The FlightCost forecast goes beyond extrapolating air passenger traffic. We examined five groups of data separately: passenger counts, population, real GDP, household consumption and departures. We then brought them together in a regional macro model. The sections below show the historical series, how we use them and the general formula.
Air passenger traffic
What we examined. Annual passenger counts from ICAO and country-level WDI series for airlines. The historical data show steady growth through 2019, followed by a crisis-induced drop that breaks the trend.
Global air passenger traffic: growth through 2019 and the break in trend
billion passengersFrom 2010 to 2019, passenger counts rose from 2.71 to 4.50 billion (an average of +5.8% per year). At the same rate, the count would have reached 5.97 billion in 2024, compared with 4.53 billion actually recorded. The orange line is an illustration of that extrapolation, rather than a FlightCost forecast.
ICAO’s global total for 2025 — 4.95 billion passengers — sets the reference level. We allocate it among countries with positive WDI shares for 2023 and roll it forward to the 2026 baseline using the model.
WDI coverage for the allocation: 149 of 249 countries and territories. Passenger counts are assigned to the country where the carrier is registered. They do not measure demand by departure airport or between individual cities.
Population trends
What we examined. Historical population and the change in passenger traffic over the same years. Population tends to grow steadily. Air traffic can shift abruptly without a comparable change in the number of residents.
Population grows more steadily than air passenger traffic
index, 2010 = 100Between 2010 and 2019, world population grew by 11.1%, while the number of air passengers rose by 66.2%. Population sets the potential scale of demand and explains little about sharp fluctuations.
Population converts forecast trips per capita to total passenger counts. For 2026–2036, we use the UN’s baseline population scenario, WPP 2024 Medium.
Population figures alone tell us little about travel frequency, transfer hubs or ticket affordability. The model therefore uses population as a separate multiplier in the passenger forecast.
Real GDP
What we examined. Growth in real economic output and passenger traffic. The series are related, but changes are not proportional. The 2020 crisis shows particularly clearly how an economic contraction can differ from the actual fall in demand for flights.
Global GDP and air passenger growth
annual change, %Both indicators generally rose before the pandemic, at different rates. In 2020, global GDP fell by 2.9%, while air passenger counts fell by 60%. A single constant GDP elasticity cannot capture shocks of this kind.
The macro model uses changes in real GDP per capita. The external GDP path comes from the IMF’s October 2026 World Economic Outlook. For 2032–2036 the model carries forward each country’s average growth rate in 2027–2031.
A 1% rise in GDP can correspond to different increases in passenger traffic. We test this relationship against historical observations and model restrictions due to war and pandemics separately.
Real consumption
What we examined. Real household spending per person complements the measure of total economic output. The two groups of countries show that purchasing power can grow at different rates even after the same global crisis has ended.
Real consumption per capita: diverging trends
index, 2010 = 100From 2010 to 2024, real consumption per capita rose by 22.0% in high-income countries and by 81.8% in upper-middle-income countries. The chart shows an economic indicator that does not measure passenger traffic.
Growth in real consumption per capita serves as a proxy for purchasing power. The forecast carries forward a growth rate estimated from historical data.
Consumption differs from a household’s disposable income and does not account for the price of any particular ticket. We study the link between personal income and flying separately. This article does not use that relationship to estimate fare elasticity.
Number of departures performed
What we examined. Departures actually flown by airlines and the number of passengers carried. The two series move differently. Departures capture aspects of air service availability that GDP and population alone cannot describe.
Departures and passengers: growth and recovery
index, 2010 = 100By 2019, departures were up 30.6% from 2010, while passenger counts were up 66.2%. The two measures capture different aspects of air travel.
Historical data on departures performed per capita help describe how the airline network developed. The baseline scenario extrapolates historical departure growth and accounts for projected population.
Flight frequency reflects both demand and airline decisions. More departures do not necessarily mean more passengers per flight. We do not interpret the modeled relationship as evidence of causation.
Crises and restrictions
What the data show. In 2020, the four main indicators differed in how much they changed and even in which direction.
What changed in 2020
Change from 2019, %The model has separate shocks for a pandemic, a financial crisis and a military conflict. Each shock is specified directly using the share of the market affected, independently of GDP changes. After the last year of decline due to a pandemic or financial crisis, the remaining loss halves every two years. This is a scenario assumption, rather than a universal rule.
General model formula
We calculate the change in passengers per capita for each country and account for population growth or decline. We sum country forecasts within each of the 22 subregions and apply the chosen scenario shock to the subregional total.
ac,y = α + βgxg + βhxh + βfxf
Here, c is the country, r the subregion, y the calculation year and t the target year. The term a expresses the annual change in air passengers per capita on a logarithmic scale.
The model uses a historical baseline and external scenarios for the future. Adjusting annual GDP growth by ±1 percentage point creates alternative scenarios. It does not produce a forecast confidence interval. The model does not estimate the opening of specific direct routes or the number of aircraft.
What the model covers and what we study separately
| Factor | Status | Notes |
|---|---|---|
| Historical passenger traffic | Baseline | ICAO and WDI, with counts assigned by country of carrier registration |
| Population | In the model | UN WPP Medium, country-level population scenarios |
| Real GDP | In the model | WDI historical data and IMF WEO projections through 2031 |
| Real consumption | In the model | Continuation of historical growth in real consumption per capita |
| Historical departures | In the model | Continuation of historical departure growth, accounting for projected population |
| Fares, visas, tourism, diaspora communities, investment | Separate studies | These factors have no universal coefficients in the global model |
| New direct routes, aircraft fleets | Later stages | Require separate observations, constraints and validation |
All historical chart values, 2010–2024
| Year | Passengers, bn | Population, bn | GDP, % YoY | High-income consumption, 2015 USD | Upper-middle-income consumption, 2015 USD | Departures, mn |
|---|---|---|---|---|---|---|
| 2010 | 2.71 | 7.001 | 4.52% | 20,441 | 2,687 | 29.70 |
| 2011 | 2.87 | 7.087 | 3.31% | 20,642 | 2,902 | 30.70 |
| 2012 | 3.01 | 7.176 | 2.75% | 20,774 | 3,048 | 30.90 |
| 2013 | 3.14 | 7.265 | 2.89% | 20,976 | 3,198 | 31.30 |
| 2014 | 3.32 | 7.354 | 3.17% | 21,302 | 3,327 | 31.90 |
| 2015 | 3.56 | 7.441 | 3.13% | 21,677 | 3,453 | 32.90 |
| 2016 | 3.80 | 7.528 | 2.79% | 22,004 | 3,594 | 34.89 |
| 2017 | 4.07 | 7.614 | 3.45% | 22,454 | 3,776 | 36.50 |
| 2018 | 4.33 | 7.696 | 3.29% | 22,894 | 3,942 | 38.09 |
| 2019 | 4.50 | 7.777 | 2.68% | 23,212 | 4,066 | 38.79 |
| 2020 | 1.80 | 7.854 | -2.89% | 21,991 | 3,892 | 22.47 |
| 2021 | 2.30 | 7.920 | 6.49% | 23,469 | 4,248 | 24.92 |
| 2022 | 3.20 | 7.989 | 3.44% | 24,286 | 4,397 | 31.21 |
| 2023 | 4.17 | 8.063 | 2.86% | 24,596 | 4,690 | 35.25 |
| 2024 | 4.53 | 8.141 | 2.90% | 24,947 | 4,884 | 37.09 |
The series shown here trace the historical factors over comparable years. The FlightCost model is trained on more detailed country-level data and checks for missing values.
Data and reproducibility
The historical charts use published observations from 2010–2024. Passenger counts for 2010–2018 come from ICAO Air Transport Statistics 2019, and those for 2019–2024 from ICAO Safety Report 2025. The departures series combines the ICAO editions listed under its chart. WDI data may also be revised. Sources, units and methodology are documented separately from the FlightCost calculations.