Section 9 - Tax subsidies for saving

9.         On tax subsidies for saving[95]

 While governments can certainly influence the ways in which people save for retirement, they seemingly cannot incentivise people to save more for retirement than they want to save.  Tax-favoured Tier 2 (compulsory) or Tier 3 schemes (voluntary and occupational) may see more financial assets accumulated than in the absence of such schemes but again, savers can and do change other aspects of their behaviour.

A set of acronyms summarises the tax treatment of financial assets, particularly in a retirement saving context.  There are three main movements of money:

  • contributions: ‘T’ means that contributions to the scheme come from after-tax income; ‘E’ that contributions reduce taxable income before tax is deducted (or attract a direct subsidy); also, in the case of occupational schemes, that the employer’s contributions are not deemed part of the employee’s taxable pay.
  • investment income on the accumulation: ‘T’ means that invested assets are taxed with the saver’s other income; ‘E’ that the assets accumulate tax-free.
  • benefits received: ‘T’ means that benefits are taxed as income in the year of receipt; ‘E’ that benefits are exempt from tax in the recipient’s hands.

Most countries treat retirement savings on EET principles – contributions are deductible or directly subsidised through the tax system and, for employees, not deemed to be part of pay (E); there is no tax on the saving scheme’s investment income (E) and the final benefits (usually pensions) are taxed as income (T).  In an expenditure tax environment, EET is relatively neutral[96].

That’s because if the government relied entirely on expenditure taxes, taxes are collected when the savings and all other assets are spent.  However, in a world where most government revenue is collected from taxes on income, EET is highly favoured[97].  Such a strategy must therefore be designed to encourage greater self-provision for retirement and, impliedly, to reduce pressure on future government-delivered age pensions.  That last justification would be part of a stronger case if the state pension were means-tested.  Few countries’ age pensions are so tested.

TTE is a ‘neutral’ treatment in an income tax environment.  A bank account is a convenient example: savings into the account come from after-tax income (T); interest earned on the account is added to the saver’s other taxable income (T) while withdrawals from the account are exempt (E).  They are not really ‘exempt’; they are withdrawals of tax-paid capital.

Countries have different shades of these mixtures and usually run both together.  Financial savings that are locked up for retirement may be EET while accessible bank accounts (another potential part of the retirement savings fabric) are TTE.  There may also be reduced tax on ‘retirement’ accounts.  Australia has ‘ttE’[98] which means lower levels of tax than ‘normal’ on contributions and investment income but, overall, retirement saving schemes are greatly favoured by comparison with, say, bank accounts.  On generous assumptions, Australia’s ttE is broadly equivalent to the more usual EET.

[95] This section is based on a submission by Michael Littlewood for the Retirement Commissioner’s 2016 Review: Ageing populations, retirement incomes and public policy: the four ‘first principles’ of policy-making - A submission to the Commission for Financial Capability (accessible here).
[96] They are ‘neutral’ as long as the marginal tax rates on retirement incomes are equivalent to the rates payable on income during the accumulation period.  However, given that retirement incomes are generally lower, average taxes on retirement income will also generally be lower.  This means that, even if all withdrawals are taxed under EET (that usually doesn’t happen), there is a natural tax bias that favours EET in an expenditure tax environment.
[97] In New Zealand, about 60% of tax revenue was income tax in the 2014/15 year: see Briefing for the Incoming Minister of Revenue – 2015, Inland Revenue (accessible here).  40% was through GST and excise duties.
[98] Ross Guest in Comparison of the New Zealand and Australian Retirement Income Systems (2013) accessible here summarises the tax treatment: in Australia, contributions are taxed at a flat rate of 15% to an annual cap of $A25,000.  Investment income is taxed at a rate that probably averages 8% and benefits are tax-free if withdrawn after age 60.  The lowest individual marginal rate of income tax is 19% after a tax-free band of $A18,200.  The tax rules changed from 1 July 2017, including reduced concessions for very high earners and deductibility for employee contributions (see here).


Page 2

Of the three money movements, the tax treatment of the investment accumulation is the most significant.  This reflects the power of investment earnings (i.e. ‘compound interest’) over the very long periods involved in retirement saving’s accumulation and decumulation periods and the difference between pre- and post-tax returns.  Even small differences between pre- and post-tax returns create large differences in the eventual size of the retirement accumulations.  Because of the relatively shorter decumulation period in retirement, even if all the benefits were taxed at the retiree’s top personal tax rate, the government will never recover the value of the concessions given on contributions to the scheme and investment income earned on the accumulating savings[99].  That makes tax incentives for retirement saving very expensive, especially over the long run[100].  That is not their only difficulty:

(a)   Tax incentives are regressive:  The rich can afford to contribute more and so capture most of the value of the concessions[101].  Poorer taxpayers, who cannot afford to save, help pay for the cost of the tax concessions (‘my tax concession is someone else’s tax cost’).

(b)  Regulations are complex: Savings that attract the concessionary treatment must be kept under EET for decades so the regulations that control the money’s entry, accumulation and exit are necessarily intricate[102].  As individuals game the system, the regulations inevitably become more complex and more expensive to administer.

(c)   Distortionary: Tax concessions ‘label’ a particular form of behaviour as preferable to other equivalent behaviour.  EET-approved retirement saving schemes are seemingly better for savers than, say, a bank account that retains the TTE treatment.  Advocates for tax incentives should show why locked-up savings are better for a country than accessible equivalents[103].

Tax incentives also distort ‘signals’.  Fund managers should aim to deliver real returns (more than inflation) to savers.  That task is much easier under EET by comparison with an environment where all ‘income’ is taxed.  Coupled with the fact that EET savings are locked-in until retirement, fund managers do not have to work as hard to achieve real returns or to retain existing business.

Also, savers themselves do not capture the full value of EET concessions.  Savers can afford to be less sensitive to the fees charged by managers of EET savings compared with their TTE equivalents.  That special treatment increases the risks of capture by managers and promoters.  Locking EET savings up until retirement increases those risks.

There is also a suggestion of an unintended consequence of New Zealand’s TTE regime.  Andrew Coleman[104] thinks it may be a cause of New Zealand’s runaway house prices.  Housing has a more favourable tax treatment than retirement saving so disadvantaging the latter.  Because it seems politically difficult to fix the tax treatment of housing it may be preferable to move to the internationally more usual EET for retirement savings.  The trouble with this argument is the absence of direct evidence of the linkage, as the author himself acknowledges.  There may be a correlation or even a coincidence of timing but, unless a direct link is established, this seems a poor justification for re-introducing the distortions of EET to retirement savings.  New Zealand may be the only country to have TTE but is not the only country with significant recent increases in house prices.

(d)  Inequitable: As with compulsion at Tier 2 (see the next section 10), a retirement income policy driven by work-based income necessarily favours higher income earners.  This is a separate point from the regressive nature of tax concessions (paragraph (a) above).  Those with higher rates of pay increases and more complete working lives tend to save more when saving rates are set in relation to pay.  They arrive at retirement with larger retirement accumulations both in money terms and as a proportion of pay.  Tax concessions that favour occupational saving schemes tend to institutionalise these inequalities.

(e)   Deadweight costs:  There are ‘deadweight’ losses to the economy of collecting the extra taxes needed to finance the more fiscally expensive, front-loaded EET environment.  These costs reflect the value of the opportunities that are effectively lost when taxation diverts labour and capital from their best uses[105]. 

(f)    Loss of flexibility: Next, individuals face costs through a loss of flexibility.  Savings might be better spent from a lifetime perspective on an earlier financial crisis (such as a health condition or housing issue) or on a more productive investment, such as buying and building a business or reducing debt.  Compulsory private provision at Tier 2 faces parallel difficulties.

(g)   Do they work? Given that all countries have tax concessions for retirement saving, we might expect studies that demonstrate the ‘value for money’ test.  Do tax incentives actually increase savings?  The answer is ‘possibly not’ despite very large sums that accumulate in tax-favoured schemes.  It’s very difficult, perhaps impossible, to work out because we do not know what might have happened in the absence of the incentives; what economists call the ‘counter-factual’[106].  Some studies suggest the overall impact on the quantum of savings and national saving rates is doubtful[107].

In fact, if households as a whole were perfectly rational, they would allow for the value of tax concessions when setting target retirement saving levels.  The annual amounts required to meet a given target are less if those savings are subsidised through favourable tax treatment.  We should therefore expect lower annual levels of household saving in a tax-favoured EET environment than under TTE because of the large value of the concessions given by taxpayers to the saver’s lifetime saving project.  Given that tax breaks seem not to ‘improve’ the quantum of savings (along with the other difficulties described above), the expensive, complex concessions in an EET environment arguably become pointless.

As a result, while tax policy (or a matching contribution that has similar characteristics to a concession such as KiwiSaver’s ‘member tax credit’) encourages contributions to a retirement saving vehicle (public, occupational or retail), we should expect EET-based incentives to have little, long-term effect on national saving.  There have been remarkably few studies as to whether tax breaks work to improve national saving levels and we have offered some international evidence on that issue.  Further work is needed both to identify overseas studies and to understand the effect of KiwiSaver subsidies on New Zealanders’ recent behaviour.  We have more to say on this in section 11 (The role of the government).

There is no doubt that the financial services industry favours tax incentives for retirement saving and that alone should give us pause for thought.  It is so much easier to generate new business when there is a time-dependent amount paid for by taxpayers.

Given that tax breaks for retirement saving are expensive, complex, inequitable, distortionary, regressive and seemingly don’t work, it’s difficult to understand why the government might be interested in using tax to increase subsidies to savings.  The Minister of Finance, Steven Joyce answered a reporter’s question about that in the 2017 Budget lockup as follows:

“That's an issue that, from my perspective, would repay further work," Mr Joyce said in the yesterday's pre-budget media and analysts’ lockup at the Beehive…
"I literally didn't have time to have a look at it in the current cycle but I'd like to have a look at it in future years – if I get the opportunity."[108]

It might be too much to ask that any review of that issue addresses, let alone answers, any of the problems we have identified with tax breaks for saving but we will suggest the questions anyway.

[99] In How to create a competitive market in pensions: the international lessons (1998), Institute of Economic Affairs, London, Michael Littlewood explains the mathematics behind this suggestion.
[100] Not many countries count the cost of tax incentives for retirement saving.  In 2009, Australia spent almost as much on tax incentives ($A24.6 bn) as it spent on the entire Tier 1 ‘Age Pension’ ($A26.7 bn) – see The great superannuation tax concession rort (2009), David Ingles, The Australia Institute (accessible here).
[101] David Ingles (op cit) suggests that in Australia, “The current concessions provide almost no benefit to low-income earners.”  Again: “The system has become so skewed that the annual cost of providing superannuation tax concessions to high-income earners is much greater than the cost of simply paying those same individuals the age pension.  Providing tax concessions for superannuation as a mechanism to help insulate the budget from the cost of providing for an ageing population is not sensible.”  In the US, about 80% of the value of tax concessions is captured by the top 20% of earners; the bottom 60% of earners capture just 7% of that value – see Tax Deferred Retirement Savings, Seth Hanlon (2011) Center for American Progress, accessible here.  Again in the US, the Congressional Budget Office estimates that the top 20% of households receive nearly twice as much in retirement tax subsidies as the bottom 80 percent combined – see The Distribution of Major Tax Expenditures in the Individual Income Tax System (2013), accessible here.  The total cost of those subsidies to the US tax system was $US137 billion (0.9% of GDP) in 2013.
[102] ‘Protecting’ the tax concessions in KiwiSaver is relatively less intricate than applies in most other jurisdictions though there is ‘leakage’ (first home concessions; disability; death, emigrants).
[103] Some suggest, for example, that “The concessional taxation of superannuation [retirement savings] is…intended to address the bias in the current taxation system against long-term saving.” Submission to the Financial System Inquiry, The Department of the Treasury, Australia, 3 April 2014 at page 44 (accessible here).  This presumes a public policy interest in the relative quality of long-term savings (‘better’) than short-term savings (‘worse’).  Expected after-tax returns on savings, from a timing perspective, should be for savers and investors to decide, not governments.
[104] Housing, the ‘Great Income Tax Experiment’, and the intergenerational consequences of the lease, Andrew Coleman, 2017 University of Otago Business School, accessible here.
[105] This is the same point as referred to in section 7 above with respect to the extra taxes needed to pay for contributions to the NZSF.
[106] Spain introduced tax incentives for retirement saving in 1988.  A report on household behaviour across their introduction concludes that “at most” only one quarter of the contributions were ‘new’ savings: see The Effects of the Introduction of Tax Incentives on Retirement Savings (2007), Juan Ayuso, Juan Jimeno and Ernesto Villanueva, Banco de España (accessible here).  That analysis took no account of the cost to the tax system of lost revenue.
[107] Alicia Munnell in Current taxation of qualified pension plans: has the time come? (1992) Federal Reserve Bank of Boston (accessible here) suggests that the costs of deferring tax on pension accumulations aren’t justified.  Instead, the “taxation of benefit accruals should be shifted to a current basis.”  In Tax Incentives to Saving and Borrowing (2003), Tullio Jappelli and Luigi Pistaferri (accessible here) say “…there is considerable empirical debate as to the effectiveness of tax incentives in promoting saving: most studies conclude that tax incentives affect the allocation of household portfolios, but the effect on the amount saved is less clear-cut.”  In The Effects of 401(k) Plans on Household Wealth (2000 – accessible here), Eric Engen and William Gale suggest that, without regard for the fiscal and regulatory costs, “between 0 and 30 percent of 401(k) balances represent net additions to private savings.”  If the fiscal and regulatory costs were also included, we think those percentages might turn negative.
[108] From the National Business Review, 26 May 2017, accessible here (paywall in place).

Page 3

Questions New Zealand needs to discuss on tax subsidies for saving:

  1. Why does our government have a particular interest in the way New Zealanders save for retirement?  That ‘interest’ is currently expressed through KiwiSaver (that we discuss in section 14 below) and through tax breaks given both to KiwiSaver (estimated cost in the 2017 year: $738 m, rising to $840 m by 2019/20) and to the concessionary tax treatment of ‘Portfolio Investment Entities’ – there is more on that in section 17 (Income tax and saving vehicles).  Where are the cost/benefit analyses to support these direct, costly interventions in New Zealanders’ saving behaviour?
  2. Are we correct that tax breaks for retirement saving are complex, distortionary, expensive, regressive and inequitable?
  3. Are we further correct that tax breaks for retirement saving probably do not ‘work’ (if by that we mean that they probably do not raise savings overall)?  Where is the evidence that costly incentives for KiwiSaver (an accumulated cost to taxpayers of $8.5 billion to date) have increased household savings?  We describe what evidence there is on KiwiSaver’s impact in section 14 below.
  4. If the answers to either question 2 or 3 above are ‘yes’, can we agree that the government should not try to ‘incentivise’ people to save particular amounts in a particular way and in particular investment vehicles for a particular purpose?
  5. Does the presence of tax incentives increase economic growth above the level that it would have been if they had not been there?