Posts filed under “Data Analysis”
What can we conclude about future market performance based on how successful or not IPOs are?
Mark Hulbert looks at that question in a column in the Sunday NYT. He reviews a study that suggests all the recent IPO postponements/cancellations are "an indication of investor pessimism that may actually turn out to be bullish for the overall stock market."
"Worldwide in June, more companies withdrew or postponed their initial
public offerings than in any other month since March 2001, according to
Dealogic, a firm based in London that monitors the new-issue market.
That March 2001 trough came less than halfway through the 2000-2 bear
market, leading many investors to worry that the current gloom in the
new-issues market is a harbinger of much lower prices for stocks.
But the stock market’s continuing decline in the months after the March 2001 I.P.O. bust was probably an anomaly, says Jay R. Ritter, a finance professor at the University of Florida who specializes in I.P.O. research.
An analysis of initial offerings market since 1980 suggests that, all else being equal over the next 12 months, the market between now and the summer of 2007 is likely to produce above-average returns."
I am compelled to point to a few issues with Prof Ritter’s methodology. First, he analyzed the IPO market from 1980 – 2000. That period includes all of the 1982-2000 secular bull market and the 2 years that preceded it. It would be more representative of broader market history if the Prof had reviewed the data going back to 1966. That way, he would have had both a secular bull and bear market.
Secondly, Prof Ritter ignored the period that seems to not support his conclusion. If we are to consider the 2001 period an "anomaly," and disregard the parallel between then and now, than we must also consider whether the 1980 to 2000 period is anomalous as well. At the very least, we have to determine if it is representative of broader market history.
That time period — the strongest bull market in history — is suspect to use as the sole basis of comparison. Including a comparable bear period would likely improve the study’s results.
All historical comparisons have flaws, but the more parallels you can draw between time periods for as many different variables, the more informative the comparison can be.
graphic courtesy of NYT
Its an interesting idea to use IPO pricing and other data points as the basis for forecasting. A broader historical period might have produced a data set that wecould have more confidence in.
If Initial Offerings Fall Short, Start Looking for Bulls
NYTimes, July 30, 2006