Bridgeway's Montgomery 'Indexes' the Market's Smallest Companies By John Spence June 17, 2003

John Montgomery is the founder of Houston-based Bridgeway Capital Management and manages six Bridgeway portfolios - including both active and passive funds. He worked with computer modeling and quantitative methods as a research engineer at MIT in the late 1970s. Later, as a student at Harvard, he investigated methods to apply modeling to portfolio management. He left his full-time position in the transportation industry at the end of 1991 to perform full-time research on his investment models, study the mutual fund industry, and write a business plan for Bridgeway.

This is the first part of a two-part interview. Below, Montgomery discusses Ultra-Small Company Tax Advantage (BRSIX), a fund that uses a passive approach to invest in the market's tiniest companies. In Part 2 he talks about Bridgeway's other passive fund, Ultra-Large 35 Index (BRLIX)

What's the rationale for indexing small-cap stocks in a supposedly inefficient market?

Right, the standard wisdom is to index large-caps and use active funds for small-caps. We believe our approach to indexing small-caps gives us a leg up over active managers, for three reasons.

1. First, the expense ratio is low at 0.75%, and we expect the expense ratio to decline in the next fiscal year. [The average expense ratio of funds in Morningstar's small company fund database is 1.60%.] 'Micro-cap' funds that invest in tiny companies are even more expensive than 'small-cap' funds.

2. Trading costs are a major factor in the micro-cap space where there are large bid/ask spreads, which can sometimes run 4% to 5% percent of the cost of the trade. This is prohibitively expensive unless you have the size and expertise to aggressively make this an advantage rather than a disadvantage. Our goal is to on average sell closer to the ask and buy closer to the bid, rather than just putting out a market order, which would in fact more than give up the return advantage of the [ultra-small] asset class. When you're trading for example 50,000 or 100,000 shares of one of these small companies, you are a bigger player, and in some cases you can provide liquidity and get some great deals [on the bid/ask spread]. We've also been trading these tiny stocks for about nine years, and we dedicate a lot of our resources to trading.

3. Finally, the asset class itself has advantages. We call them 'ultra-small' stocks - CRSP [Center for Research in Secruity Prices, University of Chicago] calls it their 10th decile stocks. The asset class has had higher historical returns, and is also a great diversifier because of its low correlation with large-caps. Ultra-small stocks individually are widely volatile, but a big part of that is company specific risk. If you own enough companies you can diversify away this company specific risk - the portfolio has 450 to 500 stocks.

The Small-Cap Effect, 1926-2002
Decile	Arithmetic mean (returns)	Standard deviation
1 - largest	11.20%	19.44%
2	12.90%	22.13%
3	13.50%	24.02%
4	14.00%	26.26%
5	14.50%	27.06%
6	14.90%	28.11%
7	15.20%	30.33%
8	16.20%	34.03%
9	17.10%	36.90%
10 - smallest	20.80%	45.37%

BRSIX annual returns since inception (July 1997)
	1998	1999	2000	2001	2002	YTD 2003 (4/30/03)
BRSIX	-1.81%	31.49%	0.67%	23.98%	4.90%	10.30%
S&P 500	28.58%	21.04%	-9.10%	-11.88%	-22.09%	4.82%
Russell 2000	-2.55%	21.26%	-3.03%	2.49%	-20.48%	4.56%

	1-year	3-year annualized	5-year annualized	5-year standard deviation (through 4/30/03)
BRSIX	28.67%	19.21%	14.73%	23.11%
S&P 500	2.16%	-10.33%	-0.06%	18.76%
Russell 2000	-8.18%	-1.16%	0.65%	24.36%

I think there are more inefficiencies in small-cap stocks compared to large-caps. However, I don't think the average active small-cap manager is going to beat a well-constructed index. I think one of the reasons managers beat the Russell 2000 is not because it's a poor index, but because there are issues with rebalancing and front-running.

One of the benefits of using a CRSP index is that it isn't widely followed - they're not even priced daily, so you don't have the front-running issue.

I would add that the recent outperformance of BRSIX is unusual. From 1994-1999, we had probably the biggest bull market in large-cap stocks, which isn't great news for a fund that specializes in tiny stocks. However, at some point that large-cap dominance had to unwind. Now small-caps are outperforming, and historically the tiny ultra-small stocks have experienced a huge upswing when small-caps outperform. They're sort of on the tail end of that whip, so to speak.

How much smaller is BRSIX than other funds in the small-cap universe?

On average, historically the CRSP 10 index is about one tenth the market capitalization of the Russell 2000. We're talking about really small companies.

Source: www.bridgewayfunds.com

Has there been significant tracking error against the CRSP Cap-based Portfolio 10 Index?

First, we don't own all 1,900 stocks in the index.

We did find out early on that we actually needed more stocks than we anticipated to track the index. Also, the CRSP indexes are rebalanced quarterly, and we didn't want that much turnover because of the resultant transaction costs and, more importantly, tax consequences.

Because these tiny stocks are very volatile, you have a much higher percentage of companies that go bankrupt every year. Of course, you also see a higher percentage of stocks that quadruple in price. So you want to own more companies to dampen this company-specific risk.

Because of this volatility and the quarterly rebalancing, occasionally you'll see these somewhat cataclysmic changes in the index. To reflect these changes and track the index, we'd have to run up trading costs and sacrifice tax efficiency.

The ultra-small asset class also has some unique advantages in terms of harvesting losses. Of course, it's easy to harvest losses in a bear market, but in a bull market it's much harder for most funds because naturally most stocks are going up.

However, in ultra-small companies you see a lot of variation from company to company. Even in a bull market, some companies are going to perform horribly. The good news is that you can harvest those losses and apply those to the gains that result from mergers and acquisitions, and also from companies simply outgrowing the index. This is one of the reasons why the fund has a perfect record on capital gains distributions.

If we operated like a true index fund, we wouldn't be able to provide that level of tax efficiency, and of course our transaction costs would be higher if we did the amount of trading necessary to keep up with the quarterly rebalancing. About three years ago we changed the name of the portfolio and removed the "index" label.

So there will some tracking error versus the index, especially during market swings. But we feel this is beneficial for long-term investors because of the fund's tax efficiency.

It's been six years and we have yet to distribute a capital gain. That's pretty easy in a bear market, but this fund has had only had one down year [in 1998].

How many stocks does the fund hold compared to the index?

The index right now has fewer stocks than normal - the average is around 2,000 stocks. The recent downturn has washed out some of the smaller companies.

We usually have 450-500 stocks in the portfolio, although number that could increase in the future. We have an absolute criteria where the fund cannot differ by more than 1.50% from the sector allocations of the index. It's a complicated process, but we also try to approximate the fundamentals of the index [p/e, median market cap, dividend yield, etc.]

Do you take any steps to discourage "hot money" from entering the fund, especially since the asset class has been performing so well over the past three years?

We just added a redemption fee [2% penalty within a six-month period] at the end of last year to further discourage market timers. Also, in the fund's prospectus we explicitly state our funds are only for long-term investors. We don't want market timers or people trading in and out of the fund. Any redemption fees collected revert to the fund and not to the advisor.

We did see some market timers move into the fund in January of 2002. This seems to happen when the fund has good recent performance against moving averages. We'll then see some money go into the fund and leave within a short time.

Ironically, we found that we could accommodate the market timers last year, because typically they were getting in and out of the fund at the wrong times! So it wasn't all that bad for the portfolio. However, it was running our staff ragged because of all the trading, so we added a redemption fee. We don't begrudge people for leaving; we just don't want them doing it too soon.

This is not the fund for people who are going to panic in a down market, so we also have the ability to charge an additional redemption fee in certain situations. The fund's board of directors has the right by prospectus to charge a 2% redemption fee if the S&P 500 is down more than 5% over the last five trading days. We're the only fund firm out there that does this, and I think it's a great idea.

It's just one more way of signaling the type of investor we're looking for. Also, we've run the numbers, and the people that have tried to time the market with this fund have not succeeded. If you're going to get antsy with a 30% drop, you shouldn't be in this fund - because within a 20-year period that can happen a few times, and you'll be bailing out at the worst possible time. If you look at the historical returns of the ultra-small asset class you'll see that there were some big down years.

So you shouldn't have a lot of money in the fund unless you have a steel stomach, but it can be a great diversifier even if it only makes up a small percentage of your entire portfolio.