Testing the global extent of the endogenous-money hypothesis: a panel vector autoregression approach

This paper examines the endogenous-money hypothesis using a panel-data set of 144 countries over the period 2001–2017. Its empirical analysis is conducted in a panel vector autoregressive framework, a hybrid econometric methodology that offers the advantage of jointly accounting for endogeneity issues (as in traditional vector autoregressive modeling) and individual/country-level heterogeneity associated with a panel-data structure. A panel version of the Granger non-causality test and an examination of orthogonalized impulse-response functions and forecast error variance decompositions are applied to test the causal ordering among loans, the money base, broad money, output, and prices. The empirical findings support causation running from loans to broad money and from broad money to the monetary base. A causal link running from loans, output, and broad money to total reserves is strongly supported and, in all cases, the causality is unidirectional from these variables to the money base. Beyond that, a complex interaction among prices, production, credit, and money is found, suggesting that the best approach to understanding the endogeneity/exogeneity issue is to rely on the hypothesis of endogenous money as reflected in the ‘liquidity preference’ interpretation.