4  Assessment of policy options

This section applies the framework developed above to conduct a qualitative assessment of stylised policy options available for Crown financial policy. Eight policies are identified and assigned either to the low, medium or high risk category (Appendix II provides the detailed qualitative analysis in support of these assignments). From each risk category, one policy option is chosen as warranting detailed empirical analysis.

The section begins with an outline showing how the framework is applied to Crown financial policy issues. This is followed by a discussion of key judgements made and then an analysis of each risk category.

4.1  Multiple objectives

The framework developed in Section 3.2 considered the case where the policy choice is defined by a single variable. The analysis assumed that the policy maker chooses whether to place high (“H”) or low (“L”) weight on the proposed policy in face of uncertainty about whether the impact on economic welfare would be significant (“s”) or insignificant (“i”). The application to Crown financial policy involves extending the framework to a multi-variate analysis. This reflects that the Crown balance sheet may impact on economic welfare through three main channels:

• Deadweight losses caused by the average level and variability of the tax rate;
• Higher debt servicing costs and reduced investment due to greater uncertainty (and losses caused by confidence crises) that could result if monetary or fiscal policy became time-inconsistent; and
• Inefficient forms of government expenditure arising as a result of the principal-agent relationship between government and citizens.

Both the absolute and relative magnitudes of these effects are subject to considerable uncertainty. In the following the true state or “true model” of the economy is described by the vector (x, x, x), where each “x” may be significant (s) or insignificant (i). For example, the state vector (s, i, i) would indicate that tax smoothing is economically significant but losses due to time-inconsistency and agency cost are insignificant.

Similarly, policy is described by the vector (X, X, X) where each “X” may be high (H) or low (L) weight. For example, the policy option (H, L, L) would indicate high weight on policy targets aimed at tax smoothing and low weights on policy targets to mitigate time-inconsistency and agency cost. This structure gives rise to eight (= 2³) possible policy options, as listed in Table 2.

Figure 4 illustrates that for each policy there are eight possible outcomes that depend on the true model of the economy. In the case where the policy vector matches the state vector the loss is zero. In all other cases the losses are positive.

Figure 4: Policy options and losses

 

4.2  Key judgements

A key judgement made in this paper is that existing institutional arrangements do not fully mitigate the risks of time-inconsistency and agency cost in all relevant circumstances. Although the risk of time-inconsistency appears low in the context of the Reserve Bank Act 1989 and Fiscal Responsibility Act 1994 and currently moderate debt levels, the level of risk could increase significantly if the Crown followed a highly leveraged strategy in pursuit of tax smoothing. Empirical evidence on tax smoothing for both NZ and the US imply debt levels in excess of 2500% of GDP (Bohn 1990 and Davis and Fabling 2002). It may be unwise to assume the current legislative arrangements would be robust to a highly leveraged strategy.[26]

In terms of agency cost, current institutional arrangements such as the New Zealand Superannuation Act 2001 are likely to help protect against “direct raiding” of funds by a future government. However, these arrangements do not preclude “indirect raiding” or “expenditure creep”, whereby the pressure for inefficient government expenditure would increase as Crown net worth increases.

Related to these issues, it is assumed that a high debt target would be an effective instrument in reducing agency cost. The proposition has been adopted as an application of Jensen’s (1986) free cash flow theory of corporate finance.

However, the inter-temporal budget constraint facing the Crown is usually thought of as less stringent than for a corporation. Except for the recent period in New Zealand, it is not clear that high levels of debt has been an effective constraint on government spending in many countries.^[27]

A further key judgement is that the potential loss from a false positive error on distortionary taxation is significantly larger than for a false positive error on time-inconsistency and agency cost. The difference is that placing high weights on the time-consistency and agency cost issues serves to constrain government policy action while a high weight on distortionary taxation would motivate an “activist” policy. Actively seeking to smooth taxes carries the significant risk of implementation failure due to highly uncertain correlations between asset returns. A further risk, due to the taxation of capital, is that tax smoothing could lead citizens to take on excessive risk in their portfolios (Coleman, 1997). Thus, incorrectly placing high weight on tax smoothing (a false positive error) would carry the risk of large losses in the absence of other constraints.

The judgement is made that a high level of debt would attract a risk premium for default risk. It is assumed that the risk premium would be “unjustified” in the sense that the country never intends to default or deflate the real value of the debt. The result of this judgement is that both distortionary taxation and time-consistency objectives motivate an upper bound on debt levels. Therefore, the two objectives tend to be substitutes for each other when evaluating the losses associated with debt targets.

A further working assumption made throughout this section (and Appendix II) is that the debt target implied by the agency cost objective conflicts with the debt target implied by the time-consistency and distortionary taxation objectives. This need not be the case as the “high” debt target motivated by agency cost considerations could be lower than the “low” debt target motivated by the other two objectives.