The Demand for Assets: Evidence from the Markov Switching Normalized Quadratic Model

In this paper, we provide and illustrate a solution to the inter‐related problems of estimation of asset demand functions, instability of money demand relations, and monetary aggregation. We use an econometric framework that very flexibly allows for changes in the coefficients of the asset demand functions and then quantifying the implications that this asset demand instability has for monetary policy and monetary aggregation. Our approach allows the estimation of asset demand functions in a systems context, using a flexible functional form for the aggregator function, based on the dual approach to demand system generation. However, instead of assuming that consumer preferences are fixed as in the neoclassical demand theory, we assume Markov regime switching, thus allowing for complicated nonlinear dynamics and sudden changes in the parameters of the asset demand functions and the underlying aggregator function. We use the monthly time series data on monetary asset quantities and their user costs, recently produced for the United States and maintained within the Center of Financial Stability, and the Normalized Quadratic (NQ) flexible functional form with Markov regime switching, to generate inference in terms of a full set of elasticities, simultaneously achieving consistency with the data‐generating process and economic and econometric regularity. We find evidence that our five‐regime NQ model provides a better fit of the actual data than a single regime model, that the asset demand specifications exhibit instability between regimes, but are rather stable within regimes, and that the assets are in general Morishima substitutes with the Morishima elasticities of substitution always being below unity.