This paper studies the implications of the use of data for advertising on firms and consumers. I develop a model of search frictions in product markets where heterogeneous firms choose size and target ads to different types of consumers. Firms endogenously adopt Big Data technologies, which lower the cost of targeting. Using firm-level balance-sheet data combined with a technology survey for France, I show that firms analyzing Big Data from social media (i) use it for advertising, (ii) are larger, (iii) are more productive, and (iv) show higher overhead costs. I calibrate the model to France in the late 2010s. The endogenous use of targeted advertising and Big Data creates a feedback loop between firm size and the use of consumer information, allowing the model to replicate empirical correlations between Big Data adoption, firm size, and cost structure. This feedback loop explains why only large firms that use Big Data benefit from a technological change that reduces the cost of processing information. From the consumer side, technological change increases welfare but reallocates consumption towards high-valuation types. Finally, I show that a tax on digital advertising improves welfare, partially internalizes search externalities from production and advertising, reduces the size of large firms that use Big Data, and reallocates consumption towards low-valuation types.

Presented at: UCL Macro Reading Group (2024), LSE Student Seminar (2024), BSE PhD Jamboree (2024), Universitat de València (2024), 22nd ZEW Conference on the Economics of ICT (2024) , XXVII Workshop on Dynamic Macroeconomics (2024), CREI Macroeconomics Lunch Workshop (2024),  SAEe (2024), VIII Winter Bellaterra Macroeconomics Workshop (2025)  

Previous title: Firms and Big Data: Adoption, Use and Impacts

We study how manufacturing establishments in India adapt to flood risk. Combining establishment-level data with geo-coded flood records and regional economic indicators, we examine how production and investment decisions respond to flood events conditional on historical exposure. We find that investment is more resilient in high-risk areas, consistent with forward-looking adaptation. To rationalize these findings, we propose a firm dynamics model featuring flood risk and private insurance to floods through a flood preventing capital. To overcome the course of dimensionality in this dynamic spatial model with aggregate uncertainty, we resort to Deep Learning techniques. We employ the model to quantify the aggregate economic impact of floods and evaluate the effectiveness of adaptation in mitigating climate-induced damages, and find that the proposed mechanism can replicate the patterns in the data.

Presented at: CREI International Lunch (2023), BSE PhD Jamboree (2023), PSE Summer School on Climate Change (2023), SAEe (2023)

Previous title: The Transmission of Climate Shocks: The Case of Floods in India

We analyze the implications of data use for firms in Europe using the EIB Investment Survey, which combines questions about technology with balance-sheet information. Firms that use data are more likely to innovate, to invest in product development and to introduce products that are new to global markets. We analyze the impact of the General Data Protection Regulation (GDPR) on European firms, finding a decrease in software investment on the extensive margin. For data-intensive firms, investment in R&D decreased after the regulation. The correlation between innovation and data use is more important for small firms, which reduce R&D by a larger extent after the implementation of the GDPR.