Debt and budget surpluses with a tax habit and balanced budget hawks.

Author:Loukoianova, Elena


In this paper, we demonstrate the theoretical implications of a tax habit and balanced budget hawks on the (log) primary budget surplus. The tax habit causes the budget surplus to respond more smoothly to the expected present value of government purchases than in a tax smoothing model. Put differently, a fiscal authority with a tax habit smoothes the adjustment of taxes to "news". Fiscal policymakers advocating contemporaneously balanced budgets, which we refer to as "balanced budget hawks", increase the sensitivity of surpluses to current government purchases. Using UK data from 1942 to 1964, we utilize Bayesian Monte Carlo Markov Chain simulation methods to gauge the importance of habit and hawks effects.

1 Introduction

The tax smoothing model (Barro 1979), a staple of graduate macroeconomic textbooks, represents the standard reference point for discussions about the behavior of budget surpluses and debt. In Barro's model, the government's desire to minimize the distortions from taxation causes (expected) taxes to be smoothed over time. A number of researchers, including (among others) Alesina and Drazen (1991) and Alesina et al (1998), maintain that political influences cause the path of budget surpluses to deviate markedly from the predictions of the tax-smoothing model. Unfortunately, time series data on the political factors that might influence fiscal policy are often difficult to obtain; these variables are routinely absent from macroeconomic databases (for example, the Federal Reserve Bank of St Louis's FRED).

In this paper, we consider an alternative approach to modeling departures from tax smoothing suitable in the absence of detailed time series information about political influences. We demonstrate that the present value condition commonly used by macro-econometricians to characterize tax-smoothing behavior (see, for example, Ghosh 1995) can be generalized in two dimensions. First, we consider the theoretical implications of a tax habit where the policymakers' loss is dependent on the reference level of previous taxes. Second, we allow for a proportion of fiscal policymakers, such as senior officials within the treasury or the cabinet, advocating contemporaneously balanced budgets. Like a tax habit, balanced budget hawks cause fiscal policy to deviate from the predictions of the tax smoothing model. We show, however, that these two effects have distinct impacts on the time paths followed by debt and budget surpluses.

Our model exploits the close relationship between government and household behavior noted by (among others) Sargent (1987, p. 385-388). The public finance problem faced by the government is analogous to that of a permanent income hypothesis household with consumption habits (see, among others, Carroll et al 2000 and Fuhrer 2000). Balanced budget hawks are equivalent to Campbell and Mankiw's (1989) "rule of thumb" consumers.

In the empirical component of the paper, we estimate a log primary budget surplus equation using UK data from 1942 to 1964 utilizing Bayesian posterior simulation methods. Our results indicate that UK fiscal policy deviated substantially from the path consistent with the tax smoothing model and that both habit and hawks effects were economically substantive. Our findings are consistent with Cooley and Ohanian's (1997) characterization of post-World War II UK fiscal policy as a departure from tax smoothing. (They attribute this departure to the legacy of Keynes, 1940).

In the following section, we set out the government's problem. We derive the present value condition for budget surpluses with a tax habit and hawks in section 3. We also describe the time path of debt for the log-linearized model. In the subsequent section, we provide Bayesian analysis of the parameters values for our 19421964 UK sample. We discuss our results and suggest areas for subsequent work in the final section.

2 The Government's Problem

As in Barro's (1979) tax smoothing model, the government minimizes the distortion from taxation by allocating taxes over time. Barro's (1979) model assumes that tax rates are distortionary; in contrast, as in Sargent's (1987) textbook treatment, we assume that tax revenues are distortionary for analytical convenience.

The government chooses a sequence of tax revenues [{[T.sub.t]}.sup.[infinity].sub.t=1] in order to finance a stream of exogenous real government purchases, [{[G.sub.t]}.sup.[infinity].sub.t=1], which follow a stochastic process. Since debt is the only alternative to taxation as a source of revenue, this implies a path for debt, {[D.sub.t+1]}, through time. Government purchases in excess of tax revenue in any period are financed by default-free (non-contingent) bonds. (1)

The policymaker has a tax habit that reflects its distaste for changes in taxation. As emphasized in the delayed fiscal stabilization literature, political influences may make policymakers reluctant to impose sudden changes in the level of taxation. Hence, the period t loss function depends on both the distortionary impact of taxation revenue, [T.sub.t], and the influence of the government's taxation habit:

[L.sub.t] = 1/1-[sigma] [[[T.sub.t]/[Z.sup.[gamma].sub.y]].sup.1-[sigma]]. (1)

The variable [Z.sub.t] is the habit "reference level" given by (see Fuhrer, 2000, for analysis of the analogous consumption habit case):

[Z.sub.t] = [rho][Z.sub.t-1] + (1-[rho])[T.sub.t-1]. (2)

The parameter [rho] captures the memory in tax revenue habits. For [rho] equals zero, only the ratio of taxation in time t to the previous period's taxation matters--the one-period habit case. Although earlier taxation levels can enter the loss function for the parameter range 0

We write the government's one-period budget constraint as:

[D.sub.t+1] + [R.sub.t+1] [[D.sub.t] - [T.sub.t]] ( 3 )

where [R.sub.t] denotes the (time-varying) gross real interest rate. Government purchases...

To continue reading