On the Estimation of Permanent Income: Note
指出Darby对永久收入假说的实证估计存在序列相关和估计不一致问题，修正后得到相反结论，但使用非线性Zellner估计法可支持Darby的假设。
Several years ago, Darby [1] presented a valuable restatement of the permanent income theory of consumption. The two most important hypotheses advanced by Darby were that the weight given current income in the autogressive permanent income generating equation should be small and that transitory income belongs as an explanatory variable in the nondurable consumption expenditure equation. Darby found empirical support for both hypotheses. In attempting to use Darby's work in a project of mine, I noticed that Darby's estimates of the current income weight, and therefore of the permanent income generating function, varied considerably with the measure of consumption chosen; of course, because the different kinds of consumption all are supposed to be driven by the same permanent income, such variation in the permanent income estimates should not occur. I also noticed that all of Darby's estimates suffered from severe serial correlation problems. As I explain below, correction of these problems produces a current income weight much larger than that found by Darby and also renders insignificant transitory income as an explanatory variable in the nondurable expenditures equation. Thus it appears that Darby's evidence, when properly handled, generally rejects Darby's two main hypotheses. Darby's hypotheses can be salvaged by using a better estimation method. Darby's grid search method turns out to be sensitive to the sample period chosen. In contrast, a nonlinear Zellner estimation of the total and nondurable consumption equations produces estimates that are consistent across both sample periods and measures of consumption and that support Darby's hypotheses.