James J. Feigenbaum, Princeton University and NBER; James Lee, Cornerstone Research; and Filippo Mezzanotti, Northwestern University

Capital Destruction and Economic Growth: The Effects of Sherman's March, 1850-1920

Using General William Sherman's 1864-65 military march through Georgia, South Carolina, and North Carolina during the American Civil War, this paper studies the effect of capital destruction on short- and long-run local economic activity, and the role of financial markets in the recovery process. Feigenbaum, Lee, and Mezzanotti match an 1865 U.S. War Department map of Sherman's march to county level demographic, agricultural, and manufacturing data from 1850-1920 U.S. Censuses. They show that the capital destruction induced by the march led to a large contraction in agricultural investment, farming asset prices, and manufacturing activity. Elements of the decline in agriculture persisted through 1920. Using information on local banks and access to credit, the researchers argue that the underdevelopment of financial markets played a role in weakening the recovery.

Using General William Sherman's 1864-65 military march through Georgia, South Carolina, and North Carolina during the American Civil War, this paper studies the effect of capital destruction on short- and long-run local economic activity, and the role of financial markets in the recovery process. Feigenbaum, Lee, and Mezzanotti match an 1865 U.S. War Department map of Sherman's march to county level demographic, agricultural, and manufacturing data from 1850-1920 U.S. Censuses. They show that the capital destruction induced by the march led to a large contraction in agricultural investment, farming asset prices, and manufacturing activity. Elements of the decline in agriculture persisted through 1920. Using information on local banks and access to credit, the researchers argue that the underdevelopment of financial markets played a role in weakening the recovery.

---

Trevon Logan, Ohio State University and NBER

Do Black Politicians Matter?

This paper exploits the unique history of Reconstruction after the American Civil War to estimate the causal effect of politician race on public finance. Drawing on an extensive review of the historical literature, Logan overcomes the endogeneity between black political leadership and local political preferences, demographics, economic conditions, and political competition using the number of free blacks in the antebellum era (1860) as an instrumental variable (IV) for black political leaders during Reconstruction (1867–1877). While the instrument is well correlated with the number of black officials, Logan shows that it is not related to electoral outcomes, the tenure of black elected officials, nor political competition and voter education campaigns during the Reconstruction era. IV estimates show that a one standard deviation increase in the number of black officials in a Southern county increased per capita county tax revenue by 0.62 standard deviations, a sizable effect. At the end of Reconstruction, however, the effect of black politicians entirely reverses — the same increase (which, after Reconstruction, is a decrease) in black politicians decreases per capita county tax revenue (1880–1870) by 0.86 standard deviations. Finally, Logan investigates whether the results are consistent with the policy objectives of black political leaders during Reconstruction, where black officials favored higher taxes to establish public education and initiate land reform. While he finds no effects of black politicians on land redistribution, estimates show that exposure to black politicians during school age increased black literacy more than 6% and decreased the black–white literacy gap by more than 7%. These results suggest that black political success during Reconstruction is an omitted factor in black human capital acquisition after the Civil War.

---

Henry S. Farber and Ilyana Kuziemko, Princeton University and NBER; Dan Herbst, Princeton University; and Suresh Naidu, Columbia University and NBER

Unions and Inequality in Historical Perspective

Despite a large literature on unions and inequality, virtually no representative microdata on union membership is available prior to the 1973 CPS. Farber, Herbst, Kuziemko, and Naidu bring a new source of data, opinion polls, primarily from Gallup (`N ~~ 900, 000`), to look at the effects of unions on inequality during the heyday of union power, from 1937–1980. First, the researchers present a new time-series of household union membership from 1937 to the present. Second, they estimate union household income premiums over this same period, finding that despite large changes in union density over this period, the premium holds steady, between 15 and 20 log points. Third, throughout this period, selection into unions with respect to predicted non-union wages was negative, especially during the 1940s and 1950s. Finally, the researchers regress state-year measures of inequality (using both annual tax data and decadal Census data) on state-year household union density. This relationship is negative and robust to a number of standard controls (e.g., state-year skill shares, state-year GDP).

---

Efraim Benmelech, Carola Frydman, and Dimitris Papanikolaou, Northwestern University and NBER

Financial Frictions and Employment during the Great Depression Financial Frictions and Employment during the Great Depression

Benmelech, Frydman, and Papanikolaou provide new evidence that a disruption in credit supply played a quantitatively significant role in the unprecedented contraction of employment during the Great Depression. To analyze the role of financing frictions in firms' employment decisions, the researchers use a novel, hand-collected dataset of large industrial firms. Their identification strategy exploits preexisting variation in the need to raise external funds at a time when public bond markets essentially froze. Local bank failures inhibited firms’ ability to substitute public debt for private debt, which exacerbated financial constraints. The researchers estimate a large and negative causal effect of financing frictions on firm employment. Interpreting the estimated elasticities through the lens of a simple structural model, they find that the lack of access to credit may have accounted for 10% to 33% of the aggregate decline in employment of large firms between 1928 and 1933.

---

Daniel K. Fetter, Wellesley College and NBER

Local Government and Old Age Support in the New Deal (NBER Working Paper No. 22760)

A key question in the design of public assistance to the needy is how allocation of responsibility for funding and decision-making across different levels of government influences the level and type of assistance provided. The New Deal era was a period in which this allocation changed significantly in the United States, as provision of public assistance shifted from local governments to states and the federal government, accompanied by a large increase in government transfer payments. Focusing on assistance to the elderly and using variation in state laws governing the division of funding between local and state governments for the Old Age assistance (OAA) Program, this paper investigates the responsiveness of OAA payments and recipiency to local government funding shares. Payments per elderly resident were significantly lower in states with higher local funding shares, driven largely by reductions in recipiency. The baseline results suggest that had local governments needed to fund half of OAA payments in 1939, on the lower end of local funding shares prior to the New Deal, the share of the elderly receiving OAA would have been 5 percent rather than 22 percent, and perhaps even lower. More speculative results suggest that greater local funding led to lower representation of blacks among OAA recipients relative to their share of the population, particularly in the South.

---

Walker Hanlon, the University of California at Los Angeles and NBER, and Katherin Sudol, Frank H. Netter MD School of Medicine

Pollution and Maternal Mortality: Evidence from the London Fog

The WHO estimates that, in 2015, over 300,000 women died in childbirth. Yet, the causes of maternal mortality remain under-studied because most of these deaths occur in data-sparse developing countries. Hanlon and Sudol provide new evidence showing that air pollution increases maternal mortality in settings where mothers do not have access to modern medical care. The researchers' evidence comes from London using data for 1866-1935, a period characterized by high levels of maternal mortality and severe pollution. They analyze newly collected weekly data covering over seven million births and 26,000 maternal deaths. For identification, they exploit week-to-week variation in fog events that trapped pollution in London. To track these events, the researchers reviewed over 24,000 daily weather reports. Because the formation of fog depends on a complex set of climatic variables, they argue that the timing of these events on a week-to-week basis is as good as random after appropriate controls are included. Their results show that the occurrence of heavy fog in a week was associated with a 5.8-7.4% increase in maternal mortality in that week. This response is 4.5-7.6 times larger, in percentage terms, than the response of total mortality or mortality among all adults. Thus, pregnancy and childbirth were associated with a substantial increase in vulnerability to the effects of air pollution. To the researchers' knowledge this is the first study to draw a direct causal link between air pollution and maternal mortality.