Who Bears the Lifetime Tax Burden?
Chapter One is a somewhat lengthy but nonetheless crisp introduction and summary. The authors begin with a careful discussion of why LI rather than annual or life cycle income data are required for determining tax burdens. The procedure employed to calculate LI in this book requires five steps. First, from the Panel Study of Income Dynamics (PSID) for the 1970-1987 period, wage functions are estimated and used to obtain wage profiles, which allows one to calculate the present value of potential LI. Individuals are then ranked by LI, and classified into twelve income groups. Second, within each LI group wage profiles are reestimated by age; also, age profiles are estimated for taxes and transfers. Third, Consumer Expenditure Survey data are used to estimate how individuals allocate consumption among specific goods and services. Fourth, a general equilibrium simulation model is constructed comprising nearly all U.S. taxes, seventeen industries (broken down by corporate and noncorporate sectors, by industry), and consumers. Each individual begins "life" with a given inheritance, a set of tax rates, wage profile, and a transfer profile. Each person plans a lifetime of labor supply, savings, demand for goods, and bequests. Industry's use of labor, capital, and intermediate inputs represent the supply side of the picture. The effects of tax changes on economic decisions are simulated through time. Welfare changes are computed using equivalent variations. Fifth, the effects of taxes are made by comparing the base case burden with that of a proportional tax which raises the same revenue. The findings are that personal income taxes are moderately progressive, while sales and excise and payroll taxes are regressive. Capital taxes affect those with low and high incomes more than individuals in the middle of the income distribution. An unexpected result is that the corporate income tax is found to be regressive.
Chapter Two discusses the theoretical basis of the lifetime incidence model. Consumers have a CES utility function with a composite good that includes leisure and a composite consumption good. Consumers allocate their income across periods to maximize utility subject to a lifetime budget constraint. From the maximization process, one obtains leisure demand, labor supply and composite good demand curves. Composite good consumption is modeled as a Stone-Geary function of the (seventeen) individual consumption goods. The production side of the economy uses annual optimization. The market structure is perfectly competitive, and firms face constant returns to scale CES production functions. Income taxes are a linear function of income for each age and LI category. General equilibrium occurs when there is equilibrium in the goods and factor markets, and government spending equals revenue.
Chapter Three reviews data sources. The benchmark year for the model is 1984. Production data are primarily from the Survey of Current Business and the Economic Report of the President. Chapter Four shows how LI is calculated. The unit of account is the individual. The PSID sample is restricted to households where the head individual and their spouse (if any) are intact from 1970-1987. Clearly this restriction leads to a biased sample. The wage functions estimated with the PSID sample are limited to those that earned positive wages. A Heckman correction for this possible bias was employed, but not found to be significant. An economic lifetime in the model is 60 years (age 20 through 79). Individuals have 4000 hours of time endowment per year. The net-of-tax rate of return is set at 4%. The value of leisure is assumed to equal the net wage, contrary to a long literature which shows that people value their leisure at up to one half of their wage. The average household income when two persons work is the average of the two individual incomes. The profiles calculated for each LI group are found to be more equal than profiles calculated from annual income data.
Chapter Five focuses on consumption patterns by age. Surprisingly, the data show that the share of discretionary income going to health care does not change with age. Federal data for 1986 show that health care consumption shares of total consumption expenditures rise monotonically with age .
Chapters Six and Seven calibrate the parameters and models, and provide the simulations and results, respectively. Chapter Eight does sensitivity tests of the results. In the base case, the authors use a wage elasticity of labor supply of .11, based on work by Hausman. T. MaCurdy and others  have found the Hausman analysis to be overly restrictive. MaCurdy shows that the wage elasticity is not significantly different from zero. When the authors lower the wage elasticity to .03, they find that the excess burden of all taxes declines by 30%. A second (of eight) sensitivity test relaxes the minimum purchase requirement associated with the Stone-Geary model. This leads to a 50% gain in efficiency compared to the base case. If (lower income) persons do not have to buy minimum amounts of goods, the personal income tax induces a switch towards leisure; thus, labor taxes are more distortionary. The upshot is that the choice of a Stone-Geary biases the welfare burden of taxes upward. This problem is exacerbated by the misestimate of the wage elasticity of labor supply. Chapter Nine tests the effect of the Tax Reform Act of 1986 on the model and finds little new information. A uniform consumption or value added tax (VAT) is also tested. Interestingly, the authors find that if the VAT replaced the sales, excise, payroll, or corporate income tax, the lower income persons would be helped the most.
This book is clearly and carefully written. It represents a great deal of work, and makes numerous contributions to tax incidence analysis. By overstating the wage elasticity of labor supply, imposing the Stone-Geary functional form, and restricting the sample of consumers to intact households only, I believe the excess burden estimates are heavily biased. But this is an empirical question that the Fullerton and Rogers model can easily be altered to test. I strongly recommend this book to all students interested in determining the burden of our taxes.
Douglas M. Brown Georgetown University
1. Torrey, Barbara and Eva Jacobs, "More than Loose Change: Household Health Spending in the United States and Canada." Health Affairs, Spring 1993, 126-31.
2. MaCurdy, Thomas; Green, Davis and Harry Paarsch, "Assessing Empirical Approaches for Analyzing Taxes and Labor Supply." The Journal of Human Resources, Summer 1990, 415-90.