It is always aesthetically pleasing to see something elegant brought down from an attic, dusted off, and found to be of continuing value as part of the home’s furniture.
Back in 1977, Edward Miller offered a hypothesis about the price move patterns distinctive to newly listed stocks. In “Risk, Uncertainty, and Divergence of Opinion,” Miller suggested that bullish investors, bullish that is in their opinions about a particular IPO, dominate the trading in the aftermath, because short sales constraints limit the ability of the bears to express their contrary views.
On the assumption that the offering price is an appropriate valuation, Miller’s explanation predicts that aftermarket price exceeds that offering price, and that the magnitude of the first-day run up is defined both by the pertinent short sales constraints and by the extent of the divergence of investor opinions.
Another prediction is that after that first day, the newly listed firms underperform, both because investor’s views over time converge and because the constraints on bearishness lessen. For example, the expiration of IPO share lockups increases the share of lendable shares, significantly easing the life of a short seller.
This is an elegant explanation, and it does meet the well-established facts about price moves. But Miller’s hypothesis (MH) fell into limbo for decades in the absence of any direct evidence that short sales constraints do have the hypothesized impact on the post-IPO market. Indeed, there seemed to be contrary evidence. In 2002, Christopher Geczy and two associates published an analysis of the equity lending market finding that most IPO stocks are available to borrow in the wake of the IPO.
MH Went Into the Attic
Other explanations of the first-day run up and subsequent fall-off have subsequently replaced the MH. For example, there is the model of Kevin Rock of the Harvard Business School. Rock proposed that the initial offering price for new shares is at a discount to true value, due to “asymmetric information and rationing.” So the initial run-up isn’t from an equilibrium price toward an unsustainably higher price, as Miller would have it, but from discount toward equilibrium. That leaves us with no explanation for why the beneficiaries of this run-up so often experience a falling-off, though.
Now, in a new study by three scholars affiliated with the University of California at Berkeley, Haas School of Business, MH has been taken down from the attic, dusted off, and given its empirical props.
The three authors are: Panos N. Patatoukas, Associate Professor, Richard G. Sloan, holder of the L.H. Penney Chair in Accounting, and Berkeley student Annika Yu Wang. They developed what they call a “Miller Score” for new issuers, based on “a parsimonious set of pre-IPO characteristics” available in the prospectus. The Miller Score is based on the type of fact about an issuer that tends to lead to divergent opinions on the same within the investment community: investments in intangible assets, for example, and with factors that correlate to short sale constraints, such as offering size. [A large offering size makes borrowing of shares, and thus shorting of shares, easier than a small one, so a smaller offering adds to and larger offering subtracts from the Miller Score. ]
The question then is: does the above described pattern in the stock price after an IPO correlate well with the Miller Score?
A Less Tentative Term
The answer is strongly affirmative. New issuers with a high Miller Score get more positive first-day returns and more negative returns associated with subsequent lockup expirations. The economic magnitude of their results, as they say, is quite striking.
They worked from a sample selection period that began in 2007, the first year for the detailed securities lending data available on a daily basis via Markit Securities Finance Data. The sample of IPOs included 867 issuers, listed on NYSE, NASDAQ, and AMEX between 2007 and 2014. They excluded from this sample new issuers with an offering price below $5 per share, and those for whom certain of the key information used to create the Miller Score is not available.
The evidence from the predictive value of the Miller Score for subsequent price behavior is confirmed by the authors’ look at securities lending data from Markit: stock loan fees, rebate rates, and active supply utilization.
Perhaps the MH should now be accepted under a less tentative term, as the Miller Theorem (MT).