A multivariate analysis of interest rate seasonality at the time of the founding of the Federal Reserve.

• Author: Chambers, Catherine M.

1. Introduction

   Miron [9;10], Clark [2], Mankiw, Miron and Weil [7], and Barsky et al. [1] have used univariate methods to demonstrate that the seasonality in interest rates changed in the period following the creation of the Federal Reserve System. They have also debated whether these changes were attributable to the founding of the Fed or to some other event such as the collapse of the gold standard following the beginning of World War I.

   This paper, in contrast, will use a vector autoregression (VAR) to determine the source of the change in interest rate seasonality. We will show that the change in interest rate seasonality is at least in part due to the transmission to the interest rate of changes in seasonality of other variables. In particular, when seasonal variation in other variables is eliminated, so that a "counterfactual path" for interest rates is determined, we find no tendency for the seasonality of these counterfactual interest rates to decline after the founding of the Fed. These results suggest that the decline in interest rate seasonality noted in simple autocorrelation functions computed before and after the end of 1914 is likely more complex than previous discussions have implied.

   The VAR used here includes five variables: the wholesale price index, the call interest rate, an index of stock prices, high powered money, and industrial production. We employ monthly data covering the period 1890 to 1929.(1) This VAR is subjected to a stability test with a hypothesized break in November 1914, which corresponds to the opening of the Federal Reserve. The literature has established from both economic and statistical perspectives that for single equation models of the interest rate, this is a reasonable break date.(2) Using a bootstrap technique, we extend this result, finding a significant shift in the system of equations at the date of the founding of the Fed. We next attempt to determine the sources of the shift by testing the equations in the system individually. The nature of the revealed shifts is explored further through the use of historical decompositions (HDCs).

   The paper is organized as follows. A brief review of the literature regarding the seasonality of interest rates at the time of the founding of the Fed is in section II. Section III contains a preliminary examination of the data. Results of the stability tests are given in section IV, and the HDC results are discussed in section V. Section VI presents conclusions.

   lI. Literature Review

   The seasonality of interest rates prior to 1914 has been widely attributed to seasonal increases in the demand for credit and currency which were not matched by increases in money supply. As early as 1910, members of the National Monetary Commission reached this conclusion and reported it to Congress [12]. In the agricultural sector, additional funds were needed for spring planting and fall crop moving. Because of the inelasticity of the money supply, this increase in demand pushed interest rates up in those seasons. Miron [9] asserts that additional funds were also needed in the corporate sector during the spring and fall because of increases in the volume of transactions and quarterly interest and dividend settlements.

   The creation of the Federal Reserve System was designed to increase the elasticity of the money supply, and thus eliminate the seasonal fluctuations in interest rates. Secretary of the Treasury McAdoo applauded the opening of the Federal Reserve Banks in the New York Times on 16 November 1914:

   It is believed that they will put an end to the annual anxiety from which the country suffered for the last generation about insufficient money and credit to move the crops each year, and will give such stability to the banking business that extreme fluctuations in interest rates and available credits which have characterized banking in the past will be destroyed permanently.

   Miron [9] argues that by adopting a seasonal money supply policy, the Fed eliminated the seasonality in interest rates that predominated prior to 1915. This increase in the stability of interest rates, in turn, decreased the frequency of financial panics. Mankiw, Miron, and Well [6] attempt to pin down the timing of the change in interest rate behavior and find that the most likely date of the shift is December 1914. Since the Fed opened for business in November 1914, this timing lends support to the view that the creation of the central bank was responsible for the change in seasonality.

   In contrast to Mankiw, Miron, and Weil, Clark [2] argues that a decline in interest rate seasonality in Britain, France and Germany at about the same time as in the U.S. makes it unlikely that an event unique to the U.S. was responsible for the change in interest rate behavior. Clark also finds that the seasonality of interest rates disappeared sometime between early 1913 and early 1915 while the seasonality of high powered money did not begin until September 1917. Thus, he argues that the change in interest rate seasonality cannot reasonably be attributed to seasonal monetary policy and that the beginning of World War I and the resulting breakdown of the gold standard were more likely the causes. He concludes that the change in interest rate behavior cannot reasonably be attributed to seasonal monetary policy. Note that, although he disagrees regarding the cause, Clark does agree with Mankiw, Miron and Weil, at least roughly, about the timing of the shift in the behavior of interest rates.

   In response to Clark, Barsky et al. [1] construct a two-country model showing that, in the presence of a gold standard, neither country can stabilize interest rates unless both have a central bank in place. This model suggests that the change in the behavior of British interest rates, for example, could only occur after the founding of the Fed. They also provide an empirically-based argument that the breakdown of the gold standard probably did not cause the change in interest rates; comparing the events of 1915 to those following the collapse of the Bretton Woods system in 1970, the authors find no changes in the latter period that correspond to those in 1915.

   Holland and Toma [4] examine another channel through which the founding of the Fed may have led to a decrease in the seasonality of interest rates: the Fed's function as a lender of last resort. This...