Part I.
Summary of analysis of 26 pre-development land properties purchased and resold by The Myers Group between January, 1978 and April, 1986, for an aggregate purchase price of $29.3 million.

1. If each property is equally weighted:
The mean return per property, continuously compounded, is 60.2 percent per year. (The corresponding number for discrete compounding is 82.6 percent.)

After allowance for length of holding period, there is no tendency for returns on properties to differ by purchase price or by time of purchase within the period studied. Returns tend to be higher for shorter holding periods. A summary statistical report for these conclusions follows:

2. The Myers performance does not appear to have been a "fluke".

In analyzing this performance record, one must ask if the 26 transactions are in some sense statistically dependent. Instead of 26 independent pieces of evidence, we are effectively seeing only the reflection of one or a small number of "lucky breaks".

As an example of the latter, suppose — contrary to fact — that the 26 properties had been nearly contiguous in a single, small geographic area, acquired at about the same time, and later sold at about the same time. Then, instead of 26 independent pieces of evidence, it would have essentially been one and there would be little assurance that this performance could be repeated in the future in the same or other areas.

The statistical analyses and detailed examination of the data suggest that the number of independent pieces of evidence is much closer to 26 than one.

A. The time-series modeling of the 26 properties suggests that the "residuals" from the regression equation shown in Part II can be regarded as nearly independent and normal through time.

B. Although in a few instances, properties were in the same metropolitan areas, geographic dispersion is substantial, within and between areas.

C. The performance shows no evidence of relationship to the broad economic changes that occurred from 1978 to 1986.

3. A likely consequence of Finding 2 is that, in addition to a high mean return and appreciation, The Myers Group offered diversification across the individual land investments.

It is unlikely that any temporally parallel set of 26 investments in stocks or other financial assets would show appreciation or returns that are substantially correlated with those of The Myers Group.

Although a systematic check of this conjecture has not been made, the fact that the residuals from the analysis of the Myers' properties appear to behave randomly through time makes it seem unlikely that a series of "coincident indicators" could be constructed from data on other financial instruments.

Even if the conjecture were not completely correct, the extraordinarily high mean appreciation and return of the Myers' properties make interest in this aspect of the investment decision virtually academic.

This assured riskiness of any portfolio that includes both the Myers' properties and other more mundane investments turns not on formal portfolio computations of the usual kind, but on one's evaluation of the evidence about Myers' performance.

Even if the correlation between Myers' and other investments were close to +1 or -1, the expected return from the Myers' component of the portfolio would be much larger than the expected return from the other components. If the Myers' investment were a substantial fraction of the portfolio, its performance would dominate the overall performance.

4. The Myers Group investments have significantly and substantially outperformed bonds, stocks, and other financial instruments.

The performance measures summarized make detailed comparisons unnecessary. For example, the mean annual return on the Myers' properties is 82.6 percent, discrete compounding. The largest annual return on stocks for any year since 1926 was 54.0 percent (in 1933). Again, stock returns were negative in nearly one-third of the years: not only are there no negatives in the 26 Myers' properties, but the smallest return was 16.9 percent. The latter is substantially larger than the mean annual stock return since 1926, which is just under 10 percent.

In order to investigate the return on investment relative to sold properties, two studies have been conducted:

a. A group of 32 randomly selected unsold properties were analyzed using current estimates of value. The results of this study were broadly similar to the results discussed for the sold properties.

b. In order to conduct an assessment of performance that did not depend on estimates of value furnished by The Myers Group, ten randomly selected unsold properties were used. MAI appraisals were conducted, using conservative R41B guidelines. Since these appraisals are expensive, it was deemed that a sample size of 10 was a reasonable compromise between the cost of sampling and the need for accuracy. The accuracy attained is sufficient to support the conclusion.

The unsold properties in the Myers' portfolio have performed in a fashion similar to the sold properties:

If each property in the MAI appraisal group is equally weighted, the mean return per property, continuously compounded, is 91.1 percent per year. (The corresponding number for discrete compounding is 148.6 percent.) This corresponds to 60.2 percent and 82.6 percent respectively for the sold properties.

The unsold properties offer the same diversification, as did the sold properties.

A summary statistical report for these conclusions follows.