David Beckworth asks a question:
Macro and Other Market Musings: The Latest Installment in the New Deal Debate: Harold Cole and Lee Ohanian... move beyond the unemployment numbers discussion and present data on hours worked, consumption per capita, and nonresidential investment....
[T]he New Deal didn't restore employment. In fact, there was even less work on average during the New Deal than before FDR took office. Total hours worked per adult, including government employees, were 18% below their 1929 level between 1930-32, but were 23% lower on average during the New Deal (1933-39)....
There is nothing new in their story, but their numbers certainly are interesting. I am looking forward to reading what Eric Rauchway, Brad DeLong, Paul Krugman and others have to say in response to this piece. Please guys, don't dissapoint me.
The immediate answer to questions about Lee and Ohanian is that all the institutions they blame for high structural unemployment during the Great Depression were still there after World War II--yet no high structural unemployment then at all.
The hasty answer to Lee and Ohanian is that they are playing two jokers:
(1) They define "before Roosevelt" as the average of 1930-32 rather than as 1932.
(2) They assume that all of the declines in hours of work per week is due to deficient demand rather than to a much-desired increase in the bargaining power of workers who then choose to bargain for more leisure. Look at:
Source: Valerie A. Ramey and Neville Francis (2007), "A Century of Work and Leisure," forthcoming in the American Economic Journal http://tinyurl.com/dl20090202, and author's calculations.
Hours of work in the early 1950s are even lower than they were in 1939. But does that mean that the economy was even more depressed in the 1950s than it was in 1939? No. You don't want to maintain that the interwar decline in hours worked tells us about cycle and not trend. Is there anyone who will say that the decline in hours worked from 1914 to 1952 tells us that the economy was performing much worse along a business-cycle dimension in 1952 than it was in 1914?
The detailed answer I will outsource to Peter Temin (2008), "Real Business Cycle Views of the Great Depression and Recent Events," Journal of Economic Literature 46:3 (September), pp. 669–684 http://tinyurl.com/dl20090202a:
Cole and Ohanian use the general equilibrium growth model to explain the Great Depression in the United States. They summarize their procedure as follows:
We conduct the analysis by assuming that the capital stock in 1929 is equal to its steady-state value, and then we feed in the sequence of observed levels of total factor productivity as measures of the technology shock . . . . The measured decrease in productivity between 1929 and 1933 generates a Depression in the model that is similar to the actual data: output in the model falls about 23 percent relative to trend in 1933, compared to the actual 38 percent decline. This similarity disappears after 1933. As a consequence of rapid productivity growth, output in the model is almost back to its trend by 1936. In contrast, actual output remained 25–30 percent below trend during the recovery (p. 31).
This is the starting point of their analysis, but it is hard to know how the comparisons in this paragraph were derived, since the footnoting is nineteenth-century in its brevity. Cole and Ohanian present their table of TFP levels in the 1930s with a footnote that says only, “Data from Kendrick 1961.” I went to John W. Kendrick (1961) and found the data on TFP in the 1930s reproduced here in table 1. I include Cole and Ohanian’s TFP estimates for comparison. Their estimates do not agree with any of the estimates by Kendrick. Cole and Ohanian state in the paragraph just quoted that productivity was back to its trend by 1936, indicating that their series is taken relative to a trend. Subtracting a 2 percent trend from Kendrick’s series, however, does not come close to reproducing their estimates. The three detrended Kendrick estimates for 1936 are 92.1, 96.8, and 73.4—far below Cole and Ohanian’s detrended estimate of 99.5.
This discrepancy is disorienting....
Cole and Ohanian also are casual in their interpretation of these observations. They say that their model generates a depression “similar to the actual data.” Output fell 23 percent relative to trend in 1933, compared to an actual 38 percent decline. Is 23 percent similar to 38 percent? It is only three fifths as large, and no metric is given by which one can judge whether this difference is large enough to refute some hypothesis.
This informal approach is evident even when there is a clear hypothesis at stake. Cole and Ohanian argue that the shocks that produced the U.S. depression were larger than the shocks that produced depressions in other countries. They infer this conclusion from a comparison of the U.S. decline with the decline of an average of seven other countries. The United States declined more; QED. As before, was the difference significant?... There is no formal test.... There also are no criteria given for inclusion in the index. Cole and Ohanian argue in the following essay that Britain... was already depressed by 1929. What do we learn about shocks after 1929 from such a comparison?
Cole and Ohanian argue in the body of their paper that various explanations for the decline in TFP are unconvincing and conclude... “There were large decreases in TFP... an exogenous productivity-shock explanation of the Depression is unsatisfactory, as it simply pushes the question of what caused the Depression one step back. Thus, developing and testing models that can explain why productivity fell so much may advance our understanding of 1929–33” (p. 47).
This conclusion does not take us very far. The use of the general equilibrium growth model does not lead to a coherent story.... Looking for an explanation of a poorly estimated magnitude does not seem like a profitable guide to future research...