4  Practical limitations of tax smoothing

Having established that the efficiency gains from tax smoothing are potentially significant, we must consider practical obstacles that might prevent its implementation or substantially erode the expected efficiency gains.

One argument employed by critics of tax smoothing is that, given the high frequency of small tax changes, it would be impossible to implement in practice. For example, the lead-time required to change income tax rates extends to some months. There are also administrative costs associated with implementing a change to the tax regime and the additional economic uncertainty to consider. While these are valid criticisms, we have shown that the need to change tax rates frequently is not unique to tax smoothing. Almost by definition, a balanced budget strategy will require frequent changes to tax rates. These considerations suggest placing a limit on the frequency of tax changes.

We implement this approach using a frequency of three years.[20] This has the effect of converting the tax rate into a step-wise function. The results are as follows:

Source: Authors’ estimates

The savings are basically the same. As expected, the volatility of tax rate changes has increased (due to the lower frequency of changes). However the increased volatility of three-year tax smoothing (over balanced budget) is proportionately lower, since this financing strategy benefits from smoothing short-run shocks to GDP.

Source: Authors’ estimates

Another consideration is that the informational requirements for tax smoothing are huge. There may be an option value in waiting due to the cost of getting it wrong (that is, maintaining taxes now to finance an expected expenditure profile that doesn’t grow as expected comes at a high cost). Furthermore, as Jensen and Nielsen (1995) point out, “even if the government’s time horizon is long, it might be tempted to enjoy the short-term benefits from the reduced pressure on public expenditures during the ‘breathing space’ period” (page 10). We model this latter scenario, assuming that the government operates a balanced budget until the baby boomers start to retire in 2010. Table 9 suggests that, while waiting till 2010 before implementing tax smoothing requires a much greater upwards adjustment in tax rates, the deferred tax smoothing strategy still generates a welfare gain relative to balanced budget.

Another reason that has been put forward against tax smoothing is the negative relationship between tax rates and GDP growth rates. While we have modelled an effect of this sort, through the substitution effect on the labour supply, the possibility remains that we have underestimated the adverse effect of tax rates on growth. A recent OECD study on the relationship between taxation and economic performance concluded that the increase in the weighted-average tax rate of about 10 percentage points over the past 35 years has been accompanied by a 0.5 percentage point reduction in the average annual rate of economic growth (OECD, 1997).

Another point to note on the feasibility of tax smoothing is the issue of time-inconsistency. Realising the efficiency gains from tax smoothing relies on future governments’ commitment to the strategy. However, in a political environment where support for a tax smoothing strategy is divided, a change in government can lead to a costly change in strategy. This problem is likely to be particularly acute in the first few years, especially if the financial asset portfolio performs poorly.

Even after a few years, the policy will remain vulnerable if political decision-makers abandon the pre-funding strategy to pursue some other objective (for example, tax cuts). However, the welfare costs of abandonment are likely to increase through time and an electorate cognisant of that fact may constrain the options of political decision-makers. In this sense, tax smoothing might become more entrenched as time goes on, assuming it gets over the hurdle of the first few years.

There is a final practical consideration that deserves more serious consideration: the critical assumption of exogenous government expenditure. Pinfield (1998) argues that tax smoothing would be undermined “if the period of operating surpluses were to trigger higher expenditure than would otherwise be the case – ‘expenditure creep’ in other words” (page 1). Historical evidence suggests that the level of inefficient public spending and investment rises when public finances are healthy (see Bohn, 1991, and Alesina and Perotti, 1995).

Since tax smoothing involves the accumulation of financial assets, it is possible that future governments might increase expenditure simply because they feel wealthier. Similarly, a future government may alter its investment strategy to pursue welfare objectives (Davis, 1998). The extent to which this sort of behaviour will occur depends on the time-consistency of the government’s financing strategy and the strength of its governing institutions. Even with strong institutions, however, some form of expenditure creep is likely.

To demonstrate the effect of expenditure creep we introduce a simple linear rule that ties government expenditure to the level of financial assets held by the government. Endogenous government expenditure is modelled by:

(3)

is adjusted government expenditure; P_t is the operating balance excluding portfolio returns; e_t is the (deterministic) forecast growth rate in government expenditure; and δ is the responsiveness of government spending to changes in net wealth.

Source: Authors’ estimates

Under this formulation, expenditure creep represents a permanent shift in the level of government expenditure. Furthermore, expenditure creep ceases to be an issue beyond the point at which the portfolio peaks (because it is increasing in wealth but always non-negative). Somewhat arbitrarily, we set δ equal to 10%.[22] The results of modelling are presented in Table 10. The main result is that expenditure creep is sufficient to completely erode the efficiency gains of the tax smoothing strategy.