Research

 Digitisation is increasing the frequency at which retailers can adjust their prices, increasing the importance of being able to identify who it is they are monitoring and responding to when they make these adjustments. In this paper I develop a general approach for solving this identification problem. The foundation of the method is a continuous-time, discrete-game model of a retailer’s strategic decision-making. I derive from this structural foundation a ‘reduced-form’ expression for the hazard rate of a retailer’s price adjustments as a function of their competitors’ prices. This hazard rate takes the form of a generalised linear model. Consequently, its estimation can be combined with l1-norm regularisation to exploit the consistent model-selection properties of the LASSO and identify the competitors whose prices define the payoff-relevant states of the retailer. Further, in implementing the method we can exploit the simplicity and efficiency of standard, highly-optimised machine learning packages. I demonstrate the method by using 30 months of minute-to-minute price data to estimate the competitive ties between gasoline stations on the southern periphery of Sydney, Australia. I find these ties connect all the stations into a single networked competition structure, which spans several geographical areas that have previously been treated as separate markets by the local competition authority in its investigations and merger authorisations.

This paper introduces a model of networked price competition and its emergence from the choices of consumers. The model micro-founds networked competition in the choices of consumers who consider heterogeneous sets of alternatives and have convex preferences over them. The demand emerging from these choices features sparse substitution, connecting firms in a network, and I prove that price competition over this network has a unique equilibrium in pure strategies. The model itself is composed of a novel combination of three classic oligopoly-modelling elements – a partitioned unit mass of consumers, heterogeneous consideration sets, and linear-quadratic utility functions. As a result, the model reveals how characterisations of price competition in classic oligopoly models are limit cases of networked competition. By analysing the model further, I connect market-level outcomes to a local measure of competition between pairs of firms: diversion ratios. Diversion ratios measure the competitive threat imposed by one firm on another, and I show that the centrality of firms in a network with diversion-ratio ties provides a direct measure of their pricing power. Thus I present a way to take a commonly-used empirical measure of pairwise competition and aggregate it to measure the constraints on firms’ price setting provided by both directly- and indirectly-connected competitors.

We investigate whether competition between the fund companies that offer mutual funds constrains individual fund fees. We document that over half of individual fund fee variation is explained by company-wide components. Moreover, we show using SEC prospectus download data that company-level attributes influence investors' consideration of companies. We connect these facts with a model of fee competition between co-considered fund companies, characterising the competitive landscape and associated equilibrium fees. Calibrating the model, we derive a testable prediction for competitively constrained fees. The prediction successfully explains cross-sectional variation in the company fee components, identifying the influence of company competition on fees.