Pawel Janas

This project’s objective is to understand the evolution and economic role of life and property insurance companies in the U.S. during the 19th and 20th centuries by creating a new, comprehensive firm-level panel database. To the best of my knowledge, disaggregated information about the insurance industry before 1980 is sparse despite the economic importance of these financial intermediaries in public and private debt markets. I digitize publicly available historical documents from regulatory filings and industry publications between 1880 and 1950 and collect income statements, balance sheets, security-level holdings, and state-level mortgage and real estate investments from all insurance companies conducting business in the state of New York (95% of aggregate).

This project’s objective is to create a comprehensive, long-run database of local public finance in California by digitizing and harmonizing municipal, county, and special district fiscal records spanning the past century. To the best of my knowledge, consistent, disaggregated, annual-frequency information on local government finances in California prior to the 2000s remains highly fragmented, despite the central role these governments play in financing infrastructure, public services, and local economic development. By collecting annual financial statements for thousands of local jurisdictions, this project will produce the first unified panel covering roughly 100 years of local fiscal activity in the state. Such a dataset would enable new research on how institutional changes, such as Proposition 13, altered local fiscal capacity and long-term infrastructure investment.

This paper documents the financial geography of U.S. life insurance between 1880 and 1940. Using a newly digitized firm-by-state panel of premiums and losses, together with the balance sheets of New York-licensed insurers, I distinguish two geographies: where insurance was sold and where the firms that collected the premiums were domiciled. The industry grew into a national market for household saving. Firm entry spread beyond the Northeast into the Midwest and South after 1900, and local insurance markets became substantially more competitive, with within-state concentration falling by half. Yet this decentralization was incomplete. Most states continued to send the majority of their premiums to firms headquartered elsewhere, and premium-weighted control remained concentrated in a small number of older financial centers. The same regional hierarchy appeared in household demand, which was persistently highest in the Northeast and lowest in the South. Life insurance was thus neither a static oligopoly nor a purely local institution, but a national savings system that combined broad geographic participation with enduring spatial concentration in financial control.

Large financial intermediaries are more stable than small ones, and the largest are also the most geographically diversified. Whether that breadth is a source of stability or a byproduct of scale is difficult to establish, since intermediaries diversify by growing large. I study how the diversified intermediary is assembled, using a newly digitized firm-by-state panel of the near-universe of U.S. life insurers from 1888 to 1940, the largest nonbank institutional investors of the period. Over these decades the number of insurers per state more than quadrupled and within-state concentration fell by half, yet a few extensive insurers held a stable share of national premiums. They entered new states before they grew large, each entry a small and favorably selected foothold with the convex payoff of a real option. Expansion was costly on two margins: a broad insurer earned less per dollar of coverage, chiefly because of where it wrote rather than how it priced, and paid 15.1 percent more in claims than an equally large insurer writing in the same state and year. Stability, by contrast, was scale, since breadth’s advantages in volatility and survival vanish once size is held fixed, and pooling was bounded, since a broad insurer’s loss-ratio volatility stayed about four times what independent markets would imply.

Forthcoming at the Financial History Review

This paper studies the long-run labor market consequences of lender-of-last-resort (LLR) intervention during the Great Depression. I exploit a natural experiment created by the Federal Reserve district border separating counties under the jurisdiction of the Atlanta and St. Louis Federal Reserve Banks. During the banking panic of 1930, the Atlanta Fed aggressively extended liquidity to distressed banks, while the St. Louis Fed largely refrained from intervention. Using newly digitized county-level manufacturing data and linked individual-level census records from 1930 and 1940, I examine how exposure to this liquidity support affected local economic activity and worker outcomes. Counties within the Atlanta district experienced fewer bank failures and stronger manufacturing performance in the early 1930s. These differences translated into persistent labor market effects: individuals in treated counties were more likely to remain in manufacturing employment and less likely to migrate across state lines by 1940. The results suggest that financial stabilization policies can shape the long-run allocation of labor across regions and sectors.

[2026] Journal of Financial Economics, 176, 104205

I examine the effects of public debt on municipal services and real outcomes during financial crises using a unique archival dataset of U.S. cities from 1924 to 1943. Unlike today’s countercyclical fiscal policies, the Great Depression provides a rare setting to observe fiscal shocks without substantial intergovernmental or Federal Reserve support. My findings show that financial market frictions – especially the need to refinance debt – led cities to sharply cut expenditures, particularly on capital projects and police services. As urban development halted during the Depression, cities with high pre-crisis debt levels faced significant austerity pressures, a decline in population growth, a rise in crime, and a departure of skilled public servants from municipal governments.

Replication Package

[2026] The Journal of Economic History, 86(2), 501–536.

Between 1910 and 1940, the high school graduation rate in the United States increased five-fold, setting the stage for human capital-led economic growth throughout the 20th century. This study examines the effects of the Great Depression’s surge in youth unemployment on educational attainment during the 1930s, with a focus on gender and socioeconomic disparities. Using data from the 1940 Census and novel city-level unemployment rates, the analysis shows that increased youth unemployment significantly boosted high school and post-secondary completion rates among young males, particularly those from higher socioeconomic backgrounds. In contrast, the effect on females and lower-income youths was negligible. I find minimal short-term labor market impacts by 1940. The results highlight the critical role of household resources in leveraging educational opportunities during the Great Depression and suggest that financial constraints may have prevented disadvantaged groups from benefiting equally from reduced opportunity costs during a crucial period during the high school movement.

Replication Package

[2025] with Scott Baker and Lorenz Kueng, Journal of Public Economics, 242, 105275.

​Empirical research in public economics, including our own, often uses variation in state and local taxes as an empirical laboratory to estimate causal relationships. A key concern is that other taxes might change at the same time. To assess this concern, we develop a dataset of state (1977–2022) and local (2000–2022) tax rates and revenue from personal income, corporate income, property, sales, and excise taxes. This new dataset generates two key results. First, we find that taxes of different types tend to co-move within a jurisdiction: a tax change of one type can more than double the likelihood of a second tax type changing in the same year. Local tax changes also co-move with tax changes enacted by the state they are located in. This positive correlation can upwardly bias elasticity estimates, but only moderately. For example, regressing state economic outcomes on the full set of state tax changes yields elasticities that are about 10%–30% smaller than those obtained from using a single tax type in isolation. Second, we document that the mix of taxes across state and local jurisdictions is very different, and that these differences have become more pronounced over time as jurisdictions have increasingly become reliant on the single tax type — sales, personal or corporate income tax — that was most prominent for them in the earliest part of our sample.

[2023] Summaries of Doctoral Dissertations for the Nevins Prize. The Journal of Economic History, 83(2).