Research

Internal and External R&D: An analysis of costs and benefits. (In progress, 2025) [solo] Paper Slide Poster Short Video

• Abstract: This paper analyzes the costs and benefits of internal and external R&D activities. Using Dutch production and innovation surveys between 2000 and 2020, focusing on the IT industry, I document an increasing trend of R&D activities across industries and present evidence suggesting that internal and external R&D are complementary. To rationalize these findings, I build and estimate a dynamic discrete choice model of R&D, which explicitly includes specific investment costs of R&D. I find that the cost of doing external R&D is ~4 times higher than the internal R&D, reflecting the transaction costs of such contract and explaining the observed small share of external R\&D firms in the data. To mimic the Dutch Tax Incentives for Innovation scheme, I simulate the effect of two types of R&D subsidization programs. I find that if the government has no preference for any particular R&D activity, the share of R&D-active firms increases the most and leads to higher change in the firm's total valuation. On the other hand, if the government prefers the firms only to perform internal R&D, the share of R&D-active firms is practically unchanged and leads to lower change in the valuation.

Mergers and Innovation: An Exploration of Firms' Direction and Scope of Innovation. (In progress, 2025). [joint with S. Dobbelaere and J. L. Moraga-Gonzalez] [draft soon!]

• Mergers are often justified as enhancing innovation through synergies, yet their effects on the scope (magnitude) and direction (composition) of firm innovation remain debated. This paper develops a theoretical framework integrating mergers with multi-project innovation decisions, showing that: mergers can increase, decrease, or leave unchanged scope and direction depending on project substitutability. Empirically, using a version of the matching estimator on U.S. public mergers from 1980 to 2020, we find no robust overall change in innovation scope but improved directional alignment to high-impact benchmarks by 0.02 (on a 0-1 cosine similarity scale). Heterogeneity reveals downsides: high technological overlap reduces scope by 21–60% and diverges direction by 0.07; inactive acquirer–active target mergers shrink scope by 47% and diverge by 0.21; both-active cases cut scope by 16% and diverge by 0.06. These findings highlight mergers' potential to contract research diversity in overlapping or innovation-intensive deals, informing antitrust policy in tech-heavy industries.

Entry into Innovation Areas. (In progress, 2025) [solo, draft soon!]

• Abstract: This paper investigates the notion of innovation direction as the outcome of firms' entry decisions into specific innovation areas. Using Latent Dirichlet Allocation, a natural language processing technique in topic modeling, on 631,000 U.S. electronics patents (1990–2019), I first optimally classify each patent and firm's innovation portfolio into 20 innovation areas. Key stylized facts reveal persistent multi-domain engagement, with patents and firms averaging 2.86 and 4.68 topics respectively, alongside rising entry in software-intensive fields like user interfaces and content services, contrasted by declines in capital-intensive areas such as memory architecture. Valuation heterogeneity is evident, with patent values increasing unevenly across domains, while competitive dynamics show varying competitor numbers over time. To explain these patterns, I build an entry model estimated via moment inequalities that provides bounds on parameters, indicating that a 1% rise in competitors reduces the firm's innovation profit by 18.44 to 56.85 million USD, an increase of one standard deviation of the portfolio diversification boosts the profit by 0.57 to 36.81 million USD, and the average entry costs are bounded at 75.93 million USD.

The Impact of Subsidy on the Firm's Direction and Scope of Innovation. (In progress, 2025). [joint with C. Cincotta]

Evaluating the congestion-reducing effects of road rationing policy: Evidence from Jakarta's odd-even policy (Previously circulated as: Does Traffic Management Matter? Evaluating Congestion Effect of Odd-Even Policy in Jakarta. (Case Studies on Transport Policy, 2025). [joint with M. Halley Yudhistira, Y. Sofiyandi, and M. Firman Hidayat] Paper

• Abstract: Travel demand restriction policies are becoming popular to reduce traffic jams in heavily congested urban areas. This paper evaluates the short-term impact of the odd–even number plate policy – a form of travel demand restriction – using Google Maps data on traffic congestion in Jakarta, one of the world’s most congested cities. The estimation result of hourly travel time data at road section levels shows that the odd–even policy statistically significantly reduced travel time by 3.0%, on average, during the first month of its implementation. The effect was higher – about 10% during the weekend and the weekly afternoon peak hours. The estimates were robust across various specifications, and no effect was found when looking for travel speed changes in untreated parts of the road network. This study also suggests that road users are likely to depart early in the morning to avoid the policy. The effect seems stagnant after the fourth week of the policy’s introduction. This study suggests that the odd–even number plate policy alone would be ineffective for the traffic congestion reduction agenda in Jakarta’s urban environment.