The Use of Financial Spreads as Indicator Variables: Evidence for the United Kingdom and Germany
  • 1 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund
  • | 2 0000000404811396https://isni.org/isni/0000000404811396International Monetary Fund

There has been growing interest in the use of financial spreads as advance indicators of real activity and inflation. Empirical evidence is marshalled on a range of spreads when these are used in vector autoregressive models of the U.K. and German economies. It is found that they do provide significant information, even after allowing for the effects of other influences upon macroeconomic activity.

Abstract

There has been growing interest in the use of financial spreads as advance indicators of real activity and inflation. Empirical evidence is marshalled on a range of spreads when these are used in vector autoregressive models of the U.K. and German economies. It is found that they do provide significant information, even after allowing for the effects of other influences upon macroeconomic activity.

Indicator models for the prediction of real activity and inflation have recently received increasing attention. A major reason for this renewed interest in indicator methods is the perceived failure of structural macroeconomic models in many countries to forecast the recent falls in, and indeed the subsequent increases in, output. Structural models are widely thought to be particularly bad at predicting turning points. Considerations such as these are at the center of the influential work by Stock and Watson (1989), whose indicator model is expressedly designed to predict turning points. Generally, indicator models are “nonstructural” approaches to prediction, so that where monetary indicators, for example, are used, they are not based on an explicit model of the transmission mechanism of monetary policy. This avoidance of specific structural hypotheses may indeed be an advantage of indicator models since it avoids the strictures of Sims (1980), who argued that structural models apply “incredible” identification restrictions.

Recent work on indicator models has focused on the possibility that there is advance information in financial markets, so that movements in financial spreads give early warnings of changes in activity and inflation.

The aim of this paper is to provide an extensive empirical evaluation of spreads as indicators, by comprehensively testing them to predict output and inflation. The tests are done using nonstructural Vector-Auto Regressive (VAR) models, which are initially quite large (six equation models are used), so that a significant amount of information is used before including the financial spreads. The tests are thus directed at whether the financial spreads add to the explanatory power of what is already a fairly extensive model.

The paper first discusses a priori reasons for using financial spreads as indicators of changes in both real and nominal aggregates. Then Section II presents the empirical results for the United Kingdom and Germany, and Section III concludes.

I. Financial Spreads

A spread is the difference between secondary market yields on financial assets. Generally speaking, spreads exist between the returns on financial assets because the assets are imperfect substitutes for each other. These spreads depend, in turn, on differences in liquidity, maturity, and risk of the different assets, modified by taxes and any portfolio regulations.

In the research reported in this paper, a wide set of spreads is used, in order to provide extensive coverage and to extend applications developed for other countries—most particularly the United States—to the countries we study. The set comprises the long-term credit quality spread, the yield spread, and two reverse yield spreads (for details, see Davis and Henry (1994)).

The long-term credit quality spread is the difference between yields on private and government bonds of the same maturity. Changes in this spread signify increases in market expectation of defaults, which may itself be correlated with downturns in economic activity. Also, this spread may widen when monetary policy is tightened, if firms shift their credit demands to the bond market.

The yield curve is the differential between long and short rates. The interpretation advanced here is that a declining yield curve signals a future slowdown in economic activity, because when short rates are relatively high this indicates restrictive monetary policy. Also, the yield curve will tend to invert if expected inflation and activity fall.

Reverse yield gaps reflect the difference in yield between bonds and equities. We use two here: the spread between the long-term bond yield and the dividend yield, and that between bond yields and the earnings yield. The mechanisms implicit in these spreads may operate partly through agency costs of lending, which Bernanke and Gertler (1989) suggest are related to firms’ net worth. Thus, if declining equity prices following a monetary tightening reduce net worth, then this in turn may make it more difficult for firms to obtain credit, because of increased moral hazard and adverse selection in lending to firms with low net worth. (There are also likely to be effects from the anticipation of lower earnings or dividend growth in a recession.) The empirical results are preceded by a brief account of the methods used.

II. Empirical Results

This section reports the results of tests on the information contained in financial spreads that account for the behavior of output and inflation. The countries studied are the United Kingdom and Germany. The same methodology is used for each country exercise to test whether spreads have significant explanatory power in equations that include other macroeconomic variables useful for predicting output and inflation. The procedure followed first estimates models by first differencing non-stationary variables. Tests on spreads then use a ten-variable VAR that is estimated using all variables in I(0) form. A second exercise provides examples that include co-integration relationships between the I(1) variables, and use a Vector Error Correction Model (VECM) (for a discussion see Davis, Henry, and Pesaran (1994)).

The data for these exercises are quarterly and are seasonally adjusted. The full sample for Germany is 1974 Q2 to 1992 Q2, and refers to West Germany only, and for the United Kingdom, 1968 Ql to 1991 Ql. The variables are: real GDP; the GDP deflator (PGDP); the real exchange rate (RXR); the current balance (BAL) (normalized on nominal GDP); the PSBR (for the United Kingdom, normalized on nominal GDP); the public sector deficit (PSD) (for Germany, normalized on nominal GDP); and interest rates (R)—the three-month interbank rate for the United Kingdom and the one-month euro-DM rate for Germany. In addition, four financial spreads are used for the United Kingdom: (1) the long-term credit quality spread—the yield on corporate bonds in the secondary market less the 20-year government bond yield (CQS); (2) the term structure or yield curve—the government bond yield in the secondary market less a short rate (this is taken to be the three-month interbank rate) (YC); (3) the reverse yield gap (earnings)—the yield on government bonds less the earnings yield on equities as measured by the FT-500 index (RYGE); and (4) the reverse yield gap (dividends)—the yield on government bonds less the dividend yield on equities (RYGD). For Germany we use three financial spreads: (1) the long-term bank bond spread—the yield on bank bonds in the secondary market less government bonds (BBS); (2) the yield curve differential (YC); and (3) the reverse yield gap (dividends) (RYG). Given its key role as an indicator of German monetary policy, we also use M3 in the German model. The empirical results are described next, starting with the United Kingdom.

Table 1.

Summary of Significant Spread Variables in VAR Equations for the United Kingdom, Estimated by OLS

(Sample 1969 Q2–1990 Q4)

article image

Signifies at least one lagged value significant at 10 percent.

Signifies at least one lagged value significant at 5 percent.

United Kingdom

The model variables, with the exception of the price deflator, were found to be I(1). The Dickey-Fuller tests for the price deflator were not clear-cut, with results being borderline I(1) or I(2). If the price deflator is treated as I(1), the VAR model can be estimated in first differences, and we make this assumption first. We later consider the effects of treating the variable as I(2).

In estimating the VAR model, tests suggest that a lag of four quarters is sufficient. The resulting OLS version of the model is summarized in Table 1. It shows spreads were significant in most equations. Standard statistical tests for autocorrelation, normality, heteroscedasticity, and block exogeneity of spreads were all highly satisfactory.

The final step is quantifying the effect of shocks to the innovations in the variables using the Sims’ triangular ordering. To save space, only the response of real output growth to shocks to the innovations to the spreads is commented on here. (For a more detailed discussion, see Davis and Henry (1994).) Shocks to ΔYC account for about 1 percent of the variance in output, while a shock to ACQS accounts for about 2.5 percent of the variance. The two reverse yield gaps have relatively larger effects on output variance, with ΔRYGE averaging 8 percent and ΔRYGD, 4 percent. Hence, it appears that the spreads contribute in a small but significant way to the explanation of the movements in output.

We next test for co-integrating relations between the variables, using a subset of the variables: the level of output (GDP), the price level (PGDP), the real exchange rate (RXR), the credit quality spread (CQS), and the yield spread (YC).

As the integration results for the price deflator were ambiguous and imply it could also be I(2), we report on a model in which PGDP is assumed to be I(2), using three I(1) variables: inflation (DLPGDP), output, and the real exchange rate, plus the two spreads (which are, or are rendered, I(0)). The LR test of the number of co-integrating vectors suggests that there are probably two distinct vectors. These vectors are used in the dynamic equations reported in Table 2; the first vector is RES1, and the second, RES2.

Table 2.

Vector Error Correction Model (United Kingdom)

(Sample 1968 Q3–1994 Q1)

article image

These results are encouraging. At least one vector is highly significant in each of the dynamic equations, and there are significant effects from the yield spread in two of these. The conclusion is that spreads have explanatory power when co-integration relationships between the variables are taken into account.

Germany

Since the methods used in this section closely follow those used for the United Kingdom, this discussion can be quite brief.

Tests of the levels of integration of all the model variables show that the bulk of the variables are I(1), with the exception of the price deflator, which could be I(2), and the public sector deficit, the yield curve differential, and the bank bond spread, all of which are stationary (I(0)). We accordingly estimate a VAR using stationary variables, and the spreads themselves are significant in virtually all the equations (Table 3). The equations—once spreads are included—are also successful overall; as in the United Kingdom, standard statistical tests were highly satisfactory, including in-sample prediction.

Table 3.

Summary of Significant Spread Variables in VAR Equations for Germany Estimated by OLS

(Sample 1974 Q2-1992 Q2)

article image

Signifies at least one lagged value significant at 10 percent.

Signifies at least one lagged value significant at 5 percent.

The final step evaluates the quantitative effects of these variables using the Sims’ triangular ordering. These show a strong effect on the variance of both output and inflation of an innovation to the bank bond spread. (Eleven percent of the output variance and nine percent of the inflation variance is accounted for by this spread.) The other spreads have smaller effects, but nonetheless are comparable to money in their effect on prices, and have a greater effect on output than does money.

A VECM was estimated using a subset of the variables used in the previous section and assuming PGDP to be I(2). The co-integration tests use a set of four I(1) variables: output, inflation (DLPGDP), the short rate, and the real exchange rate, together with the two spreads, which are taken to be I(0).

The Johansen procedure indicates the presence of a single co-integrating vector, which is included in the dynamic equations for the four I(1) variables reported in Table 4 (denoted RES(-l)).

Table 4.

Vector Error Correction Model (Germany)

(Sample 1974 Q1-1990 Q4)

article image

These results, albeit highly tentative, are again encouraging. They offer evidence that spreads have explanatory power even when co-integrating relationships between other variables are taken into account.

III. Forecasting the Recession

The final exercise we report returns to the issue raised in the introduction: the usefulness of these indicator models in predicting turning points. Both economies have recently moved into recession—the United Kingdom in 1990 and Germany in 1992—and this last test considers how well the models cope with these downturns. Full ex ante predictions are made over the early part of the recessionary phase using the VAR model estimated on data up to the start of the recession. We then compare the predictions made by the model that does not use financial spreads with the model that includes them.

In the U.K. case, the model predicts output growth to be very sluggish in the first half of 1990, then negative, and there is very slight positive growth predicted in 1991 Q2 only before growth falls again. If this model had been used in 1989 to make forecasts, it would have given a clear message about an impending recession. In contrast, the model without spreads does not predict a recession (see Davis and Henry (1994)).

A similar exercise is then done using the German model for the period 1991 Q2 to 1992 Q2. Three forecasts were made: using the basic six-equation VAR; using the model including the money supply only as a financial indicator; and using the full ten-equation model with money and spreads. The main results show that the spreads model outperforms the more restricted models. It captures the rise in GDP growth year on year followed by a sharp fall, whereas the others predict that growth remains rapid in mid-1992.

The conclusions drawn from such a short exercise must, of course, be tentative. But in this case there does seem to be evidence that the use of financial spreads in the VAR models improves their ability to track downturns.

IV. Conclusions

This paper has considered the empirical evidence that financial spreads have information accounting, in part, for changes in output and inflation. We have taken two approaches to this assessment for each country: one using a stationary model, which extends much similar work on this topic in the United States, and an alternative, which embodies co-integrating relations. There is, we believe, considerable evidence from both of the approaches illustrated here that financial spreads contribute information in joint models of output and other macro variables including inflation.

REFERENCES

  • Bernanke, Ben, and Mark Gertler, “Agency Costs, Net Worth, and Business Fluctuations,” American Economic Review, Vol. 79 (March 1989), pp. 1431.

    • Search Google Scholar
    • Export Citation
  • Davis, E.P., and S.G.B. Henry, “The Use of Financial Spreads as Indicator Variables: Evidence for the U. K. and Germany,” IMF Working Paper 94/31 (Washington: International Monetary Fund, 1994).

    • Search Google Scholar
    • Export Citation
  • Davis, E.P., S.G.B. Henry, and B. Pesaran, “The Role of Financial Spreads; Empirical Analysis of Spreads and Real Activity” (Manchester: forthcoming in The Manchester School, 1994).

    • Search Google Scholar
    • Export Citation
  • Sims, Christopher A., “Macroeconomics and Reality,” Econometrica, Vol. 48 (January 1980), pp. 148.

  • Stock, James H., and Mark W. Watson, “New Indexes of Coincident and Leading Economic Indicators,” in NBER Macroeconomics Annual 1989, ed. by Olivier Jean Blanchard and Stanley Fischer (Cambridge, Massachusetts: MIT Press, 1989), pp. 35194.

    • Search Google Scholar
    • Export Citation
*

E.P. Davis is an economist at the European Monetary Institute. S.G.B. Henry is an Advisor in the IMF’s European I Department.