increased more sharply than coach over the past three years, with the result that demand for first-class service has shrunken sharply, while at the same time the airlines have been reducing the size of their first-class compartments.137 This supports the arguments made earlier that if (or when) first-class fares reflect full long-run first-class costs, the demand for first-class service will shrink to nearly zero.

Meanwhile, reduction of the cross-subsidy between first class and coach has narrowed the margin between actual and optimal coach fares. Thus, as previously stated, the markup of actual interstate coach fare over hypothetical unregulated fare on twenty-eight top interstate routes declined from 20-90 per cent in 1968 to 30-56 per cent in 1974. Nevertheless, a markup of 30-56 per cent is not a trivial one, so that the reduction of this cross-subsidy which occurred between regu1968 and 1974 did not come near eliminating the welfare cost of lation to coach passengers. Furthermore, recent work by Pustay has shown that even if the cross-subsidy were eliminated completely, there would still be considerable welfare cost from nonoptimal fare structure due to CAB regulation. It is to his results which we now turn.

Costs of a Non-Optimal Fare Structure by Route Type Pustay's analysis of the costs of CAB regulation is more accurate than those of previously mentioned studies, in that he calculates optimal fares and service qualities by route type (based on the Douglas-Miller schedule delay function), compares actual and optimal fares and service qualities on each route type, and adds up welfare losses over all routes. Thus, he estimates the cost of a nonoptimal fare structure over a wide range of routes, as well as the costs of a nonoptimal fare level. Unfortunately, Pustay's welfare loss estimates are downward-biased relative to those presented in the subsection above, for in these calculations he has ignored the costs of an inefficient differential between first-class and coach service, and has also used the CAB's long-run cost model, which probably overstates the operating costs (at a given load factor) which a carrier of the intrastate type could achieve. 138 (Pustay does acknowledge, indeed, asserts strongly, that CAB regulation extracts higher costs than his estimates because of these two factors-he devotes two chapters of his thesis to them; but he does not quantify these costs). Nevertheless, Pustay estimates that as of 1969, CAB regulation extracted a welfare cost of $660 million at a $5 per hour value of schedule delay time, and $490 million with a $10 value.

Ideally, it would be desirable to integrate the cost model used for welfare cost calculations earlier in this section with route-by-route demand data, to get a more accurate estimate of the welfare cost of regulation. However, sufficient data and resources were not available at the time of the writing to make this possible.

THE IMPACT OF CAB REGULATION ON PRODUCERS

It is agreed by most that the airline industry operates subject to constant returns to scale. Furthermore, it earns normal profits, and appears that it would earn normal profits in a deregulated world, as

137 Ibid., pp. 25-26.

138 The Effects of Regulation on Resource Allocation in the Domestic Trunk Airline Industry, ch. 4.

well. It is tempting to argue, then, that "in the long run" there would be no loss to producers in the airline industry from deregulation. But this is not a satisfying conclusion, for two reasons: first, there could be considerable short-run losses of profits moving from the regulated equilibrium to the deregulated one; furthermore, the demand for labor in the industry could shift substantially, increasing or reducing the number of jobs in the industry. Secondly, even if firms were no better or worse off from deregulation, some might be helped, and others hurt. In other words, the transition period to a deregulated environment could involve two types of gains or losses: overall shortterm gains or losses to specific factors or production in the industry, and short-term gains and losses to specific firms (or factors hired to specific firms). We consider both.

It can be argued that the overall gains and losses to both capital and labor in the industry resulting from deregulation are likely to be tied quite closely to the relationship between fares and the derived demand for aircraft capacity. In other words, if a decrease in fares tends to increase the total number of flights supplied by the industry in the US, that will tend to increase the return both to capital and labor: since aircraft capacity is the most important type of capital, an increase in the number of flights will increase the demand for aircraft. This will benefit both holders of existing aircraft and producers of new ones. Similarly, it will benefit flight crews. The opposite will hold if deregulation reduced the demand for aircraft capacity.

Now, it has been asserted by Douglas and Miller that existing CAB policies have set fares at a level so as to maximize the number of seats provided for air trips in the United States, and that the higher load factors resulting from deregulation would reduce the number of flights and the derived demand for aircraft capacity."

139

However, the evidence from intrastate routes in California and Texas suggests a very different story. In nearly every case studied, the total number of flights grew after the institution of low-fare service provided by intrastate carriers. (The load factor rose because the number of passengers rose more rapidly than the number of flights, and on the basis of evidence from previous studies, it is difficult to find anything which would cause traffic to grow so rapidly on these routes save for the lower fares.140 Thus, the few instances we have seen of dramatic fare reductions would seem to contradict Douglas and Miller's assertion that a reduction in fares would reduce the demand for aircraft capacity.

The Appendix (Section I) challenges the assertion that CAB fares are currently set to maximize the number of flights. The details are too technical to go into here, but the basic idea is that this result depends crucially on the empirical value of a specific demand parameter (the amount by which a one per cent fare cut would increase traffic on a typical route). We simply do not have sufficiently accurate. estimates of this parameter for the trunk air routes of the US to know the impact of a fare cut on the amount of capacity provided.

Nevertheless, the method presented earlier in this section, and described in detail in the Appendix, can be used to predict the impact of deregulation on the number of flights offered. As can be seen from

139 Economic Regulation of Domestic Air Transport, p. 176.

140 See Kennedy Report, pp. 46-48.

Table I of the Appendix, the result is ambiguous, and depends on what one assumes to be an accurate measure of the fare under current CAB regulation. If the revenue yield is an accurate measure of the fare, then the number of flights would fall with deregulation. If, on the other hand, the unrestricted fare is the accurate measure, the number of flights would go up. The basic question is whether the flightincreasing tendency of an increase in demand resulting from a lower, less-regulated fare will offset the flight-reducing effect of an increase in load factor. Although the results of the theoretical model are ambiguous, the actual results from deregulated routes indicate that the number of flights could easily rise from deregulation.

The roughness of these results is obvious. However, what will happen on any one route is even more ambiguous: the few studies of demand on individual routes which have been made (by observing changes in fares and traffic over time and trying to net out other factors) would seem to indicate that the fare elasticity of demand (and, hence, the extra traffic which a fare cut would induce) varies widely from route to route, depending on numerous factors, including available surface alternative transportation, percentage of traffic made up of business versus pleasure travelers, the attractiveness of the route for conventions or vacations, etc.141

As a result of this, it is likely that deregulation would increase the demand for flights on some routes, while decreasing it on others. This means that if a carrier served routes with mainly low demand elasticities, it could, indeed, suffer losses from deregulation, although it is difficult to predict specific carriers which would lose. Other carriers would gain.

Route structure, however, is not the only reason why some carriers could suffer losses (and lay off workers) as a result of deregulation. In any competitive market, it is to be expected that new firms will attempt to enter, and not all will survive in the long run. CAB regulation has suppressed both the entry and the exit which is necessary for these market forces to function. Thus, it would not be surprising if, under deregulation, a few of the less efficient firms were competed out of business by more efficient firms. One would expect this process to go on continually, renewed by firm entry.

The point of all this is that while deregulation could benefit holders of capital invested in the airline industry, it is likely that some airlines, some investors, and some workers will lose in the short run.

THE IMPACT OF REGULATION ON MANAGERIAL EFFICIENCY

Because the trunk airlines are subject to entry regulation, and because they are free to seek fare increases if their profit levels are "inadequate," one might be inclined to believe that managerial inefficiencies could creep into their operations, might fail to minimize costs for the outputs they produce.

Whether such inefficiencies do or do not exist in the airline industry is a matter of some controversy, and, before presenting any conclusions on this matter, it is worth briefly summarizing alternative viewpoints on this matter.

141 See, for example, Philip K. Verleger, Jr., "Models of Demand for Air Transportation," Bell Journal of Economics 3 (Autumn, 1972), pp. 437-457.

www

Let us first consider those arguments, theoretical and empirical, which support the notion that the airline industry should possess managerial inefficiencies. This argument most logically stems from the notion of "sheltering" from competition mentioned above, and has been supported directly with some empirical evidence. For example, Robert J. Gordon, some time ago, found that trunk air carriers with more "competitive" route structures (i.e., more service rivals on their routes) had lower costs than those carriers with fewer rivals.142 More recently, evidence presented by William Jordan has indicated that the California intrastate carriers achieve higher levels of productivity in all their cost accounts than do interstate carriers, and that even their higher load factors may be due partly to better scheduling practices, i.e., more planes at peak periods than the trunk carriers schedule.148 Finally, theoretical economic models have been developed by notable economists (Oliver Williamson, Richard Cyert, and Herbert Simon, to name a few) describing why, in a sheltered market environment, one would expect managers to pursue goals other than maximum profits, and why, as a result, inefficiencies could creep into firms.144 There is some evidence, then, both theoretical and empirical, to make us believe that trunk airline managements are indeed inefficient as a result of the sheltered environment afforded by the CAB.

However, there is also evidence to make us doubt that managerial inefficiency is a serious source of welfare loss from CAB regulation (above and beyond the losses stemming from service competition of various sorts, which is socially inefficient, but not from the viewpoint of a non-colluding, profit-maximizing firm).

First, and more importantly, although airlines do not compete in price, they do compete in service quality. Thus, on a given route, if one airline is less efficient than another, it will not be able to provide so good a service quality as others for a given fare level (this can involve any of the various dimensions of service quality discussed previously in this section). There is every reason to believe that the competitive pressures to provide high service qualities are every bit as great in the trunk airline market as are the pressures to operate at the lowest possible price in a price-competitive market.

There is also more direct empirical evidence that casts doubt on the hypothesis that a more sheltered regulated environment causes managerial inefficiency. Using data for the late 1960's Douglas and Miller attempted to redo the Gordon study comparing airline costs with the number of "competitive" routes each airline faced.145 The DouglasMiller results indicated no meaningful relationship, unlike the results of the earlier Gordon study. And Douglas and Miller's statistical techniques were more sophisticated than Gordon's.

As regards the intrastate carriers, it is difficult to separate completely the extent to which their lower costs stem from a different service quality relative to the trunk carriers and to what extent they stem from superior managerial efficiency. It will be recalled that the 1972

142 "Airline Costs and Managerial Efficiency."

143 Airline Regulations in America, chapter 11.

144 See, for example, H. Simon, "Theories of Decision-Making in Economics and Behavioral Science," American Economic Seview 49 (June 1959), pp. 253-283; 0. E. Williamson, "Managerial Discretion and Business Behavior," American Economic Review 53 (December 1963); pp. 1032-1057; and R. M. Cyert and J. G. March, A Behavioral Theory of the Firm (Englewood Cliffs, N.J., Prentice-Hall, 1963).

145 Economic Regulation of Domestic Air Transport, pp. 141-144

study done by the present writer (based on 1968 data) used a cost model estimated for the trunk airlines, but that model accurately predicted fares on the California intrastate routes. It is true that this model did attempt to correct for some aspects of perceived managerial inefficiency on the trunk carriers,146 and that may be why it managed to predict California intrastate fares using cost data for the trunks. However, it would appear from this work that a very large part of the difference between trunk and intrastate costs can be explained by such basic differences as seating configurations and load factors.

Thus, while there is some evidence that CAB regulation has allowed managerial inefficiencies to creep into the trunk airline industry, that evidence is anything but conclusive.

REGULATION AND TECHNOLOGICAL CHANGE

So far, our discussion has taken technology as given. That is, it has assumed that the "state of the art" in aircraft, ground equipment, pilot training, and firm organization are given, and are unaffected by regulatory policies. However, over the long run, regulation can affect the very development of technology. Although the costs and benefits of regulation as it relates to technological change cannot be discussed in the same quantitative context as the static costs and benefits (i.e., those calculated with technology assumed to be given), nevertheless, it is worth saying something about the impact of CAB regulation on technological progress in the airline industry.

Because CAB regulation has tended to hold fares at an artificially high level, forcing competition to occur in the form of product rivalry, one would a priori expect part of the rivalry to take the form of product sophistication. That is to say, the airlines might be expected to invest in more advanced and sophisticated equipment as a result of CAB price regulation than they otherwise would do.

Quantitative evidence on this is difficult to find, but it is possible to tell stories which illustrate that this is indeed true. Perhaps the most famous and striking such story is that of the Douglas DC-7, first told in detail by Aaron Gellman.147 It will be recalled that especially during the early and mid-1950's, air fares were set to yield much higher profits on long hauls than on short hauls. The net impact of this was more frequency competition and lower load factors on long hauls than on sort hauls. But competition took a different twist here in the 1950's; before the DC-7 there had been no aircraft with a transcontinental, nonstop capability-at least none which were anywhere near so economical in operation as the Douglas DC-6B, which required a refueling stop to cross the continent. Furthermore, it was known throughout most of the 1950's that by the end of the decade, commercial jet aircraft would be available, making for lower operating costs on long hauls than even the DC-6B. Nevertheless, prices

146 Specifically, the 1968 calculations selected coefficients for the nationwide airline with the lowest indirect operating costs, American in that year. Furthermore, the coefficients of the indirect cost equation may have been downward-biased in another way, as well; all passenger-mile figures were converted to ton-mile figures and added to freight ton-mile figures for the ton-mile variable. Because the coefficient for ton-miles reflects both passenger and freight operations, and given that freight operations are probably cheaper to provide than passenger operations in indirect costs, this biases downwards the coefficient if it is used to predict passenger costs, as it was in my paper.

147 Aaron J. Gellman, The Effect of Regulation of Aircraft Choice (Ph. D. dissertation, Massachusetts Institute of Technology, 1968).