Analyzing the impact of COVID-19 on Barcelona’s Airbnb market
As a citizen of Barcelona, I am quite interested in knowing how the pandemic has affected the touristic sector. So, I decided to do a bit of research about the topic! And in order to do that, I will use data from the Airbnb platform.
Airbnb is a digital platform that offers a digital marketplace for lodging and recreational activities. Is a global company and operates in most of the major cities in the world, including, of course, Barcelona. Not official data is provided about the activity of Airbnb’s hosts, but there are a number of unofficial sources. The data I will use in this article is provided by the webpage Inside Airbnb and is obtained through web scraping (so it is publicly available data). The main period analyzed will be April 12 of 2021 but I will also use data of May 2020 and December 2019.
Throughout this analysis I will try to answer several questions: First, how the activity has been affected by the pandemic, the evolution of prices during the last year, and lastly, which kind of listings have been more impacted by the pandemic. Therefore, I will look into the two most important characteristics of a market: quantity and price.
Impact on the activity of the platform
First of all, the easiest way to assess the strenght of the market is to look at the number of listings. We can see that there were 20428 listings on December 2019, 20858 listings on May 2020, and 18226 on April 2021. The number of listings fell significantly in this last period. Secondly, we can look at the number of reviews, since they can be considered a good proxy of the booking activity.
There are three things we can infer from the plot below:
- The number of reviews increased very rapidly during the last years, especially in 2019, but the pandemic has changed this evolution dramatically. Just after March of 2021 (red line) the number of reviews dropped to zero –as you can see with the blue line–, as the restrictions made impossible traveling.
- For all the period analysed (2012–2021 approximately), the snapshot of searchable listings from April 2021 shows the lowest number of reviews. That is probably because a lot of listings have been eliminated this last year, as it has been already mentioned.
- Though it is not related to the pandemic, in the chart we can see that there is a lot of seasonality in the number of reviews. Each year during the winter months the activity drops significantly.
Impact on the listing prices
It is straightforward to see the change in listing prices. In general, prices listed on April 2021 are lower than in the other two periods. Furthermore, as we can see in the picture below, this decline is caused mainly because of the drop of upper-middle price listings (in the range from 70 to 100 EUR) and the increase of cheap lodging. In short, the price distribution analysed on April 2021 shifted to the left with respect to the previous ones. Notwithstanding this fact, high prices have remained steady, as you can see by looking at the percentiles 90, 95 and 99 of the table.
In any case, even considering the significant drop in prices of the last year, the median price listed on April 2021 is still a considerable amount, and it can be as high as 1,650 EUR per month.
Going deeper, the price listed for hotel rooms has been clearly the most affected. While the median price was 160 EUR in December 2019, on April 2021 it was as little as 103 EUR.
However, it is worth mentioning the fact that until now we have done partial analyses, we should consider the composition effect and other variables. Here it is when it comes into play the last section.
What kind of listings have been more likely to survive during the pandemic
Merging the dataset from April 2021 with the dataset from May 2020 we can see which listings remained during both periods and which listing disappeared from the platform (so they did not survive the pandemic).
Here I show the results of a model in which the dependent variable is 1 if the listing of May 2020 remained in April 2021 and 0 if it did not. The explanatory variables are the price, historic number of reviews, room type, number of beedroms of the listing and its availability in the next 90 days.
As I do not wish to make this post very much longer I will summarise briefly the results. First of all, the number of bedrooms and the price of a listing do not have such a big impact on the likelihood of surviving the pandemic. Secondly, when the listing offers a private room the probability of surviving the pandemic is lower, and the same happens when the availability increases (there is not much demand expected). The opposite happens when the number of reviews of a listing is high and when the room is from a hotel.
Now, after having seen how hard the impact has been, the real question is: How fast is the recovery going to be? It is really difficult to say, though most experts agree that it will depend a lot on the evolution of the vaccination process. What do you think?
