Joint work with Santiago Pereda Fernández. Published in: Econometric Reviews, 44(9), September 2025 (pp. 1321-1360).
Abstract. Conventional methods for the estimation of peer, social or network effects are invalid if individual unobservables and covariates correlate across observations. In this paper we characterize the identification conditions for consistently estimating all the parameters of a spatially autoregressive or linear-in-means model when the structure of social or peer effects is exogenous, but the observed and unobserved characteristics of agents are cross-correlated over some given metric space. We show that identification is possible if the network of social interactions is non-overlapping up to enough degrees of separation, and the spatial matrix that characterizes the co-dependence of individual unobservables and covariates is known up to a multiplicative constant. We propose a GMM approach for the estimation of the model’s parameters, and we evaluate its performance through Monte Carlo simulations. Finally, we revisit an empirical application about classmates in college. Contrasting with conventional methods, our methodology can estimate zero, non-significant peer effects on both academic performance and major choice.
Abstract. In this paper I examine episodes in which superstar inventors relocate to a new city. In particular, in order to assess whether the beneficial effects of physical proximity to a superstar have a restricted network dimension or a wider spatial breadth (spillovers), I estimate changes in patterns of patenting activity following these events for two different groups of inventors: the superstar’s close collaborators, and all the other inventors in a given urban area, for both the locality where the superstar moves to and for the one that is left behind. In the case of collaborators, I restrict the attention to patents realized independently from the superstar. The results from the event study register a large and persistent positive effect on the collaborators in the city of destination, as well as a simultaneous negative trend affecting those still residing in the previous location. In the long run, these effects translate into an increased difference between the two groups of about 0.16 patents per inventor. Conversely, no city-wide spillover effect can be attested, offering little support to place-based policies aimed at inducing a positive influx of top innovators in urban areas.
Abstract. We build a model of production network formation that enables econometric estimation of the determinants of supplier choice, like trade costs or matching frictions. The model informs an estimator obtained from a transformation of the multinomial logit likelihood function that conditions on two network statistics: the out-degree of sellers (a sufficient statistic for the seller marginal costs) and the in-degree of buyers (which is determined by decisions of buyers, like “make-or-buy”). In an empirical application, this estimator shows that a major Costa Rican highway contributed to shuffle the spatial distribution of firm-to-firm linkages.
Abstract. We explore the effect of a reduction in overall labor costs, indirectly induced by an Italian reform that weakened employment protection legislation, on the productivity distribution of manufacturing firms. Due to the unique institutional features of the Italian collective bargaining system, in the manufacturing sector the reform led to a clean reduction in average worker compensation, without altering the average structure of employment relationships. This decrease in labor cost resulted in a reduction in average total factor productivity (TFP) among less productive firms, and an increase at the upper end of the distribution. We pair these findings with increased entry and exit dynamics among low-productivity firms, suggesting the presence of an adverse selection mechanism at the bottom of the TFP distribution, enhanced by the reform. We formalize this concept via a general equilibrium model that links productivity to frictions in the markets for inputs.
Abstract. In a heterogeneous firm economy with monopolistic competition, could informational asymmetries between entrepreneurs and financial intermediaries sometimes improve welfare? We study this question by introducing informational financial frictions into a Melitz economy where banks finance entrepreneurs under asymmetric information. While aggregate productivity monotonically decreases with these frictions, welfare can be maximized at intermediate levels of information asymmetry due to a trade-off between productivity and product variety. We show that the decentralized equilibrium is generically inefficient, as banks do not internalize the variety benefits of entry. Additionally, input cost distortions can improve welfare when financial frictions are severe by tightening firm selection.
Abstract. We study the effects of acquisitions on firms and their production networks in Türkiye using rich administrative firm-to-firm transaction data. Leveraging a staggered event-study design, we compare post-acquisition outcomes of target firms and their trading partners to matched controls. Acquisitions increase the intangible intensity of target firms but have no consistent effects on conventional performance measures. A key finding is that the network consequences of acquisitions depend on the acquirer’s origin. Domestic acquisitions lead to tangible capital deepening and strengthen existing buyer-supplier relationships along the intensive margin, while foreign acquisitions tend to shift production toward outsourcing and diversify network connections. We argue that these differences stem from variation in firms’ relationship capability: their ability to sustain productive links in a network governed by incomplete contracts.
Abstract. This paper introduces an estimator for quadratic forms based on the linear parameters of a semi-parametric model. The leading example is the workhorse model of wage determination by Abowd, Kramarz, and Margolis (1999, AKM): our estimator targets standard variance components while allowing for a nonparametric treatment of both worker- and firm-level observable characteristics. We propose a bias-corrected estimator robust to heteroskedasticity that controls for approximating functions of the covariates. We show that this estimator is asymptotically unbiased and consistent when the number of linear parameters (e.g. the AKM fixed effects) is proportional to the sample size. In particular, consistency hinges on a strengthened smoothness condition (which we discuss for the first time) on the nonparametric component’s functional class. In an empirical application, we show that adding a rich set of controls to the standard AKM model yields implausibly large firm effects. Our method addresses this issue, yielding estimates of variance components that aremore robust relative to conventional approaches. Confounding—not functional-form choice—drives the standard model’s instability.
Abstract. How much do worker skills, firm pay policies, and their interaction contribute to wage inequality? Standard approaches rely on latent fixed effects identified through worker mobility, but sparse networks inflate variance estimates, additivity assumptions rule out complementarities, and the resulting decompositions lack interpretability. We propose TWICE—Tree-based Wage Inference with Clustering and Estimation—a framework that models the conditional wage function directly from observables using gradient-boosted trees, replacing latent effects with interpretable, observable-anchored partitions. This trades off the ability to capture idiosyncratic unobservables for robustness to sampling noise and out-of-sample portability. Applied to Portuguese administrative data, TWICE outperforms linear benchmarks out of sample and reveals that sorting and non-additive interactions explain substantially more wage dispersion than implied by standard AKM estimates.
Abstract. Social and economic interactions on networks are inherently complex, and econometric methods offer only a limited toolkit for empirically studying highly nonlinear theories of social behavior, such as learning models. To address this challenge, we bridge Graph Neural Networks with the adversarial framework for structural econometric estimation by Kaji, Manresa, and Pouliot (2023). Specifically, we adapt adversarial estimation to graph-structured data by integrating geometric deep learning into the discriminator and introducing a surrogate model for the outer minimization problem. We demonstrate the feasibility of our approach on the canonical linear-in-means model of peer effects, and provide an open-source Python implementation available as a lightweight prototype on GitHub and PyPI.