Fare Elasticities of Demand for Passenger Air Travel in Nigeria: A Temporal Analysis 1 Ejem, E. A., 2 Ibe, C. C., 3 Okeudo, G. N., 4 Dike, D. N. and 5 Ikeogu C. C. 1,2,3,4,5 Department of Transport Management Technology, Federal University of Technology, Owerri, Imo State, Nigeria ejem.ejem@futo.edu.ng Abstract -- This paper developed an air travel demand model for Nigeria domestic for predicting changes in passenger demand with respect to fare. Using panel data from the period of 2009 to 2013, the model captured both time series and cross-sectional variation of air travel demand. The empirical analysis explicitly modelled fare as a variable by a log-linear demand model using OLS estimation. Empirical finding shows that at market level, the fare elasticities from the estimates indicate inelastic market demand. Except for routes with connecting flights, which is highly elastic with fare elasticity value of -2.932. No specific time trends for fare effects are found. This study provides a framework for the airlines to estimate demand on domestic routes operated with respect to fare structure to ensure business sustainability. Keywords -- fare; elasticity; sensitivity; air-travel and demand I. INTRODUCTION The overall trend of demand for air transport has been consistently increasing worldwide. In 1950s and 1960s, annual growth rates of 15% or more were common. Annual growth rate of 5-6% persisted through the 1980s and 1990s [3]. The continuing growth in the number of passengers and aircraft movements necessitates a rise in investments in airport and aircraft capacity in Nigeria. But even with these new investments, safety concerns and the environmental impact of aviation remain problematic in the Nigerian Aviation sector. Air transport is apparently a burdened with externalities. Increasing demand for air transportation service has compelled airline management to take advantage of opportunities in different markets. At the same time, increasing competition among airlines necessitates that airline management seek efficiency in all their decisions to promote their profit. It is no surprise that many airlines throughout aviation history have been unable to remain in business, and in most cases, it is agreed that the demise of these airlines has been attributable to deficient management. In the deregulated aviation sector, aviation management bodies therefore play a critical role in protecting the flying public from excessive pricing and in safeguarding the consumer against excessive usage of market power. One potential instrument the government has at its disposal is the fare. Quantitative study on passengers travel demand in air transport network has seldom been seen either in research literature or in industry practice in Nigeria, and it is urgently called for now when airlines need to make a decision on whether they should and how to adjust their routing structures. The purpose of this paper is to build an aggregate demand models for air passenger traffic that can be used by airlines to evaluate the overall passenger demand and to see how demand changes if fares are adjusted. It can also be used by airport and government to study the impact of airport capacity on passenger service quality and to evaluate the benefit of airport expansion projects. Demand forecasting is the process of estimating the expected number of travelers on each flight in the schedule, given the flight schedules of all competitors in the different travel market. The demand forecasting and modeling process has to predict the passenger counts on each flight in the schedule and also estimate the possible changes in demand due to changes in pricing, competition, and so on. The demand forecasting process also estimates the airline share in each city-pair (O-D market). II. LITERATURE REVIEW According to economic theory, price plays a major role in demand. This is also true in air travel demand models, although some studies omitted this factor due to data unavailability or econometric issues. From a traveler s perspective, the full price (total monetary costs) of an air alternative may include the air fare and the access costs-for instance, paying for the transit ticket or parking fees- for the alternative. Since air fare is usually the dominant component of these costs, especially for long markets, it is used to capture the effect of price on air route demand. This variable is measured by the premium economy fare of the airline in constant naira. The air fare variable may be endogenous, because of supply and demand simultaneity and/or omitted variables. As a result, the coefficients estimated by ordinary least squares (OLS) method may be biased. In air travel demand models, 2321-3264/Copyright 2016, IJRMST, August 2016 27
the fare coefficient is more likely biased towards zero [5]. Although the access costs may also affect travelers decisions on routes, particularly for the airport choice in multiple airport systems, this research does not explicitly specify the access cost variables in the model mainly due to the data availability. Totally omitting the access cost variables may affect the estimated coefficients of other specified variables if (1) the impacts of access cost on route choice is substantial, and (2) the omitted and specified variables are correlated. For example, a route starting from an airport closer to a city center may have higher air fare than a route starting from an airport far from the city center, all other factors being equal. If this is the case, air fare is negatively correlated with access costs. Since access costs are expected to have negative impacts on demand, the estimated coefficient of air fare is expected to be biased towards zero, if the model excludes access cost. In this research, the effects of access costs are implicitly captured by the fixed effect dummy variables, such as time and (origin and destination) airport dummy variables. A direct demand estimation was done using ordinary least squares (OLS). It takes into account usual drivers affecting passenger air transportation demand. Literature provides many studies where determinants of air travel demand are investigated and methodologies to assess their influence are proposed. Important works in this field include, among others: [2], [6], [10], [7], [1], [4]. Reference [10] carried out a survey on the state-ofthe-art of the research in the estimation of transport demand price elasticity. After a theoretical introduction of the concepts of elasticity, a survey on price elasticity of demand for various transportation modes is presented and the influence of some factors on demand is discussed. It turns out that demand for business travel is less elastic with respect to prices than demand for leisure travel and that price elasticity estimates from cross-section data generally are higher than those from time-series data. Reference [1] provide an econometric analysis of international air travel demand in Saudi Arabia. As explanatory variables they consider only macroeconomic and demographic indicators and a detailed description of the steps followed for the development of the model is given. Results suggest that population size and total expenditures are the main determinants of international demand in Saudi Arabia. Reference [4] present a meta-analysis of the price elasticity estimates of demand for passenger air travel. After a description of determinants of demand for passenger air transport, they carry out a comparative reevaluation of previous research on price elasticities for passenger air transport. They find an overall demand mean price elasticity of -1.146 with passengers becoming more price sensitive over time. Business passengers show lower price sensitivity, with an average price elasticity of -0.8. Passengers are becoming more price sensitive over time. Applying the Elasticity Estimates and Multipliers in Table 1 provides a guideline for the estimated fare demand elasticity by level of aggregation and by region. It multiplies the estimate for the relevant level of aggregation by the relevant short and geographic elasticity multipliers. TABLE 1: ESTIMATED FARE ELASTICITIES OF PASSENGER DEMAND a N Ame rica a Eur ope a- Asia a Sub- Saha ran Afri ca a S Ame rica Tra ns- Atla ntic Tra ns- Paci fic Eur ope- Asia Route/Market level National level Supranational level -1.5-1.4-0.9-0.8-0.7-0.6-2 -2-1.2-1.1-0.9-0.8-1.5-1.3-0.8-0.8-0.6-0.6-0.9-0.8-0.5-0.5-0.4-0.4-1.9-1.8-1.1-1 -0.8-0.8-1.9-1.7-1.1-1 -0.8-0.7-0.9-0.8-0.5-0.5-0.4-0.4-1.4-1.3-0.8-0.7-0.6-0.5 Compiled from: IATA Economics Briefing No 9: Air Travel Demand. IATA, April 2008 Later, reference [5] presented two models for air passenger volume forecasting between city-pairs in Nigeria. The models used mainly geo-economics variables rather than service related factors. The model moderately showed a good fit to the observed data which contains 2954 quarters- all quarters between year 2009 and 2013. This model assumes that the marginal effects of each variable on demand are not constant but depend on both the value of the variable and the values of all other variables in the demand function [3]. In other words, the explanatory variables affect demand in multiplicative manner. 2321-3264/Copyright 2016, IJRMST, August 2016 28
Partial derivation of any independent variable proves aforementioned relationship. However, this model was made suitable for multiple regressions by applying logarithmic transformation [11]. It is obvious that interdependency is resolved in this form so that multiple regression model can be applied. Following the demand function used by [3], this study specifies the demand for air travel service as a function of geo-economics and industry related factors [11]. III. ESTIMATION OF DEMAND ELASTICITIES WITH RESPECT TO FARES Demand elasticities with respect to different variables, among which fare is particularly of interest, are calculated and discussed first. The following tables present the elasticity estimates of selected airlines in Nigeria. Elasticity is a useful tool in demand analysis. As a result, many estimates of air travel demand elasticities, especially those with respect to fare, can be found in the literature on transportation. Comparing demand elasticities from our models to previous estimates helps us assess model validity. Elasticity, since it is dimensionless, also provides a convenient way to compare the relative importance of causal factors. This is particularly useful for log-linear models, since the estimated coefficients are elasticity values. A direct demand estimation was done using ordinary least squares (OLS). The estimated parameters of models (Table 2) are values of fare elasticities in the sample. In addition, the estimated elasticities with respect to fare and income are compared with their counterparts in the literature. The demand elasticity with respect to a variable is determined by calculating the percentage change in demand resulting from one percent increase in the variable, holding all other independent variables fixed. This method is used to find route demand elasticities with respect to fare. Fare elasticities of route demand are summarized in Table 2. Since potential travelers have more choices at route, fare elasticities of route demand are expected to be larger (in absolute values). While the fare elasticities calculated from OLS estimates, are consistent with the expectation. In addition, when market size (measured by the number of passengers) is taken into account, the elasticities generally become smaller in absolute values. Details of these elasticities are discussed by disaggregation level below. TABLE 2: FARE ELASTICITY OF VARIOUS AIRLINES Fare Remark Airline Elasticity ARIK -0.151 inelastic AERO -0.240 inelastic ASSOCIATED -0.097 inelastic CHANGCHANGI -0.387 inelastic DANA -1.159 elastic IRS -0.357 inelastic OVERLAND -0.280 inelastic AIR NIGERIA -1.740 elastic Aggregate (panel) -0.319 inelastic Source: Computed by authors The fare elasticities can be further investigated by their distributions and compared with other estimates in the literature. The fare elasticities from the OLS estimates indicate inelastic market demand except for Dana and Air Nigeria which showed elastic values. This indicates that fare elasticities of low traffic markets are higher than those of high traffic markets. A possible reason is that current fares in the low traffic markets are relatively high. Thus, a proportional fare increase reduces more service attractiveness in these markets. Direct comparisons of estimates from literature and this research cannot be made because most fare elasticities available in the literature are estimates for air market demand or for airline demand. However, some guidelines for ranges of fare elasticities are available. One would expect that elasticities of route demand should be larger (in absolute values) than those of market demand, since people generally have higher flexibility in air routes as long as they can arrive their destinations, and changing to other modes or trip cancelations are less likely to be their choices. Summarizing from the literature on air market demand, reference [8] reported that the medians of the fare elasticities for different trip lengths and trip purposes range from -0.70 to -1.52. The fare elasticities of route demand from the OLS estimates are, like those of market demand, too low- most of them are smaller (in absolute values) than those of market demand from reference [8]. For example at the market aggregated level, the elasticities of -0.062 and -0.319 were computed at Model 1 and Model 2 respectively [5]. Fare elasticities of route demand are comparable to some degree to elasticities of airline demand (in a particular market) reported in the literature. First, for monopolistic routes, route demands are equivalent to airline demands. For example, airlines serving the same market generally offer competing routes each connecting at their respective hubs, so that each route corresponds to one airline. When airlines compete with each other on the same routes, elasticities of airline demand should be higher than those of route demand. This may be offset, however, by airline brand loyalty (e.g. due to frequent flyer programme or low carrier model operated by airline such as Aero Contractor in Nigeria), which reduces airline demand elasticities. The fare elasticities of route demand from the OLS estimates from various airlines in Nigeria shown in Table 2 are consistent with these expectations. For Dana and Air Nigeria airlines, the estimated absolute fare elasticity is larger than those of market demand, and close to smaller than those of airline demand, 2321-3264/Copyright 2016, IJRMST, August 2016 29
compared with the estimates of [10]. Reference [10] estimated fare elasticities of market demand and computed airline specific fare elasticities using the estimated conduct parameters. The medians of their fare elasticities are -1.54 for market demand and -2.99 for airline demand. IV. SENSITIVITIES TO FARE AND MARKET DISTANCE As shown in Table 3, OLS estimation generally yield the elasticities of market demand with some sensitivities with distance. Distance effects from the OLS estimates for markets with similar distance may vary, since they also depend on inclusive values, which represent service levels of air routes. Demand elasticities with respect to market distance can help to understand the distance effects of individual markets. Three main generalizations emerge from the Table 3. First, markets with distance less than 900 km have positive fare elasticities. That is, for two markets with distance less than 900 km, the longer distance market is expected to have higher sensitivities with air travel demand with respect to fare, all else being equal. Second, for markets with distance longer than 900 km fare elasticities are negative in line with theoretical construct. This indicates that the influence of propensity to travel, as opposed to mode shift, is more likely to be observed in longer markets. Third, within each distance category the percentage of passengers with positive fare elasticities is higher than the percentage of markets with negative fare elasticities. This implies that higher fare elasticities are more likely to be found in markets with long with connecting flights possibilities as shown in Table 4 with a value of - 2.932. All else being equal, while the influence of declining propensity to travel is more pronounced in better served markets, the influence of mode competition is stronger in minor markets. TABLE 4: PANEL DATA ESTIMATION RESULTS- MODEL 2 ESTIMATES WITH ROUTING BIAS Direct Connecting Variable Flight Flight Frequency (flights per quarter) Scheduled Fight Time (minutes) On-time Performance (minutes per flight) 1.352*** [0.013] 1.878*** [0.148] -0.176 [0.193] Fare (in Naira) -0.324*** [0.097] Income (measured in GDP per 0.147** capita in Naira) [0.057] Constant 7.360*** [1.045] 0.774*** [0.116] 0.495 [3.968] -3.127 [3.172] -2.932*** [0.842] 0.281 [0.285] -6.829 [6.032] R 2 0.823 0.384 Adjusted R 2 0.822 0.322 F 1001.840 6.133fd 1. Model 2: Dependent variable = In (Passenger-Kilometers); 2. Standard errors in brackets are robust to heteroskedasticity and serial correlation; 3. * p < 0.05, ** p < 0.01, *** p < 0.001; Statistics of the first stage. V. CONCLUSION Many factors may cause this wide range of elasticity estimates: failure to consider some specification problems, neglecting intermodal competition, data used, variables definitions, sample period and the variety of models used in estimation and their shortcomings. The literature reveals different behavior by different types of travelers (business and leisure travelers), by trip distance (short, medium and long trips) and by destination whether domestic or international trips. The estimated demand model should distinguish between these distinct market s segments, and estimate the fare elasticity separately for each segment. The estimated fare elasticity of each market segment will be more precise and reliable than the overall fare elasticity of the air travel market. Also, the model used should be built on solid theoretical basis, and yield consistent model estimation when using aggregate data or household data. The distributions of the fare elasticities clearly show that estimation method (OLS) create much larger differences of fare elasticities than the direct liner model form do. At market level, the fare elasticities from the OLS estimates indicate inelastic market demand. REFERENCES [1] Abed, S. Y., Ba-Fail, A. O., and Jasimuddin, S. M. (2001), An econometric analysis of international air travel demand in Saudi Arabia, Journal of Air Transport Management, Vol.7, 143 148. [2] Abrahams, M. (1989), A service quality model of air travel demand: An empirical study, Transportation Research, Part B, Vol. 23B, 213-223. [3] Adebayo, A. J. (2010), Demand for Air Transport in Nigeria, International Journal of Economics, 1 (1): 23-31 [4] Brons, M, Pels, E., Nijkamp, P. and Rietveld, P, (2002), Price elasticities of demand for passenger air travel, Journal of Air Transport Management, Vol. 8, 165-175. [5] Ejem, E. A. (2014). Estimating Air Travel Demand in Nigeria. An Unpublished Ph.D thesis submitted to the Post Graduate School, Federal University of Technology, Owerri, Imo State, Nigeria. [6] Fridstrom, L., H. Thune-Larsen, H. (1983), An econometric air travel demand model for the entire conventional domestic network: the case of Norway", Transportation Research, Part A Vol. 17A, 385-393. [7] Ghobrial, A., Kanafani, A. (1995), "Quality-of-Service Model of Intercity Air-Travel Demand", Journal of Transportation Engineering, Vol.121, 135-140. [8] Gillen, Morrison and Stewart (2002), Air Travel Demand Elasticities: Concepts, Issues and Measurement.Jorge- Calderon, J. D. (1997), A demand model for scheduled airline services on international European routes, Journal of Air Transport Management, Vol.3, 23-35. [9] Tae Hoon Oum, W. G. Waters 1I, Jong-Say Yong, (1992), Concepts of price elasticities of transport demand and recent empirical estimates an interpretative survey, Journal of Transport Economics and Policy, Vol.26, 139-154 and 164-169, 1992 2321-3264/Copyright 2016, IJRMST, August 2016 30
[10] Oum, Zhang, and Zhang (1993), Inter-Firm Rivalry and Firm-Specific Fare Elasticities in Deregulated Airline Markets., Pergamon Press, Oxford. [11] Sivrikaya, O. and E. Tunç, E. (2013), Demand Forecasting for Domestic Air Transportation in Turkey, The Open Transportation Journal, 2013, 7, 20-26 TABLE 3: PANEL DATA ESTIMATION RESULTS FOR AGGREGATE AIR TRAVEL DEMAND WITH DISTANCE BIAS 601- Variable 0-300KM 301-600KM >900KM Panel 900KM Frequency (flights per quarter) Fare (in Naira) 1.357*** [0.030] 0.419* [0.188] 0.212 a [0.118] 1.351*** [0.013] 1.296*** [0.045] 0.649 a [0.361] 0.832*** [0.140] -0.299 [1.182] 1.337*** [0.013] -0.319*** [0.048] R 2 0.810 0.877 0.676 0.620 0.806 Adjusted R 2 0.806 0.876 0.667 0.579 0.805 F 204.886 974.309 81.073 12.401 871.302 1. Model 2: Dependent variable = In (Passenger-Kilometers); 2. Standard errors in brackets are robust to heteroskedasticity and serial correlation; 3. * p < 0.05, ** p < 0.01, *** p < 0.001; Statistics of the first stage. a Parameters estimates for the income variable are significant at p <0.100 2321-3264/Copyright 2016, IJRMST, August 2016 31