The Use of Financial Spreads as Indicator Variables: Evidence for the U.K. and Germany


WP/94/31-EA
The Use of Financial Spreads as Indicator Variables: Evidence for the U.K.
and Germany by E.P. Davis and S.G.B. Henry


There has been considerable attention paid to the possibility that
financial spreads might be useful for predicting cyclical movements in
aggregate output. Spreads are thought to contain timely information for a
number of reasons, including the sensitivity of spreads to changes in
expectations of cyclical changes. Spreads may widen due to anticipated
increases in default risk as an economy slows down, or as monetary policy is
tightened. It is natural to ask what additional explanatory power financial
spreads have for predicting changes in aggregate output when other
information, including lagged changes in output and other macro economic
variables, are taken into account?

In this paper, empirical models are used to investigate this question
with reference to the United Kingdom and Germany. The same methodology is
followed for each case. Four financial spreads are used. The first is the
long-term credit quality spread, which is the difference between yields on
corporate and government bonds of the same maturity. The second is the
yield curve, the differential between long and short rates. The final two
are both reverse yield gaps. One is defined as the spread between the
long-term bond yield and the dividend yield, the other is between bond
yields and the earnings yield. The empirical tests consist of first
estimating a robust and well-fitting Vector-Auto Regression (VAR) model of
the economy, including output, the output deflator, and other macrovariables
(six are used in total). Each model is then augmented by the four spreads,
and a ten-equation VAR is estimated for each country. A variety of tests of
the information contained in the spreads shows that they clearly contribute
significantly to explaining changes in activity. In addition, ex-post tests
of the predictive properties of the models indicate that the models that
include spreads are more accurate in anticipating the recent cyclical
downturn that occurred in the United Kingdom and Germany than the VAR models
that do not include spreads.
The paper concludes that, on the basis of these tests, financial
spreads may have an important empirical role in accounting for cyclical
movements in aggregate activity.