Service done from time to time in the same spirit of the work that I have done, but that work is constrained to a relatively minor role within the statistical operations of the Internal Revenue Service.

What can be done to increase our knowledge of wealth? I would say that the major thing that needs to be done is to support efforts to improve our ability to measure, and the kinds of things that we should be measuring are not so much the things I have been measuring the static distributions of wealth-but the processes by which wealth is accumulated. We ought to be very concerned about the processes by which one invests, and we ought to try to distinguish how much of the unevenness in the distribution of wealth comes about by chance.

So if we are to improve our understanding, we need to do a great deal of work with respect to improving our ability to measure, improving the survey research techniques and our ability to analyze tax data, and that brings me to another area, and that is to try to understand what we are doing with respect to confidentiality legislation of all types.

What we would like to do is to protect people from harm because someone knows something about them that can be used to their detriment. The data used by social scientists-economists, in particular— really are fairly innocuous. The use of income tax returns, for sampling, as Mr. Morgan mentioned, the information that he has collected, even if it were rather widely available, are unlikely to be detrimental.

I am not denigrating the importance of providing privacy to individuals, but the types of legislation that we have engaged in with respect to privacy has done very little to protect us from really important invasions of privacy that have been carried on by the executive branch of Government through the IRS, or through the CIA, or through the FBI.

No one is going to, I would suggest, effectively stop the administration from acquiring your tax return, sir, if it so desires, and it doesn't matter which party is in power. If they desire the tax return, the chances are very good that they can get it. They certainly could get mine. But if we wanted to use tax returns, not so that we could harass people, but to improve the fairness of the tax system or the transfer system, we have a great deal of difficulty getting access to those tax

returns.

It took 7 years in a more favorable climate than this for me to get access to the Federal estate tax returns on computer tapes without any names associated with individuals' information. During those 7 years the IRS put up every kind of conceivable argument how detrimental it would be to people if that information was available. The information was finally made available to me with the support of a number of Members of Congress and some members of the administration, and the Republic still stands. In fact, the IRS now distributes the data to anybody who wants that information. Nothing has come about that I know of that has been injurious to anyone.

We certainly should protect citizens from serious invasion of privacy, but we should also provide researchers and policymakers the data needed to understand how society functions so we can make it work better.

That is all the comments I have this morning.

Mr. FRASER. Good; thank you very much, Professor.
Ms. Projector.

STATEMENT OF DOROTHY S. PROJECTOR, DIRECTOR, DIVISION OF ECONOMIC AND LONG-RANGE RESEARCH, SOCIAL SECURITY ADMINISTRATION

MS. PROJECTOR. Mr. Chairman, I appreciate this opportunity to talk about the important subject of data on the distribution of wealth in the United States. I would like to deal with two of your questions: Why do we need to know about the distribution of wealth? Is it important? How would such knowledge linkup with policy choices? The second area, in what areas do we lack data? What more do we need to know, and why?

As I see it, and I think this is consistent with the statements of both Mr. Morgan and Mr. Smith, we need to know about the distribution of wealth for two main purposes. One is as one measure of social and economic well-being, and, second, as a variable which is important in determining other measures of social and economic well-being, such as income, consumption or saving, and labor supply.

The first of these purposes is widely understood and accepted; in other words, the kind of thing that was being discussed when we talked about the concentration of wealth, whether it had changed over time. The second one, I think, is probably less so, and that is the thing we talk about when we say why do we have a particular concentration of wealth. Is it because the saving rate has differed among people in such a way that some people are richer than others, or does it have to do with asset appreciation, and so on?

I would like to talk about the second of these, in other words, the use of wealth as a variable in explaining other measures of economic and social well-being.

For many years wealth has played a central role in analyses of the factors determining consumption or saving. Robert Ferber's survey article provides a good summary of the state-of-the-art as of the early 1970's. That is listed in the list of references.

Wealth is a particularly crucial variable in analyses which focus on the allocation of consumption over the lifespan. These analyses may be formulated in terms of consumers allocating their resources over the lifespan in some optimum manner, as in the work of Modigliani and his associates. Or the analysis may be cast in terms of permanent income which translates resources into a current flow concept as in the work of Friedman. For the past 20 to 25 years, most of the empirical work on factors determining consumption or saving has been formulated in terms of a lifecycle model, either explicitly or implicitly through the use of permanent income. More recently, work in the lifecycle area has been oriented toward the simultaneous determination of consumption and labor supply.

Lifetime resources, of course, reflect an individual's future earnings stream, as well as any wealth which he has already accumulated, and both of these concepts post difficult measurement problems. The difficulties of measuring future earnings are obvious. Moreover, many of the problems of measuring accumulated wealth for individuals are well known-for example, response error, problems of sample designbecause certain forms of wealth are so highly concentrated, and so on. However, less attention has been paid to the problems of measure

ment of wealth in the form of public and private pension rights. Thus, until a few years ago, the empirical analyses of lifecycle theories of consumption did not incorporate wealth in the form of rights to public pension income. In recent years attention has been focused on efforts to measure those rights in the form of social security benefits and to incorporate that measure in a lifecycle model of consumption, as in the work of Feldstein and Munnell.

One of the findings of the Feldstein work-that social security has lowered the saving rate and hence capital formation by a sizeable amount since its inception-has received much attention. The finding obviously has important policy implications. The Feldstein work which is based on U.S. time series data, has been challenged by others. Barro, for example, argues that Feldstein's social security wealth variable is so highly correlated with an unemployment variable that it is not possible to distinguish the effect of social security wealth from that of the unemployment rate over time.

It is not my intention to discuss this matter substantively. Rather, it is my purpose to point out that the general question of the factors determining the saving rate has been and continues to be an important policy question. The effort to improve the empirical work in this area by incorporating wealth in the form of pension rights is to be applauded.

However, it is also obvious that empirical analyses based on time series data are subject to serious limitations. Because of the importance of the policy questions, other approaches should be pursued. In particular, analyses based on a cross section of individuals or families should be pursued. Thus, if we are to engage in collection of data on wealth and its distribution, an extremely expensive and difficult task, the thrust of the effort should be to measure wealth in the form of public and private pensions as well as the forms of wealth which have been covered in the past. To put the matter more strongly, I would question whether an effort to measure wealth as it has been defined in past surveys would add much to our knowledge of consumer saving behavior, even if the difficult problems of measuring such wealth were resolved. Note there I am talking about behavior, not so much the idea of wealth as a measure of economic and social well-being.

Information on wealth in the form of pension rights is also important in analyzing policy questions relating to retirement decisions. For example, in some recent work I attempted to test a simple incomeleisure model for the aged population. In that analysis the aged individual is assumed to choose an optimum combination of leisure and other goods and services, subject to the income available to him. In testing the model, the individual's constraint should incorporate the pension income which would be available to him, if he should choose to work less than full-time. It will be possible to incorporate social security benefits in the analysis because of work which matches the data source the current population survey-to administrative data. However, information on private pension rights is not available, and the analysis suffers from that deficiency.

To summarize, information on the distribution of wealth is important and can be shown to link up directly with policy issues. In any new data collection I strongly urge that the concept of wealth cover public and private pension rights, as well as the information on liquid

98-815 O-78-2

assets, securities, real estate, consumer durables, and business holdings which has been covered in past surveys. Thank you.

[Testimony resumes on p. 29.]

[The prepared statement of Ms. Projector follows:]

PREPARED STATEMENT OF DOROTHY S. PROJECTOR

Mr. Chairman, I appreciate this opportunity to testify on the important subject of data on the distribution of wealth in the United States. My statement is primarily directed to two of your sets of questions:

Why do we need to know about the distribution of wealth? Is it important? How would such knowledge link up with policy choices?

In what areas are we lacking in data? What more do we need to know? Why? We need to know about the distribution of wealth for two main purposes: (1) as one measure of social and economic well-being; (2) as a variable which is important in determining other measures of social and economic well-being such as income, consumption or saving, and labor supply.

The first of these purposes is widely understood and accepted, while the second is probably less so. I should like to concentrate my testimony on the second purpose.

For many years wealth has played a central role in analyses of the factors determining consumption or saving. Robert Ferber's survey article provides a good summary of the state of the art as of the early 1970's [4, 1973].

Wealth is a particularly crucial variable in analyses which focus on the allocation of consumption over the lifespan. Such analyses may be formulated in terms of consumers allocating their resources over the lifespan in some optimum manner, as in the work of Modigliani and his associates [1, 1963; 8, 1954]. Or the analysis may be cast in terms of permanent income which translates resources into a current flow concept as in the work of Friedman [6, 1957]. For the past 20 to 25 years most of the empirical work on factors determining consumption or saving has been formulated in terms of a life-cycle model, either explicitly or implicitly through the use of permanent income. More recently, work in the life-cycle area has been oriented toward the simultaneous determination of consumption and labor supply [7, 1975].

Lifetime resources, of course, reflect an individual's future earnings stream, as well as any wealth which he has accumulated, and both of these concepts pose difficult measurement problems. The difficulties of measuring future earnings are obvious. Moreover, many of the problems of measuring accumulated wealth for individuals are well-known, e.g., response, error, problems of sample design because certain forms of wealth are so highly concentrated, and so on [5, 1966; 10, 1966]. However, less attention has been paid to the problems of measurement of wealth in the form of public and private pension rights. Thus, until a few years ago the empirical analyses of life-cycle theories of consumption did not incorporate wealth in the form of rights to public pension income. In recent years attention has been focused on efforts to measure those rights in the form of social security benefits and to incorporate that measure in a life-cycle model of consumption, as in the work of Feldstein [3, 1974] and Munnell [9, 1974]. One of the findings of the Feldstein work-that social security has lowered the saving rate and hence capital formation by a sizable amount since its inception has received much attention. The finding obviously has important policy implications. The Feldstein work which is based on U.S. time series data has been challenged by others. Barro [2, 1977], for example, argues that Feldstein's social security wealth variable is so highly correlated with an unemployment variable that it is not possible to distinguish the effect of social security wealth from that of the unemployment rate.

It is not my intention to discuss this matter substantially. Rather, it is my purpose to point out that the general question of the factors determining the saving rate has been and continues to be an important policy question. The effort to improve the empirical work in this area by incorporating wealth in the form of pension rights is to be applauded. However, it is also obvious that empirical analyses based on time series data are subject to serious limitations. Because of the importance of the policy questions, other approaches should be pursued. In particular, analyses based on a cross-section of individuals or families should be pursued. Thus, if we are to engage in collection of data on wealth and its distribution, an extremely expensive and difficult task, the thrust of the effort should be to measure wealth in the form of public and private pensions as

well as the forms of wealth which have been covered in the past. To put the matter more strongly, I would question whether an effort to measure weath as it has been defined in past surveys would add much to our knowledge of consumer saving behavior, even if the difficult problems of measuring such wealth were resolved.

Information on wealth in the form of pension rights is also important in analyzing policy questions relating to retirement decisions. For example, in some recent work I attempted to test a simple income-leisure model for the aged population [11, 1977]. In that analysis the aged individual is assumed to choose an optimum combination of leisure and other goods and services, subject to the income available to him. In testing the model the individual's constraint should incoporate the pension income which would be available to him, if he should choose to work less than full-time. It will be possible to incorporate social security benefits in the analysis because of work which matches the data source the Current Population Survey-to administrative data [12, 1975]. However, information on private pension rights is not available and the analysis suffers from that deficiency.

To summarize, information on the distribution of wealth is important and can be shown to link up directly with policy issues. In any new data collection I strongly urge that the concept of wealth cover public and private pension rights, as well as the information on liquid assets, securities, real estate, consumer durables and business holdings which has been covered in past survey. Thank you Mr. Chairman.

REFERENCES

1. Ando, A. and Modigliani, F. "The 'Life Cycle' Hypothesis of Saving: Aggregate Implications and Tests," American Economic Review, March 1963, 53, pp. 55-84: also "Correction," March 1964, 54, pp. 111-12.

2. Barro, Robert J. "Social Security and Private Savings-Evidence from the U.S. Time Series," unpublished manuscript, April 1977.

3. Feldstein, M. "Social Security, Induced Retirement and Aggregate Capital Accumulation," Journal of Political Economy, Sept./Oct. 1974, pp. 905-26.

4. Ferber, R. "Consumer Economics, A Survey," Journal of Economic Literature, Dec. 1973, Vol. XI, No. 4, pp. 1303–42.

5. The Reliability of Consumer Reports of Financial Assets and Debts, University of Illinois Bureau of Economic and Business Research, Studies in Consumer Savings, No. 6, 1966.

6. Friedman, M. A Theory of the Consumption Function. Princeton: National Bureau of Economic Research, 1957.

7. Ghez, G. R. and Becker, G. S. The Allocation of Time and Goods Over the Life Cycle, National Bureau of Economic Research, 1975.

8. Modigliani, F. and Brumberg, R. "Utility Analysis and the Consumption Function: An Interpretation of Cross-Section Data." In Kurihara, K.K., ed. Post-Keynesian Economics. New Brunswick, N.J.: Rutgers University Press, 1954. 9. Munnell, A. H. "The Impact of Social Security on Personal Savings," National Tax Journal, Dec. 1974, pp. 553-67.

10. Projector, D. S. and Weiss, G. S. Survey of Financial Characteristics of Consumers, Washington, D.C.: Federal Reserve Board, 1966.

11. and Knoblauch, T. "Participation of the Aged in Welfare Programs," paper presented to the 15th General Conference of the International Association for Research in Income and Wealth, 1977.

12. Scheuren, F. and Herriot, R. "The Role of the Social Security Number in Matching Administrative and Survey Records," Studies from Interagency Data Linkages, Report No. 4, U.S. Department of HEW, 1975.

[The following is additional information supplied for the record by Ms. Projector:]