In addition, because their model used three correlated variables, one may influence the other. Therefore, it may be more appropriate to consider all three elasticity components instead of the price elasticity alone. It is also possible that, before deregulation, air passengers displayed less price-sensitive behavior because of the limited fare choices available to them.

The relationship between the two above estimates indicates that the business price elasticity can be 50 percent lower than the nonbusiness value.

• Straszheim, 1978 (see citation number 39 in the references). This research estimated the air traveler price and income demand functions for various classes of service. Straszheim examined the role of pricing and the impact of introducing special services at discount rates on the growth of the airline market. Data from North Atlantic routes were used because that was the only market with data available on disaggregated levels of service. Straszheim conducted regressions on the U.S. GDP and the fares for different classes of service for the period 1952 through 1973. The results of his study are:

These results show that first class traveler price elasticity is 60 percent to 70 percent lower than the economy travelers' values. If it is assumed that first class passengers represent the same price elasticity behavior as business travelers, and discount economy passengers are representative of nonbusiness travelers, then the relationship between the two passenger price elasticities is consistent—business travelers exhibit less price-elastic behavior than nonbusiness travelers. Straszheim stated that, "from examination of historical data it is evident that a large tourist market exists which is responsive to lower priced air service..." in spite of the various restrictions placed on their travel. To investigate the economy traveler further, he concentrated on the period from 1963 through 1973, when different types of economy fares were introduced. Those results follow:

For this pre-deregulation period, data showed relatively high price elasticities for economy travelers-predominantly nonbusiness travelers, according to expert panelists and several literature sources. This study adds to the evidence that nonbusiness price elasticities may be greater in magnitude than -2.0. Although this study of the North Atlantic region consisted of international traffic between the U.S. and Europe, the driving force behind the model was the U.S. GDP, so the elasticities may be biased toward U.S. travelers.

SUMMARY OF PRICE ELASTICITY LITERATURE FINDINGS

None of the literature reviewed contained price elasticity estimates for Family Travel Units (FTUs). Some price elasticity estimates were available, however, for nonbusiness travelers, the passenger group thought to be similar to FTUS in price sensitivity. To determine values for FTU price elasticity estimates, FTUs were assumed to be the most price-sensitive nonbusiness travelers.

areas:

The literature review provided price elasticity information in the following four

The overall range of price elasticity estimates

Price elasticities for business and nonbusiness travelers

Price elasticities for different trip lengths

Price elasticities by income level

The price elasticities in the literature ranged from -0.6 to 4.5. However, most of the studies presented values in the range of -0.8 to -2.7. Five of the studies provided evidence that nonbusiness price elasticities may be greater in magnitude than -2.0 [Oum, Zhang, and Zhang; Oum, Waters, and Yong; Abrahams; Straszheim; Directions]. Six studies demonstrated that business travelers were less price-elastic than nonbusiness travelers the approximate difference was 50 percent lower values for business travelers (Oum, Waters, and Yong; Oum, Gillen, and Noble; Morrison and Winston; Straszheim; Directions, Apogee 1994].

Results of two studies demonstrated that differences in price elasticities corresponded to different trip lengths. However, both were eliminated because they both used pre-deregulation data, one of them only considered monopolistic markets [Ippolito], and the other involved two correlated variables and did not differentiate between the types of travelers [De Vany).

Only one study provided any information regarding price elasticity differences for passengers of different income levels [Directions). This study demonstrated that lowincome travelers had greater price elasticities than high-income travelers.

The authors note that these values are low. When these values are combined with the two time components developed by the authors, the nonbusiness elasticity estimate becomes -0.859.

The studies most applicable to this analysis were then selected from Exhibit G-1.
Several studies were not used in the analysis for the following reasons:

Oum, Zhang, and Zhang-The Chicago hub was not indicative of the entire U.S. nor of business or pleasure travelers.

Oum, Waters, and Yong-the types of passengers subject to these values were unknown.

Directions-There was no clear distinction between price elasticity values for short and long trips. The relative differences between business and nonbusiness travel and passenger incomes verified panelists' opinions.

Apogee-The estimates were used to determine the relative difference between business and nonbusiness traveler price elasticities, but did not segment passengers by distance or trip purpose.

Morrison and Winston-Panelists' opinion was that these values were inconsistent with the level of price elasticity for U.S. air travelers. However, these values did exhibit the relative price elasticity differences between business and nonbusiness travel.

Ippolito-This study did not differentiate between business and nonbusiness travel. The dominant number of business travelers on short trips may have made those price elasticities more inelastic, and the greater number of nonbusiness travelers on longer trips may have made those price elasticity values more elastic.

De Vany-This study did not differentiate between business and nonbusiness travel. The dominant number of business travelers on short trips may have made those price elasticities more inelastic, and the greater number of nonbusiness travelers on longer trips may have made those price elasticity values more elastic.

The fact that many studies were not considered directly applicable to the CRS analysis illustrates the difficulty of determining values for use in this study. Exhibit G-2 lists the studies determined to be applicable to this analysis.

The price elasticities shown for first class travel were not used because they are generally indicative of business travel. In addition, the transcontinental price elasticity obtained from Abrahams was not used because there was no apparent differentiation between business and nonbusiness travelers.

The studies used provide a conservative range of price elasticities because they are not based on the most current data and the panelists suggested that air fare price elasticities may have been increasing over time.

Business travel. Business travel price elasticities were developed but not used in the seven scenarios in the CRS analysis. They were not used because none of the scenarios pass the increased air carrier CRS costs on to business travelers. However, the analysis can accommodate such a scenario, so the values were developed. The business traveler price elasticities are approximately 50 percent of the nonbusiness elasticities, based on a review of five sources [Oum, Waters, and Yong; Oum, Gillen, and Noble, Morrison and Winston; Straszheim; Directions]. This proportion was applied to the estimates for nonbusiness travelers to develop the following price elasticity values for business travelers.

EXHIBIT G-3

Price Elasticity of Demand Parameter Values for Business Travelers (and percentage of business enplanements)

Source of price elasticity estimates: Oum, Waters, and Yong; Oum, Gillen, and Noble;
Momson and Winston; Straszheim; Directions.

Source of passenger percentages: U.S. Travel Data Center

The simple average of these four business traveler price elasticities is -0.9; the weighted average -0.8.

LINEAR DEMAND, ELASTICITY, AND REVENUE

This study assumes a linear functional form for the demand for air travel. This form was selected over a non-linear, constant elasticity model, because of the FAA's concerns that the elasticity of demand was unlikely to remain constant over the large increases in air fares considered in some scenarios. Linear demand curves, on the other hand, have the property that demand becomes more elastic as the price level increases.

As noted in Chapter 2, the elasticity estimates documented in this report were used to estimate the slopes of the demand curves and, in turn, the impact on demand of increases in air fares resulting from certain CRS policies. However, given that demand will become more elastic as price increases (assuming a linear form), the elasticity values documented here should be thought of as "initial” values. In other words, these point estimates depict the elasticity value at that particular point on the demand curve (i.e., corresponding to that combination of price and quantity) in effect prior to the increase in air fares. The elasticity will be greater following an increase in price (given the combination of higher price and lower quantity demanded).

A further consequence of selecting the linear function form for the demand curve is that air carriers will not experience a zero net revenue gain when the initial elasticity is set to -1.0 (except for infinitesimally small increases in air fares). In fact, given this functional form, air carriers will also experience a loss in revenues for some initial elasticity values between 0.0 and -1.0 if the price increases are sufficiently large. An explanation for this phenomenon is provided by the following proof.

Let

Proof that a zero net change in air carrier revenues for a linear
model implies an initial elasticity value, E, such that -1 < E < 0.

Assume a zero net change in air carrier revenues following an increase in air fares. Hence,

R1 = R2 or R1 - R2 = 0

Proof:

Rewriting the expression for the new revenue following the price increase, R2, we get the following.

Or.

R2 = P2 Q2 = (P, + AP) (Q, + AQ)

R2 = P2 Q2 = P, Q, + P, AQ + APQ, + APAQ.

Note, however, that P2 Q2 = P,Q,. Hence we can subtract these terms from both sides of the expression to get the following:

P, AQ + APQ, + APAQ = 0