20 percent, and for ten consecutive years, it was greater than 10 percent.
Even these statistics do not fully portray the incidence of unemployment. Annual average statistics disguise periods of unemployment in excess of those averages, as will be demonstrated shortly. Beyond that, however, the burden of unemployment varied considerably between various demographic groups and geographic areas. For example, it has been estimated that unemployment among female blacks in the city of Detroit in January 1931 was around 75 percent, at a time when the national rate was probably about 14 percent. Similarly, teenage unemployment in 1937 was estimated at 36.5 percent, more than double the national rate. Moreover, the average duration of unemployment was substantial. For example, of unemployed men in Massachusetts at the beginning of 1934, a large majority (62 percent) had been unemployed for one year or more.2
The misery created by the Depression was so substantial that it is hard for Americans under the age of sixty to imagine. With the absence of a comprehensive governmentally provided safety net, it is undeniable that millions of Americans suffered a great deal. At the same time, however, the distribution of the burden of the Depression was very unequal, and millions of Americans lived normal, even prosperous lives. Sales of cars in 1937 were the second highest in history, and more radios and refrigerators were sold in 1935 than in 1929.3 The tragedy of the Great Depression derived as much from the distribution of the fall in income as the size of the decline itself. While the decrease in agricultural income is one reason for the unevenness of the distribution of the misery, the single most important problem was unemployment. Many Americans suffered little or no income loss, while others had their income fall drastically to near-starvation levels.
Monthly Estimates of Unemployment Rates
The annual data on unemployment disguise important intrayear variations in the incidence of joblessness, and fail to give us details on the timing of the downturn and subsequent recovery. Unfortunately, monthly or quarterly data were not collected until the Depression was nearly over.
To deal with this data inadequacy, we have developed monthly estimates of unemployment, which are presented in table 5.1.4 The procedure for constructing the estimates is straightforward. First, we formed a regression model of the annual unemployment rate for the 1923–37 era, utilizing as explanatory variables manufacturing employment, an index of help-wanted advertising (a proxy for job vacancies), freight carloadings, and the Babson index of business activity. The model’s explanatory power was quite high (R2 = .9947),
Monthly values of the various explanatory variables were then used with the regression results derived from annual data to calculate monthly estimates for the unemployment rate. We altered the results in two ways. First, we corrected the estimates for the deviation of the predicted annual estimate (from the regression discussed above) from the official BLS estimates. Second, we used the Department of Commerce’s X-ll seasonal adjustment procedure to correct for seasonal variations in the data.
In addition to the table, figure 5.1 shows the unemployment variations visually and relates them to some of the major happenings of the Depression era. What do the results indicate? The unemployment rate before the beginning of the Depression was very low, below 2.5 percent.5 The unemployment rate at the bottom of the Depression, in March 1933, was 28.3 percent, suggesting that unemployment rose by about 26 percentage points from 1929 to the 1933 trough. The increase in unemployment came in four phases separated by periods of stable or even falling unemployment.
About one-fourth of the increase in unemployment came with a rush, in the last quarter of 1929. The December 1929 increase was the second largest recorded during the Depression decade. The first effects were substantial, more than some symbolic beginning.6
The initial increase in unemployment was followed by a long period of stable or even falling unemployment, lasting through the first ten months of 1930. Suppose that the natural rate of unemployment in 1929 or 1930 was around 4 percent. The decline in unemployment from 9 percent in December 1929 to about 7 percent in the third quarter of 1930 meant that unemployment had recovered about 40 percent of the way back to the natural or normal unemployment rate.
The second phase in the development of the Great Depression was short, lasting but two months, but it witnessed an increase in unemployment that exceeded the initial shock following the stock-market crash in October 1929. The rise in unemployment in November 1930 is almost certainly the largest single monthly increase in unemployment in the history of the nation. That surge in unemployment turned a recession into a depression. Thus, explaining the fall 1930 phase is important in explaining why the Great Depression developed.
TABLE 5.1 ESTIMATED MONTHLY UNEMPLOYMENT RATES, NOVEMBER 1929-DECEMBER 1939
FIGURE 5.1 UNEMPLOYMENT AND MAJOR EVENTS DURING THE GREAT DEPRESSION
After four months of relative stability in early 1931, the third phase of the downturn began. Unlike the earlier periods, the 1931–32 episode was a long, continuous decline rather than a short sharp movement. In fourteen of the seventeen months from April 1931 to September 1932, unemployment increased. The accumulated effect was to double the unemployment rate, accounting for fully half of the increase in unemployment observed from peak to trough. The third phase was the biggest, the longest, and the one that made a depression into the Great Depression.
The 1931–32 slide began rather slowly and then accelerated. If our estimates are correct, unemployment rose by an average of .51 percentage points per month from April to August 1931, but at a brisker 0.95-point rate from August to December 1931. After some slowdown in the rate of decline (but not a reversal) in the first quarter of 1932, unemployment rose at an extraordinary 1.13 percentage points per month from March to September 1932.
The fourth quarter of 1932 saw still another reversal, with a significant reduction in unemployment being recorded. Conditions deteriorated again in the final phase of the downturn during the first quarter of 1933, with unemployment peaking in March at a slightly higher level than observed the previous September.
While the post-1933 recovery is taken up in detail in chapter 7, it is striking to note that a fairly brisk recovery seemed to be under way in the summer and early fall of 1933, but that it simply died after October of that year, not to resume with any real vigor until 1936, the first full year of really sustained recovery. An important question is: why did the upsurge of 1933 stop for more than two years?
Unemployment and the Adjusted Real Wage, 1929–1933
The adjusted real wage model developed in chapter 3 does an excellent job of explaining the 1929–33 rise in unemployment. We recognize, of course, that there is some dispute regarding the magnitude of unemployment during this era. Both the Coen and the Darby estimates place unemployment at lower levels in the early thirties than the official Bureau of Labor Statistics estimate derived by Stanley Lebergott.7 We are inclined to agree with Gene Smiley that the Lebergott/BLS estimates are probably as good as any.8 Most of our analysis, including our monthly estimates presented above, is centered around the BLS numbers, although we also performed some additional sensitivity analysis, using the alternative estimates, to ascertain whether our basic wages model is sensitive to differences in data sources.
TABLE
