Money supply endogeneity and bank stock returns

Abstract

This article presents results of tests on two related hypotheses on money supply. The first relates to an unresolved issue of money endogeneity while the second centres on the yet-explored relationship between money supply and bank stock returns if money is found to be endogenous. Our results, using long-horizon data of Group of Seven (G-7) economies, supports causality in money supply as running from bank lending to bank deposits, a result that is predicted by the post-Keynesian money supply endogeneity (bank-credit-driven) theory. Thus, the result is not consistent with exogeneity proposition. A new evidence of positive relationship between endogenous money supply and aggregate bank stock return is statistically significant on this hitherto unexplored topic. These findings are consistent with the post-Keynesian money supply theory and the dividend valuation theory, which predicts money supply changes to induce changes in bank earnings, so bank share prices change.

Keywords:

• money supply endogeneity
• bank stock returns
• credit market
• liquidity provision
• post-Keynesian theory
• dividend valuation theory
• G-7 countries

JEL Classification::

• E12
• E51
• G21

Acknowledgements

We record our sincere appreciation for the detailed comments of the Editor of the journal, Mark Taylor, and an anonymous reviewer for useful comments incorporated in our revised submission. We remain responsible for any errors.

Notes

¹ Sample period varies for each country depending on data availability. Data on economic time-series prior to 1992 for Russia and financial statistics on banks in Russia are not available. Hence, this study will focus on the remaining seven countries instead of the G-8.

² In some literature, this is known as credit-money or money derived from credit supply.

³ This is now commonly known as Gordon–Shapiro Dividend Valuation Model.

⁴ Also see Mazharul (2008).

⁵ Howells and Hussein (1998) used the same money definitions.

⁶ Unit root tests investigate if the order of integration of variables is similar: The Phillips and Perron (1988) test is preferred over the Dickey–Fuller (ADF) test since this test allows for serial correlation and heteroscedasticity. We find the variables to be I(1) stationary. There is evidence that the Phillips–Perron test has more power than the Augmented Dickey–Fuller (ADF) test (Davidson and MacKinnon, 1993). Next, we test for cointegration using a more sensitive Vector Autoregression (VAR) and Bayesian Information Criteria (BIC). The Johansen (1988) multivariate cointegration test is used in this article to test for any long-term relationship between the series.

⁷ Akaike Information Criterion (AIC) and BIC suggest VAR (1) as the most appropriate model to test for cointegration between bank loans and money supply in most of the samples: Canada 1 and 2, France, UK 1 and 2 and US 1 and 2. However, lag lengths two and five are also preferred in some cases. Due to space constraints, these results are not reported in this article but are available from the corresponding author upon request.

⁸ Huang (2003) found existence of the bank lending channel in the UK, while Romer and Romer (1990) concluded that the money view (exogenous money) exists in the US.

⁹ Before the VECM test can be conducted, a preliminary test was performed; unit root test for stationarity of bank price index (Appendix Table A4). Variables are found to be nonstationary in levels but stationary in first differences. The lag length of one is chosen by the Schwarz Bayesian Criterion (SBC) as the optimal lag length in Canada period-2, France, Germany, UK period-1, UK period-2, US period-1 and US period-2. VAR(2) is the optimal lag length preferred in Canada, Canada period-1, Japan, UK and US. The lag length of five is chosen for Italy.

¹⁰ Prior studies have shown that the bank stock returns in the US with a very competitive banking system is dissimilar to the results anywhere else (Ariff and Lamba, 2007).