The General and Market-Specific Impacts of Airport De-Hubbing on Airfares: A Research Paper
Over the past three months, as a summer project, I wrote a research paper that analyzes the effects that de-hubbing has on airfares departing from former hub cities. Although I am an undergraduate student at the University of Chicago, I did write this paper independently from the University. This paper is different from anything else I’ve written on this website in that it’s written in a much more academic style and attempts to perform a much more empirical analysis of data. In the paper, I use publicly available data about airfares and flight segments from the Bureau of Transportation Statistics to measure the effects of hub closures by airlines on airfares at such former hub airports. Specifically, I look at the changes in airfare that occurred at the airports of Cleveland, Cincinnati, Memphis, Pittsburgh, and St. Louis.
I at least try and explain my research very precisely in the actual research paper itself (which perhaps made it a bit longer), so I will refrain from repeating myself here even more than I already have. To better explain the paper, I’ve posted the abstract and introduction to my paper directly below in this article, and attached the full research paper below. Thank you for reading, and I hope that this paper perhaps provides useful insights.
Abstract
This paper studies the effects of de-hubbing, the process by which an airline closes hub operations at a certain airport, on airfares. In studying five cases of de-hubbing that occured between 2000 and 2019, I first examine the consequences that de-hubbing has on airfares from flights departing from the former hub airports. Second, I analyze the effects on airfares in markets out of former hub airports that lost nonstop commercial air service due to de-hubbing. Largely consistent with previous studies, I find that average airfares decrease post-de-hubbing at former hub airports where there is an increase in low-cost carrier presence post-de-hubbing. However, I find no statistically significant shift in average airfare at hub airports where there is no substantial increase in low-cost carrier operations post-de-hubbing. Additionally in city-pair markets where nonstop departing service is discontinued post de-hubbing, compared with overall airfare trends for the former hub, I find either no significant difference or a negative change in average airfare, indicating that downward trends in airfare at former hubs may be less prevalent in markets that lose nonstop service.
Introduction
Since the Airline Deregulation Act of 1978, many legacy air carriers have established considerable parts of their route networks across hub-and-spoke systems. Legacy airlines, also known as network or full-service carriers, include airlines such as American Airlines and United Airlines, while also historically included airlines such as TWA or Northwest. Colloquially, airline hubs are often characterized as airports with sufficiently large levels of traffic, or perhaps as airports of a sufficient size dominated by one carrier. However, airline hubs are often also characterized by the role that they play within the airline’s larger route network, acting as nodes that connect passengers from one destination to another. In the hub-and-spoke mode, airlines route passengers from “spoke” origin airports through hub airports that serve as intermediate stops before sending them on another flight to then arrive at their final destination airport, other “spokes” of the route network.
Although some airlines only operate one single hub, the largest legacy US carriers historically have functioned with multiple hubs. A key benefit of hub-and-spoke networks is that they open up numerous combinations of one-stop city pairs that are usually economically unfeasible to serve nonstop. Thus, legacy airlines often optimize their key operating airports to be tailored toward connecting passengers. Accordingly, airlines often create “banks” of flights at their hubs that arrive and depart within a relatively short period of time to minimize connecting times for passengers. In this paper, I examine the consequences on airfares for passengers originating from former hub airports that occur when airlines shut down specific hubs within their network. This process is often called de-hubbing.
De-hubbing can occur for various reasons, such as the unprofitability of a hub, the shutdown or restructuring of an airline, or because of an overlap with another hub of the same airline that makes the presence of two geographically similar hubs inefficient. The latter case often arises after two airlines merge, when operations from two previously competing hubs in a relatively close geographic area are, largely for efficiency purposes, consolidated at a single airport. Although local traffic plays a significant role at most hubs, because of the importance of connecting traffic, airport hubs, by definition, most often have more departures and available seats than can be filled by local traffic alone. Thus, when airlines choose to shut down a hub at airports that have disproportionally large quantities of connecting passengers (compared with the number of local passengers) unless another airline decides to fill in and create a new hub in its place or expand an existing hub at the same airport, the total supply of flights and seats offered out of the former hub airport will generally decrease. In this paper, I largely attempt to analyze the effects of this apparent negative supply shock that occurs during de-hubbing.
Not all airlines operate hub-and-spoke networks; instead, some airlines operate point-to-point networks. In pure point-to-point route networks, airlines do not operate any connecting traffic. All passengers fly directly between their originating city and destination, all on one nonstop flight. Most airlines in the United States that operate point-to-point networks are low-cost carriers (LCCs), which generally operate with lower operating costs than legacy carriers and often offer lower fares than legacy carriers. Nevertheless, the contrast between hub-and-spoke networks and point-to-point networks is often unclear. Airlines, such as Southwest, with primarily point-to-point route networks often have airports with significant operations that can counter the size of many airport hubs of legacy carriers. Historically, such airports, often termed “focus cities,” have not primarily had network schedules predominantly established around providing connections. Regardless, many low-cost airlines eventually began allowing passengers to book connecting flights on a single reservation, thus blurring the difference between hub-and-spoke and point-to-point networks.
In this paper, while I only examine airports that were hubs of legacy carriers, I do consider the entry of low-cost carriers into airports that formerly were hubs of legacy carriers. After de-hubbing occurs, low-cost carriers commonly increase service at the former hub airport (Tan and Samuel, 2016) and sometimes operate routes that were discontinued by the former hub carrier. However, given that low-cost carriers generally carry less connecting traffic, most new routes taken up by LCCs are to destinations with only the most nonstop demand from the former hub airport.
For this paper, I will be broadly defining airport de-hubbing as when a legacy airline significantly reduces its presence, in both the number of departures and the number of departing seats, at a major airport where the airline had a significant market share. In a later section, I provide a more detailed definition of de-hubbing for this paper. I consider five instances where airlines shut down hub services at airports between the years 2000 and 2019. These former airline-airport hub operations include United Airlines at Cleveland Hopkins International Airport (CLE), Delta Air Lines at Cincinnati/Northern Kentucky International Airport (CVG), Delta Air Lines at Memphis International Airport (MEM), US Airways at Pittsburgh International Airport (PIT), and American Airlines at St. Louis Lambert International Airport (STL). Although I do not specifically explore the causes of each airline to de-hub these particular airports, four of the five instances occur within five years after the hub airline merged with another airline. In each case of de-hubbing, the surviving airline had at least one other hub located within 400 miles of the hub that was shut down. Thus, it is conceivable that newfound proximity to other hubs gained after mergers play some role in the decision to shut down airport hubs.
I perform two analyses that measure the consequences of de-hubbing on airfares. The first trial attempts to determine the monetary consequences of de-hubbing on passengers in the hub city by comparing average airfares from flights departing the hub airport of interest before and after de-hubbing. The second trial seeks to determine whether airfare changes occur in markets where nonstop flights were discontinued post-de-hubbing and how this price difference compares with the results from Trial 1. While before de-hubbing, passengers flying on routes considered for Trial 2 would have the option of flying nonstop to the destination city, post-de-hubbing, such travelers must fly through a separate hub to reach the destination.
Inherently, de-hubbing results in greater inconveniences to passengers who live in the metro area of the hub airport due to the discontinuation of some nonstop flights from the hub airport, which is a characteristic of de-hubbing (Redondi, Malighetti, and Paleari, 2010). However, to my knowledge, there has been no paper that has attempted to quantify the consequences on market airfares caused by the discontinuation of nonstop service. Specifically, in Trial 2, I randomly choose ten airports that lost nonstop service to each of the five hub airports and then measure the weighted average market fares from each hub airport to their respective ten destination airports both before and after de-hubbing. Given that LCCs generally operate point-to-point networks and did not take up nonstop service in any of the markets considered in Trial 2, this trial is furthermore meant to analyze the impacts of de-hubbing in markets only minimally influenced by LCC service or competition. In both trials I use difference-in-difference (DID) regressions that compare the respective average airfares from flights originating at the former hub airport with national airfare averages as the control. Also, I further compare difference-in-difference analysis results from both trials with one another to ascertain whether there are substantially different trends in airfare in markets that lose nonstop service after de-hubbing.
In Trial 1, I find that de-hubbing contributes either to a non-significant change or a decrease in average airfares at all former hub airports considered. Specifically, at CLE, MEM, and PIT, de-hubbing contributes to an overall statistically significant decrease in average airfares, whereas at CVG and STL, I was not able to ascertain a statistically significant increase or decrease in average airfares caused by de-hubbing. I additionally find a direct correlation between airports with a decrease in average airfares after de-hubbing and airports with an increase in LCC operations post-de-hubbing. This indicates that LCC market entry into airports with former hub operations may help cause a decrease in average fares at such airports. In Trial 2 I find that de-hubbing contributes to a statistically significant decline in average market airfares to airports that lost nonstop service at CLE and PIT, a statistically significant increase in average airfare at STL, and no statistically significant change in average airfare at CVG and MEM. These results reveal that in markets from hubs to airports that lose nonstop service post-de-hubbing (and thus are also only minimally influenced by LCC market entry), changes in average airfare are either not significantly different from or less negative than the total average changes in airfare out of the hub airport (found in Trial 1).