|
on Industrial Organization |
| By: | Anton Korinek; Jai Vipra |
| Abstract: | This paper examines the evolving structure and competition dynamics of the rapidly growing market for foundation models, with a focus on large language models (LLMs). We describe the technological characteristics that shape the AI industry and have given rise to fierce competition among the leading players. The paper analyzes the cost structure of foundation models, emphasizing the importance of key inputs such as computational resources, data, and talent, and identifies significant economies of scale and scope that may create a tendency towards greater market concentration in the future. We explore two concerns for competition, the risk of market tipping and the implications of vertical integration, and we evaluate policy remedies that aim to maintain a competitive landscape. Looking ahead to increasingly transformative AI systems, we discuss how market concentration could translate into unprecedented accumulation of power, highlighting the broader societal stakes of competition policy. |
| JEL: | D43 K21 L4 L86 O33 |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:33139 |
| By: | Nigar Hashimzade; Limor Hatsor; Artyom Jelnov |
| Abstract: | Recent antitrust regulations in several countries have granted exemptions for collusion aimed at achieving environmental goals. Firms can apply for exemptions if collusion helps to develop or to implement costly clean technology, particularly in sectors like renewable energy, where capital costs are high and economies of scale are significant. However, if the cost of the green transition is unknown to the competition regulator, firms might exploit the exemption by fixing prices higher than necessary. The regulator faces the decision of whether to permit collusion and whether to commission an investigation of potential price fixing, which incurs costs. We fully characterise the equilibria in this scenario that depend on the regulator's belief about the high cost of green transition. If the belief is high enough, collusion will be allowed. We also identify conditions under which a regulator's commitment to always investigate price fixing is preferable to making discretionary decisions. |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:arx:papers:2411.06095 |
| By: | Francis Annan |
| Abstract: | We study the direct and indirect effects of randomized entry. In partnership with the two largest service providers in Ghana, we implement a three-step design that randomizes the entry of new financial mobile money vendors, who also sell non-financial goods/services, across local markets. This mixed financial and non-financial services setting is widespread and naturally emerges as the market entry approach for several real-world financial markets. Randomized entry increases firm conduct and service quality and decreases price-cost markups, indicating positive consumer surplus. We find evidence of within-market revenue reallocation and expansion for mobile money and a large services multiplier: revenues for non-financial goods/services increased (+20%), with aggregate service industry revenues increasing. These improvements emphasize the “real effects” of financial markets on the local economy, and come from adoption externalities and aggregate increase in household expenses. Entry increases local economic activity, and it does so not only by changing markets for digital financial services, but also by transforming the non-financial services sector. These effects are key ingredients for advancing basic and applied knowledge on firm entry in industry equilibrium. |
| JEL: | D18 D62 G20 G50 L22 L26 O12 |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:33134 |
| By: | Jack Fisher |
| Abstract: | Many workers provide services for customers via digital platforms that may exert monopsony power. Typical expositions of this phenomenon are inapplicable because platforms post prices to both sides of a two-sided market, and platform-specific labor supply is hard to measure when workers multi-app. This paper develops a model of a typical gig labor market that deals with these issues. Platforms exploit monopsony power to markup their commission rate and reduce equilibrium wages. A worker union sets the first-best commission rate when the customer market is competitive. I estimate the model using public data, including causal estimates from the literature on Uber’s US ridesharing marketplace. The results imply the platform exploits labor market power to depress drivers’ earnings but faces competition for passengers. An optimally set commission cap raises wages by 14 percent, but minimum wages on utilized hours harm workers. |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_11444 |
| By: | Dietmar Harhoff; David Heller; Paul P. Momtaz |
| Abstract: | We show that firm and industry, rather than inventor and invention factors, explain more than half of the variation in inventor returns in administrative employer-inventor-patent-linked data from Germany. Between-firm variation in inventive rents is strongly associated with inventor mobility. Inventors are more likely to make a move just before a patent is filed than shortly thereafter and benefit from their move through a mobility-related marginal inventor return. Employers that pay inventor returns in excess of the expected return gain a favorable position in the market for inventive labor with subsequent increases in patent quality and quantity. Consistent with theoretical arguments, effect sizes also depend on employer-inventor technological complementarity, degree of competition, and invention quality. |
| Keywords: | inventor returns, labor mobility, patents, inventive productivity |
| JEL: | O31 J24 J62 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_11449 |
| By: | Zhiguo He; Jing Huang; Cecilia Parlatore |
| Abstract: | We develop a credit market competition model that distinguishes between the information span (breadth) and signal precision (quality), capturing the emerging trend in fintech/non-bank lending where traditionally subjective (“soft”) information becomes more objective and concrete (“hard”). In a model with multidimensional fundamentals, two banks equipped with similar data processing systems possess hard signals about the borrower's hard fundamentals, and the specialized bank, who further interacts with the borrower, can also assess the borrower's soft fundamentals. Increasing the span of the hard information hardens soft information, enabling the data processing systems of both lenders to evaluate some of the borrower's soft fundamentals. We show that hardening soft information levels the playing field for the non-specialized bank by reducing its winner's curse. In contrast, increasing the precision or correlation of hard signals often strengthens the informational advantage of the specialized bank. |
| JEL: | G21 L13 L52 O33 O36 |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:33141 |
| By: | Panle Jia Barwick; Hyuk-Soo Kwon; Shanjun Li; Yucheng Wang; Nahim B. Zahur |
| Abstract: | This paper examines the impact of industrial policies (IPs) on innovation in the global automobile industry. We compile the first comprehensive dataset linking global IPs with patent data related to the auto industry from 2008 to 2023. We document a major shift in policy focus: by 2022, nearly half of all IPs targeted electric vehicles (EV)-related sectors, up from almost none in 2008. In the meantime, there has been a clear technological transition from internal combustion engine (GV) technologies to EV innovations. Our analysis finds a positive relationship between policy support and innovation activity. At the country level, a one-standard-deviation increase in five-year cumulative EV-targeted IPs is associated with a four-percent rise in new EV patent applications. Firm-level analyses (using OLS, IV, and PPML) indicate that a ten-percent increase in EV financial incentives received by automakers and EV battery producers leads to a similar four-percent increase in EV innovations. We confirm the importance of path dependence in the direction of technology change in the automobile industry but find no evidence that EV-targeted IPs stimulate innovation in GV technologies. |
| JEL: | H20 L5 L60 L62 O3 |
| Date: | 2024–11 |
| URL: | https://d.repec.org/n?u=RePEc:nbr:nberwo:33138 |
| By: | Thomas Licht; Klaus Wohlrabe |
| Abstract: | This paper examines the adoption of Artificial Intelligence (AI) among German firms, leveraging firm-level data from the ifo Business Survey. We analyze the diffusion of AI across sectors and firm sizes, showing a significant increase in AI usage from 2023 to 2024, particularly in manufacturing and services. The survey data allows us to explore not only sectoral patterns of adoption but also the drivers and barriers that firms face, including firm-specific characteristics and industry dynamics. Additionally, we investigate the role of managerial traits, such as risk tolerance and patience, in shaping AI adoption decisions. Finally, we assess the potential pro-ductivity impacts of AI at the firm level, with a focus on the expected long-term benefits of AI for different sectors of the German economy. Our findings contribute to the growing body of research on AI adoption by providing new evidence from a non-US context, offering valuable insights for both academia and politics. |
| Keywords: | artificial intelligence, AI, ifo business survey, productivity |
| JEL: | M15 O30 C83 L20 |
| Date: | 2024 |
| URL: | https://d.repec.org/n?u=RePEc:ces:ceswps:_11459 |