Markets for Ideas and the Verification Bottleneck

This paper studies markets for ideas where editors screen proposals that innovators submit and reviewers validate. Market equilibrium can be distorted by positive externalities from ideas and reviewer altruism that suppresses verification prices, but advances in drafting technology nevertheless increase welfare under optimal anonymous editorial rules when altruism is sufficiently small. In contrast, under pure peer review—when reviewers are motivated by their impact alone—technology improvements generally reduce welfare under anonymous rules: easier drafting lengthens submissions from lower-efficiency innovators which degrades reviewer impact. To ration supply, editorial standards must tighten, which exacerbates the problem under anonymous rules. The analysis identifies verification capacity—rather than idea supply—as a fundamental constraint on the rate of scientific progress.

Explaining Anomalies (joint with Leon Kowalke and Marcel Prokopczuk)

More than three-fifths of anomaly premia can be explained by a small set of proxy variables that capture key economic concepts. Using a decomposition approach that incorporates both the cross-sectional and time-series dimensions of potential explanations, we analyze the return premia of a broad set of prominent anomalies in the cross-section of stock returns. The main drivers of anomalies are mispricing, distress, trading costs, and limits to arbitrage. Accruals, debt issuance, investment, low-risk, and value anomalies are explained relatively well, whereas profit growth, profitability, seasonality, and short-term reversal anomalies are accounted for only to a limited extent.

Integrating Mixed Integer Programming and Neural Networks: Optimization of Stochastic Manufacturing Systems

Simulation is widely used for evaluating stochastic manufacturing systems, due to its adaptable modeling capabilities and the absence of exact or even approximate analytical solutions. Solving optimization problems with integrated simulation models can be time-consuming, due to the need for many replications and the combinatorial complexity of the problems. To increase the speed of performance evaluation, the existing literature suggests replacing simulation with Artificial Neural Networks pretrained with simulation results. We train a Convolutional Neural Network (CNN) with flowline feature data to predict flowline throughputs for different system configurations. We then solve the buffer allocation problem by linearizing the trained CNN, thereby transforming the non-linear problem into a mixed-integer linear problem (MILP), which is tractable by standard solvers. We show the impact of different sampling methods for generating training data and regularization methods. Furthermore, we demonstrate the flexibility of our approach by applying it to different production networks and optimization problems.

Prof. Dr. Oscar Stolper / Philipps University of Marburg