Security Analysis-A Mission Impossible?

In light of what we have just seen, you might be wondering if you can depend on current investment methods. Before jumping to any conclusions, lets watch an analyst evaluate a company using fundamental analysis. Suppose our analyst decides to examine Hewlett Packard. How will he or she go about it?

Hewlett designs and manufactures precision electronic products and systems for measurement and computation. Examining the companys financial statements, one sees a gigantic operation. In 1996, it had revenues of over $38 billion and net profit of $2,675 billion, and employed 102,300 people domestically and abroad. Foreign sales in 75 countries account for 56 percent of total revenues.

The analyst will probably start with eamings power, that most important determinant of value. To do his job properly, hell have to look at dozens or hundreds of inputs. He will review in detail the recent history and prospects of each major product line.

If the analyst is the thorough type-and he said he was on his resume-he will go further, possibly getting breakdowns from management of how each of these groups is doing in every major market.

But were not finished with our friend yet. In fact, his most difficult innings lie ahead. Multiplying this information by many other HP products extends the length and perhaps tediousness of the analysis many-fold. If tediousness were his only concem, however, the analyst might happily accept it. A far more serious problem is that much of the information he is able to ferret out is highly uncertain.

Facts provided to him by management as the basis for his various estimates are partial or incomplete, and sometimes incorrect. A company officer may tell him that a new laser printer is up nicely in revenues but that sales of fax machines are so-so year-to-date. Ifhe asks how to translate a "so-so" or an "up nicely" into a reasonable eamings estimate, as

Moreover, what the brokers said about how they analyzed the various inputs differed significandy from what they did. For example, a broker who said the trend of eamings per share was most important might actually place greater emphasis on near-term prospects. Finally, the more experienced the brokers, the less accurate their assessment of their own scales of weighting appeared to be. All in all, the evidence indicates that most people are weak configural processors, in or out of the marketplace.

How Much Is Too Much?

Under conditions of complexity and uncertainty, experts demand as much information as possible to assist them in their decision-making.

often as not the officer may shrug and say, "Company poHcy does not permit me to divulge the information." Or, "Thats your problem."

In any case, our friend is left on his own. He must rely on his own judgment, deciding whether "nicely" means up 10% or 30%, or "so-so" means flat or down sharply. He can check with trade sources and the competition, but the information provided will also be qualitative and sometimes misleading. Thus, the analysts judgment is taxed at every stage of the analysis.

Whatever the company or industry, the analyst is bombarded with vast amounts of difficult-to-quantify information on competitive conditions, capacity utilization rates, and pricing. What will be the effect on IBM or Compaq of a fast new laptop computer from Toshiba? How badly will Chrysler, GM, or Toyota be hurt by a price cut in a popular model by Ford? All pertinent information must somehow be synthesized and evaluated in order to arrive at an eamings estimate.

On top of the forecasting problems, the harried analyst must also assess management, expansion plans, finances, probable dividends, accounting, and dozens of other vital factors. All estimates are contingent on general economic conditions, which means correctly gauging interest rates, unemployment, inflation, industrial production, and other economic variables here and abroad. Economists themselves are often wrong in these estimates, as weve already seen. Eamings forecasting, then, depends on large numbers of underlying assumptions, many of which are rapidly changing and hard to quantify-which means their accuracy is always in doubt.

The theory is anything but undemanding of its poor adherents. The amount of information they are expected to process is staggering. Even if our poor analyst has won his spurs in the super-computer category, the demands on him do not end here. Contemporary investment theorys final requirement is the most difficult of all. It requires him to correcdy weigh all these factors against each other.

Weve already seen that man is simply not a good configural processor of information. The reach of conventional investment theory may very well exceed the grasp of many of us to use it properly. What it will do is bring us well into the range of information overload.

Figure 4-2

Bad News for the Handicappers

Average changes in confidence and accuracy with increasing amounts of information

I 20

§ 10

10 20

Items of information

Source: Adapted from Paul Slovic, "Behavioral Problems Adhering to a Decision Policy/ IGRF speech, May, 1973.

Seems logical. Naturally, there is tremendous demand for such incremental information on the Street, because investors believe the increased dosage gives them a shot at the big money. But as I have indicated earlier, that information "edge" may not help you. A large number of studies show rather conclusively that giving an expert more information doesnt do much to improve his judgment.*

In a study of what appears to be a favored class of guinea pigs, clinical psychologists were given background information on a large number of cases and asked what they thought their chances were of being right on each one. As information increased, the diagnosticians confidence rose dramatically, but their accuracy continued to be low. At low information, psychologists estimated they would be correct 33% of the time; their accuracy was actually 26%. When the information was increased fourfold, they expected to be correct in 53% of the cases; in fact they were right 28% of the time, an increase of only 2%.

Interestingly, the finding seems universal-no improvement with more information. The same results were obtained using track handi-cappers. Eight veterans of the racing form were progressively given 5 to 40 pieces of the information they considered important in picking