Many years ago, when I was playing SimCity 4, I remember one of the advisors insisting that I should build an airport to attract business to my city. Then I went on a trip to Cumberland Maryland, and I realized that, while they have an airport serving the area (Cumberland and the Potomac Highlands of West Virginia), it was strictly for general aviation. There was no commercial air service for many, many miles. Knowing Cumberland's status as a fairly isolated Appalachian city that has fallen on hard times, I was curious if its lack of nearby commercial air service was a contributing factor to its economic issues.
To perform this very rudimentary study, I downloaded a file of airports in the United States and its metropolitan statistical areas. I then went into QGIS and ran the vector analysis tool, "distance to nearest hub (line to hub)" using the MSA as the source "points" (causing these calculations to be based on the geographic center of the MSA) and the airport as the hub. The resulting map (including metropolitan and micropolitan statistical areas) can be viewed here.
Using MSA personal income data from the Bureau of Economic Analysis and entering the data into LibreOffice Calc, a regression was created using the data. The file can be downloaded in XLSX format here. The main sheets of interest are AllAirports, 200kAirports (airports with 200,000+ enplanements), PersonalIncome, and AllAirportsTest (regression results). The pivot tables mainly exist just to get a sense of the airport data. In the interest of full disclosure, the regressions described below do not include income for micropolitan statistical areas. The only include income for metropolitan areas, as described by the Census Bureau.
As it turns out, there is no significant relationship between an MSA's airport distance and its personal income. The regression of Metropolitan Income = a + b * Airport Distance shows a = 48199.2334 and b = -43.2840, meaning that being 1 mile further away from an airport decreases a metropolitan area's personal income, on average, by $43.28. However, the R^2 is 0.0024, meaning distance from an airport, if it is partially the cause of the region's income levels, only explains 0.24% of it. Continuing, the p-value associated with b is 0.3664, which is nowhere near statistical significance. This means that, based on all airports in the GIS dataset, there is no significant relationship between a region's income levels and their distance from an airport.
To make another attempt to determine if distance from an airport can affect an MSA's income, only airports with at least 200,000 enplanements were included in the next regression. The regression is the same before, but instead the nearest airport with at least 200,000 enplanements was used as the "x." 200,000 enplanements was chosen as the Wilkes-Barre/Scranton International Airport has approximately that many enplanements. The reason for choosing that airport is because Wilkes-Barre/Scranton is on the smaller side of metropolitan areas in the country as a whole, but among the larger Appalachian regions, which were the areas I was particularly interested in when researching this topic.
By restricting to only airports with at least 200,000 enplanements, a = 50238.2855 and b = -60.3394; this time meaning that being 1 mile further away from an airport decreases a metropolitan area's personal income, on average, by $60.33. While the R^2 is still very low, at 0.0388 (meaning 3.88% of variation in an MSA's income could be explained by distance from an airport), this time the p-value for b is well below 0.01. This means that, with the more restricted dataset, there is a significant negative correlation between an MSA's distance from an airport with at least 200,000 enplanements and its average personal income.