Given the importance of the P/E ratio as a stock attribute, we decided to go further and try to use the forecasted P/E ratio to predict stock returns. Based on this, we designed our asset management learning model based on the forecasted and historical P/E ratios for a total of 43 countries. We compared the ability of these two factors to help us select stocks with higher returns. We did this by ranking the stocks from high to low P/E ratio and then looking at the returns in the top and bottom quartiles. This was repeated for both forecasted and historical P/E's.

Methodology:

1. Data Collection

- Forecasted P/E ratios: IBES database

- Developed countries data (returns and historical P/E) : MSCI

- Emerging markets data (returns and historical P/E) : IFC

- The data was collected for 43 countries from January 1988 to December 1999.

The following steps were performed on both the historical and forecasted P/E data.

2. Each month, we sorted the countries by their P/E ratios in descending order and then ranked them.

3. Due to the fact that some of the data for the earlier years was only available to two decimal places several of the countries would frequently tie in the rankings. In order to prevent this we added a randomly generated number to each P/E ratio. To ensure that the addition of this number would not effect our overall results we used small numbers between .00001 and .000001.

4. We then counted the total number of data points available in each month and divided the countries into quartiles according to their assigned rankings.

5. We used several lookup functions to automatically select the countries in the top and bottom quartiles and their corresponding historical returns.

6. The model then calculated monthly and annual returns.

Analysis:

12 Month Forecasted P/E's

We did not the results we were expecting from the sorting program using the 12 month forecasted data. Initially, we guessed we would experience much greater returns building a portfolio based on analyst forecasts relative to portfolios built with current, actual data. The buy portfolio's annual average return is slightly greater than the market return, however, it comes with a much higher level of risk as shown by the Sharpe ratio. Even though the sell portfolio experienced an average excess return of -8%, there was a high skewness towards great returns. Putting these two portfolios together, the spread portfolio exhibited a healthy 11% return with a 0 beta and a high positive skewness.

Looking at this chart, we see that $100 invested in the beginning of 1988 in the buy portfolio yields a greater return by the end of 1999. However, the buy portfolio only surpasses the market portfolio in 1999 and, with the level of standard deviation in the buy portfolio, this investment is not a wise one from a risk-return standpoint.

Historical P/E's

We were surprised to learn that building portfolios based on the historical P/E's yielded greater success than those built using the forecasted data. In the buy portfolio, we see a tremendous average annual return of 30% with a relative risk level better than that of the market portfolio (Sharpe ratio better than that of market). In addition, the buy portfolio experienced a positive return in 71% of the 144 observations with a minimum return of only -6%. The sell portfolio was a success, as well as, experiencing an average excess return of -7%.

With the success of both the buy and sell portfolios standing alone, the combination of the two in a 0 beta portfolio was an unqualified success. The spread portfolio averaged a return 7% greater than that of the market and experienced a positive return 63% of the time. With a maximum return of 131% and a minimum return of only -10%, this portfolio is highly positively skewed.

This exercise helped to reinforce the theory that growth investment strategies dominated value investment strategies during the past 10 years. The buy portfolio (growth strategy) chose the already-lofty P/E's, while the sell portfolio (value strategy) chose the bottom-dwellers in the array. This investment approach was most successful over the last five years where we see a major divergence begin to take place.

Looking at the chart, we see the buy portfolio returning an amazing $2,100 from the $100 investment 12 years prior. The spread portfolio returned $780, well above the market's $600 return (even though most of that success was concentrated in the last 2 years).

Conclusion:

We are still in the midst of this growth environment and, with the popularity of internet-related growth assets, it seems as there is no end in sight. The historical P/E buy portfolio and spread portfolio we created demonstrated great returns and should continue to experience solid gains while the gap between growth asset returns and value asset returns should continue to widen. However, we must dynamically manage this portfolio – closely monitoring the larger trends that occur between the value and growth investment camps.

Other Charts:

- 12 Month P/E Forecasted vs. Actual Buy Portfolios

- 12 Month P/E Forecasted vs. Actual Sell Portfolios

- 12 Month P/E Forecasted vs. Actual Spread Portfolios

Models:

- Model for Historical P/E Ratio (2.5MB, Excel File)

- Model for Forecasted P/E Ratio (2.5MB, Excel File)

Power Point Slides

Back to Other Group Projects

Back to BA453 Index Page