Michael Krapp’s research while affiliated with University of Augsburg and other places

We study the influence of stricter rules for determining performance measures for compensation contracts on managers' choice between real and accounting earnings management. Constraints, like accounting regulation or corporate governance, limit managers' influence on performance measures. We find that tighter constraints intensify real earnings manipulation, because they reduce incentives for managers to supply effort on investment activities. In turn, discretion allows managers to anticipate future benefits of investment and reduces real earnings management. The results hold when contracts include forward‐looking information and suggest that constraints on managers' influence on performance measures drive the choice between accounting and real earnings management.

Many real-world problems are inherently stochastic, complicating or even precluding the use of analytical methods. These problems are often characterized by high dimensionality, large solution spaces, and numerous local optima, which make finding optimal solutions challenging. Therefore, simulation optimization is frequently employed. This paper specifically focuses on the discrete case, also known as discrete optimization via simulation. Despite their adaptions for stochastic problems, previous evolutionary algorithms face a major limitation in these problems. They discard all information about solutions that are not involved in the most recent population. However, this is ineffective, as each simulation observation gathered over the course of iterations provides valuable information that should guide the selection of subsequent solutions. Inspired by the domain of reinforcement learning, we propose a novel memory concept for evolutionary algorithms that ensures global convergence and significantly improves their finite time performance. Unlike previous evolutionary algorithms, our approach permanently preserves simulation observations to progressively improve the accuracy of sample means when revisiting solutions in later iterations. Moreover, the selection of new solutions is based on the entire memory rather than just the last population. The numerical experiments demonstrate that this novel approach, which combines a genetic algorithm with such memory, consistently outperforms popular convergent state-of-the-art benchmark algorithms in a large variety of established test problems while requiring considerably less computational effort.This marks the so-called genetic multi-armed bandit as one of the currently most powerful algorithms for solving stochastic problems.

This paper studies the influence of stricter rules for determining performance measures used in compensation contracts on managers' choice between real and accounting earnings management. We examine a two-period agency model in which the manager's efforts during the first period have short-and long-term consequences. Constraints, such as accounting regulation or corporate governance, limit the manager's discretion in affecting performance measures. We find that tighter constraints intensify real earnings manipulation, because they increase the manager's personal costs of bringing forward investment benefits, which reduces incentives for the manager to supply effort on investment activities. In turn, discretion in affecting performance measures allows managers to anticipate future benefits of investment decisions in contemporaneous performance measures and reduces real earnings management. The results also hold when contracts include forward-looking performance measures and suggest that constraints on the man-ager's ability to affect performance measures drive the choice between accounting and real earnings management.

Zusammenfassung Langfrist-Prognosen sind typischerweise problematischer als Kurzfrist-Prognosen. Im Kapitalmarktkontext ist die Sachlage jedoch umgekehrt, da verlässliche Kurzfrist-Prognosen durch Arbitrageure sofort zunichte gemacht würden. Ex ante ist der Performance-Index am Prognosehorizont eine extrem rechtsschief verteilte Zufallsvariable. Prognosen, die auf dessen Modalwert abzielen, sind daher viel zu pessimistisch. Prognosen, die auf den Erwartungswert abzielen, sind dagegen zu optimistisch. Von den drei prominenten Lagemaßen ist nur der Median in der Lage, als Basis für eine verlässliche Prognose zu dienen. Es werden einige Praktiker-Verfahren untereinander und mit einem neuen Prognoseverfahren verglichen, welches auf der erwartungstreuen Schätzung des Medians beruht. Zur Illustration der Verfahren und der resultierenden Prognosen werden Daten des DAX bis 2022 verwendet. Es zeigt sich unter anderem, dass der erwartungstreue Median-Schätzer bessere Prognosen als das beste ‚Praktiker-Verfahren‘ liefert.

We study the relative efficiency of centralized vs. decentralized organizational forms given optimized managerial performance evaluation within an incomplete contracting framework with risk-averse agents under moral hazard. Decentralization and performance evaluation are complementary control choices and the efficiency of an organizational form depends on the design of performance evaluation. Divisions can make relationship-specific investments that improve firm performance, but also increase compensation risk. We find that pure divisional performance evaluation is optimal under centralization, whereas under decentralization, optimal compensation contracts include a combination of divisional and firm-wide performance evaluation. When comparing both organizational forms, we find that the optimal form depends on managers' degree of risk-aversion and the uncertainty of the business environment. Contrary to previous literature, we find that centralization dominates in many situations, particularly at high degrees of risk-aversion and high uncertainty.

This paper proposes a new algorithm, referred to as GMAB, that combines concepts from the reinforcement learning domain of multi-armed bandits and random search strategies from the domain of genetic algorithms to solve discrete stochastic optimization problems via simulation. In particular, the focus is on noisy large-scale problems, which often involve a multitude of dimensions as well as multiple local optima. Our aim is to combine the property of multi-armed bandits to cope with volatile simulation observations with the ability of genetic algorithms to handle high-dimensional solution spaces accompanied by an enormous number of feasible solutions. For this purpose, a multi-armed bandit framework serves as a foundation, where each observed simulation is incorporated into the memory of GMAB. Based on this memory, genetic operators guide the search, as they provide powerful tools for exploration as well as exploitation. The empirical results demonstrate that GMAB achieves superior performance compared to benchmark algorithms from the literature in a large variety of test problems. In all experiments, GMAB required considerably fewer simulations to achieve similar or (far) better solutions than those generated by existing methods. At the same time, GMAB's overhead with regard to the required runtime is extremely small due to the suggested tree-based implementation of its memory. Furthermore, we prove its convergence to the set of global optima as the simulation effort goes to infinity.

Even though base-stock policies are per se straightforward, determining them in complex, stochastic multi-echelon supply chains is often cumbersome or even analytically impossible. Therefore, a wide range of heuristics has been proposed for this purpose. This is the first study considering the problem as a multi-armed bandit problem. In this context, we investigate two algorithms: first, we propose an approach that is based on upper confidence bounds and priority queues. This so-called PQ-UCB algorithm allows us to drastically reduce the runtime of upper confidence bound allocation strategies in problems with large action spaces. Subsequently, we apply the parameter-free sequential halving (SH) algorithm. We investigate various scenarios to compare the performance of both algorithms with the performance of a genetic algorithm and a simulated annealing algorithm taken from the literature. PQ-UCB as well as SH outperform both benchmark metaheuristics and require substantially less effort related to parameter tuning (or even no effort in the case of SH). As multi-armed bandits are not common in inventory optimisation so far, we aim to emphasise their strengths and hope to promote their dissemination also in other domains of supply chain management.

Das Newsvendor-Modell ist ein klassisches Entscheidungsproblem aus dem Bereich des Produktionsmanagements. Es zielt auf die Ermittlung optimaler Bestellmengen bzw. optimaler Losgrößen bei stochastischer Nachfrage ab. Üblicherweise wird unterstellt, dass der Entscheidungsträger, der je nach Interpretation des Problems Händler oder Produzent ist, seinen erwarteten Gewinn maximiert und somit risikoneutral eingestellt ist. Insbesondere wenn die Spanne zwischen der risikoneutralen Lösung und der extreme Risikoaversion ausdrückenden Maximin-Lösung groß ist, erscheint jedoch die Berücksichtigung von Risikoaversion angebracht. Wir analysieren deshalb zwei etablierte Modellrahmen, um das Newsvendor-Problem unter der Annahme risikoaverser Akteure zu lösen, nämlich das Bernoulli-Prinzip sowie die Theorie spektraler Präferenzen. Das Kalkül auf Basis spektraler Präferenzen ist mithilfe einer approximativen Lösung mit geringem Aufwand durchführbar. Der Rechenaufwand für das Bernoulli-Prinzip ist im Vergleich dazu wesentlich größer, was letztlich der Preis für eine exakte Lösung ist. Darüber hinaus diskutieren wir die relative Lage von Maximin- und risikoneutraler Lösung sowie die Einschachtelung der exakten Lösung durch diese beiden markanten (und leicht zu berechnenden) Benchmark-Lösungen.