Agricultural Shocks in the Interwar Economy: A Structural VAR Analysis

Matthew A. Martin, Kent State University

NOTE: The summary which appeared in the October 1997 issue of The Newsletter of The Cliometric Society contained graphics and formatting which cannot be accurately reproduced in this version.

In this paper, I examine the macroeconomic impact of agricultural events in the interwar period. These include overproduction, technological change, innovative government aid, and severe drought. Of particular interest is the possibility that the agricultural sector significantly affected the course of the recovery in the 1930s. Investment into new farm technologies, which had stalled during the Great Depression, resumed as government assistance provided farmers with the income needed to purchase new farm technology. This investment into farm capital constituted a beneficial supply shock that promoted industrial production and aided general recovery, despite severe drought.
U. S. farmers were slow to adjust production in the wake of diminished wartime needs. As Figure 1 shows, the prices of farm commodities fell considerably after World War I. The price of wheat, for example, fell from $2.16 per bushel in 1919 to just 97 cents per bushel in 19221. Farmers were left with excessive debt and diminished cash incomes. Farm bankruptcies, which had averaged 1200 per year in the second decade of this century, averaged 5186 per year in the 1920s.2 Despite these circumstances, farmers were buying tractors by the thousands. The tractor had been in existence for over a decade, but improvements in the early 1920s made the tractor more efficient and cost effective. Figure 2 shows that tractor sales reached a pre-depression peak of 140,000 in 1920.
During the Great Depression, farm prices fell more quickly than did the general price level, exacerbating the difficulties of farmers. Providing farm relief was a cornerstone of the New Deal and the Agricultural Adjustment Administration (AAA) was organized in 1933. The primary goal of the AAA was to provide relief farm through production control. This was accomplished through benefit payments to farmers who agreed to limit crop production.
In addition to the production control programs of the AAA, the New Deal provided aid to farmers through the Commodity Credit Corporation (CCC) and the Farm Credit Administration (FCA). The first of these programs put cash into the hands of farmers by loaning them money based on the amount of a crop in storage multiplied by a fixed price per bushel. If the market price of the crop climbed above the loan price, farmers could sell the crop, pay off the loan and pocket the difference. The FCA enabled farmers to borrow new funds and refinance existing debt. Thus, farmers were able to reduce the burden of the debt accumulated in the previous decade at the same time that they received security regarding their cash incomes in the near future.
In 1933, the first of a series of droughts afflicted the important growing regions in the Midwest. Since the drought seemed likely to accomplish production control, beginning in 1934 Secretary Wallace removed many of the planting restrictions that contract signers had agreed to previously. However, farmers still received the benefit payments promised by the government. A similar production control program was in effect for 1935.
As farmers began to receive cash income and other assistance through government programs, the number of tractors they purchased increased from 25,000 in 1933 to 221,000 by 19373. Similarly, the purchase of other farm implements, especially those pulled by tractors, increased dramatically. Only 15,000 tractor drawn cultivators were purchased in 1931, a figure that increased to a record 127,000 by 19374. 1937 was also a record year for the purchase of corn pickers and tractor plows.
In all, the AAA distributed 1.4 billion dollars as benefit payments to farmers under the production control programs (1933 to 1936). Additional payments were made after 1936 under a soil conservation program. The FCA was credited with helping half a million farmers keep their homes (Saloutos, 1982). Rucker and Alston (1987) estimate that farm credit programs alone prevented approximately 77 thousand farm failures, while all forms of government aid are estimated to have prevented approximately 186 thousand failures. Additionally, Commodity Credit Corporation loans totaled nearly 1.5 billion dollars for the years 1933 through 1939.
The business cycle literature from the early part of this century describes several ways in which agricultural shocks can affect the national economy. These mechanisms are appropriate for an economy with a large agricultural sector, which would include the interwar U.S. economy. Pigou (1927) discusses a psychological effect related to harvests. A good harvest buoys consumer expectations, increasing consumption, while a bad harvest has the opposite effect. Timoshenko (1930) lists several mechanisms through which agricultural shocks might affect the supply side of the economy. First among these is the fact that farm products are an input to production so that an increase in their prices, due to drought or some restriction on production, diminishes firms' profit margins.
The events of the Great Depression highlight another mechanism for the propagation of agricultural shocks. As noted by Schumpeter (1939) at the time, and more recently discussed by Clarke (1991), agricultural was in the midst of a mechanization process when the depression began. Clarke argues that the New Deal programs provided farmers with the security and cash flow needed to purchase a tractor, which had already proven to be a cost saving device for many farms. This is reflected in the number of tractors purchased in the Depression. As Figure 2 shows, only 25 thousand tractors were purchased in 1932 and 1933. Beginning in 1934, this number rises steadily to a peak of 221 thousand in 1937, which was considerably greater than the 1929 peak of 137 thousand tractors purchased. This increase in the purchase of tractors occurred during repeatedly severe droughts in the Midwest.
If the government enabled farmers to adopt tractors and related farm implements more quickly, then the 1930s should have been a period of substantial gains in productivity. This is indeed the case. Clarke notes that productivity increases averaged only 0.8% in the 1920s. For the years 1934 to 1939, the years in which the government programs would have enabled farmers to purchase more equipment, the average gain in productivity is 3%.
The migration of labor during the interwar years provides additional evidence that labor productivity was increasing. From the early 1920s until 1931, there was a steady migration of labor away from the farm. This migration was slowed in 1932 and actually reversed in 1933. These are the years in which investment into farm machinery is the lowest for the interwar years. In fact, investment into farm machinery was not keeping pace with depreciation, so that the value of farm capital decreased in the early 1930s.
The years 1936 and 1937 provide perhaps the best example of the increases in farm productivity. The value of farm output rose 11.6% in 1937 relative to the drought stricken year of 1936.5 Furthermore, the year 1937 was also a record year for the purchases of tractors. In that same year, 690,000 persons migrated away from farms, the most in any year since 1927.6 The increase in farm output would have tended to increase farm employment. The fact that migration away from the farm was large supports the view that the increased use of labor saving devices was enhancing farm productivity.
Any empirical analysis of agricultural events must be able to account for the competing effects within the farm sector in the presence of other macroeconomic shocks. Of particular interest is the relative effects of agricultural developments in the context of monetary shocks due to bank failures and gold flows. Friedman and Schwartz (1963), Temin and Wigmore (1990), Romer (1992) and Eichengreen (1992) have analyzed various aspects of monetary developments. In this study, I use a structural vector autoregression (VAR) method developed by Bernanke (1986) to evaluate the importance of agricultural shocks. The structural VAR is well suited to this type of analysis for two primary reasons. First, it permits the identification of unobserved, independent shocks, such as an agricultural supply shock and a money stock shock. Second, VAR representations can track the effects of these shocks over time. Of particular interest, impulse response functions with bootstrapped confidence intervals can assess the direction and significance of individual shocks over various time horizons.
There are several ways to model the interwar years. Based on Temin and Wigmore (1990), I model the interwar years (1921-1939) with a trend break in the second quarter of 1933. Temin and Wigmore argue that the abandonment of the gold standard by the newly inaugurated Roosevelt administration signaled a policy regime change that altered agents expectations. This is compatible with the notion that farmers began buying more tractors when the government provided them with some amount of income security and loan assistance.
I estimate two identifications. The first includes a grain price index, the gold stock, the monetary base, the money stock, industrial production, and the wholesale price index. The second identification replaces the gold stock with the treasury bill rate. The results from the empirical analysis are variance decompositions and impulse response functions. For both identifications, a positive shock to the grain price series increases industrial production and the price level.
To understand the significance of the empirical results, I provide examples of the size and frequency of the effect of large, positive shocks to grain prices. Table 1 presents farm expenditures as a ratio of the two measures of economic activity. Expenditures on farm equipment comprise as much as 20.5% of all expenditures on producers equipment, and between 0.62% and 1.68% of industrial production. These amounts are consistent with the results of the structural VAR analysis. Using the results from the first structural model, the largest significant response of industrial production to a one standard deviation shock to grain prices occurs at the three quarter horizon and has an approximate value of 0.025. Since the series are in natural logarithm form, this can be interpreted as a percentage response in decimal form. The mean value for industrial production index for this sample is 184.9. This value would increase to 189.6 based on the response due to a grain price shock three quarters ahead. Industrial production would have increased by approximately 3% to 190.4 over the four quarter period after the price shock, based on the point values from the IRFs.
The preceding discussion is predicated on the assumption that there were grain price shocks of a large enough magnitude to alter industrial production as described. In levels, a one standard deviation change in the grain price index is a change of 22.59, or a 27.5% change for the mean value of 82.12. During the interwar period, there were seven quarters where grain prices changes by approximately one standard deviation from the previous quarter (greater than 20% change). These are listed in Table 2. Five of these shocks are positive; and four occur during the recovery years.
The repeated shocks that are clustered in the years 1933 to 1936 are concomitant with the large increases in farm implement expenditures. If each increase in grain prices led to a three percent increase in industrial production over the next year, then these shocks account for approximately 12.5% of the total increase in industrial production that occurred from 1933 to 1937. The total increase in industrial production over this time period was 79.8%, so nearly one-sixth was due to agricultural developments.
This study suggests three conclusions. First, contrary to generally held perceptions, the economic recovery of the 1930s was aided by government agricultural assistance programs. Or, put another way, the recovery would have been slower without the security and income provided to farmers during the droughts. With the government programs, farmers were able to invest in new capital equipment that effectively increased farm productivity. Under conditions of uncertainty and low income during the depression, farmers had postponed investment in new equipment, creating a backlog of investment into new technologies such as the tractor. It is a remarkable fact that more tractors were purchased in the 1930s than in the 1920s, despite the depression.7
Second, most of the largest, positive shocks to the farm sector occur during the early recovery years. Individually, these shocks may not have been enough to produce large changes in the macroeconomy. The clustering of the largest price shocks during the recovery years, however, was enough to help promote general recovery. The timing of the shocks provides evidence that agriculture, aided by the government, contributed to the early recovery.
Third, this explanation does not interfere with the monetary explanations for the recovery. In fact, it complements the monetary story by providing an explanation for the initial stages of the recovery before the influence of the gold inflow could have been realized in the economy. Specifically, two of the large, positive price shocks occur in 1933, before the revaluation of the dollar price of gold in early 1934 and the strong gold inflow that followed. Thus, the initial turning point in the Great Depression may be traced to the change in policy regime, as described by Temin and Wigmore; while farm sector developments provided for some gains in industrial production until monetary developments were able to have their full effect on the economy.

1 Agricultural Statistics (1941), 9.
2 Agricultural Statistics (1938),457.
3 Clarke (1991), Table 1.
4 Agricultural Statistics (1940), 561.
5 National Income and Product Accounts, Table 1.18.
6 Historical Statistics of the United States, 47.
7 There were 1,077,000 tractors purchased from 1920 to 1929. There were 1,107,000 tractors purchased from 1930 to 1939, 883,000 of which were purchased from 1934 to 1939. [Clarke (1991)].

Bernanke, Ben S. "Alternative Explanations of the Money-Income Correlation," Carnegie-Rochester Conference Series on Public Policy 25 (1986): 49-100.
Clark, Sally. "New Deal Regulation and the Revolution in American Farm Productivity: A Case Study of the Diffusion of the Tractor in the Corn Belt, 1920-1940," Journal of Economic History 51 (March 1991): 101-123.
Eichengreen, Barry. Golden Fetters. Oxford: Oxford Univ. Press, 1992.
Freidman, Milton and Anna J. Schwartz. A Monetary History of the United States, 1867-1960. Princeton: Princeton Univ. Press, 1963.
Nourse, Edwin G., Joseph S. Davis, and John D. Black. Three Years of the Agricultural Adjustment Administration. Washington: Brookings Institute, 1937.
Perkins, Van L. Crisis in Agriculture: The Agricultural Adjustment Administration and the New Deal, 1933. Berkeley, 1969.
Pigou, A. C. Industrial Fluctuations. London: MacMillan, 1927.
Romer, Christina D. "What Ended the Great Depression?," Journal of Economic History 52 (December 1992):19-39.
Rucker, Randal R. and Lee J. Alston. "Farm Failures and Government Intervention: A Case Study of the 1930s," American Economic Review 77 (1987): 724-730.
Saloutos, Theodore. The American Farmer and the New Deal. New York: Ames, 1982.
Schumpeter, Joseph A. Business Cycles. New York: McGraw-Hill, 1939.
Temin, Peter and Barry Wigmore, "The End of One Big Deflation," Explorations in Economic History 27 (Oct. 1990): 483-502.
Timoshenko, Vladimir P. "The Role of Agricultural Fluctuations in the Business Cycle," Michigan Business Studies 2 (June 1930), no. 9.
U.S. Department of Agriculture. Agricultural Statistics. Washington, DC: Government Printing Office, various years.

Figure 1, Figure 2, Table 1 Ratios of Farm Implement Expenditures to Producers Equipment, Finished Commodities, and Industrial Production, Table 2 The Quarter and Size of the Largest Grain Price Increases