NBER WORKING PAPER SERIES
THE CHANGING
BEHAVIOR
OF THE
TERM STRUCTURE OF INTEREST MTES
N. Gregory Mankiw
Jeffrey A. Miron
Working Paper No. 1669
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
July 1985
We are grateful to Rudiger Dornbusch, Roger Gordon, Phil Howrey,
David Romer, Lawrence Sunners, Peter Temin, and three referees for
helpful connents. The research reported here Is part of the NBER's
research programs In Economic fluctuations and Financial Markets
and Monetary Economics. Any opinions expressed are those of the
authors and not those of the National Bureau of Economic Research.
N8ER Working Paper #1669
July 1935
The Changing Behavior of the
Term Structure of Interest Rates
ABSTRACT
We reexamine the expectations theory of the term structure
using data at
the short end of the maturity spectrum. We find that prior to the
founding of
the Federal Reserve System in 1915, the spread between
long rates and short
rates has substantial predictive power for the path of interest
rates; after
1915, however, the spread contains much less predictive
power. We then show
that the short rate is approximately a random walk after the
founding of the
Fed but not before. This latter fact, coupled with even
slight variation in
the term premium, can explain the observed change in 1915 in the
performance
of the expectations theory. We suggest that the random walk character
of the
short rate may be attributable to the Federal Reserve's commitment
to
stabilizing interest rates.
N. Gregory Mankiw
Jeffrey A. Miron
National Bureau of
University of Michigan
Economic Research
Department of Economics
1050 Massachusetts Ave.
Ann Arbor, MI 48109
Cambridge, NA 02138
I. Introduction
The most prevalent explanation of fluctuations in the
yield curve is
the expectations theory, which posits that the slope of the
yield curve
reflects the market expectation of the future change in interest
rates.
Numerous studies, however, present evidence that the data
are Inconsistent
with the joint hypothesis of the expectations theory and rational
expectations.1 Indeed, the rejections of the expectations
theory date back
at least to Macaulay (1938, p. 33]. who pointed out the implications of
the
theory but concluded that "experience is more nearly the opposite."
Perhaps the most striking rejections use data at only the short end of
the maturity spectrum. Recently, Fain (1984], Jones and
Roley (1903],
Hankiw and Summers (1984], and Shiller, Campbell, and Schoenholtz
(1983]
all conclude that yields on Treasury bills of less than
one year do not
obey the expectations theory. While stories of highly variable risk
premiums, changing asset supplies, or segmented markets might explain the
failure of very long—term yields to behave according to the
theory, such
stories seem less plausible applied to the markets for three—month and
six—month bills.2
Although the number of studies rejecting the theory Is large, the
results of these studies are not independent. Indeed, they examine almost
identical periods of history, primarily the 1960s and 1970s. the
period
during which an active market in three—month and six-month Treasury bills
existed. It is reassuring that these studies reach the same conclusion.
—2—
but confirmation requires examination of truly independent data.
In this paper we examine the term structure of interest rates at the
short end of the •aturity spectru. for the period from 1090 to 1979. We
divide our sample into different monetary "regimes" to examine whether the
failure of the expectations theory is robust. Our goal is to identify the
conditions under which the expectations theory works badly and the
conditions, if any, under which it works well.
In Section II we briefly review the expectations theory. The theory
posits that there are no expected profit opportunities. it implies that
the spread between the long rate and the short rate predicts the path of
the short rate.
We discuss the data in Section III. Prior to the founding of the
Federal Reserve System, the National Monetary Commission in 1910 collected
extensive data on interest rates and banking. We have extended the data on
three—month and six—month time rates through 1958. This data set provides
an opportunity both to reexamine findings based on more recent data and to
expand our understanding of the earlier historical period. We argue that
it provides a good data set with which to examine the expectations theory.
In Section IV we present tests of the expectations theory both
with the older data and with Treasury bill data from the 1960s and 1970s.
The results are surprising. While we confirm the failure of the
expectations theory using recent data, we find that the expectations theory
works •uch better during some previous monetary regimes. In particular,
for data prior to the founding of the Federal Reserve, the slope of the
yield curve has substantial predictive power for the path of the short
—3—
rate.
In Sections V and VI we propose an explanation for the difference in
the perfor.artce of the expectations theory in the different periods. If
the term premium varies through time, then the expectations
theory will be
rejected using the standard test. The extent of the failure, however,
depends on the variance of predicted changes in the short rate. We
argue
that the relative success of the theory with the data from before the
founding of the Fed is attributable to the greater predictable changes in
the short rate.
In Section VII we discuss the role of the Federal Reserve and its
relation to the performance of the expectations theory. With the creation
and increased activism of the Fed, changes in the short rate became less
predictable and the expectations theory performed more poorly. We
speculate that the failure of the expectations theory using post—Fed data
may be due to the Federal Reserve's commitment to stabilizing interest
rates.
We conclude in Section VIII by discussing the implications of our
results for the expectations theory of the term structure under recent
monetary and fiscal regimes.
II. The Expectations Theory of the Term Structure
In this section we briefly review the expectations theory for one—
period and two—period bills. Let rt be the yield on a one-period bill, and
let Rt be the yield on a two—period bill. The expectations theory
posits that
