Air travel mostly resumes for tourists but not business passengers: An airline case study
Authors' Note: For valuable comments, we thank Scott Marcus and Francesco Papadia at Bruegel, as well as Olivier Blanchard, Chad P. Bown, Martin Chorzempa, Cullen Hendrix, Mary Lovely, Marcus Noland, Maurice Obstfeld, David Wilcox, and Alan Wolff at the Peterson Institute for International Economics. We also thank Madi Sarsenbayev and Egor Gornostay for meticulous data assistance. Because the source and data underlying this analysis are confidential, they are not disclosed at the request of the airline. The replication code is provided at the end of the blog.
A summer of chaos has gripped airlines and airports globally, as international travel bounces back after a two-year pandemic-induced slump. Soaring fares, staff shortages and strikes, lost baggage, and other travel nightmares are widespread in some of the busiest airports and airlines in the EU, the UK, and the US. Airlines fear crippling people's summer travel plans and are rehiring and offering pay raises to their staff to keep planes in the air, while also simply cancelling many flights—at Amsterdam's Schiphol airport, for example, up to 11 percent of the flights were cancelled in a single weekend in early June. Labor strikes have also erupted at the height of the busiest holiday season. Airports and airlines that had cut back during the pandemic are ill-prepared to deal with the stark activity rebound after two years of travel anemia.
The COVID-19 pandemic indeed led to an unprecedented shutdown of international air traffic, and the ongoing chaos is a direct backlash of this crash. Global passenger traffic is estimated to have dropped 60 percent in 2020, 50 percent in 2021, and about 25 percent in the first half of 2022, compared with 2019 levels—a cumulated loss of more than 5 billion passengers over the period.
To what extent has this slump affected different destinations? Has it disrupted business ties as well as tourism? Will the pandemic shock leave a permanent legacy on international aviation? Detailed, proprietary data from a major European airline give a sense of the depth of the aviation crisis and reveal the uneven recovery of the sector. The data show that tourist destinations have caught up on average 20 percentage points more than business destinations and around 12 percentage points more than other destinations (not necessarily business or tourist). Two and a half years after the pandemic, tourism is finally on its way to completing its recovery.
The rebound in business destinations, however, consistently trails others. This is concerning because face-to-face interactions are known to be necessary to foster business dynamics, and many such interactions cannot be substituted with videoconferencing. Another crucial finding is that China's zero-COVID policy has isolated the country from international air traffic in a dramatic and potentially long-lasting way. International traffic there has remained nearly nonexistent since March 2020. This air traffic shutdown illustrates the extent to which the pandemic has redefined China's outward orientation.
The paradox of traffic gluts
"Air traffic gluts" created by the ongoing travel rebound mirror persistent disruptions in global value chains. This is despite the fact that neither overall US air traffic nor airport employment has recovered to its pre-pandemic level, as shown in figure 1. Furthermore, it is paradoxical to observe an actual decline in air transport system capacity in a context where airport employment is very close to precrisis level, while traffic remains significantly below.
Detailed data would be needed to explore the reasons for this apparent paradox. A possibility is that employment does not display airline employment, which also declined. More generally, the overall traffic capacity of the air transportation system appears to have been reduced by labor shortages in some key maintenance positions, in both airports and airlines, for instance, crucial ground handlers at Heathrow airport. The chaos has thrown off balance an intricate system whose smooth working requires the coordination of hundreds or thousands of hardly substitutable tasks. In technical terms, the shock has resulted in a persistent outward shift of the Beveridge curve. Hysteresis effects will surely dissipate, but the process is taking time.
Investigating turbulence in tourist and business air traffic
The most obvious channel funneling the aviation shock is the immediate and well-documented impact it has had on international tourism and countries that depend massively on it. The contribution of tourism to GDP is indeed the best predictor of the 2020 growth shortfall triggered by the crisis. Tourism is by nature a face-to-face activity, and this summer—the third since the pandemic—will be crucial for its full recovery.
The less obvious channel is the longer-term impact of the decrease in business interactions. Evidence indeed suggests that as business ties are fueled by face-to-face contacts, air traffic shutdown may have nontrivial hysteresis effects on overall investment and, ultimately, trade. The magnitude of these effects is hard to grasp and could differ dramatically from region to region, depending on the thickness of existing business ties and on the effectiveness of virtual substitutes. Surely, some meetings in the future will be replaced by videoconferences. But in the long term, scarce individual contact could weigh on the transfer of know-how and the geography of global value chains.
To find out if such effects are already observable, it is important to start from detailed data on how the crisis affected touristic and business travel separately. A major European airline provided us with a thorough dataset on their flights, covering a period spanning from January 2019 to April 2022. Crucially, the dataset provides monthly details on the number of passengers for around 300 different destinations, of which about 200 are international flights.
A glimpse into the crisis
Figure 2 plots month-to-month changes with respect to 2019 in the number of passengers for all destinations serviced by the major airline. It provides a graphic picture of the depth of the air traffic shutdown, with a nearly 100 percent drop in overall number of passengers in March-April 2020. Since the dataset is from a European airline (with 44 percent of international destinations in Europe), the March-April 2020 crash corresponds to the first lockdowns in European countries, which were implemented simultaneously and lifted roughly concurrently as well.
As of April 2022, the number of passengers had not recovered to its pre-pandemic level (figure 2). Considering that overall activity and GDP levels had largely recovered from the pandemic, this lag in air travel is both an indicator of how quickly the world adapted to the absence of flights and a hint that the crisis may have hysteresis effects through air traffic reduction, if the latter is indeed associated with weaker activity through reduced business travel.
A "peak" selection of business and tourist destinations
To disentangle the short- and long-term effects of the aviation crisis, we ranked destinations from the most business-intensive to the most tourist-intensive in the airline's dataset, based on a simple methodology.
To avoid selection problems such as relying misleadingly on fare categories or wrongly picking winter tourist destinations by using a summer-season approach, we sorted tourist intensity of destinations by the size of "peaks" in their monthly number of passengers in 2019. Business destinations should indeed display smaller peaks than tourist destinations, which conversely have much bigger peaks in short periods—to briefly rephrase the underlying assumption, "You go to Ibiza only in the summer but to Frankfurt for professional reasons whatever the season."
We therefore ranked destinations using their biggest peak in 2019 monthly traffic. We computed the difference between the number of passengers in the busiest month in 2019 and the mean number of passengers in 2019, divided by the mean—i.e., (max – mean) / mean. The higher this peak indicator, the likelier a destination is touristic, while the closer it is to 0, the more likely it is business intensive.
In all, we got quite a consistent distribution of 197 international destinations, excluding outliers such as Olbia (Italy) and Heraklion (Greece), which sport summer peaks amounting to eight times their mean. Figure 3 maps out the distribution, with the likely business destinations on the left side and likely tourist destinations on the right.
Most tourist destinations in the dataset are in Europe, and most Asian destinations are business routes. The US displays a more even distribution, with strikingly few likely business destinations. This matters when considering the rebound in each region. Since tourist destinations bounced back more rapidly, it could lead to overestimation of the overall European recovery and underestimation of the Asian recovery. This regional distribution arises from the fact that the airline we look at is not low-cost: Consumers likely use it to fly both to close business destinations—as opposed to close tourist destinations, which are serviced more by low-cost airlines offering deals—and to farther away tourist destinations, which are overall more costly and less serviced by low-cost airlines. Hence, our method allows us to properly account for this feature of the airline, though it introduces a bias in terms of regional interpretation.
Business lags the tourism rebound
Most tourist routes in the airline's dataset rebounded faster and business ones slower. To highlight this result, figure 4 plots the 2019 peak indicator against the rebound intensity of each destination, which is computed as the difference (in percentage points) between the March or April 2022 year-over-year passenger growth (whichever is higher) and the April or May 2020 growth (whichever is lower), both expressed in percentage relative to respective months in 2019.
Doing so reduces the number of destinations in this sample (since not all had flights scheduled in each of those months) and tends to eliminate extremely touristic destinations for which flights are scheduled only during a couple of summer months (indeed, many destinations were simply not scheduled in April 2020). This leaves us with a "conservative" set of destinations, with 2019 peak difference from the mean mostly smaller than the mean (i.e., peak indicator smaller than 1).
We regard the overall positive correlation between the two as evidence that most tourist destinations experienced a greater rebound and that business destinations lagged consistently. The pattern is even more clear in figure 5, which plots the same data by regions of interest.
To explore this result over time, we created two subsets of destinations using quintiles of the distribution to define business and tourist destinations. This method allows us to isolate the bottom (most business-intensive) quintile—i.e., the first 20 percent of the distribution, which does not vary much over the year (maximum distance to the mean in 2019 smaller than a tenth of the mean)—and the top (most touristic) quintile—i.e., the last 20 percent of the distribution, which varies quite a bit over the year (with maximum distance to the mean in 2019 up to five times the mean).
Figure 6 shows for each subset the month-to-month change in passenger traffic with respect to 2019, weighted by the number of passengers in each destination. This yields an even clearer result, as business destinations lagged quite strongly, and tourist destinations caught up more rapidly. This touristic advantage in rebound averages 20 percentage points, since business destinations had an average 55 percent loss in number of passengers with respect to 2019 levels from January 2020 up to April 2022, while most tourist destinations had an average 35 percent loss over the same period and, unlike other destinations, caught up with 2019 levels as early as summer 2021. This is a striking finding, considering that all destinations displayed very comparable negative shock from the pandemic around April 2020.
Yet, the story is a bit more nuanced when looking at the regional level (figure 7). Tourist destinations indeed recovered strongly in Europe and non-China Asia, ending up higher than and nearly at the pre-pandemic level by late 2021 and early 2022, respectively. The pattern is less clear for North American destinations, which do not fit this quintile set-up well—the line for business travel is excluded as only a single destination (Calgary) was classified as a likely business destination and is thus not relevant.
More importantly, the data do not allow us to distinguish the origin of the traveler, so we may identify as touristic some places that are not but from where people travel to Europe for touristic reasons in the same period—e.g., summer in the US: If Bostonians leave for European vacations in the same summer month and return home, Boston will be identified as touristic. Yet, the overall result is not affected since it still is tourism whatever the direction, and since the European bases of the airline that provided the dataset are way larger hubs than most destinations—so that "outflows" from those hubs most likely dominate "inflows" from most other destinations, rendering this mismatch unlikely. On the other hand, it potentially leads to underestimating business travel recovery in some regions, such as the US.
The conclusion is that European tourist destinations had a way faster rebound than any other subset. This may be regarded as evidence of European preference for proximity in the midst and aftermath of the pandemic. Business destinations, however, are lagging everywhere, which strongly confirms the suspicion of a lasting effect.
The permanent air traffic shutdown in China
China stands out radically: European air traffic there stalled roughly at lockdown levels. China in fact is an outlier because of the way it handled air traffic restrictions. For the airline that provided data, after a complete—94 percent—loss of traffic in March 2020 with respect to March 2019, European passenger traffic remained basically shut down, with an average 84 percent loss in April 2022 with respect to April 2019. This dramatic closure of Chinese international air traffic results from Beijing's very stringent zero-COVID strategy, which has imposed strict restrictions and controls at the border aimed at containing foreign contamination.
Other countries in the Asia-Pacific region also shut down international routes for longer periods than the Europe or the US. However, traffic eventually recovered. China is the only country that deliberately chose to keep its air space nearly closed to foreign flights. This international isolation is the starkest of all COVID policy choices and is likely to have the strongest hysteresis effects.
Strikingly, Chinese domestic flights rebounded faster than US domestic flights before plunging back due to recent COVID waves, with the same intensity as they did during the first wave, while international traffic never took off again (figure 8). Basically, China was trying to recover activity without recovering its international ties. But recent data show that internal activity is also dwindling due to stringent COVID policies.
The consequences of this choice are being debated in China. There is indeed early evidence of an "exodus of foreign talent" from the country, an extreme case of how the travel shock may impact overall activity through a near-collapse of face-to-face interactions. Should this isolation persist, it is likely to weigh on Chinese scientific performance, access to foreign technologies, and participation in global value chains—thereby intensifying its growing decoupling from the Western economic system.
A lasting legacy?
To conclude, the pandemic shock has left a major and lasting mark on international air transportation. Two and a half years after it occurred, it is still early to assess the legacy of the shock. Tourism is on its way to full recovery—hiccups notwithstanding, key routes are by now close to recouping the losses caused by the pandemic.
Business routes have suffered more lastingly, possibly prompting more persistent consequences. Data suggest that a 20 percent hit to business traffic is well within the range of possible outcomes. If so, individual face-to-face contact would likely be permanently substituted by videoconferencing, which could result in gains in productivity and business efficiency.
Together with growing concerns over supply chain resilience and, in recent months, the geopolitical vulnerability of the just-in-time delivery model, the traumatic experience of the pandemic crisis may, however, also prompt a reexamination of the prevailing globalization pattern. A new international division of labor is in the making. After it emerges, it will likely be hard to disentangle the causes and to determine what exactly explains what.
China finally has made use of the shock to close down and accelerate a wholesale rethink of its economic model. What could have taken many years to decide and implement was suddenly made possible. The question for the future is what it will trigger. What seems likely, at least, is that the very close integration of Chinese and global production networks will suffer. The pandemic shock may not cause the Chinese decoupling from the Western economy but will certainly accelerate it.
Appendix: Regional distribution of air passenger traffic
We divided the total sample of 197 destinations into two subsets when ranked depending on the size of their passenger peaks in 2019: Most likely business destinations (50) are in the first quartile of the distribution and most likely tourist destinations (49) are in the fourth quartile of the distribution (figure 9). The rest are in the middle of this distribution.
1. For more detail, see "Delays, shortages and strikes: Can the aviation industry get airborne by summer?," Financial Times, June 10, 2022.
2. See "Heathrow boss blames airlines for staffing shortages hitting travel," Financial Times, July 15, 2022.
4. See the survey of the literature in Olivier Blanchard and Jean Pisani-Ferry, Persistent COVID-19: Exploring potential economic implications, PIIE RealTime Economic Issues Watch blog, March 12, 2021.
5. We calculate changes in the number of passengers actually flying and not in available seats, revenue per person, or load factor of the airline, which, other than being more sensitive company information, are less indicative of overall activity levels.
6. We decided to rely neither on fare categories (first/business/economy), because they are poor indicators of the motivation for flying, nor on the surveys provided by the company, since they do not comprise the whole set of routes, among other issues.
7. Excluding one African destination (Oran in Algeria), which had an indicator greater than one.
8. We also find that the difference between tourist destinations and the rest of the sample without business destinations is bigger than the difference between business destinations and the rest of the sample without touristic ones—on the order of 12 percentage points against 7 percentage points on average during the period. This finding should be taken with a grain of salt, since there is no way to know how much virtual links (namely videoconferencing) were able to substitute face-to-face interactions, so that an unknown share of this business drop may not be associated with a loss of activity.
9. Figure 9 in the appendix shows detailed regional distribution of destinations.
10. See Huiyao Wang, "China's zero-Covid policy has damaged its stock of global talent," Financial Times, July 17, 2022.
Thomas Belaich is a student at the Paris School of Economics.
The data underlying this analysis are confidential and not disclosed. The replication code is available here.