The major strength of the paper is its approach to the evaluation of fiscal policies on saving, and for this reason my comments have centered almost exclusively on the parts where this approach has not been consistently applied. In particular, there are two aspects of the approach to investigating the determinants of the national saving rate that lead me to advocate that future work build upon the framework suggested here. The first is that the estimated saving rate functions attempt to account for the interrelationships between the saving decisions of the household, corporate, and government sectors. The second is that the estimated saving functions allow for the different impacts of anticipated and unanticipated fiscal changes.
However, any future empirical evaluations of the effects of fiscal policy on saving using the approach of this paper would be considerably strengthened if two modifications were made. First, I advocate switching the focus from the reaction function for the government saving rate to separate reaction functions for government expenditures and taxes, perhaps with some type of constraint on the surplus imposed through the loss function. Such a modification of the approach would overcome what I identified as a potential pitfall, since now tax and spending changes could be more accurately decomposed into their anticipated and unanticipated components.
Second, I advocate more explicit modeling of the behavior of the household and corporate sectors to take account of the effects of the an ticipated as well as the unanticipated components of government actions. In particular, the role of anticipated fiscal policy actions in the formation of expectations concerning future income and interest rates should be made more explicit. Further, government expenditures could be included in the households' utility function with current taxes (and future tax liabilities to the extent that they are discounted) included in the budget constraint. Or household ownership of firms could be explicitly included in the household budget constraint so that the personal and the corporate saving decisions were simultaneously determined. In addition to decreasing the ad hoc nature of the current specification, more explicit modeling might also allow the evaluation of the major institutional changes in the application of fiscal policy that result from decisions to modify the form of the government's reaction functions. The current specifications do not have this capability.