured as 1899 cross-sectional versions of the variables used in Tables 9 and 10.
The 1899 survey also provides a measure of human capital in the form of the share of a village’s male population that was literate. I include this variable in both male and female regressions to test whether higher returns to human capital were found off the farm, and whether these opportunities were more prominent through migration.72 For the female regressions, the variable may be considered a proxy for female human capital. Other evidence suggests that female literacy was quite low, and so this variable may be capturing some wealth effects if education and literacy are normal goods. I interact all of these variables with whether the village was populated by former serfs to see if the patterns differed between the two types of communities.
Consistent with the cross-tabulations presented earlier, the results here show little evidence that, ceteris paribus, former serfs were more involved non-agricultural activities in 1899 than were former state peasants. This is true for both types of non-agricultural work. However, several factors did have a differential impact between the two types of villages and the two potential locations of non-agricultural work. An increase in the share of literate males was associated with a lower share of non-agricultural workers in the village and a higher share of migratory workers. This relationship was only significant for males and differed little between former serf and state peasants villages. This finding is consistent with better rewards for human capital outside the village, although it could simply reflect the relative ease with which educated peasants found such opportunities.
Distance to the nearest railroad does not seem to be an important factor for where or whether males worked off the land. However, it was negatively related to the share of female former serfs employed outside the village. The sign of this (interaction) effect with regards to employment within the village is positive but insignificant. These results support the possible presence of some sort of institutional constraint that kept females in former serf villages from fully accessing external employment opportunities.
The estimated coefficients on population density and its interaction with former serf status indicate that this variable was positively associated with non-agricultural work in the village for males in former state peasant communities. However, this effect was muted or even reversed for former serf villages, where population density had approximately zero net-effect on the share working in either location. The results are similar for women, although their decisions to work outside the village were only slightly related to population density.73 It appears that even by 1899, former state peasants continued to be more diversified among local farm and non-farm activities. This lends support to the idea that former serfs exhibited less flexibility in their allocation of labor off the farm. However, these results are not entirely consistent with there being more relatively less external non-agricultural opportunities for former serfs.
Higher levels of obligations were uncorrelated with the share of males working in nonagricultural occupations in 1899, both inside and outside the village. But obligation levels had a (weak) negative association with non-agricultural work for females, and this was offset in former serf villages, especially for work inside the village (the coefficients are not jointly significant for work outside the village). This implies that higher obligations were more likely to push women into non-agricultural occupations within former serf villages.
In combination with the findings for railroad distance and population density, these results imply that there were differences in the location of non-agricultural activity between the two types of communities. By drawing on local employment opportunities, women in former serf villages did offer their households a margin for dealing with higher obligation levels. Overall, however, these preliminary findings appear consistent with the existence of factors that restrained members of former serf villages from being freely able to engage in non-agricultural employment outside of the village.
4 Serfdom and Emancipation in the Long Run: A District-Level Analysis (Planned)
The village-level analysis explored in the previous section offers a useful starting point for considering the long-run economic implications of serfdom along one dimension – labor market flexibility. While the serf/non-serf village comparison offers a possibly quasi-random natural experiment, the allocation of serfs across districts was not really random, as Section 2 suggested. When aggregated up, it is especially difficult to claim any sort of causal relationship between serfdom and economic outcomes in the decades that followed due to the potential endogeneity of serfdom (including the variants). In future work, I will be exploring some possibilities at getting around these problems using more aggregate data from across European Russia. The district-level data offer many more possible avenues to explore the implications of serfdom and emancipation, including post-1861 trends in the allocation of (privately owned) land, education and literacy, occupational choice, political rights, and wage or income levels. I hope to present some of these ideas at the conference.
5 Concluding Thoughts
To be discussed and written!