Universiteit Leiden

nl en

Dissertation

Judgmental Adjustments in Revenue Management

On 12 November 2019, Larissa Koupriuchina defended her thesis 'Judgmental Adjustments in Revenue Management'. The doctoral research was supervised by Prof. J.I. Van der Rest.

Author
Larissa Koupriouchina
Date
12 November 2019
Links
Leiden Repository

Nowadays, computerized systems with complex algorithms determine the prices of hotel rooms. However, the occupancy rate forecasts based on which the systems offer room rates do not always correspond to the actual hotel occupancy. Sometimes the computer is wrong. If hoteliers think they can foresee this, they will therefore adjust their computer system forecasts manually. Whether these adjustments actually have the desired effect has not been investigated before. This dissertation explores the influence of manual adjustments on the accuracy improvement of system-generated hotel occupancy rate forecasts.

The dissertation first looked at whether it matters which standard is used to assess the accuracy of occupancy rate forecasts. It was then investigated whether the moment of manual adjustments influences the forecast accuracy. The characteristics of manual adjustments, such as adjustments up / down, large / small, often / few, and their effects on hotel occupancy rate forecast accuracy were also zoomed in. The study used (longitudinal) multilevel regression analysis (with repeated measurements) and data from 1,752 hotels worldwide. The dissertation shows that (1) choosing the accuracy measure is a very complex issue, that (2) the accuracy of occupancy rate forecasts improves considerably as the forecast horizon becomes shorter, and (3) the effect of manual adjustments differs and differs per hotel .

The PhD research is scientifically important because manual adjustments can result in an accumulation of suboptimal computer decisions. The dissertation shows how timing and other adjustment characteristics influence the accuracy of computer forecasts, and shows how and when adjustments improve computer forecasts.

This website uses cookies.  More information.