7.2  Tax policy (continued)

7.2.3  Income tax incentives – can they increase national saving?

Under current tax law, returns to savings are generally taxed under the personal income tax rate structure (10.5%, 17.5%, 30% and 33%). The application of income tax reduces returns to savings by the tax rate applied.

Economic theory alone is unable to determine whether or not a cut in taxes on savings would increase household saving. This is due to offsetting “substitution” and “income” effects. While increasing the return from saving will increase the reward from saving which by itself will promote saving (the “substitution” effect), it will increase disposable incomes for those with positive amounts of savings and allow them to reach retirement savings goal with less saving (the “income” effect). Thus, the substitution and income effects will work in opposite directions.

In addition, when analysing the effect that reduced income tax rates can have on national saving, household saving is only one side of the story. Government saving is the other side. A reduction in income taxes on savings will generally decrease tax collections. Therefore, for a reduction in income tax on savings to increase national saving, any additional household saving needs to be greater than the revenue cost to the government. For example, if the reduction in income tax is matched by an increase in other taxes (e.g., GST), then there is no income effect and there is thus likely to be an increase in national saving.

The literature in this area falls into at least two categories. The first concerns tax incentives for particular forms of retirement savings and their effect on aggregate savings. The second concerns cuts in the tax rates for capital more generally and their effect on national saving.

Tax incentives for retirement savings

There is a large but somewhat inconclusive literature on the extent to which tax incentives for particular forms of retirement savings increase aggregate savings. As well as the difficulties untangling the “substitution” and “income” effects described above, there are also difficulties determining the extent to which tax incentives motivate people to reallocate existing savings to take advantage of the tax incentive and to what extent they create new savings. Globally, there are a number of data-related reasons why conclusions are difficult to draw, including:

1. Complete and reliable data describing saving and consumption behaviour are available only at a few points in time and good quality time-series data are very difficult to obtain.
2. Aggregate analysis may be biased by individual characteristics affecting the decision-making process, such as age, gender and marital status.
3. Investigations of household behaviour are based on very short data periods that may not be sufficient to provide a clear picture of the effects of incentives.
4. Cross-country comparisons are often inconclusive - largely due to the problems described above.
5. Many researchers' interpretations of the data depend on assumptions of perfect information and rationality, which are unrealistic in practice.

While these difficulties exist, two tentative conclusions can be drawn from the available literature (Antolin and Ponton 2007). The first is that tax incentives increase retirement saving mostly by reallocating existing savings. Secondly, while there is some evidence that tax incentives for retirement savings may produce a small amount of new saving, the increase is much lower than supporters of tax incentives often advocate. However, tax incentives may reduce the relative tax advantage of other classes of investments (such as owner-occupied housing, or investments made for capital gain) and thus improve the overall allocation of savings.

Reducing tax on capital more broadly

Another way of analysing the issue is to consider the effect that reducing tax on capital more broadly may have on national saving. We include here tax on all investment returns – interest, dividends, rent and capital gains. The SWG received advice from officials based upon two surveys (Attanasio and Wakefield 2010, Bernheim 2002) of the recent literature using life-cycle models. Officials from Inland Revenue performed simulations using a life-cycle model based upon those surveys – calibrated to New Zealand.

A key parameter in determining the responsiveness of savings to changes in after-tax interest rates is the “intertemporal elasticity of substitution.” This measures the willingness of people to change their spending from one period to another. Bernheim argues that this lies somewhere between 0 and 1 and Attanasio and Wakefield argued that this lies somewhere between 0.4 and 1 (the higher the number the more people increase saving in response to changes in the after-tax interest rate).

Attanasio and Wakefield choose a figure of 1 for their base case which keeps the analysis as simple as possible but seems biased towards assuming the most optimistic responsiveness of saving. They found that retirement wealth increased, but concluded that there was little effect on national saving. In their study, over 90% of the increase in private wealth was offset by reduced tax revenue.

Officials chose 0.6 when calibrating the life-cycle model for New Zealand. This is in the middle of the feasible range cited by Bernheim and Attanasio and Wakefield. They also took account of a number of key features of New Zealand's tax system which will lower the responsiveness of savings to changes in capital taxes relative to what has been modelled elsewhere, because households can invest in assets which have low or zero capital income tax. For instance, most savers will acquire an owner-occupied house over their lifetimes. Every dollar spent repaying their mortgage effectively earns the pre-tax interest rate and thus is not taxed. This is in contrast to the United States where mortgage interest is deductible from taxable income. Under New Zealand assumptions, the increase in private saving from cutting capital income taxation was only about 85% of the foregone revenue cost to the government. That means the long-run stock of national savings would fall.

These numbers should, of course, be used with caution. The model is highly stylised and does not attempt to take into account all the determinants of savings levels.

For anything other than a revenue-neutral shift away from income tax, there is a high degree of uncertainty over whether or not cutting the tax rate on private saving alone will increase or decrease national saving. Thus, decisions on whether or not to cut taxes on capital income should not be made primarily in terms of their effects on the quantity of national savings.

Why reducing income tax on savings is still important

Irrespective of the effect on national saving, there is a reasonable concern that high rates of tax on the returns to saving have undesirable economic effects. High tax rates on savings can be particularly distorting because of the bias they create towards consuming earlier rather than later in life. They also distort investment choices towards assets that generate income subject to low effective tax rates such as rental property. Moreover, the size of the inefficiency – and therefore its economic cost – rises with the length of time between consumption now and later and hence affects the incentives for sustained wealth accumulation. Therefore, reductions to the income tax rate on capital can be justified on economic efficiency grounds – particularly if they can be achieved in a fiscally responsible manner.

7.2.4  Increasing the level of national saving by changing the structure of the tax system

Continuation of the switch towards consumption tax and away from income tax

Because GST is less distorting than income tax on the saving decision, the SWG supports a continuation of the switch from income tax to GST (as occurred in the 2010 Budget), and recommends that consideration be given to an increase in GST from 15% to 17.5% (together with compensating income tax and benefit changes for lower income earners). For most people with the means to save, this switch will keep their real income roughly constant. Hence they will not experience an “income effect” that would otherwise act against their pro-saving “substitution” effect. Such switching should continue, but not go so far as to:

• Compromise the current broad-base, one-rate integrity of GST (by, for example, the introduction of a lower rate for food).
• Demonstrably push too much further activity into the so-called black economy.
• Exceed the tolerance of the highly mobile element of New Zealand’s labour force.

The SWG notes that any switch from income tax to GST does effectively reduce the purchasing power of existing savings. While this is inevitable and unfortunate (as several submissions said), most people with current savings will benefit from the lower rates of income tax on all their income (including returns on savings). This is true for people over 65, who pay relatively high average taxes on their capital income as they are wealthier on average than working age people and hold a disproportionate amount of bank loans.[18]

If the government raises GST, it will be able to either reduce its fiscal deficit (and therefore its debt) and/or further reduce the marginal rates of tax on income (from both labour and capital/investment). Either change should be positive in terms of New Zealand's level of national savings.

To the extent that the government needs tax revenue above what can be obtained through GST - including for specific (savings-issue-related) purposes - the SWG recommends that consideration be given to (savings-neutral) income taxes levied only on income from labour.

Many countries effectively achieve a full switch from income tax to consumption tax (EET) for certain retirement savings (in qualifying vehicles akin to KiwiSaver) by exempting contributions from income tax, and exempting the investment income earned while the funds are in the vehicle, but taxing the money when withdrawn in retirement. Another variation is to apply a lower but not zero rate of tax on investment income, i.e., TtE. the SWG notes that these treatments would not necessarily increase the level of national savings (although it would undoubtedly increase the level of savings undertaken through these vehicles).