Kalman filter approach to estimate the demand for international reserves.

摘要

The reserve demand literature demonstrates that the demand for international reserves by central banks is a function of the propensity to import (Heller, 1966), money supply (Johnson, 1965), opportunity cost (Heller, 1966), the level (Triffin, 1961) and variability (Machlup, 1966) of trade, gold price (Bahmani, 1984, 1985a, 1987a), and exchange rates (Bahmani and Malaxi, 1987, 1988; Bahmani and Niroomand, 1988). Except for the opportunity cost measure, all of these right-hand side variables have been shown to be statistically significant by a large number of empiricists for both LDCs and DCs. Speed of adjustment disequilibrium models are successfully estimated by all researchers who have employed them. There has been only sparse success in the literature empirically identifying a negative relation between reserve demand and an opportunity cost measure. Grimes (1993) believes that the inability of researchers to broadly isolate an opportunity cost measure implies that reserve demand is stable through time. Presumably because central banks aren't income maximizers.; Economies of scale (Olivera, 1969) in reserve holding have been shown to exist by Makin (1974), Frenkel (1980), Bahmani and Malaxi (1987), and Bahmani (1987, 1988). Heller and Khan (1978), Bahmani (1985a, 1987a, 1988a), and Landell-Mills (1989) all demonstrate that economies ensue a Keynesian adjustment path.; Crockett (1978) has said the issue of stability is key. The demand for international reserves has been shown to be unstable by Heller and Khan (1978), Frenkel (1978, 1980, 1983), Bahmani (1988b) and others. Frenkel (1980a) has called for an approach that allows the parameter vector beta, in the international reserve demand equation, to vary over time.; In summary, this thesis answers Crockett's (1978) key question, as to whether international reserve holding behavior by central banks was unstable, in the affirmative. In addition, Frenkel's (1980a) call for an approach that allows the Beta vector in our reserve demand equation to vary over time has been answered. We have responded to this call by successfully estimating the Kalman Filter Random Walk Betas. Our modified CUSUM test for structural change indicates that all nineteen countries, do indeed, exhibit extreme instability in their reserve holding behavior.; We present to the literature, the results of the Kalman Filter two-step procedure to estimate PHI so that a direct formal nested hypothesis test on the behavior of PHI can be conducted. Moreover, with this formal test, we have shown overwhelmingly, that the Beta vector of the demand for international reserves, by nineteen central banks, is time-varying (RTN or RW) without exception. Finally, with few exceptions, the Beta vector in the demand for international reserves equation, follows a random walk.