Revenue management is the science and art of offering the right product, to the right customer, at the right time, at the right price, and through the right channel to maximize profits. Revenue management has evolved into a successful and necessary practice in traditional industries like airline, hotel, car rental, cruise line, etc. However, in the apartment industry, revenue management practice is still in its infancy -- only about 9% of the apartment units in the United States use some form of revenue management. Apartment industry business is simple but has the unique aspects of renewals and expiration management.;Apartment residents often request and have a strict preference to a lease term (number of months). Also, operators often accept leases with little attention to their expiration times, just to get the apartment units filled and start generating rental revenues. Several operators also set leases to automatically expire either at the end or middle of an expiring month. Both the residents' preference and operators' shortsighted approach together usually result in leases expiring at a future time when the demand from reoccupying residents is insufficient. As a result, when the leases expire, the apartment units remain vacant for a long time before they are reoccupied, thus incurring vacancy loss, additional turn and marketing costs, revenue dilution, and even displacing higher-rent paying residents in the long run. Vacancy loss arises from a mismatch between demand seasonality and current lease expirations, risky renewal behavior, not factoring the price-demand relationship, and a high number of month-to-month leases. Optimally compensating for these risks and costs while pricing an apartment unit for the requested lease term is revenue-critical. In the apartment industry, expiration management is a revenue management function that optimally prices every available apartment unit based on the lease term and move-in date requested by a resident and the expiration time in the expiring month desired by an operator. Poor expiration management can lead to up to 2% loss in revenues, usually first in the form of lower occupancy.;In this work, we introduce the concept of simultaneous renewal pricing and formalize the concept of expiration management in the apartment industry, both of which are understudied problems. We develop three efficient deterministic optimization models for simultaneous renewal pricing and expiration management. Our first model, the forecasted renewal demand (FRD) model -- a nonlinear pure integer program -- treats renewal demand as an independent stream of demand and formalizes the application of traditional models for simultaneous renewal pricing. Additionally, it introduces the concept of modeling expiration management as constraints with integer variables. The FRD model is a computationally hard problem with a non-linear objective function and linear constraints. Based on the working knowledge of the apartment industry, we add pricing constraints that significantly reduce the search space. This reduces the computational time to a fraction of a second, while retaining business-optimal solutions. Our second model, the variable renewal demand (VRD) model -- a non-linear mixed integer program -- extends the FRD model by treating renewal demand as a function of new resident demand and lease expirations. Our third model, the dynamic demand (DD) model -- a non-linear mixed integer program -- extends the VRD model to account for the dynamic nature of demand over the booking period. All models optimize rents by dynamically looking into the long-term future with network effects while allowing us to embed corporate strategies. The models compensate for the risks and costs associated with demand seasonality, current lease expirations, renewal behavior, price-demand relationship, month-to-month leases, turn-costs, and vacancy loss. We present the data requirements and a characterization of the models. Subsequently, we develop a heuristic algorithm that solves randomized instances of efficient models at an aggregated data level several times to model data uncertainty and produce a converged distribution of optimal prices at the aggregated level. The heuristic algorithm then decomposes the distribution of optimal prices at the aggregated level back into optimal rents by lease term, unit type, move-in week requested by a resident, and move-out week desired by an operator. We introduce statistical heuristics to estimate the best-fit price-demand relationship functions for the three models and the joint probability mass functions of their parameters. We empirically establish the convergence criteria for our models. Using real transactional data from the apartment industry, we conduct a randomized case study of our models and the heuristic algorithm and present the results. (Abstract shortened by UMI.)
展开▼
