| In this dissertation, we construct a broader concept of US Divisia monetary aggregates including bond mutual funds. Both simple-sum aggregation and "CCAPM special case model" from Barnett et al. (1997) are used in constructing new monetary aggregates.; The demand for this work is found in two literatures: Missing M2 and Divisia monetary aggregate. The empirical failure of the official M2 in the 1990s and the enhanced liquidity of risky assets from the development of financial markets support the broadening of the definition of M2. Duca (1995) has shown that the inflow to bond mutual fund from time deposit played a major role in the missing M2 episode. The Divisia monetary aggregate was proposed by Barnett (1980) as an alternative method to the currently used official monetary aggregate. Instead of simply summing up all different monetary components, it captures their different liquidities and aggregates them over the assets.; After constructing newly defined monetary aggregates, we compare their empirical performance from two empirical tests: the money demand model based on cointegration and the forecast model on GDP and inflation. From the money demand model, we find that the M2 demand system is subjected to the inconsistent elasticity changes since 1990s. When we estimate the U.S. money demand function with monetary aggregates with bond mutual funds, this broken relation of money with other economic activities is restored. From the forecast model on GDP and inflation, we find that the monetary aggregates with bond mutual funds improve the forecasting power of the model, especially, for the post-missing M2 period.; Therefore, we strongly suggest that the monetary aggregates with bond mutual funds, both in simple-sum and Divisia index, should be closely monitored as a possible alternative measure of money. |
