| Monetary aggregates are typically generated using the sample sum approach, in which total holdings of monetary assets are simply added together. However, this implicitly assumes that the assets (currency, demand deposits, savings accounts, etc.) are perfect substitutes; clearly, this method is adding up "apples and oranges," since these assets are far from perfect substitutes. Thus, the simple sum procedure renders a distorted aggregate.; The purpose of this study is to create a new set of monetary aggregates for the Mexican economy which sidesteps this problem, using aggregation theory and statistical index number theory. These aggregates are then compared to Mexican simple sum aggregates using several standard tests for policy usefulness. The bases for comparison are (1) the ability of each aggregate to provide meaningful estimates of the ordinary demand for money function, (2) the forecasting properties of these demand for money functions, (3) the "controllability" of the aggregates, and (4) the aggregates' ability to explain changes in the velocity of money. Since this aggregation approach also generates measurements of the "price" of money, the demand for money and consumption goods is then simultaneously estimated using the microeconomic "system wide" approach.; The data used in this study consist of Mexican quarterly time series (1972:2-1987:4) on various monetary instruments and their corresponding rates of interest. |
