The Mistake of 1937A General Equilibrium Analysis: The Mistake of 1937 A General Equilibrium Analysis Gauti Eggertsson Federal Reserve Bank of NY Disclaimer: The views expressed are those of the authors and do not reflect those of the Federal Reserve Bank of New York or the Federal Reserve System. Benjamin Pugsley University of Chicago November 2006


Introduction: Introduction The economic conditions this paper analyzes: Signs that the economic slowdown may finally be over The short-term interest rates is zero There have been large increases in the monetary base in recent years and record accumulation of “excess reserves” by banks. Growing concern over excessive inflation on the horizon. This describes the US in the spring of 1937, just before a major policy mistake: The Mistake of 1937


Introduction: Introduction What happened in 1937? One of the biggest recessions in US economic history. What was the mistake of 1937? We argue that it was bad communication policy. During the recovery phase 1933-37 the administration committed to reflating the price level to pre-depression levels, often referred to as price level of 1926. In the spring of 1937 administration officials, President Roosevelt, Federal Chairman Eccles and others warned of too much inflation, even if prices had not risen to pre-depression levels. Our hypothesis: This was interpreted as a regime shift and had a large negative effect expectations, and through expectations on inflation and output.


Slide4: Industrial Production Subtitle Text X-Axis Text Index Index Source: Federal Reserve Board Contact Person Text (1929=100)


Slide5: Price Indices (1929=100) X-Axis Text Index Index Source: NBER Macrohistory Database Note Text Contact Person Text CPI (1929=100) WPI (1929=100)


Slide6: Some other reasons to be interested Subtitle Text X-Axis Text Index Index Source: Federal Reserve Board Contact Person Text (1929=100)


Previous literature on 1937: Previous literature on 1937 Some difference with previous literature (e.g. Friedman and Schwartz (1963)), who emphasize increase in reserve requirement in 1936 and January 1937. In this paper the money supply was irrelevant during this period because interest rates were close to zero. Expectations about future money supply (once deflationary pressures had subsided and interest rates would be positive again), however, were crucial. Monetary policy was extremely important during this period but for reasons that are different that the “monetarists” claim – expectation channel the key. The increases in the reserve requirement in 1936 and 1937 were unimportant except to the extent that they were interpreted as signals about future policy.


Slide8: At zero interest rates inflation and output are extremely sensitive to expectations about future policy. “Bad communication”, i.e. signaling excessive tightening, can have dramatic negative effect due to vicious feedback effects at zero interest rates, especially if expectations are conditioned on a reflationary regime. Much bigger effect of “communication policy” at zero interest rates than under regular circumstances. Speculative conclusion: Since the effect is asymmetric a conservative approach would be to err on the side of inflation (Japan?!!). Central Conclusions of the paper


The ModelHousehold maximization: The Model Household maximization Utility s.t. budget constraint Consumption and price indices


The ModelHouseholds: The Model Households Necessary conditions for optimum


The ModelFirm maximization: The Model Firm maximization Firm maximize profits, but as in Calvo (1982) can only adjust prices with probability 1-δ giving rise to the FOC Equilibrium: Set of stochastic processes for {Yt, it, Pt} for a given path of {bt} initial condition for P-1


The ModelApproximation: The Model Approximation


Policy Regimesdetermination of it: Policy Regimes determination of it Probability= arbitrarily small Probability = Changes in reflect changes in beliefs about future policy


The source of the Great DepressionTemin shocks: The source of the Great Depression Temin shocks -- Purely intertemporaral shock


Output Collapseunder the deflationary regime: Why output collapse? Output Collapse under the deflationary regime Government targets zero inflation


Recovery and excess sensitivityunder the Reflationary Regime: Recovery and excess sensitivity under the Reflationary Regime Government targets positive inflation target with imperfect credibility Tries to set π=4% when possible Each period in the reflationary regime, there is probability γt that government will abandon its policy in favor a deflationary regime in all future periods We interpret γt as varying with the public’s change in belief due to policy communication If gamma increases, means that the public beliefs monetary policy will be tighter in the future. Little bit tricky to solve because of non-linearity of the zero bound and the time-varying property of the probability of regime change makes the model also highly non-linear. Sneaky simplifying assumption: Absorbing state for both processes Can solve the system of equations backwards, starting from the absorbing state.


The ModelReflationary Regime: The Model Reflationary Regime


Small change in beliefs have a big effect Output and inflation in the low state for a given set of beliefs-- no change in interest rates needed: Small change in beliefs have a big effect Output and inflation in the low state for a given set of beliefs -- no change in interest rates needed Output Inflation percent


Asymmetric effects of inflation vs. deflationary signals: Asymmetric effects of inflation vs. deflationary signals Show in paper that sending signals of too much inflation has much smaller effect on output or inflation Reason: Persistent deflationary expectations interact viciously with the zero bound, deflationary signals imply deflation in all states of the world in which the zero bound is binding which creates propagation that may not even converge. One “speculative” interpretation of this finding: It is prudent to err on the side of “too much inflation” in a transition phase out of zero interest rates especially when deflationary expectations may be hard to reverse. Main point: Extreme volatility of the model with respect to small changes in beliefs can explain the recession of 1937 without any changes in interest rates!


The Mistake of 1937: The Mistake of 1937 Series of confusing policy communications from key administration officials Signals of future policy change which implies a strong reaction beyond static changes in interest rates would imply. The market started to believe that the government was going to target zero inflation. Reneging on previous announcements to increase the price level to pre-depression levels. Where do we find evidence for this hypothesis? Our approach  Read the newspapers!  Study futures data (in progress)


Slide21: Reminder! (1929=100) X-Axis Text Index Index Source: NBER Macrohistory Database Note Text Contact Person Text CPI (1929=100) WPI (1929=100)


Slide23: Intensity of Policy Discussion Mentions of Inflation by Eccles, Morgenthau, Roosevelt, or his Cabinet X-Axis Text Number of Matches Number of Matches Source: Proquest Historical Newspapers Note Text Contact Person Text


Slide24: Commodity Prices (February 1937 = 100) X-Axis Text Index Index Source: NBER Macro History Database Note Text Contact Person Text wheat cotton corn cattle hogs copper gasoline


Slide25: Stock Prices SP500 Index X-Axis Text Index Index Source: Wall Street Journal Note Text Contact Person Text


The Mistake of 1937Data through the prism of the model: The Mistake of 1937 Data through the prism of the model Question: Can movements in the parameter γt generate movements in output and inflation of the order observed? Pick free parameter (γt) to minimize the squared deviations from monthly inflation and output (industrial production)


The Reversal of 1938: The Reversal of 1938 February 1938: FDR and administration begin communicating a clear strategy to resume an inflationary policy


Slide31: 1938


Slide32: Commodity Prices (February 1938=100) X-Axis Text Index Index Source: NBER Macro History Database Note Text Contact Person Text wheat cotton corn cattle hogs copper gasoline


Slide33: Industrial Production Subtitle Text X-Axis Text Index Index Source: Federal Reserve Board Contact Person Text (1929=100)


Conclusions: Conclusions In an economy with maximum policy accommodation of zero interest rates Local to credible reflationary regime, output and inflation are extremely sensitive to communications about future policy The effect of communicating excessive tightening has much worse implications than overshooting due to interaction between the zero bound and expectations that it will be binding in the future with associated deflation. Implications for policymakers Need to communicate their objectives clearly and consistently? Err on the side of more inflation?


The Next Step:Commodity future prices (future research): The Next Step: Commodity future prices (future research) September futures of corn, wheat, lard, oats, rye.


Long rates rose around the mistake of 1937 -- beyond what was implied by short rates increases: Long rates rose around the mistake of 1937 -- beyond what was implied by short rates increases