May 2000

Background

The letters that CEOs write each year for their annual reports to shareholders are considered to be the most important of all corporate communications. Over the years, we’ve developed a proprietary model, LanguageWorks for business^SM, (LWA), to measure and score the content, style, and candor of the 100 company shareholder letters in our survey.. Why are shareholder letter language scores important? How are language scores correlated with stock price performance? Here is a summary of our findings:

I. Correlating Stock Price Performance and LWA Scores

We find a probability of 95% or better that companies with high language scores are more likely to be correlated with higher stock price performance and low scoring companies are more likely to show low stock price appreciation. Considering just the top and bottom scoring companies in our survey, we find an even stronger correlation between language scores and stock price performance of 97.5%.

While we do not claim that high language scores cause higher stock price performance, we do believe that higher language scores are indicative of strong management and leadership capabilities which are more likely to produce superior performance.

II. Comparing Actual Stock Price Performance and LWA Scores

The average price appreciation of the top scoring 25 companies in the 1998 LWA survey between the beginning of 1998 and the beginning of 1999 was 22.5% compared to a decline of 3% in the average stock price appreciation of the bottom scoring companies. During this same period the S&P 500 Index increased 19.5%.

From the period beginning in 1997 until the beginning of 1999, the average share price appreciation of this same group of the top scoring LWA companies was 89.4% compared to 3.4% for the bottom scoring companies.

Over the years, we have consistently found a positive relation between high language scores and superior stock price performance and low language scores and lower stock price performance.

III. Developing a Distribution Curve for Language Scores

The distribution of the LWA scores in our survey are weighted at the lower end of the curve. In fact, 38% of the companies in our survey fall in the lowest quadrant of LWA scores compared to only 13% of the companies which cluster in the top quadrant of LWA scores. This result validates the opinions of investors and others who either can’t understand or do not believe what is reported in the shareholder letter and typically don’t read them. It also supports the views of readers who look for informative and candid letters finding that their scarcity makes them noteworthy examples of leadership.

IV. Constructing a Reliable Sample

Because statistical results are highly dependent on the nature of the sample of companies to be surveyed, we have followed these guidelines in constructing the LWA survey sample:

Sample Size - We focus on large capitalized and well-known companies. Our sample of 100 companies represents 20% of the companies found in the Fortune or S&P 500 groups of companies. This is a statistically significant size for this universe. The companies chosen in the sample represent 39 of the 53 Fortune 500 industry groupings.

Randomness of Sample - Rather than selecting companies at random among the lists of these top 500 companies, we constructed a representative sample of companies based on 1) capitalization of $1.0 billion or more, 2) broad industry representation, and 3) placement on Fortune’s 1999 list of most admired companies. The 1998 LWA survey includes 42 companies which were ranked #1 or #2 most admired in their industry groups and 32 companies which ranked last or second last among the 39 Fortune industry groups. Eighteen companies placed in the middle of their industry rankings. Eight companies not included in the Fortune lists nevertheless qualify for the survey based on their capitalization and industry representation: AES Corp., Qwest, Amazon, UtiliCorp United, Keyspan Energy, General Mills, Quaker Oats and AFLAC.

V. Correlation of LWA Scores and Most Admired Rankings

Of the companies scoring highest language in 1998, 15 companies were ranked #1 or #2 most admired in their industry groups. Only four companies which produced a top ranked letter were listed at the bottom of their industry groups. These companies in this top quadrant were mid-ranked in their groups and three companies were from the group not listed in Fortune.

Of the 25 companies scoring lowest in language, 16 companies were ranked at the bottom of their industry groups. Only six companies in this lowest quadrant of language scores were ranked high in admiration. The remaining three companies were mid-ranked in their industry groups.

The strength of this association shows the importance of high LWA scores in identifying companies that are more likely to be well regarded in their industries. Conversely low language scores indicate companies that are less likely to be admired and have low perceptual value.