AMERICAN UNIVERSITY WORKING PAPER SERIES CAPITAL-SKILL COMPLEMENTARITY AND THE EMERGENCE OF LABOR EMANCIPATION Boris Gershman Quamrul H. Ashraf Fancesco Cinnirella Oded Galor Erik Hornung Working Paper 2017-04 http://www.edu/cas/economics/research/upload/2017-04.pdf JEL No.: J24, J47,N13, N33, O14,O15, O43 AMERICAN UNIVERSITY 4400 Massachusetts Ave. 20016-8029 ABSTRACT This paper advances a novel hypothesis regarding the historical roots of labor emancipation. It argues that the decline of coercive labor institutions in the industrial phase of development has been an inevitable by-product of the intensification of capital-skill complementarity in the production process. In light of the growing significance of skilled labor for fostering the return to physical capital, elites in society were induced to relinquish their historically profitable coercion of labor in favor of employing free skilled workers, thereby incentivizing the masses to engage in broad-based human capital acquisition, without fear of losing their skill premium to expropriation. In line with the proposed hypothesis, exploiting a plausibly exogenous source of variation in early industrialization across regions of nineteenth-century Prussia, capital abundance is shown to have contributed to the subsequent intensity of de facto serf emancipation.17606/q4ev-w066 Capital-Skill Complementarity and the Emergence of Labor Emancipation∗ Quamrul H. Ashraf, Francesco Cinnirella, Oded Galor, Boris Gershman, and Erik Hornung† March 2017 Abstract This paper advances a novel hypothesis regarding the historical roots of labor emancipation. It argues that the decline of coercive labor institutions in the industrial phase of development has been an inevitable by-product of the intensification of capital-skill complementarity in the production process. In light of the growing significance of skilled labor for fostering the return to physical capital, elites in society were induced to relinquish their historically profitable coercion of labor in favor of employing free skilled workers, thereby incentivizing the masses to engage in broad-based human capital acquisition, without fear of losing their skill premium to expropria- tion. In line with the proposed hypothesis, exploiting a plausibly exogenous source of variation in early industrialization across regions of nineteenth-century Prussia, capital abundance is shown to have contributed to the subsequent intensity of de facto serf emancipation. Keywords: Labor coercion, serfdom, emancipation, industrialization, physical capital accumu- lation, capital-skill complementarity, demand for human capital, nineteenth-century Prussia JEL classification codes: J24, J47, N13, N33, O14, O15, O43 ∗ We thank Philipp Ager, Toke Aidt, Sascha Becker, Matteo Cervellati, Nicholas Crafts, Carl-Johan Dalgaard, Brad DeLong, Christian Dippel, James Fenske, Noel Johnson, Mark Koyama, Keith Kuester, James Kung, Nippe Lagerlöf, David Mitch, Steven Nafziger, Thijs van Rens, Paul Sharp, Uwe Sunde, Joachim Voth, Nico Voigtländer, Jacob Weisdorf, Niko Wolf, Ludger Wößmann, Noam Yuchtman, and seminar participants at Bonn, Caltech, Copenhagen, Higher School of Economics, HKUST, Hohenheim, HU Berlin, the Ifo Institute, Innsbruck, the Max Planck Institute for Tax Law and Public Finance, New Economic School, Passau, Southern Denmark, UC Berkeley, UCLA, Warwick, Williams, the 2015 World Economic History Congress, the 2015 CAS Workshop on Institutions, Culture, and Long-run Development, and the 2016 Summer Workshop on Economic History and Development at GMU for valuable comments. Ashraf acknowledges research support from the NSF (SES-1338738), the Hellman Fellows Program, and the Oakley Center for Humanities and Social Sciences at Williams College. Galor acknowledges research support from the NSF (SES-1338426). † Ashraf: Williams College, Department of Economics, 24 Hopkins Hall Dr.edu); Cinnirella: Ifo Institute for Economic Research, Center for the Economics of Education, Poschingerstr. 5, 81679 Munich, Germany (email: cinnirella@ifo.de); Galor: Brown University, Department of Economics, 64 Waterman St. (email: Oded Galor@brown.edu); Gershman: American University, Department of Economics, 4400 Massachusetts Ave. NW, Washington, DC 20016, U.edu); Hor- nung: University of Bayreuth, Department of Law and Economics, Universitaetsstr. 30, 95447 Bayreuth, Germany (email: erik.hornung@uni-bayreuth. 1 Introduction The rise and fall of coercive labor market institutions over the course of human history has been a subject of intensive exploration across a broad range of academic disciplines. The emergence of labor coercion has generally been attributed to the increased demand for agricultural labor as well as the deepening of class stratification and the scope for domination within societies in the aftermath of the transition from hunting and gathering to agriculture during the Neolithic Revolution. In contrast, the origins of labor emancipation and the universal protection of individual property rights have been largely ascribed to sociopolitical forces in the early modern period, such as the influence of the Age of Enlightenment on the perceived value of personal freedoms or attempts by the elites in society to appease the masses in order to mitigate the adverse consequences of popular revolts.1 Nevertheless, the advent of coercive institutions following the Neolithic Revolution and their decline over the course of the Industrial Revolution suggest that the economic forces underlying these profound changes in the structure of production were also at the roots of the associated transformations in labor market institutions. In particular, while the agricultural revolution was associated with the rise of unskilled labor-intensive production, the Industrial Revolution marked a systematic transition to a skill-intensive production structure. This paper advances a novel hypothesis regarding the historical roots of labor emancipation. It suggests that the decline of labor coercion in the industrial stage of development has been an inevitable by-product of the emergence of capital-skill complementarity in the production process. In the earlier agricultural stage of development, production was largely intensive in unskilled la- bor, and landowning elites had an incentive to promote and maintain coercive institutions that limited labor mobility and boosted land rents.2 Over the course of industrialization, however, the accumulation of physical capital by the landed elites coupled with the advent of capital-skill com- plementarity altered their viewpoint regarding the profitability of exploiting coerced labor, thus bringing an end to this oppressive institutional regime. Specifically, in light of the growing sig- nificance of skilled labor for fostering the return to physical capital, landowning capitalist elites were induced to relinquish their historically advantageous coercion of labor in favor of employing free skilled workers. These uncoordinated actions of the elites, in turn, incentivized the masses to engage in broad-based human capital acquisition, without fear of losing their skill premium to expropriation.3 The posited mechanism therefore suggests that willingness to abandon the extant 1 In particular, the Age of Enlightenment is known to have triggered major institutional transformations (e., de jure agrarian reforms and the abolition of serfdom) throughout regions of Europe that were occupied by Napoleon’s French Revolutionary Army, paving the way for subsequent economic development (Acemoglu et al. 2 It is generally agreed, however, that because feudal relations prevented labor from being allocated efficiently, the process of society-wide economic development was held back in regions characterized by coercive institutions (North and Thomas, 1973; Brenner, 1976; Acemoglu and Robinson, 2012). 3 The role of alternative economic forces in the emergence of labor emancipation has been explored by Lagerlöf (2009). In his theory, the elites choose between imposing serfdom and freedom in order to maximize their payoffs, as determined by population density and the level of technology. 1 regime of labor coercion was more pronounced amongst elites whose stakes in the ownership of physical capital were larger in early stages of industrialization.4 The proposed hypothesis is examined empirically by exploiting both cross-sectional and temporal variations in de facto serf emancipation in nineteenth-century Prussia. Although serf- dom was de jure abolished throughout Prussia in 1807, the process of de facto emancipation from coercive institutions evolved heterogeneously across regions, extending well into the second half of the nineteenth century. In particular, at the individual level, the termination of feudal labor relations was the outcome of a bilateral negotiation regarding the compensation owed to the mano- rial landlord for the redemption of lifetime servile duties by the serf, thus introducing sufficient scope for variation in the duration of the de facto emancipation process.5 The empirical analysis therefore links the differential timing of de facto emancipation with variation in the initial extent of proto-industrialization across regions of Prussia. Exploiting unique data on emancipation cases, originally collected and reported at the county level by the Prussian state agency that supervised these settlements, the analysis finds a positive and significant relationship across counties between the initial stock of physical capital per capita, as proxied by the prevalence of water mills in 1819, and the share of serfs that were de facto emancipated between 1821 and 1848.6 This relationship proves to be remarkably robust to accounting for a wide range of potentially confounding factors, including observed heterogeneity across regions in geographical, cultural, and institutional characteristics and in various dimensions of historical development. The analysis implements two alternative strategies to mitigate potential concerns regarding endogeneity in the relationship between early industrialization and the subsequent intensity of de facto serf emancipation. The first strategy, implemented in the baseline cross-sectional setting, exploits terrain slope as a plausibly exogenous source of variation in the prevalence of water mills across counties in 1819, conditional on a sizable vector of other geographical factors. This strategy yields 2SLS estimates for the influence of the initial stock of physical capital that are quantitatively similar to the baseline OLS estimates, suggesting that the latter are not marred by any substantial 4 In contrast to the notion that average worker skills were generally unimportant until later stages of industrial- ization, Franck and Galor (2017) uncover new evidence linking early industrialization with the rise in the demand for human capital in mid-nineteenth-century France, consistent with an earlier emergence of technology-skill com- plementarity. Moreover, Squicciarini and Voigtländer (2015) show that although the density of “upper tail” human capital, as proxied by the prevalence of knowledge elites, was instrumental for productivity growth (innovation) and urbanization during the first phase of the French Industrial Revolution, average worker skills, as proxied by literacy rates, were nevertheless important for the level of productivity under the existing technologies, in accordance with the complementarity of average skills in industrial production. Further, in the context of nineteenth-century Prussia, Becker et al. (2011) document that basic education was significantly associated with industrialization in its early stages. Similarly, Cinnirella and Streb (forthcoming) find that literacy rates were positively related to productivity during the second phase of Prussian industrialization. 5 In addition, state supervision provided a commitment device for both serfs and their manorial landlords to honor the redemption agreement (Ogilvie, 2014). 6 As will become evident, for the purposes of empirically examining the proposed hypothesis, water mills are particularly suitable as a proxy measure of early industrialization, first, because their ownership was institutionally restricted to the landed elites, and second, because they foreshadowed the adoption of steam engines and related skill-intensive methods of industrial production. 2 amount of endogeneity bias, given the identifying assumption regarding the conditional excludabil- ity of terrain slope. The second strategy exploits time-varying district-level data on the number of settled redemption cases during the latter half of the nineteenth century, implementing a flexible panel setting in which the initial stock of physical capital is permitted to possess a time-varying relationship with the average annual flow of de facto emancipated serfs in different five-year peri- ods. This strategy is able to account for unobserved heterogeneity across districts in time-invariant geographical, cultural, and institutional characteristics, in addition to the time-varying influence of sociopolitical and demographic observables at the district level as well as region-specific time trends in emancipation flows at higher levels of spatial aggregation. The empirical analysis also explores the extent to which the main findings may reflect alternative theories of the historical decline in coercive labor institutions.7 For instance, one class of theories emphasizes that coercion can decline either with an increase in labor abundance, due to the resulting depression in market wages, or when an increase in outside options for workers poses the threat of labor scarcity (Postan, 1966; Domar, 1970; Brenner, 1976; Acemoglu and Wolitzky, 2011).8 Another alternative explanation is that the elites strategically relinquished political (and coercive economic) power over the masses in order to avert social unrest (Acemoglu and Robinson, 2000; Aidt and Franck, 2015). The analysis accounts for these particular alternative mechanisms by introducing controls for labor abundance and the prevalence of social uprisings.
