Bridging Quasi-Experimental and Structural Approaches for Robust Evaluation of US Airline Mergers

summary

The gist

Bridging quasi-experimental and structural approaches for robust evaluation of US airline mergers shows that apparent price stability masks offsetting efficiency and conduct effects in airline…

In short

This study compares quasi-experimental and structural methods to evaluate US airline mergers. Standard methods suggested price reductions, but structural analysis showed these effects were driven by efficiency gains and strategic conduct changes, not necessarily pro-competitive outcomes. A new synthetic GMM method improves detection of conduct effects missed by traditional models.

Key concepts

Difference-in-Differences (DiD)
This is a quasi-experimental method used to estimate the impact of a merger by comparing changes in outcomes before and after the merger in treated markets versus control markets. The paper shows how this simple approach relates to more complex structural models, revealing its limitations when analyzing mergers.
Structural Model
This framework decomposes the price effect from a merger into distinct economic components: efficiency, market power, and coordination. Unlike simple reduced-form analysis, this model is more robust because it explicitly accounts for underlying economic mechanisms driving the observed changes.
Synthetic GMM
This is a modified econometric technique that uses 'synthetic DiD weights' to reweight control observations in Generalized Method of Moments (GMM). This adjustment helps reduce bias in estimates, especially when only a few markets are treated, improving the detection of subtle conduct changes.
Conduct Effects
These refer to the strategic behaviors or coordination among airlines after a merger. The paper finds that apparent price stability can mask these effects; for instance, increased coordination or reduced efficiency due to strategic conduct can offset potential pro-competitive benefits.

Terminology used across episodes

This episode discusses

The paper

Bridging Quasi-Experimental and Structural Approaches for Robust Evaluation of US Airline Mergers · Read on arXiv

Boston University · Analysis Group · University of Virginia

We bridge quasi-experimental and structural approaches to merger retrospectives. First, we derive the difference-in-differences (DiD) equation as a reduced form of a demand-and-supply model, decomposing price effects into changes in efficiency and strategic conduct. We show how pre-merger conduct affects parallel trends and when DiD effects generalize beyond treated markets. Second, we introduce Synthetic GMM, using synthetic-DiD weights to reweight controls, reducing bias relative to standard GMM. Monte Carlo experiments support this result. Across four airline mergers, we find no efficiency gains, no systematic conduct changes, and no price reductions. In one merger, more competitive conduct partly offsets rising costs.

Transcript

Introduction to the show: ident: Economics. The day in economics: theory, mechanisms and the evidence beneath them.

Ruth: We're looking at the paper: "Bridging Quasi-Experimental and Structural Approaches for Robust Evaluation of US Airline Mergers".

Warren: Here's the gist of the paper. Bridging quasi-experimental and structural approaches for robust evaluation of US airline mergers shows that apparent price stability masks offsetting efficiency and conduct effects in airline mergers The synthetic GMM approach can improve our ability to detect changes in conduct that might be missed with standard GMM This paper establishes a connection between quasi-experimental and structural approaches to merger evaluation Our empirical analysis of three major airline mergers reveals that apparent price stability masks offsetting efficiency and conduct effects Methodological Bridge Development The paper develops a two-way methodological bridge between quasi-experimental and structural approaches for robust retrospective merger evaluation For the first bridge, it shows that the difference-in-differences (DiD) specification is the “reduced form” of a structural model, clarifying when DiD may fail to identify merger effects while the structural approach remains reliable For the second bridge, it proposes modifying GMM by weighting each observation’s contribution to the moment conditions using synthetic DiD weights–synthetic GMM to improve estimates when only a few treated markets are available Quasi-Experimental Findings The standard DiD approach with fixed effects suggests all three mergers reduced prices by 4-8% and expanded capacity by 16-29% However, the pro-competitive effects vanish when using the first-difference method The analysis of DL-NW exhibited decreased coordination (−0.141), but this change is statistically insignificant Subsequent mergers increase coordination, with AA-US demonstrating significant competitive deterioration as its conduct increases by 0.107 (at the 5% level) Structural Model Insights The structural framework allows for the decomposition of the price effect into efficiency, market power, and coordination effects The structural approach recovers the same interpretable coefficient while being robust to the fundamental identification challenges that may render the standard DiD estimate unreliable The structural decomposition provides crucial insights that pure reduced-form analysis cannot capture, demonstrating why merger evaluation requires attention to underlying economic mechanisms Synthetic GMM Results The synthetic GMM approach suggests that airline mergers were not pro-competitive as they generated limited efficiency gains while facilitating coordination The UA-CO merger increased competitor coordination and reduced efficiency, resulting in approximately 5% higher prices The AA-US merger shows the largest coordination effects but because the merger-induced efficiency is imprecisely estimated, price changes are also statistically insignificant The structural decomposition provides crucial insights that pure reduced-form analysis cannot capture, demonstrating why merger evaluation requires attention to underlying economic mechanisms The methodological framework provides a foundation for robust merger retrospectives that can inform both academic research and antitrust policy Our structural methodology avoids these limitations by explicitly modeling the conduct parameter and using different identifying variation The sensitivity of DiD estimates to specification choices raises questions about existing merger retrospectives that rely solely on quasi-experimental methods References Abadie, Alberto and Jaume Vives-i-Bastida, “Synthetic Controls in Action,” Working Paper, 2022 Arkhangelsky, Dimitry, Susan Athey, David A.

Ruth: Hirshberg, Guido Wager and Stefan Wager “Synthetic Difference-in-Differences,” American Economic Review Asker, John and Volker Nocke “Collusion, Mergers, and Releated Antitrust Issues,” in Kate Ho, Ali Horta¸csu, and Alessandro Lizzeri eds., Handbook of Industrial Organization Carlton et al., Dennis Mark Israel Ian MacSwain Eugene Orlov “Are Legacy Airline Mergers Pro- or Anti-competitive?

Warren: Evidence from Recent U.S.

Ruth: Airline Mergers” International Journal of Industrial Organization Ciliberto Federico and Jonathan W Williams “Does Multimarket Contact Facilitate Tacit Collusion?

Paper summary: Ruth: So, to summarize what this paper is about, they are bridging quasi-experimental and structural approaches for robust merger evaluation in airline mergers. The main thesis is that apparent price stability in these mergers actually masks offsetting efficiency and conduct effects >

Warren: They claim that the synthetic GMM refinement sharpens the findings, uncovering anti-competitive effects that standard approaches miss when evaluating these mergers >

Iris: They're testing this methodology across three major airline mergers. The structural framework allows them to decompose the price effect into efficiency, market power, and coordination effects which pure reduced-form analysis doesn't capture >

Ruth: And they found that modest efficiency gains are entirely offset by increased coordination across the analyzed mergers >

Warren: Specifically, for one merger, more competitive conduct partly offsets rising costs >

Iris: The synthetic GMM results suggest that airline mergers were not pro-competitive because they generated limited efficiency gains while facilitating coordination >

Conclusion: Ruth: So, looking at the title of this paper and who wrote it, "Bridging Quasi-Experimental and Structural Approaches for Robust Evaluation of US Airline Mergers", what this means is that we need to pay attention to the underlying economic mechanisms when we evaluate mergers >

Warren: It shows that relying only on quasi-experimental methods like DiD can be misleading if you don't account for the structural elements, like coordination and conduct changes >

Iris: The authors' methodological framework provides a foundation for robust merger retrospectives that can inform both academic research and antitrust policy because they explicitly model the conduct parameter >

More episodes

← Home