Welfare has several problems that need this information. The Internal Revenue Service obviously needs it. Congress in making all kinds of other policy, including tariff policy, may need it. The Federal Reserve Board needs it in dealing with monetary policy; and there is no particular one place where this thing focuses; and perhaps the reason why we haven't got any decent data is because it isn't anybody's sole responsibility. Mr. FRASER. But none of you have any reservations about the Government getting into the business of trying to learn more about wealth distribution? You see it as a thoroughly appropriate governmental responsibility? Mr. MORGAN. Surely. Mr. SMITH. Yes. MS. PROJECTOR. Yes. Mr. FRASER. We are coming close to the time when the House will be going into session. I want to express my appreciation to the three of you for your testimony today. This is a subject that may benefit from exploration if there is a way to encourage the executive branch to move ahead on this. This problem that you have just adverted to a moment ago about who has the responsibility, may be our biggest problem. Mr. MORGAN. It is possible the National Science Foundation is the proper locus for a thing like that. Whatever we might think about. Rand and its future, there was a vision some years ago that there were problems exactly like this that weren't really clearly located in some governmental agency that did have policy relevance that needed funding, and the National Science Foundation was a proper place through which to channel the funds for that kind of work. Mr. SMITH. I would join in that view, that the National Science Foundation is probably an excellent place to vest the central organization of such an effort, and those funds could then flow back to governmental agencies as well as to nongovernmental agencies for purposes of carrying out what is viewed to be the important aspects of understanding distribution processes by which wealth moves. Mr. MORGAN. And they have easy mechanisms for involving the scholarly community and funding also a panoply of smaller studies that supplement the whole field. Mr. FRASER. Good. Unless there is any other comment that anyone wants to make for the record, we will adjourn the hearing at this time. Thank you very much. [The following material was submitted for the record by James N. Morgan:] ALTERNATIVE DESIGNS FOR STUDIES OF THE DISTRIBUTION OF HOUSEHOLD WEALTH Collecting data on the distribution of wealth is difficult and expensive, so some prior consideration of the benefit-cost ratios of alternative designs is in order. These alternative designs result in part from differences in the purposes foreseen for the data. For measuring and monitoring changes in the distribution of wealth, in order to assess the equity of the system and of its recent functioning, one would need large and repeated (or even reinterviewed) samples, and high precisions in the dollar amounts. One would also have to oversample very heavily at the top of the wealth distribution, a requirement particularly expensive in money costs and delicate arrangements. Furthermore, there are conceptual difficulties because of changes in the way wealth is owned and the proliferation of new institutional forms which hold or convey rights without the legal reporting of ownership. While distributional equity is an important issue, we do not think it is sufficiently important to justify such a focus, or such an investment. For measuring short-run saving as a process of asset accumulation, one would need a substantial sample, almost as heavily concentrated at the top, plus additional questions about savings flows (and/or reinterviews to get two asset measurements). Again expenses are high and there are conceptual difficulties since there are wealth increments that would not show up in annual income-unrealized capital gains, accruals of rights in life insurance reserves, annuities, retirement funds. Some attention to asset accumulation (saving if you will) over middle-range periods may well be worth it, and would require, for adequate precision, reinterviewing people at five-year intervals. Costs would be higher than without the focus on saving but the informational benefits would be much greater. For most other purposes, including tax policies, Social Security policies, monetary policies, and understanding our economic system better, a much less costly design is appropriate. It would utilize a smaller sample, and devote less time and money to striving for precise dollar amounts. It would oversample the larger wealth holders, but perhaps by sampling households in proportion to shares of income rather than of wealth, which is easier. Most important of all, such a study would seek information about how present stocks of wealth were accumulated, why they are invested as they are, and what their owners expect or plan to do with them. The information sources, attitudes, and plans of wealth holders are far more important inputs into policymaking than precise details of their portfolios. The difference in emphasis is well illustrated in the comparison of two studies done in the early sixties, our Economic Behavior of the Affluent and the Federal Reserve Board's Survey of Financial Characteristics of Consumers. The first secured broad bracket estimates of wealth components, then asked about sources, policies about investment, and future plans, purposely narrowing down late in the interview on the effects of taxes and the extent of tax avoidance. The Federal Reserve Board Study secured much more detail about portfolios, and related the amounts of each type of assets to total family wealth or income. A reinterview a year later provided estimates of oneyear saving. But there is a great deal of random variance in one-year savings. entirely apart from conceptual problems of excluded components like unrealized capital gains, and accrued pension and other rights. There are also definitional problems with components of wealth such as which ones should have the debts subtracted from them. And finally, it is very difficult to infer either the accumulation process or the owners' purposes from the dollar facts alone. It seems more appropriate and more economical to focus on why as well as what, and in addition we think that respondents would find a study on the type of Economic Behavior of the Affluent less burdensome and also less threatening to their privacy. Indeed, given the present distrust of the government and particularly of government guarantees of privacy and confidentiality, it might be more appropriate to entrust such studies to non-government organizations, so that the results would not be a government data file subject to the Freedom of Information Act, nor the respondents liable to reinteriew as part of an "audit” by the Government Accounting Office. Guarantees of confidentiality like those the National Institutes of Health can provide are needed. The optimal sample design for a study of this kind calls for such a disproportionate selection of wealthier families that it would be almost essential to have samples drawn from Internal Revenue Service files, properly safeguarded of course. There is a range of possibilities here. One extreme would be a sample selected in proportion to aggregate dollars of income represented without revealing the income or wealth of any unit in the sample. (The sampling unit would be the tax return rather than the household, of course.) At the other extreme, the tax return information on total average income from each type of wealth, would be merged with the interview information both to assure proper weighting for differential sampling and response rates, and as part of the information for substantive analysis. A middle compromise would allow departing from restrictive self-weighting samples without the imprecisions in inferring sampling fractions, by releasing with the sample an indication as to which of three broad income classes the unit was originally selected from. There is precedent for at least the most masked arrangement of an IRS sample, since it was used in both of the early wealth studies, and in a recent study we conducted for the Commission on Private Philanthropy and Public Needs. Finally, although we prefer, at least for the first set of studies, to concentrate on the third alternative-focusing on approximate dollar amounts combined with a rich set of accompanying information about the past, present, and future of those assets-there is need for a range of studies, including still cheaper attempts to get something from already available statistics, such as the asset income reported to Internal Revenue Service. There should be a planned remeasurement both of the distribution of wealth and of what people report as to its sources, reasons for its present investment, and expected future disposition. Once every five years should be enough for such monitoring. There is also a need for continual methodological improvements and tests, not so much the traditional comparisons with aggregate data, nor even validity studies using records, but experiments and research studies on interviewing as a process. The Survey Research Center has a current project trying out a system of asking for access to routine household financial records, as a relatively errorfree source. After some additional testing and development of methods, a major study of saving and asset change-over five years rather than a single year-might well be worth its great cost. Even if such a single massive study is undertaken, there should be some smaller companion studies by other organizations both to test the stability of the main findings under variation in method, and to relate them to other material more easily collected in smaller studies, such as the attitudes and purposes, information and insights of wealth holders. There are problems with locating responsibility for such studies in any of the relevant operating agencies like Internal Revenue, Social Security Administration, National Income Division of the Commerce Department, or the Office of Management and Budget. The Federal Reserve Board should also have a vital interest in the results, but has not taken any initiative and is technically outside the government. The basic scientific importance of such studies and the need for some independence from immediate pressures and focuses, and the far superior mechanism for allocating research funds all suggest the National Science Foundation at the responsible agency. Some cost sharing by operating agencies of government might be involved, of course. Actually, the RANN section of NSF already has in its statement of focus a section on equity and the effects of government programs on income distribution. Surely studies of wealth distribution and its components and how they have been accumulated would be a logical part of such an area of study. An important reason for suggesting the National Science Foundation is a belief that their peer review and research grant process is far superior to the Request for Proposal system now required of government agencies by OMB. That process requires advance specification in great detail of research designs and focuses on the lowest bidder, often in areas where the costs are difficult to estimate or can be cut in ways that injure the research without violating the specifications. If Congress has serious concerns about efficiency, competition, and financial control, there are better ways to achieve it. They could, for example, suggest that NSF give three competitive grants for three smaller parallel studies. (If IRS drew a sample, they could easily cut it into three independent subsamples.) Three organizations could be selected on the basis of what they proposed to do with a given budget, and the grants could require that each of them had to evaluate the findings of the other two in comparison with their own findings. The result would be better research, built-in balanced evaluation free from conflict of interest, and far more freedom for the researchers to use their ingenuity doing good research rather than meeting rigid and often unnecessary requirements. In answer to the question about expected cost, one can always tailor survey research to available funds, within limits, by altering the sample size or the amount of content. There are, however, minimum sizes to projects if one wishes to get a return on the investment in design and analysis. If samples could be drawn from official records, I still believe a good national study could be done for less than a million dollars. The ideal way to proceed would be to fix the amounts and see what competing research groups could offer to do. It is a far easier scientific exercise to design an optimal study within a reasonable but fixed budget, than compromise good design in an attempt to meet arbitrary requirements and still make the lowest bid. REFERENCES Barlow, Robin; Brazer, Harvey E.; and Morgan, James N. Economic Behavior of the Affluent. Washington, D.C.: The Brookings Institution, 1966. Morgan, James N.; Dye, Richard F.; and Hybels, Judith H. "Results from Two National Studies of Philanthropic Activity." In Research Papers, Volume I, pp 157-223. Sponsored by The Commission on Private Philanthropy and Public Needs. Projector, Dorothy S. and Weiss, Gertrude S. Survey of Financial Characteristics of Consumers, Board of Governors of the Federal Reserve System, Washington D.C., 1966. Projector, Dorothy S. Survey of Changes in Family Finances. Board of Governors of the Federal Reserve System, Washington D.C., 1968. [Whereupon, at 11:55 a.m., the hearing was adjourned until Thursday, September 29, at 10:30 a.m.] DATA ON DISTRIBUTION OF WEALTH IN THE UNITED STATES THURSDAY, SEPTEMBER 29, 1977 TASK FORCE ON DISTRIBUTIVE IMPACTS The task force met, pursuant to notice, at 10:50 a.m. in room 210, Cannon House Office Building, Hon. Donald M. Fraser, chairman of the task force, presiding. Mr. FRASER. This morning we are holding the second of a two-part hearing concerned with data on the distribution of wealth in the United States. Our focus on Monday was with the importance of wealth data and with what is known at present about the distribution of wealth in the United States. At today's hearing we will address three main questions: One: What data will be collected by present or already planned Federal efforts? Two: What more is needed? What alternative approaches might be taken? Three: What are the problems, costs, and benefits of these alternatives? Ideally, we will leave today's hearing knowing not only the nature of the problem, but knowing what should be done about it. Our witnesses this morning are especially well qualified to give us the answers we seek. They are: Richard Ruggles, Department of Economics, Yale University; Joseph Duncan, Deputy Associate Director for Statistical Policy, Office of Management and Budget; Vito Natrella, Director, Statistics Division, Internal Revenue Service; and Nelson McClung, Assistant Director, Office of Tax Analysis, Department of the Treasury. Our first witness will be Mr. Ruggles. We will hold off questions until all of the witnesses have completed their prepared statements. When we do ask questions, I would encourage the witnesses to feel free to respond to statements by the other witnesses. Mr. Ruggles. STATEMENT OF RICHARD RUGGLES, DEPARTMENT OF ECONOMICS, YALE UNIVERSITY Mr. RUGGLES. Thank you very much. In inviting me to testify on the subject of Federal data relating to the distribution of wealth, I assume that the task force is looking for (45) 98-815 O-78-4 |