The Great Depression was a watershed moment not only for the U.S. economy but for economic policy and economics proper. Unsurprisingly then, the Depression has been intensely studied; still, many questions remain. Part of the reason for this is a lack of the "right" data. In particular, almost all studies have had to rely on data aggregated to at least the industry or state-levels. This makes it difficult to answer "micro'' questions about, for example, collusion under the National Recovery Administration (NRA). It also makes it difficult to test certain "macro" hypotheses about the Depression such as whether banking failures led to the collapse of the economy or the reverse. This proposal aims to remedy this shortcoming by creating a dataset of firm-level observations covering the first half of the Depression from 1929 through 1935 involving both a random sample and complete enumerations of specific industries and states.
Collecting this firm-level data has potentially large social benefits far exceeding any that come from the proposed work. It is a pure public good .The Atack-Bateman sample of the 19th century Censuses of Manufactures has been used in a variety of studies by the creators of the sample to examine capital deepening of (Atack et al. 2004a) and trends in wage inequality (Atack et al. 2004b). It also has been used by a number of other people for a variety of projects such as studies of industrialization and urbanization (Kim 2005) and in international comparisons (Inwood 2008). This demonstrates the range of studies that this data has been used for. While the Atack-Bateman sample comes from an interesting point in time in American economic history and spans 30 years, the sample here would cover the most turbulent period in all of American economic history. It seems certain that this sample would be used in a similarly wide variety of studies to understand the Depression and in conjunction with the Atack-Bateman sample to study long-run changes in the American economy.
There are a number of projects to be undertaken with this dataset. One project relevant to policy debates today is the effect of what was a "random" banking crisis in Mississippi over the first two years of the Depression. In a paper by Richardson and Troost (2005), the authors argue that for historical reasons, the St. Louis Fed, which covered the northern half of the state, believed that credit should expand and contract with the business cycle. This left them very reluctant to extend cash to banks that were in trouble. On the other hand, the Atlanta Fed, which covers the southern half of the state, was much more aggressive in stemming incipient bank runs at the beginning of the Depression. The authors show how this quasi-random difference had a major impact on how many banks failed in these two regions over the first part of the Depression. Unsurprisingly, banks in the St. Louis district failed at a much higher rate than those in the Atlanta district. After watching a huge number of their member banks fail in the first two years of the Depression, finally, in July 1931, the St. Louis Fed relented and started funneling cash to struggling banks. The question though that still plagues policy debates today is whether the money given to the banks ends up in the real economy or does it just sit as reserves back at the Fed. This question can be answered with this data at least with regards to the manufacturing sector. Did these discount loans foster output growth? Hiring? Maybe even productivity growth?
This proposal intends to build a random sample of manufacturing firms over the first six years of the Depression as well as collect all the data from a number of selected industries. In the process, it aims to address a number of the outstanding issues surrounding the Depression. With that, the results can shed light on some of the knotty policy issues that are so pressing today. Should governments bailout Wall Street to help Main Street? Is government spending effective in stimulating the economy? What role does competitive policy play in helping or hindering productivity growth? The data collected here will surely prove to be exceedingly valuable for understanding firm-level dynamics in this most interesting period in American economic history for the economics community and the public at large.
The research funded under this award has enriched our understanding of the Great Depression through the use of establishment-level data. First, the research demonstrated a tight causal link between bank failures and outcomes in the manufacturing sector. In areas of high bank distress, manufacturing plants tended to do worse in terms of output produced and hours of its workers. What is most striking about these results is the lack of persistence in the effects. Rather than having to deal with negative effects for many years, the effects on output and labor disappear after 2 years. Second, the research found that competition policies implemented in the New Deal served to reduce competition in a two distinct industries: cement and macaroni. Previously, economists had thought the macaroni industry provided a paradigmatic example for why these New Deal policies had little effect. Upon reconsideration using the plant-level data, the industry offers clear evidence that here too, the ``Codes of Fair Conduct" promulgated by the New Deal decreased competition. Third, in ongoing work, the PI has found evidence for the broader effects of the financial crisis during the Great Depression. In particular, the collapse of banking sector led to a decline in aggregate productivity through an increase in the misallocation of resources. Put simply, financial markets serve to channel resources to the most productive ends, and when those markets are dysfunctional, resources get stuck in less productive ends hurting efficiency overall. More broadly, the work supported by this grant has large social benefits by providing a public good, the digitization of these records, to be used by the PI and other researchers interested in this period of time.
