Researchers studied the impact of the operations of Airbnb in San Francisco - where the room-sharing company has a high penetration rate - on the city's hotels.

From the researchers:

"Our results, based on a mixed-model analysis using a saturated, unstructured covariance matrix, show that overall hotel RevPAR is unrelated to total Airbnb supply. Interestingly however, in certain segments, RevPAR is affected by the average price of Airbnb listings. More importantly, hotel sales performance is impacted by Airbnb customer reviews, which points to nuanced and contextual complementary and substitution effects. "

In plain English: overall (an important qualifier, as not all hotels were affected equally) the number of Airbnb units available for rent didn't have any measurable impact on hotels' revenue per available room (RevPAR). But in some hotel segments - specifically lower-scale - rates charged for Airbnb units tended to depress hotel RevPAR.

Favorable consumer ratings of Airbnb properties also tended to depress hotel RevPAR, suggesting that substitution of Airbnb units for hotel rooms is in fact occurring in San Francisco, at least for some hotel segments.

So, an actionable finding for hoteliers: if nearby Airbnb properties have daily rates similar to or lower than yours, and better reviews, you need to be really competitive on both rates and guest satisfaction.