Amid growing concerns about climate change, the airline industry’s carbon emissions have gained increasing attention. The industry’s CO2 contribution is not very large in percentage terms, accounting for around 2.5% of total CO2 emissions as of 2006 (Kwan and Rutherford, 2015), but the impact per kilogram of its highaltitude emissions on climate change is about double that of ground-level emissions, as explained by Lee et al. (2004). Recognizing the importance of airline emissions, the European Union in 2012 formulated a controversial plan to require all airlines serving EU airports to acquire emissions allowances under its emissions trading system (ETS). But in the face of substantial opposition, this plan was never fully implemented, with the required charges currently limited to Europe’s own airlines (see Albers et al., 2009, and Scheelhasse and Grimme, 2007). Following on the EU’s efforts, the UN’s International Civil Aviation Organization in 2016 proposed two emission-reduction programs with an international scope. The first is an explicit fuel efficiency standard for new aircraft, intended to take effect in 2028 (Mouawad and Davenport, 2016). Whileenvironmentalists were disappointed by this plan, which would affect airline emissions only gradually (Mooney, 2016), the ICAO subsequently announced a carbon-offset program, titled Carbon Offsetting and Reduction Scheme for International Aviation (CORSIA). Under this program, airlines are required to purchase carbon offsets (sold by other entities undertaking carbon-reduction activities) to compensate for increases in their carbon emissions beyond a 2020 baseline. Although structured differently, this plan is equivalent to requiring the purchase of allowances under an ETSstyle system.1

Given the new focus on airline emissions, it is important to gain a better understanding of the link between the emissions volumes of individual airlines and the characteristics of their fleets and flight operations. Since emissions are approximately proportional to fuel consumption, this link can be studied by exploring the connection between an airline’s total fuel usage and its characteristics. The present paper carries out such an exploration. Using annual data on individual US airlines over the 1995e2015 period, the paper presents regression results relating an airline’s total fuel usage to seven variables: the available ton miles of capacity (passengers plus freight and mail) provided by the airline; the average seat capacity of its aircraft, average stagelength (flight distance); average load factor (measured by weight); the average vintage (construction year) of its aircraft; the percentage of the airline’s flights that are delayed; and the average annual fuel price. The regression model containing these variables is generated from a theoretical framework, which has elements in common with the earlier model of Brueckner and Zhang (2010). 2

ang (2010). 2 The results give answers to the following questions: How would an airline’s fuel usage (and thus emissions) respond to modernization of its fleet? To a shift to longer flights? To a reduction in flight delays? To fuller planes? In addition, the estimated effect of the fuel price on fuel usage captures the impact of greater fuel-conservation effort, holding fixed the characteristics of the airline’s fleet and route structure. While the fuel price effect thus only captures shortrun usage adjustments, its magnitude can be used to predict the short-run impact of airline emission charges, an exercise that is carried out in the paper.