An Overview

This chapter considers some of the broad issues confronting the continued development of economic and social statistics and some initiatives for dealing with these issues in the coming decade. The first section approaches the subject of statistical development as a problem of information processing in the broadest sense. Following this, a few of the major Framework recommendations are highlighted to give an integrated overview. Then a brief action strategy is presented. A listing of major recommendations from the chapters on subject matter areas and crosscutting issues is presented at the conclusion of this chapter.

Statistics as an Information Process

Viewed from this perspective, the basic objective of the Federal Statistical System is to establish and maintain a process whereby a flow of useful and timely statistical information can be sustained.' This flow of information must provide objective and comprehensive data of acceptable precision relating to current conditions and emerging developments in each of the major areas of social concern-the economy and the areas of health, education, welfare, employment, income, public expenditures, public safety, housing, and the like. To be pertinent, these data must satisfy a dual requirement: they must reflect as fully as possible an underlying model of the real world circumstances affecting the phenomena in question and they must also address the concerns and aspirations of policymakers and the public. Two points emerge from this perspective: (1) Any information process must be sufficiently flexible to respond effectively to changing needs and interests of the users of the information; and (2) It must be oriented toward developing a monitoring capability with respect to trends and changes occurring in the

'On this point and on the more general problem addressed in this paper, see Edgar S. Dunn, Jr., Social Information Processing and Statistical Systems-Change and Reform (New York: John Wiley & Sons, Inc., 1974), pp. 19ff. For a broad philosophical perspective, see C. West Churchman, The Design of Inquiring Systems: Basic Concepts of Systems and Organization (New York: Basic Books, Inc., 1971).

different areas of interest, rather than focusing exclusively on measures of status at particular points in time.

In her recent article on income distribution (which was discussed in the Framework chapter on income, wealth, and consumption), Alice Rivlin listed four questions which policymakers (and the general public) might reasonably be expected to ask of economists:2

(1) What is happening?

(2)

(3)

Why are things as they are?
What could be better?

(4) What can we do about it?

These four questions, or close approximations, are fairly representative of the practical interests of both policymakers and the public and private groups and individuals who turn to statistics for assistance in making better judgments and arriving at sounder decisions. Thus, following Rivlin, one may reasonably expect such questions to arise with respect to every area of economic and social concern. From the users' point of view, the chief value of any statistical information system lies in the extent to which it can assist in developing reasonably reliable answers to these questions.

For any area of concern, an answer to the question "What is happening?" calls for descriptive data which depict both the current situation and underlying trends in its development in the amount of detail required to fit the specifications accompanying the question. It is evident that the bulk of the statistical data that are routinely collected are intended primarily to answer this type of question.

To deal with the second question, "Why are things as they are?", it is necessary to devise and utilize analytic measures which can provide some clues as to

'Alice M. Rivlin, "Income Distribution-Can Economists Help?" American Economic Review, Vol. LXV, Number 2, (May 1975). Also see Alice M. Rivlin, "Measuring Performance in Education," in Milton Moss (ed.), The Measurement of Economic and Social Performance (New York: National Bureau of Economic Research, 1973), pp. 411-437.

the factors which account, at least in part, for the conditions and developments of interest. In the absence of such measures, it is of course possible to obtain some insight as to "what is going on" by examining and trying to interpret descriptive information, especially when that information is available in some detail for comparative purposes. If the "why" question is to be answered in quantitative terms, it becomes necessary to examine the interactions between the variable of interest and some selection of independent variables that are thought to influence it in some way. Statistical measures, of course, do not identify causation; they help identify important associations. Descriptive statistics cannot meet this need unless they are provided in a form that permits this level of analysis.

The further question, "What could be better?" entails considerations that are outside the purview of statistical information processes. Questions of this type call for normative answers, reflecting our notions of good and bad, better or worse. In formulating such answers, we are required to move beyond factual descriptions and analysis into the realm of social goals, values, and aspirations. If, in doing so, we seek practical guidance in decisionmaking, we cannot abandon what we know about our actual condition. On the contrary, it is only by examining our current situation that we can detect specific directions of change whereby that situation might be made "better." This descriptive input

should be based on sound statistical information. For example, given the goal of improving the health status of the population, information on the prevalence of certain disabling conditions among different population groups affords some basis for determining where further medical resources or preventive measures could be applied in order to bring the higher prevalence rates of some groups down toward the lower levels manifested among other groups. In short, observed differentials and trends in these differentials, when they relate to normatively significant conditions and processes, provide useful clues with respect to "What could be better."

Finally, a response to the question "What can we do about it?" obviously requires the exercise of both analytic skill and normative judgment. In dealing with this kind of question, the role of a statistical information process is twofold: (1) to describe the existing condition from which movement in any desired direction must begin, and (2) to delineate (ideally, in cost-benefit terms) possible alternative paths from current reality toward the goals that are stipulated.

In summary, the statistical information process can be relevant to every phase of deliberation for a particular area of concern, beginning with descriptions of current conditions and their past evolution and ending with evidence relating to the costs, benefits, or trade-offs associated with alternative courses of remedial action. It must be stressed, however, that only the first two of Rivlin's four questions can be dealt with in terms of the conventional activities of statistical data collection and analysis. The latter two questions arise at the juncture between statistical information processes and political decisionmaking. Hence their treatment demands an information system which is also capable of illuminating the gaps between our current conditions and our social aspirations. In essence, good statistics with easy access are a cornerstone of the democratic process of defining alternatives and choosing the preferred direction.

Priorities and Constraints

The current status of the Federal Statistical System may be seen as an outgrowth of the substantial expansion which has occurred over a period of five decades of development which began during the depression of the 1930's and which was accelerated by the passage of the Employment Act of 1946 and the accompanying establishment of the Council of Economic Advisers and the Joint Economic Comfrom the President to the Congress and the general mittee in the Congress. By requiring annual reports public on the state of the economy, this Act spurred the development of agreed-upon economic indicators which could reflect current trends in the performance of the Nation's economy and signal emerging imbalances or problems before they assumed crisis proportions.'

The accumulation of experience in both the development and use of these economic indicators has given rise to substantial improvements in the underlying data base and to growing sophistication among technicians and policymakers alike with respect to the significance of different sets of economic data. Unfortunately, our system of social (i.e., noneconomic) information has not enjoyed a parallel course of development. It is true that the "great society" programs initiated in the early 1960's were accompanied by considerable demand for new and differerent kinds of information-information

'Philip M. Hauser, Social Statistics in Use (New York: Russell Sage Foundation, 1975), pp. 349ff. For a detailed account of the historical development of U.S. government statistics, see Joseph W. Duncan and William C. Shelton, Revolution in United States Government Statistics, 1926-76 (Washington, D.C.: U.S. Government Printing Office, 1978).

that would enable program directors to gauge the impact of their executive directives as well as information that would identify areas of critical need; however, this goal has not been addressed in a coherent fashion. The recent calls for improved public administration and greater accountability in the expenditure of public funds are, in essence, demands for a more explicit and regularized practice of evaluating the outcomes of governmental programs. Although such pressures are not new, they have clearly become more visible. A statistical information process which measures both the immediate outputs or products of government programs, as well as the longer range consequences of such initiatives, could contribute significantly to meeting the demands for more adequate and informative evaluations of public programs.

The crises and disturbances of the late 1960's offered pointed reminders to many policymakers that they could no longer rely exclusively on economic indicators in assessing the condition of the society or in formulating socioeconomic policies. Despite great efforts we still lack, in social statistics, the social accounting framework which would be equivalent to the national income accounting framework which is so useful in economic statistics. Many analysts maintain that the lack of a common denominator for measuring social programs means that such an accounting framework will never be achieved. Others believe that a structure can be developed within which important interactions can be identified and evaluated.

Further impetus toward the improvement of both social and economic information processes stems from the still more recent proliferation of complex problems such as urban decay, crime and delinquency, inter-group conflicts and the equally complex "environmental" problems of pollution and growing concern with limited natural resources. It has become increasingly evident that none of these problem areas can be understood or dealt with adequately if they are considered purely "economic❞ or purely "technological" or purely "social" in nature. It follows, therefore, that the statistical information process that is called for in coping with these areas of concern must itself be broadly interdisciplinary.

'This realization was an important motivation for the establishment of the social indicators development program within the Organization for Economic Cooperation and Development. See David E. Christian, "Social Indicators: The OECD Experience," (Paris: OECD, June 1974). For an account of progress achieved thus far in this effort, see OECD, Measuring Social Well-Being-A Progress Report on the Development of Social Indicators (Paris: OECD, 1976).

In his penetrating study of the existing forms of social information (wherein the term "social" relates to economic data as well), Edgar S. Dunn, Jr. identifies six major types of inadequacy among these bodies of data:5

1. There is a relative paucity of data adequate to represent external operating environments, as compared with the amount of (administrative) data relating to internal operations. In other words, agencies and organizations, both public and private, tend to develop more complete bodies of information and more efficient flows of information with respect to their internal organization and administration than the environment in which they must operate.

2. Data relating to developmental functions (i.e., learning or adaptive processes) are deficient relative to data reflecting performance. Thus we know more about what an agency does than about what kinds and sources of information it relies upon both in deciding what to do and in gauging the effectiveness of what it has done.

3. Sources of nonrecurrent data are deficient relative to sources of recurrent data. Periodic observations of selected phenomena are of course essential in order to analyze trends and changes over time, but such measures will often fail to provide advance warnings of emerging difficulties. Many of the "shocks" (such as the oil embargo) which account for major turning points in our standard time series of observations are themselves the culmination of a substantial history of development, but that history has remained statistically invisible because the underlying processes have taken place outside our framework of measurement.

4. There is a paucity of data elements that can be employed in constructing a variety of statistical entities, relative to aggregate data provided in fixed format. Despite the great strides that have been made in recent years, the provision of "micro" data elements that could be combined to satisfy a variety of research needs has not kept pace with the technological advances which facilitate the multipurpose use of such data elements.

'Edgar S. Dunn, Jr., op. cit., pp.158ff. Still pertinent in this regard is Oskar Morgenstern, On the Accuracy of Economic Observations (Princeton, N.J.: Princeton University Press, second edition, 1963).