Abstract

This paper examines the long-run economic consequences of Russian serfdom. Employing data on the intensity of labor coercion just prior to emancipation in 1861, we document that a 25 percentage point increase in historical serfdom (1 SD) reduces household expenditure today by up to 17%. We then provide evidence on the persistence of this relationship by studying city populations over the period 1800 to 2002. Exploring mechanisms, our findings suggest that less urban agglomeration and slower industrial development in areas with a greater degree of serfdom perpetuated the negative effects of forced labor before, during, and after the Soviet period.

I. Introduction

TWENTY-FIVE years after the fall of the Soviet Union, the economies of Eastern Europe still lag behind. A large body of research has attributed the slow rate of their convergence with advanced economies to the legacy of Soviet-era institutions and the difficulties in transitioning to a market economy. The relatively slow pace of development of the former Soviet member countries may also have deeper historical roots. At the turn of the nineteenth century, Imperial Russia was one of the poorest economies in Europe. In 1900, per capita incomes in the countries that would later comprise the USSR were only about a third of those in Western Europe ($1,196 versus $3,155).1 While it has been argued that low levels of economic development today could reflect persistent legacies of the Imperial period (Roland, 2012), this hypothesis remains largely untested and the possible mechanisms unexplored.

Figure 1.

Peasant Emancipation and Long-Run Development in Europe

This figure plots log GDP per capita in 2014 against the year of peasant emancipation in European countries. See the appendix for data description.

Figure 1.

Peasant Emancipation and Long-Run Development in Europe

This figure plots log GDP per capita in 2014 against the year of peasant emancipation in European countries. See the appendix for data description.

In this paper, we examine whether Russian serfdom generated such long-term economic consequences. Serfdom was not only one of the most prominent institutions of forced labor in history, but it is frequently regarded as a crucial factor behind Imperial Russian (under-) development (Acemoglu & Robinson, 2012; Markevich & Zhuravskaya, 2018). Figure 1 provides suggestive evidence of the legacy of similar institutions across several European countries. The figure depicts a striking negative correlation between the timing of peasant emancipation and the level of development today, which suggests that Imperial Russia's retention of serfdom until the 1860s may have contributed to lower income levels in the longrun. Clearly, the societies in figure 1 differ across many historical and contemporary dimensions, making it difficult to isolate the importance of serfdom. Therefore, to test whether and how this correlation may be indicative of an underlying causal relationship, this study investigates the economic effects of serfdom within the area of the former Russian Empire, making use of disaggregate data on labor coercion at the level of the district (uezd) just prior to formal emancipation in 1861. Our main estimates document a significant negative relationship between this institutional heritage and measures of economic development today. Critically, we complement this finding with a careful exploration of the possible mechanisms that generated this pattern. Rather than direct institutional, cultural, or human capital channels, the evidence suggests that initial economic differences interacted with evolving but high restrictions on labor mobility, delayed industrialization, and Soviet-era geographically differentiated policies to generate long-run structural impediments for development in former serf areas.

Russian serfdom as a system of labor coercion existed from the sixteenth century until 1861, when about 45% of peasants (and 38% of the total population) in the European provinces were obliged to work for the landowning nobility and/or pay them a portion of their income in the form of quitrent.2 Formal emancipation was followed by a drawn-out process of land reform that transferred property rights (generally assigned to the communal village) and associated mortgage-like obligations to the newly freed peasants. The experience of these privately “owned” serfs may be contrasted with what happened to rest of the peasantry, who resided on either state or Imperial family-owned lands prior to 1861. Serfs possessed less land and faced more restrictions on their labor, education, and entrepreneurial decisions prior to the 1860s, and the emancipation reforms solidified these differences in the short and medium term.

In this paper, we leverage this heterogeneity, defined across districts (uezdy), within the pre-1861 peasantry to identify the long-run consequences of serfdom. Assessing the potential determinants of serfdom's geography, we find that serfdom was more prominent in districts closer to Moscow (conditional on historical province fixed effects), consistent with the spread of the Imperial state, but was only weakly associated with biogeographic conditions such as agricultural suitability. To investigate subsequent economic differences across districts with heterogenous levels of historical serfdom, we link our measure of labor coercion to rich data on modern outcomes and on outcomes from intermediate dates in the Imperial, Soviet, and post-Soviet periods. Our main results document that households in districts where serfdom was widespread before 1861 are poorer today: a standard deviation increase in the share of the population who were serfs (about 25 percentage points) is associated with a 9% to 17% lower average household consumption among modern households.3 Further, we document the persistence of this pattern of differential economic activity for the period 1800 to 2002 and find that cities were significantly smaller in locations with more historical serfdom prior to emancipation, that this gap did not fully close after 1861, and that, if anything, this difference widened during the Soviet period.

Given this pattern of persistence, it is imperative to investigate the underlying mechanisms. The literature from other contexts emphasizes how persistent cultural, ethnic, racial, or institutional characteristics of a previously coerced population help generate long-run divergent outcomes.4 In contrast, our Russian context shuts down these channels, as such differences were largely nonexistent between former serfs and the rest of the population, especially after the Bolshevik Revolution dismantled institutional and social structures.5 Instead, we focus on evolving constraints on factor mobility that reinforced initial gaps between low- and high-serf areas to generate path dependencies for local structural change. We hypothesize that structural change and industrial agglomeration were less prominent in former serf areas throughout the period, and the gap solidified with the emergence of more modern sectors and reinforcing labor, migration, investment, and resource allocation policies in the Soviet and post-Soviet periods.6

To empirically evaluate this hypothesis, we draw on novel district-level data on urbanization, infrastructure, industrial development, property holdings, human capital, and policy preferences across our period. We establish that the incidence of serfdom was negatively associated with the level of urbanization, industrialization, and tertiary sector employment in Imperial Russia; road densities and the presence of firms in the Soviet period; and population density and nighttime luminosity after 1990. We also find that the greater prevalence of quitrent obligations—for which serfs enjoyed more autonomy to engage in nonagricultural activities away from the estate—was associated with lower employment in agricultural occupations and greater employment in industry in the late Imperial period. While documenting that serfdom was associated with fewer industrial establishments over the period 1939 to 1989, we also show that firms in former serf areas were smaller, less productive, and more likely to be in agriculture than manufacturing at the end of the Soviet period. Although schooling outcomes were similar between areas with more and fewer serfs by the end of the Imperial period, we estimate substantial gaps in educational attainment in modern data, consistent with the demand-side consequences of a growing complementarity between labor skills and modernizing industry in the Soviet Union. Considering plausible alternative mechanisms, we find little support for a direct channel of persistence working through economic inequality, political structures, or reduced public good provision.7 The evidence also suggests that serfdom is not associated with modern cultural differences, such as trust, xenophobia, institutional preferences, and political participation or with Communist party membership during the Soviet period.8 Overall, our results identify a set of theoretically and historically consistent linkages between the incidence of past serfdom and the current spatial distribution of economic activity across the former Russian Empire.

Robust empirical work linking labor coercion in Imperial Russia to subsequent or contemporaneous economic outcomes is limited.9 Two exceptions are Markevich and Zhuravskaya (2018), who estimate that provinces with above-average levels of serfdom grew relatively faster after emancipation, and Nafziger (2013), who shows that the emancipation and land reform processes homogenized institutional structures but fixed differences in factor endowments and prices between formerly serf and nonserf areas into the twentieth century. These empirical analyses suggest that serfdom imposed meaningful constraints on the rural economy, that some of these were relieved by the reforms of the 1860s, but that former serf areas continued to face persistent differences in land and labor market conditions until the Soviet period. Our study is the first to examine whether economic differences between high- and low-serf regions lasted beyond 1917, and we provide new evidence on how institutional legacies can constrain structural change, thereby generating the persistence of spatial development patterns.10

The paper proceeds as follows. Section II describes the historical background. Section III examines the effect of serfdom on long-run development. Section IV documents the nature of persistence in this pattern. Section V investigates mechanisms, and section VI concludes.

II. Historical Background

A. Serfdom and Emancipation in the Russian Empire

In return for military service to the czars during Muscovite state expansion in the sixteenth and seventeeth centuries, the elite received land grants that came with the right to draw on the labor of the resident peasantry. However, with competition among the servitors and the ease of fleeing to open land, the high land-labor ratio motivated the landowning nobility to impose increasingly coercive controls on their peasants. These attempts were reinforced by the state through a series of decrees, culminating in the 1649 Ulozhenie that sharply constrained peasant mobility and formalized the legal rights of the serf-owning nobility. Further eighteenth-century measures solidified the authority of the nobility, so that by 1800, the legal and institutional structure of Russian serfdom was firmly in place.

Serfdom varied widely across estates but possessed certain common characteristics. First, serfs constituted a distinct social estate, with the nobility holding ultimate authority over the daily lives of their property that allowed them to intervene in marriage, employment, educational, religious, judicial, and other matters (Wirtschafter, 1997).11 Second, serf owners demanded seigniorial obligations: labor services, cash or in-kind payments, or a combination. On many estates, owners actively managed the labor decisions of their serfs, either in person or through managerial staff. Such estates often possessed demesnes, with serf labor on the owner's land compensated by the granting of use rights to other property. On other estates, serfs were granted substantial freedom to allocate their labor as they saw fit. This variant was more common in less agriculturally productive regions, where owners tended to transfer the use of all estate land to the serfs in return for cash or in-kind payments (Moon, 1999).

Thus, the labor and property decisions of serfs were relatively constrained, which created disincentives for investment (of all sorts), impeded the adoption of better agricultural techniques, and led to the misallocation of labor and other resources in and across sectors. Many contemporary observers argued for serfdom's negative growth implications, while supporters of the status quo defended the institution less in economic terms than to maintain the Imperial regime and support elite tutelage over masses ill equipped for freedom (Field, 1976; Khristoforov, 2011).

However, there remains relatively little causal evidence on the economic impact of Russian serfdom or emancipation. Domar and Machina (1984) used information on the price of land with and without resident peasants to argue that serfdom was profitable to the nobility up to 1861. Based on evidence from a single large estate, Dennison (2011) argues that serfdom generated adverse distributional and growth effects.12 An important recent contribution by Markevich and Zhuravskaya (2018) evaluates the impact of serfdom by looking at differential economic changes between provinces with more or fewer serfs before and after 1861. The results from the study suggest strongly negative effects of serfdom, although the authors do not explicitly identify a mechanism behind their findings. Overall, most scholarship on Russian serfdom asserts that the institution undermined economic development while it existed.

More attention has been paid to the short- and medium-term consequences of emancipation in the half century before the Bolshevik Revolution. Soviet studies (Litvak, 1972) argued that emancipation and the accompanying land reforms actually worsened former serf landholdings and property rights (by reinforcing communal ownership) and imposed considerable new tax and payment burdens on the rural economy. In contrast, some more recent research (Hoch, 2004) asserts that the majority of former serfs were made better off—at least in terms of land and obligations.13 In his influential interpretation, Gerschenkron (1966) emphasized the negative implications of communal property rights (and associated joint liability for land and tax payments) for agricultural productivity and labor mobility after 1861. Gerschenkron and others writing in this vein (Allen, 2003) tend to focus on broader institutional impediments that characterized all peasants. By the 1880s, the different types of peasants were administratively unified and possessed similar institutions of communal self-governance, (generally) collective property rights, and joint liability for taxes and land payments. However, despite apparent nominal institutional similarities across peasants, Nafziger (2013) shows, using more disaggregate data than previous studies, that landholdings were smaller, land inequality was greater, and the associated land and tax obligations were higher in districts with relatively more former serfs, well into the twentieth century.

Gerschenkron (1966) also argued that the Stolypin land reforms of the early twentieth century improved incentives in peasant agriculture by offering mechanisms for consolidating plots and exiting the commune, thus alleviating some constraints on labor mobility and agricultural productivity (Chernina, Castaneda Dower, & Markevich, 2014; Castaneda Dower & Markevich, 2017). These measures were just the first in a series of dramatic developments in twentieth-century rural Russian society: the Bolshevik Revolution, wars, collectivization, famine, industrial policies, and the slow collapse of the agricultural sector from the 1970s onward. None of these changes explicitly or differentially targeted former serfs, but as we develop further below, they may have built on and reinforced geographic, institutional, and economic differences in ways that perpetuated existing gaps in economic development between former serf and nonserf areas.

B. Measuring Nineteenth-Century Serfdom

Serfdom was a defining feature of Russian society in the early nineteenth century, but by the 1850s, peasants residing on noble–owned land were a minority. The share of such serfs in the Imperial population crested at just over 50% at the turn of the eighteenth century, before falling to roughly 35% just before emancipation (Hoch & Augustine, 1979; Kabuzan, 2002). In contrast, peasants on state or Romanov family-owned land were governed by specific government ministries, typically possessed more land and greater freedom to engage in contracts, and were generally liable only for direct (and lower) tax-like obligations (Nafziger, 2013). As noted above, factor endowment differences persisted in the decades after 1861, while these different groups of peasants experienced at least nominal administrative and legal convergence following serf emancipation.

We study serfdom in the European part of the Russian Empire at the administrative level of the district (uezd), the largest subunit of a province. Relying on the tenth tax census of 1858, as reported in Troinitskii (1982 [1861]), we construct our main indicator of serfdom's intensity, Serfs % (1858), which divides the total number of serfs by the total district population.14 The resulting measure covers roughly 490 historical districts in fifty provinces of European Russia, without Poland or Finland. In our data, serfs averaged 38% of a district's population.15 Appendix figure A2 shows the underlying variation in serfdom across European Russia just before emancipation, suggesting that the institution was largely concentrated in a band from Kiev to the upper Volga. However, even within high-serfdom provinces, there was considerable variation in the share of the population subjugated to the nobility.

C. Correlates of Serfdom

We first examine the extent to which districts with a greater prevalence of serfdom were systematically different. We focus on what the historical record and economic logic would suggest were important factors underlying the geographic incidence of serfdom just prior to emancipation.16 If the prevalence of serfdom was associated with many district characteristics, we would be concerned about the influence of unobservables that are themselves correlated with our observable covariates. We address such concerns in our analysis.

Table 1.
Determinants of Serfdom
Serfs % (1858)Types of Serfs: Share
All DistrictsLiTS DistrictsQuit-rentCorvéeHousehold
(1)(2)(3)(4)(5)(6)(7)
Latitude −0.100 −3.236 −1.333 3.526 1.084 −1.798 −0.783 
 (0.648) (2.026) (1.967) (3.448) (3.627) (3.057) (2.978) 
Longitude −0.652** −1.299* −1.364** −0.642 −0.390 −0.225 0.655 
 (0.322) (0.672) (0.659) (0.910) (0.824) (0.872) (0.471) 
Distance to Moscow −3.724*** −3.360*** −2.887** −2.772 −1.485 −0.452 1.848 
 (0.849) (0.844) (1.147) (2.558) (2.156) (2.241) (1.152) 
Cereal Suitability 4.471** 3.940** 2.267* 4.005* −4.628* 3.254 1.975 
 (1.725) (1.955) (1.166) (2.242) (2.400) (2.188) (1.181) 
Distance to Coast  2.203** 1.735 0.749 2.595 −1.880 −0.816 
  (0.955) (1.218) (1.819) (1.780) (2.448) (2.125) 
Distance City in 1600  12.718 7.397 −8.723 −20.344 −10.379 9.727 
  (13.986) (13.063) (18.656) (15.759) (25.314) (20.260) 
Distance Provincial Capital  1.005 −0.092 0.983 0.008 0.026 −1.050 
  (1.460) (1.217) (1.599) (1.609) (1.159) (0.973) 
Additional geography  ✓ ✓ ✓ ✓ ✓ ✓ 
Fixed effects   Province Province Province Province Province 
R2 0.37 0.46 0.71 0.78 0.72 0.80 0.38 
Observations 490 490 490 185 472 472 490 
Number of clusters 50 50 50 45 49 49 50 
F-statistic joint significance 21.94 9.72 2.27 2.19 1.58 2.19 1.83 
p-value joint significance 0.00 0.00 0.02 0.03 0.13 0.02 0.06 
Serfs % (1858)Types of Serfs: Share
All DistrictsLiTS DistrictsQuit-rentCorvéeHousehold
(1)(2)(3)(4)(5)(6)(7)
Latitude −0.100 −3.236 −1.333 3.526 1.084 −1.798 −0.783 
 (0.648) (2.026) (1.967) (3.448) (3.627) (3.057) (2.978) 
Longitude −0.652** −1.299* −1.364** −0.642 −0.390 −0.225 0.655 
 (0.322) (0.672) (0.659) (0.910) (0.824) (0.872) (0.471) 
Distance to Moscow −3.724*** −3.360*** −2.887** −2.772 −1.485 −0.452 1.848 
 (0.849) (0.844) (1.147) (2.558) (2.156) (2.241) (1.152) 
Cereal Suitability 4.471** 3.940** 2.267* 4.005* −4.628* 3.254 1.975 
 (1.725) (1.955) (1.166) (2.242) (2.400) (2.188) (1.181) 
Distance to Coast  2.203** 1.735 0.749 2.595 −1.880 −0.816 
  (0.955) (1.218) (1.819) (1.780) (2.448) (2.125) 
Distance City in 1600  12.718 7.397 −8.723 −20.344 −10.379 9.727 
  (13.986) (13.063) (18.656) (15.759) (25.314) (20.260) 
Distance Provincial Capital  1.005 −0.092 0.983 0.008 0.026 −1.050 
  (1.460) (1.217) (1.599) (1.609) (1.159) (0.973) 
Additional geography  ✓ ✓ ✓ ✓ ✓ ✓ 
Fixed effects   Province Province Province Province Province 
R2 0.37 0.46 0.71 0.78 0.72 0.80 0.38 
Observations 490 490 490 185 472 472 490 
Number of clusters 50 50 50 45 49 49 50 
F-statistic joint significance 21.94 9.72 2.27 2.19 1.58 2.19 1.83 
p-value joint significance 0.00 0.00 0.02 0.03 0.13 0.02 0.06 

The unit of observation is the district. The dependent variable in columns 1 to 4 is the share of serfs in a district population, c. 1858. For columns 5 to 7, the dependent variable is the share of such serfs in the total number of serfs. Additional geographic controls are forest cover, ruggedness, river density, mean temperature, mean precipitation, and the share of podzol soils. Heteroskedastic-robust standard errors in parentheses, clustered at the province. *p<0.10, **p<0.05, and ***p<0.01.

As Muscovy expanded away from Moscow before 1700, state service was often rewarded with the allocation of land in newly incorporated areas. Therefore, we consider the direct distance from each district centroid to Moscow and historical provincial fixed effects to get at the broad geopolitical nature of this expansionary process.17 Apart from this historical process, variation in land productivity might have led to differences in the demand for coerced labor or in the desirability of land in return for state service. An important proxy for agricultural productivity is the quality of the soil for growing crops. As grains were dominant in the agriculturally productive areas to the south of Moscow, we use modern geospatial data to produce a time-invariant measure of the land's suitability for growing cereals.18 Other environmental and geographic conditions might have affected local agricultural productivity, the mobility of the population, and local nonagricultural opportunities (and, hence, outside options and the incentives for maintaining serfdom, as in Acemoglu and Wolitzky, 2011). Therefore, we also include the latitude and longitude of each district's centroid, the fraction of land covered with forest, the share of podzol soil (relatively poor for agriculture), the slope of the terrain, distance to the coast, the density of rivers in the district, and temperature and precipitation in the growing season (averaged over the period 1901 to 2000).19 Together, these variables constitute the base set of geographic controls for our analyses.

Table 1 provides results from our investigation of the distribution of serfdom across the European part of Imperial Russia. In a cross-sectional specification focusing on location and grain suitability (column 1), the coefficients on longitude and distance to Moscow are negative and statistically significant, consistent with the Muscovite expansion from west to east influencing the eventual extent of serfdom. However, as we illustrate empirically below, controlling for the distance to Moscow does not explain away the relationship between historical serfdom and modern outcomes.20 The suitability for growing cereals is also a strong and positive predictor of serfdom's intensity in column 1, which is consistent with the incentive to employ coerced labor in relatively agriculturally productive areas.

Column 2 adds the rest of the geographic variables, as well as the distance of a district to the nearest city in 1600 (reported in the data of Bairoch, Batou, & Chèvre, 1988) and the distance to the district in which the capital of the province is located.21 These variables take into account “preexisting” urbanization as a measure of past economic development, since districts in close proximity to cities and provincial capitals were likely characterized by higher population densities. In the absence of suitable early data, these measures also help account for the incentive to adopt serfdom in areas with high land-labor ratios (Domar, 1970). Moreover, the distance to a city is also indicative of the availability of noncoercive outside options for the serf population.22 However, neither variable is a significant predictor of serfdom. To further take into account unobserved historical and geographic determinants of serfdom as noted above, column 3 includes provincial fixed effects (defined for Imperial guberniia). While the sizes of the coefficients on the main variables in column 1 remain similar, we find that a district's province explains a large part of serfdom's intensity. Moving from the cross-district specification in column 2 to the provincial fixed-effect model of column 3 increases the R2 from 0.46 to 0.71, while soaking up much of the impact of other geographic variables. Thus, much of the spatial variation in serfdom was determined at the broader regional level.

Column 4 estimates the same regression as in column 3 over the districts for which our modern household survey data (see below) are available. We find similar balance in terms of the covariates considered, with the exception of a positive association between serfdom and cereal suitability. An even larger share of the variation of serfdom in this subsample can be explained by our geographic controls and province fixed effects (R2 of 0.78). Columns 5 to 7 investigate the share of different types of serfs. The estimates indicate that the quitrent form of obligations (obrok) was relatively less prominent in areas that were more suitable for cereal agriculture, that corvée (barshchina) areas were more riverine, and that nonpeasant (household) serfs were located in less fertile regions. With provincial fixed effects included, the other variables explain relatively little of the overall variation in the type of serfdom.

Overall, we do not find any strong association of these covariates with serfdom, once we control for province dummies that subsume many relevant geographic and historical characteristics.23 These results help mitigate concerns that the historical emergence of serfdom in Russia was related to unobservable factors that could bias our empirical estimates of the long-run development effects of coercive labor. While our data are indicative of balance in observable characteristics between areas with more and fewer serf areas, in our empirical work below, we do control for various fixed effects and a baseline set of possible geographic confounders, particularly the distance of a district to Moscow.

III. Documenting the Long-Run Impact of Russian Serfdom

A. Data

Constructing outcomes for our long-run investigation is challenging because income per capita is not available at a unit of analysis comparable to our historical data on serfdom, and because our analysis spans several current countries. To circumvent these limitations, we construct our main outcome variables from the three waves of the Life in Transition Survey (LiTS).24 Our main indicator for modern economic development is equivalent per capita household expenditure, defined as the sum of spending on food, clothing, education, health, and durables (in USD) and adjusted for the size of the household.25 In addition to our main outcome, we draw on the LiTS to document consumer good ownership (mobile phone, car, computer) the importance of farming and land cultivation, education, public goods provision, and preferences. The geolocation of each primary sampling unit (PSU) allows us to precisely match households to historical districts.26

B. Baseline Empirical Strategy

To assess whether the historical incidence of serfdom was associated with modern socioeconomic outcomes, we estimate the following model:
log(Expenditure)i,d,p,c=α+βSerfdomd,p,c+Hi,d,p,c'λ+Xd,p,c'δ+Γp,c+εi,d,p,c,
(1)

where i represents the household, d refers to the historical district, p indicates the historical province, and c contemporary country. Serfdomd,p,c denotes our variable of concern, the share of serfs (from 0 to 1) out of the total population in a (historical) district d, located in province p, and contemporary country c. The coefficient of interest is β, which gives the effect of serfdom on modern outcomes. Hi,d,p,c is a vector of household and survey controls that includes household size, the share of the household aged 0 to 18, the share aged 60 and over, the share of males in the household, the household head's religion, and indicators for LiTS waves. Xd,p,c is a vector of the district-level controls that we link to the PSUs.27 Our preferred specification incorporates a subset of these baseline district characteristics in a more flexible way by including a set of eight dummies for each class of cereal suitability; quartile dummies for river density, temperature, podzol soil, and precipitation; and linear controls for the remaining variables.

Our measure of serfdom is spatially correlated. In most specifications, we include fixed effects for administrative units, denoted by Γp,c, which can be modern countries or historically provinces.28 To account for spatial correlation, we use a conservative approach and cluster either at the level of the province (to account for correlation within a province) or compute Conley (1999) standard errors.

Table 2.
Estimating the Long-Run Effects of Serfdom
(ln) Equivalent Expenditures per Capita
(1)(2)(3)(4)(5)(6)
Serfs % (1858) −0.373*** −0.431*** −0.379*** −0.677*** −0.694*** −0.644*** 
 (0.117) (0.111) (0.104) (0.185) (0.190) (0.185) 
Distance City in 1600   20.544   −54.487 
   (21.718)   (40.700) 
Distance Provincial Capital   −0.062   −0.055 
   (0.038)   (0.045) 
Household controls ✓ ✓ ✓ ✓ ✓ ✓ 
Linear controls ✓   ✓   
Flexible controls  ✓ ✓  ✓ ✓ 
Fixed effects Country Country Country Province Province Province 
Observations 17,155 17,155 17,155 17,155 17,155 17,155 
R2 0.39 0.40 0.41 0.40 0.41 0.41 
Number of clusters 45 45 45 45 45 45 
δ for β=0 16.126 9.856 2.486 2.772 1.518 1.166 
Lower bound estimates −0.424 −0.552 −0.432 −0.591 −0.639 −0.517 
Conley S.E. 300 km       
Serfs % (1858) [0.118]*** [0.126]*** [0.123]*** [0.196]*** [0.195]*** [0.180]*** 
(ln) Equivalent Expenditures per Capita
(1)(2)(3)(4)(5)(6)
Serfs % (1858) −0.373*** −0.431*** −0.379*** −0.677*** −0.694*** −0.644*** 
 (0.117) (0.111) (0.104) (0.185) (0.190) (0.185) 
Distance City in 1600   20.544   −54.487 
   (21.718)   (40.700) 
Distance Provincial Capital   −0.062   −0.055 
   (0.038)   (0.045) 
Household controls ✓ ✓ ✓ ✓ ✓ ✓ 
Linear controls ✓   ✓   
Flexible controls  ✓ ✓  ✓ ✓ 
Fixed effects Country Country Country Province Province Province 
Observations 17,155 17,155 17,155 17,155 17,155 17,155 
R2 0.39 0.40 0.41 0.40 0.41 0.41 
Number of clusters 45 45 45 45 45 45 
δ for β=0 16.126 9.856 2.486 2.772 1.518 1.166 
Lower bound estimates −0.424 −0.552 −0.432 −0.591 −0.639 −0.517 
Conley S.E. 300 km       
Serfs % (1858) [0.118]*** [0.126]*** [0.123]*** [0.196]*** [0.195]*** [0.180]*** 

The unit of observation is the household. Household controls include the household size, the share of household members aged 0 to 18, the share of household members aged 60 and over, the share of male household members, the religious denomination of the household respondent, and LiTS wave fixed effects. Linear controls include latitude and longitude of the district, the area covered by forest, ruggedness, cereal suitability, growing-season temperature and precipitation, river density, share of podzol soil, the distance to the coast, and the distance to Moscow. Flexible controls include eight dummies for cereal suitability and four dummies for quartiles of growing season temperature, growing-season precipitation, the share of podzol soil, and river density, as well as the remaining linear controls. The restricted model used to compute δ and the lower-bound estimates controls for country/province fixed effects. Standard errors clustered at the province are in parentheses. *p<0.10, **p<0.05, and ***p<0.01.

C. Results

We present our main results in table 2.29 The estimates from equation (1) with the log of household expenditure as the dependent variable are reported under different strategies regarding the use of fixed effects and controls. Overall, we find a large, negative, and statistically significant relationship between serfdom's intensity and our main measure of economic well-being, conditional on household controls, base geographic controls, and fixed effects. The estimated coefficient is negative and significant with either country or provincial fixed effects and with fully linear or more flexible versions of the geoclimatic controls. In columns 3 and 6, we add our controls for pre-1861 economic development: the distance to the nearest city of more than 5,000 inhabitants in 1600 and the distance to the provincial capital. The coefficient decreases slightly in absolute terms but stays significant.

Overall, these estimates are economically meaningful. A 1 standard deviation increase in the prevalence of serfdom (around 25 percentage points or 0.25 here) is associated with a substantially lower level of per capita expenditure in the modern data of between 9% and 17%. This finding is robust to the way we control for geography, to the type of administrative unit fixed effects, and to taking into account spatial correlation of errors with a cutoff distance of 300 km (reported at the bottom of table 2).30

D. Assessing Selection on Unobservables

The negative effects of serfdom on contemporary development are robust to an exhaustive set of controls and fixed effects that together explain about 78% of variation in the main independent variable (see table 1). Nevertheless, it is possible that unobservables bias our estimates. To assess the scale of any such bias, we employ the methodology of Oster (2019) to evaluate how strong selection on unobservables has to be to explain away the negative coefficient. The method studies coefficient stability by comparing estimated coefficients and R2's in models with full controls relative to a model with a restricted set of controls.31 At the bottom of table 2, we present both the δ estimate of the proportional bias due to unobservables that would have to exist to drive the coefficient of serfdom to 0, along with a lower-bound coefficient estimate of the impact of serfdom under equal selection. Both calculations assume a maximal R2 that is 30% larger than the R2 from the controlled regression, as Oster (2019) suggested. A δ equal to 1 means selection on unobservables would have an equal impact on the coefficient estimate as selection on observables, and so values exceeding 1 imply that selection on unobservables must be significantly stronger than selection on observables to explain away our result. Indeed, the δ values that we compute are consistently larger than 1. The implied lower-bound estimates are negative and of large magnitudes that are economically significant. These findings imply that a bias of our estimates by unobservables is unlikely and suggest a causal interpretation of the effect of serfdom on contemporary development.32

E. Robustness and Extensions

We undertook a series of robustness exercises for our main estimates, reported in table C3 of the appendix. These control for additional geographic determinants of sectoral productivity or economic development—in particular, the within-district variation in land quality, differences in the length of the growing period, climatic risk (i.e., the year-to-year variability during the growing-season months), the presence of coal deposits measured during the Soviet period, the distance from St. Petersburg, and pre-emancipation population density.33 We find only weak relationships between these variables and long-run outcomes, along with stable coefficients on our serfdom measure, suggesting that the baseline controls and fixed effects already absorb the most important geographic factors. To alleviate concerns about other factors possibly driving persistence, in particular religious differences, we also control for the district-level share of adherents to major religions in 1870, in addition to the religious affiliations of respondents today. This does not affect the main result. Finally, we show that our finding is unchanged if we employ ownership of various durable consumer goods (mobile phone, car, and computer) in the household as an outcome: there is a negative and significant relationship between the historical incidence of serfdom and this modern measure of household wealth.34

Another way to investigate the long-run impact of serfdom is to differentiate the effects of observable characteristics on economic outcomes in areas where peasants were or were not subjected to the institution. We report the results of such an exercise in appendix table C9.35 If we consider only provinces without serfdom in 1861, land suitability for cereals, wheat, rye, barley, and oat shows the expected positive (and statistically significant) correlation with modern per capita expenditures. If one considers the remaining provinces of Imperial Russia, the coefficients of grain suitability turn negative. With lower labor coercion, a greater share of suitable land would be conducive to economic development for many reasons, even if the agricultural sector was lagging (as has been the case in the post-Soviet period). However, in areas where Russian serfdom existed prior to 1861, the positive effects of land quality on long-run economic outcomes are limited by persistent effects of labor coercion, since serfdom was more prevalent on more productive land. Overall, while the nonserf provinces are admittedly a small group, this evidence on the differential long-run effects of agricultural suitability is highly suggestive that a legacy of serfdom gave rise to persistent constraints on subsequent Russian economic development.

F. Heterogeneous Effects by Type of Serfdom

Did local heterogeneity in serfdom matter for long-run outcomes? Employing data from the late 1850s (see the appendix for details), we differentiate between the share of serfs required to pay only quitrent in cash or kind to their landlords (obrok) and the share with at least some labor obligations (corvée or barshchina). Consistent with their greater prevalence in less fertile regions (table 1), the historical literature emphasizes that serfs on obrok generated more of their income from nonagricultural activities, from factory wage work to self-employment. The resulting relative autonomy of economic decision making among obrok serfs could have led to more favorable long-term outcomes compared to the conditions faced by serfs subject to the more directly coercive barshchina.

Table 3 presents regression results in which we include the population shares of quitrent, corvée, and household serfs (the residual category is the remaining population). The standardized coefficients in columns 1 and 2 suggest that the negative effects of serfdom on contemporary household expenditure and consumer good ownership are more pronounced in areas with a larger share of corvée serfs and attenuated where there was a larger share of serfs on quitrent. Columns 3 to 6 document that areas with more serfdom are, on average, more agricultural, but this effect is driven by corvée serfs. This is consistent with quitrent serfs maintaining relatively more economic autonomy before (and after) 1861, which translated over time into a greater degree of transition from agriculture in those areas. Thus, these heterogeneous effects are suggestive of structural change as a mechanism underlying the persistent impact of serfdom, a possibility that we explore further below.

IV. Tracing the Persistence of Serfdom's Effects

The previous section documented a long-run association between serfdom and economic outcomes today. To understand the mechanisms behind this relationship, it is crucial to identify when formerly serf areas fell behind and whether there are processes of divergence or convergence over time across historical districts. Unfortunately, generating a consistent indicator of economic development over the entire period is complicated by the changes in regimes and the lack of disaggregate data. We therefore focus on a sample of cities for which we can follow their population during the nineteenth and twentieth centuries. City population as a measure of economic activity has been used extensively in the development, urban, and history literatures (Glaeser, Scheinkman, & Shleifer, 1995).36 This measure allows us to estimate the effect of serfdom before and after emancipation in cross-sectional and panel frameworks.

Table 3.
Heterogeneity in Long-Run Outcomes by Type of Serfdom
(ln) Equivalent Expenditures per CapitaConsumer GoodsSale Farm ProductsLand Cultivation
(1)(2)(3)(4)(5)(6)
Serfs % (1858)   0.166***  0.427***  
   (0.061)  (0.152)  
Corvée % (1858) −0.126*** −0.114***  0.053***  0.097** 
 (0.037) (0.032)  (0.014)  (0.046) 
Quitrent % (1858) −0.073** −0.002  0.018**  0.018 
 (0.034) (0.021)  (0.008)  (0.011) 
Household Serfs % (1858) 0.003 −0.027  −0.001  0.049* 
 (0.042) (0.027)  (0.012)  (0.025) 
H0: Corvée = Quitrent (p-value) 0.20 0.00  0.01  0.09 
Household controls ✓ ✓ ✓ ✓ ✓ ✓ 
Flexible controls ✓ ✓ ✓ ✓ ✓ ✓ 
Distances: City and provincial capital ✓ ✓ ✓ ✓ ✓ ✓ 
Fixed Effects Province Province Province Province Province Province 
Observations 14,736 18,609 13,011 11,196 6,171 5,291 
R2 0.38 0.43 0.08 0.09 0.18 0.19 
Number of clusters 44 44 45 44 38 35 
(ln) Equivalent Expenditures per CapitaConsumer GoodsSale Farm ProductsLand Cultivation
(1)(2)(3)(4)(5)(6)
Serfs % (1858)   0.166***  0.427***  
   (0.061)  (0.152)  
Corvée % (1858) −0.126*** −0.114***  0.053***  0.097** 
 (0.037) (0.032)  (0.014)  (0.046) 
Quitrent % (1858) −0.073** −0.002  0.018**  0.018 
 (0.034) (0.021)  (0.008)  (0.011) 
Household Serfs % (1858) 0.003 −0.027  −0.001  0.049* 
 (0.042) (0.027)  (0.012)  (0.025) 
H0: Corvée = Quitrent (p-value) 0.20 0.00  0.01  0.09 
Household controls ✓ ✓ ✓ ✓ ✓ ✓ 
Flexible controls ✓ ✓ ✓ ✓ ✓ ✓ 
Distances: City and provincial capital ✓ ✓ ✓ ✓ ✓ ✓ 
Fixed Effects Province Province Province Province Province Province 
Observations 14,736 18,609 13,011 11,196 6,171 5,291 
R2 0.38 0.43 0.08 0.09 0.18 0.19 
Number of clusters 44 44 45 44 38 35 

The unit of observation is the household. Corvée, Quitrent, and Household Serf shares are standardized (mean = 0, SD = 1). See the appendix for further information. Household controls include the household size, share of household members aged 0 to 18, share of household members aged 60 and over, share of male household members, religious denomination of the household respondent, LiTS Survey Wave fixed effects. Flexible controls include eight dummies for cereal suitability and four dummies for quartiles of growing-season temperature, growing-season precipitation, the share of podzol soil, and river density, as well as the remaining linear controls. Distances are the distance to the nearest city in 1600 and the distance to the provincial capital. Sample sizes vary due to the number of LiTS waves reporting the dependent variables. Standard errors clustered at the province in parentheses. *p<0.10, **p<0.05, and ***p<0.01.

We begin with a cross-section of 366 cities.37 After locating these cities in our historical districts d and provinces p, we regress the log population of city i in each year on the measure of serfdom and our standard controls, including variables that proxy for the pre-1861 level of economic development. We therefore estimate regressions of the form
log(Population)i,d,p=α+βSerfdomd,p+Xd,p'δ+Γp+εi,d,p.
(2)
The estimated coefficients on historical serfdom from this repeated cross-sectional exercise are plotted in figure 2 (see appendix table D1 for the regression results). We find a negative association between city population and the incidence of historical serfdom in the surrounding district for every year. Increasing serfdom by 1 standard deviation was associated with 25% to 38% lower city population on average.38 The magnitude of the coefficient is slightly smaller in 1939 before becoming larger in the later years. While this pattern could be due to urban-biased Soviet policies interacting with initial conditions (see below), the consistently negative coefficient is in line with longer-run processes originating in the Imperial period. Once we control for population in 1897, the coefficient on serfdom becomes insignificant, while the R2 jumps from 0.31 to 0.67 (column 8, appendix table D1). Thus, about 36% of the variation in city population in 2002 is explained by the distribution of population at the end of the Imperial period. Serfdom affected the Imperial spatial economic equilibrium, which then persisted and was even reinforced through the Soviet era.
Figure 2.

Serfdom and City Population, 1897–2002

This figure plots the coefficient from regressions of (log) city population on serfdom, conditional on province fixed effects and flexible controls. See appendix table D1 for the corresponding regression results.

Figure 2.

Serfdom and City Population, 1897–2002

This figure plots the coefficient from regressions of (log) city population on serfdom, conditional on province fixed effects and flexible controls. See appendix table D1 for the corresponding regression results.

Despite the large number of controls and the consistency of the estimates, the cross-sectional results may be subject to residual unobserved factors. Therefore, we also studied a panel of cities observed over the entire period 1800 to 2002 (a subset of the cities considered above) and estimate variants of the following model, where i denotes cities, d districts, and t time:
log(Population)i,d,t=αi+γt+β(Serfdomd×Post1861)+(Xd'×Post1861)δ+εi,d,t.
(3)

If cities in districts with a relatively high incidence of serfdom experienced catch-up growth after emancipation, we would expect the coefficient β to be positive and significant. In the case of the persistence of initial conditions under serfdom, β should be essentially 0, and if serf areas were falling further behind over time after 1861, β would be negative. The advantage of this specification is that it allows us to control for time fixed effects, γt, and for time-invariant but unobservable characteristics of cities, αi (since each district contains only one city, city fixed effects are equivalent to district fixed effects) that were potentially associated with the strength of historical serfdom and influenced economic development in the long run. We allow for time-variant effects of the fixed observable district characteristics, Xd×Post1861, and cluster standard errors at the level of the city.

The results reported in appendix table D4 indicate little catch-up in former serf areas in terms of city growth. Considering the full sample (1800–2002), we estimate a negative coefficient that is, however, not significantly different from 0 once we control for geographic and economic controls interacted with the post-emancipation dummy. When we restrict the sample to the years 1800, 1850, and 1897, we find a positive coefficient that is very small and statistically insignificant. Furthermore, the widening of the economic gap during the Soviet period that we saw in the cross-sectional results is confirmed by the negative interaction between serfdom and the Soviet dummy.39 Overall, the evidence presented in this section indicates the persistence of initial conditions, which is inconsistent with a process of convergence in economic development across localities with different exposure to serfdom.40

V. Mechanisms

This leads us to the essential question: How did historical Russian serfdom generate persistent impediments to economic development? Our findings document a negative association with development outcomes, but we have also shown that the strength of this relationship varied over time, with some escalation from the 1930s. This section considers several possible mechanisms: impediments to structural change and local agglomeration, human capital, inequality and public goods provision, and cultural differences. Overall, the evidence suggests that slower structural change and less urban agglomeration generated the persistent pattern of spatial inequality in economic outcomes.

A. Structural Change, Urban Development, and Local (Dis-)Agglomeration

Russian serfdom may have had a negative impact on subsequent well-being through the generation of persistent constraints on the interconnected processes of urbanization and structural change (i.e., the transfer of factors from relatively low-productivity agriculture to higher-productivity industry). As Bleakley and Lin (2012), Davis and Weinstein (2002), Michaels and Rauch (2016), and others argued, geographic factors and historical institutions (such as serfdom) can generate initial spatial patterns in economic activity or urbanization that, through the forces of economic diversification, increasing returns, and (dis-)agglomeration, give rise to path dependencies in urban and industrial development over the long run. In this sense, the pre-1861 distribution of Russian serfdom may have generated local and long-lasting effects on urbanization, industrial growth, infrastructure, and labor productivity, as well as knock-on consequences for inequality and the demand for human capital.

The relatively more burdensome emancipation settlements and land reforms experienced by the former serfs likely made their experience of the post-1861 institutional regime, centered on the communal ownership of property and collective liabilities for taxes and land payments, more constrained than among other peasants. Critically, this occurred within the context of relatively high transportation costs and an Imperial internal passport system that imposed costs on migration out of the countryside and to more distant employment opportunities.41 These frictions plausibly perpetuated initial conditions by constraining former serfs when it came to moving off the farm into urban settings or industrial employment opportunities.42 Such relative immobility may have also disincentivized local human capital and technological development, with implications for the path dependency of early differences. Thus, while there was labor migration in the period and some growth in urbanization and industry was evident by the end of the nineteenth century, former serf areas may have participated less in these processes.

Table 4.
Structural Change and Urbanization
Pre-SovietSovietPost-Soviet
Urbanization RateFactories per 1,000 People, 1868Log Production per Worker, 1868Road DensityGULAGLog Population Density, 2000Log Light Density, 2008
18631913
(1)(2)(3)(4)(5)(6)(7)(8)
Serfs % (1858) −15.559*** −15.618*** −50.056 −0.651* −0.009*** −0.296** −1.001*** −0.820*** 
 (5.051) (5.524) (34.049) (0.338) (0.003) (0.124) (0.339) (0.295) 
Flexible controls ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 
Distances: City and provincial capital ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 
Fixed effects Province Province Province Province Province Province Province Province 
Observations 483 490 483 434 490 490 490 490 
R2 0.39 0.39 0.35 0.33 0.54 0.40 0.64 0.57 
Number of clusters 50 50 50 50 50 50 50 50 
δ for β=0 5.544 4.818 4.400 5.786 32.208 2.376 19.789 17.248 
Lower-bound estimates −14.851 −14.146 −49.582 −0.652 −0.010 −0.243 −1.094 −0.888 
Conley S.E. 300 km         
Serfs % (1858) [4.828]*** [5.049]*** [33.099] [0.264]** [0.003]*** [0.136]** [0.350]*** [0.303]*** 
Pre-SovietSovietPost-Soviet
Urbanization RateFactories per 1,000 People, 1868Log Production per Worker, 1868Road DensityGULAGLog Population Density, 2000Log Light Density, 2008
18631913
(1)(2)(3)(4)(5)(6)(7)(8)
Serfs % (1858) −15.559*** −15.618*** −50.056 −0.651* −0.009*** −0.296** −1.001*** −0.820*** 
 (5.051) (5.524) (34.049) (0.338) (0.003) (0.124) (0.339) (0.295) 
Flexible controls ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 
Distances: City and provincial capital ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 
Fixed effects Province Province Province Province Province Province Province Province 
Observations 483 490 483 434 490 490 490 490 
R2 0.39 0.39 0.35 0.33 0.54 0.40 0.64 0.57 
Number of clusters 50 50 50 50 50 50 50 50 
δ for β=0 5.544 4.818 4.400 5.786 32.208 2.376 19.789 17.248 
Lower-bound estimates −14.851 −14.146 −49.582 −0.652 −0.010 −0.243 −1.094 −0.888 
Conley S.E. 300 km         
Serfs % (1858) [4.828]*** [5.049]*** [33.099] [0.264]** [0.003]*** [0.136]** [0.350]*** [0.303]*** 

The unit of observation is a district. Flexible controls include eight dummies for cereal suitability and four dummies for quartiles of growing season temperature, growing-season precipitation, the share of podzol soil, and river density, as well as linear controls of latitude and longitude of the district, the area covered by forest, ruggedness, the distance to the coast, and the distance to Moscow. Distances are the distance to the nearest city in 1600 and the distance to the provincial capital. The restricted model used to compute δ and the lower-bound estimates controls for province fixed effects. Standard errors clustered at the province in parentheses. *p<0.10, **p<0.05, and ***p<0.01.

Were such structural frictions reinforced or even strengthened by Soviet policies? At first glance, the massive population movements and institutional reforms of the Communist period would seem to preclude any sort of persistence.43 However, several features of Soviet society likely contributed toward persistence of the prior spatial pattern of economic development. The locations and sizes of GULAG camps, “special cities,” and many Soviet industrial centers were often driven by noneconomic concerns that may have reinforced existing spatial patterns.44 Moreover, political whims, planning objectives, and nonmarket mechanisms for allocating goods, capital, and land inhibited spatial arbitrage and subsequent convergence.

While the New Economic Policy in the 1920s saw a general abandonment of Imperial passport restrictions, Soviet authorities quickly adopted a draconian system of internal passports (propiski) and residency restrictions to control the allocation of labor and limit social unrest (Buckley, 1995; Kessler, 2001).45 Even when and where mobility was possible, the shortfall of housing and other disamenities of Soviet urban life likely generated additional frictions in the allocation of labor across sectors. Housing inadequacies, internal passports, and various residency restrictions continued to generate impediments to labor mobility after 1991 in the Russian Federation, Ukraine, and Belarus.46

These impediments to factor mobility from serfdom onward likely translated existing differences into persistent productivity gaps across space, which would have limited agglomeration economies and constrained urban development in the negatively affected areas. While the historical literature suggests the plausibility of this set of mechanisms, we can go further in documenting its validity using novel empirical evidence.

B. Documenting Serfdom's Constraints on Structural Change and Urbanization

As table 4 shows, historical serfdom was strongly associated with lower rates of urbanization (as opposed to city size) before the Revolution (columns 1 and 2). The reduction in the 1913 urbanization rate of about 3.8 percentage points implied by a standard deviation increase in serfdom is a large effect, given a mean of 10.1 and a standard deviation of 12.2 for the former. Columns 3 and 4 investigate industrial production using newly digitized district-level data from just after emancipation. We find a negative, albeit not statistically significant, association between serfdom and the number of firms per capita, but when we divide the ruble value of factory turnover in a district by the number of firms or factory workers, we find that worker productivity was significantly lower in areas with higher levels of serfdom. A 1 standard deviation increase in serfdom corresponded to about 16% lower industrial productivity.47

Table 5.
Structural Change, 1897 Employment, and the Heterogeneous Effects of Serfdom
Primary Employment 1897Secondary Employment 1897Industry Employment 1897Log Light Density, 2008
(1)(2)(3)(4)(5)(6)(7)(8)
Serfs % (1858) 0.077*  −0.008  −0.024  −0.820***  
 (0.043)  (0.023)  (0.022)  (0.295)  
Corvée % (1858)  0.035**  −0.016*  −0.016*  −0.240*** 
  (0.014)  (0.009)  (0.008)  (0.081) 
Quitrent % (1858)  −0.015  0.026***  0.018**  −0.020 
  (0.010)  (0.008)  (0.008)  (0.080) 
Household Serfs % (1858)  0.006  −0.008  −0.009*  −0.102 
  (0.012)  (0.006)  (0.005)  (0.091) 
H0: Corvée = Quitrent (p-value)  0.00  0.00  0.01  0.02 
Flexible controls ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 
Distances: City and provincial capital ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 
Fixed effects Province Province Province Province Province Province Province Province 
Observations 490 468 490 468 490 468 490 468 
R2 0.51 0.54 0.60 0.64 0.57 0.60 0.57 0.59 
Number of clusters 50 49 50 49 50 49 50 49 
Primary Employment 1897Secondary Employment 1897Industry Employment 1897Log Light Density, 2008
(1)(2)(3)(4)(5)(6)(7)(8)
Serfs % (1858) 0.077*  −0.008  −0.024  −0.820***  
 (0.043)  (0.023)  (0.022)  (0.295)  
Corvée % (1858)  0.035**  −0.016*  −0.016*  −0.240*** 
  (0.014)  (0.009)  (0.008)  (0.081) 
Quitrent % (1858)  −0.015  0.026***  0.018**  −0.020 
  (0.010)  (0.008)  (0.008)  (0.080) 
Household Serfs % (1858)  0.006  −0.008  −0.009*  −0.102 
  (0.012)  (0.006)  (0.005)  (0.091) 
H0: Corvée = Quitrent (p-value)  0.00  0.00  0.01  0.02 
Flexible controls ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 
Distances: City and provincial capital ✓ ✓ ✓ ✓ ✓ ✓ ✓ ✓ 
Fixed effects Province Province Province Province Province Province Province Province 
Observations 490 468 490 468 490 468 490 468 
R2 0.51 0.54 0.60 0.64 0.57 0.60 0.57 0.59 
Number of clusters 50 49 50 49 50 49 50 49 

The unit of observation is a district. Corvée, Quitrent, and Household Serfs are standardized variables (mean = 0, SD = 1). Flexible controls include eight dummies for cereal suitability and four dummies for quartiles of growing-season temperature, growing-season precipitation, the share of podzol soil, and river density, as well as linear controls of latitude and longitude of the district, the area covered by forest, ruggedness, the distance to the coast, and the distance to Moscow. Distances are the distance to the nearest city in 1600 and the distance to the provincial capital. Standard errors clustered at the province in parentheses. *p<0.10, **p<0.05, and ***p<0.01.

Table 4 also reports the estimated relationships between serfdom and two indicators of Soviet place-based policies: road density and the location of GULAG camps. We find that road density shortly after 1991 was more limited in former serf areas (column 5). This was likely both an effect of, and a contributing factor toward, the slower pace of structural change in such districts. At the same time, several authors have argued for the positive local economic impact of the GULAG camps through employment or productivity channels (Gregory & Lazarev, 2003), both in remote regions and in close proximity to already urbanized localities. Column 6 documents a lower incidence of camps where serfdom was more prevalent, suggesting that this possible source of factor redistribution does not drive our findings. Finally, columns 7 and 8 employ other measures of modern economic development as outcomes: population density in 2000 and satellite light intensity in 2008. The latter variable can be seen as an indicator of structural change, since industry tends to generate much more nighttime illumination than agriculture. Indeed, we find that modern population density and light density are much lower in former serf areas.48

Overall, the results in table 4 are suggestive that a legacy of serfdom may have constrained labor and population mobility into more economically dynamic sectors through the late-Imperial, Soviet, and post-Soviet periods. We can turn to additional data to further shed light on this mechanism. In table 5, we examine the occupational structure of the late Imperial period as another way of documenting structural change.49 Relating sectoral employment shares to our measure of historical serfdom, we find a positive and significant difference in primary employment (column 1), and negative, but not significant, differences in secondary or industrial employment (columns 3 and 5). These average effects mask heterogeneity by the types of serfdom: areas with more serfs on quitrent, who paid their obligations in cash or kind, had a lower share of the population working in agriculture and had larger employment in the secondary and industrial sector at the end of the nineteenth century (columns 2, 4, and 6).50 Column 8 shows that the light intensity result from table 4 stems from particular structural constraints in formerly corvée areas. Overall, we find that districts with relatively more quitrent serfs (with more economic autonomy) saw a greater shift toward higher productivity, nonagricultural activities and locations, with long-run consequences for the level of income today (as in table 3). Therefore, these results provide additional evidence on the role played by impediments to structural change and urbanization as a key channel underlying serfdom's persistent impact.51
Figure 3.

Serfdom and the Location of Soviet Industry

This figure plots the coefficient from regressions of the number of defense establishments per city on serfdom, conditional on province fixed effects and flexible controls. See appendix table E1 for the corresponding regression results.

Figure 3.

Serfdom and the Location of Soviet Industry

This figure plots the coefficient from regressions of the number of defense establishments per city on serfdom, conditional on province fixed effects and flexible controls. See appendix table E1 for the corresponding regression results.

Did industrial production during the Soviet period reflect the same patterns? We first consider the number of Soviet defense factories in a sample of Russian and Ukrainian cities, as compiled by Acemoglu, Hassan, and Robinson (2011) using an early version of the data from Dexter and Rodionov (2016). We observe the number of such factories at six points in time: 1939, 1945, 1959, 1970, 1979, and 1989. While defense factories in a command economy were certainly not allocated through market mechanisms, the geographic variation in this type of establishment could be indicative of structural transformation if Soviet authorities made location decisions to take advantage of preexisting or complementary industrial activities.52

The estimated cross-sectional coefficients plotted in figure 3 show a statistically significant and increasingly negative relationship between historical serfdom and the number of defense plants in observed cities (see appendix table E1 for regression results).53 The fact that coefficients increase over time is consistent with localized agglomeration economies that accelerated the allocation of firms away from areas with greater levels of historical serfdom. A 1 standard deviation increase in historical serfdom results in 3.75 fewer defense firms in 1989 (or 0.22 of a standard deviation). When we include the “initial” level of defense production in 1939, we find no residual effect of historical serfdom on firms in 1989.54 Similar to the results on city populations, this implies that the structural impact of serfdom's legacy was present prior to World War II and then persisted over subsequent decades.55

While defense plant data offer a consistent indicator of industrial activity over the Soviet period, we also study the location and number of nonmilitary firms from the 1989 Soviet Census of Manufacturers. As we show in the appendix (column 8 of table E1), a 1 standard deviation increase in historical serfdom translates into two fewer firms in a district in 1989 (or 0.11 of a standard deviation). Besides the aggregate number of firms, these data allow us to investigate differences in characteristics across the approximately 14,000 Russian firms observed in that year. Table 6 displays these firm-level regression results. Consistent with persistent constraints on structural change (and agglomeration), we find that 1989 firms in areas where serfdom was historically important were (a) more likely to be in the agricultural sector and less likely to be in manufacturing; (b) employed fewer people; (c) had smaller turnover; and (d) were less productive.

C. Alternative Mechanism I: Human Capital

The implications of coercive labor institutions for long-run human capital accumulation have been cited as another mechanism behind historical persistence. Lower health or education investments under historical institutions can persist through intergenerational mechanisms, local preferences, or supply-side factors (perhaps driven by an unequal political structure; see below).56 At the same time, low levels of human capital could be fostered by slower structural change (Rocha, Ferraz, & Soares, 2017).57 To identify whether human capital mechanisms were an independent source of historical persistence, we study schooling and literacy before and after the Soviet period (equivalent data at the relevant unit of observation are not available for the Soviet period) using a variety of novel sources.

Table 6.
A Firm-Level Analysis for 1989
AgricultureManufacturing(log) Employment(log) Turnover(log) Turnover per Worker
(1)(2)(3)(4)(5)
Serfs % (1858) 0.065*** −0.062*** −0.290** −0.381** −0.096*** 
 (0.016) (0.020) (0.119) (0.142) (0.034) 
Flexible controls ✓ ✓ ✓ ✓ ✓ 
Distances: City and provincial capital ✓ ✓ ✓ ✓ ✓ 
SIC fixed effects   ✓ ✓ ✓ 
Fixed effects Province Province Province Province Province 
Observations 14,154 14,154 14,055 13,933 13,923 
R2 0.03 0.02 0.57 0.63 0.77 
Number of clusters 36 36 36 36 36 
δ for β=0 −4.422 2.761 2.178 5.115 −1.892 
Lower bound estimates 0.096 −0.085 −0.240 −0.379 −0.139 
Conley S.E. 300 km      
Serfs % (1858) [0.012]*** [0.018]*** [0.142]** [0.165]** [0.037]** 
AgricultureManufacturing(log) Employment(log) Turnover(log) Turnover per Worker
(1)(2)(3)(4)(5)
Serfs % (1858) 0.065*** −0.062*** −0.290** −0.381** −0.096*** 
 (0.016) (0.020) (0.119) (0.142) (0.034) 
Flexible controls ✓ ✓ ✓ ✓ ✓ 
Distances: City and provincial capital ✓ ✓ ✓ ✓ ✓ 
SIC fixed effects   ✓ ✓ ✓ 
Fixed effects Province Province Province Province Province 
Observations 14,154 14,154 14,055 13,933 13,923 
R2 0.03 0.02 0.57 0.63 0.77 
Number of clusters 36 36 36 36 36 
δ for β=0 −4.422 2.761 2.178 5.115 −1.892 
Lower bound estimates 0.096 −0.085 −0.240 −0.379 −0.139 
Conley S.E. 300 km      
Serfs % (1858) [0.012]*** [0.018]*** [0.142]** [0.165]** [0.037]** 

The unit of observation is a firm. Flexible controls include eight dummies for cereal suitability and four dummies for quartiles of growing-season temperature, growing-season precipitation, the share of podzol soil, and river density, as well as linear controls of latitude and longitude of the district, the area covered by forest, ruggedness, the distance to the coast, and the distance to Moscow. Distances are the distance to the nearest city in 1600 and the distance to the provincial capital. SIC fixed effects are dummies for industrial classifications of firms using the five-digit Standard Industrial Classification Codes. The restricted model used to compute δ and the lower-bound estimates controls for province fixed effects. Standard errors clustered at the province in parentheses. *p<0.10, **p<0.05, and ***p<0.01.

Appendix table H1 presents estimates of the relationship between historical serfdom and human capital outcomes in the late Imperial period (columns 1 and 2). Confirming historical accounts (Eklof, 1986), column 1 finds that districts with more serfdom had fewer schools per thousand inhabitants before emancipation.58 However, if we consider educational outcomes about fifty years after emancipation (column 2), we do not find significant differences in educational attainment measured by the density of schools. These results suggest a convergence of basic educational attainment after emancipation.59 Historical serfdom does, however, matter for measures of educational attainment taken from the LiTS (columns 3 to 8). For adults above the age of 25, we find that serfdom is significantly associated with lower educational achievement by the respondent (column 3) and a reduced completion of postsecondary and tertiary schooling (columns 4 and 5).60 In addition, we investigate the implications of serfdom for the level of education of respondents' parents to shed more light on impediments to human capital accumulation during the Soviet period (more than 3,000 respondents in the sample are born before 1945). As shown in columns 6 and 7, the estimated coefficients suggest a similar reduction in average education of parents.61 Finally, column 8 investigates (and rejects) the idea that these differences in contemporary human capital stem from lower demand for government-provided education.62

Together, the findings in table H1 provide little support for a direct human capital channel of persistence, given the convergence after 1861.63 Rather, these results likely reflect less demand for skilled labor in former serf areas during the acceleration of industrial development in the Soviet Union. This is another piece of evidence suggesting a mechanism of historical persistence closely connected to constraints on factor mobility—particularly the allocation of labor out of less productive agriculture—and structural change that emerged under serfdom but were reinforced and even strengthened over the subsequent periods.64

D. Alternative Mechanism II: Inequality, Institutions, and Public Goods Provision

A large literature posits a relationship between labor coercion, income or wealth inequality, political institutions, and the subsequent provision of public goods, including basic schooling (Engerman & Sokoloff, 1997; Nunn, 2008a; Dell, 2010).65 While there are various possible linkages between inequality and development outcomes (e.g., financial access; differential incentives to save and invest), many of these are relatively transitory. For inequality to be a channel of long-run persistence, there must be some sort of underlying structure that perpetuates the unequal distribution of resources (and its possible impact on public goods) over time, leading to worse development outcomes. Generally, Engerman and Sokoloff (1997) and others highlight the role of institutions and the reinforcing feedback between the unequal distribution of wealth and political power.66

Are such mechanisms relevant in the Russian case? The demolition of Imperial institutions, Soviet expropriations and transfers, and the dramatic governance reforms of the Soviet and post-Soviet periods would seem to preclude a direct inequality/political economy mechanism differentiating former serf and nonserf areas. Despite this, it might have been the case that some sort of legacy of Imperial inequality did generate longer-term outcomes through less visible informal structures. The evidence presented in appendix table C10 shows that the incidence of serfdom was strongly related to measures of land concentration prior to 1905, suggesting that the end of serfdom did not fully equalize the distribution of this key asset.67 However, as we have already seen in table H1, there is little evidence that schooling can similarly be linked to serfdom by the late Imperial period, and we lack adequate data to evaluate other types of locally provided public goods in the Imperial or Soviet periods.

Using measures of contemporary inequality that we can construct from LiTS data, we do not find a significant effect of serfdom, nor do we find that past inequality predicts contemporary inequality (appendix table C10). Persistent inequality therefore seems to be an unlikely direct mechanism. Furthermore, we also investigated the connection between the historical incidence of serfdom and access to a variety of public goods in the modern data. Given the massive changes imposed by Soviet authorities, especially at the local level, it would perhaps be surprising to see the historical legacy of serfdom affecting the level of locally provided public goods today. Indeed, we find little relationship between serfdom and locally determined public goods such as water and sanitation access (see appendix table C11). However, we do find lower levels of more centrally provided (i.e., financed) public goods, such as road infrastructure, in areas with a higher incidence of serfdom (see table 4). Overall, we interpret these results to suggest that underlying structural constraints related to serfdom played a key role in determining the long-run provision of public goods (perhaps through income effects).

E. Alternative Mechanism III: A Long-Run Culture of Serfdom?

Did economic exploitation over several centuries shape beliefs and attitudes, perhaps fostering a “culture of serfdom” (Schooler, 1976), with persistent implications for economic development? Several recent studies have documented that initial institutions can have an impact on cultural norms in the longrun, which can persist through transmission within households or through formal institutions (for an overview, see Nunn, 2012). When we consider measures of trust, preferences for economic and political institutions, xenophobia, or membership in the Communist Party from LiTS, we find little evidence that serfdom was associated with differences in cultural attitudes today (see appendix tables C12, C13, and C14). This is consistent with the absence of racial, ethnic, or class markers for the descendants of serfs and with the sharp break in political and social life created by the Soviet regime. Similarly, we do not find a link between serfdom and discontent with the government, as manifested in protests or political participation (appendix table C16).

However, we do find that individuals in areas with a greater prevalence of serfdom are more likely to see luck as the reason for prosperity or poverty, have stronger preferences for redistribution, and prefer to equalize incomes rather than widen income inequality (appendix table C15). Rather than reflecting direct cultural mechanisms of persistence, we interpret these results as indicative of underlying spatial differences in income and economic activity that derived from the persistent impediments to urban development and structural change.68

VI. Conclusion

In this paper, we explore whether variation in the experience of coercive labor institutions, which existed for centuries in the Russian Empire, generated persistent differences in economic development that persist even today. Our findings confirm the adverse medium and long-run economic consequences of Russian serfdom that has often been assumed but never definitively proven. This is the case across a wide variety of specifications and robustness checks, and we argue that it cannot be driven by unobservable factors associated with both historical serfdom and modern development. In the absence of cultural, racial, or ethnic markers of past labor coercion, we show that the experience of Russian serfdom and emancipation generated persistent constraints on urbanization and structural change. This lasted through the late Imperial and Soviet periods to today, resulting in slower city growth, lower industrial development, weaker infrastructure development, and, eventually, lower educational attainment and income levels. Thus, our results imply that early industrial development and subsequent agglomeration effects can be important channels of persistence of the effects of historical serfdom, even through periods of dramatic social and economic change.

The failure to develop adequate institutions to support market and political development has been a theme of research into Eastern Europe's transition since the fall of the Soviet Union (Aslund, 2013). Our study points to possible deeper historical roots for the difficulties that the Russian Federation and other former members of the Russian Empire currently face in their efforts at economic reform and modernization, a hypothesis that has been proposed but remains relatively untested (Castaneda Dower & Markevich, 2014; Roland, 2012). Along these lines, a number of interesting questions remain open for further research. How did specific Imperial policies, institutions, or economic shocks translate different experiences of serfdom into economic variation prior to 1917? In what ways did specific Soviet and post-Soviet policies differ across space? Are local policymakers and residents aware of a prior legacy of serfdom? We hope that the development of new empirical sources in Russian and Soviet economic history can shed light on these and related questions.

Notes

1

Estimates from the Maddison project (Bolt & Zanden, 2014).

2

Slavery and debt servitude were prevalent earlier in Kievan and Muscovite Russia (Hellie, 1982).

3

These findings are robust to controlling for a large set of controls and various fixed effects and to testing for selection on unobservables (Oster, 2019).

4

See Engerman and Sokoloff (1997); Dell (2010); Nunn (2008b); Acharya, Blackwell, and Sen (2018); and Lowes and Montero (2018).

5

Different from the African context where ethnicity, religion, and race play central roles as mechanisms (Michalopoulos & Papaioannou, 2013; Nunn & Wantchekon, 2011), Russian serfs differed little from their masters with respect to race, ethnicity, or religion, and they tended to enjoy considerable autonomy in how they allocated their time unlike, for example, the majority of American slaves.

6

This hypothesis is consistent with studies by Bleakley and Lin (2012), Davis and Weinstein (2002), and others. Delays in structural change can also rationalize the cross-country relationship between incomes and the timing of peasant emancipation depicted in figure 1. Appendix figure I1 illustrates that a later emancipation of peasants is strongly associated with a larger share of labor in agriculture in 1900, and even in 2000.

7

While there was an association of serfdom with late-Imperial land inequality, there is little effect of serfdom on contemporary measures of inequality or on the provision of local public goods today. For such proposed mechanisms in other contexts, see Engerman and Sokoloff (1997), Galor et al. (2009), and Galor and Moav (2006).

8

While we find that preferences for redistributive policies are elevated in former serf areas, we view these differences as reflective of the persistent spatial inequalities driven by differential structural change. Given the disruptions of the twentieth century, our main findings are also unlikely to be driven by a specific culture of serfdom (Schooler, 1976).

9

For historical perspectives, see Dennison (2011), Gerschenkron (1966), and Lenin (1911).

10

We also document that coercive labor institutions have economic consequences in the absence of racial or ethnic markers and outside the context of European colonialism. In addition, we provide new evidence on the long-run effects of different forms of labor coercion (corvée versus quitrent), a distinction that has received little attention in the prior literature (summarized in Nunn, 2013).

11

Many of these constraints were explicit under Russian law, especially with regard to restrictions on landownership, the freedom to contract over labor, and schooling. From the beginning, the nobility's autonomy included the possibility of emancipating their serfs on their own terms. This option was exercised relatively infrequently.

12

Soviet Marxist works (e.g., Koval'chenko, 1967) did marshal considerable data to argue that the serf economy was in decline prior to 1861.

13

Such revisionist studies have relied on empirical evidence that is not necessarily representative, is too aggregate to identify differences, or covers an intermediate stage of a complicated and drawn-out reform process.

14

In the absence of district-level data on the total population from the tax census, we draw on Bushen (1863), who provides the total population in 1863.

15

While over 90% of districts contained some serfs just before emancipation, in only a few did the share of the total population exceed 80%. See figure A1 in the appendix.

16

The variables mentioned here are described and summarized in appendix table A1.

17

The inclusion of provincial fixed effects helps us to address many strategic and economic considerations for the location of military populations, estates, and serfs, a concern highlighted by Matranga and Natkhov (2019).

18

We also consider soil suitability for growing specific grains in robustness checks.

19

Many of these variables are measured today. Soviet authorities did engage in policies that may have affected agricultural conditions. Such changes were relatively small, likely uncorrelated with incidence of serfdom, and occurred largely outside European Russia.

20

A priori, it is not clear how proximity to Moscow of high-serf areas would directly relate to long-run economic outcomes. There may have been positive development spillovers from the economic center, but being close to the heart of an extractive state might generate other negative consequences. We find that, if anything, places close to Moscow are more developed, suggesting that any negative impact of serfdom on economic development might be underestimated.

21

For the coefficients of all controls, see appendix table B1.

22

As argued by Acemoglu and Wolitzky (2011), the depression of outside options can enable stronger coercion. Other available indicators of outside options prior to 1861, such as the presence of factories, are more likely to have been endogenous to the location of serfdom. There are no district-level data on industrial activity prior to 1861.

23

See appendix table B2 for results of an omnibus test indicating that the variation in development outcomes today that is predictable from covariates is unrelated to the historical incidence of serfdom.

24

The LiTS is collected by the European Bank for Reconstruction and Development to assess household and individual well-being in transition countries. The appendix contains additional information on the survey and relevant variables.

25

Although this variable relies on recall, the accuracy is remarkably good when compared to directly measured household consumption (Zaidi et al., 2009).

26

Appendix figure A3 shows the PSU locations.

27

Besides the latitude and longitude of the district, we control for the area covered by forest, ruggedness, land suitability for growing cereals, average temperature and precipitation during the growing season, river density, the share of land with podzol soils, the distance to the coast, and the distance to Moscow.

28

In our preferred specifications, when the sample size permits, we utilize historical province fixed effects, which rules out that the results are driven by provinces without serfdom in 1860, such as the Baltics. This is a demanding specification, since in some provinces, the number of households sampled in the LiTS is small.

29

Appendix table C1 displays these results with coefficients for all the control variables.

30

To compute spatially adjusted standard errors, we use the routine of Colella et al. (2019) that calculates p-values assuming a normal distribution of errors. Appendix table C2 shows that the 300 km cutoff produces the largest standard errors.

31

See Oster (2019) for the formal details. Our restricted model controls for only province or country effects.

32

A previous version of this paper explored the number of monasteries expropriated (with monastic serfs transferred to the state peasantry) by Catherine the Great in the mid-eighteenth century as an instrumental variable. However, the location of monasteries may be related to unobservable local factors that could plausibly drive longer-run economic outcomes. Therefore, we have excluded this strategy from this paper. A version of our earlier instrumental variable strategy was recently adopted in Markevich and Zhuravskaya (2018).

33

Specifications that include pre-emancipation population density not only take into account one possible driver of the incidence of serfdom (labor scarcity, as suggested in Domar, 1970), but this variable also soaks up many other (potentially unobservable) geographic and other channels of long-run persistence.

34

In the appendix, we also document robustness of the main results to varying household controls (table C4); to controlling for wheat, rye, barley, and oat suitability separately (table C5); to using measures of consumer good ownership as dependent variables (table C6) or to the waves of LiTS (table C7); and to controlling for (potentially endogenous) rural/urban status of the PSU (table C8).

35

See appendix section C.2. We thank Katia Zhuravskaya for suggesting this exercise.

36

This is especially useful in agrarian economies, in which urbanization and agricultural productivity are tightly linked (Bairoch et al., 1988).

37

We rely on population data from Mikhailova (2012), which, in turn, is derived from the Imperial, Soviet, and post-Soviet censuses of 1897, 1926, 1939, 1959, 1970, 1989, and 2002.

38

Following Oster (2019) again, we find that our estimates are, if anything, possibly biased downward. We get similar results when we control for the geographic environment in a linear fashion, although the effects are slightly smaller and less precisely estimated (see appendix table D2). The pattern of persistent differences in urbanization levels according to the experience of serfdom is also consistent with the results obtained using city growth as the dependent variable, controlling for the initial level of population in each subperiod (see appendix table D3 and figure D3).

39

Appendix figure D4 illustrates the evolution of this urban gap between areas above and below the median level of serfdom during the Soviet period. Also see the appendix for (a) a flexible version of equation (3) that finds negative and increasing coefficients over time on the interaction of serfdom and year dummies (table D5), (b) a visual illustration of the estimated interaction coefficients of serfdom with each time dummy (figure D5); (c) estimates with standard errors clustered at the level of the province (table D6); (d) a discussion of possible sample selection bias in the balanced panel (section D.3); and (e) estimates using data aggregated to the level of the province (sections D.2 and D.4).

40

This conclusion echoes the results in Markevich and Zhuravskaya (2018), who find large, positive economic effects of emancipation in areas with a greater prevalence of serfs. This does not imply that full convergence ever took place.

41

The main goals of the passport system, run by the Ministry of Internal Affairs through the local police, were to maintain control over mobility to prevent social instability and ensure tax payments (Burds, 1998; Chernukha, 2007). To the best of our knowledge, there are currently no district-level data on Imperial passports.

42

Collective communal tax liabilities, urban disamenities, and the relatively rigid social estate system also generated frictions in the reallocation of labor that may have affected former serfs to a greater degree.

43

Collectivization aimed to break up traditional institutions and factor relations in the Soviet countryside, and it at least partially succeeded. Although we have found no evidence that collectivization differentially targeted former serf villages, if such settlements were relatively more unequal by the 1930s, then related “dekulakization” campaigns aimed at wealthier peasants may differentially have affected them. This would be consistent with the findings of Naumenko (2019), who shows that Soviet areas more adversely affected by famine in 1933 saw slower subsequent urban growth. In our view, the policies (e.g., procurement, collectivization) that drove famine may have compounded initial differences between serf and nonserf areas.

44

The geostrategic shift of resources eastward (and largely out of our study region) before, during, and after World War II, along with a continual emphasis on cross-regional “equalization” policies, generated significant spatial distortions in the Soviet economy (Markevich & Mikhailova, 2013). These may have perpetuated the initial lag in development in former serf areas.

45

Over time, further restrictions were imposed on migration to “secret” or closed cities. Such cities, while initially larger, grew more slowly over the Soviet (and post-Soviet) period. However, if we control for their presence using data reported in Gang and Stuart (1999), our main findings and extensions are unchanged (results available upon request).

46

On post-1991 labor mobility constraints across the former Soviet Union, see Buckley (1995), Koettl et al. (2014), and Markevich and Mikhailova (2013).

47

After 1861, Markevich and Zhuravskaya (2018) find a large, positive industrial productivity increase for provinces with relatively more serfdom, but the mechanisms behind this sudden change are not directly observed, and such growth did not fully offset preexisting differences between serf and nonserf areas.

48

The δ values of the Oster (2019) test reported in table 4 indicate that selection on unobservables cannot explain these findings. Appendix table G5 documents that the negative relationship of serfdom and light density can also be found in other years between 1994 and 2012 and when conditioning on historical population density in 1858. We also find very similar results when we use linear rather than flexible controls; see table G1.

49

We do this by drawing on district-level occupational data from the 1897 national census, using occupational totals across the 65 specified in the original source to define employment shares of different sectors. See the appendix.

50

Districts with higher levels of serfdom also displayed significantly lower 1897 employment in service occupations, such as those related to education and commerce. See appendix table G3.

51

Appendix table G4 reports similar heterogeneous effects for additional indicators of structural change, such as urbanization, factory employment, and contemporary population density.

52

This might also be due to localized upstream and downstream linkages related to the defense factories, many of which also produced consumer goods. Other than the eastward shift in World War II and the creation of “closed” cities (see above), we have found little evidence for a particular spatial allocation rule when it came to defense plants. If anything, the sector's priority status meant that the location of the Soviet defense industry likely built on and reinforced existing spatial patterns (Markevich, 2008).

53

The negative coefficient prior to World War II suggests that our findings are not indicative of the wartime movement of production eastward.

54

For 1959, 1970, and 1989, when we have population data to define a per capita measure of defense plants, we find similarly negative effects for 1959 but not for the latter two years. This matches our finding of greater urban population growth in areas with fewer serfs, leading to some convergence in defense plants per capita. These results are available on request.

55

These dynamics are depicted in appendix figure E1. We find similar results when we use linear controls (table E2) or when estimating negative binomial regressions (table E3). In table E6, we document that the negative effects of serfdom on city population and city growth are attenuated in cities with a higher number of factories. Finally, we find a lack of catch-up growth in industrial production (appendix table E4 and figure E3).

56

Discussions of long-run persistence through human capital channels in other contexts include Chen, Kung, and Ma (2020), Margo (2016), and Sacerdote (2005). These and other studies tend to emphasize the role of intergenerational transmission via households or local culture.

57

Ivanov (2016) argues that areas with higher human capital levels prior to 1991—for nonmarket reasons under the planned economy—saw greater gains in human-capital intensive activities afterward. This is consistent with our findings.

58

A 1 standard deviation increase in serfdom reduces schooling by 0.12 of a standard deviation.

59

We have studied district-level measures of enrollment and schooling from 1880, 1894, and 1911 and literacy rates in 1897. Our finding that human capital “gaps” vanished by the end of the nineteenth century is supported in these exercises. See appendix table H2.

60

A 1 standard deviation increase in serfdom reduces the likelihood that the respondent has completed tertiary education by 6 percentage points (from a mean of 28%).

61

A 1 standard deviation increase in serfdom decreases parents' schooling by 0.6 of a year (or 0.15 of a standard deviation) and the number of parents with tertiary education by 0.11 of a standard deviation.

62

None of the significant educational differences are likely to be driven by unobservables, as documented by the Oster (2019) test results at the bottom of table H1.

63

This convergence is consistent with late-Imperial efforts to improve the provision of basic schooling, especially in rural areas (Eklof, 1986; Kaser, 2006; Nafziger, 2012).

64

Cheremukhin et al. (2017) argue that industrial market power and entry barriers constrained pre-Soviet growth. We cannot directly examine their hypothesis in our spatial analyses. Our emphasis on labor market frictions is complementary to their interpretation.

65

Similarly, Galor et al. (2009) assert that elites in largely agrarian societies may have little interest in funding public goods that have limited direct payoffs to themselves. For empirical evidence from Prussia, see also Cinnirella and Hornung (2016).

66

Acharya, Blackwell, and Sen (2018) recently have argued for the long-run political consequences of slavery in the American South, whereby historical coercion was associated with persistent racist beliefs, inequality, and subsequent institutional capture by whites, both of which shaped black and white political behavior for generations after emancipation.

67

For a longer discussion of public good provision amid political and economic inequality in late Imperial Russia, see Nafziger (2011).

68

These preference differences are broadly consistent with the view that a high-redistribution equilibrium in Europe (relative to the United States) may reflect the continent's experience with feudalism (Alesina & Glaeser, 2004).

REFERENCES

Acemoglu
,
Daron
,
Tarek A.
Hassan
, and
James A.
Robinson
, “
Social Structure and Development: A Legacy of the Holocaust in Russia,
Quarterly Journal of Economics
126
(
2011
),
895
946
.
Acemoglu
,
Daron
, and
James
Robinson
,
Why Nations Fail: The Origins of Power, Prosperity and Poverty
(
New York
:
Crown Books
,
2012
).
Acemoglu
,
Daron
, and
Alexander
Wolitzky
, “
The Economics of Labor Coercion,
Econometrica
79
(
2011
),
555
600
.
Acharya
,
Avidit
,
Matthew
Blackwell
, and
Maya
Sen
,
Deep Roots: How Slavery Still Shapes Southern Politics
(
Princeton, NJ
:
Princeton University Press
,
2018
).
Alesina
,
Alberto
, and
Edward Ludwig
Glaeser
,
Fighting Poverty in the US and Europe: A World of Difference
(
New York
:
Oxford University Press
,
2004
).
Allen
,
Robert C.
,
Farm to Factory: A Reinterpretation of the Soviet Industrial Revolution
(
Princeton, NJ
:
Princeton University Press
,
2003
).
Aslund
,
Anders
,
How Capitalism Was Built: The Transformation of Central and Eastern Europe, Russia, the Caucasus, and Central Asia
, 2nd ed. (
Cambridge
:
Cambridge University Press
,
2013
).
Bairoch
,
Paul
,
Jean
Batou
, and
Pierre
Chèvre
,
Population des villes européennes de 800 à 1850: Banque de données et analyse sommaire des résultats (la)
(
Paris
:
Librairie Droz
,
1988
).
Bleakley
,
Hoyt
, and
Jeffrey
Lin
, “
Portage and Path Dependence,
Quarterly Journal of Economics
127
(
2012
),
587
644
.
Bolt
,
Jutta
, and
Jan Luiten
Zanden
, “
The Maddison Project: Collaborative Research on Historical National Accounts,
Economic History Review
67
(
2014
),
627
651
.
Buckley
,
Cynthia
, “
The Myth of Managed Migration: Migration Control and Market in the Soviet Period,
Slavic Review
54
(
1995
),
896
916
.
Burds
,
Jeffrey
,
Peasant Dreams and Market Politics: Labor Migration and the Russian Village, 1861–1905
(
Pittsburgh
:
University of Pittsburgh Press
,
1998
).
Bushen
,
Artur
, ed.,
Nalichnoe naselenie Imperii za 1858 god
, vol.
2
of Statisticheskiia tablitsy Rossiiskoi Imperii
(
1863
).
Castaneda
Dower
,
Paul, and Andrei
Markevich
, “
A History of Resistance to Privatization in Russia,
Journal of Comparative Economics
42
(
2014
),
855
873
.
Castaneda
Dower
,
Paul, and Andrei
Markevich
The Stolypin Reform and Agricultural Productivity in Late Imperial Russia,
Center for Economic and Financial Research working paper w0239
(
2017
).
Chen
,
Ting
,
James Kai-sing
Kung
, and
Chicheng
Ma
, “
Long Live Keju! The Persistent Effects of China's Civil Examination System,
Economic Journal
130
(
2020
),
2010
2064
.
Cheremukhin
,
Anton
,
Mikhail
Golosov
,
Sergei
Guriev
, and
Aleh
Tsyvinski
, “
The Industrialization and Economic Development of Russia through the Lens of a Neoclassical Growth Model,
Review of Economic Studies
84
(
2017
),
613
649
.
Chernina
,
Eugenia
,
Paul Castaneda
Dower
, and
Andrei
Markevich
, “
Property Rights, Land Liquidity and Internal Migration,
Journal of Development Economics
110
(
2014
),
191
215
.
Chernukha
,
Valentina
,
Pasport v Rossii 1719–1917 gg
(
St. Petersburg
:
Liki Rossii
,
2007
).
Cinnirella
,
Francesco
, and
Erik
Hornung
, “
Landownership Concentration and the Expansion of Education,
Journal of Development Economics
121
(
2016
),
135
152
.
Colella
,
Fabrizio
,
Rafael
Lalive
,
Seyhun Orcan
Sakalli
, and
Mathias
Thoenig
, “
Inference with Arbitrary Clustering
,”
IZA discussion paper
12584
(
2019
).
Conley
,
Timothy G.
, “
GMM Estimation with Cross Sectional Dependence,
Journal of Econometrics
92
(
1999
)
1
45
.
Davis
,
Donald
, and
David
Weinstein
, “
Bones, Bombs, and Break Points: The Geography of Economic Activity,
American Economic Review
92
(
2002
),
1269
1289
.
Dell
,
Melissa
, “
The Persistent Effects of Peru's Mining Mita,
Econometrica
78
(
2010
),
1863
1903
.
Dennison
,
Tracy
,
The Institutional Framework of Russian Serfdom
(
Cambridge
:
Cambridge University Press
,
2011
).
Dexter
,
Keith
, and
Ivan
Rodionov
, “
The Factories, Research and Design Establishments of the Soviet Defence Industry: A Guide (version 17)
” (
2016
), https://warwick.ac.uk/fac/soc/economics/staff/mharrison/vpk/history/preface_17.pdf.
Domar
,
Evsey D.
, “
The Causes of Slavery or Serfdom: A Hypothesis,
Journal of Economic History
30
(
1970
),
18
32
.
Domar
,
Evsey
, and
Mark
Machina
, “
On the Profitability of Russian Serfdom,
Journal of Economic History
44
(
1984
),
919
955
.
Eklof
,
Benjamin
,
Russian Peasant Schools: Officialdom, Village Culture, and Popular Pedagogy, 1861–1914
(
Berkeley
:
University of California Press
,
1986
).
Engerman
,
Stanley
, and
Kenneth
Sokoloff
, “
Factor Endowments, Institutions and Differential Paths of Growth among the New World Economies,
” in
Stephen
Haber
, ed.
How Latin America Fell Behind
(
Stanford
:
Stanford University Press
,
1997
).
Field
,
Daniel
,
The End of Serfdom: Nobility and Bureaucracy in Russia, 1855–1861
(
Cambridge, MA
:
Harvard University Press
,
1976
).
Galor
,
Oded
, and
Omer
Moav
, “
Das Human-Kapital: A Theory of the Demise of the Class Structure,
Review of Economic Studies
73
(
2006
),
85
117
.
Galor
,
Oded
,
Omer
Moav
, and
Dietrich
Vollrath
, “
Inequality in Landownership the Emergence of Human-Capital Promoting Institutions, and the Great Divergence,
Review of Economic Studies
76
(
2009
),
143
179
.
Gang
,
Ira
, and
Robert C.
Stuart
, “
Mobility Where Mobility Is Illegal: Internal Migration and City Growth in the Soviet Union,
Journal of Population Economics
12
(
1999
),
117
134
.
Gerschenkron
,
Alexander
, “
Agrarian Policies and Industrialization: Russia 1861–1917,
” in
Hrothgar J.
Habakkuk
and
Michael M.
Postan
, eds.,
The Cambridge Economic History of Europe
(
Cambridge
:
Cambridge University Press
,
1966
).
Glaeser
,
Edward L.
,
José A.
Scheinkman
, and
Andrei
Shleifer
, “
Economic Growth in a Cross-Section of Cities,
Journal of Monetary Economics
36
(
1995
),
117
143
.
Gregory
,
Paul R.
, and
Valery
Lazarev
, eds.,
The Economics of Forced Labor: The Soviet Gulag
(
Palo Alto, CA
:
Hoover Press
,
2003
).
Hellie
,
Richard
,
Slavery in Russia, 1450–1725
(
Chicago
:
University of Chicago Press
,
1982
).
Hoch
,
Steven L.
, “
Did Russia's Emancipated Serfs Really Pay Too Much for Too Little Land? Statistical Anomalies and Long-Tailed Distributions,
Slavic Review
63
(
2004
),
247
274
.
Hoch
,
Steven L.
, and
Wilson
Augustine
, “
The Tax Censuses and the Decline of the Serf Population in Imperial Russia, 1833–1858,
Slavic Review
38
(
1979
),
403
425
.
Ivanov
,
Denis
, “
Human Capital and Knowledge-Intensive Industries Location: Evidence from Soviet Legacy in Russia,
Journal of Economic History
76
(
2016
),
736
768
.
Kabuzan
,
Vladimir Maksimovich
,
Krepostnoe naselenie Rossii v XVIII v. – 50-x godakh
.
XIX v.
(
St. Petersburg
:
BLITS
,
2002
).
Kaser
,
Michael
, “
Resources for Russian Education: Soviet Strategies in Historical Context,
History of Education
35
(
2006
),
561
583
.
Kessler
,
Gijs
, “
The Passport System and State Control over Population Flows in the Soviet Union, 1932–1940,
Cahiers du Monde russe
42
(
2001
),
477
504
.
Khristoforov
,
Igor
,
Sud'ba Reformy: Russkoe krest'ianstvo i pravitel'stvennoi politike do i posle otmeny krepostnogo prava (1830-1890-e gg.)
(
Sobranie
,
2011
).
Koettl
,
Johannes
,
Olga
Kupets
,
Anna
Olefir
, and
Indhira
Santos
, “
In Search of Opportunities? The Barriers to More Efficient Internal Labor Mobility in Ukraine
,”
IZA Journal of Labor and Development
3
(
2014
).
Koval'chenko
,
Ivan
,
Russkoe krepostnoe krest'ianstvo v pervoi polovine XIX veka
(
Moscow
:
Moscow University Press
(
1967
).
Lenin
,
Vladimir I.
, “
The Peasant Reform and the Proletarian-Peasant Revolution,
” in
George
Hanna
, ed.,
Collected Works
, vol.
17
(
Moscow
:
Progress Publisher
,
1911
).
Litvak
,
Boris
,
Russkaia derevnia v reforme 1861 goda
(
Moscow
:
Nauka
,
1972
).
Lowes
,
Sara
, and
Eduardo
Montero
, “
Blood Rubber,
Harvard University working paper
(
2018
).
Margo
,
Robert
, “
Obama, Katrina and the Persistence of Racial Inequality,
Journal of Economic History
76
(
2016
),
301
341
.
Markevich
,
Andrei
, “
Planning the Supply of Weapons: The 1930s
” (pp.
78
117
) in
Mark
Harrison
, ed.,
Guns and Rubles: The Defense Industry in the Stalinist State
(
New Haven, CT
:
Yale University Press
,
2008
).
Markevich
,
Andrei
, and
Tatiana
Mikhailova
, “Economic Geography of Russia” (pp.
617
642
), in
Michael
Alekseev
and
Shlomo
Weber
, eds.,
The Oxford Handbook of the Russian Economy
(
New York
:
Oxford University Press
,
2013
).
Markevich
,
Andrei
, and
Ekaterina
Zhuravskaya
, “
Economic Effects of the Abolition of Serfdom: Evidence from the Russian Empire,
American Economic Review
108
(
2018
),
1074
1117
.
Matranga
,
Andrea
, and
Timur
Natkhov
, “
All along the Watchtower: Defense Lines and the Origins of Russian Serfdom,
University of California, Davis, working paper
(
2019
).
Michaels
,
Guy
, and
Ferdinand
Rauch
, “
Resetting the Urban Network: 117–2012,
Economic Journal
128
(
2016
),
378
412
.
Michalopoulos
,
Stelios
, and
Elias
Papaioannou
, “
Pre-Colonial Ethnic Institutions and Contemporary African Development,
Econometrica
81
(
2013
),
113
152
.
Mikhailova
,
Tatiana
, “
Gulag, WWII and the Long-Run Patterns of Soviet City Growth,
working paper
(
2012
).
Moon
,
David
,
The Russian Peasantry 1600–1930: The World the Peasants Made
(
London
:
Longman
1999
).
Nafziger
,
Steven
, “
Did Ivan's Vote Matter? The Political Economy of Local Democracy in Tsarist Russia,
European Review of Economic History
15
(
2011
),
393
441
.
Nafziger
,
Steven
The Political Economy of Primary Education in Tsarist Russia,
Williams College working paper
(
2012
).
Nafziger
,
Steven
Russian Serfdom, Emancipation, and Land Inequality: New Evidence,
Williams College working paper
(
2013
).
Naumenko
,
Natalya
, “
Economic Consequences of the 1933 Soviet Famine,
George Mason University working paper
(
2019
).
Nunn
,
Nathan
, “
The Long-Term Effects of Africa's Slave Trades,
Quarterly Journal of Economics
123
(
2008a
),
139
176
.
Nunn
,
Nathan
Slavery, Inequality and Economic Development in the Americas,
Institutions and Economic Performance
15
(
2008b
),
148
180
.
Nunn
,
Nathan
Culture and the Historical Process,
Economic History of Developing Regions
27
(
2012
),
S108
S126
.
Nunn
,
Nathan
Historical Development,
” in
Philippe
Aghion
and
Steven
Durlauf
, eds.,
Handbook of Economic Growth
, vol.
2
(
Amsterdam
:
North-Holland
,
2013
).
Nunn
,
Nathan
, and
Leonard
Wantchekon
, “
The Slave Trade and the Origins of Mistrust in Africa,
American Economic Review
101
(
2011
),
3221
3252
.
Oster
,
Emily
, “
Unobservable Selection and Coefficient Stability: Theory and Evidence,
Journal of Business and Economic Statistics
37
(
2019
),
187
204
.
Rocha
,
Rudi
,
Claudio
Ferraz
, and
Rodrigo R.
Soares
, “
Human Capital Persistence and Development,
American Economic Journal: Applied Economics
9
(
2017
),
105
136
.
Roland
,
Gerard
, “The Long Run Weight of Communism or the Weight of Long Run History” (pp.
153
171
), in
Gerard
Roland
, ed.,
Economies in Transition: The Long-Run View
(
London
:
Palgrave Macmillan
,
2012
).
Sacerdote
,
Bruce
, “
Slavery and the Intergenerational Transmission of Human Capital,
this review 87
(
2005
),
217
234
.
Schooler
,
Carmi
, “
Serfdom's Legacy: An Ethnic Continuum,
American Journal of Sociology
81
(
1976
),
1265
1286
.
Troinitskii
,
Aleksandr G.
,
The Serf Population in Russia According to the 10th National Census
.
Oriental Research Partners
(
1982
;
1861
).
Wirtschafter
,
Elise
,
Social Identity in Imperial Russia
(
DeKalb, IL
:
Northern Illinois University Press
,
1997
).
Zaidi
,
Salman
,
Asad
Alam
,
Pradeep
Mitra
, and
Ramya
Sundaram
, “
Satisfaction with Life and Service Delivery in Eastern Europe and the Former Soviet Union: Some Insights from the 2006 Life in Transition Survey
,”
World Bank working paper
162
(
2009
).

Author notes

We thank the editor and four anonymous referees, as well as Yann Algan, Quamrul Ashraf, Roger Bartlett, Sascha Becker, Elena Nikolova, Oded Galor, Sergei Guriev, Stephanos Vlachos, Joachim Voth, Katia Zhuravskaya, and participants at numerous seminars and conferences for useful discussions and comments. We also thank the EBRD for providing the geocoded LiTS survey waves and Ekaterina Borislova, Theocharis Grigoriadis, Dmitry Kofanov, Tatiana Mikhailova, Alexander Skorobogatov, and Marvin Suesse for sharing additional data. Ivan Badinsky and Gabrielle McPhaul-Gurrier provided excellent research assistance. J.B. acknowledges financial support from ERC Starting Grant GRIEVANCES-313327. The online appendix contains additional material and results. All errors remain our own.

A supplemental appendix is available online at https://doi.org/10.1162/rest_a_00857.

Supplementary data