Tnwpn h. 0 Papnon
Vol. 12. ~9. 313-319 Press Ltd.. 1978. Printed in Great Britain
INCOME AND COST IMPACT ON GENERAL AVIATION HOURS FLOWN BY INDIVIDUAL OWNERS STEPHEN G. VAHOVICH FederalAviation Administration,Officeof EnvironmentalQuality,Washington, D.C.. U.S.A.
(Received 28 June 19ll; in reoisedform 16 Febntary 1978) Ahstraet-Using data from the FederalAviationAdministration’snationalsampleof general aviation(GA) aircraft owners (1975). this study explores the impactof changes in family income, hourly operating cost, and other variableson live diiereat measures of hoursgown for aircraftowned by individuals(versus companies). Results for total, itinerant,instrument, and visualhoursgown supportthe theoreticalnotion that hours tlown rise until unit costs reach some critically high level, beyondwhich hours gown begin to decmase. For itinerant, visual, and total hours, the critical costs very closely approximatethe mean total hourly cost (operating plus annualixediixed) of flying.Local hoursrespond negatively to operatingcosts increasesover all positive unit cost levels. In all cases, the income elasticities of demand suggest thathours gown are a necessity. The low income and price elasticities of demandcoupled with the insensitivity of ownersto the availabilityof a substitutemode of transportationsuggest that GA aircraftowners are strongly committedto theirflyingactivities,exhibitingconvention4 but weak, demand responsesto economic stimulus. l.mTRoDucTIoN
General Aviation (GA) includes all aircraft except certified air carriers, supplemental air carriers and military. GA represents approximately 98% of all civil aircraft operations at Federal Aviation Administration (FM) towered airports (U.S. Federal Aviation Administration, 1977). The definition of GA suggests a large heterogeneous grouping, encompassing diverse types of aircraft, user groups, and utilization rates. This characterization of the GA community is supported by a fairly comprehensive analysis of the characteristics of a representative sample of GA aircraft owners (Vahovich, 1977). Findings from the latter study also support the---uniqueness -.of .-- the GA community as compared to the U.S. population. When the 3.S. is divided into 11 distinct geographical sreas (FM regions), both the median and mean incomes for aircraft owners are substantially higher than those of the U.S. population, and the distribution of aircraft owners across the U.S. is less skewed toward the East Coast than that for the general population. Given that almost all previous studies (e.g. Baxter and Howrey, 196& and Urban Systems Research and Engineering, 1975) dealing with GA utilize so&economic data that is not specific to the GA community, the latter findings suggest that an urgent need exists to conduct and present analysis utilizing GA specific data. The results presented in this paper work toward that end. A particularly critical problem area, faced in this paper, is the demand responses of GA aircraft owners to income and price changes. FAA published hours flown figures (U.S. Federal Aviation Administration, 1975) are estimates, imputed from the information gathered on the optional portion of the Aircraft Registration Eligibility, Identification, and Activity Report. The problem of nonrespondent bias is not addressed. Thus, the present research is unique in that it utilixes both GA specific socioeconomic sample data and accurate (within samp-
ling error) information on GA hours llown. An additional concern is that several studies (Baxter and Howrey, 1968, and Urban Systems Research and Engineering, 1975) utilizing income data for the U.S. population have arrived at GA activity income elasticities of demand that are greater than 1.0. These results suggest that GA flying is a luxury good. However, given the author’s earlier findings (Vahovich, 1977) that significant differences exist between the income level of the U.S. population and of GA aircraft owners, the luxury good hypothesis will be further scmtinixed in this paper. 2.SAMPI.E
The data source for the analysis is a strati&d random sample of the universe of 177,641 GA aircraft as of 1 January 1975 (aircraft owned by persons not residing in the. U.S.< nonengine-propelled craft, and govemmentowned aircraft are excluded). The Bureau of the Census collected the data for the FM. The primary method of data collection was telephone interviews (with personal visits and mail follow-up for telephone non-respondents). The usable response rate was 94.2% (N-.9,286). To enhance the homogeneity of the good consumed assumption, only fixed-wing aircraft owned by individuals are included. In addition, several screening criteria were established to eliminant deviant members of the sample. Five diflerent measures of hours flown are utilized in the present study. Respondents are excluded from each of the hours flown estimates if they flew zero hours for the relevant measure, if their aircraft was not presently flyable (i.e. at the time of the survey), or if they owned a twin/muRi-engine piston aircraft weighing 12.5OOlbor more. The 12,XlOlb and over piston aircraft have not been manufactured since 1960 and almost onehalf of these aircraft could not be classified in any of the user categories util&d in this paper. The remaining subsample consisted of 4,435 respondents. -- --_ .-___-.--. A com~~arisonofthis screened subsample with the full 315
STEPHENG. VAHOVICH
316
individual owner sample respondents by their distribution across the eleven FAA regions, nine primary use categories, and by age and type of aircraft reveals virtually no differences on any of these background variables. It should be noted that aircraft owned by individuals represent 65% of all aircraft owned, and about 63% of the screened subsample (vs 65% for the full sample) indicated that the primary use of their aircraft is personal use. 3.TElmmcAL.FRAMEwoRK
theory of consumer behavior provides the basic conceptual framework for this study. The fundamental assumption underlying the model is that GA aircraft hours flown (HRS) are significantly affected by owner’s income (Y), aircraft operating costs (C), availability of a substitute mode of travel (SUB), and a vector of variables (V) intended to reflect intensity of use. Because the demand function is usually formulated in a nonlinear fashion, and for reasons presented below, the following representation (mixed double- and semi-logarithmic forms) is estimated for all hours flown measures, except local hours (CSQ is square of C): The
In(HRS) = BO
+
B, ln( Y) + B2C t BICSQ t B4 ln(SUB) tB,V (1)
The double-logarithmic form estimated for local hours is: ln(HRS) t BO+ B, ln( Y) t Bz In(C)t B, ln(SUB) t Bg V. (2) HBS, Y, C and SUB are defined above and V represents the vector of avionics, age, and primary use dummy variables. All monetary variables are deflated by a regional price index (U.S. Department of Labor, 1975). Thus, geographically determined variations in cost and income are controlled. Several different measures of hours flown are employed in this study. Total hours flown, whose components are local (takeoff and landing at the same airport) and itinerant hours (origin of flight different than its destination), represents the aggregate hours measure. Itinerant hours flown are further dichotomized by the instrument (itinerant flight requiring periodic reporting to an FAA facility) and visual flight hours (itinerant hours, reporting not required) components. Although the same aircraft frequently is used to fly more than one type of hours, each component of total hours flown represents a distinct attribute of total GA activity. For example, the definition of local hours suggests a pleasure flight of relatively short duration with discretionary trip costs, while the definition of itinerant hours suggests a purposeful transportation tlight (for business or pleasure travel) of longer duration entailing a certain minimum trip cost. Substantive differences among these types of hours flown suggest that income and cost.changes impact differently on each of these activity measures. Thus, separate demand estimates are conducted for each hours flown measure. Family income is included in the estimating equation
on the expectation that as income increases, GA flying activity increases. Since the GA sample did not include data on family size, age of children, or families with more than one wage earner, the extent to which such factors influence hours flown could not be controlled. The operating cost variable (fuel and oil costs, airframe, avionics, and engine overhah and maintenance costs; U.S. Federal Aviation Administration, 1976) is included in the estimating equation on the expectation that fewer hours are flown as cost per hour increases. Except for local hours, preliminary estimates (using both the linear and log forms) yielded the improbable result that hours flown increase as variable costs increase. Yet when hours flown was regressed on the ratio of operating costs to income, the estimated coefficient was negative. The latter result suggests that as variable cost takes a larger proportion of individual owner’s income, hours flown decrease. While the estimated coefficient for this ratio variable was statistically signiticant (F-test at 0.95 level) in most of the hours flown estimating equations, its incremental R-square was very low and it was highly collinear with other variables in the equation. In an attempt to capture the suspected negative impact of high levels of unit operating costs on hours flown, a squared cost term (CSQ) was introduced into the estimating equations (except for local hours) along with the linear cost term (C). The hypothesis being tested is that because of the high initial investment cost of purchasing an aircraft, hours flown are insensitive to increasing unit costs at low operatingcost levels; only at some high level of operating costs, will hours gown decrease. The latter would be indicated by a positive sign for C and a negative parameter estimate for CSQ. The implicit assumption is that hours gown by different types of aircraft are homogenous commodities. The variable reflecting the impact of the most closely related substitute mode of transportation, air carrier, on GA hours flown is defined for each FM region as the product of the ratio of ah carrierto total operations and the number of airports having air carrier operations. The number of air carrier airports component of this variable describes the level of available service (i.e. the number of service points) in each FM region; the ratio component describes the frequency of air carrier service, adjusted for inter-regional differences in total service. The greater the availability of air carrier service for a region, the fewer the number of GA hours flown. Since almost all of the airports included in this variable are FAA towered airports (i.e. airports whose construction was highly subsidized by Federal Government approved funds), this variable is exogenous and not determined by the number of GA hours flown. Five user group dummy variables are expected to account for the varying impacts of the different uses of aircraft on GA hours flown. The business use category is held constant. Except for local flying, increased hours flown by business users may be associated with reducing nonproductive travel time and lowering opportunity costs as compared to other travel modes. _.__. Because _ .Ithese ex&it-Zd implicit costs are likely to be greater for
Income and cost impacton generalaviationhours flown by individual owners
business users than for personal and the combined aerial application-industrial/special users, negative parameter estimates are expected for these user group variables in all equations, except for local hours. The instructional and the combined air taxi and rental categories utilize their aircraft as revenue generating sources. Since these groups realize the full impact of reduced utilization rates on their marginal revenues, whereas reduced consumption of hours flown by business users is only partially reflected in reduced revenues, a positive sign is expected for these user group variables. Local hours, having the same orgin and destination point, are generally associated with the pure pleasure of Ait experience and with certain productive activities. Since, at best, local hours may be considered incidental to the business use of aircraft, a positive parameter estimate is expected for the personal and the aerial application-industrial/special user groups for local hours. That is, personal users may at least derive a certain level of satisfaction from recreational flying, and most aerial application and industrial/special user activities (e.g. crop dusting, survey work, pipe line patrol) are compatible with local hours. The final set of variables relate to aircraft characteristics-age and level of aeronautical sophistication. The age-related variables test the hours flown response of owners whose aircraft were manufactured prior to 1960 and those of the post-l%9 vintage against owners whose aircraft are of the middle vintage (year of manufacture 1960-1%9). Because of the deleterious effect of age/use on an aircraft’s operating condition and performance, aircraft manufactured prior to 1960 are expected to fly fewer hours than those manufactured between 1960 and 1969. For the opposite reason, aircraft of the post-l%9 vintage are expected to fly more hours than those in the category held constant. The level of aeronautical sophistication is expected to vary directly with hours flown. Since the GA survey did not include information on pilot certification, his aircraft avionics equipage is used as a surrogate measure of sophistication. That is, either because of necessity of use or because of a greater preceived level of satisfaction derived from the consumption of hours flown, the GA owner has achieved a level of proficiency over and above the minimum level required or attained by other owners and has correspondingly equipped his aircraft with the appropriate avionics. In particular, a dummy variable representing equipage with an automatic direction finder is the selected measure of aircraft sophistication. Because of its function, equipage with this instrumentation is indicative of longer inter-city flights. However, since this device is also one of the several types of avionics that satisfy instrument flight requirements, it did not serve well to dtierentiate among owners’ hours flown in the instrument hours equation. In the latter equation, equipage with an altitude encoding transponder was used to differentiate among owners flying at the most sophisticated flight procedural level. Equipage with this type of avionics, required for high altitude flights (18,000ft and above) and in airspace around most major cities, suggests a high number of.instrument hours flown.
317
4. EMPIRICAL RESULTS
Table 1 presents the estimated demand equations for each of the hours flown measures. The estimated coefficient for the income variable represents the income elasticity of demand. As shown in Table 1, except for local hours, the parameter estimates for family income are statistically significant (at the 0.95 level or better), positive, and less than one. Income elasticities of demand less than one imply that hours flown are a necessity. This result is in contrast to the popular conception of GA as a highly discretionary, pure pleasure sport. The hours flown categories exhibiting income inelasticity are precisely those associated with purposeful transportation-takeoff and landing at diierent airports. The convenience of use and speed of travel advantages offered by GA aircraft for purposeful transportation help to explain the apparent insensitivity of hours flown to decreases in family income. On the other hand, it appears that, in the absence of other factors, an increase in income is not a sufficient incentive to induce disproportionately larger increases in hours flown. In contrast, two fairly recent studies, one using income of the population for a specific geographical area (Baxter and Howrey, 1968), another using income for the U.S. population (Urban Systems Research and Engineering, 1975). obtain income elasticities greater than one. The principal difference is that the present study uses GA aircraft owner’s income. Further, comparing the income and price elasticities of demand (see Table 2 for price elasticities), results from the present study concur with Weld’s (1953) conclusion: “ . . .as a rule, income elasticities of necessities are smaller than their price elasticities, whereas income elasticities of luxuries are grearer than their price elasticities.” For local hours the income elasticity is negative and less than one, suggesting that local hours is an inferior good. However, the parameter estimate is not statistically significant. The parameter estimates for the cost terms are consistent with conventional economic theory-variable cost is a significant determinant of hours flown. However, except for local hours, the results lend credence to the theoretical notion that hours flown rise until costs reach some .critically hi point, beyond which hours flown begin to decrease-de demand curve bends backward. This is evidenced by the positive and negative signs for the parameter estimates of C and CSQ, respectively. Information for a more detailed analysis of those cases exhibiting this sign pattern, with statistically significant (at the 0.99 level or bet&) coefficients, is presented in Table 2. Results from Table 2 imply that visual hours flown decrease when hourly operating costs exceed $144. itinerant hours flown decrease when unit costs exceed $275, and total hours flown decrease beyond $211. However, a breakout by types of hours flown reveals that less than 1.0% of the aircraft have hourly operating costs greater than or equal to the relevant critical point. One plausible explanation for the high critical point values is that the operating cost coefficients are picking up the impact of annualized investment cost in addition
318
STEPHENG. VAHOVICH
Table 1. Regressionresults Dependent
Independent
Tqtal (N-3,910)
Variables In
Pemily InCone
125:@f9"
colrt
.OOBla (15.49)
coat
_.2.10-'a (6.90)
Squared
--
In Coclt
1~ Subetitution Variable
-.0657b (5.OE)
Variable*: In of Iioure Flown Mea8ure8 1tinermlt Inat-t ’Vimal Local (N=3,039l (N=3,334) (N-3,237) (N-790)
(9.49)
__
-.4055a (41.86)
1.806a (292.83)
Air TaxiRental
.624ia (42.25)
Aerial Applice~~~~;d=etriel/
.3690a (23.641
.2423a (23.07)
-.4530a (98.25)
.0583b (4.61) .0115' (14.25) -.4*a0-4a
(io.49i
__ -.0864b (5.56) -.1269 (2.07)
-.4509' (90.42)
.1003 t.931
.2852 (2.03)
.1712 t.401
1.081" (71.15)
.4388a (15.151
,6662" (16.47)
.3402' (8.17)
1.587a (262.30)
-.5178a (17.72)
.4269 (1.79)
-.4962' (15.301
-.3612a (115.34)
-.1114' (6.821
-.2949a (50.20)
-.1878C (3.46)
-.2622' (37.60)
.1202' (7.17)
.2913' (25.73)
.1419a (6.75)
.2183b (4.931
.132Eb (5.45)
Avionics*
__
__
.5399a (167.52)
CONSTANT
4.04
4.43
3.43
l
1i.io-4
t.011
-.0691b (3.78)
--
.7062a (68.73)
R-SQUARE
.0056 (2.56)
-.2.lo-4a.
--
Inetructioaa1 use .
.19
.18
F
(17.75)
.0110a (17.65)
-.3393a (78.38)
Post-1969
(13.82)
--
Personal Use
Pre-1960
.2293a
.09774’
-.0239 C.79)
e5.92a
96.12"
.23 04.ma
1.341" (21.01)
.4425' (104.12) 3.89
.54
.lS
.16 14.41a
59.08"
Repreeents altitudeencoding treneponderfor the iaetrument hours equation, In all other equetlonsit represent?equipege with en automatic directionfinder.
( ) = F-values
Signikance levels: a = 0.99, b = 0.95, c = 0.90 - = Variable omitted because of irrelevanceor because of multicolinearity. Table 2. Inflectionpoints (criticalcosts), mean costs, and price elasticities Tyw
ROurS
of Flown
Total Itinerant Vima.al
Critical Operating comts (S/hr. 1 I
mean Hourly Operating Costa
$210.50 275.00 143.62
to operating cost. Under this scenario, an aircraft owner is viewed as renting the airplane to himself-both the operating cost and the annualized portion of the total fixed cost are considered important factors determining the use of the durable good. The aircraft owner’s hourly use charge to himself is the sum of the annualized fixed cost divided by the number of hours flown plus the hourly operating cost. This rental price describes the opportunity cost of the owner’s use since it represents the income he foregoes which could otherwise be obtained from renting the plane rather than using it himself (see, e.g. Rat&ford, 1974, Jorgenson and Stephenson, 1967, and Aaron, 1970). The aircraft owners’
$16.95 17.42 16.93
Mean iiourly User Charge $229.17 333.01 164.43
Price
Ela8ticity .13 .19 .17
mean hourly use charge for each category of hours flown is presented in Table 2. The annualized average fixed cost component, by type of aircraft, includes annualized investment cost (straight line depreciation is used); hull insurance; liability and medical insurance; hangar, storage and tie down charges; federal registration fee and weight taxes; and miscellaneous costs such as state registration fees, maps, charts, etc. (U.S. Federal Aviation Administration, 1976). Without exception, the critical cost, at which hours flown decrease, appears to be closely related to owners’ mean use charge. Thus, the wide discrepancy between mean hourly operating cost and the critical point can be explained. That is, aircraft
Income and cost impacton generalaviationhours flownby individual owners
owners decrease hours flown approximately at the point at which the price of an additional hour equals the total hourly cost (operating plus annualized fixed) of flying. A considerably different picture emerges from the local hours equation. Local hours respond negatively to increases in operating costs at all positive unit cost levels. However, the parameter estimate for the cost variable (representing the price elasticity) implies that for a 1.0% increase in operating cost, hours flown decrease by only 0.41%. Thus, while operating cost have a negative impact o,n local hours flown over the entire relevant range of the function, the demand response to price change is weak. Results for the other variables support the hypotheses presented in Section 3. All user categories fly more local hours than business users. Personal users fly fewer total, itinerant, instrument, and visual hours than business users. On the other hand, instructional and the combined air taxi and rental categories fly more of each of the latter types of hours than do business users. While the combined aerial application and industrial special user group is similar to personal users-negative parameter estimates in the itinerant and visual hours equationsthey fly more total hours than business users. The positive impact of the aerial application-industrial/special group in the total hours equation, versus its negative sign in the itinerant hours equation, probably results from the large positive influence of the local hours component of total flight hours-i.e. the activities of this user category are highly compatible with local flying. Parameter estimates for the variable representing the substitution effect suggest that substantial switching to air carrier service is unlikely, even if such availability increases significantly. In part, this may stem from the “value” GA aircraft owners place on the comfort and/or joy of tlying their craft, and in part from the lack of compatibility between the routes/service offered and the destination of current users of GA aircraft. Parameter estimates for the agerelated variables suggest that, compared to aircraft of the MO-l%9 vintage, older aircraft fly fewer and more recent vintage aircraft fly more of each type of hours. Avionics equipage has a statistically significant (0.99 level) and positive impact, suggesting higher utilization rates for more sophisticated aircraft.
5. CONCLUSIONS
findings for individual GA aircraft owners are consistent with conventional economic theory-income and costs are sign&ant determinants of the demand for aircraft hours flown. Contrary to findings obtained elsewhere (Baxter and Howrey, 1968, and Urban Systems Research and Engineering, 1975), parameter estimates for family income suggest that hours flown are a necessity rather than a luxury good. The present study uses GA aircraft owner’s income, whereas other studies use The
TRVol.12,No.5-B
319
income for the general population. Further, there is some weak evidence supporting the contention that local hours flown is an inferior good. The parameter estimates for the hourly operating cost terms suggest that, except for local hours, hours flown rise until costs reach some critically high level, beyond which hours flown begin to decrease. For itinerant, visual, and total hours, the critical costs very closely approximate the mean total hourly cost (operating plus annualized fixed) of flying. On the other hand, results for local hours suggest that unit operating costs have a negative impact on flying time over all positive unit price levels. For each type of hours flown, when the price elasticity is evaluated at the mean, the demand response to changes in operating cost is weak (less than proportional). Comparing the price and the income elasticities of demand reveals that the price elasticities are larger than the income elasticities. This result concurs with Wold’s (1953) conclusion for goods that are necessities. The low income and price elasticities of demand coupled with the apparent insensitivity of h6urs nown to increases in air carrier service presents d consistent picture of individuals who own their own GA aircraft. In general, the results suggest that GA aircraft owners are strongly committed to their flying activities, exhibiting conventional, but weak, demand responses to economic stimuhls. REFERENCES
Aaron H. (1970) Income &es and housing. Am. Econ. Rev. 60, 789-806. Baxter N. D. and Howrey E. P. (196@ The determinants of general aviation activity: a cross-sectional analysis. Transpn Res. 2,73-M. Jorgenson D. aad Stephenson J. (1967) Investment behavior in U.S. manufacturing, 1947-1960.Econometrica. 35, 169-220. Rat&ford B. T. (1974) A model for estimating the demand for general aviation. Transpn Res. 8,193-203. Urban Systems Research and Engineering (1975). Models for Auiatiott Actioity Fomcusting. Report prepared for Federal Aviation Administration, Cambridge, Mass. U.S. Code of Federal Regulations. Title M-Aeronautics and Space, part 91.33 (1977). G&e of Federal Register, National Archives and Records Service. Washington, D.C. U.S. Department of Labor, Bureau of Labor Statistics (1975). Handbook of Labor Statistics. (Table 145) U.S. Government Printing Office. Washington, D.C. U.S. Federal Aviation Administration, Mice of Aviation Policy (1977). Aviation Forecasts fiscal Years 197~1%9. National Technical Information Service. Springtield, Viiginia. U.S. Federal Aviation Administration, G&e of Aviation Policy (1976). Selected Statistics United States Geneml Aviation 1959-197s. National Technical Information Service. SpringBeld, Virginia. U.S. Federal Aviation Administration, Gfiice of Management Systems (1975). Census of U.S. CM Aircmft. National Technical Information Service. Spriagtield, Virginia. Vahovich S. G. (1977) General Aviation: Aircmft Owner and Utilization Clromcterfstics. National Technical Information Service. Springfield, Vii. Wold H. (1953)hand Analysis. Wiley, New York.