When this year's Nobel Laureate, Robert Solow, and Ely Lecturer, Alan Blinder, teamed their impressive talents several years ago to ask, Does Fiscal Policy Matter?, the answer they gave was a resounding yes. Working in a then sophisticated neo-Keynesian IS-LM tradition, Blinder and Solow presented a parsimonious macroeconomic model and some statistical and historical evidence suggesting that changes in the weighted standardized government surplus did indeed substantially affect real economic activity. The profession was pretty much convinced that the permanent income hypothesis (PIH) or life cycle hypothesis (LCH) almost provided a sufficient framework for analyzing consumption and saving and hence the effects of fiscal policy. Curiously, the large Keynesian effect of a tax-induced rise in current disposable income was reconciled with the very small effect predicted by the longerhorizon models with more of a whimper than a bang. Since then, the profession has moved some distance from complete acceptance of the life cycle and permanent income hypotheses and perhaps even further from the traditional consumption function specification in vogue at that time. Indeed, Robert Barro (1974) rekindled the notion that a tax-fordebt swap would have no real effects. An avalanche of analytical and empirical research has sharpened our understanding of the issues involved, the econometric difficulties in estimating the relevant parameters, and even the care necessary in defining what is meant when one asks whether fiscal policy has any real economic effects. Surprisingly, despite numerous caveats, new and' improved data and estimation techniques, and improved perspectives offered by analytical insights not yet prevalent when Blinder and Solow wrote their paper, my conclusion is that their answer is essentially correct: fiscal policy does matter, both for short-run stabilization purposes and for long-run capital accumulation. I believe the preponderance of the evidence strongly supports this view, although the empirical research suggests that the impact of, say, tax cuts on consumption is perhaps only one-third as large as the typical Keynesian estimate of two decades ago, but much larger than the neutrality predicted by Ricardian equivalence or the very small effect predicted by the PIH or LCH.

Much attention has been focused recently on the effects of unions on economic stability, resource allocation and income distribution. Almost always, the discussion begins with the effects of unions on labor earnings or wages.' Yet substantial agreement on the magnitude of the effect of unions on wages or earnings hardly seems close at hand. Among other studies on this subject, it is noted that the classic study by H. Gregg Lewis estimates a union/nonunion wage differential of about 10-15 percent in 1957-58; Leonard Weiss estimates about the same differential as Lewis; and Victor Fuchs, Frank Stafford, Adrian Throop, and Orley Ashenfelter and George Johnson estimate a much larger differential. The question such studies should attempt to answer is whether and how much union membership increases wages facing individuals, holding constant other things such as education, race, sex, age, and occupation. The studies mentioned above are not entirely appropriate to answer this question. For example, some suffer from a potentially severe aggregation bias in examining average wages or earnings and the percentage of the labor force unionized and/or fail to disaggregate by race and sex. Those that attempt to examine opportunities facing individuals are forced to employ data on earnings rather than wages and thereby build (at least partially) voluntary labor supply and demand decisions into their estimates. The purpose of this paper is to present new evidence on the relative wages of union and nonunion workers by applying recent advances in the hedonic method of price measurement to a new and rich source of data on individual workers. In Section I, an equation relating wages to personal characteristics is developed which focuses on union membership and its interaction with race, sex, occupation, and geographical area. The equation extends work in this area by Robert Hall. In Section II, a brief discussion of the data is presented together with empirical estimates of the union/nonunion wage differential. Formal tests are made of some interesting hypotheses about the pattern of the relative wages of union and nonunion workers by race, sex, occupation, and geographical area. The results are in much closer accord with the estimates of Lewis and Weiss than those of Fuchs, Stafford, and Throop. In Section III, some concluding remarks are offered, including some observations on the limitations of this type of study.

the great inventions of the 20th century. He was right. It is difficult to imagine modern economics and even public discourse on the economy without them. The NIPAs (and related accounts) provide the basic set of estimates on a wide range of economic variables of interest to economists, citizens, policy makers, firms, investors, workers and consumers. They enforce important economic and statistical properties and reveal many of the most important features of the evolution of the economy. In short, it would be difficult to imagine where our understanding of recent economic events and economic history would be without the NIPAs. Even a short list of some of the major improvements of recent decades, of important historical changes, is impressive. That list includes: 1. The development and implementation of chained Fisher ideal indexes – with their superlative properties (W. Erwin Diewert 1976). 2. The highlighting of the difference between government consumption and investment; 3. The reclassification of software as investment; 4. Computer price hedonics. Add to these an array of improvements in source data and the changes are impressive indeed1. However, the economy evolves very rapidly, so our statisticians are constantly playing catch-up. Thus, it is potentially quite important when a major new architecture (NA) is developed and implemented for such a

The national income accounts, together with the source data which they use, form the core of our economic measurement system. The development of the concepts and measurements of national income are among the most important achievements of modern

ing from all the difficulties of forecasting, let us focus our attention, despite, perhaps, its looser ties to the future performance of the economy than surprises anticipated future deficits, on what would be involved on a conceptual and accounting basis in estimating the federal government deficit for a recent year. I leave aside all issues about the extent to which we need to worry about measuring the cumulative state and local government surplus, given the wide range of accounting procedures and features employed therein, and refer to R. G. Penner for those who desire further discussions concerning the budgetary process and forecasting. As a brief reminder, however, agency and OMB preliminary budget analyses are conducted approximately at least one year prior to the beginning of the fiscal year under consideration; the president issues his budget message in January; congressional review occurs throughout the balance of the fiscal year under consideration (and sometimes continues into it); the fiscal year commences on October 1 and runs through the following September 30; and auditing of income and expenditures proceeds thereafter. It is not surprising, therefore, that forecasting spending, revenues, and thus the deficit, is a nontrivial statistical matter subject to all sorts of problems. My point is simply that we do not even know what the budget deficit, in any reasonable sense, was in any recent previous year; and that if we cannot know this, it may mean that our econometric analyses of the impact of the budgetary deficit are based on analytically inappropriate concepts, or substantial measurement error; and that a large amount of disagreement concerning the extent to which consumers anticipate future taxes, the Fed monetizes the deficit, and a bevy of other issues at the core of macroeconomic analysis and policy are not being analysed or tested in an appropriate manner. Let us now turn to a discussion of some of the vagaries in estimating the budget deficit for a recent period, for example, last year. Conceptually, if all accounting was done appropriately, the budget deficit D would merely be the difference between the government's income Yg, and outlays 0, as indicated in equation (1), (1) Dt t+ IYgt,t+I-Ogt,t+i where the subscript t denotes the beginning of the period, and t + 1, the end of the period, however long the period under consideration. Problems occur because it is difficult to measure both Y and 0. These problems include classic problems in accrual versus realization accounting; inflation accounting; This content downloaded from 40.77.167.90 on Sat, 06 Aug 2016 05:17:22 UTC All use subject to http://about.jstor.org/terms 298 A EA PAPERS AND PROCEEDINGS MA Y 1982 developing appropriate price indices for various categories; valuing a variety of types of services and goods which are not freely traded on well-defined markets, as usual, with public goods types of issues; etc. As with the private sector national income accounts, we have substantial problems in measuring capital gains and losses. Unfortunately, these swings in the change of the federal government's outstanding obligations are quite large in some periods and can dominate or at least equal the regular deficit figures. Further, a large number of government operations during a period, say, between t and t + 1, are explicit or implicit contingent promises to deliver cash or commodities under certain conditions, for example, loan guarantees, insurance, social insurance benefit payments, etc. While I shall return to these points in more detail below, how should one treat accruing implicit forecasted Social Security benefit payments? Or new loan guarantee commitments combined with possible changes in the probabilities that previously issued loan guarantees will have to be paid? Let us return to the original budget deficit estimate and work our way toward a more comprehensive measureincluding large estimated standard errors of estimate-of the federal government deficit, indicating some of the land mines along the way in terms of difficult conceptual issues, data demands, measurement problems, etc. Last year the federal government ran an estimated budgetary deficit of $58 billion. The exact figure will not be known for quite some time, until after auditing occurs. This is the most widely quoted figure, and when discussions of the budget deficit and/or projected deficits are made, they are with reference to the budget deficits. It is important to note that this figure must not only be adjusted for a variety of reasons, but it is not at all comprehensive. First of all, the federal government does not keep a separate capital account; the official budget figures exclude certain items pursuant to the mid1970's budget reform process, for example, so-called off-budget activities and the activities of federally sponsored agencies. Many of these activities involve subsidized lending or loan guarantees. In any event the Treasury must borrow to finance the deficits of these entities. I will return to one or two of these items below. A. Tangible Capital and Investment If we maintained a separate and conceptually correct current and capital account system, the deficit on current account would be the true deficit, although the government would be in a position to change the composition of asset purchases and sales in order to change the net surplus. The basic point is that for capital items, any excess of expenditures over receipts on capital account does not change the net asset position of the government since the new debt is matched by a new government asset. In the capital account, the purchase of assets would be included as expenditures and sales of government assets, and funds transferred from the current account to cover depreciation receipts would be counted as receipts. Depreciation charges on government assets would appear as an expenditure in the current