Common Ownership and Competition in the Ready-to-Eat Cereal Industry

An exciting and controversial idea at the intersection of corporate governance, economics, and finance is what is known as the “Common Ownership Hypothesis”. Simply put, it states that because investors hold portfolios of stocks which include horizontal competitors, for managers to do right by their investors, they may want to internalize some of the effects their actions have on their competitors. This is in many ways an old idea, with its roots going back to the 1980’s and the analysis of joint ventures, but there is renewed interest today for several reasons: (a) that investors have become increasingly diversified and hold portfolios that are similar both to the index and to one another; and (b) the largest institutional investors (Vanguard, BlackRock, and State Street) are often among the largest investors in most publicly traded firms—holding 4-6% of most S&P 500 constituents.

While data are generally available on how much investors hold in various stocks via SEC 13f filings, less is known about how corporate governance actually works, and how managers choose which investors to pay attention to (and which to ignore). This is particularly challenging when investors disagree about the direction they wish management to pursue (such as expanding output and reducing price or reducing output and increasing prices). Our previous work examines how to map assumptions on corporate governance into “profit weights” which measure how one firm values the profits of another firm relative to $1.00 of its own profits. To frame ideas, a merger of two firms would be represented by a profit weight of one on a rival’s profits. In that work, we showed that typical S&P500 constituents might value $1.00 of competitor profits as close to $0.20 of their own in 1980, but closer to $0.70 in 2017 given the rise in common ownership. At the same time, we documented substantial asymmetries in these relationships: Kellogg’s might value the profits of General Mills as $0.22 of their own profits, but General Mills might simultaneously treat one dollar of Kellogg’s profits as $0.60 of their own. In some sense these asymmetric relationships are one of the unique predictions of the theory of common ownership.

The most important piece of the puzzle is whether or not these overlapping investor positions actually lead to reduced competition, higher prices and less output in product markets. Most of the early literature has focused on the Modified Herfindahl-Hirschman Index (MHHI) which multiplies the market shares of each firm with one another and the aforementioned profit weights, and then sums them to form a single market-level index. The early literature used regression analysis to test whether or not prices were correlated with this MHHI measure and interpreted a positive relationship as evidence in favor of the “Common Ownership Hypothesis”. An influential paper examining airline prices found that a 1000 point increase in the MHHI measure was associated with a 2-3% increase in fares. When we examined the relationship between prices of ready-to-eat (RTE) breakfast cereals and the MHHI measure we found a 1000 point increase in MHHI was associated with a 3-4% decrease in prices, which is definitely inconsistent with consumer harms from common ownership.

To us, this highlighted one of the main challenges with the MHHI measure, namely that it places strong a priori restrictions on the nature of competition: it assumes that firms sell homogenous products and compete by simultaneously choosing quantities in what is known as Cournot competition. The motivation behind our work was to devise a test for firm conduct (including possible common ownership effects) that allowed for products to be differentiated and for manufacturers to compete by setting prices.

One challenge of our approach is that it requires estimating the degree of substitutability between products in the market. While data intensive, this is an area where economists have made much progress, and these kinds of analyses are now a common input into the evaluation of horizontal mergers. Otherwise, our test is relatively straightforward. We use the estimated patterns in demand to infer elasticities and estimate markups under different assumptions for firm conduct (such as joint profit maximization of all firms, own-profit maximization, and common ownership—a hybrid of the first two). We use data on observed prices and our estimated markups to infer marginal costs. We then compare these marginal costs to variables that should affect costs (the input prices of rice for Rice Krispies, the price of corn for Corn Flakes) and things that should not affect costs, but might affect markups (the income of consumers in a particular store or area, the number of similar sugary kids cereals sold at the same retailer, the inputs costs of rice for products that don’t use it like Wheaties, and so on). One technical challenge is that while we require information on which variables do or do not affect costs, we are agnostic about how different variables affect costs.

In our paper, we find strong evidence in favor of cereal manufacturers maximizing their own profits and largely ignoring the profits of their competitors. There are some nice features of this market that make it amenable to testing: Kellogg’s has a large undiversified shareholder and should behave more competitively than its rivals—instead it has relatively high profit margins. Quaker Oats (a division of Pepsi) should place relatively high weight on competitor profits and our estimates suggest it instead has low prices and profit margins. And finally, Post experiences several shocks to its ownership (it is spun-off, involved in an IPO, and enters and exists various indices) and in general the pricing patterns do not align with these ownership changes.

Our results are obviously not the final word on the common ownership debate. It may be that these effects simply don’t manifest in the prices of RTE breakfast cereals. At the same time, the most important aspect of common ownership is that it is so ubiquitous. Therefore we might ask, if common ownership effects do not show up in the prices of cereal, why not and does this reflect a failure (or success) of corporate governance?

The complete paper is available for download here.