A review of The Warring Forties: The Economic Consequences of World War II by Taylor Jaworski.
The economic history that matters for this review started in 1958 with the publication of The Economics of Slavery in the Antebellum South by Alfred H. Conrad and John R. Meyer. It was thoroughly a work of economics and history combining painstaking data collection with sharp economic analysis. It spawned what I would consider the “heroic” generation of economic historians (at least to economists generally) such as Douglass North, Robert Fogel, Albert Fishlow, and William Parker to name a very few. All of these pioneers took their marching orders from Conrad and Meyer to go and count. What made their work so unique and powerful was that while it often tended to focus on seemingly parochial topics (who knew ocean shipping could be that interesting?), it spoke to deep questions in economics on the nature of the firm, the role of property rights, and the sources of economic growth.
And the decline came. While some people such as Fogel continued to produce first rate students that earned prestigious positions, much of the field struggled to attract the best and brightest graduate students. In fact, many of the leading graduate schools dropped their economic history requirement altogether. This was not necessarily the fault of economic historians per se. Times changed. Econometric technique advanced, or, at least, got more involved, and the data collected by the leading generation could not bear it. In the end to borrow a line from Marx, “the class which has the means of production at its disposal has control at the same time over the means of mental production.”
There is a renaissance going on in economic history right now. Over the last 10 to 15 years, we have seen the precipitous decline in the costs of data collection and entry. Joseph Ferrie has told me that now 20 some years after his dissertation on intergenerational mobility, he could collect all of those data in an afternoon. He assures me that it took him appreciably longer than that for the original dissertation. Jeremey Atack and Fred Bateman, in collecting the 19th Century of Manufactures, never collected the name of the business because it wouldn’t fit on the punch cards they used! The rise of “big data” in economic history has made these stories seem quaint, and it has resulted in work admired by the profession as a whole for the sophistication of its technique and breadth of its data. I have in mind here work by Richard Hornbeck on the Dust Bowl or Dave Donaldson on the railroads in India.
Taylor Jaworski’s dissertation fits quite snugly in this “new” new economic history that uses the language of modern economics with a focus on formal models and (fetish for?) identification. His dissertation written under the direction of the indomitable Price Fishback begins the process of writing a new economic history of the effects of WWII in setting the stage for the post-war development of American economy. Jaworski focuses on two questions: the role of conscription for women’s educational attainment and the location of manufacturing and people.
Most refreshingly, Jaworski’s work seems free of the “helicopter” feel that much empirical work in economics suffers from. In these works, it seems like the author helicoptered in, ran a couple regressions, claimed everyone else didn’t understand causality, and left. While this mindset is problematic for economics in general, it is particularly troubling for economic history because it leads to a cramped view of the role of history. It tends to reduce the valuable parts of history to those that can be cleanly identified. In the end, there is no history, no narrative instead history is a rummage bin to be rifled through in search of clever natural experiments (I admit my guilt in this pillaging at times).
But Jaworski knows and cares about history. Because of this, the dissertation seems comfortable in its own skin knowing precisely how it fits into a much bigger picture. This is nicely reflected in the second chapter (the first chapter is a summary of literature). In it, Jaworski examines the effects of the conscription of men on the educational attainment of women. Many including Claudia Goldin have commented on the slowdown in educational attainment for those in the cohort, men and women, directly affected by conscription. Jaworski provides additional evidence using variation across states and years in number of callups focusing on the effects for women. He finds substantial negative effects on educational attainment presumably due to women working in munitions factories rather than sitting in class.
The results here partially accord with more recent work by Matt Notowidigdo and co-authors on the Great Recession. They found that, at the height of the housing bubble, many men, rather than go to college, took construction jobs and then never went back to school when the jobs disappeared. Jaworski, on the other hand, finds that the negative effects disappear by 1970 and interprets this as evidence for women returning to school. He includes “state” fixed effects, which controls for the state in which a particular cohort of women were born. However, as Jaworski’s third essay points out, the post war period was a time of great geographic churning with record numbers of people moving to different states of residence. So while controlling for whether or not a cohort was “treated” with high conscription rates in 1945 is important, it seems equally crucial to control for where a particular woman is living in 1970 since the labor market in that state should be a main determining factor in the potential returns from going back to school. I hope that in the future, Jaworski will connect these two essays to draw out the influence of geography and the shifting location on labor market opportunities for women.
The next two essays explore the effects of WWII government spending on the geographic location of manufacturing establishments and people. The question loitering in the background here is the relative rise in industrialization of the US south after WWII. The third chapter has the feature of specifying an economic model of location choice by business establishments and then actually estimating it. The model is a very standard one from industrial organization based on a random coefficients specification for “utility” or profit across locations. With a particular assumption on the distribution of random components, the model gives rise to a simple logit specification for the share of total manufacturing in a particular location. To estimate this, he collects data on spending contracts and matches that with information on the number of manufacturing establishments at the 2-digit. This allows him to not only study the overall level of industrialization but also the mix of manufacturing establishments.
The first set of results show very minor effects of supply contracts on the location of industry. These were contracts to produce a particular type of product for use in the war effort. On the other hand, a different kind of government spending in the form of capital investment in new or existing plants did appear to have some positive effects, particularly relative to size of the preexisting manufacturing sector. Jaworski seems to hint that factories that won supply contracts simply converted their production to what was needed and then reconverted back after the war. On the other hand, capital investment is by its nature could be more permanent leading to meaningful effects.
These are important estimates, but I think the implications are not completely drawn out. At one point, I was arguing with Jaworski’s adviser that we needed more structural work in economic history. Well-identified regressions and structural models provide causal estimates and allow us to conduct counterfactual exercises. Structural models help us interpret those counterfactuals by linking the estimates to particular parameters in an economic model. They also allow us to make welfare comparisons in many cases. Both of these upshots of structural models are quite fascinating here. What was the welfare implications of these policies? What do these results on investment relate to other studies that attempt to estimate elasticities of investment with respect to tax changes? I hope that in future revisions, Jaworksi explores some of these questions and connections to other literatures future.
The fourth essay continues this thread by looking at the effects of government spending on the location of people. He develops a simple model of the spatial distribution of workers and wages with different levels of productivities. While these different types of workers cannot be substituted for one another, they share a common resource in short supply: housing. This means that in-migration of, say, low productivity types has a direct negative effect on high productivity types for the simple reason that demand and, hence, prices for housing are higher. So high productivity real wages decline as do real and nominal wages for low productivity workers.
Jaworski is quick to point out that calculating real wages requires deflating nominal wages by local housing prices, which he does not have. While this is (sort of) a problem for the empirical exercise, I do not want to belabor that. Instead I would note that the economics of the model, while completely standard, are also rather fragile. However, this plagues all studies of the effects of immigration.
In the empirical analysis as before, Jaworski draws on variation across locations (technically SEAs) in government spending on supply contracts. Now the dependent variables are employment and nominal wages for various demographic groups, which are proxying for skill groups. He finds some effects on the relative labor supplies across and wages across these groups with the male-female wage ratio falling, for example. Note that because in his model, high and low productivity workers consume the same type of housing, the ratio of nominal wages will be equal to the ratio of real wages. So these results are consistent with the model. Such an observation leads the reader to speculate on why there should be effects on nominal wages as well.
Even though the dissertation is nominally about the effects of WWII, it is really about whether temporary shocks can have long-run effects. Now what Jaworski is looking for is not as silly as some of these papers purporting to find a link between the number of spontaneous cheese festivals in the Middle Ages and brie consumption in 2014. Still I hope to see in future revisions a sharper link between the effects estimated here and the expectations of economic actors to these seismic shifts in the US economy.
Nicolas L. Ziebarth
Department of Economics
University of Iowa
Integrated Public Use Microdata samples from the 1960 and 1970 censuses
Quotas, Calls, and Inductions volumes of the Special Monographs of the Selective Service
Census of Manufactures 1939, 1947, 1958, 1963, and 1967
Market Data Handbook of United States
Industrial Market Data Handbook of the United States
Haines, M. R. (2010). Historical, Demographic, Economic, and Social Data: The United States, 1790-2000, volume ICPSR02896-v3.
University of Arizona. 2014. 158 pp. Primary Advisor: Price Fishback.
Image: Chow is served to American Infantrymen on their way to La Roche. Wikimedia.