Seeking Alpha
Long/short equity, research analyst, portfolio strategy, newsletter provider
Profile| Send Message|
( followers)

By Carlos Guillen

Unearthing cheap stocks can often be very time consuming work that requires looking at many facets of valuation. In analyzing stocks, one has to look at companies both qualitatively and quantitatively. Qualitatively, one has to get a feel for the effectiveness of companies' management teams and one has to get an intuition of the overall business; that is, will customers like the products being offered, will management teams have the connections and know how to bring products to market, and will they be able to run operations efficiently enough to maximize profits and long run growth. Quantitatively, one has to analyze trends in revenues, margins, and earnings, not only for the company being analyzed but also for competition and the industry, and one has to get a grasp of the forces acting on the overall economy. However, at the end of the day, after considering everything, it all comes down to deciding whether a stock is cheap or not. But how does one decide if a stock is a bargain?

Perhaps the best way to find a bargain is to incorporate all the research done into what is called a discounted cash flow model. Discounted cash flow models are driven by the company's fundamentals, which includes projections of cash flows, growth, and risk. However, this task is much easier said than done. These models require numerous inputs to determine the fundamentals, which also require many assumptions about the distant future. So, while this is theoretically the best way to find the intrinsic value of a company, it is not the most practical.

A more practical way to determine the value of stocks is to use multiple analysis, or relative analysis. While there are many types of multiple analysis, including revenue multiples, book value multiples, and earnings multiples, I will focus on the latter, particularly in PE ratios. The reason why this valuation method is attractive is that it requires fewer assumptions than discounted cash flows and is much quicker to do. It is also much easier to present and understand. It should be noted, however, that this relative analysis does not reflect absolute value of stocks, but rather as the name implies, it reflects relative value. So if the entire market is overvalued, relative analysis will yield higher values for stocks than discounted cash flow analysis would, and vice versa.

To get an understanding of how relative analysis can be implemented to pick stocks, I have compiled data on some semiconductor companies that have somewhat similar businesses or at least face similar business risks. The data for each stock includes consensus estimate earnings for the following four quarters, the latest closing prices, and consensus estimate five-year earnings growth rates. The table below includes all the data mentioned and also has a column for the forward price-earnings (PE) ratio, which is just the price of the stock divided by the sum of the expected earnings for the next four quarters.

Click to enlarge images

Source: yahoo.com and nasdaq.com

This methodology takes into account the different growth prospects of all the comparable firms and compares them to their respective forward PE ratios. The process involves performing a linear regression between the forward PEs and the five year EPS growth estimates of all comparable firms. It should be noted that, although I'm choosing to do a linear regression between these variables, one could perform other types of regressions. In fact, the relationship between PEs and growth rates is not linear, but for the sake of simplicity I assume linearity.

After plotting the data and its corresponding regression line (Y = 113X + 2.88), as in the chart below, I look for the points that are the farthest from the regression line; those points that are the farthest below the line are stocks that are undervalued, and those points that are significantly above the line are overvalued. Since I'm looking for cheap stocks, I look below the line, and I find two stocks that seem undervalued (marked in red and also highlighted in the table above); these two stocks are Cirrus Logic Inc. (NASDAQ:CRUS) and On Semiconductor Corp. (ONNN). Using the regression equation, I calculate a predicted fair PE value based on consensus growth rate estimates.

PE CRUS = 113(0.20) + 2.88 = 25.6

PE CRUS = 113(0.20) + 2.88 = 25.6

Since the actual PE ratio for Cirrus Logic and On Semiconductor are 15.0 and 12.6, respectively, the predicted fair values suggest that these stocks are undervalued by 41.3% and 36.4%, respectively. Of course, this is given how all the other companies selected are currently priced by investors.

I should point out that this analysis is a first order estimate, that is, I am assuming that other important fundamentals are similar from company to company and do not need to be taken into account. One good example of this is that I'm assuming that the list of companies in this analysis has similar risk characteristics, which is very likely not the case. A better analysis would have attempted to control for differences in risk characteristics between companies, perhaps by including beta or standard deviation data for each stock in the regression. However, this would have introduced new issues to worry about such as co-linearity between independent variables that tends to amplify error. For example, high growth firms tend to have high risk, so risk and growth are somewhat correlated, which amplifies the resulting coefficients in the regression equation, increasing the error of the estimated PE. So, for the sake of simplicity, I chose to control for just one variable, that being growth.

While the method discussed here is certainly attractive in terms of its simplicity and understandability, it has lots of room for error. As I have mentioned, failing to control for differences in fundamentals can lead to significant error. Not only that, but also not knowing or understanding the fundamentals driving a multiple can lead a valuation down the wrong track. In addition, failing to define the multiple correctly also leads to errors. For instance, it makes no sense to mix up trailing PEs, forward PEs, and current PEs. Moreover, failing to recognize outliers can also skew the results. Consequently, this regression method should be viewed as one more instrument in the quest to dig for cheap stocks.

Source: Two Cheap Semiconductor Stocks