多房型飯店預訂政策：理論模型與個案研究

Abstract

大部分飯店在面對旅行社訂單時，只要尚有剩餘空房間就會選擇接受，但旅行社訂單往往因為需求房間數多而使價格偏低，拒絕旅行社訂單而將房間賣給未來可能出 現的散客或公司客可能可以賺得更多收益。此外，實務上當低房型賣完後，飯店可能 會考慮將低房型需求做免費升級，以高房型房間滿足低房型房間需求。在本研究中，我們將未來預期收入與房型升級納入考量，建構動態規劃模型幫飯店計算接受旅行社訂單的價格底線，最大化飯店預期收益。研究結果指出，採用本模型後，預期收入會隨著剩餘房間數增加或距離入住日期越久而上升，但價格底線就較沒有一定的規則，會根據旅行社訂單和剩餘房間數而上升或下降。我們進行數值實驗，發現當旅行社議價能力低、市場狀況較差或旅行社訂房機率較小的時候，考慮房型升級對預期營收有正面影響，當旅行社議價能力高或訂房間數多的時候，考量未來收益的效果較好。最後，我們拿台灣某飯店業者的歷史資料作為探討，利用歷史訂單資料做需求估計，測試動態規劃與房型升級能增加多少飯店營收，發現採用我們所提出的考慮房型升級之動態規劃模型，能提升總體營收的的 9.5%。

Facing a travel agent’s order, most of the hotels will accept the order if they have enough rooms. However, because it orders a large number of rooms, a travel agent usually asks for a lower price. Rejecting a travel agent’s order can reserve those rooms to individuals or business customers for a higher price, which may allow a hotel to earn more money. Furthermore, if a hotel does not have enough low-type rooms, it may consider to freely upgrade the request to high-level rooms to accept the order in practice. In our study, we take future demands and room upgrade into consideration, build a dynamic programming model to help hotels find the bottom line price of accepting a travel agent’s order to maximize expected revenue. We show that if a hotel adopts our model, its expected profit will increase when there are more unsold rooms or more days before check in. However, the bottom line cannot be predicted easily and may be affected by the numbers of unsold rooms and days before check in in various ways. We have done some numerical experiments and found that when travel agents’ bargaining power becomes lower, market condition becomes worse, and the probability for travel agents’ orders to appear becomes lower, considering room upgrade has a positive impact on the expected revenue. When travel agents’ bargaining power becomes higher or the average number of rooms travel agents order becomes higher, considering future demand increases the expected revenue. Finally, we use a Taiwanese hotel’s historical sales data to examine the amount of expected revenue that our model may increase for the hotel. We find that using our proposed model can increase the hotel’s revenue by 9.5%.