National Academies Press: OpenBook

Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies (2019)

Chapter: Chapter 3 Sources of Disaggregated Socioeconomic Data

« Previous: Chapter 2 Survey of Past Analyses of Air Passenger Demand
Page 25
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 25
Page 26
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 26
Page 27
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 27
Page 28
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 28
Page 29
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 29
Page 30
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 30
Page 31
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 31
Page 32
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 32
Page 33
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 33
Page 34
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 34
Page 35
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 35
Page 36
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 36
Page 37
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 37
Page 38
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 38
Page 39
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 39
Page 40
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 40
Page 41
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 41
Page 42
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 42
Page 43
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 43
Page 44
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 44
Page 45
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 45
Page 46
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 46
Page 47
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 47
Page 48
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 48
Page 49
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 49
Page 50
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 50
Page 51
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 51
Page 52
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 52
Page 53
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 53
Page 54
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 54
Page 55
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 55
Page 56
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 56
Page 57
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 57
Page 58
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 58
Page 59
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 59
Page 60
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 60
Page 61
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 61
Page 62
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 62
Page 63
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 63
Page 64
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 64
Page 65
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 65
Page 66
Suggested Citation:"Chapter 3 Sources of Disaggregated Socioeconomic Data." National Academies of Sciences, Engineering, and Medicine. 2019. Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies. Washington, DC: The National Academies Press. doi: 10.17226/25411.
×
Page 66

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

25 C H A P T E R 3 Two broad categories of disaggregated socioeconomic data are especially important for modeling and forecasting air passenger activity at airports or larger regions. This chapter is a summary of these data and their sources. A more detailed discussion is available in Appendix C available on the TRB website. The first sort of disaggregated socioeconomic data includes the distribution of socio­ economic characteristics within a national, regional, or airport catchment area population. This can include age distributions, income distributions for households or individuals, levels of educational attainment, identification of ethnic or racial cohorts, and others. Cataloging such characteristics for a population provides not only a richer picture of that population, but also important information for businesses and public service organizations for understanding how to best serve (or market towards) that population. However, it is important to bear in mind that this second source of value from awareness of the distributions of socioeconomic characteristics within populations can truly provide this value only if these different cohorts exhibit different tastes and preferences for goods and services. The second sort of valuable disaggregated socioeconomic data in the context of passenger aviation is do different socioeconomic cohorts tend to travel by air more or less than other cohorts? An important way to collect this type of data is the passenger or consumer survey. In these surveys, interviewers capture the socioeconomic data of the respondent (age, annual household income, educational background, ethnic, racial, or gender identification, and so forth) and (in the case of an air travel survey) information about the respondent’s use of air travel (how frequently air trips are taken, travel party size, mode of transportation used to travel to the airport, destination information, use of business or economy class, use of airport terminal facilities and amenities, and others). An effective survey of this type can provide an airport or a regional transportation planning organization with a nuanced picture of the air traveling public it serves. These data may be useful in the near term because they provide airport managers with information about the preferences and expectations of those using the airport. The survey data may also identify differ­ ences in these tastes, expectations, and propensities to fly between socioeconomic cohorts. With a survey­based picture of these differing cohort preferences and an understanding of how the distribution of cohorts is expected to change over time, airport and other aviation analysts may be able to develop more informative models of air passenger demand, and better forecasts of future passenger activity, compared to those results using aggregate socioeconomic data alone. This chapter summarizes sources of disaggregated socioeconomic data about communities and regions served by airports and recent socioeconomic trends, which provide information about how distributions of socioeconomic traits appear to be changing over time. It also pres­ ents an analysis of several air passenger and travel survey efforts that have provided data on how Sources of Disaggregated Socioeconomic Data

26 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies different socioeconomic groups choose to travel by air, contributing to the understanding of how general societal characteristics are distributed within the air traveling public—which is also the airport using public. Future research needs that could help improve and focus the use of disaggregated socioeconomic data in the analysis of air travel demand also is discussed. Availability of Disaggregated Socioeconomic Data This section presents an overview of sources for socioeconomic data. Aggregate national socioeconomic data is reported by federal sources and includes such variables as population, GNP, price indices (such as CPI), employment and unemployment levels and rates, and measures of economic activity in specific sectors of the economy. These data are reported on an annual, quarterly, and sometimes monthly basis. These aggregate data are compiled by specific government agencies and departments; data for employment statistics, household incomes and price indices are gathered by the BLS (part of the Department of Labor), GDP data are gathered by the BEA (part of the Department of Commerce), and population data is developed by the Bureau of the Census (also part of Commerce). Historical values for these aggregate national measures are collected and reported by a number of government organizations with economic responsibilities, such as the President’s Council of Economic Advisers and the Board of Gov­ ernors of the Federal Reserve. These historical data series are readily available at the websites of these and other agencies. The agencies that originate data for these socioeconomic variables also report the data for regional, demographic, and socioeconomic subsections of the country. These federally pro­ vided data are free to the user and available for download on the internet. In recent years, the level of detail reported by these federal data sources has grown and expanded as the ability of the internet to host larger and richer databases has improved. The availability of a greater breadth of socioeconomic data has made the processes used to identify and use more refined datasets more complicated, but the agency webpages provide explanatory and largely user­ friendly support. Important examples of the categories of data that are available, and their sources, include • Population: The Bureau of the Census 2010 Census Tool is an interactive tool that allows the user to examine and extract data on population size and characteristics (age, ethnicity, and household composition) at the county or parish level for all states. The provision of these data at the local level makes the Bureau of the Census a fundamental source of many categories of disaggregated socioeconomic data, especially demographic data. Interactive Population Maps and other data tools make it possible to examine relatively complex population dynamics, such as migration patterns across counties in the United States. These capabilities represent the application of recently developed visualization tools to recently collected population and demographic data, which limits their availability to past Census data in many ways. The Bureau of the Census web pages also provide detailed instructions for users regarding data availability, data acquisition, and data interpretation. Bureau of the Census databases can be browsed and searched at http://www.census.gov/data.html and data that is developed from the most recent 2010 Census can be accessed at http://www.census.gov/2010census/data/. • Employment: BLS maintains an extensive online database for employment and compen­ sation for 100 industries at the national, state, county, and local (selected) levels. The BLS databases also report values and indexes for consumer and producer prices, as well as data on labor productivity, occupational trends, and consumer expenditure patterns. Like the demographic data available from the Bureau of the Census for population characteristics, the BLS data provides a detailed picture of labor market conditions in aggregate and by industry sectors. BLS data can be accessed at http://www.bls.gov/data/. Much of the BLS regional data

Sources of Disaggregated Socioeconomic Data 27 is available at links related to each of the eight BLS Regional Offices, which can be found at http://www.bls.gov/data/#regions. • Economic Activity: BEA develops and maintains databases for GDP, personal income and personal consumption expenditures at the national, state, county, and MSA levels. GDP esti­ mates for individual industries are provided at the state level, beginning in the early 2000s. Explorations of these data can be fine­tuned and focused by the user and extracted using BEA’s online interactive data tools. BEA’s national, industry specific, international accounts, and regional data can be accessed at http://www.bea.gov/itable/index.cfm with an additional set of pages devoted to regional economic data available at http://www.bea.gov/regional/. • Economic and Socioeconomic Data: Another source of socioeconomic data is the Federal Reserve System. The Federal Reserve tracks socioeconomic data as part of its management and oversight of the nation’s financial system. The board of governors of the Federal Reserve maintains databases on financial system economic data, and these data may be of limited use for air passenger demand studies. Each Federal Reserve regional bank, however, is a source for regional economic and demographic data at the state and MSA level. The website of the Federal Reserve Bank for each District (listed with links at https://www.federalreserve.gov/ otherfrb.htm) includes a section on Research and Data where extensive data about the states and metropolitan areas within a district can be downloaded (although most of these data come from the data services of the Bureau of the Census, the BLS, or the BEA described above). While there is similar content in the pages of each District bank, the websites are not identical from district to district. Finally, the Federal Reserve Bank of St. Louis maintains a uniquely comprehensive collection of socioeconomic data, the Federal Reserve Economic Data, or FRED. These data cover economic variables in the United States and the international economy, and include more than 350,000 individual data series. As with the other federal data sites described here, FRED data can be downloaded, and it is also possible to create new data vari­ ables and data graphs. Accessing FRED data requires the creation of a user account, which can be done at https://research.stlouisfed.org/fred2/. Many of these aggregate socioeconomic data are also provided at metropolitan, regional and state levels by the individual states at demographic data pages hosted by the states themselves. These state level data pages typically, though not in all cases, report data taken from federal data sources, presenting the data from the perspective of the individual state. Some regional organizations, such as councils of governments and metropolitan planning organization areas also maintain online databases of regional information and data, including demographic and socioeconomic data, also typically but not always derived from federal data sources. While they do provide data at no cost, the federal, state and local data sources identified here may be cumbersome or difficult to use in some cases or for some users, depending on the complexity or specificity of the data of interest. However, there are commercial firms that provide access at a cost to these local and regional socioeconomic data, in what is often a more user­friendly format. These commercial providers also frequently provide projections of future values for some variables. Having such projections of socioeconomic variables is necessary for developing forecasts of air passenger demand from econometric models that estimate historical relationships between these socioeconomic data elements and air passenger activity, such as annual enplanements. A widely used commercial private provider of socioeconomic data is Woods & Poole Eco­ nomics, Inc. Woods & Poole provides historical data and future projections for population, income, employment, retail sales and households for states, regions, metropolitan and micro­ politan statistical areas, and the nation as a whole. These data are downloadable and are provided at a cost to the user. Data from Woods & Poole were used for the case study analysis described below.

28 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies The Woods & Poole socioeconomic data for a given geographical area are reported starting with 1970 values. After the most recent year of actual data (2015 for the case studies conducted as part of this project), smooth projections for future years are calculated by Woods & Poole through the year 2050. An area’s socioeconomic data is reported with several distributional details, such as • Total area population and area population by age cohort, gender, and ethnicity; • Total area households and the numbers of area households within a span of income ranges; • Area mean household income; • Total area employment and area employment by industry or sector; • Total area earnings and area earnings by type of compensation and by industry; • Total area income and income by type of income, area per capita income; • Total area retail sales and retail sales by industry; and • Gross regional product. Socioeconomic Trends The U.S. socioeconomic trends reported in this section show recent changes in some of the disaggregated socioeconomic factors that air travelers in passenger surveys commonly report about themselves or their households. The trends represent an important piece of the puzzle that may link changes in social demographics to changes in air travel demand. A major objective of the research undertaken in this project is improving understanding of these links, the ways in which the future evolution of these socioeconomic trends may influence future air travel demand, and the ways consumers spend on air travel. Demographic Trends and the Distribution of Household Incomes To estimate passenger demand models, either for forecasting future demand or for analyzing and modeling the factors that influence that demand, it is necessary to rely on historical socio­ economic data at different degrees of aggregation. The U.S. Census Bureau is an important source of these historical data, and the online availability of these data has improved in recent years. At the national level, Census sources provide a baseline for demographic and socioeconomic variables that influence air passenger demand, such as household income, revealing national trends in household income stratification along a variety of economic and demographic factors. Figure 2 shows recent trends in household income distribution, reporting the upper limits of each household annual income quintile since 1967, as reported by the U.S. Census Bureau (U.S. Census Bureau 2017). Household income values are adjusted for inflation and reported in 2016 dollars. Shown are the inflation­adjusted household incomes received by the households at the 20th, 40th, 60th, 80th, and 95th percentiles in the national income distribution. For U.S. households at the 20th percentile of the distribution, annual income in 2016 increased by 27% since 1967, from $18,856 to $24,002, and households at the 40th percentile had 2016 incomes of $45,600, a 24% increase from the inflation adjusted level of $36,768 in 1967. The income received by U.S. households at the 60th percentile was 47% higher in 2016, having grown to $74,869 from its inflation­adjusted 1967 level of $52,186. Similarly, the income received by U.S. households at the 80th percentile was 63% higher in 2016, having grown to $121,018 from its inflation­adjusted 1967 level of $74,417. Finally, the income received by U.S. households at the 95th percentile was 89% higher in 2016, having grown to $225,251 from its inflation­ adjusted 1967 level of $19,419. This income distribution trend expresses a slowly but steadily growing separation in incomes across the economy, with the real incomes of higher income households growing more rapidly than those of households with lower incomes. This trend,

Sources of Disaggregated Socioeconomic Data 29 which in this form and others has been widely discussed and examined as evidence of growing income inequality in the United States, may also have important implications for understand­ ing trends in air passenger demand to the extent that those with higher incomes have a greater propensity to use air transportation. These national trends in income distribution can be related to other socioeconomic and demographic trends that can also be depicted with data from the Census Bureau. Figure 3 shows the number of U.S. households headed by someone in the reported age ranges for each year since 1960 (U.S. Census Bureau 2017). An important trend in these data is the increasing aver­ age age of heads of households. In 1960 around 45% of households were headed by someone aged 44 or younger. By 2016 this had fallen to less than 38% of households, with over 62% headed by someone 45 or older. It is possible to see the “baby boom” generational bulge (people Figure 2. Quintile distribution of annual household income, 1967 to 2016 (Source: U.S. Census Bureau). Figure 3. U.S. Households by age of head of household, 1960 to 2016 (Source: U.S. Census Bureau).

30 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies born between 1946 and 1964) moving through Figure 3, where they first show up as the bulge beginning around 1971 in households headed by someone between 25 and 34. This bulge moves along in time to other age cohorts, reflecting the aging of the boomers, until they begin swelling the ranks of households headed by someone aged 65 or older in around 2011. This aging trend in the U.S. population is also related to air passenger demand trends. Household incomes customarily increase as the head of household ages, at least until the householder nears the end of his or her work career or has retired. When household income is correlated with the age of the head of the household in this way, it may be difficult to deter­ mine whether a household’s increasing propensity to travel by air should be contributed to its growing income or its increasing age. This correlation will be seen in the analysis of air passenger survey results. Figure 3 illustrates the changing distribution of the age of U.S. heads of households, which can be summarized or simplified as an increase in the average or median age of heads of household. One way this aging trend in the population may affect air travel demand is through the related changes in household income, since an older head of household is likely to be more experi­ enced as a worker, with longer job tenure. Socioeconomic aspects of these changes in the age distribution are shown in Figure 4, which reports trends and recent history in average household income according to the age of the household head. The figure also shows a widening dispersion of incomes across these households, at least through the “age 45 to 54” cohort, at which point the growth in household median income begins to be reversed as some householders retire. Finally, the steady growth in median household income for those headed by someone 65 or older, although at lower levels of annual income, is in striking contrast with the household income trends for households headed by persons younger than 25. For these households, there was little if any growth in median household income since 1967. These figures come together to tell a somewhat complicated story about how disaggregated socioeconomic factors can contribute to air passenger demand. Analysis of passenger surveys may conclude that wealthier households tend to travel by air more frequently and also that at least up to a point, households headed by a person in middle age (from 45 through 64) may also be more likely to travel by air. Since these two socioeconomic factors, when reported in a Figure 4. U.S. Household median income by age of head of household, 1967 to 2016 (Source: U.S. Census Bureau)

Sources of Disaggregated Socioeconomic Data 31 more disaggregated way are clearly positively correlated with one another, it may not be easily understood whether the observed increased propensity to travel by air is due to the household’s greater wealth or its greater age. Other similarly correlated disaggregated socioeconomic characteristics, such as educational attainment and household income, present similar analytic challenges for understanding and modeling how individual socioeconomic characteristics contribute to passenger demand for air travel. Analysis of Air Passenger and Travel Survey Data Surveys of air passengers and households provide the most disaggregated information on the way the use of air travel varies with household and air traveler characteristics, since the response data to these surveys provide information on air travel and respondent or household characteristics for individual travelers or households. The extent to which such surveys can be used to study how air travel varies with traveler or household characteristics obviously depends on what questions are asked in any given survey. Surveys that ask how many air trips a respondent or household made in a previous year, or some other long period, allow the survey responses to be used to analyze how air travel propen­ sity (expressed as air trips per year) varies with those respondent characteristics reported in the survey. When surveys ask how many air trips respondents have made in a period shorter than a year, care is needed in converting these air trip rates to an annual basis, since many people make relatively few air trips per year. The fact that a household reported no air trips in the previous three months does not mean that the household had not made any air trips in the previous year. Other difficulties can arise from the way that survey questions are worded, such as whether respondents are asked how many air trips they have made or all members of their household have made. In the case of air passengers surveyed at airports, obviously the respondents have made at least one air trip in the past year (the trip they were making when they were surveyed). However, such surveys will not include anyone who has not made an air trip in over a year (or ever). Only household surveys will collect data on those who have not made an air trip in the past year. In order to collect data on infrequent air travelers, it would be desirable to ask those who report not having made an air trip in the past year when they last made an air trip, if ever. However, this is rarely done. It would also be helpful for surveys to ask how many of the air trips that a respondent reported having made in the previous year were made for business purposes and how many for personal purposes. These data can then be used to compare the responses to the trip purpose proportions reported for the current trip by air travelers surveyed at airports. Unfortunately, this too is rarely done. Of course, many air passenger surveys performed at airports are done for purposes other than understanding the determinants of air travel demand, such as to measure traveler satisfaction with airport facilities and services or to collect information on ground transportation mode use. Even so, given the cost of performing such surveys, it would make sense to also use them to help improve the understanding of the factors that determine air travel demand. Similarly, household travel surveys are usually primarily interested in local travel rather than long­distance travel in general and air travel in particular. Although adding questions about long­distance travel does increase the length and complexity of the survey, the information that this provides significantly increases the value of the survey. This is particularly valuable since household surveys typically collect more data on the household characteristics than air passenger surveys at airports and also include households that rarely or never make air trips, as noted above.

32 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies The current project analyzed response data from a range of surveys of various types: • General socioeconomic household surveys – CES • Air passenger surveys – Airport intercept surveys – SIAT – Online surveys of air travelers • Household travel surveys – National household travel surveys – State household travel surveys Consumer Expenditure Survey The Public­Use Microdata (PUMD) dataset of the U.S. Bureau of Labor Statistics CES includes detailed disaggregated socioeconomic information at the household and/or individual level, as well as data on household expenditures in a wide range of categories, including travel. The travel data include information on airfare expenditures. A significant attraction of the CES data is that it is national in scope, with annual data going back many years using a consistent survey methodology. Thus, these data can be used to explore whether the relationships among air­ fare expenditures and household characteristics have been changing over time. Although the CES survey only covers consumer expenditures, and therefore excludes reimbursed travel or business travel paid directly by employers, personal travel has been an increasing proportion of all air travel and currently accounts for about two­thirds of all air trips. However, the survey also covers non­reimbursed expenditures on trips largely paid for by employers or others, so some business­related travel is included. In addition, respondents report the number of fully reimbursed air trips that they have taken, although the survey does not collect expenditure data for those trips. The actual data collection for the CES is performed for the BLS by the U.S. Census Bureau. The CES data is collected in two ways: a quarterly telephone interview of selected households that asks about detailed household characteristics and major expenditures over the previous quarter and a more detailed diary survey that collects expenditure data recorded by the sample households over a one­week period using an expenditure diary provided by the Census Bureau together with detailed household characteristics of the participating households. The households par­ ticipating in the two different surveys are not the same. Data on air travel and vacations are primarily collected in the quarterly interview survey. A given household will generally be surveyed five times in successive quarters and report expenditures in the last four interviews. Each interview covers household expenditures in the previous three months. Since house­ holds are interviewed throughout the quarter, the reported expenditures are not always for a calendar quarter. In addition, households rotate through the panel with new households being added in each quarter and those that have completed five quarterly surveys (or would have if they had been reached in each quarter) are dropped. Thus a given household’s four quarters for which they report expenditure data are not necessarily for a calendar year. For the purpose of analyzing the number of reported air trips, only households that reported four quarters of data were included. Data for a given year included interviews conducted during the first quarter of the following year, since many of those interviews would have included air trips made during the fourth quarter of the previous year. The CES expenditure files provide information on each air trip or vacation taken by a household, including the number of trips and the duration of each trip, as well as the airfare

Sources of Disaggregated Socioeconomic Data 33 (and other) expenditures for each trip. This allows expenditures on air travel to be expressed in terms of the average number of air trips involved, which can then be compared to the data on air trips from the U.S. DOT 10% O&D survey. Thus the CES data provide a complementary perspective to more traditional sources of air passenger data that considers air travel expenditures rather than air trips per se. In addition to the expenditures on airfares, the CES expenditure files also include the expenditures on a range of other travel cost categories, including lodging, meals in restaurants, alcoholic beverages in restaurants and bars, purchased food and beverages, and local trans­ portation. Since most people do not keep detailed records of their expenditures while making personal trips and to the extent that they do have records (e.g., credit card statements) may not take the time to review these records when responding to the survey, it is almost certain that the reported figures are at best highly approximate. Nonetheless, the CES provides one of the few sources of data on the non­airfare costs of trips taken by air, and hence provides a way to study how these costs have changed relative to airfares over time. In order to determine how the results from the CES compare to the air passenger traffic statis­ tics reported by the U.S. DOT, the number of air trips reported by respondents to the 2014 CES was compared to the corresponding number of air trips estimated from the U.S. DOT data. This comparison required a number of assumptions, since there are significant differences between the air trips reported by CES survey respondents and the air passenger traffic data reported by the U.S. DOT, as discussed in Appendix C. The principal difference is that CES respondents reported the number of air party trips made by household members whereas the U.S. DOT data report air passenger trips. The CES data do not indicate how many household members went on each trip. In order to convert the U.S. DOT air passenger enplanement data to air party trips by U.S. residents, it was necessary to adjust the data for the average air party size, as well as the average number of enplanements per round trip, and the percentage of U.S. domestic enplanements made by foreign visitors. In summary it appears that the CES survey is missing at least half of the air trips being made by all U.S. households and quite possibly as much as 60% for reasons that are not clear. It is possible that CES respondents failed to report all the air trips made by members of their household. However, although the overall air trip propensity values from the CES should be viewed with some caution, the relative air trip propensities for different segments of the popula­ tion may still be valid. Trends in Air Travel by Household Characteristics To examine changes of air trip making and household characteristics over time, the same analysis was undertaken of the relevant data from the 2006, 2010, and 2014 CES. Household Income. The changes in the average number of air trips per household for survey respondents in different household income ranges are shown in Table 2, together with the percentage of respondents in each income range who reported making no air trips during the four quarters for which they reported expenditures. As could be expected, the average number of air trips per year generally increased with increas­ ing household income in each survey. However, the average number of air trips across all house­ holds declined progressively from 2006 to 2014, although this pattern does not appear for every income range, with some showing an increase or decrease from 2006 to 2010 that reversed in 2014 or even a progressive increase. There appears to be no obvious pattern in the changes in the average number of air trips for a given household income range, other than a progressive increase in the average number of air trips for survey respondents in households with an annual income of $300,00 or more.

34 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies The percentage of households that reported no air trips in the past year declines with increas­ ing income, as could be expected. Surprisingly, about 10% of households with incomes less than $20,000 reported making at least one air trip in the past year, with those households reporting an average of about 1.4 trips in 2006 and 2010 and 1.7 trips in 2014. In contrast, only about 25% of households with incomes of $300,000 or more reported making no air trips in the past year, with those households making air trips reporting an average of 2.6 trips in 2006, 2.7 trips in 2010, and 4.0 trips in 2014. However, the results for respondents in the higher income ranges may be influenced by the relatively small number of survey respondents in these income ranges, as shown in Table 3. The distribution of survey respondents across the different income ranges for each year shows a small progressive decrease in the percentage of survey respondents in households with an annual income in each range below $80,000. This is to be expected as real incomes increase, moving some households into a higher income range. However, there is no obvious pattern in the changes in the distribution of survey respondents in income ranges of $80,000 and over, although the percentage of respondents in each income range above $100,000 increased from 2006 to 2014 with the exception of respondents with household incomes from $250,000 to $299,999, the percentage of whom was essentially unchanged. Annual Household Income 2006 2010 2014 Percent w/ No Air Trips Air Trips per H/H (All) Percent w/ No Air Trips Air Trips per H/H (All) Percent w/ No Air Trips Air Trips per H/H (All) Less than $20,000 89 0.15 88 0.17 92 0.14 $20,000 - $39,999 84 0.28 88 0.17 88 0.19 $40,000 - $59,999 77 0.38 77 0.40 81 0.32 $60,000 - $79,999 71 0.53 78 0.39 75 0.37 $80,000 - $99,999 61 0.67 64 0.66 66 0.55 $100,000 - $124,999 55 0.68 58 0.78 69 0.55 $125,000 - $149,999 43 1.24 49 1.37 48 0.87 $150,000 - $199,999 42 1.49 50 1.03 47 1.20 $200,000 - $249,999 22 2.03 44 1.40 46 1.46 $250,000 - $299,999 24 1.76 19 2.22 38 1.77 $300,000 or more 25 1.94 24 2.04 26 2.97 Total 72 0.533 75 0.476 76 0.459 Table 2. Air trips per household by annual household income. Annual Household Income 2006 2010 2014 All Households Percent All H/H All Households Percent All H/H All Households Percent All H/H Less than $20,000 496 19.3 489 19.1 397 18.2 $20,000 - $39,999 575 22.3 552 21.5 467 21.4 $40,000 - $59,999 440 17.1 434 16.9 360 16.5 $60,000 - $79,999 335 13.0 333 13.0 268 12.3 $80,000 - $99,999 241 9.4 257 10.0 181 8.3 $100,000 - $124,999 179 6.9 174 6.8 199 9.1 $125,000 - $149,999 87 3.4 119 4.6 83 3.8 $150,000 - $199,999 88 3.4 86 3.4 85 3.9 $200,000 - $249,999 78 3.0 69 2.7 84 3.9 $250,000 - $299,999 32 1.2 34 1.3 26 1.2 $300,000 or more 25 1.0 16 0.6 31 1.4 Total 2,576 100 2,563 100.0 2,181 100 Table 3. Survey respondents by annual household income.

Sources of Disaggregated Socioeconomic Data 35 Since the household income ranges are expressed in current dollars for each year, it is unclear to what extent the changes in the distribution from 2006 to 2014 reflect a shift into higher income categories as real incomes rise versus a shift in the distribution of real income from lower income to higher income households. Respondent Age. The changes in the average number of air trips per household for survey respondents in different age ranges are shown in Table 4. Although the analysis found no consistent pattern in the average number of air trips per household for survey respondents in the individual age ranges, when grouped into three broader age ranges there appears to be a somewhat different pattern for each range. For survey respondents ages 25 to 44, the average number of trips declined from 2006 to 2010, then remained unchanged in 2014. For survey respondents ages 45 to 64, the average number of air trips declined sharply from 2006 to 2010, then recovered slightly in 2014. For survey respondents ages 65 or over, the average number of air trips remained unchanged from 2006 to 2010, but declined in 2014. The percentage of households with respondents who reported making no air trips in the past year was generally similar across the different age ranges up to about age 70 then generally increased in older age ranges, as could be expected. The number of households with survey respondents in each age range is shown in Table 5. There were significantly more respondents in the 20­year age range from 45 to 64 than the 20­year age range from 25 to 44. It appears that this reflects the changing age distribution in the population as a result of the baby boom following the World War II, as can be seen from the progression of age ranges with the highest percentage of respondents into older age ranges over the period from 2006 to 2014. The distribution of survey respondent ages over the three surveys shows the percentage of respondents in the age ranges below age 55 generally declined from 2006 to 2014, except for the age range 18 to 24, where the percentage increased. The percentage of respondents in the age range 55 to 59 declined from 2006 to 2010 but then recovered in 2014, while the percentage of respondents in age ranges of 60 or over generally increased from 2006 to 2014. Respondent Age 2006 2010 2014 Percent w/ No Air Trips Air Trips per H/H (All) Percent w/ No Air Trips Air Trips per H/H (All) Percent w/ No Air Trips Air Trips per H/H (All) 18 - 24 77 0.30 79 0.35 82 0.34 25 - 29 72 0.54 73 0.60 72 0.45 30 - 34 77 0.43 73 0.43 76 0.57 35 - 39 72 0.51 73 0.46 76 0.48 40 - 44 65 0.66 74 0.49 74 0.43 45 - 49 69 0.67 73 0.51 77 0.38 50 - 54 67 0.61 69 0.59 72 0.52 55 - 59 65 0.69 75 0.44 75 0.63 60 - 64 69 0.53 76 0.57 73 0.64 65 - 69 68 0.59 74 0.49 76 0.41 70 - 74 80 0.43 83 0.43 77 0.34 75 - 79 82 0.25 74 0.45 85 0.31 80 or over 88 0.23 90 0.15 88 0.14 Total 71 0.538 75 0.476 76 0.459 25 - 44 71 0.55 74 0.49 75 0.49 45 - 64 67 0.63 73 0.53 74 0.55 65 or over 79 0.38 80 0.38 81 0.31 Table 4. Air trips per household by respondent age.

36 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies The survey findings shown in Tables 4 and 5 have interesting implications for future trends in air travel demand. Table 5 shows that the percentage of the population aged over 64, which make the lowest average number of air trips per household, can be expected to continue to increase while the percentage between ages 45 and 64, which make the highest average number of air trips per household, can be expected to continue to decline. Furthermore, within broad groups of age ranges the average number of air trips per household has declined from 2006 to 2014 despite increases in real incomes and a decline in real airfares over the period. Race and Ethnicity. Differences in the average number of air trips per year per household by respondent race and ethnicity and changes in these air trip propensities over the period from 2006 to 2014 are shown in Table 6. Households with an Asian respondent reported a significantly higher average number of air trips than those with a white respondent. In contrast, households with a Hispanic respondent reported Respondent Age 2006 2010 2014 All Households Percent All H/H All Households Percent All H/H All Households Percent All H/H 18 - 24 61 2.4 66 2.6 67 3.1 25 - 29 151 5.9 130 5.0 122 5.6 30 - 34 194 7.6 196 7.6 169 7.7 35 - 39 254 9.9 217 8.4 168 7.7 40 - 44 283 11.0 232 9.0 169 7.7 45 - 49 289 11.3 298 11.5 182 8.3 50 - 54 289 11.3 293 11.4 236 10.8 55 - 59 271 10.6 254 9.8 229 10.5 60 - 64 211 8.2 255 9.9 226 10.3 65 - 69 165 6.4 202 7.8 215 9.8 70 - 74 110 4.3 140 5.4 146 6.7 75 - 79 121 4.7 138 5.3 103 4.7 80 or over 168 6.5 160 6.2 152 7.0 Total 2,567 100 2,581 100 2,184 100 25 - 44 882 34.4 775 30.0 628 28.8 45 - 64 1,060 41.3 1,100 42.6 873 40.0 65 or over 564 22.0 640 24.8 616 28.2 Table 5. Survey households by respondent age. Respondent Race/Ethnicity 2006 2010 2014 Percent w/ No Air Trips Air Trips per H/H (All) Percent w/ No Air Trips Air Trips per H/H (All) Percent w/ No Air Trips Air Trips per H/H (All) White 69 0.61 74 0.49 73 0.54 Hispanic 79 0.28 83 0.28 86 0.23 Black 87 0.19 88 0.26 90 0.15 Asian 49 0.85 45 1.12 62 0.68 Native American 60 0.70 78 0.22 100 Pacific Islander 50 1.13 57 0.71 40 1.00 Multiple races 63 0.58 69 0.59 77 0.55 Total 71 0.537 75 0.476 76 0.459 NA / PI / Multiple (1) 60 0.71 69 0.54 76 0.52 Note 1. Native American, Pacific Islander, and multiple races combined. Table 6. Air trips per household by respondent race/ethnicity.

Sources of Disaggregated Socioeconomic Data 37 an average number of air trips less than half of that by households with a white respondent in 2006 and 2014, and slightly over half in 2010, while households with a black respondent reported an even lower average number of air trips than those with a Hispanic respondent. In 2006 and 2014, households with a black respondent reported an average number of air trips less than a third of that by those with a white respondent, although in 2010 they reported a significantly higher average number of air trips that was over half that reported by households with white respondents. Households with Pacific Islander respondents reported the highest average number of air trips of all race/ethnicity categories in 2006 and 2014 and the second highest in 2010, although, as noted below, the number of respondents was too small to give reliable results. The changes in the average number of air trips per year across all households participating in the survey for each of the race/ethnicity categories over the three surveys does not show any obvious pattern other than an apparently progressive decrease in the air trip propensity of the combined category of Native American, Pacific Islander, and multiple race respondents. However, that pattern is heavily influenced by differences between the three categories. Changes in the distribution of survey households by respondent race/ethnicity are shown in Table 7. Although accounting for over two­thirds of the households participating in the survey, the percentage of households with white respondents declined progressively over the three surveys. In contrast, the percentage of households with a Hispanic or Asian respondent increased, although the increase from 2010 to 2014 was not a great as from 2006 to 2010. The percentage of house­ holds participating in the survey with black respondents varied, decreasing from 2006 to 2010 then increasing from 2010 to 2014 to the highest level over the three surveys. The changes in the percentages for the other race/ethnicity categories are not statistically significant due to the small number of such households in the survey. As with the findings for the changes in the composition of the population by age and the associated average travel propensity, the survey findings shown in Table 6 and Table 7 have inter­ esting implications for future trends in air travel demand, although the potential consequences are less clear. The percentage of Asian and Hispanic households in the population is increasing, while the percentage of white households is declining. Although households with a Hispanic respondent reported a much lower average travel propensity than those with a white respondent, those with an Asian respondent reported a higher average travel propensity. However, the percentage of households with a Hispanic respondent in 2014 was almost two and a half times larger than the percentage with an Asian respondent, so on balance these trends are likely to lead to a continuing decline in overall air travel propensity, at least for some time. Respondent Race/Ethnicity 2006 2010 2014 All Households Percent All H/H All Households Percent All H/H All Households Percent All H/H White 1,867 72.6 1,803 69.8 1,484 67.9 Hispanic 283 11.0 320 12.4 290 13.3 Black 291 11.3 281 10.9 261 11.9 Asian 87 3.4 130 5.0 117 5.4 Native American 10 0.4 9 0.3 6 0.3 Pacific Islander 8 0.3 7 0.3 5 0.2 Multiple races 24 0.9 32 1.2 22 1.0 Total 2,570 100 2,582 100 2,185 100 NA / PI / Multiple (1) 42 1.6 48 1.9 33 1.5 Note 1. Native American, Pacific Islander, and multiple races combined. Table 7. Survey households by respondent race/ethnicity.

38 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies It should also be noted that differences in air travel propensity between the different race/ ethnicity categories are likely to be at least partly influenced by differences in household income rather than race/ethnicity per se, so care is needed to avoid double­counting the influence of both factors. Overall Air Travel Propensity The decline in the average number of air trips per year across all households from 2010 to 2014 (a decrease of about 3.6%) is somewhat unexpected in the light of the increase in total enplane­ ments by U.S. commercial air carriers over the same period, which increased from 712 million passengers in fiscal year 2010 to 757 million passengers in fiscal year 2014 (FAA 2016), an increase of about 6.3%. However, during the same period the number of households increased from 117.5 million to 123.9 million (Proctor, et al. 2016), an increase of about 5.4%. The combined effect of the increase in the number of households and the decline in the average number of air trips per household would have given a net increase of about 1.8%, still well below the increase in air passenger enplanements. However, it should be noted that the CES may not include all business trips and the number of business trips may have grown faster from 2010 to 2014 than the number of personal trips, as the economy recovered from the 2007–2009 recession. This suggests the importance of accounting for changes in the percentage of business trips compared to personal trips when analyzing changes in air passenger traffic. Lodging, Food and Beverage Expenses for Air Trips The analysis also examined the changes in lodging expenses per night and food and beverage expenses per night for air trips, expressed in constant dollars. Details are provided in Appendix C. For both categories of expenditures the expenses per night declined from 2006 to 2010 and then increased in 2014, although food and beverage expenses in 2014 were still below their level in 2006 in constant dollars. The average expenditure on lodging (for those air trips where lodging expenses were incurred) was $105 per night in 2006, declining to $92 per night in 2010, and recovering to $107 per night in 2014 (all in 2015 dollars), whereas the average expenditure on food and beverages (for all air trips of one night or more where food and beverage expenses were incurred) was $56 per night in 2006, declining to $41 per night in 2010, and recovering to $51 per night in 2014 (all in 2015 dollars). Air passenger demand models rarely consider the costs of making an air trip other than air­ fares. These data show that lodging and food and beverage expenses can be a significant part of the cost of making an air trip and that these costs change over time at different rates from airfares. Summary and Conclusions The analysis of the CES data has shown that air travel propensity not only varies with household income, as would be expected, but also that there are significant differences by survey respondent age and race/ethnicity. Current trends in the composition of the population by age and race/ethnicity combined with differences in air travel propensity across different age ranges and race/ethnicity categories appear likely to result in a decline in air travel propensity after allowing for changes in real household incomes and the real costs of air travel (including airfares, lodging, and other costs). This suggests that air travel demand studies need to take the changing composition of the population into account, although how to do this is less clear. Although the CES provides a valuable resource to help improve the understanding of air travel demand that has received little previous attention, there are two issues that deserve further study as a matter of some urgency. The first is the apparent discrepancy between the number of air party trips reported by CES respondents and the number of air passenger trips reported in U.S. DOT data. The second is the extent to which differences in air travel propensity by different

Sources of Disaggregated Socioeconomic Data 39 household income ranges and age ranges and race/ethnicity of survey respondents are inter­ related. Ideally, one would like to know how average air travel propensity varies by households of a given income, age, and race/ethnicity of household members. Although the CES is a fairly large survey, the sample size is not large enough to support a three­way tabulation of the average number of air trips per year with sufficient resolution and reliability, so another analysis approach is needed. Air Passenger Surveys Air passenger surveys can provide disaggregated data on the socioeconomic characteristics of air travelers as well as their air travel frequencies, depending on the questions asked in the survey. The response­level data from these surveys allows an analysis to be undertaken of how air travel propensity (air trips per year) varies with socioeconomic characteristics. Airport Intercept Surveys Airport intercept surveys are the most common form of air passenger surveys and are typically performed at a specific airport or group of airports over a relative short period of time, such as one or two weeks, although they may be repeated at different times of the year to capture seasonal differences in air traveler characteristics. The research team identified 10 air passenger surveys conducted at eight airports in various years from 2006 to 2015. In addition, smaller sample surveys have been undertaken on a quarterly basis at a large number of airports as part of the Airport Service Quality (ASQ) survey program run by the Airports Council International (ACI). Detailed survey response data have been obtained for recent surveys at seven of the eight airports, as well as for a number of surveys undertaken in prior years when similar surveys were undertaken at each airport. In addition, airport staff for the eighth airport, Boston Logan Inter­ national Airport, provided survey response data for selected questions from a recent survey in place of providing complete survey response data, and ACI staff provided detailed ASQ survey response data for surveys undertaken at 28 de­identified airports in 2015. Appendix C presents a detailed analysis of survey data from the following eight airports in addition to data from the 2015 ASQ survey: • Boston Logan International Airport (BOS) – 2013 survey • Los Angeles International Airport (LAX) – 2006 survey – 2015 survey • Metropolitan Washington Council of Governments (MWCOG) surveys undertaken at Washington Reagan National Airport (DCA), Washington Dulles International Airport (IAD), and Baltimore­Washington International Airport (BWI) – 2011 survey – 2013 survey • Oakland International Airport (OAK) – 2006 survey undertaken by the Metropolitan Transportation Commission – 2014/15 survey • San Francisco International Airport (SFO) – 2006 survey undertaken by the Metropolitan Transportation Commission – 2014/15 survey • Tulsa International Airport (TUL) – 2016 survey

40 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies The surveys were undertaken by the airport authority, except where noted. The MWCOG surveys were each undertaken using the same survey questions at each airport and at the same time of year. Since these surveys were performed at all three commercial service airports in the Baltimore/Washington metropolitan region, the survey results were combined to give a regional profile of air travel propensity. The 2006 and 2014/15 surveys at OAK and SFO were also each undertaken using essentially the same survey questions, differing only in response options relevant to each airport, and at the same time. Although there were some changes in the survey questions from 2006 to 2014/15, most questions were largely unchanged. However, the two LAX surveys differed in that the 2006 survey asked respondents how many air trips they had made in the past year from each commercial service airport in the Southern California region, while the 2015 survey only asked how many air trips they had made from LAX. In order to explore how air travel propensity varies with respondents characteristics, an analysis was limited to responses by residents of each region, since visitors could (and doubtless many did) make many air trips to other destinations that would not have been reported in the surveys. In the case of the 2015 LAX survey, the analysis was restricted to residents of the West Side of the Los Angeles basin, the area surrounding LAX for which LAX is the closest airport. Since LAX has by far the most air service of any of the commercial service airports in the region, it was assumed that West Side residents would have had little reason to use other airports and thus their use of LAX was a good indicator of their total air travel. The analysis of the airport intercept surveys gave broadly consistent results after making allowances for the range of years over which they were performed. The variation in air travel propensity (average total number of air trips per year) with household income for the nine surveys that asked both how many air trips the respondents had made in the past year and their household income is shown in Tables 8 and 9. The TUL survey did not ask how many air trips the respondents had made in the past year and the ACI ASQ survey did not ask the respondents’ household income. Since the nine surveys used different ranges of household income, the responses were adjusted to a consistent set of income ranges for comparison by assuming that the respondents within each income range were uniformly distributed across the incomes in the range and made the same number of air trips per year as the average for the income range. Since the surveys were performed over a wide range of years, the household income ranges used in each survey were converted to constant 2015 dollars, assuming that survey respondents reported their Household Income (2015 $) Average Annual Trips LAX 2006 Survey MWCOG 2011 Survey OAK 2006 Survey SFO 2006 Survey Weighted Average (1) Under $15,000 3.4 5.8 5.0 5.6 4.4 $15,000 - $24,999 3.4 5.1 5.0 5.6 3.8 $25,000 - $49,999 3.9 4.9 5.7 5.0 4.3 $50,000 - $99,999 5.3 6.3 7.8 6.8 5.8 $100,000 - $149,999 7.3 7.7 9.1 9.6 7.7 $150,000 - $199,999 9.3 9.0 12.9 10.6 9.5 $200,000 and over 13.4 13.8 18.5 17.1 14.4 Total (2) 6.87 9.15 10.99 10.68 8.33 Valid responses 8,685 6,118 1,186 1,083 17,072 Notes: (1) Weighted by valid survey responses. (2) Average annual trips exclude respondents who did not indicate their household income. Table 8. Average air trips in past 12 months by air passenger survey respondents by annual household income—earlier surveys.

Sources of Disaggregated Socioeconomic Data 41 income for the previous calendar year, before adjusting the responses to a consistent set of household income ranges. Although the general pattern is broadly consistent across the nine surveys, the average numbers of annual trips for each income range and in total for each survey differ consider­ ably. In particular, the values for the surveys at the two Bay Area airports (OAK and SFO) are significantly higher than for the other surveys. This is apparent for the two LAX surveys; the first was performed in the same year as the first two Bay Area surveys, and the second was performed less than a year after the second two Bay Area surveys. The reason for the large difference in values between the LAX surveys and the four Bay Area surveys is unclear and deserving of further research. Although both OAK and SFO serve the same region, it is notable that Bay Area residents using OAK had a slightly higher overall air travel propensity than those using SFO in both the earlier and more recent surveys. Although this was not the case for all income ranges, it was true for higher income respondents with household incomes over $150,000 (in 2015 dollars) in both sets of surveys, who of course made the highest average number of air trips. It is possible that this reflects a greater use of OAK for frequent business travel in West Coast markets, which are particularly well served from OAK. If so, this has interesting implications for use of secondary airports in multi­airport regions such as the Bay Area and is deserving of further research. Another interesting finding from the results shown in Tables 8 and 9 is that the overall average air trip propensity declined from the earlier to the later surveys in all regions for which surveys were analyzed in both periods. This decline does not appear in the weighted average results across each set of surveys because the surveys for LAX, which had the lowest overall average air trip propensity, had very different sample sizes in the two periods. The 2006 LAX survey had the largest number of valid responses of all the earlier surveys, accounting for almost half the responses, which depressed the weighted average. The 2015 LAX survey had the lowest number of valid responses used to calculate the average number of annual air trips in the more recent surveys and had very little influence on the weighted average. This decline in average air travel propensity appears to have occurred in almost all income ranges for the MWCOG and Bay Area surveys. The situation is less clear in the case of the LAX surveys but, as noted above, the number of valid responses used in the analysis of the 2015 survey was very small so differences for a given income range are less reliable. Household Income (2015 $) Average Annual Trips BOS 2013 Survey LAX 2015 Survey (1) MWCOG 2013 Survey OAK 2014/15 Survey SFO 2014/15 Survey Weighted Average (2) Under $15,000 3.5 4.3 4.7 4.1 3.9 4.2 $15,000 - $24,999 3.5 4.1 4.0 4.1 3.9 3.9 $25,000 - $49,999 3.4 3.8 4.8 4.6 4.7 4.4 $50,000 - $99,999 4.1 7.1 5.7 6.5 6.7 6.0 $100,000 - $149,999 5.7 7.5 7.3 9.4 9.7 8.0 $150,000 - $199,999 8.0 7.5 9.0 12.4 10.5 9.8 $200,000 and over 12.6 10.3 12.7 17.2 15.7 14.2 Total (3) 7.07 6.79 8.30 9.81 9.51 8.69 Valid responses 3,394 480 5,034 3,039 4,852 17,069 Notes: (1) Los Angeles West Side residents. (2) Weighted by valid survey responses. (3) Average annual trips include respondents who did not indicate their household income. Table 9. Average air trips in past 12 months by air passenger survey respondents by annual household income—more recent surveys.

42 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies The average number of air trips in the past 12 months reported by survey respondents to five recent air passenger surveys by respondent age range is shown in Table 10. Respondents in the age range 45 to 54 reported the highest average number of air trips in the past year for all surveys except the MWCOG 2013 survey, where respondents in the age range 55 to 64 reported a slightly higher average number of air trips. Respondents in the age range with the highest average number of air trips generally reported about twice as many air trips on average as those in the age range with the lowest average number of air trips. This was generally respondents in the age range 18 to 24, except for the LAX 2015 survey, where respondents 65 or older reported slightly fewer air trips on average than those in the age range 18 to 24. The weighted average number of air trips per year across the five surveys gave respondents in the age range 45 to 54 having over twice as many air trips as those in the age range 18 to 24. It is clear from the analysis of the various surveys that respondents who were making a business trip when they were surveyed typically report a higher average number of air trips in the previous year than respondents who were making a personal trip. This is not surprising, since most if not all business air travelers will also make personal air trips, while the reverse is not always true. Also, it seems reasonable that some business travelers make a large number of air trips each year for business, while relatively few travelers make a large number of personal air trips each year. However, it is quite likely that the number of annual air trips reported by many of the survey respondents making a personal trip who reported making several air trips included both business and personal trips. The data from each of the surveys consistently show that a fairly small number of air passengers report making a large number of air trips per year (say more than 12, or an average of one a month) and that this small number of passengers accounts for a relatively large proportion of total air trips. More detailed comparative analysis of each of these surveys that explored how air travel propensity varied with additional respondent characteristics is presented in Appendix C. Tulsa International Airport. Tulsa International Airport (TUL) has undertaken annual airport customer surveys for the past 5 years. The findings of these surveys provide a useful comparison to those from the surveys at larger airports. Airport staff provided summary data for the surveys from 2012 to 2016 and detailed survey response data for the 2016 survey. The surveys covered both arriving and departing passengers as well as meeters and well­wishers. In addition to collecting statistical data on each type of airport customer, the survey asked questions about the customer experience at the airport and improvements they would like to see, as well as other Age Range Average Annual Trips BOS 2013 Survey LAX 2015 Survey (1) MWCOG 2013 Survey OAK 2014/15 Survey SFO 2014/15 Survey Weighted Average (2) 18 - 24 3.5 5.7 5.1 4.9 5.4 4.9 25 - 34 6.1 7.0 7.9 8.7 9.3 8.2 35 - 44 8.3 7.1 8.9 11.4 11.3 10.0 45 - 54 9.5 7.9 9.2 12.4 11.7 10.6 55 - 64 8.2 6.5 9.5 11.0 10.1 9.7 65 or older 4.7 5.3 6.5 6.9 5.9 6.0 Total (3) 7.08 6.79 8.30 9.81 9.51 8.69 Valid responses 3,938 701 6,164 4,558 6,614 21,975 Notes: (1) Los Angeles West Side residents. (2) Weighted by valid survey responses. (3) Average annual trips include respondents who did not indicate their age. Table 10. Average air trips in past 12 months by air passenger survey respondents by age range—recent surveys.

Sources of Disaggregated Socioeconomic Data 43 airports in the region that they had used. However, the survey did not ask respondents how many air trips they had made in the past year. The composition of survey respondents and the trip purposes of each type of air passenger are shown in Table 11. The proportion of respondents making business and personal trips varies considerably from year and also differs between departing and arriving passengers in a given year. This could reflect the relatively small size of the sample, as well as the timing of the surveys. Nonetheless, for the three most recent surveys the percentages of respondents making business trips were higher than for those making personal trips, in contrast to the survey results for 2013 and the findings of air passenger surveys at larger airports. The survey did not ask how many people were in each travel party, so the percentages shown in Table 11 (and the following table) are of air parties, not total air passengers. Since the aver­ age air travel party size of those making business trips is usually considerably lower than those making personal trips, the percentages of air passengers making a business trip would be lower than shown in the table. A more detailed analysis of the 2016 survey response data was undertaken to determine how air passenger characteristics varied by trip purpose. Table 12 shows the profile of the air passenger respondents by trip purpose. A much higher percentage of respondents on a business trip were male, while a higher percentage of those on a personal trip were female. While the higher percentage of male business travelers is not surprising, the higher percentage of those making a personal trip who were female is surprising. Although other air passenger surveys have also found a somewhat higher percentage of respondents making personal trips to be female, the difference has not been as great as suggested by the TUL survey. The age profile of the survey respondents is significantly different for respondents who were making a business trip and those making a personal trip. Of those making a business trip, 80% were between the ages of 30 and 59, with only 6% below the age of 30 and over a third in the age range from 50 to 59. In contrast, 23% of those making a personal trip were below the age of 30, with 33% aged 60 or more. As might be expected, survey respondents who were making business trips had significantly higher annual household incomes than those who were making personal trips, with 43% having incomes of $150,000 or more compared to only 12% for those making personal trips. Survey respondents who were making personal trips had household incomes more evenly distributed between $25,000 and $150,000, with the largest percentage (25%) having incomes between $50,000 Customer Type 2013 2014 2015 2016 Departing passenger 40% 51% 47% 40% Arriving passenger 41% 49% 47% 40% Meeter/well-wisher 19% 5% 20% 100% 100% 100% 100% Survey sample size 261 204 213 252 Departing passengers Business 42% 65% 55% 65% Personal 58% 35% 45% 35% 100% 100% 100% 100% Arriving passengers Business 32% 51% 61% 53% Personal 68% 49% 39% 47% 100% 100% 100% 100% Table 11. Sample size and composition of TUL customer survey respondents.

44 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies and $75,000. No survey respondents who were making a business trip reported a household income below $25,000, while only 7% of those making a personal trip reported an income below $25,000. The 2016 survey was the first one to ask respondents about the highest level of education they had attained, as shown in Table 12. Not surprisingly, survey respondents who were making business trips had a fairly high level of education with two­thirds having a bachelor’s or post­ graduate or professional degree and over 90% having some college education. However, survey respondents who were making personal trips also had a fairly high level of education, with 60% having a bachelor’s, or post­graduate or professional degree and a further 24% having some college education. In summary, the results of the TUL customer surveys provide a useful indication of the profile of air travelers using a smaller airport, although the small sample size both limits the ability to explore how the respondent characteristics vary with more than one variable and may have introduced a fair amount of error in the results. ACI Airport Service Quality Surveys. For many years the ACI has maintained a program of customer satisfaction surveys, ASQ survey, in which many of its member airports partici­ pate (http://www.aci.aero/Airport­Service­Quality/ASQ­Home). These surveys are generally undertaken four times per year using a standardized survey questionnaire with a minimum of 350 respondents for each survey. As of 2015, 28 U.S. airports participated in this program. The ASQ survey asks respondents to rate 30 aspects of their airport experience at the airport where they were surveyed. In addition, the ASQ survey includes a number of questions about the Characteristics Business Trip Personal Trip Gender Male 79% 29% Female 21% 71% 100% 100% Age 18 - 19 4% 20 - 29 6% 19% 30 - 39 20% 14% 40 - 49 24% 15% 50 - 59 36% 16% 60 or more 15% 33% 100% 100% Household Income Less than $25,000 7% $25,000 - 49,999 5% 18% $50,000 - 74,999 10% 25% $75,000 - 99,999 15% 18% $100,000 - 149,999 28% 21% $150,000 or more 43% 12% 100% 100% Educational attainment Some high school 1% High school graduate 3% 9% Trade/tech/vocational 4% 6% Some college credit 15% 12% Associate’s degree 10% 12% Bachelor’s degree 45% 41% Post-graduate degree 22% 19% 100% 100% Table 12. Distribution of TUL 2016 air passenger respondent characteristics.

Sources of Disaggregated Socioeconomic Data 45 current air trip being taken by respondents and traveler characteristics, although these do not include household income. ACI staff provided ASQ survey response data for surveys undertaken at 28 de­identified airports in 2015 for the following questions: • Trip purpose • Number of return air trips made in the past 12 months • Nationality • Country of residence • Postal code • Gender • Age The dataset comprised 68,484 survey responses, of which 58,448 were by U.S. residents. Analysis of the data for U.S. residents was undertaken to explore how the number of air trips in the past 12 months varied by trip purpose, age, and gender. The number of air trips was reported by survey respondents in five ranges: 1–2, 3–5, 6–10, 11–20, and 21 or more. The number of respondents by trip purpose and the proportion that reported making air trips in each range is shown in Table 13. It should be noted that these trip frequency distributions are based on the purpose of the current trip and do not show the number of trips by each purpose in the past 12 months. Survey respondents who were making a business trip would be more likely to have made more air trips in the past 12 months (many of which would most likely have been for business) than survey respondents making a non­business trip and may not have made any business trips in the past 12 months. The average number of air trips for each trip frequency range shown differs from the mid­point of the range, reflecting the curvature of the cumulative distribution curves. They were obtained from an analysis of the frequency distribution of air trips in the past 12 months by San Francisco Bay Area residents in the 2014/15 air passenger survey under­ taken at San Francisco International Airport. Respondents reported the actual number of trips they had made. There does not appear to be a large difference in the trip frequency distribution for those respondents making a leisure trip or other non­business type of trip, although there was a large difference between those respondents and those making a business trip. Therefore subsequent analysis combined respondents who reported making a leisure trip or other non­business type of trip. The overall average number of trips per year is somewhat lower than found in air passenger surveys at large airports performed around the same time, as shown above in Table 9, but not significantly so. This could reflect the inclusion of smaller airports in the survey responses where local residents may have less reason to make air trips and may well have lower average household Trip Purpose Survey Responses Percent Air Trips in Past 12 Months 1-2 3-5 6-10 11-20 21+ Total Business 21,105 36.1 17% 26% 21% 17% 20% 100% Leisure 30,893 52.9 41% 36% 15% 5% 3% 100% Other 6,450 11.0 44% 35% 14% 4% 3% 100% Total 58,448 100 33% 32% 17% 9% 9% 100% Average Trips Business 1.6 4.0 7.8 15.0 38 12.9 Leisure 1.6 3.8 7.5 14.3 33 4.8 Other 1.6 3.8 7.5 14.3 33 4.6 Total 7.7 Table 13. Distribution of air trips per year by trip purpose—ASQ surveys.

46 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies incomes. The difference in average annual air trips between survey respondents making business and non­business trips is similar to that found in the surveys at large airports. Table 14 shows the number of respondents by age range for those making business and personal trips, as well as the average number of air trips in the past year. The average number of trips per year increases with age until the age range 55 to 64 in the case of those making a business trip and age range 45 to 54 for personal trips. The average number of trips per year then declines with age. For those making personal trips, the average number of trips per year by those in the highest age range of 76 and over is the lowest of any age range, at 3.4 trips per year on average, although only 3% of those making personal trips are in this age range. Even so, this is still a respectable number of trips per year for someone in this age range. As found in the analysis of other air passenger surveys, the variation in the average number of air trips per year with age for those making a business trip is much wider than for those making a personal trip. Table 15 shows the number of respondents by gender for those making business and personal trips, as well as the average number of air trips in the past year. As might be expected, a higher proportion of those making business trips were male and those travelers reported a somewhat higher average number of air trips in the past year than female respondents making a business trip. Perhaps more surprisingly, although consistent with findings from other air passenger surveys, a higher proportion of those making personal trips were female, although of those making a personal trip, male respondents reported a higher average number of air trips in the past year than female respondents. This most likely reflects the fact that male respondents making a personal trip may have made a larger number of business trips in the past year on average than female respondents making a personal trip. More details of the analysis of the ASQ survey data are provided in Appendix C. Age Range Business Trips Personal Trips Survey Responses Percent Avg. Air Trips in Past 12 mo. Survey Responses Percent Avg. Air Trips in Past 12 mo. 16-21 340 1.6 5.5 2,675 7.3 4.0 22-25 1,106 5.3 8.3 2,893 7.9 4.3 26-34 3,540 17.0 11.0 5,279 14.4 5.1 35-44 4,521 21.7 12.9 4,246 11.5 5.0 45-54 5,587 26.8 14.3 6,146 16.7 5.2 55-64 4,491 21.6 14.5 8,431 22.9 5.0 65-74 1,142 5.5 12.5 6,029 16.4 4.4 76 and over 100 0.5 9.7 1,079 2.9 3.4 Total 20,827 100 12.9 36,778 100 4.8 Table 14. Distribution of air trips per year by age and trip purpose— ASQ surveys. Gender Business Trips Personal Trips Survey Responses Percent Avg. Air Trips in Past 12 mo. Survey Responses Percent Avg. Air Trips in Past 12 mo. Male 13,305 64.4 14.1 14,007 38.5 5.4 Female 7,360 35.6 10.6 22,348 61.5 4.4 Total 20,665 100 12.1 36,355 100 4.8 Table 15. Distribution of air trips per year by gender and trip purpose—ASQ surveys.

Sources of Disaggregated Socioeconomic Data 47 Survey of International Air Travelers SIAT is undertaken annually by the National Travel & Tourism Office (NTTO) (formerly the Office of Travel and Tourism Industries) of the U.S. Department of Commerce. This survey provides a large sample of both outbound U.S. residents and inbound foreign visitors traveling between the United States and overseas countries (i.e., excluding Canada and, until recently, Mexico). The survey collects detailed trip purpose information and the number of air trips to/from the United States by the respondent in the prior 12 months (and until 2012 the prior 5 years), as well as a range of socioeconomic data. The major value of this survey lies in the fact that it has been performed using a consistent survey instrument every year since 1983, with some changes from 2012 on. Therefore, it provides a unique resource to study how international air travel propensity and air traveler characteristics have changed over time. Individual survey response data for U.S. residents was obtained from the NTTO for the surveys performed in 2005, 2010, and 2015. These data were analyzed as described in detail in Appendix C. A number of changes were made to the SIAT survey instrument for the 2012 and subsequent surveys. The principal change that affects the analysis presented in this report is that prior to 2012 respondents reported their annual household income by selecting from 11 income ranges, the highest of which was $200,000 or more, whereas from 2012 respondents reported their actual household income. This has two implications for the analysis. The first is that for surveys before 2012 it prevents any analysis of how the number of international air trips varies with income for households with an annual income above $200,000. In the 2015 survey, 18.6% of survey respondents reported an annual household income of $200,000 or more, with 2.4% reporting a household income of $500,000 or more. The second implication is that the analysis of the 2015 survey data could exclude respondents who reported an annual household income below $10,000, since it was felt that these respondents included an unknown number who misreported their household income (it is unclear how someone with an income under $10,000 could afford to make an international air trip). However, in the survey data prior to 2012 the lowest household income range was under $20,000 so it was not possible to exclude respondents who would have reported an annual household income of less than $10,000 had this been an option. Therefore all respondents were included in the analysis. In the 2015 survey data, about 3% of those reporting a household income reported an income below $10,000, while about 2.4% reported an income between $10,000 and $19,999. Thus is appears that a high proportion of survey respondents reporting a household income below $20,000 may have misreported their income. The surveys for the 3 years over a 10­year period showed broadly similar patterns in the variation of the average number of international air trips per year with household income for each year as shown in Table 16. The overall average number of international air trips per year declined from 2005 to 2010, then recovered by 2015 to a level somewhat below that in 2005. Although the decline from 2005 to 2010 and the recovery from 2010 to 2015 occurred in all household income ranges, the change from 2005 to 2015 differed between the income ranges, with the average number of air trips generally increasing from 2005 to 2015 for survey respon­ dents from households with incomes below $120,000 per year but decreasing significantly for respondents from households with incomes above $120,000 per year. Of course, the household incomes reported by survey respondents are expressed in current dollars and the same income ranges were used for each of the surveys prior to 2012 and also used for the 2015 data in Table 16 for consistency. Therefore, if expressed in constant dollars, the income ranges would change over time. Nonetheless, the distribution of survey responses by household income does not appear to have changed significantly between the three surveys.

48 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies The percentage of respondents reporting annual household incomes under $20,000 in the 2015 survey excluded those reporting incomes under $10,000, so it is to be expected that this per­ centage would be lower than for 2005 and 2010 (and the percentages for other income ranges correspondingly higher, although the effect in each income range would be relatively small). The percentage of survey respondents who reported a household income of $160,000 or more declined from about 27% in 2005 to about 24% in 2015. This is somewhat surprising, given the recent trends in household income distribution in the United States and the increase in household incomes in current dollars that occurred from 2005 to 2015, the 2007 recession notwithstanding. In contrast, the data from the three surveys for the average number of international air trips per year by the age of the survey respondent shows a clear trend, with an increasing percent­ age of survey respondents in the age ranges below 35 and age 60 or over, and a corresponding decline in the percentage of survey respondents in the age ranges between 35 and 59, as shown in Table 17. There was also a clear trend in the trip purpose split over the three surveys, as shown in Table 18. The average number of air trips per year declined for both business and personal purposes from 2005 to 2010 and subsequently increased from 2010 to 2015. However, whereas the average number of air trips by respondents making a business trip in 2015 was lower than in 2005, the average number of air trips by respondents making a personal trip was higher in 2015 than in 2005. The percentage of survey respondents that were making a business trip declined steadily from 2005 to 2015. By 2015, the share of international air trips that were made for business was almost half that in 2005. Perhaps the most striking finding from the analysis of the SIAT data is that although the average number of international air trips in the past year reported by survey respondents increases with increasing household income, as would be expected, the difference between the lowest income ranges and the highest is quite small. In 2015 the average number of international air trips per year by respondents with a household income of $200,000 or more was less than twice that by respondents with a household income under $20,000. Similarly, although survey respondents aged between 40 and 59 reported a somewhat higher average number of inter­ national air trips per year than those under 40 or over 59, the difference between the average number of international air trips per year by respondents in the age range with the lowest average Household Income 2005 2010 2015 Percent Responses Average Annual Trips Percent Responses Average Annual Trips Percent Responses Average Annual Trips Under $20,000 3.7 1.73 5.8 1.62 2.2 2.04 $20,000 - $39,999 7.4 1.75 8.3 1.74 8.8 2.06 $40,000 - $59,999 11.4 2.02 11.3 1.92 12.3 2.15 $60,000 - $79,999 12.3 2.21 11.9 2.01 12.7 2.37 $80,000 - $99,999 11.8 2.49 11.5 2.16 10.4 2.48 $100,000 - $119,999 11.6 2.70 11.7 2.35 13.7 2.73 $120,000 - $139,999 8.2 3.24 7.7 2.61 7.6 2.83 $140,000 - $159,999 6.4 3.39 6.0 2.82 8.3 3.17 $160,000 - $179,999 4.3 3.80 4.2 2.95 3.1 3.10 $180,000 - $199,999 3.3 3.85 3.4 3.12 2.3 3.25 $200,000 or more 19.6 4.52 18.1 3.85 18.6 3.95 Total 100 2.90 (1) 100 2.45 (1) 100 2.67 (1) Valid responses 21,584 22,980 18,274 Note: (1) Average annual trips include respondents who did not indicate their household income. Table 16. Average international air trips in past 12 months by U.S. residents by annual household income.

Sources of Disaggregated Socioeconomic Data 49 air travel propensity (those aged 18 to 24) and those in the age range with the highest propensity (those aged 45 to 49 in 2005 and 50 to 54 in 2010 and 2015) declined from a factor of 2.1 in 2005 to 1.6 in 2015. There is a practical limit on how many international air trips any household is likely to make in a year, irrespective of their household income, and every survey respondent will have made at least one such trip, so the relatively small range of the average number of inter­ national air trips with household income or respondent age is perhaps not so surprising. The overall average number of international air trips per year by survey respondents was about 2.7 in 2015. Although this is a decrease from the value in 2005, it was an increase from the value in 2010. It is not clear whether the increase from 2010 to 2015 represents an ongoing trend that will result in higher values in the future. Additional analysis of the SIAT data exploring differences in the average number of inter­ national air trips per year by gender and race/ethnicity of the survey respondent is documented in Appendix C. Online Surveys of Air Travelers Online surveys of air travelers are sometimes undertaken in support of research projects or other studies and can provide a complementary perspective to data collected through intercept Age Range 2005 2010 2015 Percent Responses Average Annual Trips Percent Responses Average Annual Trips Percent Responses Average Annual Trips 18 - 24 6.7 1.72 9.7 1.66 12.6 1.99 25 - 29 8.1 2.17 9.6 2.03 11.1 2.31 30 - 34 9.6 2.72 9.2 2.42 9.9 2.60 35 - 39 10.8 3.08 9.2 2.69 8.5 2.88 40 - 44 12.5 3.31 10.8 2.73 9.1 3.10 45 - 49 12.5 3.59 10.8 2.89 9.4 3.06 50 - 54 12.2 3.35 11.0 2.95 10.5 3.16 55 - 59 10.6 3.17 9.6 2.77 8.4 3.08 60 - 64 7.4 2.84 8.6 2.36 7.9 2.73 65 - 69 4.9 2.50 5.9 2.35 6.5 2.39 70 - 74 2.8 2.17 3.3 2.23 3.8 2.33 75 - 79 1.4 2.04 1.4 1.87 1.7 2.17 80 - 84 0.4 1.92 0.7 1.75 0.5 2.09 85+ 0.1 2.53 0.2 1.89 0.1 2.61 Total 100 2.90 (1) 100 2.45 (1) 100 2.67 (1) Valid responses 22,972 12,925 27,508 Note: (1) Average annual trips include respondents who did not indicate their age range. Table 17. Average international air trips in past 12 months by U.S. residents by age range. Trip Purpose 2005 2010 2015 Percent Responses Average Annual Trips Percent Responses Average Annual Trips Percent Responses Average Annual Trips Business 29.9 4.87 23.0 4.18 16.9 4.62 Personal 70.1 2.06 77.0 1.94 83.1 2.27 Total 100 2.90 (1) 100 2.45 (1) 100 2.67 (1) Valid responses 23,963 26,435 28,016 Note: (1) Average annual trips include respondents who did not indicate their trip purpose. Table 18. Average international air trips in past 12 months by U.S. residents by trip purpose.

50 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies surveys of air passengers at airports. Although online surveys of air travelers typically collect information on air trips made by respondents, they are really a form of a household survey in that each respondent usually represents one household (or one individual within a household) and the probability of a respondent participating in the survey does not vary with the number of annual air trips made (although this information may be reported in the survey). In contrast, an intercept survey of air passengers conducted at an airport is essentially a survey of air passenger trips, not of individual travelers. The research has identified three such surveys, one undertaken as part of ACRP WOD 22: Passenger Value of Time, Benefit Cost Analysis, and Airport Capital Investment Decisions, and two more recent surveys that were performed in late 2015 and early 2017 for Airlines for America (A4A). A detailed analysis of the data from each of these surveys is presented in Appendix C. ACRP 03-19 Air Traveler Survey. As part of ACRP WOD 22 a web­based survey was per­ formed in early 2013 of individuals who had made a paid domestic air trip in the prior 6 months. The survey respondents were recruited from a commercial firm that maintains panels of indi­ viduals willing to participate in online surveys and provided details about their most recent trip, including the trip purpose and air party size, as well as the number of air trips they had made in the previous 12 months for business and non­business (personal) purposes. The survey also collected information on various socioeconomic characteristics of the respondents, includ­ ing personal and household income, gender, age, and household composition. The distribution of the O&D of the most recent domestic air trip reported by survey respon­ dents was compared to the overall pattern of domestic O&D travel and found to be reasonably representative of domestic air travel in general (Resource Systems Group, Inc. 2015). Although the sample size of the survey was fairly small (1,171 respondents) the respondents were distrib­ uted throughout the United States. Thus, the survey provides a more representative sample of national air passengers compared to surveys conducted at specific airports. Although the survey respondents were on average older and with higher incomes than the general U.S. population, this may partly reflect the air traveler population, which other surveys have shown to be older and have a higher average income than the population in general. The proportion of the survey respondents who were female (55%) was somewhat higher than the population in general. The extent to which this reflects the domestic air traveler population is unclear. Although many airport air passenger surveys collect data on the respondent gender, they typically only obtain one response from each air party and thus only report the gender of the person completing the survey or answering the survey questions on behalf of the party. In the case of a family or couple traveling together, it is possible that a male member of the party would be more likely to respond to the survey. Also, the comparison is complicated by the fact that men make more air trips per year on average than women, which would reduce the percentage of female respondents in airport air passenger surveys. A detailed analysis of the gender, age, household income, and household composition profile of the survey respondents compared to the U.S. population is presented in Appendix C. The purpose of the most recent air trip taken by survey respondents is shown in Table 19 with the six reported purposes reclassified into business and personal trips. Overall, 26% of the trips were taken for business purposes, of which a little less than a third (8% of all trips) were to attend a conference or similar event. Personal trips were fairly evenly split between vacations and visiting friends or relatives. However, this could be influenced by the time of year when the survey was performed. Air trips taken in the previous six months would have included visits to family at Thanksgiving or Christmas, but would not have included any trips taken the previous summer. The low percentage of trips to attend college or school (less than 1%) reflects the age profile of the survey respondents, with only 4% in the age range from 18 to 24.

Sources of Disaggregated Socioeconomic Data 51 The proportions of trips by different purposes are significantly different for male and female respondents. Only 19% of female respondents made a trip for business purposes, compared to 35% for male respondents, although both male and female respondents made a similar percentage of trips to attend a conference or similar event. Female respondents made slightly more trips to visit friends or relatives than for vacations, whereas the reverse was true for male respondents. An analysis of the average air party size reported for the most recent air trip was undertaken, as discussed in Appendix C. Although the average air party sizes for business and personal trips given by the analysis were both higher than those typically found in airport air passenger surveys, the discrepancy was larger for business trips. The difference may be due to the higher proportion of older respondents in the survey respondents, particularly male respondents, who may be more likely to take a spouse on a business trip than younger air travelers, who are more likely to be single, or with a working spouse or children. The average number of trips in the past year reported by survey respondents is shown in Table 20 by household income, gender, and trip purpose, together with the estimated number of air passengers on these trips based on the analysis of average air party size for the most recent Trip Purpose Total Male Female Responses Percent Responses Percent Responses Percent Business 206 17.6 139 26.1 67 10.5 Attend conference 91 7.8 43 8.1 48 7.5 Vacation 386 33.0 160 30.0 226 35.4 Visit friends/relatives 391 33.4 144 27.0 247 38.7 Attend school/college 9 0.8 3 0.6 6 0.9 Other 88 7.5 44 8.3 44 6.9 Total 1,171 100 533 100 638 100 Business 305 26.0 187 35.1 118 18.5 Personal 866 74.0 346 64.9 520 81.5 Total 1,171 100 533 100 638 100 Table 19. Purpose of most recent trip taken by ACRP 03-19 survey respondents. Household Income Total Business Trips Personal Trips Total Male Female Total Male Female Under $10,000 7.2 3.6 4.5 3.0 3.6 3.5 3.6 $10,000 - 19,999 3.4 1.1 3.5 0.6 2.4 2.5 2.1 $20,000 - 29,999 3.4 0.8 0.6 2.6 2.8 $30,000 - 39,999 4.1 1.6 4.0 0.7 2.5 2.3 2.6 $40,000 - 49,999 4.5 1.7 2.7 1.4 2.7 3.4 2.4 $50,000 - 74,999 4.5 1.7 2.4 1.2 2.8 2.7 2.9 $75,000 - 99,999 5.8 2.7 3.7 1.9 3.1 2.9 3.2 $100,000 - 149,999 6.1 3.0 3.9 1.9 3.2 3.0 3.3 $150,000 - 199,999 7.5 4.1 5.4 2.2 3.4 3.4 3.3 $200,000 - 249,999 8.5 4.5 5.4 3.3 4.1 3.3 5.0 $250,000 or more 10.2 5.7 8.1 2.9 4.5 4.5 4.4 Total 6.0 2.9 4.2 1.7 3.2 3.1 3.2 Average air party size 1.79 1.55 1.48 1.70 2.01 2.02 2.00 Air passengers 12,609 5,201 3,312 1,889 7,408 3,354 4,054 Percent business 41.3% Table 20. Average air trips in past year by ACRP 03-19 survey respondents by household income and gender.

52 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies trip. It was assumed that the average air party sizes for business and personal trips by male and female survey respondents in the past year were the same as reported for the most recent trip. On this basis, for the trips in the past year reported by the survey respondents the number of air passengers making a business air trip accounted for 41% of all air passengers trips. This percentage is considerably higher than the percentage typically found in air passenger surveys at airports, which may be partly due to the average air party size assumptions for business trips. The average air trips in the past year generally increase with increasing household income, as would be expected, and are generally larger for personal trips than for business trips. The percentages for both business and personal trips by males with household incomes between $10,000 and $30,000 have been combined due to the small number of male respondents with incomes in this range. Even so, the number of male survey respondents with household incomes below $50,000 and the number of female respondents with household incomes below $30,000 are both fairly small, so the average annual trips for respondents in these income ranges could be distorted by a few respondents who made an atypically large number of trips. As has been found with other surveys, the average number of reported air trips by survey respondents in the lowest income range (under $10,000) is surprisingly large, and higher than the values found for most income ranges below $200,000 and all income ranges below $150,000 (in the case of personal trips by male respondents all income ranges below $250,000). The average number of business trips by female respondents is significantly lower than for male respondents, particularly for female respondents with household incomes between $10,000 and $40,000. However there appears to be no systematic difference in the average number of personal trips between male and female respondents. The difference for business trips does not appear to reflect differences in the proportion of female respondents who were employed. In fact almost the same proportion of male and female survey respondents (52%) reported that they were employed full­time or self­employed, so the difference is more likely to reflect the different types of employment. Female respondents who reported that they were self­employed made almost as many business trips on average (5.1 per year) as self­employed male respondents (5.3 per year), while female respondents who were employed full­time made only 2.1 business trips per year on average compared to 6.9 for male respondents employed full–time. Analysis of the response data from the ACRP 03­19 survey suggests that the survey respon­ dents are neither fully representative of the U.S. population nor of air travelers in general. How­ ever, given these limitations, the survey response data do provide several useful findings. The first relates the average number of air trips per year to various household characteristics, includ­ ing household income and age and gender of the respondent. In particular, female respondents made fewer air trips per year for business than male respondents, although there appears to be little difference between male and female respondents in terms of the number of personal air trips per year that they make. As expected, the average number of air trips per year increased with household income, although the increase was not proportional to income. The second useful contribution of the survey is that survey respondents reported the number of business and personal air trips they had made in the past year. This allows some analysis of air travel propensity by trip purpose. Although the average number of both business trips and personal air trips per year increased with income, the average number of business trips increased much faster than personal trips. A4A Air Travel Survey. A4A has commissioned two recent online surveys of the U.S. adult population (age 18 or over and resident in the continental U.S., Alaska, and Hawaii) that were performed by the polling firm Ipsos Public Affairs. The first survey was performed from December 14 to 22, 2015, and collected data from 3,019 adults. The second survey was performed from January 6 to 13, 2017, and collected data from 5,047 adults. The survey respondents were

Sources of Disaggregated Socioeconomic Data 53 drawn from an Ipsos panel, supplemented with panels from partner organizations, and other sources. The sample was selected to reflect the demographics of the U.S. population based on the U.S. Census American Community Survey and post hoc weights were applied to reflect population characteristics on gender, age, region, race/ethnicity, and income. The surveys addressed a wide range of air travel issues, but collected data on the demographic and socioeconomic characteristics of the respondents, as well as the number of air trips that the respondents had made in the previous year and (for those respondents who had not flown in the previous year) whether they had ever flown on a commercial flight. For the 2015 survey, respondents were asked how many round trips by airline they had already made in 2015 or planned to make in the remaining weeks of 2015. For the 2017 survey, respondents were asked how many round trips by airline they had made in 2016. In both cases respondents were asked to give separate totals for trips that were primarily for business purposes, primarily for personal leisure, and primarily for personal non­leisure purposes (such as traveling to/from college, family event, job interview, or medical reasons). Respondents were also asked how many of their total trips combined business and personal purposes. Detailed survey response data were not made available. However, A4A staff has shared aggre­ gate results (Heimlich 2016; Heimlich and Jackson 2017) and topline statistics for each question have been published on the Ipsos website (ipsos­na.com/news­polls/pressrelease.aspx?id=7208 and id=7585). The surveys found that 19% of 2015 respondents and 11% of 2017 respondents indicated that they had never flown on an airline. The percentage of 2015 respondents who had never flown on an airline was similar to the percentage found in a previous survey in 1997 performed by the Gallup Organization for the Air Transport Association (the former name of A4A). The percentage of 2015 respondents who had never flown on an airline seems too high, both relative to the 1997 percentage as well as the much lower percentage a year later. Of those respondents who indicated that they had flown before, 55% of 2015 respondents indicated that they had made no air trips that year, while 51% of 2017 respondents indicated that they had made no air trips in 2016. The distribution of the number of air trips taken by survey respondents who made at least one air trip in each year is shown in Table 21 together with the percentage of air trips for each trip purpose and the average number of trips per respondent who had made an air trip during the year. The decline in the percentage of survey respondents who indicated that they made nine or more air trips in the past year from 2015 to 2016 explains the drop in the total average trips per respondent from 4.8 to 4.5. Given the larger sample size in the 2017 survey and the fact that the overall level of air travel increased by about 3.5% from 2015 to 2016, based on U.S. Department of Transportation data, it seems likely that the difference is more an artifact of the sampling than an actual trend and that the 2016 data are probably more accurate. More detailed analysis of the survey findings that examined differences in the average number of reported air trips per year with respondent race/ethnicity, age, annual household income, and highest level of educational attainment is presented in Appendix C. Differences in the average number of air trips in the past year by respondent race/ethnicity and age is shown in Table 22. The larger sample size of the 2017 survey (2016 air trips) probably makes those results somewhat more reliable. The survey findings show considerable variation in the average number of air trip per year across respondents of each race/ethnicity, with Asian respondents reporting an average number of air trips per year over three times the average trip rate reported by black respondents and over one and a half times the average trip rate reported by white respondents. The age group with the

54 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies highest average number of air trips, those age 25 to 44, reported an average trip rate three times that reported by respondents age 65 and over. The A4A surveys provides a useful complement to the other air passenger surveys discussed earlier and to the household travel surveys discussed in the next section, since they are surveys of all American adults, whether or not they made an air trip in the past year. The survey results thus allow a comparison between air travelers (those who made at least one air trip in the past year) and the larger population. The surveys also collected data on the race/ethnicity and educa­ tional attainment of the respondents, characteristics that are not typically available from air pas­ senger surveys (although sometimes available from other surveys, such as the CES). However, Air Trips 2015 2016 1 31% 27% 2 20% 20% 3 10% 13% 4 9% 13% 5 5% 5% 6 6% 6% 7 2% 3% 8 2% 2% 9 or more 15% 11% 100% 100% Trip Purpose Business 31% 31% Personal Leisure 48% 51% Personal Other 21% 18% 100% 100% Average Trips Business 1.5 1.4 Personal Leisure 2.3 2.3 Personal Other 1.0 0.8 Total 4.8 4.5 Table 21. Distribution of air trips per year by A4A survey respondents who made air trips in each year. Air Trips 2015 2016 Race/Ethnicity White 2.0 2.2 Black 1.4 1.1 Hispanic 3.4 2.6 Asian 3.2 3.4 Other 1.0 1.1 Total 2.1 2.2 Age 18-24 2.1 1.3 25-44 3.2 3.3 45-64 1.6 1.5 65+ 1.1 1.1 Total 2.1 2.2 Table 22. Average air trips per year by all A4A survey respondents by race/ethnicity and age.

Sources of Disaggregated Socioeconomic Data 55 some of the distributions of respondent characteristics reported in presentations by A4A staff suggest that the survey may have undersampled some population segments or the weighting of the survey data may have distorted the results. Summary and Conclusions The analysis of air passenger surveys has shown that air travel propensity, expressed as the average number of air trip per year, varies widely with a broad range of respondent socio­ economic characteristics, including household income, age, race/ethnicity, and educational attainment. It can be expected that changes in the distribution of any of these characteristics across the population will have an effect on air travel demand. It also follows that the common practice in air passenger demand models of using aggregate or average measures of household income will fail to reflect the effect of changes in the distribution of household incomes as a percentage of the average income level. Although household income shows the widest range of air travel propensity, with survey respondents from households with incomes over $200,000 in 2015 dollars reporting making an average of over three times as many air trips per year as respondents from households with incomes under $25,000 in 2015 dollars, there is also a significant range in air travel propensity for different age ranges and race/ethnicity categories. Survey respondents to recent surveys aged between 45 and 54 generally reported an air travel propensity of about twice as many air trips per year as those aged between 18 and 24 and somewhat under twice as many as those 65 and over. Fewer air passenger surveys have collected data on respondent race/ethnicity, but recent surveys by A4A have found that Asian and Hispanic respondents reported an average number of air trips per year about one and a half times that reported by white respondents and about three times that reported by black respondents. However, it is quite likely that at least part of the differences in air travel propensity by factors such as age, race/ethnicity, and educational attainment reflect differences in household income. The analysis undertaken in the course of the project has not attempted to separate these effects but this forms an important topic for future research. Another important dimension of disaggregated measures of air travel propensity is air trip purpose. Survey results show that the distribution of the average number of business air trips per year with various socioeconomic factors is significantly different from the corresponding distributions for personal air trips. Since the relative proportions of business and personal air travel can be expected to vary in different markets, and may well change at different rates over time, developing separate air passenger demand relationships for different trip purposes is likely to lead to more robust air passenger demand models. Household Travel Surveys Household travel surveys can provide an alternative perspective on air travel to that provided by intercept surveys of air travelers undertaken at airports or through online surveys of air travelers. These surveys commonly take the form of an interview survey that asks household members to report the details of their travel on a particular day or days as well as the house­ hold socioeconomic characteristics. Since most respondents will not have taken a long­distance trip on the particular day they were asked to report, some household travel surveys include a long distance travel component that asks respondents to provide details of long­distance travel undertaken within a longer recent recall period. However, the definition of a long­distance trip and the length of the recall period typically vary between different surveys. One valuable feature of most household travel surveys that address long­distance trips is that they typically include long­distance trips by all modes, not just air trips. While air trips are the

56 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies primary focus of the current research, analysis of household travel surveys can contribute to a better understanding of how mode choice decisions in making long­distance trips affect the demand for air travel. The research team examined four such household travel surveys, two national surveys and two statewide surveys. The following is a summary of the findings from the analysis of the data from these surveys; more detail is presented in Appendix C. National Household Travel Survey The 2001–2002 National Household Travel Survey (NHTS) collected data on all trips over 50 miles made by members of responding households during a four­week recall period. Obviously many households would not have made any air trips in a given four­week period and for those that did there is no information on how many air trips they made in the previous year. However, the relationship between household characteristics and the average number of air trips and long­distance trips by other modes made during the four­week recall period can give an indication how air travel propensity and long­distance travel mode split vary with household characteristics. An analysis of data from the 2001/2002 NHTS is described in more detail in Appendix C. This analysis examined the increase in the number of air trips per year with household income based on the long­distance trips reported by survey respondents. The results of this analysis indicate that in 2001 U.S. households made approximately 191 million air trips, of which about 79% were made by households with incomes of $50,000 or more. These data allow a comparison of the increase in air travel propensity with income based on the survey responses. For respon­ dents in all income ranges the data imply an average of about 1.75 air trips per household, with a 10­fold difference between households with incomes below $25,000 and households with incomes of $50,000 or more. As discussed in more detail in Appendix C, this overall air travel propensity is well below the values found in more recent surveys, such as the Michigan Travel Counts Survey discussed below. Furthermore, the difference in average air trips per household between the three household income ranges is much wider than found in other surveys. Until these differences are resolved it would appear that the NHTS data on air trips are not only quite dated but should be regarded with considerable caution. Omnibus Household Travel Survey The U.S. Department of Transportation Bureau of Transportation Statistics (BTS) under­ took a monthly or bi­monthly Omnibus Household Survey of approximately 1,000 randomly selected households from August 2000 to October 2003, with an additional survey in October 2009. The survey included information on air travel as well as respondent and household socio­ economic characteristics and thus provides additional information on air travel to that given by the 2001–2002 NHTS, which was undertaken at approximately the same time. However, unlike the NHTS, the Omnibus Household Travel Survey not only asked about air travel in the previ­ ous month, but the surveys from February 2002 asked how long ago the respondents made a commercial flight. This allows some analysis of how air travel in the previous month is related to annual air travel propensity. Although each survey only included about 1,000 respondents, over the period from August 2000 to October 2003 surveys were performed in 29 months, giving a total of 29,705 responses, allowing fairly detailed analysis of subsets of the survey responses, including seasonal and regional differences. The survey questions relevant to air travel and the socioeconomics of the survey respondents evolved somewhat over the course of the survey period. The first survey, in

Sources of Disaggregated Socioeconomic Data 57 August 2000, only asked about the number of days on which a respondent flew in the previous 30 days, without distinguishing between commercial flights and general aviation flights. How­ ever, subsequent surveys had separate questions for commercial and general aviation flights. Prior to July 2001, the questions asked about air travel in the previous 30 days. From July 2001, subsequent surveys asked about air travel in the previous month. While the distinction is fairly minor, in calculating air travel propensity some adjustment needs to be made for months with fewer or more days. Also from July 2001, the survey asked how many days in the previous month the respondent made an air trip for business or work. An analysis of responses to the survey for the one­year period from August 2001 to July 2002 was initially performed. Survey data was collected for 11 months of this period, excluding September 2001, when air travel was disrupted by the terrorist attacks of September 11. During this period survey responses were obtained from 11,325 households, of which 1,293 (11%) reported making at least one air trip on a commercial flight during the previous month. Those survey respondents who reported taking one or more commercial flights during the previous month took flights on an average of 2.64 days in the month. Additional survey data were then analyzed for 9 months from August 2002 to October 2003. During this 15­month period, survey responses were received from 9,502 households, of which 1,070 (11%) reported taking one or more com­ mercial flights during the previous month. Those survey respondents took commercial flights on an average of 2.76 days in the month, a slightly higher use of air travel than reported in the survey for the period from August 2001 to July 2002, which seems reasonable given that the earlier period included the immediate aftermath of the terrorist attacks of September 11, 2001. The results of these analyses are shown in Table 23. Household Income Used Commercial Air Travel Average Days Used Commercial Air Travel Percent Made Business Air Trips Average Days Used Air Travel on Business Percent of Air Travel Days Making Business Trips August 2001 to July 2002 Under $15,000 2.2% 2.14 13.6 1.67 10.6 $15,000 - $29,999 5.0% 2.08 23.6 2.86 32.4 $30,000 - $49,999 6.7% 2.42 35.4 3.12 45.6 $50,000 - $74,999 12.1% 2.36 32.0 2.99 40.6 $75,000 - $99,999 18.5% 2.64 45.1 3.15 53.9 $100,000 or more 31.9% 3.19 56.3 3.73 65.7 Total 11.4% (1) 2.64 (1) 47.6 (1) 3.32 59.2 Valid responses (2) 9,717 Total responses 11,323 August 2002 to October 2003 Under $15,000 2.5% 3.10 23.8 2.00 15.4 $15,000 - $29,999 4.3% 2.25 17.5 2.36 18.3 $30,000 - $49,999 7.3% 2.32 22.1 2.88 27.5 $50,000 - $74,999 12.1% 2.70 40.8 2.95 44.6 $75,000 - $99,999 17.3% 2.91 45.2 3.59 55.8 $100,000 or more 31.0% 3.09 49.4 3.59 57.4 Total 11.3% (1) 2.76 (1) 45.6 (1) 3.30 53.9 Valid responses (2) 7,959 Total responses 9,499 Notes: (1) Includes respondents who did not indicate their household income. (2) Valid survey responses gave both household income and number of days on which they used commercial air travel in previous month (including none). Table 23. Use of commercial air travel in previous month by omnibus travel survey respondents by annual household income.

58 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies Further details of the analysis of these survey responses are presented in Appendix C. The most striking findings of the analysis of the survey results for both periods is that although the percentage of survey respondents who reported taking one or more commercial flights in the previous month increased strongly with income, with only about 2% to 3% of survey respondents with a household income under $15,000 taking at least one flight but over 30% of survey respondents with a household income of $100,000 or more taking at least one flight. The average number of days during the previous month when those who reported taking at least one flight used commercial air travel (which is a surrogate measure for the number of air trips taken) did not vary very much by household income, increasing slightly with income from a little over 2 days for survey respondents with household incomes between $15,000 and $30,000 to a little over 3 days for survey respondents with household incomes of $100,000 or more. This is perhaps not all that surprising, since even people who make many air trips per year may only make one or two in a given month. In addition to household income, the survey collected data on the respondent age, gender, ethnic group, and education level, as well as the number of household members age 18 or over. These data could be further analyzed to explore how the use of commercial air travel reported in the survey varied with these socioeconomic characteristics, although this was not done as part of the current project. State Household Travel Surveys The research team identified three recent statewide household travel surveys that included a long­distance travel component, although the survey for Utah had a much smaller sample size than the other two surveys as well as other limitations. Therefore it was decided to limit the analysis of statewide household travel surveys to the surveys for California and Michigan. Survey response data were obtained for the 2012–2013 California Household Travel Survey and the corresponding data for the long­distance component of the 2015 Michigan Travel Counts survey were provided by the staff at the Michigan Department of Transportation. California Household Travel Survey. The California Department of Transportation (Caltrans) undertakes the statewide California Household Travel Survey (CHTS) every 10 years (Caltrans 2016). The most recent survey was undertaken from January 2012 following a pre­ test in late fall 2011, and ending on January 31, 2013. The survey obtained complete travel data from 42,431 households (NuStats 2013). Travel data were collected from households in all 58 California counties as well as parts of three adjacent Nevada counties. Data was collected by a combination of computer assisted telephone interviewing (CATI), an online survey, and three types of global positioning system (GPS) devices. All participating households were asked to record their travel for a pre­assigned 24­hour period and complete a long­distance travel log covering trips by household members to a location 50 miles or more away in the 8 weeks prior to the assigned travel day. An initial recruitment survey collected an extensive range of information about each house­ hold that agreed to participate in the survey, including the following: • Household size • Total household income for the past year • Name, age, gender, and race of each household member • Relationship among household members • Country of birth of each household member and year moved to United States if not born in the United States • Employment status of each household member, together with details of employment • Student status and highest education level of each household member

Sources of Disaggregated Socioeconomic Data 59 The long­distance travel log provided space for recording up to eight long­distance trips of 50 or more miles made by any household member during the 8 weeks preceding the travel diary day, starting with the most recent trip. Thus if the household members made more than eight such trips, only the most recent eight trips could be recorded on the long­distance travel log that was provided. The information requested for each trip comprised: • The date of departure • The place name and address where the trip started • The place name and address of the final destination • The main purpose of the trip (using trip purpose codes provided) • The number of people traveling with the respondent • The number of household members traveling with the respondent and which ones • The method of travel that was used for the longest distance (using codes provided for each mode of travel) Respondents were instructed to treat each direction as a separate trip. This effectively limited the long­distance travel log that was provided to a maximum of four round trips. Since it covered all trips of 50 miles or more, it is quite likely that many households made far more long­distance trips than this during an 8­week period. Respondents who made more than eight long­distance trips during the 8­week period were instructed to record the details of the additional trips on a separate sheet of paper. An additional problem with the long­distance travel log is the wording of the instructions for entering the number of people traveling together. This refers to the number of people traveling with “you” (the person completing the log), i.e., excluding the respondent. However, this implies that the respondent was traveling on all the trips, which may not have been the case. The long­distance travel file contained records for 68,193 trip legs reported by 18,012 house­ holds, four of which did not report their household characteristics. Some records were missing data on the mode used or the number of people on the trip. These trip leg records were dropped from the analysis, which resulted in 67,125 trip legs reported by 17,736 households. The number of reported trips dropped off sharply above eight trip legs, suggesting that many households did not attempt to record more than eight trip legs. Although no household should report only one long­distance trip leg, since any round trip has to involve at least two legs, some 20% of households did so. This suggests that many households may have only reported one direction of a round trip, which is borne out by an examination of the trip data. The largest number of reported trip legs by a given household in the long­distance travel file was 50, by 10 different households. Although this is a very small fraction of the total households reporting long­distance trips (less than 0.1%) only one or two households in each case reported a number of trip legs between 41 and 49, suggesting that the number of reported trip legs recorded in the retrieval survey may have been limited to 50. These trip legs were then classified as outbound, intermediate, or return legs, based on the O&D zip code of each leg, where an intermediate leg was one that neither began not ended at the respondents home. This allowed the number of outbound legs of round trips to be counted, which gave a total of 38,115 outbound legs by 17,480 households. The lower number of house­ holds than those reporting long­distance trips was because a few households did not report any outbound trips. This could have resulted from the outbound leg of a trip taking place before the 8­week recall period, but more likely was because the respondents failed to report the outbound leg. Households reporting making long­distance trips during the recall period comprised 42% of all households participating in the survey. Of those households that reported making outbound long­distance trips during the recall period, 49% reported only making one such outbound trip.

60 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies Because some respondents reported air trips beginning or ending at an airport rather than their home zip code, those trip legs were classified as outbound or return legs if the airport in questions was within 150 miles of the home zip code. Examination of the data showed that a small number of reported air trips were made by general aviation, military flights, or were not in fact air trips and these were excluded from valid airline trips. Trips involving valid airline travel were reported by 3,666 households, or 20.4% of all house­ holds reporting long­distance trips and 8.6% of the 42,436 households participating in the survey. Nineteen households did not report the air travel party size for all their airline trip legs and these households were excluded from the analysis, since it is not clear how many air passenger trips those households made. The 3,647 households that reported the air travel party size for all their airline trips reported 5,117 air trips, or an average of about 1.40 air trips per household. The great majority of these households (about 70%) only made one air trip in the 8­week recall period. The largest number of air trips reported by a household was nine. Less than 10% of households reporting making air trips made more than two air trips and less than 5% made more than three air trips. The distribution of the number of households that reported making long­distance trips during the 8­week recall period and the number that reported making air trips during this period by household income is shown in Table 24. About 8% of households that reported making long­distance trips and 10% of households that reported making air trips did not report their household income. Table 24 also shows the average number of air trips reported by households in each income category that reported making air trips. As expected, it can be seen from Table 24 that a larger proportion of households that reported making air trips are in the higher income categories. The average number of air trips reported per household generally increases with household income, as expected, except for the lowest income category below $10,000. However, this income category had a fairly small number of households that reported making air trips. It is also consistent with the findings of air passenger surveys performed at airports that respondents in the lowest income category often report a higher rate of making air trips than those in the next income categories above them. Although the average air trips per household shows an increase with increasing household income, the difference between households with annual incomes below $50,000 and those with annual incomes above $250,000 is not particularly large. However, these averages should be treated with some caution, since they only reflect air trips made over an 8­week period, so the Household Income All Long-Distance Trips Air Trips Avg Air Trips per Household (1) Households Percent Households Percent Up to $9,999 348 2.1 31 0.9 0.17 $10,000 - $24,999 1,173 7.1 96 2.9 0.17 $25,000 - $34,999 1,079 6.5 123 3.7 0.33 $35,000 - $49,999 1,688 10.2 214 6.5 0.49 $50,000 - $74,999 3,025 18.2 446 13.5 0.71 $75,000 - $99,999 2,905 17.5 523 15.8 1.09 $100,000 - $149,999 3,466 20.9 799 24.1 1.67 $150,000 - $199,999 1,488 9.0 458 13.8 2.40 $200,000 - $249,999 682 4.1 263 7.9 3.37 $250,000 or more 750 4.5 361 10.9 5.17 16,604 100 3,314 100 1.13 Total (2) 18,012 3,666 1.15 Note: (1) Air passenger trips per year. (2) Includes respondents who did not indicate their household income. Table 24. Long-distance trips reported by California household travel survey respondents by household income.

Sources of Disaggregated Socioeconomic Data 61 likelihood of the members of any given household making more than one air trip in this period is quite small, even if the household makes several air trips per year. This is reflected both in the high percentage of households (70%) that reported making air trips that reported only one such trip and the high percentage of households (80%) that reported making long­distance trips in the 8­week period that did not report making any air trips at all. Although the survey provides a direct measure of the number of air trips reported for the 8­week period prior to the assigned travel day, in order to estimate the average number of annual air trips made by households with given characteristics, it is necessary to convert the number of air trips in an 8­week period to the number in a year. Since the survey was performed over the course of a little over a year from January 2012 to January 2013, the survey responses cover air travel undertaken over the 15­month period from November 2012 to January 2013, although this should not affect the analysis of air travel propensity of households with different charac­ teristics, since each household only reported the number of air trips in an 8­week recall period. Assuming that the number of air trips made in the 8­week recall period is representative of the average rate of air travel over the year for any given subset of households or for the survey respondents in total, the average number of air trips per household given by the survey responses was multiplied by 6.5 (52/8) to give the average number of air trips per year shown in Table 24. The overall average number of annual air passenger trips per household of 1.15 is well below the average number of annual air trips reported by respondents to the air passenger intercept surveys discussed above. There are several factors that could, at least partially, account for this difference. The first is that it is not known how many of the households that did not report any air trips in the 8­week recall period in fact made no air trips in the previous year. Obviously such households would not show up in an air passenger intercept survey, which would tend to reduce the average number of air trips per household compared to those found in an air passenger survey. The second factor is that air passenger intercept surveys are surveys of air passenger trips, not of households. An air traveler who makes a large number of air trips per year is more likely to be interviewed in an airport intercept survey than an air traveler who only makes one or two air trips per year. This will tend to inflate the average of the reported number of air trips per year. On the other hand, survey respondents are typically reporting the number of air trips that they personally made during the previous year, which will obviously be less than the number of air passenger trips made by their household in total. We were not able to come up with a definitive explanation for this discrepancy, but this is an important topic for future research in order to better understand these differences in reported air travel propensity between the findings of household travel surveys and airport intercept surveys. Leaving aside the issue of the actual air travel propensity values, because the CHTS collected a large amount of detailed socioeconomic data about each household that participated in the survey, the survey data can be used to explore relative differences in air travel propensity across different household characteristics. A more detailed analysis of the CHTS data is contained in Appendix C, including an analysis of the differences in air travel propensity by age and race/ ethnicity of the survey respondent. Michigan Travel Counts Survey. Starting in 2004, the Michigan Department of Trans­ portation has undertaken a series of statewide household travel surveys termed MI Travel Counts. The first survey was undertaken in 2004–2005 and updated in 2009. The latest survey was undertaken in 2015, starting in January and continuing throughout the year. Travel data was obtained from 16,276 households across the state. Participating households provided information on household composition and other characteristics and each household member completed a travel log for an assigned day. In addition, each household completed a long­distance travel

62 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies log that covered all trips over 100 miles made by household members in the 3 months prior to the assigned travel day. Information on household composition and characteristics collected included the age, gender, and employment status of each household member, the vehicles owned by the household members, and the household income. The long­distance travel log included the following information for each long­distance trip: • Destination city and state • Household members traveling on trip • Dates of departure and return • Main reason to making the trip • How the travel party got to the destination and got around at the destination • The number of times that this trip was made in the previous 3 months The reason for the trip and means of travel were free­form entry fields on the travel log, although these were coded when the data were reported. If a trip involved multiple destinations, respondents were instructed to report the furthest destination. Of the 16,276 households responding to the survey, 9,961 (61%) reported making long­distance trips. The long­distance travel file contains one record for each long­distance trip reported by survey respondents. In addition to collecting details in each long­distance trip undertaken in the previous 3 months reported by survey respondents, the survey asked if any of those trips were made multiple times during the 3­month period and if so, how many times. If respondents reported that a particular trip was made multiple times in the previous 3 months, the survey also asked how many times the trip had been made in the previous 12 months. One potential problem with this approach is that it is unclear whether the same number of household members went on each trip in the case where the same trip was made multiple times. Indeed, it is unclear how the “same” trip was defined. Was a trip considered the same if it was to the same destination, or did other aspects of the trip, such as the main mode used, also have to be the same? Obviously, it does not matter if survey respondents reported multiple trips to the same destination as separate trips, since the information for each of these trips would be included in the long­distance travel file. However, if multiple trips to the same destination were reported as the “same” trip when in fact they involved different numbers of household members or even the use of a different main mode, then the resulting number of air passenger trips could be under­ or overestimated. A second problem with estimating the number of air passenger trips reported by survey respondents results from the fact that survey respondents did not identify which household members went on 23% of all the reported air trips or, in a few cases, how many times a given trip was made in the previous 3 months. The missing travel party size data can be corrected for some analyses by assuming that the average number of household members who went on trips where this was not stated is given by the average number of household members who went on air trips where the travel party size was reported. Analysis of the survey data on repeated trips suggests that on average survey respondents make somewhat more than four times the number of reported air trips in a 3­month period on an annual basis. The survey respondents reported 738 repeated trips in the previous 3 months and 3,132 repeated trips in the previous 12 months, or about 6% more than four times the number in the previous 3 months. Taking account of the number of household members traveling on each air trip and the number of times a given trip was repeated in the previous 3­month period, the average number of air passenger trips made in the previous three months and the average number of air trips for each household income range are shown in Table 25.

Sources of Disaggregated Socioeconomic Data 63 The overall air travel propensity shown in Table 25 is equivalent to 9.6 passenger trips per household on an annual basis, assuming that the numbers of annual trips are four times the numbers reported for the previous 3­month period. This is slightly higher than the weighted average of about 8.7 air trips in the previous 12 months reported in the airport air passenger surveys shown in Table 9. However, it should be noted that the air travel propensity given in Table 25 is the number of air passenger trips per household, while that given in Table 9 is the average number of annual trips reported by each survey respondent. Presumably the survey respondents were reporting the number of trips that they had made personally, not the number made by all members of their household. The average number of air trips per household shown in Table 25 appears to decline with income for annual household incomes below $35,000 and then generally increase thereafter. However, the increase in the average number of air trips with income is not as great as found in airport air passenger surveys, as shown in Table 9. This most likely reflects the differences between surveys of households and intercept surveys of air passengers discussed earlier. The average air travel party size also tends to increase somewhat with household income, as shown in Table 25. As a result, the average number of air passenger trips per household shows a stronger increase with household income than the number of air party trips and the increase in average air party size largely offsets the drop in the average number of air trips with income below an annual household income of $35,000. A more detailed analysis of the Michigan Travel Counts data is presented in Appendix C. Summary and Conclusions Although the household travel surveys analyzed in the course of the project provide useful information in how air travel propensity varies with household characteristics, limitations in the way that each survey collected data on air trips make it difficult to compare the results with air passenger surveys that ask how many air trips respondents had made in the previous year. The findings of the Omnibus Travel Survey are particularly interesting because this was a nationwide survey that covered all households, whether or not they had made air trips. Some 11% of all households reported that they had used commercial air travel in the previous month. Of those households that had used commercial air travel in the previous month during the period from August 2002 to October 2003, they had used air travel for an average of about 2.8 days. As could be expected, the use of air travel increased with household income, both in terms of the percentage of households that had used commercial air travel during the previous Household Income Air Passenger Trips Air Party Trips Avg Air Party Size Households Percent Pax Trips Avg per h/h Air Trips Avg per h/h Less than $15,000 14 1.3 24 1.71 19 1.36 1.26 $15,000-$24,999 25 2.3 46 1.84 32 1.28 1.44 $25,000-$34,999 40 3.7 73 1.83 44 1.10 1.66 $35,000-$49,999 68 6.4 153 2.25 79 1.16 1.94 $50,000-$74,999 221 20.7 420 1.90 248 1.12 1.69 $75,000-$99,999 190 17.8 407 2.14 232 1.22 1.75 $100,000-$124,999 205 19.2 434 2.12 252 1.23 1.72 $125,000-$149,999 111 10.4 267 2.41 150 1.35 1.78 $150,000 or more 195 18.2 710 3.64 349 1.79 2.03 Total 1,069 100 2,534 2.40 (1) 1,405 1.31 (1) 1.83 (1) Note: (1) Average air passenger trips, air trips per household, and average air party size for all households include respondents who did not indicate their household income. Table 25. Air passenger trips in previous 3 months reported by Michigan Travel Counts survey respondents by household income.

64 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies month and the average number of days in which they had used air travel. Also, as could be expected, of those households that had made air trips, the percentage that had made business air trips increased strongly with income, as did the average number of days that those respondents used air travel for business trips. As a result of these two effects, the percentage of the average number of air travel days on which respondents made business trips increased strongly with income, from about 15% for households with incomes under $15,000 to about 57% for house­ holds with incomes of $100,000 or more. The average number of air travel days that involved business trips was about 54%. Although this suggests that somewhat over half of air travel was for business, this could be misleading for two reasons. The first is that personal air trips typically have a higher average air party size than business trips, while the second is that a higher proportion of business trips are likely to be same­day trips than personal trips. The Michigan Travel Counts survey results are broadly consistent with the findings from airport air passenger surveys in terms of the overall average air travel propensity, although the average number of air trips per household reporting air trips in the previous 3 months does not increase with household income as much as typically found in airport air passenger surveys. As noted earlier, this is likely to result from only considering air trips in the previous 3 months. In order to effectively compare the findings from the various household travel surveys with those from airport air passenger surveys, the use of air travel reported in the household travel surveys needs to be adjusted for the fact that the period reported is much less than a full year. Further research is needed to establish a robust procedure for making these adjustments. Summary and Implications for Air Passenger Demand Studies The findings of the analysis of air passenger and household travel surveys described earlier have important implications for air passenger demand studies. Not surprisingly, the survey results show that air travel propensity, expressed as the average number of air trips in the previ­ ous year, varies considerably by household socioeconomic characteristics, including income, age of the survey respondent, gender, and race/ethnicity. Naturally, some of the characteristics are correlated, such as household income and age of the head of household (or at least the household member responding to the survey), so some of the differences in air travel propensity with any one of these characteristics may be partly due to correlation with other characteristics. Ideally, it would be desirable to develop a model of air travel propensity that considers all the relevant household characteristics simultaneously. The survey findings also show, again not surprisingly, that respondents making a business trip tend to have a higher air travel propensity on average than those making a personal trip. Surveys that have asked how many trips respondents have made in the past year by trip purpose clearly show that the average number of air trips for business purposes by those who made at least one business air trip is considerably higher than the average number of personal trips by those who made at least one personal trip (which includes almost all air travelers). However, many surveys only ask how many air trips respondents made in the past year in total (if they ask about previous air travel at all). Even so, because travelers who have made at least one business air trip in the past year make more air trips per year on average than travelers who have only made personal air trips, the average number of total trips per year by travelers making a business trip when they were surveyed is higher than for those making a personal trip. The analysis of air passenger and household travel survey data also shows that air travel pro­ pensity varies with the age of the survey respondent. In the case of multiperson households, it is assumed that the age of the survey respondent is a good proxy for the respondent’s spouse or

Sources of Disaggregated Socioeconomic Data 65 partner, although of course this is not always the case. The situation is a little more complicated in the case of households with more than two adults, since the other adults could be adult children or parents or other relatives of the survey respondent and hence could be significantly younger or older than the survey respondent. Nonetheless, it seems reasonable that air travel propensity would change at different points in a person’s life cycle for any given income level. Therefore changes in the age distribution of the population will affect air travel demand, separately for the effects of changes in household income. Appendix C provides a more detailed discussion of the implications for air passenger demand studies of the findings of the analysis of the different disaggregated socioeconomic data sources summarized in this section. Based on the analysis of air passenger and household travel survey data, the principal household characteristic from the perspective of the influence of these characteristics on air travel propensity is annual household income. The survey results show that although air travel propensity increases progressively with household income, it is not directly proportional to household income (at least as reported by survey respondents). Appendix C presents functional relationships that have been fitted to the data from five airport air passenger surveys for the increase in average annual air trips with household income for respondents making business and personal air trips. These relationships are generally consistent across the five surveys after expressing household incomes in constant dollars to account for the growth in real incomes over the period covered by the surveys, although the relationship for one survey in each case shows a somewhat different pattern from the other four surveys for reasons that are unclear. All five surveys show a decline in the rate of increase of average air trip propensity with increasing income, suggesting that incorporation of disaggregated household income in air passenger demand studies needs to be done in a way that reflects this declining marginal propensity at higher income levels. These fitted functional relationships for the change in the average number of annual air trips (AAAT) with annual household income (AHI) used an inverse tangent mathematical function: AAAT atan AHI[ ]( )= × −a b c where a, b, and c are estimated parameters. This function was found to give a good fit to the data while reflecting both a diminishing rate of growth of average annual air trips with increasing income, as shown by the survey data, and a continuously increasing relationship between average annual air trips and income, as required by logic. However, these are purely empirical relationships and not based on any underlying causal explanation, such as a relationship between disposable household income and the percentage of disposable income used for air travel. The functions for survey respondents making a business trip were significantly higher than for those making a personal trip (approximately twice as high), which seems inherently reasonable. However, it should be noted that these functions are for total annual trips in each case, not for business and personal trips separately. The varying slope of the air travel propensity functions with household income suggests that simply using an aggregate measure of household income (or similar measure of overall economic activity such as gross regional product) in air travel demand models will fail to fully capture the effect of changes in real income distribution and the associated changes in air travel propensity. The functional relationships of air travel propensity with household income have two other important implications. The declining slope of the functions with increasing household income means that if real household incomes rise over time, the resulting change in air travel propensity for any given increase in income becomes less. Therefore it can be expected that the relationship between air travel and aggregate household income (or similar measures of overall economic

66 Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies activity) will tend to decline over time as real incomes rise unless offset by reductions in the real cost of air travel. The second implication arises from the significant difference in the average number of annual air trips at any given level of household income by those making a business trip and those making a personal trip. This means that the total amount of air travel at a given airport generated by any particular distribution of annual household incomes in the region served by the airport will depend on the relative proportions of business and personal air trips by the residents of the region, which is likely to vary from region to region, depending on the composition of the local economy. Given the effect on overall air travel propensity of changes in household income and the age distribution of the population, as well as the potential effect of changes in other household characteristics and the significant impact on overall air travel propensity of the relative propor­ tions of business and personal trips in any given market, it seems unlikely that these complex interactions can be adequately reflected in air travel demand models by including just one or two socioeconomic variables that measure a limited number of dimensions of the underlying distributions. Instead, it may be more effective, and more flexible, to include an estimate of aver­ age air travel propensity in the models directly and develop a procedure to calculate the average air travel propensity from the underlying distributions of household income, age of the head of households, and other relevant socioeconomic characteristics, based on relationships estimated from air passenger and household travel surveys. Historical data on the underlying distributions of household income, age of the head of households, and other household characteristics are generally readily available at a national and regional level. Projections for changes in the distri­ butions for future years are also available for some data, or scenarios for future changes in the distributions can be developed from current trends in the distributions and assumptions about potential changes in those trends. Future Research Needs It is clear from the analysis undertaken in the current project that existing surveys often do not ask questions that would be helpful to a more comprehensive analysis of air travel propensity. Such limitations include failing to record the gender of the other adult members of the air travel party, only asking about total air trips in the past year by survey respondents, rather than distinguishing between business and personal trips, limiting the ranges of household income, particularly higher incomes, when presenting respondents with ranges of income to select or recording responses using income ranges, and omitting questions about race and ethnicity or household size. One socioeconomic characteristic that is difficult to address in a survey is the sector of the economy in which business travelers are employed. However, this information would be extremely valuable to improve the understanding of how the extent of business air travel differs across different sectors of the economy. Developing appropriate and practical survey questions to collect this information, as well as the necessary analysis techniques to translate survey responses to standard codes, would be a valuable future research activity. Although there are practical limits on how many questions can be added to an air passenger intercept survey before it becomes too long to expect travelers to answer, there are ways to address this problem through the design of the survey. For example, there could be different versions of the survey questionnaire that address different issues or follow­up questions could be asked depending on the response to given questions. The increasing use of programmable devices to collect survey data makes this fairly straightforward. Guidance on the appropriate design of survey methodology and question wording would be very helpful in ensuring that air passenger surveys generate the necessary data to improve our understanding of air travel propen­ sity and better support future air passenger demand studies.

Next: Chapter 4 Case Studies in Modeling Airport Passenger Enplanements Using Disaggregated Socioeconomic Data »
Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s Airport Cooperative Research Program (ACRP) Research Report: 194: Using Disaggregated Socioeconomic Data in Air Passenger Demand Studies explores the potential benefits of using disaggregated socioeconomic data, such as regional household income distributions and air passenger and travel survey data, for air passenger demand studies.

Aviation demand is strongly correlated to socioeconomic activity, and analysts typically use aggregate socioeconomic data, such as gross regional product or average regional household income, to better understand current and potential future aviation demand.

Since the United States is experiencing significant and ongoing demographic trends there is a question as to whether traditional methods and data sources will adequately capture these trends or would more detailed, disaggregated socioeconomic data, or even nontraditional data provide more accurate results.

This report summarizes long-term socioeconomic trends, attempts to understand their potential impact, and provides guidance for incorporating disaggregated socioeconomic data into air passenger demand studies.

The following appendices to ACRP Research Report 194 are available online:

Appendix A: Detailed Survey of Past Analyses of Air Passenger Demand

Appendix B: Airport Industry Use of Socioeconomic Data for Air Passenger Demand Studies

Appendix C: Additional Material on Sources of Disaggregated Socioeconomic Data

Appendix D: Detailed Case Study Analysis Results

Appendix E: Background on Other Analytic Approaches

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!