RDP 2007-08: The Effect of the Australian Superannuation Guarantee on Household Saving Behaviour 1. Introduction

Individual pension accounts are growing in importance as a pillar of retirement incomes policy in developed economies. However, the framework adopted differs from country to country. Australia, Sweden and Switzerland have introduced compulsory accounts, while in the United Kingdom employees automatically receive accounts with an option to opt-out. At the same time, the debate on social security reform in the United States centres on whether funds should be diverted into individual accounts. Policy-makers around the world have generally assumed that introducing such pre-funded pension schemes improves retirement incomes and increases household wealth. However, there has been relatively little empirical work that tests this assumption.

This paper focuses on the effect that Australia's system of compulsory pension accounts (the ‘Superannuation Guarantee’) has had on household saving. This scheme, which was introduced over 1986–1992, requires employers to pay a percentage of their employees' earnings into individual pension accounts.^[1] The accounts are managed by private sector pension funds and cannot be accessed until the employee retires after the age of 55.^[2] The initial contribution rate was set at 3 per cent of labour income in 1986 and gradually increased to 9 per cent by 2002. The scheme has raised pension plan coverage from around 40 per cent of the workforce in 1983 to 90 per cent since the mid 1990s.^[3]

Australia's system provides a natural experiment since a small share of employees do not receive employer contributions to pension accounts. As a result, it is possible to compare households that received contributions with those that did not. Using data from the Household, Income and Labour Dynamics in Australia (HILDA) Survey, this paper attempts to answer three related questions:

1. Have compulsory pension accounts increased household wealth?
2. What effect do compulsory pension accounts have on voluntary saving for retirement?
3. Do compulsory pension accounts influence the timing of retirement?

Turning to the first question, compulsory pension accounts will raise wealth if households do not increase consumption to fully offset the growth of their pension accounts. In a life-cycle model with no financial constraints, households who view their pension accounts as perfect substitutes for other assets would reduce those other assets (or, if need be, borrow) so as to offset compulsory contributions to their pension accounts, leaving their net wealth unchanged. However, if some households are financially constrained, they may not be able to do this. Furthermore, households might not view their pension accounts as perfect substitutes for other forms of saving. For instance, households may value the balance in their retirement account less than a more liquid investment if they are financially constrained. Finally, tax incentives could lead households to voluntarily save a higher share of their wealth in pension accounts, although the effect on net wealth is theoretically ambiguous due to offsetting income and substitution effects.

While this first question has not been addressed directly in the literature, there are several related findings. When the scheme was introduced, the Australian Government projected that for every dollar contributed to the pension accounts, other savings would fall (be ‘offset’) by 30 to 50 cents. Using macroeconomic data, Connolly and Kohler (2004) find that the offset is likely to be in this range, with a point estimate of 38 cents. However, the balance in pension accounts reflects not only contributions, but investment returns. Dvornak and Kohler (2007) examine the effect of increases in wealth (including investment returns) on consumption, although they do not examine the impact of compulsory pension accounts in particular. According to Dvornak and Kohler, a one dollar increase in stock market wealth (including pension accounts) results in consumption rising by between six and nine cents. Since compulsory pension account balances reflect both contributions and investment returns, these two papers can provide us with reasonable upper- and lower-bound estimates of the effect of the accounts on household wealth. One could expect that compulsory pension accounts would increase wealth by between 62 cents in the dollar (the effect of compulsory pension contributions in Connolly and Kohler) and 94 cents in the dollar (the effect of stock market wealth in Dvornak and Kohler). Broadly consistent with this, aggregate pension (superannuation) assets have increased significantly over the past two decades, rising by almost 100 percentage points as a share of household income, while financial assets and total net wealth have increased by around 130 and 230 percentage points respectively as a share of household income (Figure 1).

For the purpose of analysing retirement incomes, it is possible to decompose household saving into three parts: compulsory pension contributions, voluntary pension contributions and non-pension saving. In answering the second question above, I define voluntary retirement saving as making voluntary contributions to pension accounts, where these funds cannot be accessed until retirement.^[4] Compulsory pension accounts will only increase retirement incomes if households do not reduce their voluntary retirement saving in response. Since compulsory and voluntary pension contributions are close substitutes, households could be expected to reduce their voluntary contributions in response to an increase in compulsory contributions. On the other hand, the implementation of a compulsory pension account scheme might actually increase voluntary saving for retirement by shifting household preferences. For instance, if these accounts alert households to the importance of retirement planning, then they may choose to save more for retirement (Mariger 1997 and Elmendorf and Liebman 2000 make similar arguments in the context of the US debate on pension reform). Furthermore, by automatically providing households with pension accounts, the compulsory system may make it more convenient for them to save.^[5] Some employers also provide employees with saving incentives such as matched contributions. In aggregate, total contributions to pension accounts (including voluntary contributions) have increased since 1986, broadly in line with the rising compulsory contribution rate (Figure 2).

The third question is whether compulsory pension accounts could also influence the timing of retirement. If contributions are not completely offset by reductions in other saving, households with compulsory pension accounts will accumulate more wealth as they approach retirement than households without accounts. Assuming that leisure is a normal good, these households would prefer to retire early (Freebairn 2004). However, Samwick (1998) and Gustman and Steinmeier (2001) find that the level of wealth has only a small effect on the retirement decision. Compulsory pension accounts could also influence the retirement decision by emphasising to households the importance of retirement planning, encouraging them to work longer to achieve an acceptable standard of living in retirement. In aggregate, the labour force participation rate of persons aged 55 to 64 has been increasing since 1986, mainly due to higher female participation, but also due to a recent turnaround in male participation among this age group (Figure 3). It is worth noting, however, that since the analysis in this paper is based on data for the 2002 to 2003 period, it says nothing about any effects on saving and retirement behaviour associated with the changes to superannuation announced in the 2006 Budget.

The remainder of the paper is structured as follows: in Section 2, the methodology is outlined; in Section 3, the dataset is introduced; in Section 4 the results are analysed; and the conclusions are drawn in Section 5.

Footnotes

The initial scheme was known as Award Superannuation from June 1986. The broader Superannuation Guarantee scheme was introduced in July 1992. [1]

The Preservation Age is 55 for persons born before 1 July 1960 and rises to 60 for persons born after 30 June 1964. [2]

For a more comprehensive description of retirement incomes policy in Australia, see Edey and Gower (2000), Connolly and Kohler (2004) and Bateman and Kingston (forthcoming). [3]

Non-pension saving can also be consumed in retirement. However, only contributions to pension accounts are guaranteed to be for the purpose of retirement saving, as a result of the preservation rules. [4]

This is analogous to the finding of Madrian and Shea (2001) that households are more likely to save in 401(k) accounts in the US if they are automatically enrolled by their employer. [5]