Estimating the cost of COVID-19 related cancellations to Airbnb and their hosts in Edinburgh
The current COVID-19 crisis has serious implications for the hospitality industry in general and Short-Term Lets (STL) in particular.
Edinburgh has seen STLs grow considerably over recent years to what many see as problematic levels, but the current crisis is having a profound impact on this part of the tourism industry.
This is the second of our blogs examining the early impact of COVID-19 on the short-term let market in Edinburgh and on Airbnb specifically. The first blog focused on the changes in Airbnb bookings based on changes in available stock and the number of reviews. This blog focuses on the cost of cancellations occurring as a result of the UK’s lockdown.
On the 14th March 2020, Airbnb announced its “extenuating circumstances policy” to lessen the impact of the COVID-19 pandemic to its short-term let business. Under the policy, customers and hosts can cancel reservations made before the 14th March 2020, and with a check-in date between the 14th March and the 31st May 2020, with a full refund and without penalty. People cannot travel (or at least should not) to stay in Airbnb properties and we would expect under the new rules that most existing bookings will be cancelled.
How can we estimate Airbnb cancellations?
By examining data obtained from monthly scrapes (provided by Inside Airbnb) we can compare booking calendars month on month to estimate the number of cancellations and new bookings made before the 21st March 2020 (the latest available data). We used eight monthly web scrapes, four in December to March 2018/19 and four in December to March 2019/20, to calculate cancellations and new bookings for January, February and March in 2019 and 2020. These data provide detailed snapshots of the booking calendars and a record of each listing's availability, allowing us to examine the impact on bookings in the run-up to the usually busy Easter/Spring holidays for 2019 and 2020. We assume a cancellation has taken place when we observe the availability status of a listing being changed between months from ‘unavailable’ (which we assume means it is booked(1)), to ‘available’. We do this for all the bookings and cancellations which fall within the period of Airbnb’s policy (bookings with a check-in date between the 21st March and the 31st May 2020).
As we start our observations in January 2020 (based on a comparison with December 2019) when the Edinburgh market was still healthy and growing, we can estimate roughly when cancellations were made(2). We also analyse the opposite phenomenon where a listing’s status is switched from ‘available’ to ‘unavailable’ indicating new bookings(3). Combining these two measures, we can estimate turnover to the 31st May. The pattern for this current year becomes clearer when we compare data from the same period in 2019.
Figure 1. 2019 and 2020 cancellations and bookings (made for bookings with a check-in date of the 21st March to the 31st May) observed in January, February, and March.
What do the estimated changes in bookings and cancellations tell us about Airbnb’s revenue?
Comparing cancellations (of bookings up to the end of May) made in January 2020 to those made in January 2019, we see a 10% decrease from last year. However, in February 2020 there is an increase of 48% in the cancellations (of bookings up to the end of May) compared to the previous year. After the announcement of Airbnb’s cancellation policy, there were more than 55,000 cancelled nights, a 108% increase from the previous year.
We calculate the lost revenue from these nights by subtracting the nights cancelled from the nights booked and multiplying by the average price per stay(4) for both periods in 2019 and 2020 (21st March to 31st May). We then subtracted this year’s net income from last year’s net income.
Based on the Inside Airbnb data scraped on the 21st March 2020, the difference between cancellations and bookings is 11,052 nights which translates to a £1.5 million loss of revenue for the period in this year. In contrast, the figure for the same period last year (2018/19) was an increase of £8.4 million. This suggests that even as early as the 21st of March 2020, there had already been a drop in revenue of nearly £10 million.
Conclusion
While our estimation is a rough figure of the loss in Airbnb revenue in the Edinburgh market, this is likely a large underestimation and the losses are probably much greater. We will have a more complete picture following an analysis of Inside Airbnb’s April data for Edinburgh, but the early signs suggest that Airbnb and their hosts will have made significant losses over this period.
Notes
- We are aware that ‘unavailable’ noted in the calendar could mean either ‘booked’ or ‘off the market’ for booking. Given the travel restrictions under COVID-19, the chances of seeing a change from ‘not available’ to bookable are assumed to be very slim.
- Here we analyse those listings that can be observed throughout the four months. They are around 93%-95% of total listings each month.
- Under COVID-19 lockdown new bookings are still possible, as accommodation can be designated to key workers for self-isolation purposes. The ‘available’ to ‘unavailable’ can also indicate the listings are temporally off the market due to the outbreak.
- Inside Airbnb suggest £133/night for Edinburgh.
Project Team
Research team (and the authors of this blog): Dr Yang Wang; Dr Mark Livingston; Dr David McArthur; Professor Nick Bailey
Data science team: Nikos Ves; Dr Andrew McHugh
