blue-chip market. Only the largest and most widely held companies were admitted through its portals. The bottom 20% of NYSE stocks are much larger than what are normally considered small companies. The Curb, later the American Stock Exchange, was where the larger of the second tier of companies traded, and after that came the over-the-counter and the various regional markets. Banz, or certainly his blue-ribbon thesis committee, should have Icnown that a representative group of smaller companies did not trade on the Big Board. At best it was a study of how the bottom 20% of larger companies performed. Error number one then was serious in itself-Banzs study simply did not measure the performance of small stocks.

But this was only the beginning of a rather remarkable string. He showed that small companies had sparkling retums near the bottom of the Great Depression. Thats right, they moved up over 100% in the 1931 to 1935 period, whipping their largest brethren easily. How was this possible, you might ask, when we Icnow this was a time that small companies had the greatest financial difficulties and highest bankmptcy rates in American history? Figure 15-1, taken from the Fortune article discussing the Banz study, shows this supposed result as well as portraying an investor happily riding the pole of the giant payoff of small stocks in the 1941 to 1945 period, dollars floating all around him.

Banz, a financial statistician (as are virtually all of the current generation of investment Ph.D.s), apparently had little idea of what was in his sample, which resulted in a second major error. Banzs computer sort automatically placed numbers of troubled companies on the NYSE in his smallest group in the 1931 to 1935 period. Near the bottom of the Great Depression, some were in or approaching Chapter 11, which resulted in a drastic shrinkage of their market values, dumping them into the smallest capitalization group. Banz was often measuring larger companies that had been whittied down to small companies in this extremely troubled time-not small companies that were growing into major con-cems by rapidly improving market share and profitability as he and many others conjectured. A fair number subsequently recovered, accompanied by dramatic increases in price.

Many capitalizations had shmnk drastically in the 1930s-the average company value for the smallest stock group on the NYSE at the beginning of 1931, for example, was a piddling $640,000-thats correct, I didnt leave out any zeros.

Some of the reasons for survival were not a little strange. For example, the Electric Boat Company, which became a part of General Dynamics, sailed through this period with giant gains, because it won an $8 million setdement in the World Court from the German govemment.

Figure 15-1

WHEN THEY WERE GOOD THEY WERE VERY, VERY GOOD

Over the last 54 years, the slocks of the largest Big Board companies provided higher returns in six of the 11 periods charted. Yet on average the five-year retums on small companies stocks were more than twice as high. Thats because the smallest stocks rarely did a lot worse than the largest group, and when they did better they really shone.

Copyright: Fortune, June 30, 1980; reprinted by permission. Chart by Parios Studio.

The reason: the Imperial German Navy had "borrowed" a number of its submarine patents without permission in 1911 and 1912, before the start of World War I. Without this windfall, the company would have gone down for the third time.

The second and more damaging error also stemmed from contemporary financial researchers near obsession with statistical tests, applied all too often with little understanding of the underlying data or market mechanisms. Banz and his thesis advisors completely overlooked how illiquid the market was in the 1930 to 1945 period. At the beginning of

* If the stock did not trade, the CRSP database (Center for Research in Security Prices, University of Chicago), which Banz and many others have used for small cap studies, substitutes the midpoint of the bid/ask spread for an actual price. However, CRSP does not list the bids and asks themselves for any stocks before 1962. We obtained them by combing through newspaper listings published at the time.

1931 some companies were in receivership, although they were still listed on the Exchange. Many traded at a fraction of a dollar, a number going as low as % or . The average price of the bottom 20% of stocks on the New York Stock Exchange was only $3.34 a share over the 1931 to 1935 period.

By 1935, 30% of these companies were delisted. It took only four survivors (Adas Tack, Spiegel May Stern, Evans Products, and United Dyewood) to improve the average performance of each remaining stock in the cheapie group by 43%.

And here is the hitch: these were companies that just couldnt be bought! The markets were so thin that a number ofthese stocks traded only a few hundred shares a week, if that. Moreover, the spread was enormous for the pint-sized stocks, averaging 45% between what the buyer was willing to pay and the seller was willing to offer the stocks at. (These are the spreads for the lowest 20% of stocks on the last day of each of the 20 quarters in the 1931 to 1935 period, taken from The New York Times.)

Thats right, if you wanted to buy a stock at market and had to pay the offering price, your cost went up 45%. How much stock could you buy if you decided to grit your teeth and pay 45% more? Very little. An investor could buy only a few hundred shares, if that, for the majority of these issues. The average volume for the 139 stocks of the smallest-sized group on the NYSE was 240 shares a day through the five-year period. (This number is inflated because a few issues traded several thousand shares.)

The median volume was 100 shares at the beginning of 1931. Moreover the market continued to be almost illiquid through the entire 1931 to 1935 period. A startling 58% of these stocks did not trade on any given day.*

Not only did the researchers believe one could easily trade these stocks, they took a price between the bid and the offer price and did not consider that with the gigantic spread and low volume, the price was completely theoretical. The Banz study simply assumed the stock could be bought or sold in the middle of the spread,* thus paying absolutely no commissions and other transaction costs.