One area in which computer-assisted reporting has been less frequent is business coverage, in large part because there aren't gobs of data readily available, such as that prepared by government agencies.
But, if you scout around, it turns out great information actually is available. With some imagination, you can use it to produce good stories. One example, humbly submitted: The San Jose Mercury News recently enjoyed success in mining trading records of the Nasdaq Stock Market to show how investors are forced to pay sharply higher trading costs.
Before you yawn, understand that Nasdaq affects virtually everyone. In Silicon Valley, the Mercury News' backyard, Nasdaq has been the coin of the realm in recent years. Hundreds of high-tech companies have gone public on Nasdaq, raising the capital that has funded the valley's world-changing innovations.
More broadly, tens of millions of Americans invest in Nasdaq stocks – on an individual stock basis, through 401(k) plans or through pension plan holdings.
Unlike its better-known cousin, the New York Stock Exchange, Nasdaq has no trading floor. Instead, Nasdaq exists in the form of a super-fast computer network connecting securities dealers across the country.
Investors' losses are stock dealers' gain
Therein lies our story. The fundamental premise of this system is that spirited competition among the dealers is supposed to keep investors' trading costs low. But, using a CAR analysis, the Mercury News found that isn't true.
As we said in our "Shortchanged on Nasdaq?" report, more often than not, trading of a company's stock is concentrated in the hands of only a few dealers. When that happens, the transaction costs for investors are sharply higher.
How much higher? Add it all up and the extra cost to the investing public could reach hundreds of millions of dollars annually.
Who profits? Mostly the stock dealers, which include such household names as Morgan Stanley, Merrill Lynch and Goldman Sachs, plus regional firms across the country.
To study Nasdaq, we analyzed about 11.2 million Nasdaq trades and market price postings. This data came from the Trade and Quote database of the New York Stock Exchange. Available on CD-ROM at nominal cost, TAQ contains trade-by-trade data on stocks listed on the NYSE, the American Stock Exchange and Nasdaq.
Widely used by academics, TAQ is a good example of the kind of lesser-known business data that's available.
Our analysis found that trading of a company's stock was concentrated in the hands of only a small number of dealers nearly two-thirds of the time. When there was the least amount of competition among dealers, the cost to investors to buy or sell those stocks was nearly twice what it was to trade when competition was vigorous.
Most important, the level of dealer competition turned out to be a bigger factor in explaining investor costs than other factors customarily looked at by Nasdaq and federal regulators.
Developing this last point was key, and it accounted for most of the heavy lifting in our CAR analysis. It turns out that assessing dealer competition by itself isn't terribly difficult. Besides dealer competition, things like trading volume also affect investor costs. It was crucial to "filter out" those other things to arrive at an apples-to-apples comparison.
Know statistical limits
We did this using a multiple regression model we developed. If you're interested in the gory details, check out the Web sites listed in the sidebar. In the meantime, suffice it to say this model accounted for various factors affecting transaction cost and allowed us to explore among all of them how significant the influence was of dealer competition. Our primary software tools were QuattroPro, Paradox and SPSS.
As we said in our story, we didn't find evidence of collusion or other deliberate anti-competitive behavior among dealers. But, some economists who reviewed our findings believe Nasdaq should make changes to improve market performance.
With any analysis like this, it's important to keep in mind that you can never prove a relationship – in statistical terms, you've got to be on guard against confusing correlation with causality. While we didn't prove that lagging dealer competition directly causes higher investor costs, the analysis clearly suggested that the level of competition plays a significant role, even when accounting for other factors.
Nasdaq executives hated what we did, calling it misguided and wildly inaccurate. In fact, they went to considerable lengths to try and prevent publication of the story.
But, the bottom line, as business people are wont to say, was this: We put CAR to work in a coverage area where data isn't as readily available as others. In doing so, we think we told a compelling pocketbook tale that affects nearly everyone. You can, too.