©2005 All content on site protected by copyright
Could ROE Provide for A Meal for a Lifetime? by Victor Niederhoffer
Fundamental relations. The growth of a company book value is a function of its rate of return on equity (ROE) and its payout ratio. For those companies with high ROE, growth should be higher, holding the payout ratio constant. One would think that over time, differences in the ROE might be translated into returns to stockholders. This relation might be expected to vary over time, industry and stage of the economic cycle. It might determine the prospective returns to stockholders of public offerings . Retailers with high returns on equity might be particularly attractive as their growth could be exponentiated up or down depending on the ease of raising capital and the infrastructure necessary to provide scalable growth.
At a more general level, the relation between ROE and returns to stockholders might be expected to vary over the economic cycle with reasonable leads and lags. It could provide a model for which industries to invest in and which to avoid, or when growth companies are moving into and out of favor versus value companies. I hypothesize that the relation might also be predictive of the overall health and prospective returns of the stock market.
One wonders what other factors besides industry -- for example, the price level of a company, its debt equity ratio, its rate of return on capital or its beta -- might affect the relations posited above. Also, whether the most fruitful avenues of study here are how the rates of return vary for an individual company over time, or how the relations hold on a cross-section of companies at a moment in time. How do predicted returns on equity correlate with performance from standard sources such as Value Line, Zacks and IBES? To what extent should the relations be normalized by the existing price-earnings ratio and interest-rate environment in the economy or the company? How relevant are the finding in the empirical finance literature on this subject, and how out of date are they? The book "No Monkey Business" by Stuart Fowler seems to have some good work on this.
I would not be surprised if my colleagues and I were to drill down on queries similar to the above. Assuming we can slip the results past the Minister, something as difficult as slipping a lamb past a wolf (when he's not on the courts), you will doubtless be hearing from us further on this subject, and I encourage your own insights on this seemingly fruitful subject.
From Prof. Charles Pennington:
The Minister will let this lamb slip by!
One can sort the 500 S&P stocks into quintiles based on their trailing return on capital, updated quarterly (and with the list of S&P components updated annually). Then one can start with $1 invested in each of the 5 quintiles, with quarterly re-allocations.
I'm finding that the $1 would grow to the following values, depending on whether invested in quintile 1 (highest return on capital), quintile 2, 3, 4, or 5 (lowest return on capital):
1 (highest return on capital) $4.12 2 $4.33 3 $3.10 4 $3.78 5 $3.65
Totally consistent with randomness, just the way I like it.
The recent book by Greenblatt advocates buying stocks with a combination of high return on capital and high earnings-to-price (or low P/E), and his back tests show that that strategy is a winner. He doesn't say though how much of the effect comes from the return on capital and how much from the earnings-to-price.
So a brute force ranking by return on capital doesn't add much. The Chair's more elaborate ideas on this topic may however have much value, and therefore I don't feel they're suitable for discussion.
I would think that the high ROE portfolio would need to be adjusted for leverage of the underlying companies. No point in having the ROE filter include companies capitalized with 99% debt. Even if those firms generate a nice ROE, once the equity builds up, the ROE declines markedly.
One of the Merrill Lyn#h strategists disseminated in the mid '90s a piece of proprietary research showing that a portfolio of high ROE companies outperformed each of the other dozen or so strategies that they followed, including value, revenue growth, price/sales, price/book, etc.
Stern Stewart & Co. has made a nice franchise for themselves using ROE as the main component for their Economic Value Added analysis and research. Essentially, they look at firms whose unlevered ROE (based on unlevered net operating profit after tax, or UNOPAT) compares favorably to their weighted average cost of capital. They claim good historic investment returns, and they also apparently make quite a bit of consulting income to securities issuers as well.
Sushil Kedia responds:
The problems of finding a definable way of inflation accounting as pointed by Prof. Haave and of ignoring leverage undoubtedly add to the randomness observation.
Also, the rates of depreciation admissible under the Taxation laws (which help postponements of taxes really and nothing else) are often more aggressive than rates of depreciation allowed in laws governing Corporate reporting. This creates a set of companies with high / rising capex that add to bloating of reported earnings in nearer time periods on a earnings forecast series, and thus another variable that adds to consistent with randomness observed possibly.
EBITDA based ratios as a surrogate for measuring efficiency of capital usage and specifically EBITDA by Networth ratio provide a better feel, but I would guess it too is likely to be unable to escape definitions of consistent with randomness, even after being adjusted for payout ratios. Markets are pricing in management capability, signalling effects and such other qualitative variables fairly often.
To turn the crystal ball around, I have been playing with plugging in the value of the current prices (market cap) as being equal to the Discounted Cash Flow value of a firm. Simply the market implied growth rate that jumps up on the sheet with each change in prices and its ongoing comparison with the historical growth rates of returns and finding deviations does provide clues to overpricing. [a fairly simple bloomberg plugin sheet that assembled for this objective and containing codes of Indian companies only, modifiable by replacing with codes of stocks you want to check will be mailed off-list upon requests].
Financial stocks where money is the raw material, end-product and yardstick of performance measure seem to provide better feeling numbers when gauged with the reverse dividend discount model as opposed to the reverse DCF for most others.
Just felt like putting up some extenstions to Chair's core theme while good questions are brewing up in stronger minds.
May be, tuning up various permutations of extended ratios as the RoE but applied to specifically similar stocks (not necessarily from similar sectors) would raise the chances of finding workable relationships of ranking stocks.