Brad DeLong catches James “DOW 36000” Glassman complaining about Paul Krugman being a stalker:
“By the way, none of the tumultuous events of the past six years has changed our minds about our thesis. In fact, despite terrorist attacks and a recession, price-to-earnings ratios have remained high, in historic terms, just as we predicted…” We all know that Dow 36000 predicted not that price-earnings ratios would remain at their ca. 1999 levels, but would triple over the next three to five years–i.e., by 2002-2004.
Were Hassett & Glassman claiming that the price-earnings ratio would triple so that it would be near 100 or were they trying to claim that earnings would triple with an unchanged price-earnings ratio?
Krugman has consistently misrepresented (i.e., lied about) the basic argument of the book – the notion that investors are solving what economists have long called the “equity premium puzzle” – the mystery of why stocks return so much more than bonds even though both are roughly equal in risk over long periods of time. By the way, none of the tumultuous events of the past six years has changed our minds about our thesis. In fact, despite terrorist attacks and a recession, price-to-earnings ratios have remained high, in historic terms, just as we predicted. I’ve gotten used to the misrepresentations of “Dow 36,000.” It was, I admit, a flamboyant title for a book with a fairly conservative proposition, and perhaps Kevin and I invited criticism from folks who either read no more than the title and formed their own conclusions or who read the whole book but decided to willfully ignore what it said. (Kevin once jokingly suggested that we call the book, “A Treatise on the Declining Equity Risk Premium.” Not a bad idea, perhaps.)
I hope the following does not misrepresent what Glassman was trying to say. If earnings had tripled and the price-earnings ratio had stayed constant, then perhaps a prediction that the DOW would reach 36000 might make sense from a pure simple multiplication perspective. But one would have to make two questionable assumptions: (1) that earnings were expected to rise by 25% per year for five years; and (2) that one’s valuation model would somehow predict that the price-earnings ratio would somehow remain near 33 by the end of this five-year period.
In thinking about this issue, I consulted the webpage of Aswath Damodaran to find his excel file labeled “Implied Equity Risk Premiums – United States”. The purpose of Damodaran’s analysis of stock market data was to provide an estimate of the equity risk premium provided by the relationship between stock valuations, earnings, and expected earnings growth. While he does not provide the price-earnings ratio over time, the following diagram does – using his data. Earnings in 2004 were about 48% greater than they were in 1999 and yet the value of the S&P has declined by 17.5%. Simply put, the price to earnings ratio in 2004 is far below what it was in 1999.
Would one have expected this ratio to decline as much it did over the past five years? While interest rates on Federal bonds (one measure of the risk-free rate) have declined, Damodaran is suggesting that the overall cost of capital has not changed as his estimate of the risk premium has actually increased. Of course, his estimate depends in part on the assumed expected earnings growth rate. Notice that in 1999, market participants were already forecasting significant medium term earnings growth, which is one explanation of the high price to earnings ratio in 1999.
Now one might wish to suggest that Damodaran’s estimate of the risk premium is not quite right and that it has not increased since 1999. But this challenge would be based on the premise that his estimate of current expected earnings growth is too aggressive. In other words, one would have to suggest that the decline in expected earnings growth has been more than Damodaran suggests. But shouldn’t that also tell Glassman that his presumption of a constant price to earnings ratio is even more flawed?
Unless Glassman wants us to believe that the steady-state growth rate is that high, it’s not clear why he expected the price to earnings ratio to remain constant. But even if he did believe high earnings growth would persist, he has a wee bit of a problem. Rapid earnings growth would require investment in new tangible assets reducing the dividend payout ratio. But then I’m only rehashing Andrew Samwick’s contribution to that old Baker-DeLong-Krugman discussion. Alas, Glassman must not have paid very careful attention to the debate and its implications for stock valuations.