The details of the calculations for the results presented here are shown in the Appendix to this study. The detailed calculations, as shown in the Appendix, are for the year 1974, perhaps the most recent year that CAB regulation continued in its fullest force, as previously argued.

Effects of CAB Regulation on Fare Levels

To estimate the potential impact of deregulation on fares, we shall first calculate how fares are likely to change with deregulation for a "typical" trip in terms of length, number of stops, and aircraft type. We shall then aggregate to all trips by multiplying the likely fare reduction for a typical trip by the total number of such trips. If anything, this will understate the consumer gain from moving from existing to optimal fares, because it ignores inefficiencies introduced by a nonoptimal fare structure of the sort mentioned in the previous subsection. We shall attempt to get a rough notion as to how much additional cost is imposed by a nonoptimal fare structure later in this section.

Another important assumption as relates to the estimated welfare loss is just what measure of actual, regulated fare should be used for the "typical" trip. The estimates of Douglas and Miller, Pustay, and others used actual revenue yields of CAB carriers. The main problem with it, as pointed out previously, is that it averages in many different discount fares with various trip restrictions. These fares provide service qualities different from the no-string-attached unregulated fare offered on intrastate routes in California and Texas. Thus, if a model predicting optimal (cost-based) fares on interstate routes also predicts accurately the unrestricted fares on intrastate routes, it is only reasonable that predicted efficient fares in interstate routes be compared with similarly unrestricted fares under CAB regulation. The importance of this is that using revenue yields as a measure of interstate fares (as Douglas and Miller have done) yields a downward-biased estimate of the welfare loss from CAB regulation.

What we shall do, then, is to predict an "efficient” interstate fare for an average interstate trunk route, based on intrastate load factors and seating configurations. This fare is then compared with the actual, unrestricted CAB fare on such an average route. This difference between the two fares is then multiplied by the total number of coach trips taken to get the full fare cost of CAB regulation to coach

passengers.

Although use of interstate revenue per passenger trip will yield a downward-biased estimate of the fare loss from CAB regulation, nevertheless, it would be useful to know what the welfare loss figure looks like based on the assumption that the regulated fare is the revenue yield. The reason is simply to get a "rock-bottom" estimate of the welfare cost of CAB regulation, and to test for the sensitivity of our results to variations in assumptions. Therefore, we present two estimates of the welfare loss to consumers from airline regulation, one based on each assumption regarding regulated fares.

The calculations are based on the cost model presented in Keeler (1972), adjusted for 1974 prices as described in the Appendix. It was assumed, furthermore, that an unregulated carrier would use a 727-200 aircraft for this typical route (the average trunk haul in 1974 was 795 miles, a distance appropriate for this aircraft type). It was assumed

121 Ibid., p. 6.

also that an unregulated carrier could achieve a load factor of 60 per cent in an all-coach configuration. This figure reflects what the California and Texas carriers have achieved, and also the calculations of Douglas and Miller, which indicate that on a route such as this one, a load factor of 60 per cent is about right in terms of efficiency.122 Because meal service would be required on a trip of this length, we allow for full galley space, consistent with a seating capacity of 153 passengers per aircraft. 123

Based on these and other assumptions discussed in the Appendix, an "unregulated" market for this typical route would result in a fare of $46.70, or a yield of 5.88 cents per mile. For a typical route of this length, a regular fare of $67.60 was charged in 1972, for a revenue yield of 8.5 cents per mile. Multiplying this over all coach trips made in 1974 yields a loss (excessive charge) to coach passengers of $2.7 billion; the fare loss so calculated is $20.90 per trip, or 31 per cent of the fare actually paid.

The alternative calculation, based on the assumption that the regulated fare is the coach revenue yield, generates a lower welfare loss estimate. Specifically, the actual 1974 coach revenue yield was 6.94 cents per passenger-mile.124 Thus, the loss on the basis of this assumption is about 1 cent per passenger-mile, or $1.1 billion overall. By these calculations, the fare loss to coach passengers is about 15 per cent of the regulated fare paid.125

Effects of CAB Regulation on Coach Service Quality

Let us now consider the service quality improvements brought on by CAB regulation. Again, we restrict our analysis to coach and economy service, for several reasons. First, there is strong evidence to indicate that first class service exists exclusively by grace of the cross-subsidy it currently receives from coach, which will be discussed in the next subsection; in other words, there is evidence to indicate that the vast majority of those who use first class service (at that now only ten percent of the total) would not be willing to use it if they had to pay the full long-run costs of providing it.126 Furthermore (and along the same line of reasoning), no intrastate carrier provides first

122 For route density of 400 passengers per day, Douglas and Miller find an optimal load factor (on a Boeing 727-200) of fifty-seven percent with a $10 per hour time value, and sixty-one percent with a $5 per hour value. For a 1,000 mile haul, they find an optimal load factor of sixty to sixty-three percent, again depending on the value of schedule delay time. Thus, for the 792 mile haul discussed here sixty percent might seem roughly correct. It must be remembered, however, that because the operating cost data used here differ from those of Douglas and Miller, optimal load factor calculations based on them are likely to be different from those of Douglas and Miller, at least by a small amount. Thus, the sixty percent assumed here is an approximation. Dorman, op. cit., provides a more detailed integration of the present cost model with the Douglas and Miller stochastic delay model.

123 The capacity of 153 with full galley space is based on an estimate by Mr. Lawrence Guske of PSA as to how much the capacity of its 727-200's (which seat 158 with two coffee bars) would be reduced if they were provided with full galley space. Mr. Guske gave this estimate in a telephone conversation with Mr. Thomas E. Dooley of the U.S. General Accounting Office.

124 Air Carrier Traffic Statistics (December 1975), p. 6, and USCAB, Air Carrier Financial Statistics (December 1975), p. 3.

125 That is (6.94-5.88)/6.94-.153. In calculating this welfare loss, we have not attempted to calculate a "welfare triangle,' which represents the extra benefits from extra air services consumed as a result of the lower fares, in addition to cost savings from existing consumption, for three reasons. First, this welfare loss is likely to be quite small compared with cost savings. Second, it is highly dependent on the demand elasticity for air transportation, and that we do not know with any certainty, as explained in the text. Third, for calculation of this value to be meaningful, it is necessary that prices of substitute goods (such as auto travel) reflect marginal costs as well. In fact, however, there is evidence that highway user charges for intercity auto transport do not, in fact, reflect their costs. See Keeler, Resource Allocation in Intercity Passenger Transportation, chapter 4.

126 See the text below, under the subheading "The First-Class-Coach Differential."

class service, so not only it is difficult to see a market justification for it, it is impossible to see what this service would be like (if it existed) in a deregulated environment. Finally, if first class service did disappear with deregulation, it is difficult to measure the full benefits which current first class users receive from the service which they would not receive in the absence of first class.

To determine the likely effects of deregulation on trunk airline service quality, we make some further simplifying assumptions. As Douglas and Miller do for their welfare loss calculations, we assume that a typical route carries 400 passengers (in both directions) per day.127 In addition, we assume that each one per cent reduction in the coach fare level will induce an additional 1.2 percent increase in passenger travel on a route. This is consistent with previous demand studies, and it incorporates two effects: the effect on additional demand from a reduction in price, and the reduction in service quality which lower fares will induce (in other words, a lower fare induces the airlines to achieve higher load factors in order to break even. This increase in load factors makes it harder to get reservations at some times, and may, as a result, drive away demand.). The demand elasticity assumed incorporates both effect, based on the most reliable available empirical evidence.128

On the basis of this assumption regarding route density, and with evidence on seating capacities of a typical coach compartment under CAB regulation, and of a coach-configured aircraft under deregulation, it is possible to calculate total schedule delay, before and after deregulation, confronted by the typical coach passenger (recall that schedule delay is the difference between the time the traveler would prefer to leave and the time he must leave, given flight frequency and seat availability on a given flight). The details for finding this schedule delay are given and discussed in Section IV of the Appendix. They are based on a functional relationship between aircraft size, passengers per aircraft, and route density developed and estimated by Douglas and Miller. Again, calculations were based on two alternative assumptions regarding the currently-regulated fare: the first is that the fare may most accurately be measured by the official, unrestricted fare on an "average route" (average in the sense of length). The second assumption is that the typical fare is best measured by the revenue collected on such a typical route, averaging in the effects of various discount (but restricted) fares.

The results of these calculations are surprising: they indicate that the typical coach passenger on regulated trunk airlines suffered as much or more schedule delay as a coach passenger in a deregulated environment would encounter. This is not so surprising as it may seem when one examines the load factor in coach service provided by the trunk carriers: in 1974 and 1975, it averaged around 58-59 per cent.129 Given that the "deregulated" or "efficient" fares calculated here are based on a load factor of 60 per cent (and that that is the load factor, in fact, achieved most often in recent years by the intrastate carriers), it is not surprising that coach passengers would find little, if any,

17 Economic Regulation of Domestic Air Transport, p. 172.

12% This value of demand elasticity is the average of the values estimated by the DeVany and Brown-Watkins studies previously cited. See Appendix footnote 31.

129 U.S. Civil Aeronautics Board, Air Carrier Operating Statistics (December 1975) p. 6.

deterioration in schedule delay with deregulation. And if one uses the official, unrestricted fare as the measure of the coach fare under CAB regulation, the decline in fare with deregulation is so dramatic that there is enough induced demand from coach alone to actually increase flight frequency with deregulation, reducing schedule delay modestly, with a value of coach air travelers of over $200 million per year in schedule delay time saved.130 Again, details are shown in Section IV of the Appendix.

On the basis of evidence from intrastate markets, the latter assumption about regulated fares seems to produce results on flight frequency more consistent with reality: in these markets, whenever an intrastate carrier has moved onto a route, flight frequency has grown much more sharply than previously, and more rapidly than on equivalent routes elsewhere.131

The calculations on which these results are based are rough, and obviously not too much should be attached to the exact numbers shown in the Appendix as regards welfare losses and schedule delay times. Nevertheless, they indicate quite clearly that if a 60 per cent load factor is representative of the likely outcome on a typical deregulated route (and evidence indicates it is), then the much-touted service benefits which CAB regulation confers on coach passengers are quite negligible, and may even be negative.

Does this mean that CAB regulation has conferred no benefits of increased service on the traveling public? Certainly not. They imply instead that practically all the service benefits of the regulation have accrued to the ten per cent who ride in first class.132 First class load factors are far lower than coach (only 39.7 per cent in 1974 and 34.7 per cent in 1975), and first class passengers receive numerous other service amenities not received in coach.

In conclusion, then, we can assert with reasonable confidence that the welfare loss from CAB regulation to trunk coach passengers in the United States in 1974 was $1-2.7 billion, excluding the costs of a nonoptimal fare structure, and excluding losses from inefficient investments in technical progress induced by regulation. These costs will be investigated in the following two subsections of the present section.

Meanwhile, it is worth repeating that the loss calculated so far is not totally deadweight in nature; an unknown fraction is redistributed in benefits to the privileged ten per cent who fly in first class. Although these benefits cannot be measured, more should be said about the current benefits of regulation to first class passengers, and that is done in the following subsection, which is on the costs of CAB regulation in nonoptimal fare structure.

Effects of Regulation on the Structure of Fares and Service Quality There are two types of inefficiency in fare structure which, it can be argued, have been brought about by CAB regulation: the first is an inefficient differential between coach and first class fares. The second

130 This assumes a $5 per hour value of schedule delay time. It is difficult to believe that the average coach air traveler in a deregulated market would have a scheduled delay time value higher than this, for reasons given in the text in the discussion of schedule delay time valuation, above.

131 Kennedy Report, pp. 46-48. The recent introduction of sharply lower fares by Southwest Airlines has had an especially strong impact on both traffic and fight frequency in the Rio Grande Valley in Texas. Since Southwest entered the market, the number of flights serving Harlingen, Tex. has risen from 37 to 61 (Kennedy Report. p. 47).

132 This is discussed in more detail in "The First-Class-Coach Differential," on p. 122.

is relationship and fare and fare and distance, which in both coach and first class does not necessarily reflect the optimal relationship.

The First Class-Coach Differential

There is considerable evidence that first class service is cross-subsidized by coach, and that this has gone on for some years. For 1968, for example, the present writer found that while a first-class seat took up 1.68 times as much space as a coach seat, first class fares yielded only 1.36 times as much as coach fares. Furthermore, in the same year, the coach load factor was 56 per cent, compared with only 43 per cent in first-class. Overall, the revenue yield of a first-class-seat-mile was 2.98 cents, whereas the yield of 1.68 coach seats (one first-class space equivalent) was 4.80 cents per mile.133 It is this sort of cross-subsidy from coach to first class which seems to be responsible for a good part of the high price-cost margins for coach passengers observed in the previous section, with no corresponding service improvements to those passengers in the coach compartment. On the basis of this evidence, it can be shown that the first-class fares collected in the late 1960's and early 1970's were lower than the long-run costs of first-class service.134 Why should the airlines favor the continuation of this sort of fare differential, given that it does not seem profitable in the long run? No certain answer can be given, but the following one is reasonable and

consistent with the statements of various airline executives.135 The airlines have excess capacity, and they have had it for a long time. In the short run, it will increase profits to use a discriminatory pricing scheme which extracts extra payments from those willing to pay extra for the space which this excess capacity allows. Even if the extra fare charged for this capacity does not totally cover its long-run costs, it will be profitable in the short run, if the airlines are stuck with the capacity. Thus, the airlines tend to move the bulkhead between first class and coach so as to keep the coach compartment relatively crowded, while the roomier seats in first-class stay relatively empty, but the airlines do extract extra payments from those willing to pay something extra for the benefits of space and easy reservations during peak periods, plus other amenities such as finer meals and complementary drinks.

Although these policies may make sense as a short-run profit-maximizing strategy on the part of the airlines, they make less sense in the long run; indeed, if this story is true, and if airlines did maximize profits, it is difficult to see why an airline would configure new aircraft to accommodate first-class passengers, unless it perpetually overestinated its capacity needs and then price-discriminated to take advantage of that.

This problem of the first-class-coach differential was one which was considered by the CAB during the DPFI. After some vacillation on the issue, the CAB decided (in 1974) that fares should reflect costs in both compartments (based on a 55 per cent load factor), and that the cross-subsidization should thus be eliminated, but gradually over a period of several years.136 As a result first-class fares have been

133 Keeler, "Airline Regulation and Market Performance," p. 419.

134 For a rough calculation of the first-class loss for 1974, see the Appendix, Section V. 135 This is also the explanation given by Douglas and Miller, pp. 102-103.

136 One fare increase came Apr. 29, 1975, and another Apr. 1, 1976. The final increase occurred on Apr. 1, 1977. See "Seat Density Changes May Signal Trend," Aviation Week 104 (Jan. 19, 1976), pp. 25-26. See also A. E. Kahn, interview, in Congressional RecordSenate, Feb. 21, 1978, p. S. 1978.