Browsing by Person "Schwalbe, Ulrich"

Now showing 1 - 8 of 8

Collusive behavior in markets: partial cartels, tacit collusion, and artificial intelligence
(2023) Grüb, Jens T.; Schwalbe, Ulrich
This cumulative dissertation thesis consists of five papers on collusive behavior. The individual research areas are (i) the link between partial cartels and mergers, (ii) the effect of price announcement letters on the cement price in Germany, (iii) limitations of the transferability of experimental results with Q-learning agents in economic environments, (iv) the strategic choice of price-setting algorithms in a game theoretic model, and (v) the influence of algorithm heterogeneity on the ability to collude. It is often assumed that a cartel consists of all firms in a market. Cartels, however, do not necessarily have to be all-inclusive. If only some firms in a market are part of a cartel, it is called a partial cartel. The structure of such a partial cartel as well as the behavior of outside firms are explained in the literature. How such a partial cartel is formed, however, is not explained. This question is addressed by establishing the link between mergers and partial cartels in a Bertrand model with heterogeneous capacities and heterogeneous discount factors. The critical change in the discount factor induced by a merger which then leads to a partial cartel is described. Thus, it is also shown that coordinated and non-coordinated effects can occur simultaneously. While the hard-core cartel in the German cement market has been analyzed in several studies, a phase of tacit collusion with parallel behavior between 2008 and 2017 has not been investigated as thoroughly. During this period, 15 firms that covered 93 percent of the cement sales in Germany had sent so-called ``price announcement letters'' to all customers. Due to the fact that the firms are customers of each other and the vertical integration on the supply side of this market, these letters were effectively sent between all firms in this market. This can be seen as a means to induce parallel behavior and reduce competition. The resulting increase in the cement price is estimated using the traditional before-and-after approach and a simple forecast. In addition, the price increase is estimated using a forecast with an autoregressive integrated moving average (ARIMA) error to check the robustness of the mentioned estimates. The price increase is estimated to be 6 percent. It is, therefore, comparable to the estimated overcharge of 7.6 percent of the hard-core cartel which shows that tacit collusion in the form of price announcement letters can be as harmful as a hard-core cartel. The third major topic is algorithmic collusion. Because of technological advances, especially in the field of artificial intelligence, more and more firms are able to use self-learning algorithms. The major concern with these algorithms is that -- without being explicitly programmed to do so -- they learn to behavior collusively. Experiments with algorithms employing reinforcement learning have been carried out that confirm such concerns. How the results of these experiments transfer to more realistic settings is extensively discussed. For these specific algorithms, technical and economic limitations hinder the direct application to real-world markets. Using the same type of algorithm, experiments are carried out to investigate how this algorithm compares against one simpler learning algorithm and two non-learning algorithms. The results of the experiments are used in a simple one-shot game where the players select from one of the algorithms instead of setting prices directly. The payoffs are generated by letting the algorithms play a Bertrand duopoly game. Analyzing the game reveals that both players choosing a self-learning algorithm is not a Nash equilibrium. Simpler yet effective pricing rules are more profitable for firms. Besides the reinforcement algorithms discussed so far, there are also deep-learning algorithms that make use of advanced methods like artificial neural networks. These do not suffer from some of the shortcomings of the above-mentioned algorithms and have a broader field of application. Such deep-learning algorithms are used in experiments to analyze how sources of heterogeneity affect the level of collusion. Heterogeneity is introduced by using two different types of algorithms and two different parameter settings. Experiments are run in various economic environments. First, the level of collusion depends on the economic environment and algorithm type. Secondly and more importantly, the level of collusion almost always decreases with heterogeneity. In even more complex markets with more firms and multiple products, this indicates that algorithmic collusion is not yet an issue.
Consumer prices
effects of learning algorithms and pandemic-related policy measures
(2023) Buchali, Katrin; Schwalbe, Ulrich
When it comes to product prices, two major topics have dominated the public debate in recent years: One is pricing with the help of artificial intelligence, and the other is the price level, which has risen more than usual with the onset of the COVID-19 pandemic. Higher prices create a loss of consumer surplus and possibly total welfare, which is the reason this topic has become ubiquitous in political discussions. This dissertation contributes to the debate by extending the existing literature on algorithmic pricing, which is said to facilitate personalized pricing, as well as collusive behavior and to enhance the general understanding of how government measures enforced during the COVID-19 pandemic contributed to (short-time) price developments. Thereby, the first part of the thesis addresses the concern that tacit collusion might occur if firms employ learning algorithms, as several simulation studies have demonstrated that algorithms using reinforcement learning are able to coordinate their pricing behavior and, as a result, achieve a collusive outcome without having been programmed for it. We discuss several conceptual challenges as well as challenges in the real-world application of algorithms and show by or own simulations that resulting market prices strongly depend on the type of algorithm or heuristic that is used by the firms to set prices. In the subsequent part of the thesis we examine how a self-learning pricing algorithm performs when faced with inequity-averse consumers. From our simulations we can conclude that consumers sense of fairness, which have prevented firms from engaging in price discrimination in the past years, can be incorporated into firms pricing decisions with the help of learning algorithms, making differential pricing strategies more feasible. The discussion surrounding the above-average price levels in many countries during the COVID-19 pandemic is extended in the third part of the thesis. We present empirical evidence for the impact of government-imposed restrictions and, as a consequence of their enforcement, reduced mobility on consumer prices during the COVID-19 pandemic. We show that the stringency of government measures had a positive and significant impact on consumer prices mainly in the food sector, which means that more stringent measures induced higher consumer prices in these categories.
Digitale Musik - Eine industrieökonomische Analyse der Musikindustrie
(2006) Raschka, Oliver D.; Schwalbe, Ulrich
This is a book about the competition in the (German) music industry. It examines the issues important to the future of the music business and especially the music industry. An in-depth study of the demand und supply side shows the current problems and issues for the record companies. First, the stability of cooperative behavior in file sharing networks. Second, the construction of optimal pricing schemes in conjunction with the optimal number of different versions of a music song in the presence of unauthorized file sharing. For the solution of these problems, the book contains two new (game) theoretic models. The models provide optimal behavior strategies for consumers and record companies as well. The results will be supported by empirical evidence. The organization of the book is as follows: Chapter 1: Introduction. Chapter 2: Product innovations and market structure in the history of the (global) music industry. Chapter 3: Characteristics and marketability of digital information goods. Chapter 4: Market analysis and competition behavior in the (German) music industry. Chapter 5: A simple game theoretic model of cooperation in peer-to-peer (p2p) file sharing networks. An evolutionary approach. Chapter 6: Pricing schemes, product quality and digital rights management in the presence of illegal copying. Chapter 7: Summary. Literature.
Effekte verschiedener Rabattformen
Überlegungen zu einem ökonomisch fundierten Ansatz
(2008) Schwalbe, Ulrich; Inderst, Roman
Der vorliegende Aufsatz diskutiert die ökonomischen Wirkungen von Treuerabatten im Ein-Produkt-Fall. Es zeigt sich, dass die in der letzten Zeit vorgebrachten Argumente bezüglich der ?prokompetitiven? Wirkungen solcher Rabatte mit Skepsis beurteilt werden müssen. Dies gilt insbesondere für die unterstellten Wirkungen von Treuerabatten bei ?doppelten Gewinnaufschlägen?, bei fallenden Durchschnittskosten, bei Größenvorteilen auf der vor- und nachgelagerten Stufe sowie im Zusammenhang mit Preisdiskriminierung und Nachfragemacht. Wir argumentieren, dass viele der behaupteten Wirkungen auch mit Rabattformen erreicht werden könnten, bei denen die Gefahr von wettbewerbsbeschränkenden Wirkungen geringer sein sollte. Unsere Skepsis beruht aber auch darauf, dass oft die Voraussetzungen für die den Argumenten zugrunde liegenden Theorien nicht oder nicht hinreichend gegeben sind?und vor allem oft nicht hinreichend explizit gemacht werden. Allerdings müssen auch die unterstellten wettbewerbsbeschränkenden Wirkungen von Treuerabatten ökonomisch besser fundiert werden. Insgesamt regt dieser Artikel die Weiterentwicklung eines an der Form des jeweiligen Rabattsystems orientierten Beurteilungsmaßstabs an.
Inter-firm R&D networks in pharmaceutical biotechnology
what determines firm's centrality-based partnering capability?
(2013) Schwalbe, Ulrich; Riedel, Nadine; Krogmann, Yin
This paper analyses the inter-firm R&D network formed in the pharmaceutical biotechnology industry during the 1990s from different perspectives: theoretical network formation, firm's structural positions and its collaborations at the entire network level, and the determinants for firm's centrality-based partnering capability. The results indicate that pharmaceutical biotechnology industry has experienced a significant evolutional change in size and structure during 1991-1998. By considering individual structural positions, the descriptive statistics show that in the 1990s, established pharmaceutical companies developed into dominant star players with multiple partnerships while holding central roles in the R&D network. In the network analysis that emphasized aggregate network level, the degree-based and betweenness-based network centralization were not high implying that the distribution of overall positional advantages in the pharmaceutical biotechnology industry is, to a large degree, not unequal and even though most firms in this sector are linked to the R&D network, some of them are more active than others. The current analysis also shows that firm's efficiency, firm's dependency on its complementary resources and firm's experiences at managing partnerships are important determinants for firm's centrality-based partnering capability, which has important managerial implications for understanding firm's strategic partnering behaviour.
Inter-firm R&D networks in the global pharmaceutical biotechnology industry during 1985?1998
a conceptual and empirical analysis
(2011) Krogmann, Yin; Schwalbe, Ulrich
This paper analyses a large database on inter-firm R&D cooperation formed in the pharmaceutical biotechnology industry during the period 1985?1998. The results indicate that network size largely grows, whereas the density of the network declines during the periods. In the network analysis that emphasizes individual structural positions, the empirical results show that small biotechnological companies had a crucial bridging role for the large pharmaceutical firms in the second half of the 1980s. In the 1990s, the bridge role of biotechnology companies became less important and established pharmaceutical companies developed into dominant start players with many collaborators while holding central roles in the research network. The current analysis also shows that degree-based and betweenness-based network centralization are both low implying that the overall positional advantages are relatively equally distributed in the inter-firm R&D network of the pharmaceutical biotechnology industry.
On collusive behavior - models of cartel formation, organizational structure, and destabilization
(2011) Fischer, Julia; Schwalbe, Ulrich
This dissertation contributes to the theoretical literature on cartel formation, organizational structure, and destabilization in Cournot competitive markets. Cartel formation in Cournot competitive markets may take place as a sequential process even if the merger paradox applies. This conclusion was reached after giving up the assumption of symmetric information in cartel formation processes: it is assumed that outside firms are not informed about new cartel agreements and face a time lag by adjusting to changing behavior of some of the market participants. Furthermore, an extension to the standard cartel stability models is presented to capture the influence of communication and organizational structure in a cartel by modeling cartels as social networks. Despite the fact that communication in cartels is costly because contacts between members might be detected by antitrust authorities, it is shown that intensive contacts are possibly stabilizing within a cartel. Both aspects, the costs and benefits of communication in cartels, contribute to the players' valuation of collusion and therefore change cartel stability conditions. Additionally, this model accounts for the influence of leniency programs and fines. A theoretical explanation is given for differences between explicit and tacit collusion on the basis of this network model. Additionally, this dissertation examines whether collusive behavior might be deterred in vertical structures if dominant firms are allowed to apply specific discount schemes. It is shown that the profit maximizing behavior of a monopolistic upstream firm might lead to the deterrence of collusive behavior of downstream firms if the upstream firm is allowed to implement all-units discount schemes. All-units discounts, despite the fact that they are sometimes considered anticompetitive, possess welfare improving effects that are not generally shared by other pricing schemes.
Strategic choice of price-setting algorithms
(2023) Schwalbe, Ulrich; Muijs, Matthias; Grüb, Jens; Buchali, Katrin
Recent experimental simulations have shown that autonomous pricing algorithms are able to learn collusive behavior and thus charge supra-competitive prices without being explicitly programmed to do so. These simulations assume, however, that both firms employ the identical price-setting algorithm based on Q-learning. Thus, the question arises whether the underlying assumption that both firms employ a Q-learning algorithm can be supported as an equilibrium in a game where firms can chose between different pricing rules. Our simulations show that when both firms use a learning algorithm, the outcome is not an equilibrium when alternative price setting rules are available. In fact, simpler price setting rules as for example meeting competition clauses yield higher payoffs compared to Q-learning algorithms.