The Art of Streetplay

Tuesday, June 27, 2006

Poking Holes in Bogle's Pro-Cap Weighting Rationale

Surprise surprise... John Bogle is putting down fundamental indexing in favor of, you guessed it, what has made him rich-- cap weighted passive indexing. And he got Burton Malkiel to back him up and give him a sense of credibility-- not too different from WisdomTree getting Siegel on board. Must be right if they've got an academic on board!

I thought I'd hone in on a few aspects of his argument and then share some personal thoughts.

Aspect #1:
"First let us put to rest the canard that the remarkable success of traditional market-weighted indexing rests on the notion that markets must be efficient. Even if our stock markets were inefficient, capitalization-weighted indexing would still be -- must be -- an optimal investment strategy. All the stocks in the market must be held by someone. Thus, investors as a whole must earn the market return when that return is measured by a capitalization-weighted total stock market index. We can not live in Garrison Keillor's Lake Wobegon, where all the children are above average. For every investor who outperforms the market, there must be another investor who underperforms. Beating the market, in principle, must be a zero-sum game."

This does make some sense-- it is true that (1) the market return from one instant to the next is, technically, the return generated by a capitalization-weighted total stock market index. It is also true that (2) beating the "market" is a zero-sum game. If one makes the claim (as they did) that the market is inefficient, though, the flaw in the logic above is that claims (1) and (2) imply the market must be the an optimal investment strategy. If, for whatever reason, there are at times deviations from intrinsic value, and one is able to, probabilistically or otherwise, construct a strategy which takes advantage of mean reversion of stocks to their intrinsic value over time, then one could theoretically outperform the market (with a few more conditions). Because beating the market return is zero-sum, yes, you would be earning a profit at the expense of another market participant, and yes, it is notoriously difficult to beat the market after fees over time. The only thing I am saying here is it is a logical fallacy to say that given (1) and (2) are true, then even if the market is inefficient, cap weighting must be an optimal investment strategy-- it is NOT a logical fallacy, however, to make the deduction that cap weighting MAY be an optimal investment strategy.

Aspect #2: Expenses-- Management Fees, Turnover, Taxes (Capacity)
"Purveyors of fundamentally weighted indexes also tend to charge management fees well above the typical index fund. While index funds also incur expenses, they are available at costs below 10 basis points. The expense ratios of publicly available fundamental index funds range from an average of 0.49% (plus brokerage commissions) to 1.14% (plus a 3.75% sales load), plus an undisclosed amount of portfolio turnover costs."

"
Fundamental weighting also fails to provide the tax efficiency of market weighting."

Later in the article they delve into some of the conditions which would allow one to make the claim that investing in cap weighted indexes is the optimal investment strategy. First off I would say that again, they are definitely right. Cap weighting has a bunch of natural advantages. They are easy to construct, they require no turnover and hence no transaction cost and no manager who takes a fee for himself, and they are tax efficient.

The question I ask again is, does this lead to the natural deduction that cap weighting MUST be the optimal strategy? Perhaps I'm wrong, but I don't believe so. What it implies to me is the crux of the argument for active management-- as one deviates from investing in the market to increase your portfolio's allocation to a security you believe to be mispriced, you run up against a number of frictions: (1) all the time you are spending to figure out whether that security is, in fact, mispriced-- the cost of time (which you may or may not outsource to a money manager for a fee); (2) transaction costs to invest in that security; (3) tax inefficiencies; (4) other (ie. market impact, etc). These are real costs.

So the fundamental question is: in the face of probabilistic inefficiency (which is all Arnott and Siegel claim), is "market noise" of a large enough magnitude and does it mean revert quickly enough for it to be worthwhile to incur the incremental costs necessary to generate those returns?

I am not saying anything which hasn't been said a million times in all likelihood. Arnott's paper in the FAJ allocated large chunks of space to the time series of the returns relative to the market returns, how "market-like" those returns were, what capacity was available to the trading strategy, the effect of the trading strategy on volatility, and the incremental costs involved assuming turnover at a certain rate.

In other words, he was making an apples to apples after-transaction-cost comparison between his strategy and the market return, and he found that his trading strategy outperformed over time robustly enough that the probability of overfitting was minimal. That is pretty valid-- this article applied no such rigor. I don't blame it (this is the WSJ we're talking about), but nevertheless, it does not conclusively disprove the assertions made in Arnott's paper.

Aspect #3: Increase in Efficiency
"We concede that there is some evidence, based on numbers compiled by Ibbotson Associates, that long-run excess returns have been earned from dividend-paying, "value" and small-cap stocks -- albeit returns that are overstated by not taking into account management fees, operating expenses, turnover costs and taxes. But to the extent that investors are persuaded by these data, the premiums offered by such stocks may well now have been "arbitraged away" in the stock market, as price-earnings multiples have become extremely compressed."

This is a valid point, not out of line with the logic in the fundamental question: in the face of probabilistic inefficiency (which is all Arnott and Siegel claim), is "market noise" of a large enough magnitude and does it mean revert quickly enough for it to be worthwhile to incur the incremental costs necessary to generate those returns?

It isn't enough to make the claim that the market is noisy-- the noise must be large enough and mean reverting enough. If, through the influx of hedge fund investment and everything else, the market is more efficient than it was, then perhaps even though a risk-reward favorable trading strategy existed in the past, we wouldn't see returns nearly as large going forward-- especially after fees.

I have commented on the outperformance of value of late in Commentary on The Trouble With Value, based on GMO's piece a while back. It is true- the run value has had of late is now getting long in the tooth. The outperformance of value relative to growth has averaged out to a certain level over time, and we are now well above that average. Reversion to the mean would imply value may have a more difficult time going forward.

Claim #4: New paradigms don't tend to last

"We never know when reversion to the mean will come to the various sectors of the stock market, but we do know that such changes in style invariably occur. Before we too easily accept that fundamental indexing -- relying on style tilts toward dividends, "value" and smallness -- is the "new paradigm," we need a longer sense of history, as well as an appreciation that capitalization-weighted indexing does not depend on efficient markets for its usefulness.

While we have witnessed many "new paradigms" over the years, none have persisted. The "concept" stocks of the Go-Go years in the 1960s came, and went. So did the "Nifty Fifty" era that soon followed. The "January Effect" of small-cap superiority came, and went. Option-income funds and "Government Plus" funds came, and went. High-tech stocks and "new economy" funds came as well, and the survivors remain far below their peaks. Intelligent investors should approach with extreme caution any claim that a "new paradigm" is here to stay. That's not the way financial markets work."

This is theoretically a slightly different, broader slant from Aspect #3. This isn't necessarily making the claim that markets have secularly gotten more efficient in general. It is more making the claim that over time, two things tend to happen at different points in time for a variety of reasons-- (1) fads develop and (2) systematic statistical patterns form. Neither persist over time, and it is so unlikely that you will be able to know when they pop or decrease to statistical insignificance that it isn't worth the costs necessary to act on the information. They give a bunch of examples of (1). I would posit an example of (2) to be the incredibly large serial autocorrelation detected in the market indices by Andy Lo. It really did exist. They pointed it out, people got excited and probably traded on it a bunch, probably made some good money, and then over time, it decreased in absolute value to the point that it is no longer profitable to trade on relative to sticking all that money in an index fund.

Fair enough. What is the "truth"? Will the magnitude of this aberration decrease over time, as more people catch on, to the point that it isn't worth the incremental costs, or not? You know, they very well could be right.

Personal Thoughts
My belief is that this is less a statistical anomaly than the Andy Lo autocorrelation phenomenon was. It is lower resolution, it takes longer to realize the abnormally positive returns, it requires patience and a smidge of contrarianism. These are all qualities which would allow it to persist longer than other phenomena would. The big irony of it all, however, is this-- if Siegel is able to convince enough investors that he is, in fact, on the right side of this debate, the influx of capital may itself cause the anomaly to disappear! This stems from one unquestionable truth-- beating the market return is a zero sum game, and the market return is the return on the cap weighted total stock market index. If the majority of investors believe they will beat the market return by investing in fundamental indexing, they will have to earn their above market return at the expense of other market participants-- but those market participants aren't anywhere to be had. Those abnormal returns exist because the "market" has allocated funds in a particular way over the history of the stock market. If the "market" were to no longer allocate funds that way, perhaps we would have the indirect benefit of an overall better functioning economic system, but directly, the market, as a whole, cannot escape the market return. If everyone believes something to be true, you cannot earn abnormal returns off of it.

The other aspect which I personally grapple with is Aspect #3. As trite as it may sound, we have seen a hell of a rally in value relative to growth. The outperformance of value is now above the mean. Has the influx of capital to professional money managers made the pricing of stocks relative to each other more efficient? If so, the returns of an investment strategy which worked when the investing landscape was not riddled with value managers may not be applicable to the world we will see over the next 20 years.

I no doubt believe that the market is noisy, as Siegel puts it. But that alone is not enough to make fundamental indexation "work". For it to "work", there needs to be sufficient noise and mean reversion to compensate for the costs incurred. From the point of view of someone today investing over the next 10 years, that is a difficult tradeoff for anyone to say definitively will go one way or the other, in my opinion.

Money Managers Have a Place in This World
A final thought is in order on the topic of market efficiency, and professional money managers. They do have a place in this world. Just think about it: if all money was invested in index funds, who would set the value of the individual stocks which comprise the S&P? We need stock pickers! More than that, they deserve compensation for providing efficiency to the price of stocks! Without individuals estimating the intrinsic value of stocks, the market system breaks down, because its whole purpose in the paradigm of the financial markets is to allow companies to raise capital efficiently. If they did not do that, there would be no need for the stock market at all.

The question is not whether they should exist or not-- the question is what is the just compensation they deserve relative to the amount of efficiency they can provide to the market.
With all this talk of index investing, I get a good feeling inside knowing I might have a place in this world-- as an allocator of efficiency capital. Great.

Tuesday, June 20, 2006

WisdomTree Update, June 20th 2006

Needless to say, a lot has happened since my last post, and since I first started writing about WisdomTree in April 2005. The 20 ETF's have officially launched. All trade on the NYSE under a variety of tickers- DTN, DLN, ..., all of which are listed on the up and running website they now have. One of the commenters on this blog completely nailed the launch date. They have brought on board yet another BGI veteran, Bruce Lavine.

Rather than spell out everything that is easily available to the public, it might be of value to analyze what is going on one level deeper.

(1) WSDT is leveraging its star power and university environment very effectively. It has done this in 3 ways- 1) it has obtained ETF's licenses much more quickly than I thought would be possible, 2) it has gotten discount or free advertising all over the WSJ, CNBC, on the floor of the NYSE, and elsewhere, and 3) it has gotten heavily discounted research and development aid from students at the University of Pennsylvania and Wharton, through a class offered called the Wharton Field Challenge.

This cost structure really doesn't need much capex at all to fuel itself. The management team is also most likely compensating itself in a call option-type fashion than anything else. We will look at the economics later.

(2) The Expense Ratios seem low to me. The expense ratios range from 28 to 58 bps, but the "bread and butter" fund, in my opinion, seems to be the Total Dividend Fund which charges 28, and DIEFA, which charges 48. The rest are probably better looking on the backtests (as weighting to small caps increases, and as they squeeze for more yield), but I am unsure about their merits relative to what is currently in the marketplace. As Luciano, their head of research said himself, what is lacking in the marketplace today are indices which are broader, more representative indices which fill larger asset allocation needs. What is not lacking are "one-off" products that may be seen as tricky, clever attempts to game the system... but a Small Cap Dividend Fund may fall into that category itself. We will factor this into the economics later.

(3) They are 100% playing the "fundamental indexation" theme, which has been beaten to death on this blog. Siegel mentioned it in his piece in the WSJ, and it has been mentioned many other times since. As such, they are essentially piggy-backing off of a wave which really truly originated with Bob Arnott, off of which 2 companies have already put out ETF's. My hypothesis is that they started with a Dividend index, and not one of the other perhaps more "expected" fundamental metrics, because Siegel, their Director of Research, has already done quite a lot of work on dividends, which means there may be cost factors involved. In a prior post on this blog, I mentioned a study that he had done a while back, but concluded that the dividend space was too crowded for this to be a likely ETF candidate (oops). My bet is they either won't have to pay a licensing fee, or the licensing fee is greatly reduced, because Siegel can claim that this is all simply an extension of prior work that he has done, which gives him a claim on said work. If this is true, then he gets all the advertising and education benefits of the "fundamental indexation" wave-- which I am sure that he, Arnott, Steinhardt, and others in the pseudo-active ETF space now intend to drive into the heads of common investors around the globe-- without having to pay for it. If it works, maybe he can release other indices based on other fundamental metrics later.


The Economics of an Investment in WSDT

I talked about the probable cost structure for WSDT in prior posts. Most specific talk of it is in the oldest post-- basically their revenue model is the expense ratio. Barclays charges something like 70 bps on a whole bunch of its indices, while the Spiders goes down to like 12 basis points. WSDT seems to be in the middle.

So you'd make an assumption on what WSDT's weighted average expense ratio is (if the split is 50/50 domestic international flagship ETF's, that would imply 38 bps). They typically get paid Monday through Friday if PowerShares was any indication, so the cash flow is a very slow and steady function of the assets under management. They may or may not have to pay a license fee (variable cost), then they pay for listing on the NYSE (fixed and variable cost), and they pay for the traders who construct the indices which most likely track computer-generated output of what it is the portfolio should look like with say a 5% leeway. They pay transaction costs (variable cost). They will also pay for a sales force (pseudo-variable cost), through which they intend to open themselves up to new investment channels. Other than that the biggest costs are for the management team. If you look at the pedigree of their management team (many guys who were heavily involved with the launch of the BGI suite), there is no way they are getting much in cash- they are probably accepting a call option-type compensation package-- variable cost. Research is probably not expensive at all because of student help, and marketing is probably much cheaper because of their star management. Main other costs I would imagine are consumer education, maintaining a website and logistical and administrative expenses.

They are "competing" against a handful of other ETF's which already have fundamental indexation products on the market-- I've talked about many of them on my blog, but they include 2 that were put out and are paying licensing fees to Arnott as well as the products put out by Powershares, now a sub of Amvescap. I know that WSDT intends to release a bunch of other non-dividend products (isn't focusing on being a dividend ETF co.), but I would be surprised if they were to sway too far from the fundamental indexation theme (aka piggybacking Arnott).

The key swing factors from my point of view are as follows:
  1. How much of mutual fund and hedge fund assets will end up in the hands of ETF products, synthetically or directly, when ETFs represent ~$420bn in assets, mutual funds ~$8T, and hedge funds ~$1.25T?
  2. Will Wisdom Tree win out over the host of other ETF products attacking the same market?
The driving force behind (1) is the sad fact that 80% of mutual funds underperform the market. And by market I mean the S&P.

More broadly speaking, the driving force behind (1) is the sad fact that it is very difficult to beat the "market", period. It takes a lot of work. And when you throw hedge funds into the equation, most of the hedge funds that do consistently outperform are either lucky or are not open to new investment. Of the hedge funds that are open to new investment, a good proportion of them are probably receiving compensation that is not in line with their ability to generate risk adjusted returns. Niederhoffer's Matador fell 30+% in the month of May alone. I am sure they are suffering from redemption issues. There were a slew of other funds which have closed after the recent market weakness. And I am sure there are many other investors who are looking at these funds closing, looking at their own investments and scratching their heads at why they are paying so much themselves (200 basis points and 20% of profits) when their hedge fund investments, which were supposed to be resilient on the downside, have fallen far more than the market has.

There is nothing new in the fact that mutual funds underperform. Academic studies have been done, etc etc. It boils down to one real question-- if 80% of mutual funds underperform the market and mutual funds charge 150 basis points, and there are ETF's which have shown an ability to outperform the market over time which also have deep capacity for investment and charge a 80% less than mutual funds, the current aggregate allocation of funds may consider changing!

So there are reasonable arguments for individuals in both camps to perhaps consider ETF's in some shape or form.

I will sound crazy for proposing the numbers below, but remember that I am looking at this from a 5 year perspective. In 5 years, either the paradigm shifts, or this company is probabilistically dead. I factor in probabilistic death into the upside through a setting, at the end, of the probability that paradigm shift does not occur. Adjust the market size, costs, margins as you wish... but I would hope that the underlying model is more or less representative.

The basic calculus-- if 25% of assets in funds right now are paying excessive, tax inefficient fees with inefficient portfolio construction, and come to the realization that they are doing so over the next 5 years, and if, in that period of time, (1) companies can release the education necessary to educate the market and (2) companies can create the platforms which can provide investors easy, tax efficient access to these products, and (3) WSDT is able to get 20% of those assets, it will have around $500bn in assets under management. At 38 bps, its top line is $1.83bn. With the SPY as a guide, transaction costs are probably around 12 bps for WSDT. Licensing fees is a wildcard. Sales commission and management expenses may be another 5 bps (or $242mm, split between ~10 hotshot (greedy) managers and a salesforce of maybe 30 highly successful guys), just to throw out a number. The other costs will probably become more variable-- research maybe $1mm, listing probably cost them $200k per ETF initially plus maybe 1bp of ongoing costs, website + non-exec admin + consumer education maybe another $80mm. Because IXDP emerged from a dead company, there may be some tax benefits, so perhaps slap on a 20% tax rate. This implies recurring net profit in the upside case of around $700mm. That profit will be a bit cyclical but in general pretty high quality so lets say slap a 15 multiple on it-- market cap of $10.5bn. Its market cap right now is $314mm, implying an annualized return of 100% for 5 years.

So what is the likelihood of this happening? Assume, for a moment, that the outcome of this company is binary (probably not too far from the truth). If the probability that they meet this admittedly extremely lofty scenario is 10%. That implies the expected value of the future market cap is $1.1bn, imply an expected annualized return of 27% from here... with some serious volatility.

Any thoughts would be appreciated.