STR: Airbnb’s impact minor on Manhattan hotels
A new analysis from STR of Airbnb-provided data shows that the alternative-accommodations site does not have a measurable effect on hotel demand in Manhattan.
HENDERSONVILLE, Tennessee—Airbnb unit sales do not materially affect hotel performance in Manhattan, according to an STR analysis that used data provided directly by Airbnb.
The data set—which aggregated Airbnb daily data (supply, demand, revenue) from 1 December 2013 to 30 November 2015—is the largest Airbnb has provided to a third-party hotel research company. STR was not remunerated in any way for its analysis; its participation was not contingent upon reporting predetermined results.
Here are the eight most important things from the analysis that you need to know.
1. Airbnb does not cannibalize hotel demand
There is no evidence that a unit occupied by an Airbnb guest is a room taken from a hotel.
Airbnb-occupied nights in 2015 (1,496,260) were 5.5% of overall lodging demand (25,788,543 hotel roomnights sold). This was up from 4.4% in 2014.
With only a few daily exceptions, Airbnb occupancy ran well below hotel occupancy.
The Manhattan hotel market reported occupancy of 87% during 2015, 52 days of which finished with occupancy of 95% or higher. This indicates the existence of large amounts of unaccommodated demand, some of which could be accommodated by Airbnb units or by new hotel supply as it continues to open in the market.
Some guests likely searched both hotel and Airbnb options and chose the latter, but there are no apparent statistical trends to suggest the degree to which this is occurring.
2. Hotel rooms outnumber Airbnb units 10 to 1
Average daily supply of Airbnb units in the market was 8,806 from 1 December 2013 to 30 November 2015. Average daily hotel room supply, by contrast, was 87,762.
During 2015, Airbnb units were available for booking on the equivalent of 3.2 million nights, representing 9.8% of the 33 million total hotel roomnights in 2015. That’s up 8.5% from 2014.
Note that Airbnb’s “effective” supply is lower than what is recorded. Many hosts list their units as available but only selectively accept bookings.
3. Airbnb does not undermine hotel pricing power
During the strong demand nights for Airbnb units, there was no pattern demonstrating that hotel average daily rate for that same period was affected adversely.
Hoteliers achieved strong rate premiums during compression nights, when occupancy was 95% or higher, compared with non-compression nights (e.g. occupancy less than 95%). By contrast, Airbnb achieved minimal rate gains (less than 4%) for the same time-period comparisons.
4. Airbnb units, on average, cost $100 less than hotel rooms
Hotel ADRs in Manhattan run at significant premiums over average Airbnb rates, particularly during times of peak demand.
Furthermore, there appears to be virtually no revenue-management strategies employed by Airbnb hosts. On New Year's Eve in 2014, for example, Airbnb units sold at an average rate 14% higher than all the other days in December. For hotels, this premium was 48%.
The correlation coefficient (a statistical measure of the relationship between two variables) between hotel ADR and Airbnb ADR was low at 0.6835.
5. Airbnb units account for 3.5% of overall lodging revenue
Airbnb units generated 3.5% of overall lodging revenue (hotels and Airbnb units) in 2015, up from 2.8% in 2014.
6. Airbnb units skew lower in the chain scale
Nearly two-thirds (60.7%) of Airbnb units in Manhattan fall in the midscale and economy classes, based on STR methodology. Conversely, 13% of hotel supply is concentrated in the midscale and economy classes.
7. Airbnb guests stay longer than hotel guests do
The majority of Airbnb guests stay more than seven days, which is similar to the stay pattern at extended-stay hotels. (Only 4% of hotel rooms in Manhattan are classified as extended stay, according to STR.)
More than four out of 10 Airbnb guests (41.4%) stayed between seven and 29 nights and 16.7% stayed for 30-plus nights.
8. Airbnb occupancy is more volatile
Airbnb occupancy fluctuated more dramatically over the observed period. While some of this fluctuation was the result of a constantly changing supply base, similar patterns were demonstrated when calculating off an average supply base.