The Value Investing Blog

Empowering value investors with stock valuation tools, tutorials and resources

A Test of Graham’s Stock Selection Criteria

September 6th, 2010

23 Comments

A great way to find ideas is to use the selection checklist of gurus. There is one by Benjamin Graham that not too many people know about. A couple of readers have been kind enough to provide the selection criteria and papers testing the process which I will go into detail here.

Graham’s Stock Selection Criteria

There are 10 criteria in total.

The first 5 criteria measure ‘reward’ and is sensitive to price and earnings changes. The focus in this group of five criteria is on stock price, earnings and dividends.

The second group of 5 offers a measure of ‘risk’ and does not change rapidly with changes in price and earnings. Criteria number 6,7 and 8 represent the financial soundness of companies.

Selection, by using the criteria, is based on the concept of maximising the ‘reward’ to ‘risk’ ratio of stock selected.

One thing I want to point out is that the reward-risk philosophy is backward. Personally, after chasing reward to risk much often and taking the damage, I have found it much better to chase after low risk to reward ratios. It may sound like the same thing, but determining low risk followed by reward is much different to finding reward and then performing the risk analysis.

List of 10 Stock Selection Criteria by Benjamin Graham

1. An earnings-to-price yield at least twice the AAA bond rate

2. P/E ratio less than 40% of the highest P/E ratio the stock had over the past 5 years

3. Dividend yield of at least 2/3 the AAA bond yield

4. Stock price below 2/3 of tangible book value per share

5. Stock price below 2/3 of Net Current Asset Value (NCAV)

6. Total debt less than book value

7. Current ratio great than 2

8. Total debt less than 2 times Net Current Asset Value (NCAV)

9. Earnings growth of prior 10 years at least at a 7% annual compound rate

10. Stability of growth of earnings in that no more than 2 declines of 5% or more in year end earnings in the prior 10 years are permissible.

Test Combination of Criteria

Not a single stock will be able to pass this filter today. When the screen reaches no. 3, only around 30 stocks make it. When you hit the fourth condition, the list becomes 0.

So to make this exercise worthwhile, I would have to run the criteria in combination with each other but not all at the same time.

The table below is from the paper with the test performed from 1977 to 1995, a span of 18 years.

Keep in mind that back in the days of Graham, most companies were industrial and the stock universe was much smaller then as well.

In order to clean up the results, I left out financial stocks, OTC stocks and ADR’s.

I do not have access to historical AAA bond yield data so I applied a static 4.5% as the AAA bond yield for all screens for all periods.

Still, it results in plenty of pickings. The additional condition that I applied was (7) Current ratio great than 2.

I also tested one that included (8) Total debt less than 2 times Net Current Asset Value (NCAV). The number of results were reduced but the screen did not perform any better so I left it out.

Graham Stock Selection Criteria Results

With the limited amount of data that I have access to, 20 years of back testing is out of the question. 20 year tests also do not serve much purpose. Such tests, I find to be distorted due to it being unrealistic. You might as well do the test for 100 years and conclude it kills the market.

I chose the period from 2004 end of August 2010 to test the results.

Why 2004?

I didn’t want to apply the test in 2001-2003 because of the dot com bust. By 2004, the recession was over, markets have stabilized and “ordinary” people would have gotten back into the market. Not quite Nobel prize worthy, but realistic.

The conditions I applied to all screens are as follows:

  • ignore financial companies, OTC and ADR’s
  • AAA bond yield was kept at 4.5%
  • max 20 stocks held
  • 1 year holding period

I would say a 6 year period is long enough to say that this strategy would beat the indexes over the long run. With $100, the screen depicted by the orange line would have netted $193.

Individual year performance will vary and fluctuate and you won’t beat the market every year, but Graham’s simple selections looks to be solid for the passive investor as well as being a good source of ideas for the enterprising.

10 Stocks that Graham Would Choose

Applying the (1),(3),(6) & (7) conditions, here is a list of stocks that even Graham would be interested in if he were alive.

Disclosure

None

About Jae Jun

Jae Jun has written 529 value investing articles.

Jae is the founder of Old School Value and has a mission to empower every investor by providing tools, tutorials and educational material so that they can invest successfully on their own.

Recent posts by Jae Jun

You may also like:

  1. Graham’s Stock Selection Screen Part 2
  2. Value Stock Investment Criteria
  3. Investment Checklist for Stock Selection
  4. How to Value a Stock with Benjamin Graham’s Formula
  5. Graham Checklist Screen

Value stocks in 30 seconds with the stock valuation spreadsheets

  • Josh

    What is meant by (1),(3),(6), etc.? Considering that there are 10 conditions, this part isn’t clear.

    Thanks

  • Mark

    Interesting results. I would be interested in seeing the response for additional conditions applied (assuming there are stocks that still qualify). Even more interested in seeing all possible combinations of the ten conditions. Would the same pattern of improved performance emerge with additional conditions applied or would this behavior break down at some point?

    Sorry if you’ve answered this before, but what screener did you use to find these results?

  • Eldinril

    So… this is referred to as Graham’s Last Will and Testament. It was written by a man named James Rae, who became friends with Graham in the last few years of his life. The bond yield should be the 5-year AAA Corporate bond yield, total debt should actually be compared with tangible equity, and earnings should simply double over a ten-year period. Graham stated in a lecture at UCLA that if an investor just used earnings yield, dividend yield, and debt to tangible equity, they would get results double the DJIA.

    Best wishes,

    Eldinril

  • http://www.oldschoolvalue.com Jae Jun

    @ Josh,
    (1),(3),(6) are the conditions 1,3 and 6 in the list of 10 criteria.

    @ Mark,
    With all 10 conditions, you get 0 results. I tried many different combinations but the results were not any better. I noticed that if too many conditions were included, it actually broke down the performance.
    I’ll see about doing a follow up with different conditions.

    @ Eldinril,
    Do you have a source from where you got this info? I’d like to check it out.

  • fandegraham

    Hi
    Could you backtest with 1,3,4,6,7 criteria ? I guess you should have better result with criteria 4, i am very interested with the result of that compared to others.

  • Roy

    What is meant by current ratio?

  • http://www.wiftsinvestment.com R

    like Eldinril said, after coming up with these 10 points, Graham said that you could achieve the same result by selecting stocks with only the following 3 out of 10 points: Dividend yield at least 2/3rd AAA bond, Earning yield at least twice the AAA yield, debt to equity less than 1. Which means 1 3 and 6 above. Other points are redundant. See how the performance of a portfolio of those companies look like. I think there was an article on a recent AAII publication.

  • Eldinril

    R is correct. The AAII recently did a write up of the checklist. The original article was published in the Summer 1977 issue of \ The Journal of Portfolio Management\. It was also mentioned in a Forbes publication with the chapter about Graham entitled \Value Avatar\.

    I will correct myself on the point about bond yield. The original article simply states \the average AAA bond yield\. I am not certain where I got in the practice of using the 5-year yield for this. Of course, 7% compounded or doubled over ten years are pretty much equivalent. That said, he definitely used the tangible book value.

    I recall reading once that Rae and/or his son attempted to run money using the ten-point checklist, but achieved lackluster results. I find it particularly intriguing to compare this checklist to the list of criteria included in the later editions of \The Intelligent Investor\. It is important to understand that this particular checklist was the result of conversations between Graham and Rae. Some of the details were the result of compromises between the two of them.

    Best wishes,

    Eldinril

  • TMurphy

    Graham also had another criteria that he discussed near the end of his life, he suggested buying a large number (20+) of stocks that fell 50% below the high of the past 2 years. Hold these shares until they have appreciated 50% or until two years have past. This buy criteria was very helpful during the Sep ’08 to Mch ’09 period. Graham said most of the stocks bought using this method would be extremely out of favor companies, the beauty of the ’08/’09 period was that great companies met this buying guideline. Greatly enjoy your blog. Thx.

  • Eldinril

    That’s right, TMurphy, and that really plays into the comment I made about this particular checklist being the result of conversations between Graham and Rae. If I recall correctly, Graham suggested using the criteria of a stock selling below 50% of its 2-year price high. Rae wanted to use 33% of the 2-year P/E high. They compromised and used 40% of the 2-year P/E high.

    It seems to me that an analyst could use any of the three, or perhaps substitute in EPV or Graham’s Intrinsic Value formula, and achieve what Graham would call “satisfactory results”.

    Best wishes,

    Eldinril

  • dan mcintosh

    I know Mark already asked this, and he was sort of answered, but I am curious of what type of screener you are using and where to find it?

    Thanks.

  • somrh

    With respect to bond yields, there does seem to be a bit of interpretation there. Historical records for government bond yields are available at the Treasury site, for example:

    http://www.ustreas.gov/offices/domestic-finance/debt-management/interest-rate/yield.shtml

    My thought would be that longer term yields would be more appropriate. I’ve been using 10 year bonds as a guide simply for the fact that I can get them easily here:

    http://www.multpl.com/interest-rate/

    This gives the following info:
    Current Yield: 2.77%
    Mean: 4.69%
    Median: 3.84%

    The 4.5% Jae uses seems reasonable according to historical averages. The only exception I would make is that the mean is somewhat inflated as a result of the high yields in 70′s/early 80′s due to high inflation. If we expect the Fed to have a more conservative fiscal policy, then the high inflation of the 70′s (hopefully) won’t return making the mean somewhat inflated.

    If we wanted to adjust for inflation by hand we can look at TIPS for inflation adjusted yields. The treasury supplies data for TIPS yields as well:

    http://www.ustreas.gov/offices/domestic-finance/debt-management/interest-rate/real_yield.shtml

  • somrh

    Also, the James Rea paper can be accessed here:
    http://www.iijournals.com/doi/abs/10.3905/jpm.1977.66

    Unfortunately, my university doesn’t have online access to this so I haven’t had a chance to read it. Anyone happen to have a copy or know where I can access it?

  • http://www.oldschoolvalue.com Jae Jun

    Part spreadsheet, part custom made database to get all the data needed for the screens. Used to use AAII historical data but too complicated. The RobotDough screener is a good start and they will soon be implementing a backtester as well.

  • Eldinril

    I would just like to say that I appreciate Somrh’s comments. I sometimes use Moody’s Seasoned AAA Corporate Bond Yield, which is available on the St. Louis Fed’s website. Oddly enough, the yield at the end of August was 4.49. Using the higher number will certainly result in a more limited number of stocks.

    I do use elements of this screen, but use alternate criteria instead of the \below 40% of the two-year P/E high\ for valuation. I cannot recall when I decided to start using the 5-year Corporate Bond rate for this screen, but have found that it provides reasonable results. For example, when I limit myself to Moody’s Seasoned AAA, I am limited to stocks such as BKE or JNJ. If I use the 5-year Corporate Bond Yield, the results also include stocks such as BDX and MSFT among others. That said, the 10-year provides similar results. So, it boils down to a matter of personal preference.

    Best wishes,

    Eldinril

  • http://www.oldschoolvalue.com Jae Jun

    @ somrh,

    Yea would be nice if someone has a copy of the James Rea article.

    I would also much prefer to use a yield of 4.5% rather than anything lower. Just helps to limit the results and a stock offering better yields than the current 2.x% is very important in my criteria. Otherwise, I might as well buy the index rather than try to screen for individual stocks.

  • Daniel Cluley

    Nice evaluation of a classic set of rules. I enjoy your analyses of value and dividend strategies.

    For Bond Rates you can find historical rates from the St. Louis Fed here: for AAA: http://research.stlouisfed.org/fred2/series/AAA/downloaddata?cid=119 and Treasury http://research.stlouisfed.org/fred2/series/GS10/downloaddata?cid=115. I have not updated in a couple months, but AAA historical Mean=5.91 Median=5.25 and 10 yr Treasury Mean=4.7 Median=3.95. Typically I take a midpoint average of the Median and Mean for both in calculating some of my formulae. According to the FRED Data over approximately 100 years the historical AAA Rates are slightly higher than your 4.5% estimate at ~5.58% and 10 Yr Treasuries are slightly lower at ~4.32% using the midpoint average of Mean and Median. Giving an expected alternate _safe_ fixed income return of ~4.95%.

    I think there are other factors you should take into account in your screen as well. The Main one is dividend payout ratio. This must be looked at both quantitatively and qualitatively to ensure that the expected dividend return will continue in the long run. Typically I look for <50% payout ratio with rising dividends and NI that coincide with a typical long term inflation rate.

    Additionally, I look for High FCF and ROE companies and always at the end take a look at Dilution in the stock that may happen through adding Shares outstanding, stock options, and pensions to ensure that future dividend revenue streams won't get spread out to others at my detriment.

    These times have other factors to take into account than just the numbers however. So relying upon qualitative decisions about a company's future prospects such as management and buffet's "wide moat" and constant demand are key as well.

  • Rob

    I find it ironic that Graham fathered a number of successful investors that took his DNA and expanded upon it. Buffett, Ruane, and Schloss all eventually differed in the style of stocks they chose, the weighting of their portfolio, and when and why they sold/held.

    I think the key is to take the best and forget the rest. I don’t have an overly strict criteria…

    For me that begins with low P/E (hopefully) coupled with consistent EPS, compared next to low P/S and consistent revenuse/sales, and ultimately the dividend. To steal a book title from Geraldine Weiss, The Dividend doesn’t lie and as long as the EPS and sales support the payout I’m generally going to buy.

    A rising dividend and/or meaningful share buyback is a strong influence.

    Insider buying/sales don’t carry as much weight.

    Debt is certainly important.

    Cash/debt means more to me than book value.

    Any inconsistencies can be forgiven even when they are cyclical. These hiccups in the data usually influence the length of time I plan to hold, the amount of shares purchased or sold.