Industrial Employment and the Policies of Herbert C. Hoover
by Douglas W. MacKenzie
Wages and cycles
The idea that President Hoover intensified and prolonged unemployment with his high wage policy depends upon five propositions:
1. He intended to keep nominal wages high during industrial depressions
2. He possessed real influence over the industries he targeted
3. He could monitor wages in these industries
4. He followed through by applying actual pressure on targeted industries
5. He pushed industrial wages above equilibrium
Did Hoover possess real influence over industry? Stone (1932) suggests that Hoover impressed his views via his regulatory authority over some businesses, while other businesses may have simply been wary of drawing his adverse opinion. As Secretary of Commerce, Hoover pressed for an eight-hour workday in the steel industry. Success in reducing hours meant that the twelve-hour day was on the way out in American industry (Hawley, 1981, p. 95). Hoover’s ability to reduce work hours indicates that he could influence industry.
Hoover’s influence as Secretary of Commerce was limited. Secretary Hoover had intended to keep industrial wages high during the depression of 1920–22, blaming this crisis on difficulties associated with readjustment following the First World War. In September 1921, Hoover invited some 300 business and labor leaders to a conference, the goals of which were to alleviate unemployment, eliminate waste, increase foreign trade, and study business cycles (Hoover, p. 44–5). Under his influence, union operators would maintain wages for coal workers well into 1923 (Hawley, 1981, pp. 63–4).
… Hoover affected the coal industry even earlier (Ibid., p. 63). Passage of the Railway Labor Act of 1926 seems to have given Hoover influence over that industry as well. The influence of Secretary Hoover appears weak. Vedder and Gallaway (1997) observe that wages and unemployment remained high only briefly, while high wages and high unemployment would be more persistent during Hoover’s presidency. The idea that Hoover held influence over industry, especially as President, appears reasonable.
Hoover reacted to the stock market crash in 1929 by holding new conferences with business leaders. Hoover’s aim was to maintain wages, stimulate counter-cyclical investments, and provide emergency relief (Hawley, p. 65). His program entailed three steps: mobilizing credit, maintaining wages, and constructing and maintaining plants and equipment (Cover, 1930). At these conferences, he obtained pledges from numerous business leaders to refrain from cutting wages (Rothbard 1972).
As far as the ability to monitor industry is concerned, the Commerce Department recorded data on wages and employment during this time. Hoover had access to this data, both as Secretary of Commerce and as President. The third proposition is therefore correct.
Hoover as secretary and as president
The evidence suggests that Secretary Hoover had relatively little influence on wages during the 1920–22 crisis. With the onset of this crisis, money wages fell, but consumer prices fell further, so productivity-adjusted real wages initially rose by 17 percent while unemployment rose to 11.7 percent (Vedder and Gallaway, 1997, p. 62).
Vedder and Gallaway find an 86 percent correlation between productivity-adjusted real wages and unemployment during the 1920–22 crisis. Subsequent declines in productivity-adjusted real wages coincided with falling unemployment during the mid 1920s.
… Money wages remained stable in 1930 and fell only slightly in 1931. Price deflation during this time caused real wages to rise by 12 percent in 1930. This trend continued until the latter part of 1932. Wage increases paralleled the trend in unemployment at this time. Unemployment during 1929 started out in the low single digits but had climbed to 9 percent by December. Unemployment ranged from about 6–8 percent throughout most of 1930, but spiked up to 14 percent at the end of the year. During 1931 it gradually climbed to 20 percent, and peaked in 1932 at one quarter of the workforce.
… There was an employment boom in the iron and steel industries from July 1927 to June 1929, and a positive correlation between real wage rates and employment can be detected in these industries at this time, with a correlation coefficient (R2) of 0.728. Given that this boom originated at a time when the Federal Reserve had set low interest rates, it is possible that this boom was bank credit induced. After President Hoover held his White House conferences in 1929, real wages for iron and steel workers rose and employment fell, and there is a negative correlation between real wages and employment in these industries between December 1929 and August 1931 (R2 = 0.823).
Data from the boom years is consistent with the interest rate-driven boom in that higher wages coincided with higher employment levels. We can see this in the first of these graphs. The points for the Y variable represent actual data. The points for “predicted Y” represent the statistical trend. The first graph indicates a rightward shift in demand.
Data from the bust years indicate that Hoover’s high wage policy was problematic. In the second graph, higher wages coincided with lower employment and lower wages coincided with higher employment. The bust likely reduced demand for iron and steel workers, but Hoover’s high wage policy drove the decline in employment.
… In contrast, from October 1929 to August 1931, iron and steel employment fell by over 50 percent while real wages on the NICB index rose from 109.3 to a peak level of 126. Real wages remained high during 1932 despite an overall decrease in employment in this industry of over 70 percent.
Other industries experienced a boom up to 1929 and higher real wages and job losses after Hoover’s conferences. The foundries and heavy equipment industry saw employment and real wages increase from late 1927 to mid 1929 with a weak but positive correlation (R2 = 0.495). After Hoover’s conferences, nominal wages rose and employment fell (a negative correlation: R2 = 0.652). … From December 1929 to January 1933 real wages in the machine tools industry rose and employment fell sharply with a very high correlation (R2 = 0.887). These data, of course, extends beyond Hoover’s presidency. Most of the employment losses and wage increases did take place between the Hoover conferences and the summer of 1932 (during Hoover’s campaign for reelection), and there was a high negative correlation between wages and employment during this period (R2 = 0.860).
The National Industrial Conference Board index for real wages for auto workers rose from 108.2 in December 1929 to 125.5 in May of 1932. This real wage increase resulted in a reduction in the auto employment index of nearly sixty points. Overall, there is a negative correlation, but the statistical correlation is low (R2 = .422). Nominal wages have a stronger correlation with auto employment (R2 = 0.600). Why? Henry Ford not only attended the Hoover conferences, he was an outspoken advocate of the “buy back the product” theory of high wages. Of course, Ford did not set wages for the entire industry, but his influence may provide a partial explanation of the importance of nominal wages in this industry.
During the first half of 1929, nominal automobile wages fell 2 percent below 1928 levels. During the second half of 1929, wages fell an additional 11 percent and employment rose by 5 percent. It therefore appears that wages were flexible leading into the Hoover conferences. Employment in the auto industry dropped 25 percent in 1930, during which time wages for auto workers fell only 3 percent. A slight fall in the price level in 1930 meant that wages for auto workers were virtually unchanged. It would seem that the industry followed Hoover’s instructions and tried to maintain nominal wages. Since overall unemployment rose about ten percentage points during this time, it is clear that the auto industry bore a disproportionately large part of the national increase in unemployment.
Nominal auto wages actually rose by 1 percent in 1931. With 9 percent deflation during that year, real wages rose 10 percent and employment fell three more points to 28 percent below its 1929 peak. Nominal wages in the automobile industry rose an additional 8 percent in 1932. An additional 10.5 percent of deflation during that year drove real wages up 30 percent. Employment then fell 17 points more to 55 percent of its 1929 peak. This drastic decline in employment greatly exceeds what we should expect from the 6 percent increase in national unemployment rate during 1932.
Employment in the machine tool industry rose 20 percent in 1929, while nominal wages rose 2 percent in 1930. Deflation caused real wages to rise 4 percent, and employment fell 17 percent. In 1931 nominal wages fell 1 percent from their 1929 peak, but real wages rose 8 percent and employment fell 21 more points, a total decline of 38 percent. Nominal wages fell by 9 percent in 1932, but a deflation of 11.5 percent more than offset this decrease in wages, and employment fell another 17 percent for a total 3 year decline of 55 percent.
This loss of 55 percent exceeded the employment losses of the auto industry, despite a decline in machine shop wages relative to auto wages. The rapid increase in machine tool employment in 1929 is consistent with a credit-driven boom. The subsequent collapse of employment in the machine tool industry also fits with the idea of an unsustainable credit driven boom. Abandonment of investment projects in a bust should cause disproportionately large employment losses in capital goods industries.
Steel and iron industry nominal wages were stable in 1929, and employment expanded 6 percent. In 1930 nominal wages fell 1 percent and real wages increased by 1.5 percent, while employment declined 13 percent from its 1929 peak. In 1931 real wage rates increased by 7.5 percent and employment fell to 68 percent of its peak value. In 1932, real wage rates fell to 4 percent below their peak level, but employment fell another 16 points to 52 percent of the peak level. [...]
The leather industry saw less wage rigidity and less employment losses. In 1929 there was a 1 percent increase in employment and no change in wages. In 1930, nominal wages fell 1 percent, and real wages rose by 1.5 percent. By 1931, nominal wage rates were 8 percent below their 1929 peak, but after adjusting for deflation, real wage rates were still 2 percent above their 1929 level. Employment dropped 17 percent below its peak, far less than in the heavy industries on which Hoover had focused so much of his attention. In 1932, nominal wages fell to 22 percent below their prior peak, but this decline was almost entirely offset by deflation.
In the meat industry, nominal wages fell slightly in 1929 and employment grew by 2 percent. In 1930, real wages rose 2.5 percent and employment fell by only 4 percent. In 1931, real wages rose to 8 percent above their peak, and employment fell by a total of 10 percent from its peak level. In 1932 real wages were 6 percent lower than peak, and employment had fallen off by 20 percent.
The paper industry exhibited a 2 percent nominal wage decrease and a 6 percent employment increase in 1929. Wages fell 2 percent in 1930 and deflation caused a slight increase in real wage rates, but employment in this industry actually rose 3 percent. In 1931, deflation pushed real wages to 7 percent above the 1928 level, and employment lost all its previous gains, falling 8 percent below its peak. In 1932 deflation pushed real wages 3 percent above their peak, and employment fell by a total of 19 percent from its peak.
Market or government failure?
… Even in the total absence of official price indexes, workers should still be able to recognize deflation as it affects the goods that they buy on a routine basis. Even if workers ignored prices as they engaged in routine purchases of staple items, they would surely notice increasingly large amounts of funds in excess of these purchases as their real wages rose. Depression era workers could not have been so ignorant of prices as to have thought that any nominal wage cuts would reduce their real income. O’Brien also overlooks the strong possibility that industrial leaders adopted Hoover’s wage maintenance policy out of fear of reprisal by the President rather than out of conviction. Finally, O’Brien must explain why employment losses tended to be the highest in industries whose leaders attended Hoover’s conferences.
… O’Brien claims specifically that belief in the “high wage-high demand” theory was commonplace. It is very unlikely that a CEO of a high profile company or the president of a major trade association would publicly contradict a sitting President, especially during the early part of his term in office. O’Brien notes that wages did break by the fall of 1931 because firms may have felt that wage cuts would not affect productivity. Rising unemployment had, after all, led to decreasing probabilities of finding another job. However, the simultaneous decline of wages and Hoover’s influence as president is probably not a coincidence.
… Industrial leaders who attended Hoover’s White House conferences tended to keep their employees’ wages high until shortly before his defeat in the 1932 election. These industries exhibited above average employment losses, while industries that did not attend Hoover’s conferences exhibited below average losses.
Unusual Historical Claims about Hoover
The Railway Labor Act of 1926 is cited as an example of Hoover’s authority, despite the fact that the act remedied the excesses of the earlier Railway Labor Board, rendering federal intervention less likely. In 1921, the Railway Labor Board and the major railways implemented a twelve percent wage reduction. In response, rail workers called a national strike, introducing major supply interruptions and provoking the Department of Justice to step in and outlaw the strike (Northrup, 1971). The Railway Labor Act of 1926 represented an attempt to avoid this disruptive interplay of corporate austerity, labor unrest, and severe federal intervention by requiring the railways and the union to exhaust a series of mediation procedures before proceeding with a strike. Certainly the law was interventionist as a piece of labor relations legislation (Northrup  calls it “the most comprehensive control of labor relations and disputes on the American scene”), but the idea that it could function as a tool for Hoover to implement federal wage policies is implausible. The Railway Labor Act of 1926 reduced the likelihood of federal dictation of wage policies, even as it formalized a particular model for the conduct of labor relations. [...]
The paucity of federal data is colorfully illustrated in an anecdote shared by Frances Perkins (Governor Roosevelt’s industrial commissioner in New York, and the future Secretary of Labor in the Roosevelt administration). Hoover’s preferred employment statistics in 1930 were extrapolated from the U.S. Employment Service (a WWI-era job placement agency) data, and in 1930 the president made an announcement in the press about improved employment conditions on the basis of the Employment Service reports. Commissioner Perkins, who had access to higher quality data from New York contradicting the president’s claims, confirmed her figures with Hoover’s own Bureau of Labor Statistics before publicly humiliating Hoover with an announcement that he was relying on inaccurate data. Duncan and Shelton (1978) note that the administration’s credibility on questions of employment “never recovered” from the Perkins revelation. Hoover’s employment data during the post-conference period was therefore widely considered to be suspect, although it was at least available in some coherent condition. Data on wages and pay scales was not as advanced as the data on employment, and was “barely beyond the pilot stage” (Duncan and Shelton, 1978). The wage data that was available certainly were not surveyed comprehensively enough to be used to monitor the conference attendees in the way that MacKenzie (2010) implies it was. Comprehensive, firm-level wage data only became available to the federal government with the inauguration of the federal unemployment tax in 1935, and it is highly doubtful these data were used to intimidate reporting firms.
… But another problem with MacKenzie’s (2010) argument is that Hoover consistently noted three activities that he believed businesses should voluntarily partake in during a depression: maintain wages, maintain employment, and maintain investment [...] If we are to accept the argument that wage maintenance was successfully implemented, why did Hoover stand idly by as his other two stated goals failed quarter after quarter for the duration of his term in office? Economists are taught to be suspicious of “corner solutions” in human behavior, and yet this is exactly what MacKenzie’s (2010) argument requires. If Hoover had the ability to maintain high wages, why would he refrain from using it to achieve the remaining two-thirds of his agenda? Even if these goals operate at cross-purposes, there is no reason to think that Hoover would pursue one to the complete exclusion of the others. Bernanke (1995) notes this problem with relying on inadequate wage adjustment to explain the Depression as well, pointing out that both parties to a wage contract have a strong incentive to renegotiate (i.e., lower the wage), something which is not true of debt contracts.
Recent Findings on the Hoover Conferences
… First, Rose (2010) compares the elapsed time until wage cuts were implemented for firms that attended the November 21 conference with the time it took large firms who did not attend to implement these cuts. While he notes a small increase in the time it takes for wage cuts to be implemented for attendees, this differential disappears when controls are added for a firm’s industry and its assets. Rose’s (2010) conclusion is that firms with characteristics that predisposed them to delaying wage cuts were disproportionately represented at Hoover’s conference, such that when these characteristics are accounted for, there is no discernable difference between attendees and non-attendees. The absence of any evidence of an impact by the conference is maintained when “early movers,” or firms who cut wages so early in the Depression that they would not have been influenced by Hoover’s efforts, were excluded from the sample.
Aggregation Bias and Real Wage Cyclicality
… Research suggests that during post-war recessions, firms generally maintain employment contracts with high-wage workers and end contracts with low-wage workers. This drives up average wages because the composition of the workforce changes over the business cycle (Bowlus, Liu, and Robinson, 2002). To address this bias in the aggregate data, macroeconomists have turned to individual level data to look at the cyclicality of wages for individuals or more homogenous groups of workers.
This literature on the cyclicality of the real wage in disaggregated data, which Abraham and Haltiwanger (1995) describe as “a small explosion of research,” begins with the work of Stockman (1983), Raisian (1983), and Coleman (1984). These authors use the Panel Study of Income Dynamics (PSID) to compare aggregated and disaggregated wage adjustment patterns. Subsequent work, including a celebrated paper by Bils (1985), uses the National Longitudinal Surveys (NLS) to explore disaggregated wage cyclicality. After Bils (1985), most studies of real wage cyclicality used one of these two datasets: the PSID or the NLS. Examples of the post-Bils (1985) literature include Mather (1987), Keane, Moffit, and Runkle (1988), Blank (1990), Tremblay (1990), Beaudry and DiNardo (1991) Solon, Barksy, and Parker (1994), and Shin (1994) although this list is not exhaustive. In contrast to studies using aggregated data, research relying on disaggregated data suggests a modest pro-cyclicality for real wages. To put it differently, the disaggregated studies demonstrate that aggregation introduces a counter-cyclical bias driven primarily by the shifts in the composition of the workforce over the business cycle. Studies which use PSID data generally find greater degrees of aggregation bias than studies using NLS data (Abraham and Haltiwanger, 1995). Other differences exist as well; Shin and Solon (2007) find that PSID data suggest that salaried worker wages are less cyclical than hourly worker wages, while this differential does not emerge in NLS data.
The classic PSID and NLS studies form the backbone of the literature on disaggregated real wage cyclicality, but other investigations have been conducted as well. Levy and Newman (1989) find that composition bias impacts aggregate wage cyclicality in developing economies as well. Messina, Strozzi, and Turunen (2009) provide evidence of composition bias in aggregate statistics in many OECD countries besides the United States. Martins, Solon, and Thomas (2010) look specifically at the cyclicality of the wages of newly hired workers. They note the reliance of recent search and matching models on the rigidity of the wages of newly hired workers, and demonstrate that even this sub-population exhibits wage cyclicality when the data are disaggregated.
Hoover and Wages in the Depression: A Comment on Douglas MacKenzie: A Rejoinder
by Richard Vedder and Lowell Gallaway
The gist of the main argument is this: Studies cited above that show relative wage rigidity use aggregate data showing that wage levels were relatively constant as the downturn unfolded, but modern studies using more disaggregated data show wage pro-cyclicality, which in this context means wage rates would fall with declining economic conditions. It is argued that compositional shifts in the labor force are not picked up in aggregate data. Specifically, as firms shed labor, they get rid of the junior, lower paid employees, maintaining average pay for the diminished work force even as total payrolls and hourly pay rates are reduced. The failure of aggregate wages during the early Depression to fall aggressively, the author opines, no doubt reflects this phenomenon. The empirical validity of this, he asserts, has been demonstrated by several modern scholars.
There are a lot of problems with all of this, but let us start with the obvious historical one. All the studies showing the pro-cyclical nature of wages come from such modern data sets as the Panel Study of Income Dynamics (PSID) or the National Longitudinal Survey (NLS). All of them deal with post-World War II labor markets. There was a revolution in labor markets between the early 1930s and the early 1960s (or even the early 1940s). In 1930 and 1931, labor unions were weak and essentially not important in huge mass production industries like steel and autos. This is before the Wagner Act of 1935 or even the wage-enhancing characteristics of the National Industrial Recovery Act of 1933.
Unions, however, were extremely important in industry when the PSID and NLS data were starting to be collected. The notion of “last hired, first fired” incorporated into collective bargaining agreements was almost certainly dramatically less prevalent in, say, 1930 or 1931. Indeed, we could see the possibility that employers in 1930 or 1931 shedding workers would disproportionately discharge more highly paid workers in order to more aggressively lower labor costs. If so, the aggregation bias runs in the other direction, and the changing composition of the labor force would lead to reductions in average reported aggregate wages, not increases. But who knows? The author has no evidence on this, and he is merely speculating that the world of 1930 was not much different than the world of, say, 1970 or 1980 with regards to this phenomenon. To base an argument based on other research for a period far removed from the one in question is very questionable, in our judgment.
… First, many contemporary economists commented on the wage rigidity. For example, Columbia economist Carter Goodrich (1931, p. 187) said “so far… the patient does not seem to have swallowed the prescribed medicine [wage reductions].” The iconic economist Joseph Schumpeter (1931, p. 180) noted that the depression “is much intensified by this factor” (high wages), a point made by others in different venues, including Robbins (1934). It seems to me that the observations of economists of that era (see also Benjamin Anderson) are more likely to be accurate than the musings of later scholars relying on research postdating the depression by decades.
Second, if there were a significant compositional shift to senior workers as newer workers were discharged, one would expect that the aggregate data would show an upsurge in labor productivity if one reasonably presumes that these higher paid senior workers were more highly paid because of their greater productivity owing to more experience, skills, etc. That productivity surge or even productivity stability, however, in fact did not occur: we estimate productivity fell 5.7 percent from the fourth quarter of 1929 to the fourth quarter of 1930. If aggregation bias impacted the true interpretation of wages, it should have also had an impact on productivity that the data do not support. In a sense, what is critical is not the real wage rate, but the real wage rate adjusted for productivity change.
Third, there is absolutely no question that certain government policies had a dramatic impact on raising wages in the middle of the Great Depression and prolonging its duration. The mammoth increase in hourly wages from June to December 1933, for example, certainly is not a reflection of “aggregation bias,” but rather of the effects of the implicit minimum wages applicable under the National Labor Recovery Act (wages in major industries increased something on the order of 20 percent). Similarly, large (double digit) wage increases in 1937 were a reflection of the delayed impact of the Wagner Act, particularly after a court decision rendering the law constitutional.
… Suffice it to say that looking at the wage behavior of those firms whose president attended the employment conference of, say, November 21, 1929, as opposed to those not attending, appears to us to be a very dubious approach, particularly since no one to our knowledge has ever claimed that Hoover’s impact was solely on the small number of industrial leaders in the room — the exhortations of Hoover and reports of the conference were well known to everyone, since they made the front pages of major newspapers. [...]
Allow us to give our interpretation of the Depression experience, which we have elaborated upon elsewhere (1997, 2000). In the latter, we demonstrate that all the major macroeconomic paradigms have in common the following relationship:
(1) E = f(W/(P, O)
where E denotes employment, W represents the money wage level, O indicates a measure of the productivity per unit of labor, and P signifies the general level of prices. This formulation embraces the actual co-ordination, or dis-coordination, among all the elements relevant to labor markets. A brief description of one of the primary data sets we collected for Out of Work is illustrative. It begins with the first quarter of 1959 and concludes with the second quarter of 1996, a span of 150 quarters. All data are expressed in index number form with the year 1992 = 100. While the subcomponents (W, P, and O) vary considerably over time, the productivity-adjusted real wage rate has a maximum value of 104.14 (in 1980.2) and a minimum of 96.24 (in 1965.4). Basically, the productivity-adjusted real wage rate constitutes a stationary time series with a plus or minus four percent range of variation. During the Depression, variations were somewhat larger, but most changes in one key variable were considerably but not completely offset by moves in another one. For example, as productivity fell, money wages tended to move downward (although not enough — creating most of the massive rise in unemployment).
This is the essence of our story. The productivity-adjusted real wage rate is the product of a set of systematic interactions between the price level, money wage rates, and the productivity of labor. What is implied here is that the real wage rate and the productivity of labor move in near lock-step. Only “near” lock-step, though. There are divergences, and these divergences are capable of generating business cycles in the American economy.
More to the point is the significance of the productivity-adjusted real wage rate for the particulars of this debate. To begin, it largely vitiates the aggregation bias criticism. It stands to reason that if employers dismiss low money wage workers first during a cyclical downturn, these workers will also be lower productivity individuals. While it may not be a perfect match, even an approximate one will alleviate greatly any aggregation bias, rendering the productivity-adjusted real wage rate a reasonably accurate approximation of the relationship between the real wage rate and productivity.
… For example, Rose’s critique treats money wage rates in isolation. What is important is the behavior of money wage rates relative to the general levels of prices and labor productivity. In such a context, we conducted a specific statistical test (1997, pp. 95–96) of Hoover’s high-wage doctrine. What we did was use the annual data for the years 1901–1929 describing money wage, price, and productivity levels to estimate a function explaining money wage rates, with prices and productivity being the major independent variables accounting for movements in money wage rates. We note that any aggregation biases in the money wage and productivity variables should be of roughly similar magnitude and thus will only affect the constant term in any linear regression. In addition to these independent variables, given the importance of immigration in this era and the enactment of the Smoot-Hawley tariff in 1930, we also included independent variables measuring the amount of immigration and the level of tariffs in the money-wage function. The full estimated relationship is:
(2) WAGES = – 3.486 + 0.225•CPI + 0.145•PRDTY
. (20.485) (16.650) (13.797)
. – 0.008•TARIFF – 0.014•IMM/POP
. (2.614) (0.224)
. R² =.9964, D-W:1.978, F-stat.:1585.358
where WAGES is an hourly wage measure, CPI represents the consumer price index, PRDTY indicates hourly output of labor, TARIFF represents the percentage tariff levied on durable goods, and IMM/POP is immigration as a proportion of the population. The values in parentheses beneath the coefficients are t-statistics.
We then employed the 1930 and 1931 values of the independent variables to calculate the level of money wage rates that would be expected to occur in these years. These were then compared to the actual values of money wage rates. The results are striking. In 1930, actual money wage levels exceeded the expected by 8.3 percent. In 1931, the overhang is 10.5 percent. Between the fourth quarters of 1929 and 1930, money wage levels fell by a mere 1.7 percent. At the same time, prices declined by 7.0 percent and labor productivity dropped by 5.7 percent. As a result, the productivity-adjusted real wage rate rose by an astounding 12.0 percent and the estimated unemployment rate broke into double digits at 10.7 percent.
Four quarters later, at the end of 1931, money wages were 7.9 percent below their fourth quarter 1929 level. However, prices were now 16.6 percent less than in 1929.4 and labor productivity was down by 7.5 percent. The productivity-adjusted real wage? Up by 19.3 percent over 1929.4. The unemployment rate? 18.4 percent. Now, perhaps Herbert Hoover’s actions in late 1929 were irrevelant, an unfortunate coincidence. However, a far more likely scenario is that Hoover was successful in implementing the “high-wage doctrine,” just as many contemporary observers opined.