Conference Agenda

We study an auction design problem, where a seller aims to sell an item to multiple bidders. Bidders’ willingness to pay (values) are independent random variables, and the seller only knows an upper bound on these values, lacking distribution knowledge. The objective is to devise a deterministic mechanism effective across various plausible distributions. We propose a second price auction with a reserve price set to half of the upper bound. Despite no deterministic mechanism can achieve a positive fraction of the maximum achievable expected revenue across all distributions, we define a distribution class G, which is extensive and contains several distributions of practical importance, and we demonstrate that our mechanism achieves at least 1/4 (1/2 under i.i.d. values) of the maximum expected revenue for distributions within G. We conduct numerical experiments to evaluate our mechanism’s performance beyond G, under randomly generated distributions, demonstrating its superior performance in approximately 95% of the generated instances compared to benchmark mechanisms from the literature. We illustrate numerically that our mechanism exhibits greater robustness against different correlations than the benchmarks when considering two non-independent bidders. We consider the scenario where the estimated upper bound is subject to errors and show that appropriately lowering the reserve price based on estimation confidence ensures a constant positive fraction of the maximum expected revenue across G. Traditional auction design strategies often propose randomized mechanisms that lack interpretability, implementability and transparency, causing trust issues among bidders. Instead our mechanism is simple, requires minimal information, and is effective in practical scenarios.

Two-sided marketplaces are online platforms that facilitate exchanges between suppliers and buyers as they reduce search costs, for example, in the context of vacation rentals, freelancing, or salon bookings. Marketplace operators do not offer this service for free but charge marketplace fees, which are the basis for revenue generation and the marketplace's long-term economic success. There are two main types of marketplace fees, which can also be adapted jointly: commissions, where the operator keeps a percentage of the total transaction amount, and subscription fees, flat fees paid by all users for accessing the marketplace.

Existing research approaches analytically evaluate optimal marketplace fees, focusing on deterministic settings. These approaches discuss how to optimally fix commissions and subscription fees a priori to maximize marketplace operator revenue or overall marketplace welfare. In practice, however, marketplace operators face multiple uncertainties regarding the market setup that complicate the fee-setting procedure, such as suppliers' and buyers' stochastic arrivals, pricing expectations, and maximum waiting times. To address this gap, we propose an adaptive design of marketplace fees based on deep reinforcement learning (DRL), enhancing the operator's ability to navigate inherent market uncertainties. As the agent, the platform learns forward-looking fee-setting policies for each transaction and each time period in an uncertain market environment under revenue maximization for all marketplace users. Exploring the intersection of DRL and the platform economy, this approach allows for non-myopic fee-setting in marketplaces by considering the impact of these fees on the overall welfare of all marketplace participants and the operator's revenue.

Innovative revenue models, e.g. as hybrid products through jointly selling a tangible asset with services to be provided later, complicate operational performance management considerably, since capacity requirements in subsequent periods are predetermined by guaranteed follow-up services. The type of the corresponding service level agreement and the conditions of the service call determine how services have to be provided, at least partly beyond the company's own control. As a consequence, the customer's stochastic call-off behaviour must be modelled for contractual design, supply planning and service management. Although there is already extensive literature that deals, e.g., with bundle pricing, congestion pricing or dynamic pricing of such hybrid products, there is no consideration of the risk of later capacity bottlenecks from the current scheduling of uncertain later service calls from new business. This effect must be taken into account when structuring the contractual conditions, esp. pricing, in order to enable optimal production and sales planning under risk. The paper develops a model for this purpose, which is investigated by simulation using a case study. In this evaluation, the sample average approximation and the scenario approach will be contrasted.

A flexible resource utilization can be beneficial for a multi-product vendor in times of increased uncertainty. One possibility is to offer flexible bundles, i.e. bundles that contain flexible products. While price bundling makes it possible to achieve the capacity utilization with the highest expected profit at the time of purchase, flexible products provide opportunities to manage demand and resource uncertainties in the time between purchase and delivery. However, both have a significant impact on customers' purchasing behavior. An empirical study was conducted to collect data on customers’ individual WTP range for flexible bundles. It turned out that customers perceive greater performance uncertainty the more flexible the bundles are. As a result, the individual WTP range becomes narrower and has a lower mean value. Since this effect is moderated by the customer's purchasing power and risk attitude, different customer segments can be formed. However, the segment-related aggregation of individual WTP ranges results in asymmetric distributions, which are difficult to handle analytically and numerically. Against this background, we propose a way of how to incorporate asymmetric WTP distributions into a bundle pricing model that can be solved by standard optimization software. A numerical study provides insights into profit, sales quantities and prices depending on the characteristics of both customer segments and capacity. Furthermore, we determine the impact of incorrect assumptions about segment-related WTP distributions on the advantageousness of pricing.