Corporate credit, stock price inflation and economic fluctuations

Abstract

This study analyses the empirical interaction between real corporate credit, real income, real stock prices, the short-term interest rate and inflation for the Netherlands and the USA. The framework is based on a five-variable structural vector error correction model which identifies the permanent and temporary shocks within the system. Erratic shocks in the real amount of corporate credit and in stock prices could potentially have some impact on inflation in the case of the USA and on real output in the Netherlands. However, the structural VAR analysis also shows that the above-mentioned erratic shocks only explain a small proportion of the variation in inflation and economic activity, and inflation objective shifts and supply side shocks are much more important determinants for economic fluctuations.

Acknowledgements

This paper was written while the author was a Post-doctoral Fellow within the Research Department of the Dutch Central Bank and he would like to thank this institution for its hospitality and support. The paper has benefited from discussions with Peter van Els, Maarten van Rooij, Peter Vlaar, and seminar participants at the Dutch Central Bank. Research assistance by Claudia Kerkhoff is gratefully acknowledged. The views expressed in this paper are those of the author and do not necessarily reflect the views of the Dutch Central Bank, Monetary Policy Committee members or the Bank of England.

Notes

Unreported unit root tests confirm this viewpoint.

For example, an improvement in the access of firms to funds on the stock market decreases the dependency of firms on banks to finance their investment projects. Also, liberalizations makes it possible for firms to borrow funds from other firms instead of banks. Developments like the aforementioned structurally decreases the demand for corporate credit.

Dutch corporate credit data were retrieved from the database which corresponds with the MORKMON macroeconometric model for the Netherlands as used at the Dutch Central Bank, and these data end in 1997. For the USA the corporate credit data were supplied by the BIS. All the remaining data series came from the IMF's International Financial Statistics.

Asymptotically, likelihood ratio tests for simultaneous restrictions upon the cointegrating vectors are distributed as a χ² random variable (see Johansen, 1996, Chapter 13). Due to the bad finite sample performance of χ² tests under a true null hypothesis a finite sample correction is used for the likelihood ratio test based on the average number of degrees-of-freedom under the alternative hypothesis instead of the number of observations (see Sims, 1980) and the null is rejected when the p-value of this corrected likelihood ratio test is smaller than 5%.

Vlaar (1998) also provides a correction for the stochastic nature of the long-run restrictions upon C based on the partial dervatives of C with respect to the elements of αβ ^′, and this correction can be used to compute the distribution of the parameters in C.

Such a distinction between ‘quantity’ and ‘price’ effects also avoids in this case the occurrence of the so-called ‘price puzzle’ (see Sims, 1992): inflation significantly increases in response to a monetary policy contraction.

All these identifying restrictions yields an overidentified system for each of the countries. This overidentification, and therefore also the ‘classical dichotomy’, is accepted for both countries as the χ²(1) p-value of the likelihood ratio statistic which tests this overidentification equals 6% for the Netherlands and 41% for the USA.