An IPO’s Impact on Rival Firms

An initial public offering (IPO) is a major event in the life of any firm. But what does an IPO imply for the industry’s future? In our paper, An IPO’s Impact on Rival Firms, which was recently made publicly available on SSRN, we take a structural approach that allows different industries to progress in different ways post IPO. If one is forced to make a sweeping generalization, then this paper finds an IPO augurs in an era of reduced profits and greater consumer mobility within an industry. Unlike a static model, a structural model’s parameters produce implications about magnitudes rather than just signs. This permits one to assess whether the estimates are economically “reasonable in a straightforward manner.”

We begin by modifying the Spiegel and Tookes (2013) continuous time model of competition in a heterogeneous product oligopoly. (A step-by-step guide detailing the model’s solution can be found here.) Our variant generally does well at capturing changes to a firm’s profits and values after a rival’s IPO. Using 3 years of data it yields an in-sample R² statistic of nearly 17% and with 5 years of data 34%. By contrast, recent purely empirical models report in-sample R² statistics of approximately 4%. The structural model’s fit comes about despite the fact that it requires only four estimated industry parameters and three firm-specific parameters. When compared with the results typically seen in empirical corporate finance, where far more independent variables are used, the fit produced here is quite good.

Overall, the model indicates that an IPO is generally bad news for rivals’ future profits per unit of market share. Depending on the estimation window used the median industry sees a long-term drop of between 10% and 25%. However, the estimated heterogeneity across industries is quite large with an interquartile range between -60% and +40%. If forced to provide a broad characterization of what happens, the hypothesis that the information released from an IPO leads firms to a more homogenous form of product competition (and thus lower profits per unit sold) appears to dominate. The parameter estimates indicate that post-IPO it becomes 3 to 4 times easier to lure away a rival’s customers. An example of this type of market evolution can be seen in the cell phone industry. As a number of articles have noted, unit sales are up but profits are down. The generally accepted reason is that, as time has passed; the product offerings have become more homogenous, increasing price pressure.

In the case of IPOs it is particularly important to retain some perspective on what is occurring and what role the IPO firm may be playing. When firms go public they are typically very small, with market shares of well under 1%. Furthermore, they remain small with most seeing changes in their absolute market share of less than 1% in the subsequent 3 years. In line with this, the interquartile range of industry value changes in our sample ranges from -4% to +3% over the long run post-IPO. This contrasts with some other papers on how IPOs impact rival firms that find strikingly large affects.

While any model can potentially be judged by whether or not its results are reasonable in magnitude; reasonable may lie in the eye of the beholder. A more concrete test is to determine whether or not the model’s estimated parameters tell us anything useful about how an industry evolves post-IPO. That is, can the model forecast industry dynamics out of sample? The short answer, here, is yes. Our tests also show that forecasting with the model and its estimated parameters does a better job out-of-sample than other empirical variables that have been used in the IPO literature.

Forecasts not only test a model against the data, but also offer a window into causality. In general, good competitive news for one firm should be bad news for its competitors and vice versa. For example, if going public makes a firm stronger, its forecasted profits per unit of market share should increase and its rivals’ profitability should decrease. Alternatively, if the IPO leads to the transmission of formerly private information useful to the firm’s rivals, then the opposite should be true. Instead, the paper finds that estimated parameter changes pre- and post-IPO look similar for both the newly public firm and its rivals, indicating the IPO is best described as a “canary in the coal mine” rather than a causal competitive event. Another way to test for causality is to look out of sample. If the IPO causes future changes to the issuer’s competitive prospects its forecast errors should be negatively related to those of its rivals. However, we find that this correlation is positive, providing further evidence that an IPO presages events rather than causes them.

To our knowledge, this is the first paper to employ structural parameter change estimates and forecasts as a way to test an event’s causal relationship to future changes within an industry. One can think of these tests as structural model analogs to a differences-in-differences (DID) analysis. In a DID test the target or event firm is matched another firm that has similar characteristics along a few dimensions but is not associated with the event. Matching criteria often include selecting from the same industry as the target firm. Unfortunately, matches within 4-digit industries are often impossible frequently leading to the use of 2-digit industry groups. However, 2-digit industries are rather broad. If the event in question arises from industry specific changes among firms competing with each other, then a match at the 2-digit industry level may not correct for that. The structural test proposed here has the potential advantage of using the firm’s own industry as its benchmark. Changes within the set of competitors are thus picked up which can help address the concern that observed effects are due to changes in a specific industry rather than from the event being studied.

The full paper is available for download here.