Essays in technological innovation & financial economics
Abstract/Contents
- Abstract
- This thesis examines the effects of technological innovation, particularly recent developments in machine learning and artificial intelligence (ML/AI), on firm growth, productivity, investment and competitiveness. It has two parts. The first chapter of my dissertation takes a broad view to ask a more fundamental question: do these technologies add value, and how can we quantify this? Academic literature is divided into two broad schools of thought. The first is that ML/AI represent general purpose technologies comparable to electricity or the steam engine, citing the extensive and expanding applications as supporting evidence. The second suggests that the utility of ML/AI is, in reality, more limited, and that the technological landscape is still evaluating added value while in the inflationary phases of a hype cycle. The major challenge associated with this literature is in measuring timing and intensity: what firms use ML/AI, and how extensively is it applied in business functions? The bulk of research in this field has focused on job postings data, which requires subjective feature construction by the researcher. Moreover, jobs data does not provide a precise time series of adoption and utilization intensity. My paper improves upon these approaches by developing a novel methodology based on cutting edge techniques from natural language processing. I adopt deep learning and topic modeling frameworks for unsupervised textual analysis to generate measures superior to more traditional scaled frequency-based approaches. I show that ML/AI utilization is associated with enhanced predictive capabilities and reduced cash flow volatility, with significantly more accurate earnings forecasts by firms. Firms using ML/AI show higher capital and labor productivity, as well as higher sales growth, profitability and market returns. My work helps shed light on the impact of ML/AI in a corporate setting, building on similar work focusing more granularly on labor markets. I show that the evidence is supportive of the general purpose technology hypothesis, and that the widespread adoption of ML/AI is correlated with positive outcomes across a range of industries and markets. Moreover, I show a substitution effect, with firms cutting back on employment and increasing investment in technological innovation. In the second chapter, I work towards understanding the effects of these new technologies on smaller firms. In particular, I study the role of democratized access to ML/AI technologies in encouraging productivity and innovation. Technological innovation has historically been a major driver of economic growth, with Schumpeterian creative destruction and subsequent resource reallocation supporting higher levels of equilibrium output. In recent decades, there has been evidence that suggests that these economic mechanisms may not be working well: increased barriers to entry, reduced business dynamism, asymmetric contributions to technological innovation, a widening gap between small and large firms, and reduced productivity growth. This has led to decreased industry competitiveness and new firm market entry, with risks of predatory pricing, reduced wage growth and consumer surplus, and diminished incentives to innovate. Larger firms have seen greatly increased R&D investment and growth in digital capital holdings, which has fueled high research productivity, product diversification and technological complements. I emphasize the role of open-source ML/AI technologies in reducing this disparity and leveling the playing field for smaller firms: specifically, I study the unexpected public release of TensorFlow. The open-source release of TensorFlow rep- resents an exogenous shock to the cost of ML/AI related digital capital: firms are able to enjoy the benefits of these technologies without prohibitive investments in high skill human capital and technological infrastructure. This natural experiment provides a unique setting to study the effect of open-source technology in supporting small firm growth. My main findings are consistent with the hypothesis that digital capital accumulation positively impacts firm growth. I show that small, TensorFlow user firms have higher ex-post sales growth, market returns, and profitability. These firms are also more likely to innovate, and the evidence is suggestive that a larger share of user firms is associated with subsequent declines in a range of industry concentration measures. My findings support the reasoning that digital capital encourages ML/AI utilization which allows for greater unstructured task automation leading to increased labor productivity. Firms are also able to better forecast demand and reduce volatility of uncertain future cash flows. My research emphasizes asymmetric gains from technological innovation as a driver of productivity slowdown and reduced wage growth. I show that open-source technologies supporting infrastructure may help enhance competition and the scope for future proprietary innovation. Finally, I relate ML/AI capital formation to a broader literature discussing the efficacy and applications of these new technologies, and their effects on labor markets and productivity growth.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2022; ©2022 |
| Publication date | 2022; 2022 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Mukerji, Abhimanyu |
|---|---|
| Degree supervisor | Seru, Amit |
| Thesis advisor | Seru, Amit |
| Thesis advisor | Rauh, Joshua |
| Thesis advisor | Robles Garcia, Claudia |
| Thesis advisor | Strebulaev, I. A. (Ilya A) |
| Degree committee member | Rauh, Joshua |
| Degree committee member | Robles Garcia, Claudia |
| Degree committee member | Strebulaev, I. A. (Ilya A) |
| Associated with | Stanford University, Graduate School of Business |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Abhimanyu Mukerji. |
|---|---|
| Note | Submitted to the Graduate School of Business. |
| Thesis | Thesis Ph.D. Stanford University 2022. |
| Location | https://purl.stanford.edu/xz167nw8839 |
Access conditions
- Copyright
- © 2022 by Abhimanyu Mukerji
Also listed in
Loading usage metrics...