Research

Abstract: We analyze the links between advances in financial technology, investors’ sophistication, and the composition and returns of their financial portfolios. We develop a portfolio choice model under asymmetric information and derive some theoretical predictions. Using detailed microdata from Banca d’Italia, we test these predictions for Italian households over the period 2004-2020. In general, heterogeneity in portfolio composition and in returns between sophisticated and unsophisticated investors grows with improvements in financial technology. This heterogeneity is reduced only if financial technology is accessible to everyone and if investors have a similar capacity to use it.

Supported by The Sandoz Family Foundation - Monique de Meuron Program for Academic Excellence 2022

Prepared for the RCFS 2022 Winter Conference and RCFS Special Issue on "Finance for the Greater Good"

Abstract: In statistics, samples are drawn from a population in a data-generating process (DGP). Standard errors measure the uncertainty in estimates of population parameters. In science, evidence is generated to test hypotheses in an evidence-generating process (EGP). We claim that EGP variation across researchers adds uncertainty—nonstandard errors (NSEs). We study NSEs by letting 164 teams test the same hypotheses on the same data. NSEs turn out to be sizable, but smaller for more reproducible or higher rated research. Adding peer-review stages reduces NSEs. We further find that this type of uncertainty is underestimated by participants.

Abstract: Artificial intelligence (AI) is transforming productivity and market structure, yet the roots of firm dominance in the modern economy remain unclear. Is market power driven by AI capabilities, access to data, or the interaction between them? We develop a dynamic model in which firms learn from data using AI, but face informational entropy: without sufficient AI, raw data has diminishing or even negative returns. The model predicts two key dynamics: (1) improvements in AI disproportionately benefit data-rich firms, reinforcing concentration; and (2) access to processed data substitutes for compute, allowing low-AI firms to compete and reducing concentration. We test these predictions using novel data from 2000–2023 and two exogenous shocks—the 2006 launch of Amazon Web Services (AWS) and the 2017 introduction of transformer-based architectures. The results confirm both mechanisms: compute access enhances the advantage of data-intensive firms, while access to processed data closes the performance gap between AI leaders and laggards. Our findings suggest that regulating data usability—not just AI models—is essential to preserving competition in the modern economy.

Supported by The Sandoz Family Foundation - Monique de Meuron Program for Academic Excellence 2022

Prepared for the Carnegie-Rochester-NYU 2025 Conference on Public Policy “The Consequences of AI use on Society and Policy”.

Abstract: Does AI make firms vulnerable or resilient to cyber risk? To answer this question, we develop a novel measure identifying AI-intensive U.S. public firms using publicly available patents and business-description data. While cyber threats typically suppress innovation, AI-intensive firms neutralize this effect. This protective effect strengthens with greater AI experience. Moreover, firms combining AI innovation and implementation exhibit a stronger buffer protecting their innovation and financial outcomes under cyber stress, whereas firms merely implementing AI without internal innovation gain no such resilience. Our results emphasize internal AI innovation as fundamental in enabling firms to effectively withstand cyber threats.

Supported by ArmaSuisse Cyberdefense Campus Grant 2024

Prepared for the RCFS 2025 Winter Conference and RCFS Special Issue on “Understanding Firms, Households and Financing in the Age of AI”

Abstract: In this paper, I build a theoretical model of investors heterogeneous in wealth who are trading under asymmetric information to study the impact of financial information technologies on capital income inequality. In the model, investors have a choice between not participating in the stock-market, investing directly in equities or indirectly through a mutual fund. All market participants can acquire private information about a stochastic asset payoff. To assess the impact of financial innovation, I reduce the cost of stock market participation and of data processing, for funds and investors, over time. I find that lower participation costs always reduce capital income inequality, but lower costs of data processing, for investors or funds, do not guarantee broad increases in household wealth. Instead, the sophisticated investors who already have relatively high levels of wealth are most likely to benefit from many of the new information technologies. Lower data acquisition costs, for investors or for funds, make wealthier investors acquire more valuable information. In turn, this makes less wealthy investors more reluctant to trade in a market with a higher degree of asymmetric information. By staying out of the market, the wealth of poor investors grows less than that of wealthier ones.

Winner of the 2021 Outstanding Dissertation Award for Inequality Category, NYU 2021

Winner of the ECB Young Economists' Competition, ECB 2020 

Winner of the Cubist Strategies Prize for Outstanding Research, WFA 2020

Abstract: We show that uncertainty significantly affects how households form expectations, but its impact depends on the source of uncertainty. Using U.S. inflation survey data, we find that households update beliefs more (lower rigidity) when they perceive their prior information as more uncertain, whereas they rely more on prior beliefs (higher rigidity) when new information is perceived as noisier. While broadly consistent with Bayesian updating, households overreact to new information relative to the Rational Expectations benchmark. We show that belief rigidity helps disentangle different sources of uncertainty, which we apply to the recent uncertainty surge. At the pandemic’s onset, belief rigidity declines while uncertainty rises, driven by volatile fundamentals making prior information less reliable. In contrast, during the high-inflation period, both rigidity and uncertainty increase, driven by noisier information. Our findings highlight the importance of distinguishing uncertainty sources for central bank policy and communication strategies.

Supported by The Sandoz Family Foundation - Monique de Meuron Program for Academic Excellence 2022

Abstract: Data security is often viewed as a compliance cost, an insurance policy against worst-case scenarios. But in today’s data-driven economy, this view is increasingly outdated. For many firms, data protection is a strategic enabler of innovation and growth, particularly when it unlocks access to sensitive, high-value customer data that would otherwise remain untapped due to trust or regulatory barriers. We introduce a novel measure of "data innovation complementarity", the degree to which firms integrate data-security expertise into their broader innovation processes, and show that this integration is a key driver of innovation-led growth in response to regulatory shocks. Using inventor-level patent data and exploiting the staggered adoption of U.S. Data Breach Notification Laws as quasi-exogenous variation in data salience, we find that high-complementarity firms increase innovation output by 115 citation-weighted patents and boost return on assets by 0.175 percentage points over five years. To explain this heterogeneity, we develop a real-options model where integration is costly but generates innovation spillovers. Counterfactuals show that mandating outsourced security reduces firm value by about a third. Our findings reframe data protection as a latent driver of competitive advantage in the modern digital economy.

Supported by The Sandoz Family Foundation - Monique de Meuron Program for Academic Excellence 2022

Abstract: The rapid rise of artificial intelligence (AI) has spurred unprecedented productivity gains, yet its energy intensity poses critical challenges to environmental sustainability. This paper explores the interplay between AI-driven growth and climate policy, focusing on the role of renewable energy in supporting the dual pursuit of technological progress and decarbonization. Leveraging a unique dataset on firm-level ICT investments and state-level renewable energy production, we identify causal linkages using regulatory policy instruments. Our findings reveal that increases in renewable energy production significantly boost firms' ICT investments, suggesting that access to clean energy mitigates potential trade-offs between AI adoption and emissions reduction. Our analysis underscores a growing strategic shift among firms, which increasingly channel energy-intensive AI development to regions with abundant renewable energy. By highlighting the policy levers that reconcile AI innovation with climate goals, this study offers a blueprint for designing regulations that foster sustainable growth.

Supported by The Sandoz Family Foundation - Monique de Meuron Program for Academic Excellence 2022

Supported by ESSEC Research Grant 2024 

Abstract: We study a model where firms accumulate data as a valuable intangible asset. Data accumulation affects firms' dynamics. It increases the skewness of the firm size distribution as large firms generate more data and invest more in active experimentation. On the other hand, small data-savvy firms can overtake more traditional incumbents, provided they can finance their initial money-losing growth. Our model can be used to estimate the market and social value of data.

Abstract: We analyze the expansion of Big Data and artificial intelligence technologies from the perspective of economic theory. We argue that these technologies can be viewed from three perspectives: (1) as an intangible asset; (2) as a search and matching technology; and (3) as a forecasting technology. These points of view shed light on how new technologies are likely to affect matching between firms and consumers, productivity growth, price discrimination, competition, inequality among firms, and inequality among workers.

Abstract: Macro-prudential policies aimed at mitigating systemic financial risks have become part of the policy toolkit in many emerging markets and some advanced countries. Their effectiveness and efficacy are not well-known, however. Using panel data regressions, we analyze how changes in balance sheets of some 2800 banks in 48 countries over 2000–2010 respond to specific policies. Controlling for endogeneity, we find that measures aimed at borrowers – caps on debt-to-income and loan-to-value ratios, and limits on credit growth and foreign currency lending – are effective in reducing leverage, asset and noncore to core liabilities growth during boom times. While countercyclical buffers (such as reserve requirements, limits on profit distribution, and dynamic provisioning) also help mitigate increases in bank leverage and assets, few policies help stop declines in adverse times, consistent with the ex-ante nature of macro-prudential tools.

Abstract: This paper investigates the effects of national culture on firm risk-taking, using a comprehensive dataset covering 50,000 firms in 400 industries in 51 countries. Risk-taking is found to be higher for domestic firms in countries with low uncertainty aversion, low tolerance for hierarchical relationships, and high individualism. Domestic firms in such countries tend to take substantially more risk in industries which are more informationally opaque (e.g., finance, mining, oil refinery, IT). Risk-taking by foreign firms is best explained by the cultural norms of their country of origin. These results hold even after controlling for legal constraints, insurance safety nets, and economic development.

Winner of the President's Prize by ACEI