We interview Jean-François “JF” Houde, professor at the University of Wisconsin–Madison, to learn about his research on the economics of collusion and competition in mortgage markets. This interview presents his perspectives on why cartels form along the supply chain, how cartels of unequal firms hold together, and how economists distinguish collusion from firms simply following a market leader.
1. You have spent much of your career examining how firms compete—or fail to compete—in complex markets. To begin, could you provide an overview of your research on collusion and cartels? Has there been a unifying theme to your approach to this topic?
I approach my research, on cartels included, by trying to understand what economics can teach us about how firms and markets work. The link between economic theory and what firms do in practice has always fascinated me, along with analyzing how firms compete.
Collusion makes that challenging. Economic theory tells us less about cartels than you might expect. It gives us broad strokes, but it does not necessarily explain whether a given arrangement makes sense. What it does provide is insight into which practices or institutions can facilitate collusion, and how firms adjust their behavior to sustain an agreement.
Every cartel case I study raises the same two questions: why these firms acted as they did, and does economic theory explain it? When the answers line up, we understand not only that firms colluded, but how the arrangement held together.
My work has focused on known cartels that have been prosecuted or alleged, where there is fairly concrete evidence of collusion. My colleagues and I examine the case and evaluate what makes collusion easier or harder in that setting, drawing on court documents. In my work on gasoline markets, for example, we obtained transcripts of wiretapped conversations from the court, which we codified and then combined with pricing and quantity data to understand how firms were coordinating.
My coauthored paper on the Canadian grocery industry, “Hub-and-Spoke Cartels: Theory and Evidence from the Grocery Industry,” published in the American Economic Review, took a similar approach. From a theory point of view, the case was interesting because hub-and-spoke cartels can be difficult to understand under standard models of competition, given that firms are colluding along the supply chain. A retailer should not want its suppliers to collude, because it will pay more for their products, and a supplier should not want retailers to collude, because they will sell fewer of its products. In this case, we found that specific contracts incentivized firms to form vertical cartels.
Every cartel case I study raises the same two questions: why these firms acted as they did, and does economic theory explain it? When the answers line up, we understand not only that firms colluded, but how the arrangement held together.
2. Collusion in gasoline markets is another focus of your research, where you have examined individual cartels case by case. Tell us about a challenge that those cartels faced, how they overcame it, and what that reveals about how cartels hold together.
Gasoline markets were my first research love, and I have written several papers about them. In “Collusion with Asymmetric Retailers: Evidence from a Gasoline Price-Fixing Case,” published in the American Economic Journal: Microeconomics, my coauthor and I examined a cartel in Quebec, Canada. The main challenge faced by the cartel we studied was that the firms were asymmetric. That is true in most markets, but gasoline is an especially stark case because consumers are so price-sensitive.
A typical gasoline market has a big-box retailer selling gas, an independent gas station that might also be a distributor, and a chain with two or three convenience stores. Each entity has different costs and incentives, so each would prefer a different price. A cartel, though, must hold one price, which tempts the low-cost members to undercut competitors and makes it hard for the others to tell who is cheating. Nearly all retail cartels face this problem.
We analyzed how the Quebec cartel got around it: they agreed on the timing of price moves. The court documents showed this mechanism in detail. The Canadian Competition Bureau had posted hundreds of pages of extracts from the wiretapped conversations, and for each price increase you could see the sequence. The cartel leader raised its price first, then told followers to raise theirs at a set time, so most stations moved together. Last came the low-cost firms, the big-box retailers, which were allowed to raise prices later in the day.
That delay is a transfer, as in manufacturing. Because consumers are so price-sensitive, lines form at the station that still has the low price, and the low-cost firm gains market share in the meantime. That gives it a reason to stay inside the agreement.
A cartel of unequal firms holds together only if it compensates its low-cost members and contains the firms outside it. This can generate patterns in price data that are best explained by the particular needs and characteristics of a cartel––a fact that can be useful in evaluating economic evidence of alleged collusion.
3. A central challenge in antitrust is telling the difference between firms that have agreed to coordinate and firms that are simply following a market leader. How does economic modeling help separate the two?
Economists are often asked to evaluate allegations of collusion when the question of the existence and participation in a collusive agreement is not settled. In those cases, we turn to economic theory.
Whether collusion is tacit or explicit, a common approach is to examine whether firms are leaving money on the table. A colluding firm is passing up profit: given its rivals’ prices, it could lower its own price, gain market share, and earn more, but it does not. In a gasoline market, for example, you can estimate demand, show that a firm could profitably undercut its rivals, and quantify the market share it gives up by following the leader instead.
That approach requires detailed data. When the data are not available, an alternative is a natural experiment: a shock to the market where theory predicts how independent firms would respond. Analyzing the observed response can shed light on whether these firms follow unilateral incentives or behave in a way that is consistent with the alleged collusive scheme.
Whether collusion is tacit or explicit, a common approach is to examine whether firms are leaving money on the table. A colluding firm is passing up profit: given its rivals’ prices, it could lower its own price, gain market share, and earn more, but it does not.
My research on a hub-and-spoke cartel in the Canadian grocery industry is an example. A challenge here was to use industry data to determine which, if any, retailers were part of the arrangement.
One way in which we approached this problem was by looking at increases in the wholesale price of bread. On certain dates, manufacturers proposed common wholesale increases of $0.07 per loaf, and retail prices were expected to increase by $0.10. Retailers are very different from one another, and under almost any economic model a large discount grocery chain would not pass through a cost increase the same way a premium grocer would. If retailers were acting independently, their responses should have varied. Instead, retail prices clustered at exactly ten cents per loaf, and within each market, low-price stores raised prices by roughly the same amount as high-price stores. That uniformity, a form of parallel conduct, was more easily understood as part of the cartel than as the result of independent profit maximization by the grocery stores.
4. As an economist who bridges the gap between theory and data, what do you see as the most critical contribution an expert witness can make in a cartel case to help a judge or jury understand the dynamics of a specific industry?
Two elements are at the heart of the economics of collusion: first, that the firms gave up profits they could have earned by deviating from the alleged agreement; and second, that a credible threat of retaliation kept them from doing so. An economist can help establish whether the empirical evidence is consistent with these two elements.
The first is the point I made above. A firm that is colluding gives up profit it could have earned by cutting its price, so the first task is to identify whether such an opportunity existed and the extent to which the firm took it. If undercutting competitors is not associated with higher profits, it is hard to argue that the firm was colluding. Economists look for this sort of evidence in a variety of ways, including through demand estimation or natural experiments.
The second is to assess whether the cartel monitors and retaliates, or threatens to retaliate, against firms that might deviate from the alleged agreement. A cartel is sustainable only if each firm believes that cheating will cost it more than it gains. In the gasoline cartels I have analyzed in my research, the threat is visible in the data: you see price wars and periods of below-cost pricing. In the grocery case I referenced earlier, firms enforced the agreement by threatening to remove products from store shelves, and the record shows trade disputes of that kind.
5. You have published several studies of competition in mortgage markets. What sets these markets apart from more traditional consumer products and how have those features shaped your understanding of competition in mortgage, and by extension, real estate markets?
My research has focused on mortgage markets, and many of the same specificities in these markets apply more broadly to other consumer finance, insurance, and real estate markets. The defining feature of both areas is that the cost of serving one consumer differs from the cost of serving the next.
Default risk is the obvious example in mortgage markets, yet lenders also worry about prepayment risk: how long will you stay in your house, or stay with the lender? Moreover, a borrower who buys several services from the same lender is cheaper to serve than one who does not. In a grocery store, every shopper costs roughly the same to serve. In a mortgage market, no two borrowers do.
This variation creates two problems for consumers. First, it is hard to know when you are getting a good deal. A rate that looks attractive may still be well above what a comparable borrower is paying, and because every borrower is priced differently, there is no public benchmark to check against. Second, without a benchmark, generating competition is costly. Prices are not posted, so consumers must shop around and know how to compare offers. Consumers influence the intensity of competition through their own time and effort, and the savings need to be large relative to that cost.
My research has focused on mortgage markets, and many of the same specificities in these markets apply to more broadly to other consumer finance, insurance, and real estate markets. The defining feature of both areas is that the cost of serving one consumer differs from the cost of serving the next.
Individualized pricing is common in real estate as well, though for a different reason: the assets are differentiated. In finance, a loan from one lender is much like a loan from another; the differentiation is in the borrowers themselves. In both settings, though, haggling is the norm. The posted price is not a good indicator of the price you could pay if you searched, and that creates a large friction in the market.
That friction has consequences. One we have studied is the role of intermediaries. Wherever search costs are large, you see brokers, mortgage specialists, and realtors. Intermediation can reduce these frictions, but intermediaries have their own incentives and do not necessarily act in the consumer’s best interest. My coauthors and I examined those agency problems in “The Role of Intermediaries in Selection Markets: Evidence from Mortgage Lending,” published in the Review of Financial Studies.
A second consequence is important for antitrust: haggling and search costs make mergers harder to evaluate. In “The Effect of Mergers in Search Markets: Evidence from the Canadian Mortgage Industry,” published in the American Economic Review, we studied a merger between two banks. The standard approach compares average prices before and after the merger. In mortgage markets, though, most consumers do not search beyond one lender. They were not benefiting from competition before the merger, so the merger did not hurt them, and the average masked its effect. We found that the average understated the true rise in market power. Consumers who felt the merger were the ones who had searched and collected more than one or two quotes.
5 Questions is a periodic feature produced by Cornerstone Research, which asks our affiliated experts and senior professionals to answer five questions related to their work.