Do Tax Incentives Increase National Saving?

01 Jul 2002

Have the tax incentives of the Individual Retirement Account (IRA) programme in the United States led to an increase in national saving? Only by a small fraction according to new research by Professors Orazio Attanasio and Thomas DeLeire, published in latest issue of the Economic Journal.


Their analysis indicates that households financed the majority of their IRA contributions not from a reduction in consumption but rather from existing saving or planned saving. These results should lead policy-makers to re-examine tax-favoured saving accounts to determine whether the additions to household saving that are induced by tax incentives are substantial enough to justify the loss in tax revenue from such programmes.


Attanasio and DeLeire examine consumption data from a sample of households that had just opened an IRA account or had an existing IRA. They test whether the IRA programme led to an increase in national saving by determining whether newly contributing households decreased their consumption on entering the programme. Results suggest that there was no substitution between consumption and saving when households began participating in the programme. The researchers interpret these results as providing evidence against the hypothesis that the IRA legislation was effective in generating new saving.


While participation in the IRA programme may have increased household saving (through reduced tax liabilities), results indicate that there was little or no increase in national saving between 1982-6. Attanasio and DeLeire also examine changes in the non-IRA financial assets of IRA contributors. They find that on average, new contributors reduced their non-IRA assets by over $1,400 compared with old contributors. They estimate that at most 9-20% of the IRA contributions of new contributors represents new national saving.


These figures are lower than the results found by studies that compared IRA contributors with non-IRA contributors, only using data on assets. The key policy question remaining is whether a 9 -20% increase is a large enough percentage to justify the IRA programme.


The authors also examine household behaviour after 1986, when eligibility to make tax deductible IRA contributions was limited and many of the tax advantages of the IRA programme were removed. They find that there was less reshuffling during this period. But the results are mixed, as they also find no evidence that these contributions constituted new saving. Future research should examine whether tax incentives for saving are more effective for lower income households.


The US national saving rate has declined dramatically over the past 10-15 years. This decline may have significant implications for the domestic capital accumulation rate, productivity growth and the ability of households to finance their retirements. The primary reason for the decrease in national saving is the decline in the US personal saving rate. There is little agreement among econo mists as to what has caused this decline or the most effective way to reverse the trend. One proposed method is the tax favoured saving account, such as the IRA, which allows households to defer paying taxes on both contributions to and interest earned from these accounts. Following the 1986 Tax Reform Act, eligibility for IRAs was restricted, leading to a sharp decrease in IRA contributions: from almost $50 billion in 1986 to about $12 billion in 1987.


Most empirical studies of IRAs have focused on whethe r contributions before the Tax Reform Act of 1986 represented new saving or merely a reshuffling of assets. These studies often compared contributors to non-contributors and used data on household assets. In contrast, Attanasio and DeLeire compare the cons umption growth of households that had recently opened an IRA account with households that had already invested in an IRA. The difference in the conduct of these two household groups indicates the effect of the tax incentives on retirement saving. The researchers also conduct a test based on changes in the non-IRA financial asset balances for these two household groups. These tests make it possible to see whether households financed their IRA contribution from existing saving, planned saving that would have been conducted anyway, or from decreases in consumption.


Attanasio and DeLeire use the Consumer Expenditure Survey (CEX), a representative sample of US households available continuously since 1980. In the CEX, households are interviewed four times a year and information is gathered on consumption expenditure, income, assets, household characteristics and contributions to IRAs. The CEX is the only data set that contains information on the consumption, IRA contributions and non-IRA financial asset growth of a large sample of US households – an ideal data set on which to examine the new saving and reshuffling hypotheses.


''The Effect of Individual Retirement Accounts on Household Consumption and National Saving'' by Orazio Attanasio and Thomas DeLeire is published in the July 2002 issue of the Economic Journal. Attanasio is Professor at University College London and Research Fellow of the Institute for Fiscal Studies, London; DeLeire is Assistant Professor at the University of Chicago Harris School.

Orazio Attanasio

+44-207-679-5880 | o.attanasio@ucl.ac.uk