Martin Feldstein, who passed away in June 2019, was one of the most important applied economists of the last half-century. His focus on the incentive effects of public sector activity – in spheres ranging from taxation to education finance to healthcare to social insurance – transformed public economics. In addition, he made important contributions to macroeconomics and international economics, and he was a pioneer in the use of large cross-sectional and panel data sets on both households and firms.
Although he only served in government for two years (as chairman of the US Council of Economic Advisers), Feldstein had a major impact on US public policy for nearly half a century. Through his three decades of leadership of the National Bureau of Economic Research (NBER), he helped to create the economics profession as we know it today.
Feldstein’s work had characteristic aspects that have been widely emulated by other researchers, including both of us. Rather than trying to measure more precisely effects that were already the subject of extensive study, such as the impact of taxes on hours worked, Feldstein focused on predictions of economic theory that had not been tested and then examined them using as many different sources of data and evidence as possible.
His interests ranged from the impact of Social Security on saving to the determinants of replacement investment, and from the relationship between inflation, taxes, and interest rates to the way in which the financing of unemployment insurance affected layoffs, and much more. His work was often at the frontier of empirical methodology, but his emphasis was always on gauging the importance of some possible incentive effect or other phenomenon.
Feldstein worked on big questions – the topics that he studied were important when he tackled them, and in many cases they have become even more important over time. He recognised the challenges of pay-as-you-go entitlement programmes for an ageing population, and was a long-time advocate for Social Security reform with a particular preference for greater reliance on private accounts to pre-fund retirement consumption. He highlighted the US tax system’s subsidy to employer-provided health insurance, and argued that the excessive health insurance that resulted was a contributor to the relatively high level of healthcare spending as a share of GDP in the US.
Feldstein was a pioneer in the first ‘big data’ revolution in empirical economics, the widespread analysis of household surveys and public and private compilations of data on corporations. His dissertation research exploited administrative data from the National Health Service (NHS) in the UK. His subsequent research used many different data sets, including household surveys such as the Current Population Survey and the Survey of Consumer Finances, the COMPUSTAT data set on firms, and administrative data on US tax returns, and it helped to make each of these data sources part of the standard toolkit for applied economists.
Feldstein's research in health economics provided foundations for the modern field. His dissertation, written in the mid-1960s while he was a graduate student at Oxford, supervised by Terence Gorman, pioneered the empirical analysis of production functions for hospitals and other healthcare providers. It exploited variation across places and over time within the NHS to develop estimates of the costs and benefits of specific medical procedures.
The findings, which were published in both medical and economics journals, helped to shift the dialogue on healthcare productivity from analysis of specific cases to systematic study of population data sets. The book based on his dissertation stands as a landmark in applied econometric analysis of the healthcare sector (Feldstein 1967).
Feldstein moved from Oxford to Cambridge, Massachusetts in 1967 to join the Harvard faculty. While the absence of a central provider of healthcare services in the US made the sort of analysis he had developed in the UK more difficult – Medicare had been introduced in the US only two years earlier, and the comprehensive data sets on Medicare beneficiary claims that today underpin health economics research were several decades in the future – Feldstein retained his active interest in the healthcare sector.
He was particularly drawn to the income tax exclusion of the value of employer-provided health insurance. He emphasised that the tax benefit associated with employer provision distorted the generosity of insurance plans, and that the moral hazard associated with excess health insurance was a driving force in the inefficient use of health resources.
Feldstein’s study of the efficiency cost of excess health insurance was a seminal contribution (Feldstein 1973). The paper used cross-state data to explain variation in the fraction of the population enrolled in a hospital insurance plan, and the generosity of the insurance coverage in the state. The study, which focused on the 1959-1965 period, before Medicare and Medicaid complicated the analysis of private health insurance markets, found that both insurance coverage and the generosity of insurance plans were positively associated with the availability of employer groups, such as public sector employers or manufacturing firms, that facilitated the provision of employment-based coverage.
Feldstein also used his estimates of the demand curve for hospital insurance to calculate the welfare loss associated with excess insurance. Arguing that moral hazard was important in many markets for medical care, he found that raising the average co-insurance rate from 30% to 50% would result in a substantial efficiency gain in the health insurance marketplace. This paper stands as an early analysis of the complexities of the healthcare sector, the role of private insurance, and the distortions associated with excess insurance.
Feldstein’s interest in health insurance and the public sector’s central role in this marketplace led naturally to a broader research agenda on social insurance programmes. Prior to his work, economists studying social insurance programmes had focused primarily on their distributional and risk-sharing effects. These were often examined by examining who received benefit checks, and who paid the taxes that financed these benefits.
Secondarily, social insurance programmes were recognised as automatic stabilisers from a macroeconomic perspective. Feldstein’s enduring contribution was to emphasise the many and varied incentive effects of social insurance programmes.
During the 1970s, he carried out path-breaking studies on the behavioural effects of Social Security and unemployment insurance, in each case calling attention to unintended consequences of these programmes. He studied how Social Security affected aggregate capital accumulation, retirement decisions, and the nature of private saving, and his research agenda and findings have played a central role in framing subsequent work on this programme.
With respect to unemployment insurance, he demonstrated that more generous benefits resulted in longer unemployment spells, a higher rate of layoffs, and a higher aggregate unemployment rate. In each case, the distortions that Feldstein identified must be traded off against the programme’s benefits – old age security and longevity insurance, and insurance against an interruption of earnings – to determine the optimal policy. Framing and measuring these distortions was his key research contribution.
Feldstein (1974) found that expansion of the US Social Security programme had crowded out private saving. The point estimates suggested that private saving was half what it would have been in the absence of this programme. This paper, along with a contemporaneous study by Alicia Munnell (1974), pointed out that the effect of a government-provided old-age benefit on private saving was ambiguous. The government annuity reduced the need for an individual to save, but if the receipt of benefits was linked to retirement, induced retirement would increase saving.
These studies spawned a substantial literature on the effects of Social Security on aggregate capital accumulation, with Robert Barro (1978) arguing that households’ recognition of the future taxes associated with the Social Security system would reduce and potentially eliminate the wealth effect, and Dean Leimer and Selig Lesnoy (1982) suggesting that correcting a computational error in the original paper yielded much weaker evidence on crowding-out of private saving.
Feldstein approached the question of how Social Security affected saving with the perspective that was characteristic of his research. He was not content to produce one set of parameter estimates from one data set, but marshalled as much evidence from as many data sets as possible.
He also examined international cross-sectional data, aggregate US time series data (Feldstein 1980), and a number of data sets with micro data on individuals (Feldstein and Pellechio 1979). His focus was on pointing to a potentially important impact that needed to be considered in policymaking, rather than on making a precise parameter estimate or specific recommendation.
Feldstein’s research programme on Social Security spanned nearly three decades. He was concerned that the aging US population made the programme unsustainable, and argued, for example, in his Ely Lecture to the American Economic Association (Feldstein 1996), that it was critical for both academic economists and policymakers to devote time and energy to developing Social Security reform proposals and studying their consequences.
He subsequently directed a major research project on both design and risk-transfer issues associated with ‘individual accounts’ types of Social Security programmes (see Feldstein and Campbell 2001 and Feldstein and Liebman 2002). In many ways, his call for greater attention to these issues anticipated the work of the President’s Commission to Strengthen Social Security, which was appointed in 2001. His research, and the follow-on studies that his research stimulated, remain central to discussions of Social Security reform today.
While Social Security is the social insurance programme that most engaged Feldstein’s interest, he recognised that the trade-offs between a fine-mesh social safety net and the distributional benefits it provides, and the efficiency costs of such a system, are ubiquitous and can be found in many settings. Economists had recognised prior to Feldstein’s work that unemployment insurance could prolong unemployment by reducing the incentive to search for work. Feldstein broke important new ground by emphasising this programme’s impact on employer behaviour in initiating separations.
Feldstein (1976) described the incentives created by the US unemployment insurance system, both in reducing the burden on workers who are laid off and, in some cases, because of detailed institutional features of the firm-specific experience rating system, subsidising firms with high layoff rates. It pointed out that the programme encouraged seasonal work by allowing individuals to work part-year on a regular basis, rotating, for example, between winter work at a ski resort and receiving unemployment insurance during summer months, with firms bearing only a fraction of the layoff costs.
A related paper (Feldstein 1978) concluded that roughly half of temporary layoff unemployment could be attributed to the incentives created by the unemployment insurance system. The discovery that temporary layoffs accounted for a significant share of overall unemployment, that they typically resulted in relatively short spells of unemployment and often in recall to the same job at the prior employer, was an important insight that called for some re-evaluation of the welfare cost of unemployment more generally.
Feldstein’s work demonstrated that social insurance programmes that were usually thought of primarily for their distributional consequences had important efficiency costs. These distortions needed to be, and could be, measured using household-level data, and the resulting findings needed to be considered in programme design. Hundreds of studies have built on the foundations that he created, and these studies have informed the applied policy analysis carried out in finance ministries and treasuries around the globe.
Feldstein recognised that the US income tax system distorted many decisions besides the purchase of health insurance, and he deployed his skills to frame the efficiency costs of various tax provisions, and to develop estimates of these costs. In many cases, his findings suggested that the incentive effects of the tax code were larger than was recognised.
Feldstein’s analysis of the income tax deduction for gifts to charity illustrates this. Feldstein and Charles Clotfelter (1976) analysed household data from the Federal Reserve Board’s 1963 Survey of Consumer Finances, and Feldstein and Amy Taylor (1976) analysed individual income tax returns from 1970. Both studies found that the elasticity of charitable giving with respect to the after-tax price of giving was between -1.0 and -1.5. These values suggested that the amount of charitable giving induced by the tax deduction exceeded the revenue cost of this deduction to the US Treasury.
Feldstein found similarly large effects of the tax code when he studied the elasticity of capital gain realizations with respect to the marginal tax rate on such gains. His study with Joel Slemrod and Shlomo Yitzhaki (1980) examined the patterns of capital gain realisations in US income tax data, where non-capital gain income creates variation in the marginal tax rate at which gains are taxed. The estimates of the realisation elasticity were large, and for some values, it appeared that the revenue generated by the capital gains tax might rise if the tax rate were lowered – a rare example of the theoretical possibility that lower rates might raise revenues.
Feldstein’s most enduring contribution to empirical tax analysis was his study of the elasticity of taxable income with respect to the marginal tax rate (Feldstein 1995). Recognising that most studies of how tax rates affect behaviour focus on just one dimension, such as hours of work but not effort or engagement at work, both of which could matter for earnings, Feldstein suggested that taxable income – a taxpayer’s gross income less deductions – was the most appropriate variable to study when considering the distortions from the tax code.
He recognised that under some conditions, a taxpayer would be indifferent between an increase in labour supply that resulted in one additional dollar of after-tax income and the effort required to generate another dollar of tax deduction that would result in the same increase in taxable income.
Feldstein’s study of the 1986 Tax Reform Act in the US, using administrative data on tax returns, found relatively large elasticities of taxable income with respect to the after-tax price of such income. This study has spawned a substantial literature, using tax return data for many time periods and spanning many periods, that has provided detailed information on taxable income elasticities in many settings.
Capital taxation and cross-country capital flows
A final strand of notable research, which emerged from research on capital income taxation but had impacts that ranged far beyond public finance, concerns the tax incidence in open economies. A number of Feldstein’s papers in the late 1970s and early 1980s (collected in Feldstein 1983) drew attention to the unintended consequences of high inflation for the effective tax rate on corporate capital income. Taxes on nominal capital gains, and the reduced value of nominal depreciation allowances in a high-inflation environment, raised tax burdens.
In a closed economy setting, raising the tax burden on corporate capital would lower the return to investors and the burden of this higher tax would likely be shared between capital owners and labourers. But in an open economy, where savers must be indifferent to investing in any location, the stock of capital in the high tax jurisdiction must contract, and the pre-tax return must rise, so that the after-tax return to investors remains equal to the return available elsewhere. This body of research suggested that determining whether the US is closer to an open economy or a closed one is a key issue for tax analysis.
While a body of research in open economy macroeconomics, particularly based on large gross capital flows between nations, suggested the open economy benchmark, Feldstein investigated this issue by studying the net flows. In a very influential study with Charles Horioka (1980), he discovered that despite significant cross-border capital flows, there was a very strong correlation between a country’s gross saving rate and its gross investment rate. This finding, which represents a challenge to much of the standard analysis in open economy macroeconomics, has been actively studied since the publication of the Feldstein-Horioka analysis.
There is no ‘Feldstein theorem’, ‘Feldstein estimator’, or ‘Feldstein model.’ Inevitably, with the passage of time, the accumulation of new data, and the development of new empirical techniques, Feldstein’s empirical papers are no longer viewed as the state-of-the-art estimates of key effects. Many of the questions he explored, such as the magnitude of the distortions induced by capital income taxation, have been considered in more sophisticated frameworks than those that he employed. Feldstein’s papers are no longer the reading list staples that they were a generation ago.
But in other, perhaps more profound ways, Martin Feldstein leaves behind a deep scholarly legacy. His approach to research has shaped not just our own, but that of many other scholars. We learned from him to “be about the economy, not the economics literature”.
Feldstein conveyed to his students and the profession the idea that “no single parameter estimate will change anybody’s mind about anything. To be convincing, an empirical conclusion has to emerge from multiple approaches and multiple data sets.”
Closely related was his emphasis on the importance of new information in gaining new insights. Most of his research papers involved the construction of new variables, the use of previously unexploited data, or the measurement of previously unmeasured effects.
Feldstein was a key figure in the change in the way economists viewed the economy that took place during the 1970s, and which found its political expression in the Reagan and Thatcher revolutions of the 1980s.
When Feldstein entered the economics profession, economists generally believed that government programmes had large income effects and small incentive effects, that stabilisation of business cycles was the principal macroeconomic task of government, that efficiency losses from tax distortions were small, and that accepting higher inflation in order to reduce unemployment was prudent in many circumstances.
Many factors – including, importantly, Feldstein’s research and advocacy – contributed to moving that post-war Keynesian belief system out of fashion. The counter-revolutions of today – calling for expanded benefits, larger deficits, and higher taxes – are a testimony to the enduring influence of Feldstein’s work.
Barro, R J (1978), The Impact of Social Security on Private Saving, American Enterprise Institute.
Feldstein, M S (1967), Economic Analysis for Health Service Efficiency: Econometric Studies of the British National Health Service, North-Holland.
Feldstein, M S (1973), “The Welfare Loss of Excess Health Insurance”, Journal of Political Economy 81: 251-80.
Feldstein, M S (1974), “Social Security, Induced Retirement and Aggregate Capital Accumulation”, Journal of Political Economy 82: 905-20.
Feldstein, M S (1976), “Temporary Layoffs in the Theory of Unemployment”, Journal of Political Economy 84: 937-57.
Feldstein, M S (1978), “The Effect of Unemployment Insurance on Temporary Layoff Unemployment”, American Economic Review 68: 737-42.
Feldstein, M S (1980), “International Differences in Social Security and Saving”, Journal of Public Economics 14: 225-44.
Feldstein, M S (1983), Inflation, Tax Rules and Capital Formation, University of Chicago Press.
Feldstein, M S (1995), “The Effect of Marginal Tax Rates on Taxable Income: A Panel Study of the 1986 Tax Reform Act”, Journal of Political Economy 103: 551-72.
Feldstein, M S (1996), “The Missing Piece in Policy Analysis: Social Security Reform”, American Economic Review 86 (May): 1-14.
Feldstein, M S, and J Y Campbell (2001), Risk Aspects of Investment-Based Social Security Reform, University of Chicago Press.
Feldstein, M S, and C T Clotfelter (1976), “Tax Incentives and Charitable Contributions in the United States: A Microeconometric Analysis”, Journal of Public Economics 5: 1-26.
Feldstein, M S, and C Y Horioka (1980), “Domestic Savings and International Capital Flows”, Economic Journal 90: 314-29.
Feldstein, M S, and J B Liebman (2002), Distributional Aspects of Social Security and Social Security Reform, University of Chicago Press.
Feldstein, M S, and A J Pellechio (1979), “Social Security and Household Wealth Accumulation: New Microeconometric Evidence”, Review of Economics and Statistics 61: 361-68.
Feldstein, M S, J Slemrod, and S Yitzhaki (1980), “The Effects of Taxation on the Selling of Corporate Stock and the Realization of Capital Gains”, Quarterly Journal of Economics 94: 777-91.
Feldstein, M S, and A Taylor (1976), “The Income Tax and Charitable Contributions”, Econometrica 44: 1201-22.
Leimer, D R, and S D Lesnoy (1982), “Social Security and Private Saving; New Time Series Evidence”, Journal of Political Economy 90: 606-29.
Munnell, A H (1974), “The Impact of Social Security on Personal Savings”, National Tax Journal 27: 553-67.