Government Debt, Government Spending, and Private Sector Behavior: Reply
回应Kormendi（1983）关于政府支出而非税收影响消费的观点，指出其方法缺陷，并基于生命周期假说重新推导消费函数，实证检验后发现李嘉图等价命题缺乏支持。
Roger Kormendi (1983) presents apparently strong evidence that, in contrast to the standard view, consumption is not reduced by taxes but is reduced by government expenditure. He interprets his results as supporting a consolidated approach to private sector behavior in which consumers effectively internalize the government budget constraint. Specifically, he claims that consumers regard government spending as the true measure of the government's claim on private resources, and so do not respond to changes in taxes, given spending. This is basically the approach advocated by Robert Barro (1974) and known also as the Ricardian Equivalence Proposition (hereafter REP). Kormendi's results appears to contradict other empirical work based on the Life Cycle Hypothesis (for example, Martin Feldstein, 1982; Modigliani 1984a; Sterling, 1985) although results similar to his have been reported (see David Aschauer, 1985; John Seater and Roberto Mariano, 1985 and the references in Kormendi). In our view, Kormendi's analysis is seriously flawed. His heuristic derivation of the consumption function leads him to a specification of the aggregate consumption function, which is not consistent with the Life Cycle Hypothesis (LCH) or with REP, and to questionable methods of estimation. Once his conceptual and methodological errors are corrected, his formulation and, more generally, the REP hypothesis are found to receive little empirical support. In the next section we rely on the LCH to derive an aggregate consumption function which shows explicitly how government expenditure and taxes should effect private consumption. This derivation helps to bring out the observable implications of REP, which are shown to be equivalent to a limiting form of the LCH in which the planning horizon is infinite. It also serves to clarify the appropriate specification of the variables appearing in the consumption function. Next, Section II reports our empirical estimates and tests. Section III compares our results with Kormendi's. Finally, Section IV reports the results of endeavors to improve the specification of fiscal variables, notably by distinguishing between permanent and transitory tax changes.