Knowledge that Transforms

Q UALITY of service has been neglected in empirical studies of air travel. Travel demand depends on travel time as well as the usual price and activity variables. Travel time includes not only average en route time, but delay. Following Douglas-Miller (1974a,b), frequency delay is the gap between one's desired and the nearest offered departure time, while stochastic delay is time lost due to the nearest offered departure being unavailable. Their sum, schedule delay, is the major element in air service quality. Despite its theoretical importance, no empirical study of air travel demand has incorporated it.' This one does.2 A second novel feature of this study is that it models demand on a service segment rather than passenger origin-destination demand. From the carriers' point of view, this is the relevant market demand. Much air travel is done in several segments rather than non-stop. Ideally, segment demand should be modelled based on passenger origin-destination demand, with service segment demand built up from all the possible routings using the segment under study. In practice this is impossible, since the origin-destination passenger data contain no information on routings. This is why previous demand studies, based on origin-destination data, have been unable to incorporate consideration of delay. The third novel feature of this study is that it is based on time series estimation embedded in a simultaneous equations model of service segment markets. Elsewhere (Anderson-Kraus, 1980), we have reported on the supply side of the model. The time chosen is 1973-76 (48 monthly observations), when fares were set according to the CAB formula, and flight scheduling rivalry dominated carrier competition. The results below indicate that segment demand can be successfully modelled with these methods. It is notable that price elasticities in excess of one in absolute value are found even in a number of business markets. It is dangerous to draw general conclusions from a specific instance of success, but the results do support further attempts to model demand incorporating delaybased quality of service. Many markets are characterized by significant waiting costs and competition over the reduction of these costs. Where feasible, the results below indicate that even crude measures of delay have a significant payoff. Part I lays out the model of demand and service quality and embeds it in the model of market equilibrium. Part II presents the results.

An index of vertical integration is constructed as a tool to examine how the average level of vertical integration has changed over time. The hypothesis that if vertical integration can lower costs or reduce risks, more firms will integrate vertically is subjected to an empirical test using the Vertical Industry Connection index. A randem sample of firms is selected from the COMPUSAT data base. The results support the contention that vertical integration has increased in contrast with Laffer's findings. 17 references, 10 figures, 3 tables.

D O the poor substitute among commodities more flexibly than the rich when relative prices change, even when compensated for the income effects of the price changes? Is it necessary to add a third dimension to the Slutsky substitution matrix to identify household income? In short, do the terms of the Slutsky elasticity matrix curve smoothly from poor households to rich households in such a way that equation (1) is meaningful?

dom Experience, Economic Journal 86 (Mar. 1974a), 32-55. Inequality Measures, Prices and Household Consumption, Review of Economic Studies 41 (Oct. 1974b), 493-504. Nordhaus, William D., and John B. Shoven, A Technique for Analyzing and Decomposing Inflation, paper presented at the National Bureau of Economic Research Conference on Price Behavior, Bethesda, Maryland, November 21-23, 1974. Phlips, Louis, A Dynamic Version of the Linear Expenditure Model, this REVIEW 54 (Nov. 1972), 450-458. Polenske, Karen R., State Estimates of the Gross National Product 1947, 1958, 1963 (Lexington, MA: Lexington Books, 1972). Prais, S. J., and Hendrik S. Houthakker, Analysis of Family Budgets (Cambridge: Cambridge University Press, 1955). Ramsey, James B., R. Rasche, and B. Allen, Analysis of the Private and Commercial Demand for Gasoline, this REVIEW 57 (Nov. 1975), 502-507. Stern, Irving, Industry Effects of Government Expenditures: An Survey of Current Business 55 (May 1975), 9-23. Stigler, George S., and Gary S. Becker, De Gaustibus Non Est Disputandum, American Economic Review 67 (Mar. 1977), 76-90. Sydenstricker, E., and W. I. King, The Measurement of the Relative Economic Status of Families, Journal of the American Statistical Association 17 (1921), 842-857. U.S. Bureau of Economic Analysis, The Structure of the U.S. Economy: 1967, Survey of Current Business 54 (Feb. 1974), 24-56. , Definitions and Conventions of the 1967 Study (Washington, D.C.: U.S. Government Printing Office, 1974). U.S. Bureau of Labor Statistics, Survey of Consumer Expenditures, 1960-61 (Washington, D.C.: U.S. Government Printing Office, 1966). , Consumer Expenditures and Income: Survey Guidelines, Bulletin No. 1686 (1971). U.S. Department of Commerce, Definitions and Conventions of the 1967 Study, Bureau of Economic Analysis, Interindustry Economics Division (BE-51). Vaccara, Beatrice N., Input-Output Coefficients for the United States, in A. P. Carter and A. Brody (eds.), Applications of Analysis (Amsterdam: North-Holland Publishing Company, 1970), 238-260.

ONE of the most important developments in government during the last two decades has been the growth of public employee unions. Because personnel expenditures comprise a substantial portion of the cost of providing public services, there has been considerable concern that unionization has severely exacerbated budgetary problems of governments, especially local governments. Consequently, a large literature on the wage effects of public employee unionism has developed in recent years. 1 In general, these studies have found rather modest union wage impacts in the public sector, with unions raising the wages of municipal employees by about 5%, on average. There are several reasons, however, why these estimates may be misleading. First, much public sector union wage impact research has focused on teachers and the demand for teacher services is considerably more elastic than the demand for other public services such as police protection, fire protection and medical care. Second, previous studies have treated public employee unionism as an exogenous variable, and studies of private sector unionization have shown that the estimated wage impacts of unions change substantially when unionism is treated as an endogenous variable.2 Finally, the earlier studies have not considered how public employee unionism has affected fringe benefits, which represent a substantial and growing component of governmental compensation. This paper attempts to deal with each of these problems. First, we focus on the effects of unionism on the wages of municipal police officers.3 Because police services are commonly thought to be among the most essential of all public services, organized police should be in an especially powerful position when bargaining with local governments. Second, we estimate a system of equations in which wages and unionism are simultaneously determined and compare these results to the estimated wage impacts from a single equation model. Finally, some estimates of the effects of unionism on police fringe benefits are presented.

THE decline in the rate of productivity growth that has occurred in the private nonfarm economy during the last decade has been particularly acute in the construction industry. Between 1950 and 1968, labor productivity in the private nonfarm sector rose at an annual rate of 2.4%; between 1968 and 1978 the yearly rise was only 1.2%. In the construction industry productivity rose by about 2.4% annually between 1950 and 1968, but has declined by 2.8% annually since then. This paper addresses seven possible explanations for these trends in construction industry productivity. These include (a) the measurement of real output, (b) shifts in the composition of construction industry output, (c) changes in capital per worker, (d) demographic changes in the workforce, (e) economies of scale, (f) regional shifts, and (g) changes in work rules or practices. The results are disconcerting. Only a small portion of the deterioration in productivity is explained by these factors. In all of the presentations below, rates of growth are measured from 1950 to 1968 and from 1968 to 1978, except where precluded by the unavailability of data. Labor productivity reached a peak in 1968 and thus this date serves as a convenient point to split the historical period. Prior to 1968 there were only a few years, depending on the measure used, in which productivity declined; after 1968 there were only one or two years in which productivity increased. The problem of declining productivity cannot be isolated to a particular year or two, but rather has persisted almost without exception for the last decade. Measures of growth in labor productivity are displayed in table 1. The measure of output in each case is real value added. While there are differences between these rates of productivity growth, they are relatively minor. This suggests that productivity differences due to (a) the distinction between hours worked and hours paid, (b) the distinction between employees and all persons, which also includes proprietors and unpaid family workers, and (c) the distinction between employment and hours, reflecting changes in the number of hours worked per employee, are not particularly important in explaining the decline in construction productivity in the last decade. In this sense the problem is not how to measure productivity.

ECONOMISTS and statisticians who construct estimates of total factor productivity or who estimate production functions or systems of consumer demand functions are often forced to aggregate subsets of their data. In order to perform this aggregation, an index number formula is generally used. A price index P(pO, pl, x?, xI) is defined to be a function P of the prices of the N commodities to be aggregated in periods 0 and 1,p?-(pll, . . . , PNO) and pl (pl,.'.. PN'), respectively, and of the corresponding quantities utilized during periods 0 and 1, x? (xi?, . . .,XNO) andX1 _ (xi', . . .,XN1), respectively. A quantity index Q(p0, pl, x?, xl) is defined to be another function Q of the price and quantity vectors for the two periods. Generally, we assume that P and Q satisfy Fisher's (1922) weak factor reversal test: