JINGYI HUANG

Jingyi Huang’s dissertation is titled “The Impact of Innovation, Regulation, and Market Power on Economic Development: Evidence from the American West.” It is an impressive study of interrelated issues in the development of the American West. In the first chapter, Huang explores the impact of the refrigerated rail car on American agriculture. This invention saved substantial shipping costs for livestock, and reduced the risk of weight loss or animal death during transport. Her empirical analysis exploits differences across counties in suitability for livestock versus grain production. Huang finds that the introduction of refrigerated rail cars caused agricultural production to shift towards areas more suitable for ranching that persisted for decades.

These are interesting findings that add to our understanding of the influence of innovation in American agriculture. Whereas previous research has focused primarily on the aggregate impacts of railroads (i.e., Fogel Reference Fogel1964; Donaldson and Hornbeck Reference Donaldson and Hornbeck2016), this paper highlights how refrigeration altered comparative advantage to reshape the geographic patterns of agricultural activity. Given that the potential gains from this technology depend not just on underlying land suitability, but also on proximity to markets, I would encourage Huang to delve more deeply into the sources of these geographic patterns.

Huang’s second chapter is a detailed exploration of Chicago’s meatpacking industry in the early twentieth century. Farmers shipped livestock to be sold to a group of five meatpackers, who openly colluded to manipulate wholesale prices. Huang highlights a fascinating, dynamic element to this collusion. Given lags in shipment, farmers had to pre-commit to sell before observing the spot price, and so were vulnerable to dynamic price manipulation. Huang documents this market collusion through both narrative evidence and descriptive statistics based on weekly price data from the major stockyards.

To estimate the costs of dynamic market collusion, Huang exploits a change in the regulatory environment that caused the cartel to stop holding their weekly price-fixing meetings. Although they could still manipulate markets through static collusion, the cartel could no longer coordinate dynamic pricing. Applying a structural model, she compares outcomes across both regimes to quantify the additional costs associated with dynamic price collusion. She finds that dynamic market manipulation hurt farmers through lower wholesale cattle prices, and imposed significant costs on downstream consumers through higher food prices. This chapter provides an example of first-rate economic history that speaks to current policy, particularly in developing countries, where small rural producers often enter relationships with a handful of powerful buyers.

In her final chapter, Huang assembles county-level data on nineteenth-century fence laws across a number of western states to study how the assignment of liability rules affects resource allocation. She compares “fence out” laws, which assign responsibility to farmers to protect land from livestock incursion to “fence in” laws, which assign responsibility to ranchers. In contrast to Ronald Coase’s (1960) classic paper, her empirical results suggest that the assignment of property rights did, in fact, influence agricultural outcomes.

PAWEL JANAS

The title of Pawel Janas’s dissertation is “Financial Crises and Growth: U.S. Cities, Counties, and School Districts during the Great Depression.” I must admit that when I first read this title, I was skeptical. Given the extensive literature on the Depression, I thought, what more can be learned? I was wrong. Janas provides many fascinating new insights into this historical episode, drawing on an impressive array of novel data sources.

In Chapter 1, Janas investigates how local governments responded to the decimation of their revenue base during the Great Depression. Drawing on newly collected data on city spending and municipal bonds, Janas compares changes in spending outcomes across more or less leveraged cities during the Depression. He also implements a novel strategy based on the timing of bond due dates to compare outcomes across cities with similar debt levels, that were more less exposed to debt repayment during the Depression. Janas finds that financially constrained cities faced downgraded credit ratings during the Depression, and were forced to make significant cuts across a broad category of public spending. Beyond the headline findings, the detailed spending data allow Janas to provide a first-ever in-depth portrait of the operations of local governments through this unique period of financial distress.

In his second paper, Janas studies how adolescents educational choices were affected by the Great Depression. There is a sizeable literature on the elasticity of schooling choices with respect to youth labor market opportunities (e.g., Atkin Reference Atkin2016; Baker, Blanchette, and Eriksson Reference Baker, Blanchette and Eriksson2020). Nevertheless, Janas highlights another mechanism at play during the Depression: the sharp drop in school funding temporarily reduced school quality, potentially lowering the incentive to stay in the classroom. Drawing on several new data sources, Janas provides convincing empirical evidence that the Depression did increase overall educational attainment, but that this effect was partially offset by worsening educational quality. Exploring how educational choices were influenced by within-household dynamics, such as parental job loss or migration, might be an interesting extension of the current analysis.

In his third chapter, Janas studies the role of the Atlanta Fed in affecting local access to credit and economic outcomes by acting as a lender-of-last-resort. His empirical strategy follows prior work (Richards and Troost 2009; Ziebarth Reference Ziebarth2013), but he brings several new sources of data, including county-level measures of pre-Depression financial constraints. He confirms the relationship between Fed policies and local credit conditions that has been documented by other researchers, but finds no evidence that the Atlanta Fed’s policies improved local economic outcomes. These are interesting and surprising findings. As he continues this line of research, I would like to see him delve more deeply into understanding why his results diverge from the prior literature.

SEBASTIAN OTTINGER

Sebastian Ottinger’s dissertation, “Essays on Political Economy and Economic Geography,” is made up of three distinct papers, two of which address fundamental questions in American economic history, while the third focuses on Europe. In the first chapter, Ottinger studies the role of immigrants in shaping the geographic patterns of American economic activity. In doing so, he links two seemingly distinct historical phenomena: the massive inflow of foreign workers during the Age of Mass Migration and the rise of the U.S. manufacturing belt. He motivates the analysis with several case studies, describing the role of immigrant entrepreneurs in establishing local hubs of economic activity.

His empirical analysis is based on the insight that immigrants arrive endowed with different skills depending on their country of origin. He assembles data on comparative advantage across 49 manufacturing industries in 13 European origin countries to capture the “embodiment” of skills among new arrivals. Exploiting the unequal spatial distribution of different immigrant groups across U.S. counties in 1850, he then explores how differences in this measure of “immigrant specialization” affected the geographic patterns of industrial development in the latter nineteenth and early twentieth centuries.

Ottinger finds that the comparative advantage embodied in new immigrants predicts the growth in specific local manufacturing industries in subsequent decades. The early establishment of these industries persisted until well into the twentieth century, potentially due to agglomeration forces that locked in an early advantage. Interestingly, he also finds evidence that greater “immigrant specialization” predicts the entry of pioneer firms into a particular county-industry, and that these new firms were disproportionately owned by immigrants.

This is fascinating research that demonstrates the critical role of immigration in the emergence of the U.S. manufacturing belt. His work also highlights the contingent nature of economic history, which, in this case, depended on the particular destination choices of millions of new arrivals to the country.

In his second paper, coauthored with Max (Winkler) Posch, Ottinger studies how local political leaders responded to the political threat of the formation of new coalitions with minority groups. They study the short-lived electoral success of the Populist Party in the 1892 presidential elections, who sought support from both poor White and Black farmers, threatening the Democratic establishment in the South. Using archival newspaper data, Ottinger finds that Democratic leaders responded to this political threat through anti-Black propaganda. The effects are particularly large in counties with high levels of wealth inequality, where elites presumably have more to lose from redistributionist policies. Sadly, it appears that this propaganda was effective, as it contributed to persistent gains for Democrats in subsequent elections. Overall, I think this work provides important new insights into the determinants of racism and political repression of African Americans in the Postbellum South.

Joshua Lewis, Université de Montréal

LONG-TERM EFFECT OF MECHANICAL REFRIGERATION ON U.S. AGRICULTURAL PRODUCTION

Despite the historical importance of refrigeration, few studies have empirically analyzed its long-term impact on American agriculture. Prior to refrigerated rail cars, livestock needed to be transported alive from farms to urban markets to be slaughtered and sold. The transportation process was costly and risked animals losing weight or dying on the trip. In 1880, meat packers in the Midwest acquired the patent to build the first refrigerated rail car. The invention significantly reduced the shipping cost of beef: carcasses could be shipped for one-third the cost of shipping live cattle. Within ten years, from 1880 to 1890, the number of cattle slaughtered in Chicago more than quadrupled, and the meatpacking industry became the second-largest manufacturing sector in terms of output value.

The research design exploits the variation in relative natural suitability for livestock versus grain production across counties to capture the impact of refrigeration. Perishable goods, such as beef, experienced a drastic reduction in transportation costs after 1880. Meanwhile, this change did not influence other non-perishable products, such as wheat. In other words, counties more suitable for livestock production were more influenced by the new technology, thus allowing an event-study analysis that compares changes in counties more or less suitable for livestock production before and after 1880.

The event study shows that after 1880, when refrigeration was commercially adopted in the meatpacking industry, counties that were relatively more suitable for ranching than farming witnessed more farmland development and higher output value. For every percentile increase in the relative suitability ranking, counties experienced a 0.1 percentage point increase in the share of land areas being developed as farmland and a 0.5 percent increase in output value and land value. The effects also differ for counties across the range of relative suitability but persist over time. The results were driven primarily by the top two quartiles, but the impact on output and land value persisted until 1960.

MONOPSONY, CARTELS, AND MARKET MANIPULATION: EVIDENCE FROM THE U.S. MEATPACKING INDUSTRY

In addition to shifting upstream agricultural production, how does the new technology affect market power and the competition structure in the manufacturing sector? In this chapter, I answer this question by analyzing the American meatpacking industry in the early twentieth century.

Due to the high fixed costs, mechanical refrigeration created a few firms with unprecedented market power in the meatpacking industry. By the early twentieth century, five firms dominated the meatpacking industry. Under weak antitrust enforcement, they formed a monopsonistic cartel to manipulate the wholesale cattle market. The cartel dominated both the input market (cattle) and product market (beef): the five packers purchased 95 percent of cattle sold at the ten largest stockyards and produced more than 80 percent of refrigerated beef for urban markets. In an era of weak antitrust enforcement, they openly colluded to manipulate the wholesale cattle market from 1893 to 1920.

Standard monopsony models focus primarily on static and immediate responses to cartel strategies. However, for markets with substantial time to ship or time to build, current market outcomes may influence future supply decisions and thus make sellers vulnerable to a more complex form of dynamic manipulation. Because sellers must make future production or shipment decisions based on current market information, they must commit to the market before observing the realized spot market price at the time of delivery. If a monopsonistic cartel incorporates the delayed supply responses in the collusive strategy, the canonical static model may fail to properly assess the cartel damage on the market.

Two factors make this historical case particularly well suited to examining the effect of a dynamic monopsonistic cartel strategy. First, because the cartel was eventually challenged in court, the resulting litigation created detailed documentation on the cartel’s manipulation strategies. The court found that the cartel members were guilty of “bidding up through their agents, the prices of livestock for a few days at a time, to induce large shipments, and then ceasing from bids, to obtain livestock thus shipped at prices much less than it would bring in the regular way.” Second, exogenous changes in the regulatory environment forced the cartel to switch from the aforementioned dynamic strategy to a static, fixed market share agreement in 1913, while other features of the market remained unchanged. Thus, I observe the market outcomes under both dynamic and static strategies, but with the same market participants. This allows me to compare the empirical outcomes under the dynamic strategy to counterfactuals suggested by the well-understood static monopsony model.

The main analyses leverage exogenous regulatory changes that forced the cartel to switch from dynamic to static strategies. I first construct and estimate a static model of the cattle wholesale market using data after 1913, when the cartel strategy coincided with the static model. I then use the estimated static model to solve for market outcomes for the dynamic period before 1913. This recovers the counterfactual cattle wholesale prices and cartel quantities as well as downstream wholesale beef prices. The difference between the observed market outcomes under the dynamic strategy and the counterfactuals suggested by the static model is, therefore, the “additional” damage of the dynamic cartel strategy not captured by standard models.

I find two sets of key results. First, regarding the wholesale cattle market, the dynamic strategy causes more damage to small sellers than what is suggested by the standard model. Without cartel manipulation, the average cattle wholesale price would increase by 23.4 percent, which would increase the profit margin by 57 percent for the sellers. The average total quantity purchased by the cartel would also increase by 14 percent, or 15,000 more heads of cattle per week sold at the four stockyards. Second, regarding the downstream wholesale beef market, the dynamic strategy hurts urban consumers by reducing the beef supply and increasing household food expenditure. However, the effects are much smaller: without cartel manipulation, downstream wholesale beef prices would reduce by 6 percent, and total household food expenditures would reduce by $3.6 per year.

FENCE LAWS: LIABILITY RULES AND AGRICULTURAL PRODUCTIVITY

This chapter uses the historical evolution of fence laws in the American West to analyze the long-term effect of liability rules on resource allocation and productivity. In his seminal article, Coase (Reference Coase1960) uses an example between a farmer and a cattle-raiser adjacent to each other to illustrate that the assignment of property damage liability does not affect the allocation of resources. Cattle may stray and destroy the crops on the farmer’s land. Regardless of whether the farmer or the cattle-raiser is legally liable for the trespassing damage, the land allocation between the two types of production should reach the same equilibrium, as long as the liability is well-defined and enforced, and the transaction is costless. However, despite the wide application, most research focuses on the effects of establishing and enforcing property rights. Few empirical works study how the assignment of property damage liability may influence resource allocation.

In the American West, counties assign the liability of protecting against animal trespassing to either farmers or ranchers. Under the “fence-out” rule, farmers can claim damage from owners of the trespassing animals only if they have enclosed the land with fences that satisfy specific regulatory requirements. Meanwhile, under the “fence-in” rule, farmers can claim damage from livestock owners regardless of whether the farms are enclosed with fences. In other words, the liability for livestock trespassing is assigned to farmers in some areas and livestock owners in others. Ranchers and farmers have long contested fence regulations. Prolonged public debates and occasional violent conflicts between farmers and ranchers suggest that this supposedly innocuous rule had profound economic implications.

I compiled data on all county-level fence-related regulation changes from the first state (or territorial) legislature until 1930. To the best of my knowledge, this is the first dataset that fully captures the legal environment for property rights protection during this period. I combine the fence law data with the decennial censuses, measuring land use, land value, output value, and productivity over the past century. This data provides a comprehensive legislative history of liability rules for the western states.

The analysis exploits the county-level fence laws variation over time to quantify the effects of liability rules on agricultural productivity. Consistent with historical accounts, the baseline difference-in-differences results show that fence-in rules incentived agricultural development. Compared to fence-out counties that required farmers to construct fences, the fence-in law increased the density of farmland and the share of improved farmland. By making ranchers liable for trespassing damages, the fence-in rule also increased grain cultivation area. This eventually translates to a higher total value of farm output for fence-in counties, although the higher productivity was not reflected in land values.

CONCLUSIONS

With newly constructed historical data and clear identification strategies, this dissertation provides new causal evidence on the persistent impact of innovation and regulation. It quantifies the long-term economic impact of some of American history’s most important technological and regulatory changes. The results also expand our understanding of the interaction between innovation and regulation and how they can influence market structure and economic development. First, mechanical refrigeration benefited upstream agricultural production, especially in places relatively more suitable for ranching. Meanwhile, the high fixed cost of the new technology also created a highly concentrated market, resulting in welfare losses for both small suppliers and consumers. Finally, while antitrust laws improved welfare by restricting market manipulation, other seemingly innocuous regulations, such as the fence laws, created persistent resource misallocation.

Jingyi Huang, Brandeis University

Jingyi Huang, Assistant Professor, Brandeis University, Lemberg Academic Center, MS 021, Brandeis University, 415 South Street, Waltham, MA 02453. E-mail: jingyih@brandeis.edu. This dissertation was completed at the University of California, Los Angeles under the supervision of Dora Costa (co-chair), John Asker (co-chair), Michela Giorcelli, and Nico Voigtländer.

PUBLIC GOODS UNDER FINANCIAL DISTRESS

In the first chapter of my dissertation, I ask: when faced with financial constraints during a crisis, how much do municipalities adjust local public goods, and how much of this curtailment is due to financial market frictions?

I shed new light on these questions by studying how U.S. cities responded during the Great Depression. Building upon the work of Siodla (Reference Siodla2020), I construct a historical dataset of local public good provision and debt from multiple archival sources on U.S. cities. I also supplement the city-level data with a second database of over 29,000 municipal bonds outstanding in 1929. To estimate the impact of leverage on local public good provision, I compare expenditure in more or less leveraged cities before and after the onset of the Great Depression using a difference-in-differences framework. I find that municipalities in the 75th percentile of pre-Depression leverage saw a 5 percentage point decrease across current expenditures and a 15 percentage point decrease in capital investment relative to cities in the 25th percentile. I find a significant effect on police and firefighting spending and capital investment and quantitatively smaller but significant effects on the other categories, such as health and sanitation.

I then unpack financial leverage into two distinct mechanisms: a causal refinancing supply-side channel and a non-causal demand-side channel. I study the refinancing channel by exploiting the quasi-exogenous timing of bonds becoming due right after the financial market crash of 1929, which led to a collapse of bond markets in the early 1930s (Hillhouse Reference Hillhouse1936). As a result, municipalities could not easily issue new debt to repay the principal owed on bonds that were becoming due during this time, and cities with more of these outstanding bonds were plausibly more constrained in allocating revenue between debt service and public goods. These bonds, however, were primarily issued well in advance of the onset of the Depression, such that the specific timing of these debt-repayment shocks was unlikely to be driven by the demand for new investment during the Depression (Almeida et al. Reference Almeida, Campello, Laranjeira and Weisbenner2009; Benmelech, Frydman, and Papanikolaou Reference Benmelech, Frydman and Papanikolaou2019).

I find that cities with more debt that matured during the Depression curtailed public good provision on capital investment and public service expenditure more than similar cities that did not face the same financial shock. Specifically, one standard deviation in the amount of bonds due is associated with a decline in current spending and capital investment of about 2.5 and 20 percent, respectively. Altogether, these results suggest that roughly 20 percent of the drop in expenditure can be explained through a re-allocation of budgets towards debt repayment. The lack of infrastructure demand from highly leveraged cities, on the other hand, can account for some of the leverage effects in the early years of the Depression but only a small portion of it after 1933. My findings suggest that financial market frictions played a significant role in the decrease in local public goods during the Great Depression in the United States.

CRISES AND INTERGENERATIONAL MOBILITY

In the second chapter of my dissertation, I shift my attention to the impact of reducing one particular public good during the Depression: public education. Specifically, I study the schooling behavior and intergenerational mobility of young urban males in the United States during the 1930s. Broadly, education is one mechanism through which younger generations climb the socioeconomic ladder and sever intergenerational links in economic outcomes—understanding the interaction between macroeconomic shocks and microeconomic mechanisms driving changes in human capital investment is essential for guiding education and labor policy (Goldin and Katz Reference Goldin and Katz1997; Black, McKinnish, and Sanders Reference Black, McKinnish and Sanders2005; Cascio and Narayan Reference Cascio and Narayan2015; Kerwin, Hurst, and Notowidigdo Reference Kerwin, Hurst and Notowidigdo2018). In this chapter, I take advantage of the decentralized public education system of the United States in the 1930s to study how schooling decisions are affected by changes in education quality and the opportunity cost of youth labor. When a crisis hits, do differences in quality and opportunity costs at the local level result in varying levels of educational attainment and, ultimately, intergenerational mobility? If so, who are the winners and losers of crises?

I study these questions by merging newly collected local data on youth unemployment and school quality during the Depression with linked full-count Census records (Abramitzky, Boustan, and Rashid Reference Abramitzky, Boustan and Rashid2020). First, I collect unemployment-by-occupation-by-age data from the Special Unemployment Census of 1931. Since the Unemployment Census canvassed only 18 regionally dispersed cities and 3 boroughs of New York City, I create estimates of youth unemployment for all other cities by taking a weighted average of regional youth unemployment-by-occupation rates, using 1930 occupation-by-city shares aggregated from the 1930 complete count Census as weights. Then, to obtain education quality measures, I digitize biennial records from the Census of Education from the U.S. Office of Education on revenues and expenditures at the city level from 1922 to 1938. I follow the economics of education literature and proxy quality with the change in the total real spending per pupil. I show that expenditure is closely related to the student-teacher ratio, average real teacher wages, and school term length.

My empirical strategy attempts to explain the within-city variation in high school graduation rates across cohorts using a difference-in-differences design using across-city variation in unemployment and public education spending cuts. The strategy thus compares the education outcomes of individuals on the cusp of making secondary schooling decisions during the Great Depression with those who graduated before the Great Depression within the same city, conditional on national trends and static city determinants of educational attainment.

I find that increases in the youth unemployment rate in 1931 significantly increased secondary school graduation rates: one standard deviation (4 percent) increase in unemployment caused a 1.2–1.6 percentage point increase in the graduation rate for the Depression-era cohorts and 0.9–0.12 more years of schooling. On the other hand, cuts in education spending decreased graduation rates by about a quarter of this effect. To put these numbers in perspective, I find that youth unemployment can account for 18–21 percent of the total increase in high school graduation during the 1930s, while education spending cuts attenuated this effect by 4–5 percent.

LENDER OF LAST RESORT AND LOCAL ECONOMIC OUTCOMES

In the last chapter of my dissertation, I focus on banking panics and local manufacturing activity. I use archival panel data on local manufacturing and banking conditions in the United States to investigate the link between policy, bank failures, and firm production and employment. I use the divergent policies enacted by the Atlanta Federal Reserve Bank, first brought to the literature by Richardson and Troost (Reference Richardson and Troost2009) with additional contributions by Jalil (Reference Jalil2014) and Ziebarth (Reference Ziebarth2013), as my empirical laboratory. Unlike the other Federal Reserve banks at the time, Atlanta acted as a lender of last resort inside its region by extending credit to solvent but illiquid banks to prevent bank runs.

Because the Federal Reserve borders bisect states and consumer markets, I can compare the local economic trajectories before and during the Great Depression using the quasi-exogenous incidence of bank failures.

In the first part of my analysis, I investigate the difference in bank failures between counties just outside the Atlanta border and those just inside it and explore its robustness. I find that banks failed less often inside the Atlanta region in 1929 and 1930, even after accounting for pre-existing differences in local banking conditions, excluding outliers and individual border segments, and measuring bank distress in various ways. I perform placebo randomization tests of the Federal Reserve’s borders and find that the baseline results are unlikely to have occurred by chance. Additionally, I do not find much evidence that bank distress differed across the borders of Federal Reserve districts that did not follow different lender-of-last-resort policies. The results, taken together, suggest that credit conditions were more favorable in the early years of the Depression in the counties of the Atlanta district.

I turn to local manufacturing outcomes in the second part of my analysis. By combining an industry-level credit survey with pre-Depression industry-by-county data, I construct proxies for the financial constraints of small and medium-sized manufacturers in each county around the Atlanta region border. The survey shows that most of these manufacturers relied on commercial banks to finance both working capital and long-term investment, but the degree of these constraints varied by industry. By exploiting the geographical differences in the types of manufacturing along the Atlanta border, I compare economic activity across counties with fewer and more banking failures. Since commercial banks survived at a higher rate inside the Atlanta region, I hypothesize that manufacturing output and employment shrank less and recovered quicker there than in counties just outside the Atlanta region. Furthermore, I test whether the results are strongest for counties with firms in highly constrained industries.

I do not find evidence to support this hypothesis, which contrasts with the results of the existing literature. First, manufacturing outcomes were worse, not better, in counties inside the Atlanta region, despite having more banking resources. Second, I find evidence that the county-level financing constraints, on average across all counties, predict worse outcomes after, but not before, the Depression. Yet, the combination of pre-Depression measures of financial constraints and banking panics during the Depression was not an important determinant of local economic outcomes, as one might expect. Even though the Atlanta Federal Reserve’s policies strengthened the banking sector early in the Depression, there was no positive downstream effect on firms, possibly due to the banking sector’s reluctance to extend credit.

Pawel W. Janas, California Institute of Technology

Pawel W. Janas, Assistant Professor, California Institute of Technology, 135 Baxter Hall, MC 228-77, Pasadena CA 91125. E-mail: pawel@caltech.edu. This dissertation was completed at the Kellogg School of Management at Northwestern University under the supervision of Paola Sapienza, Carola Frydman, Joel Mokyr, and Scott Baker.

IMMIGRANTS, INDUSTRIES, AND PATH DEPENDENCE

Manufacturing industries cluster in space. Examples abound—in the past and present, in the United States and elsewhere. Detroit emerged as the nation’s center of the automobile industry and remains so today. In the twentieth century, San Francisco Bay became “Silicon Valley.” In the twenty-first century, Shenzhen—on a bay on the other side of the Pacific—became a global center of smartphone production. Jena, a medium-sized city in eastern Germany, arose as a center of the optics industry in the nineteenth century and remains so today.

Why do specific manufacturing industries end up clustering in particular regions? Economists have advanced three explanations and tested two empirically. The first emphasizes the natural resources of an area that might render it particularly suitable for specific industries. Second, a region’s market access to consumers and suppliers of crucial inputs might do so as well. However, while these explanations can account to some extent for why particular regions specialize in specific industries, a considerable amount of regional specialization remains unexplained by both. A third, hitherto untested explanation is commonly advanced in a diverse plethora of case studies for specific industries (Krugman Reference Krugman1992), supported by established economic theory (Arthur Reference Arthur1990). Take two identical regions. Suppose, randomly, one of them gets an early start in a particular industry, and agglomeration forces are sufficiently strong. In that case, this region will emerge as a center of that industry and remain so. Co-agglomeration of industries along this industry’s supply chain further cements the “lock-in” of this industry there. However, empirical evidence supporting these theories and anecdotal accounts is scarce.

In the first chapter of the dissertation, I provide such empirical evidence, leveraging the co-occurrence of two major developments in U.S. economic history in the nineteenth century. From this century on, the United States emerged as the most technologically advanced nation in the world, with a high degree of geographic specialization across its regions. Simultaneously, the United States attracted large numbers of immigrants from all over Europe. I create a measure of the potentially embodied industries of early or later immigrants that affect local employment patterns. This measure, henceforth immigrant specialization, combines the settlement patterns of all European origins in 1850 with the immigrants’ origin specialization in all manufacturing industries based on their international trading patterns.

Immigrant specialization was a significant, sizable, and robust factor in employment patterns across U.S. counties in the early twentieth century, holding fixed initial employment patterns as of 1850, as well as other well-documented determinants of local specialization, such as market access or natural resource endowments. This baseline association is stronger in the novel industries of the Second Industrial Revolution and for immigrants from Germany, which was emerging as an industrial powerhouse in those industries only after 1850.

I show that this was more than an episode of immigrants bringing their industries immediately. Rather, early immigrants induced local specialization through exchange with their origins after arrival and by attracting later immigrants. Several pieces of evidence support this. First, the baseline association is even stronger for some industries that did not exist by 1850, such as electrical manufacturing or the auto industry. Second, on the frontier of settlement around 1850, the association between immigrant specialization and employment patterns is negative, indicating, if anything, negative selective migration to those counties. An instrumental variable strategy based on the movement of the frontier and the aggregate arrivals of European origin in the prior decade supports this. So does the historical record. Early immigrants to this frontier mainly were farmers, and major cities and industry specialization in these only formed decades after 1850.

The rich data available in the complete count census allows me to provide direct evidence for the mechanism underlying the baseline association of how early immigrants induced local specialization. To do this, I identify the pioneer entrepreneurs in each county and industry, that is, the first individual recorded as an owner or manager in a particular industry for each county. Immigrant specialization predicts the pioneers’ entry, mostly after 1880. Their early entry helped select equilibria in county industries, shaping their destinations’ specializations for generations.

The traces of the early immigrants can still be found in specialization patterns today. Immigrant specialization remains a strong predictor of employment patterns in 2000 for those industries that emerged after 1850, particularly on the frontier of settlement then, where early immigrants found a tabula rasa and had the most opportunity to imprint local specialization.

This paper exploits only one of potentially many shocks that might lead some locations to have an early start in an industry and build their comparative advantage in those to form a cluster. It lends direct empirical support to the importance of such early shocks and subsequent path dependence in shaping regional specialization in manufacturing.

THE POLITICAL ECONOMY OF PROPAGANDA: EVIDENCE FROM U.S. NEWSPAPERS

The second chapter of the dissertation, co-authored with Max (Winkler) Posch, studies the origin of hateful media content, drawing on the U.S. South after Reconstruction.

The People’s Party, with its aim of redistributing from rich to poor, posed a threat to the Democratic establishment in the South. The Populists, as the People’s Party was known, too, remained ambiguous regarding their position on race. This provided the segregationist Democratic establishment in the South with an incentive to fan racial outrage to alienate white voters from the Populists.

We use text data from local newspapers to identify hateful anti-Black content and show that its prevalence increased after the 1892 Presidential election, at which the Populists first ran on a nationwide platform, and in counties in which the Populists likely gained votes at the expense of the Democrats.

Additional results suggest a strong political motive behind this increase, indicating that Democratic political elites supplied this hateful content to further their electoral prospects and remain in control of local politics. All of the increase is driven by newspapers affiliated with the Democrats, and triple-difference results document that even within counties where at least one Democrat-affiliated newpaper and other newspapers are available, this increase is present in Democrat newspapers only. On the other hand, we find no effect for independent newspapers, which should have (profitably) reacted to local swings in demand as well. This and a battery of other results suggest that political elites used newspapers affiliated with them to supply propaganda to split their competitors’ potential electorate.

While politics may remain a source of hateful media content in the present day and other contexts, the setting of the U.S. South after Reconstruction, with newspapers as tightly politically controlled media (Gentzkow et al. 2015), the lack of alternative media forms—since radio and TV had yet to be invented—and the fact that newspapers served highly local markets, offers an optimal testing ground for theories of divide and rule by elites for political gain (Glaeser Reference Glaeser2005).

HISTORY’S MASTERS: THE EFFECT OF EUROPEAN MONARCHS ON STATE PERFORMANCE

The third and final chapter of my dissertation, co-authored with Nico Voigtländer, is set in European economic history and studies the effect of national leaders on the performance of the states they govern. While scholars since the nineteenth century have pondered the importance of national leaders, economists have brought identification to this debate (Jones and Olken Reference Jones and Olken2005), exploiting the accidental deaths of modern heads of state. We turn to the historical leaders before Napoleon and advance a new identification strategy to identify the effect of national leaders’ cognitive ability on the performance of their states.

We create a novel reign-level dataset for European monarchs, covering all major European states between the tenth and eighteenth centuries. We document a strong positive relationship between rulers’ cognitive ability and state-level outcomes. To address endogeneity issues, we exploit the facts that (i) rulers were appointed according to hereditary succession, independent of their ability, and (ii) the widespread inbreeding among the ruling dynasties of Europe over centuries led to quasi-random variation in ruler ability. We code the degree of blood relationship between the parents of rulers, which also reflects “hidden” layers of inbreeding from previous generations. The “coefficient of inbreeding” is a strong (negative) predictor of ruler ability, and the corresponding instrumental variable results imply that ruler ability had a sizeable effect on the performance of states and their borders.

This supports the view that “leaders made history,” shaping the European map until its consolidation into nation-states. We also show that rulers mattered only where their power was largely unconstrained. In reigns where parliaments checked the power of monarchs, a ruler’s ability no longer affected their state’s performance. Thus, the strengthening of parliaments in Northern European states (where kin marriage of dynasties was particularly widespread) may have shielded them from the detrimental effects of inbreeding.

Sebastian Ottinger, Assistant Professor, CERGE-EI, a joint workplace of Charles University and the Economics Institute of the Czech Academy of Sciences, Politickych veznu 7, 111 21 Prague, Czech Republic. E-mail: sebastian.ottinger@cerge-ei.cz. This dissertation was completed under the supervision of Nico Voigtländer, Christian Dippel, Paola Giuliano, and Romain Wacziarg at the Anderson School of Management, University of California, Los Angeles.

THE CRITERIA

First off, it is important to emphasize that every dissertation submitted for the 2022 prize constituted a package of high-quality research and industrious efforts on the part of the candidates. For that, I congratulate each and every candidate. In order to begin to differentiate among the submissions, I set up a rubric of sorts and then rated each dissertation on a scale of 1 to 5.

My rubric assessed the originality of the topic or question being asked, the novelty of the data collection, empirical and analytical rigor, and quality of writing. I paid special attention to the demonstration of intellectual breadth and independent work and looked for dissertations that employed the methodological approaches of both economics and history and, preferably, would answer questions of wide interest. In other words, the ideal dissertation would be a sole-authored piece of work that took up an enduring dilemma or raised a new question that the field should be grappling with but has not. This model dissertation would then develop an integrated approach based on the latest methods from economics, combined with rich, historical, and institutional detail, and then proceed to unearth and analyze novel data sources. Indeed, it is a tall order to even conceive of such a dissertation, never mind execute it successfully. More than three of the dissertations submitted actually cleared this hurdle, but three of them stood out as truly outstanding examples of the best of what economic history research can offer.

THE FINALISTS

I give a very brief overview of the major contributions of each finalist’s dissertation, presented in alphabetical order.

COMMON ELEMENTS OF EXCELLENCE

To summarize briefly the accomplishments of the three finalists for the 2022 Gerschenkron Prize, I underscore the common elements of excellence across the three works. All of the finalists asked big and deep questions, demonstrating intellectual curiosity and an unusual level of independent and richly-detailed work. Only one chapter of one of the three dissertations had co-authorship, and that stemmed from an enormous, externally-funded, collaborative data collection effort. The three dissertations all opened up a “black box” of aggregate, long-run patterns with the painstaking collection of truly novel micro- or municipal-level data and the use of rigorous analysis, both statistical and institutional. By coincidence, all three dissertations related to taxation, either as the topic itself or as the evidentiary basis that elucidated the topic of concern, such as wealth inequality. This fact points to the treasure trove of original source material that likely awaits diligent and energetic economic historians willing to undertake the painstaking work of unearthing, cleaning, and analyzing such records in myriad countries over a span of many centuries.

It was a real challenge to decide on and then among the finalists for the 2022 Gerschenkron Prize. I extend my congratulations to all 24 candidates who submitted their work, the three finalists, and especially to Dr. Hanzhi Deng as the ultimate recipient of the prize.

Caroline Fohlin, Emory University

FORCED DECENTRALIZATION: TAIPING REBELLION AND LOCAL INDIRECT TAXATION

The early-nineteenth-century Qing public finance had suffered a chronic malaise, and the abrupt Taiping Rebellion (1851–64) triggered the unprecedented transitions. The rebellious regime failed to attract gentry elites and mass people, whereas the Qing state took this opportunity to strengthen its legitimacy and capacity. The precarious central court abandoned the ancien régime and granted local governments the greatest fiscal-military autonomy in exchange for dynastic longevity. Hence local governments not only established new armies and militias but also introduced a novel tax, the lijin, to finance their military actions. Lijin was an indirect tax on transported goods, usually levied at the transportation hubs on key roads and waterways.Footnote ¹⁴ The Qing regime survived the Taiping crisis with timely lijin income, and local governments firmly preserved their lijin funds in the postwar decades.

This chapter quantifies the prefectural intensity of Taiping warfare with archival materials;Footnote ¹⁵ it uses available surveys to map the rise, spread, and postwar persistence of the lijin institutions at the local level, too.Footnote ¹⁶ It presents an internal “war making state” story by constructing the link between the Taiping Rebellion and the rise of indirect taxation, through which the de facto local fiscal system gradually took shape.

INDIRECT TAXATION: DOMESTIC CUSTOMS, MARITIME CUSTOMS, AND LIJIN INSTITUTIONS

This chapter continues discussing indirect taxation and expands the scope to all types of late imperial indirect taxation, namely lijin, domestic customs (changguan) and maritime customs (yangguan).Footnote ¹⁷

Before 1850, the Qing state had maintained a domestic customs network, but it was only able to capture very few long-distance flows of goods, thereby making a limited fiscal contribution. After 1850, two novel institutions—lijin by local governments, and maritime customs as an independent apparatus under Sir Robert HartFootnote ¹⁸ —formed an unintentional “duopoly” and served as two new cash cows for the empire because of their distinct tax bases and clearcut accountability, while the old domestic customs were crowded out. From 1850 to 1900, the overall Qing indirect tax revenue achieved a steady growth, the share of which in total government income grew from less than 30 percent to over 60 percent. The ratio of tax to GDP increased significantly, too.

Furthermore, this chapter finds an interesting competition between lijin and maritime customs due to the commutation tax (zikoubanshui) scheme, in which the merchants engaged in import-export trade could choose to pay either lijin (to local agents) or commutation tax (to maritime customs). This competition fully manifested the rationality and pragmatism of lijin and maritime customs, as both applied great efforts to attract more taxpayers, such as lowering tax rates and simplifying clearance procedures.

FOREIGN BORROWING: LOCAL FISCAL CAPACITY AND THE ADOPTION OF DEBT FINANCE

This chapter connects taxation to debt finance. How did the Qing state tackle fiscal shortfalls? Before 1850, the central court was responsible for all local exigency management and interregional resource redistribution. However, after 1850, the steadily growing tax revenue and the access to foreign capitals via the treaty port system enabled local governments to employ foreign loans, secured by local future revenue, to mitigate their fiscal constraints. This chapter studies the fundamentals of over 230 late Qing foreign loans during 1853–1912;Footnote ¹⁹ it links the foreign borrowing data to the late Qing taxation data of previous chapters and finds a strong relationship between taxation capacity and borrowing capacity at the provincial level.

At the initial stage, local officials raised small short-term loans without ex ante central consent for military exigencies, but in the long term, as local governments became more accountable to a wide range of public affairs, they began to launch developmental projects such as telegraph and railway construction and intentionally sought funds from foreign banks. This chapter finds that such “developmental” loans accounted for the bottom-up introduction of China’s modern infrastructure and expanded the scale of public budgets permanently. Through a typology, this chapter deduces that most loans with long-term prospects for public interest were initiated by local governments, rather than the central court, in the late Qing era.Footnote ²⁰

PUBLIC SPENDING: LOCAL FISCAL AUTONOMY AND SELF-STRENGTHENING INDUSTRIALIZATION

Expedient transitions during the Taiping crisis unexpectedly reshaped the Qing intergovernmental fiscal relations. From then on, the central court refrained from monitoring the national budget and focused only on a limited number of affairs, such as central official salary payment and royal household consumption, for which it required provinces to send fixed annual portions to Beijing. Once the latter had fulfilled their obligations under this apportionment scheme (tanpaizhi), they could control excessive revenues, and their local budgets were no longer supervised by Beijing. Hence, they were incentivized not only to cultivate new fiscal sources but also to rationalize local spending structure. During 1850–1911, the share of central spending in the total government budget declined sharply, while local spending on modern military investments, local infrastructure, etc. grew steadily in both amount and share. Diverse spending patterns across provinces well manifest the nature of fiscal decentralization.

Among various spending accounts, this chapter studies an important aspect—local industrial investment since 1860—by compiling relevant sources and constructing a dataset of over 4,000 enterprises.Footnote ²¹ Local governors with local funds introduced modern industries for both defense purposes and long-term economic gains. This Self-Strengthening Movement covered a series of heavy and light industries and marked the start of China’s native industrial modernization. Official-led arsenals and shipyards intentionally brought in Western technological and managerial expertise and thus offered an external economy of scale for private firms. Improvements of infrastructure, such as railways and telegraphs, proposed and financed mainly by local governments, were also conducive to a conceivable industrial takeoff for not only the late Qing but also the early republican periods. In a word, when the central court remained conservative, local governments resorted to a more dynamic and developmental statecraft.

CONCLUDING REMARKS

Decentralization during the Taiping crisis was a triggering condition for late Qing fiscal transitions. It reshaped the objectives, tasks, endowments, and constraints of local actors and hence initiated a series of irrevocable transitions in taxation, borrowing, and spending. The central court gradually refrained from fiscal-military imperatives and acquiesced to such transitions;Footnote ²² by contrast, local governments as “stationary bandits” were sufficiently incentivized to undertake such reforms towards a modern state.

From a quantitative perspective, transitions engendered substantial changes to the fiscal structure at the aggregated level. By 1850, the national fiscal revenue had been approximately 40 million silver taels,Footnote ²³ over 70 percent of which had been land tax revenue. Regarding expenditure, the center had monitored over 75 percent of the national budget, while local governments had been stipulated to retain a rather small fund. After five decades, the national revenue more than doubled to nearly 100 million taels in 1900, and the share of land taxation declined to 40 percent. Novel indirect taxation became the mainstay of Qing finance and contributed to half of total government revenue. Furthermore, foreign borrowing became increasingly important: the ratio of annual borrowing to tax revenue reached 13 percent in the 1900s.Footnote ²⁴ In terms of spending, the central court handled less than 15 percent of the budget, with the remainder in local hands. Their new spending items accounted for 50 percent of the national budget.

The post-1850 fiscal expansion was more than evident, and local governments made the major contribution. During 1850–1900, per capita tax burden rose from 0.1 to 0.25 silver taels, and the ratio of taxation to GDP increased from 0.8 percent to 1.4 percent. Although it was below the levels of the Western benchmarks, the progress per se was notable and accelerated in the twentieth century.Footnote ²⁵ China embarked on an onerous but irreversible path of state modernization, and a small, rigid, and agrarian state in line with Confucian statecraft would never have a second opportunity.

Hanzhi Deng, Fudan University

Hanzhi Deng, Postdoctoral Researcher, Department of Political Science, School of International Relations and Public Affairs, Fudan University, 220 Handan Road, Shanghai, 200433, China. E-mail: hanzhideng@fudan.edu.cn. This dissertation was completed under the supervision of Kent G. Deng, London School of Economics and Political Science.

THE INEQUALITY DEBATE: REVISING THE KUZNETS CURVE?

Chapter 1 reviews the evidence and debate surrounding inequality and its causes in the past. The most widely accepted explanation of rising inequality in the past has been the “Kuznets curve” (Kuznets Reference Kuznets1955). Named after economist Simon Kuznets, who first hypothesized that rising inequality in the first phases of the industrial revolution and declining inequality thereafter were driven by sectoral productivity differences, the Kuznets curve had become the dominant paradigm for understanding economic inequality well into the 1980s (Atkinson Reference Atkinson1997). While Kuznets’ model seemed to hold true for much of the post-war period, the recent resurgence of inequality in Western countries has led to questions about its validity. Studies of wealth and income inequality show that inequality was already high and rising during the early modern period, before the onset of modern economic growth (Alfani Reference Alfani2015; Ryckbosch Reference Ryckbosch2016; Santiago-Caballero Reference Santiago-Caballero2011; Van Zanden 1995). Epidemics, warfare, revolt, and state collapse have been suggested as the main forces reducing inequality in pre-industrial times (Scheidel Reference Scheidel2017).

Five Hundred Years of Inequality Data

Chapter 2 introduces the two databases collected from archival and published sources that form the foundation of this dissertation. The first database, Wealth Inequality in Pre-Industrial Germany (WIPIG), comprises more than 100,000 household-level observations of wealth tax payments across 30 cities and 70 rural communities in the period 1300–1850. Data have been retrieved from a number of archives in Germany as well as published records of archival sources, including those found in studies by the German Historical School. These data were collected in close collaboration with Felix Schaff (EUI) and Guido Alfani (Bocconi University). The second database, Germany City Budgets (GCB), comprises 2,200 city-year observations across more than 40 cities in the period 1350–1800. It is based on records of city budgets, detailing revenue and expense streams at the aggregate and disaggregate level—that is, including a breakdown of income streams from taxation, credit, and other revenue sources as well as a breakdown of expenditure on interest payments, construction, administration, and warfare. It also includes 1,000 observations on urban wealth tax rates between 1350 and 1800. To assess the representativeness of these datasets, I constructed a reference universe of towns based on data from Bairoch, Batou, and Chèvre (1988), De Vries (1984), Wahl (Reference Wahl2016), and the Deutsche Städtebuch (Keyser Reference Keyser1939–1974). I show that both datasets are broadly representative of the larger universe of towns in pre-industrial Germany along characteristics such as population size, geography, religion, political institutions, and status within the wider Empire.

THE FOUR PHASES OF WEALTH INEQUALITY AND POVERTY IN PRE-INDUSTRIAL GERMANY

Chapter 3 is a joint work with Felix Schaff (EUI) and Guido Alfani (Bocconi University) and has been published in this Journal (Alfani, Gierok, and Schaff Reference Alfani, Gierok and Schaff2022). It presents the first estimates of wealth inequality for pre-industrial Germany for the period between 1300 and 1850. It shows that wealth inequality followed four phases: a decline in inequality after the Black Death around 1350 that lasted until circa 1450, a considerable rise in inequality over the long sixteenth century, reaching its peak before the outbreak of the Thirty Years’ War in 1618–1648, a precipitous fall in inequality during and after this devastating conflict, and again a slow rise starting in 1750. It contrasts these findings with available evidence for other European regions, such as pre-industrial Italy, the Netherlands, and Spain. The Thirty Years’ War distinguishes Germany from Italy, the Netherlands, and Spain, which experienced consistently rising inequality after the Black Death. The chapter not only shows that this war and the plagues and famines that followed in its wake were particularly destructive and long-lasting compared to other pre-industrial wars. It also provides novel evidence of the severe economic impact of the war and the broader seventeenth-century crisis.

Given the considerable effect of the Thirty Years’ War on inequality in Germany, Chapter 4 further investigates this effect by tracing the development of the poorest stratum of the population. It provides the first consistent poverty estimates for pre-industrial Germany for the period between 1300 and 1800. It argues that measuring poverty via three measures—absolute poverty, relative poverty, and the Poverty Gap Index—overcomes a roadblock that has prevented researchers from producing consistent long-run estimates previously. Poverty moved largely in tandem with inequality throughout the pre-industrial period. However, the Thirty Years’ War led to a sharp rise in poverty rates while inequality declined. Urban communities experienced this spike first, whereas rural communities saw poverty rise towards the end of the conflict. Considerable migration into the cities, which provided a safe haven for many rural dwellers, is one possible explanation for this effect. While cities were spared destruction, the countryside suffered considerably. Destruction of cultivated land, the confiscation of livestock and capital, and the extortion of financial aid are potential explanations for the rising poverty rates in rural Germany. The war had long-lasting effects on the German economy, however, and although inequality and poverty rates declined after the war, the full consequences for the economy are still underexplored. This chapter concludes by observing that from circa 1750 onward, poverty started to increase again—much earlier than the onset of modern economic growth and industrialization.

THE ROLE OF GUILDS AND GENDER: A CASE STUDY OF FREIBURG

Having established the broad outlines of inequality and poverty, Chapter 5 explores the role of gender and guilds by focusing on the case study of Freiburg between 1481 and 1675. Both factors have been highlighted in the literature as important factors influencing economic growth and the distribution of resources. Freiburg presents an ideal case study to explore both aspects because it is a mid-size town in southwestern Germany characterized by strong guilds. An analysis of Freiburg’s tax registers between 1481 and 1675 reveals the following: First, female headship rates in Freiburg were on the lower end of the spectrum compared to similar cities such as Wildberg. This suggests that guild-dominated cities limited female independence. Only during and in the aftermath of the Thirty Years’ War did female headship rates increase. This increase is notable across all guilds, but guilds in the textile trade report particularly high female headship rates. Second, among the non-guild households, women make up a disproportionally high percentage. These non-guild households had very limited economic opportunities; women lived more precarious lives than men and were more frequently affected by poverty. This is further confirmed by comparing the distribution of tax payments along gender dimensions. Third, an analysis of tax revenues by guild shows that their contributions remained fairly stable over time, with the exception of the tailors and butchers guilds, whose contributions declined substantially. In sum, this suggests that strong guild dominance influenced the distribution of resources within a city—often to the detriment of women.

CITIES OF EXTRACTION: FISCAL PRESSURE IN URBAN GERMANY

Chapter 6 investigates the role of taxation in public finances and provides the first estimates of fiscal extraction for a representative sample of German cities. Prior studies of fiscal capacity have focused on Prussia and Habsburg Austria and measured fiscal extraction solely through per capita tax revenues. In this chapter, I argue that we should measure fiscal capacity along a broader range of measures: (1) type and number of taxes, (2) tax design and coverage, (3) tax rates, (4) the share of expenses on public administration relative to the entire budget, (5) the tax share relative to total revenues, and (6) per capita tax pressure in local currencies as well as gold. The latter measure involves finding conversion rates for a large number of local currencies—another contribution of this chapter. The resulting estimates show that fiscal extraction rose considerably from circa 1550 onward. This was achieved by raising additional wealth and consumption taxes and increasing tax rates. While extraction peaked during the Thirty Years’ War (1618–1648), this rise in extraction six decades before the Thirty Years’ War questions the state-and-war-making paradigm, which explains state-making primarily as a function of warfare. Finally, this chapter shows that the tax pressure on citizens in urban Germany was considerable compared to the tax pressure in other European regions that were fiscally more advanced European regions, such as England and the Netherlands.

Victoria Gierok, University of Oxford

Victoria Gierok, British Academy Post-Doctoral Fellow, Department of Economics, Manor Road Building, Manor Road, Oxford, OX1 3UQ, United Kingdom. E-mail: victoria.gierok@economics.ox.ac.uk. This dissertation was completed under the supervision of Stephen Broadberry at Oxford University.