Optimal Stabilization Policies and the Government Budget Constraint

Abstract

This paper examines the implications of the government financing requirement for policies designed to stabilize fluctuations in national income. Because of the government budget constraint (which has not been considered in the control setting hitherto), the control instruments (money, bonds and government expenditure in our case) have to be used in combination. Thus, for example, instead of using government expenditure alone as a control variable, it is necessary to specify how this expenditure will be financed-whether through money creation, bond issues or additional tax revenues. Furthermore, in a stochastic framework with income-linked tax revenues, the budget surplus/deficit is itself random. Accordingly, if two instruments are to be used for stabilization purposes, it is not possible to regard both instruments as control variables: one instrument may be used as the control variable, while the other is used as a residual instrument to balance the budget. (Mixed policies may be devised, in which the control and residual instruments are orthogonal linear combinations of the basic instruments. Such policies may be regarded as linear combinations of the simple policies, for which the control and residual are themselves basic instruments.) As we note in this paper, the efficiency of stabilization policy is not affected by the choice of control variable, but rather by the choice of residual instrument. For simple policies, we find that the use of high-powered money as the residual instrument is certainly superior to the use of government expenditure, and is also superior to the use of bonds except in the perverse case where open-market bond sales are expansionary. The same conclusion holds for mixed policies (considered in the Appendix), when attention is confined to convex combinations of basic instruments. For non-convex combinations, perfect stabilization may be achieved, although a consideration of adjustment costs, neglected in this paper, would clearly modify this result. In addition, we find that, for some of the simple stabilization policies considered, the instruments follow non-stationary processes, despite the fact that national income is itself stationary. This can, under conditions derived later, be true of four of the six simple policies considered here. For the remaining two policies-the use of government expenditure as the control with high-powered money as the residual, and the use of high-powered money as the control with government expenditure as the residual-it is shown that the instruments always follow stationary processes. Thus, except in the case where open-market bond sales are expansionary, the optimal stabilization policy that also generates stationary series for the instruments is the use of government expenditure as the control with high-powered money as the residual. In the exceptional case where bond finance as the residual instrument is optimal from the' point of view of stabilization, we show that its use in conjunction with high-powered money as the control may not generate stationary instruments.