Investing in Low/High Debt Companies

  • somrhsomrh
    Posts: 984
    So this discussion started over here and I figure it deserves its own topic. @Gammastyle

    I had pointed to GMO's article that indicates that (contrary to theory) low debt companies outperform high debt companies.

    I noticed that @Antacular had found some research that mid-level debt(?) outperforms low debt (as theory would predict) but not high debt. I'd be interested where you found the study.
  • 47 Comments sorted by
  • somrhsomrh
    Posts: 984
    And I'm not sure where I got the data from but here's the data.

    The EBIT/Assets and EBIT/Sales are calculated based on some simplifying assumptions. (See the definitions for why they are only crude approximations.)

    I brought it up because there's a correlation between Bond rating and Leverage and another correlation between Bond Rating and Profitability. Correlation, however, is not transitive but it would be a reasonable hypothesis to suggest that companies that are more Leveraged may very well have lower Profitability (as measured by my crude EBIT/Assets and EBIT/Sales).
  • AntacularAntacular
    Posts: 159

    Measured via Debt/Equity ratio, What Works on Wall Street, Quantitative Strategies for Achieving Alpha, and Quantitative Value all show that firms employing a reasonable amount of debt outperform both those that use none or are very highly leveraged.

  • GammastyleGammastyle
    Posts: 210
    @Antacular @somrh

    I'd appreciate any input on how I should approach this research.  Right now, I'm pulling in all the financials with Jae's sheet.  I'm a little fuzzy on how I should use that data.

    I think the factor I want to use is Long-Term Debt + Short-Term Debt to Equity.  The goal would be to see two things.  1.) Is there a correlation between total share returns (stock price + dividends paid per share) and 2.) Is there a correlation between the volatility of the share price (incorporating dividends somehow) and the debt level.

    I hypothesis is that there is a inverse relationship between debt levels and return (more debt; lower returns). My second hypothesis is that there is an inverse relationship between the debt levels and volatility.

    I am a little concerned about using stock prices as my base line since it brings in Mr. Market.  I think I'll use that first and then run the same experiment with Revenue, Income, FCF, Owner's Earnings, Dividend's paid and see what comes out.

    The question is now what companies do I include.  

    For this study, I think for a company to qualify they must meet the following criteria:
    1) They have been publicly traded for at least 5 year. 2008.  I'm going to run the study on a 5 and 10 year basis.
    2)  Not a trust (fail the perpetual going concern criteria), closed end funds (they are a function of investing in other companies which could have debt), and banks (entire business is around debt and the financials are goofy).
    3.) USA companies only to ensure the financials are all following the same rules.  


    I'm going to use Jae's sheet to pull the data into a large spreadsheet to run the experiment.

    I don't have the time to pull all 6000 companies in so I'm going to have to sample.  What do you think is a good sample size.  I'm thinking 100 companies debt free, 100 companies with debt over 50% and 100 companies with debt of 0 to 50%.  

    Can you all provide some feedback on this criteria?  Am I missing something?  If can get a couple "I agree"'s, I'll start pulling.  


  • somrhsomrh
    Posts: 984
    Thanks @Antacular

    Did the studies look at both book equity and market equity or are they all one (other the other)?

    @gammastyle

    If you're lazy, you might start with Damodaran's data. That would give you 13 years of US data and quite a few years for some of the international companies.

    One consideration would be if you're looking at book equity, to use book assets instead. Some companies have extremely low or negative equity and I think Debt/Assets would probably work better.

    As far as measuring debt, I suppose there are some other things you might consider. What about preferred shares (even though they classify them as equity, I still see them as debt-like)? A couple of other categories would be things like capital lease obligation and underfunded pensions.

    I would guess you'll get similar results regardless of what you use but some companies are able to utilize other forms of financing (such as their pension funds) which adds leverage to the mix.

    Regarding volatility, are you referring to the stock price volatility or something else? I'd be curious how it relates with something like ROIC volatility. Actually I can probably look at that one since I already did the calculations via Damodaran's data.
  • @somrh
    Thanks.  I'll look at it for a Debt/Assets.  I agree that Preferred should count as debt.  Especially since its something gets paid before the shareholders.

    When I said volatility, yes I meant share price, but I will want to look at other things as well like FCF, Revenue, Net Income, etc to see if these companies are more stable as companies as well.  I know there is a growth element so it will have to be the r-square calculation.
  • @Gammastyle IIRC, most of the "good debt" firms have D/E between .4 and .8, so might want to adjust the parameters a bit. Perhaps use quintiles, with "No debt," "Up to .25," "Up to .5", "Up to .75," and "Over .75."

     

    @Somrh Yes book value for both, although they mention that if you do have market prices for debt, should use those, but book is a generally acceptable proxy.

  • You may want to add some coverage ratios into your analysis to seperate the firms that can take on leverage and those who cannot.  A directional shift in the coverage ratios would also be useful as in most of the failure situations I have seen there is not only low coverage but worsening coverage over time.


    Packer
  • Thanks guys. I think the analysis over various methods will be quick once I have the right sample set. Anyone want to be my objective sample creator?
    I set up Jae's sheet to pull the company data into a database format so it is copy paste after I know what companies to use. 6000 would take way too long so I'm thinking about 100 no debt and 200 debt companies that I can split into quartiles.

    I tried using Finviz, but I get a lot of funds and trusts that I don't think apply. Thoughts?
  • somrhsomrh
    Posts: 984
    @gammastyle

    I'd predict there isn't a decent linear correlation between them so using something like quintiles will work better.

    Finviz has an "ex-fund" filter under industry. Also, if you put them into excel, you can use Excel's filter to remove some you don't want (do you want to include financial companies?)

    In terms of selection, Excel works for that too. Put all the names in one column and put '=rand()' in another column. Then have it sort from smallest to largest (or largest to smallest). Pick the first 100.
  • somrhsomrh
    Posts: 984
    Anyway, since I mentioned Damodaran's data and since I already had ROIC calculated for the 13 years, I did some deciles (I eliminated some that had goofy numbers like negative Debt/Equity... he was using market equity so I'm not sure how they ended up with negative values.) The results are not what I expected:

    These are the median values within deciles (outliers would make some of the mean values somewhat goofy):

    image

    I had to scale the axes a bit (the highest D/E decile is 125%) so you can see the results better.

    image

    ROIC stability is the mean divided by the standard deviation. It represents how many standard deviations the mean is from 0.

    The results are not exactly as expected. I was curious if the Moody's result would carry over (high ROIC for low debt companies). Some of the lower debt companies had fairly low ROICs. Maybe this indicates that companies that have low ROICs are unwilling to finance with debt (since the margin over the cost of debt wouldn't be high) versus companies with high ROICs would be better capable of taking on the debt.

    Definitions:

    ROIC = EBIT(1-t)/(Book Value of Debt + Book Value of Equity - Cash)
    ROIC Mean = Average ROIC for years 2001-2013
    ROIC Median = Median ROIC for years 2001-2013
    ROIC STD = Standard Deviation of ROIC for years 2001-2013
    ROIC Stability = ROIC Mean / ROIC STD (measures how many STD from ROIC mean to 0)

    As a side note, I still don't like his definition of invested capital (sometimes that number goes negative) but since he's already done the work I can't complain too much :)
  • Interesting. One thing though. Part of the idea is that a debt free company does not have interest costs. I think interest expense would change this equation. Even though they would have lower EBIT ratios, they would be helped by no interest. I think you have to go much further down the statements to get a good comparison. Stopping at EBIT takes away a huge advantage of a debt free firm.
    FCF or Net Income would be better places to compare since capital structure matters in this discussion.
  • somrhsomrh
    Posts: 984
    @gammastyle

    I agree that would get different results but that's because netting out interest would already get you a new leverage number. If you use, say, Net Income/Invested Capital, you would already be a measure of leverage since it looks at Equity Earnings / Firm Invested Capital (debt finances it as well)

    What I found interesting in the Moody's calculation (post above) was that there were differences in the unlevered numbers (EBIT/Sales). That's not something you would necessarily expect. Of course the levered numbers will differ as you vary leverage; that's to be expected. A company with leverage will have lower profit margins (net income/sales) than a company without leverage, all else being equal.

    But why would the unlevered metrics change?
  • @somrh
    I'm not sure I follow you or maybe we're looking at this differently.

    As a shareholder, my claim on the cash flows of the business is diminished by the interest expense.  In addition, my claim also is subordinate to interest expense.  Therefore, I have to now overcome an additional fixed cost before there is anything left for me. I think you have to take into account that there is this additional hurdle to you getting any return.

    Excluding that from the equation by using EBIT defeats the purpose of the study before it starts.  A debt laden company SHOULD always have a higher EBIT/XXXX than a debt free company all else being equal.  The point of the debt is to increase return before interest.  Whether they do after interest is what we are trying to prove.

    The other important part will be the variance in share price, FCF, OE.  
  • imageimageimageimageimage
    I took a look at this in a different way.  I screened on FinViz for the following:
    Market Cap: >$2bln
    Country:  USA
    Industry: Stocks only (ex-funds)
    Date 9/2/2013

    I got a total of 1167 companies.

    I split those companies into 10 groups based upon Capital Structure so that it in 10% increments:
    No Debt
    Less 10%
    10%-20%
    20%-30%
    etc.
    Over 80%

    I looked at a bunch of different factors:
    Gross Margin
    Operating Margin
    Net Profit
    1 Year Performance
    1 Month Volatility
    EPS growth last year
    EPS growth next year
    EPS growth Past 5 years
    EPS growth Next 5 years
    Sales Growth Past 5 years
    ROA
    ROE
    ROI
    Beta
    P/E
    Forward P/E
    PEG
    P/S
    P/B
    P/Cash
    P/FCF

    All the data is based upon the median in the group for the reasons somrh stated (mean values came out goofy).

    Margins:


    No debt companies had the highest Gross Margins, and Net Margin.  They were in the middle on Operating Margin.  This seems to show that No Debt companies are able to make up for higher OpEx by not having interest expense.  The trend hold true down the line.  The more debt, the smaller Net Profit Margin is.  This leads me to believe that the more debt a company had, the more it is taking on less profitable operations (because they can with the additional capital). 

    Conclusion:  The more debt a company takes on, the more risky operations it is willing to handle.  A smaller margin means more competition and more risk.  This combined with a higher debt level seems to make them more risky.

    So the question is now whether or not the debt laden companies use that capital to grow the business faster.  In other words, do they have a smaller margin on a larger pie.

    EPS Growth and Sales Growth:



    Over the last 5 years, No Debt companies have seen significantly more EPS growth and Sales growth and there is a fairly strong correlation between debt load and the rate of growth for each.

    So this means that a company taking on debt does not in fact result in faster growth to either the top or bottom line.

    Conclusion: Debt cannot be justified by using the growth argument.

    Returns (ROA, ROE and ROI)


    So the argument could now be that you receive a higher rate of return on investment by utilizing debt and levering up.  That is the common theory right?

    I was surprised by this.  As you can see above, No debt companies win in all three categories.  Also, there is a very strong correlation between debt load and the returns in all 3.  ROE is the weakest one though where middling companies show the weakest results.

    No debt companies find assets that yield the highest return.  This leads me to believe management is much more prudent with capital when they can't fall back on debt. 

    Conclusion:  The leverage concept is flawed in practice.  Higher risk (more debt) does not lead to higher returns.

    Volatility in price:



    As much as I hate Beta in valuations, I thought is was worthwhile here since it is a measure of how much your investments liquidation value fluctuates.  Can you feel safe your investment will hold its value in the marketplace.

    Again, No debt companies had the lowest Beta.  However, this measure didn't seem to follow a straight pattern like the others.  The highest Betas were in the middle.  Also, highly levered firms saw some low Beta groups as well which may be due to the fact that fairly stable companies, like utilities and cash cows can get away with a lot more debt and do so.

    Conclusion:  No debt companies are somewhat more stable investments.

    Price ratios:


    I want to know whether or not I have to pay a premium on the earnings.  While No debt may seem good, I don't want to over pay.

    I appears that there is a slight premium for for lower debt companies, especially for FCF. 

    Conclusion: Investors are willing to pay more for a debt free company's cash flow, but once debt is taken on, there is not much of a difference.


    Overall conclusion:
    Debt Free companies, have larger margins, more growth (both top and bottom line) and have less market volatility.  However, you may have to pay a premium for them.

    When I ranked the groups in each category, I see:



    There is strong performance in a multitude of categories for no to low debt companies.  The more debt you take on, the further down the rankings you go in various categories.

    Keep in mind, this is FinViz data and I do not claim it is accurate or conclusive.  I didn't pull separate company data and took it as is. 

    Also, without doing this study in buckets, I did not find a strong correlation in the data.  The only point that came back with a strong correlation to Debt/Equity is Price/Book Value.  Other than that, I didn't find a correlation over .1 or under -.1.  I would imagine this means we the more debt a company has, the more people think the book assets have value. It could also be that these higher debt laden companies are older and have written down a significant portion of their assets. 

    It is a straight data dump from FinViz based upon the filters above.  I used larger cap companies since I wanted to avoid strange startups from skewing the data.  I figured if it was over $2B, it has been around for a while and the data is probably more accurate.  Small/Micro Caps could have very different results. 

    Here is the data I used in excel.


    Post edited by Gammastyle at 2013-09-02 14:46:40
  • Charts did not pull into the post?!?!  Nor did the file

    They are in there when I go to edit the post.  Anyone have any ideas. My tech skills are just not there today.




  • somrhsomrh
    Posts: 984
    @gammastyle

    As a shareholder, my claim on the cash flows of the business is
    diminished by the interest expense.  In addition, my claim also is
    subordinate to interest expense.  Therefore, I have to now overcome an
    additional fixed cost before there is anything left for me. I think you
    have to take into account that there is this additional hurdle to you
    getting any return.
    So far we're in agreement. But at this point we're talking about equity earnings/cash flows. The problem with "invested capital" is it's not a purely equity denominator. The invested capital is financed by both equity and debt. So you're comparing equity earnings to firm level investment. The appropriate comparisons are either ROE (earnings/BV of Equity) or ROIC EBIT/Invested Capital (or EBIT adjusted for taxes in some way). This gives an appropriate metric.

    This is all related to my recent blog post, this forum thread and Damodaran's blog post. A debt laden company SHOULD always have a higher EBIT/XXXX than a debt free company all else being equal.

     A debt laden company SHOULD always have a higher EBIT/XXXX than a debt free company all else being equal.
    This isn't true. Debt doesn't effect the income statement until after interest charges are taken off. EBIT is before interest charges. So if you have two firms that are identical in every respect except one is financed purely with equity and the other is financed with some debt/equity mix, EBIT will be identical but earnings will obviously be lower for the levered firm. So EBIT margins should be the same but net income margins should be lower for the firm financed partially with debt, all else being equal.

    My overall point is that looking at, say, earnings margin you'll likely get the conclusion that you expect, firms with debt will (all else being equal) have lower earnings margins.

    All else being equal, they should all have the same EBIT margins but the evidence above suggests that they don't. That's interesting. That suggests there may be some other link between firms being more/less profitable and firms deciding to take on debt. (For example, perhaps firms with higher margins may find it easier to pay down their debt than firms with lower margins.)

    Does that make sense?
    Post edited by somrh at 2013-09-02 14:53:57
  • imageimage
    @somrh
    We are in agreement that the debt flows are after EBIT. My poorly articulated point is that a company should see a higher EBIT/XXXX because the point of the debt is to increase returns (leverage).  If a company does not have a higher EBIT/XXXX then it took on additional risk without showing returns for its efforts.  If EBIT is not higher because you took on the debt, then why do it?  You have made the firm more risky and did not yield a return in exchange. 

    That was what I was driving at. 
  • somrhsomrh
    Posts: 984
    @gammastyle

    Is there a link to your file? Or graphics? I'm not seeing either if there is.

    1) Regarding the above, the gross margins is an unlevered number so all else being equal, you'd expect there to be no difference between high/low debt regarding gross margins. So I think it's interesting that lower debt companies have higher margins.

    So if there's excess return there, it may partly come from the higher gross margins instead of lower debt.

    2) The sales growth is interesting. I'd have to think about the earnings number a bit.

    3) ROA I think is useless as commonly measured. ROE should be higher for higher for debt companies. I'm not sure how finvis measures ROI.

    4) One issue with beta is that firms with lower correlations with the market have lower betas since beta is proportional to correlation. Beta is just a worthless metric imo :)

    For example, IBB (biotech ETF) has a beta of like 0.8 in spite of being more volatile than S&P 500.

    5) In theory, the equity based multiples should be lower for higher debt firms. If there's not much difference between the debt groups may be interesting.

    . . .

    Regarding correlations, you may end up needing to flip some figures to get the correlations right. That's why in academic studies they look at "book to market" instead of P/B.
  • somrhsomrh
    Posts: 984
    @gammastyle

    I'm not sure about your point.

    Let's suppose we both find a rental property (perhaps they are duplexes so they're basically the same). You finance purely with equity whereas I put 20% down and take out a mortgage.

    I'm claiming that the EBIT multiples will be identical since they are the same properties. So their rents will be the same, their expenses will be the same and as a result their EBIT's will be the same.

    So if I'm comparing EBIT/Invested Capital (invested capital would be like the cost of the house), those should be the same. I'm not understanding why you think they would be different. Are you saying EBIT would be different or are you saying Invested Capital would be different? I think they ought to be the same.

    Where the difference will be is with ROE. Your ROE will be more akin to the Cap Rate. My ROE will be higher (provided my borrowing costs are lower than cap rate).

    So if the cap rate is 6%, you'll earn about 6%. If I borrow at 4% and put 20% down, I'll obviously a higher rate of return (14%) for taking on higher risk.
  • somrhsomrh
    Posts: 984
    @gammastyle

    Do you use the "Attach a file" bit just below the post? I've used it once and I think it worked.

    There's something goofy with the image URL. Here's a link to the first one. That's probably why it's not showing up. Check the link on the image and try replacing it maybe.

    All of the rest of your <img> links say this at the beginning: "denied:data:image/png;base64". I'm not sure why.
  • Those images are Excel Charts. I copy pasted them into the post. I guess that doesn't work. How would I put them in? The file attachment wasn't working either. Tried to attach the data and it never loaded. Strange. Any help would be appreciated.
  • somrhsomrh
    Posts: 984
    @gammastyle

    What I do is I copy the image and paste it into paint. Then I save the image and upload it to image hosting site. I'm not sure about the file attachment thing though.
    Post edited by somrh at 2013-09-02 17:07:08
  • somrhsomrh
    Posts: 984
    @gammastyle
    If a company does not have a higher EBIT/XXXX then it took on additional
    risk without showing returns for its efforts.  If EBIT is not higher
    because you took on the debt, then why do it?  You have made the firm
    more risky and did not yield a return in exchange.
    OK, so I wanted to get back to this. I think the first thing you need to do is look at the unlevered numbers (which I'm still claiming ought to be the same) and then see if there are differences. If there are differences, then it indicates that debt levels is correlated with that difference for some reason and you might start asking why. (example below).

    In terms of the above, I would think ROIC would simply need to be higher than cost of debt (ideally higher than WACC). If I can borrow money at 6% and put it into an investment that earns 8%, I'll be OK and be adding value from the spread (assuming that the 8% is stable enough to not cause me to dip to that 6% or below level too often).

    To take an example, take the ROIC stability chart I posted. The unlevered numbers are different so we might suspect that there's some relationship between debt and ROIC stability. So here's a couple potential hypotheses that might explain the relationship.

    So companies with lower ROIC stability are going to be more likely to have negative EBITs than companies with higher ones. So one possible explanation for part of the data is that companies with low ROIC stability are playing conservative and not adding too much debt because they don't have the stability to weather a storm. Companies with high ROIC stability would be more capable of taking on a higher debt load. I think we talked about real estate assets being a good example of that. So real estate investments can take on a higher debt load and still be OK due to the overall stability of their profitability.

    That hypothesis might explain the left side, but the right side conflicts with that. So either the hypothesis is wrong or there's more going on here. One possibility is that some companies don't think like the conservative ones on the right side and only take on higher debt levels when ROIC stability is high. So some of these non-conservative companies take on high debt in spite of low ROIC stability. The run into trouble doing this and need more debt to stay alive.

    So those two hypotheses might explain why there appears to be a relationship between debt and ROIC stability. (Granted, we'd probably need to collect more data and actually test those hypotheses.)
  • imageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageSo let's collaborate and figure out what the best way to test our ideas is. 

    I think we are talking around each other on this EBIT/XXXX idea.  Same idea from different viewpoints.  I still think using EBIT when talking about debt levels is not correct.  Interest Expense is a critical part of the discussion.  Are firms getter better at a high enough rate to overcome the taking on of debt?  Interest Expense is a very real cost in this discussion.  It's almost like talking about Sales without examining the costs on the Income Statement. 

    How can we test our hypothesis?  What is our hypothesis?

    I think we agree that borrowing at 6% and getting 8% is good in theory.  The caveat that ROIC has to be stable enough to do that is where I think the problem lies. 

    I think (hypothesize)  that companies do not take a large enough spread between their debt level and expected return to overcome the lack of stability.  The risk taken on with debt overcomes the spread difference. 

    How do you prove that?  If the leverage concept were true we should see several things happening potentially.

    Margins would be higher.  The debt laden company would be able to reduce costs by borrowing to invest in more efficient operations.  Also bigger would mean they would have more bargaining power with their suppliers.

    Growth would be higher:  By borrowing they should be growing both sales and/or EPS.  They are able to capture a larger piece of the market so sales should increase as debt increases.  EPS should increase because they are capturing the sales with borrowed money which should be cheaper than organic growth.  This could happen with a shrinking margin since they are growing the pie.

    Returns would be higher:  ROE and ROI should be increased.  As I borrow, I have to put in less equity for the same amount of capital.  Even if the operations were identical (your real estate example), ROE would heavily favor the debt laden company.  ROI should increase as well (depending on how its calculated).

    For me, the burden of proof is on the debt companies to show me something I am getting for taking on the additional risk.  One of those things above needs to happen or else why borrow.

    Either the company is growing more, has higher margins or is giving me a higher return. 

    The data I pulled (it still won't let me attach) shows that a debt free company outperforms in all those metrics.  That runs counter to traditional finance. 






    Debt Level Study Post.pdf
    194K
    Post edited by JaeJun at 2013-09-04 15:19:27
  • somrhsomrh
    Posts: 984
    Do you have a google account? Maybe you can upload the spreadsheet there and share it via link (don't allow editing of course).

    One reason I like looking at the enlevered numbers like EBIT/Invested Capital is it already provides a quasi-answer to "how much more return should I be getting for the added risk of leverage."

    I can't see your data but let's suppose the following happen. Suppose that you have a low debt firm that earns 12% ROE and a high debt firm that earns 14%. So the debt is obviously enhancing returns. That's good. But how much higher should it be given the added risk?

    The nice thing about looking at ROIC (and I mean the unlevered calculation, I don't like the version with net income in the numerator) is that it allows us to compare the two as if they were unlevered. That doesn't directly answer the question of how much ROE they should be getting but it does allow for comparison.

    I think if I'm understanding you, your data suggests that levered firms have lower ROEs which is clearly not good. Obviously it should be higher - how much? we may not know - but we do know they should be higher, certainly not lower. If they are lower that gives one clear reason to prefer the unlevered firm.

    That also might explain why returns end up being higher for the lower debt firms. Perhaps the correct causal relationship isn't low leverage = high returns but high ROE = high returns.

    Part of looking at the hypotheses will be to see how relationships are correlated and determining which of the various correlations (if any) is the causally contributing.

    So let me propose this hypothesis more specifically: The higher returns of the low debt firms is related to their higher ROE, not the lower leverage. That would be something you could test for.

    In that case, if you're screening for stocks, should you look for low debt or high ROE? Which one is the causal variable?

    The higher ROE might explain other things like sales growth as well. Growth can be expressed as:

    Growth = Investment Rate x Return on Capital

    Where ROC is either ROE or ROIC. More ROE here would be causing the higher growth.

    Of course we'd have to test all of these relationships. Perhaps we can sort the debt declies into ROE deciles. Or perhaps high ROE is a better indicator of returns, growth, etc than debt is. Or perhaps there's some other variable at play that we haven't considered.

    But sorting for ROE might show that higher debt firms do outperform lower debt firms when we first sort for ROE. Obviously you'd have to try it out and see what happens.
  • somrhsomrh
    Posts: 984
    So it looks like Jae got your file up.

    Here's a potential study (and made up results I had in mind). Suppose I sorted stocks into 5 ROE quintiles and 5 Leverage quintiles. And here are my hypothesized returns for these stocks:

    Chart Includes Numbers I made up; no study was done :)
    image

    So in my theoretical study, standard finance is vindicated; higher debt results in higher returns. The reason why it didn't work out that way when we sorted only for debt is a lot of high debt firms happen to have low ROEs so their returns are lousy (say 6-8%) while a lot of low debt firms have high ROEs so their returns are decent (say 12-16%).

    Hopefully that makes sense. That would be a potential hypothesis and test that could well vindicate standard finance theory. We'd have to run the test of course (and not just use some made-up numbers).

    As a side note, when I have more time I'll take a closer look at the file. I did want to make one comment. Regarding the P/E ratio chart, P/E should actually be lower for higher debt firm (high risk = lower P/E). So the fact that they're all almost the same indicates you're paying a premium for high debt.

    That's why I've been preferring metrics like EV/EBIT over P/E metrics because the unlevered metrics should be identical whereas the levered metrics (like P/E) should be higher for the low debt firms (lower risk).
    Post edited by somrh at 2013-09-04 16:21:16
  • I think one aspect you may want to consider is comparing RoEs and returns in the same industry, Otherwise you may have other non-leverage factors influencing results (like the industry dynamics the firm competes in a and how bankable the assets are that the firm holds).

    Packer
  • somrhsomrh
    Posts: 984
    @packer

    Yeah that's another good consideration. That brings up another issue I've often wondered about: how good is depreciation at measuring maintenance CAPEX? I'm guessing the answer probably varies by industry.

    I don't think there's any way around any of these problems either. Well, there it is but it's not practically possible. The proper way to do a study would be to randomly assign individuals (in our case, individual company's) to the "treatments" we want to test for (varying levels of debt) and then check the result. The problem is that the company's themselves decide which group to belong to so we end up having to play games of controller for other effects since we really don't have control over the variable we're testing.

    *shrug*

    I guess we just make do with what we have and adjust accordingly.
  • imageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimage@somrh
    I did your chart on all companies above $1B Market Cap (the prior post was with companies over $2B.  I wanted a larger sample since I'm doing so many buckets.  The problem with going lower is almost 25% of the companies were in the debt free arena.  That made it tough to put them in quintiles.

    If I understood your chart above, I think I did the right.  I took the median for each of the 25 groups.  So in the chart, a company in the bottom 20% of ROE and in the lowest 20% in debt load (smallest amount of debt) is 5/5 (-7.7%).  Highest Debt/Highest ROE is 1/1 (43.5%).


    (edit:  chart didn't pull again so I did it by hand)
                                                       ROE
    Debt Level      High   Mid-High   Mid   Mid Low    Low        Overall
    High              43.5%    16.1%   11.7%    6.3%    -15.8%     12.8%
    MId High        29.2%    17.1%   11.2%    6.4%     -2.3%     10.5%
    Mid               30.1%    16.9%   11.3%    6.8%     -1.2%      10.8%
    Mid Low         29.9%    17.3%   11.4%    7.0%     -1.4%     11.1%
    Low               27.9%    16.5%   11.6%    6.3%     -7.7%     13.0%
    Overall           31.3%    16.9%   11.4%    6.7%     -4.7%     11.4%


    I think this shows what you wanted.  It is as expected for High ROE companies.  Companies with high debt have a much higher ROE, but once you get out of the top 20%, it sort of levels off.  There is not much difference between debt level and ROE in the middle 60%.  In the lowest quintile (worst performers) the high debt companies are clearly worse and then the lowest debt ones. 

    Overall, Low debt companies outperform by a hair which is surprising given traditional theory. 
                                                       ROI
    Debt Level      High   Mid-High   Mid   Mid Low    Low        Overall
    High              21.7%    12.0%    8.1%   5.0%     0.03%     6.4%
    Mid High        18.3%   12.3%     8.4%   5.5%     0.02%     6.9%
    Mid               21.2%    12.4%    8.5%   5.0%    -0.05%     8.5%
    Mid Low         20.4%   12.8%     8.4%  5.1%     1.1%      10.5%
    Low               22.1%   13.0%     9.3%   5.1%    -7.8%     13.1%
    Overall           21.1%   12.6%     8.5%   5.1%    -.03%      8.6%

    ROI shows a little different view.  There is no real changes between the quintiles with the exception of Low debt/Low ROI.  (edit: I added the overall columns to this and ROE).
     
    I'm still digesting this, but I thought you would be interested as would anyone else.

    Jae has been pretty quiet on this one.  I'd be curious what he has to say.

    imageimage
    Post edited by Gammastyle at 2013-09-06 15:29:20
  • somrhsomrh
    Posts: 984
    @gammastyle

    That wasn't exactly what I had in mind (I was considering stock returns) but I think it may end up illustrating a related point. Part of what's happening is that many of the high debt firms have low quality assets. Leverage ends up working against you there so you see that ROE declines with high debt but increases in the highest quintile where the asses are better quality.

    And going back to @packer 's comment regarding industry, I wonder what the industry makeup looks like. My guess is that the higher debt firms have a lot of low return on assets type businesses (finance, real estate, utilities). But these cash flows are more stable so they can typically lever them up. So they may only earn, say, 6% on assets but if they're borrowing at 4% and lever up, they can increase their equity. The other issue is that spreads may tighten which could decrease ROE which gets to this. . .

    I put a blog together with a related topic (mainly because I needed to derive some equations) which illustrate part of what I'm getting at. Here's a link to the blog: Relating ROE with ROA and Leverage

    The main result is this:

    ROE = A/E ROA - I/E

    where A is assets, E is equity, and I is interest charges.

    The main suggestion is that, yes, leverage increases ROE when the assets are quality but high debt firms may, for various reasons, be starting out with lower quality assets in which ROEs won't be higher (or may end up lower than their unlevered counterparts). Since the leverage works based on the spread (ROA - interest rate), if you start out with lower quality assets, it reduces spread (or turns it negative). Also, higher debt firms tend to pay higher interest rates (increased risk) which further reduces the spread.

    Generally what I'm thinking here is not so much in terms of low/high debt but also related to the quality of the assets. Low debt + low quality assets will be low returns. High debt with high quality assets will make for a better investment.

    The quality of the assets would have to be related to both the returns and the stability of the assets.

    I'll get back to this when I have some more time.
  • I think this is my point as well.  Instead of increasing returns, companies are using debt to buy inferior assets.  They are taking on more risk for a diminishing return.  I hypothesize that this is due to a less prudent attitude when you have access to debt.  Companies are much more loose with the cash the more access they have to it.  The debt free companies are much more careful about where they allocate capital.  This is a behavior matter and not one that can be proven with some academic formula.  In theory, they should enhance returns when they lever up.  In practice it seems they do not.

    I think the table shows that companies taking on debt does not, in fact, create higher returns as it should and how it is sold.  It simply becomes a balancing act where the interest payments wipe out any additional returns the investor receives.  Of course there are going to be specific companies that buck the trend, but on the whole, there seems to be no correlation between growth and debt levels.  There also doesn't seem to be a correlation between returns and debt levels.  These are two fundamental reasons to borrow in the first place.  

    As an investor that can buy a myriad of stocks and move from one to another with ease, what exactly am I buying (with my additional risk) by allowing management to borrow against my assets.  It doesn't appear to be growth, or enhanced returns or increased margin.  If I'm not getting any of those, why would I want management to borrow.  Keep the cash flow coming to me, not the bank.  They're not enhancing my returns.

  • somrhsomrh
    Posts: 984
    @gammastyle

    Oh wait, are those the 1 year performance figures from finviz? I guess I'm not sure what calculation you're doing. I wouldn't give much consideration to 1 year performance as it's quite noisy. If you only look at 1 year, depending upon which data set, you might find a number of things that don't make any sense. Longer term horizons would be required.

    But I think we are in agreement about the asset quality. My hypothesis is that higher debt firms, in aggregate, tend to have lower quality asset. Ultimately, I suspect that it's not debt that matters but that it's asset quality matters. That's what I find appealing about Greenblatt's "magic formula" (certainly not the name :) ) since it gives you an asset quality metric and an unlevered valuation metric.

    I guess I don't see anything wrong with filtering out high debt firms but I'm not convinced that doing so should be required. I think if you filter by quality, you get rid of the poor performing ones and then you'll see some enhancement in returns. I think part of what's happening (and that's being seen by some of the data) is that there is correlation between the debt and asset quality. But I'd be just as inclined to filter out the low quality low debt ones as the high debt ones as far as stock selection.

    In terms of sales growth, I would emphasize asset quality there as well since growth really only adds value when return on capital > cost of capital, regardless of whether it's financed with cash flows, debt issuance or equity issuance.

    I guess in terms of ownership, would you object to the company issuing shares of stock to finance instead of debt? Because that would lower your claim on the assets as well?

    I also suspect we may view debt slightly differently. I view bondholders as a kind of owner of the business. So the firm's assets aren't purely owned by the equity stakeholders; the equity owners have a residual claim on the firm's assets. Debt holders also own the assets in a different sense (they typically have little say over the company and if the company performs very well, they don't benefit from it other than making sure the contractual payments come in.)

    But otherwise I agree with you, financing low quality assets with debt (or equity) is not something I would want to invest in. But I'm perfectly OK with firms that use debt to enhance returns.
  • somrhsomrh
    Posts: 984
    To use another example of the aggregation problem, consider high book to market stocks ("value"). In Piotriski's paper, he notes that high B/M stocks tend to be, in aggregate, in financial stress. So his goal was to come up with a metric which separates stocks that are in distress from those that aren't.

    He was essentially responding to Fama's claim that high B/M stocks outperform due to higher risk.

    I'm basically suggesting the same thing here. I think part of the reason low debt firms, in aggregate, underperform, is that they tend to own lower quality assets to begin with. Focusing on firms with higher quality asset (whether we look at low debt or high debt) should (I hypothesize) give greater returns.
  • somrhsomrh
    Posts: 984
    Another quick thought on low debt investing. Let's suppose you have a company with a debt load. Let's suppose the debt load is D and the market value of equity is E with a firm value of V = E + D.

    Now you happen to not want a high debt company. Let's suppose you want to take a $100 unlevered position in the company. Why not do this?

    Buy D / V * $100 of the firm's bonds and E / V*  $100 of the firm's equity.

    Now none of those debt payments would be going to someone else because you're effectively the bond holder here.


  • JaeJunJaeJun
    Posts: 2,503
    My theory is that it will depend on the industry.
    This really needs some data mining to come to a conclusive answer because I'm not surprised by Gammastyle's chart. There are leveraged companies that earn great returns, but at the same time, not all companies can achieve good returns. Subpar returns on high debt is quite normal.

    While in theory they should be earning higher rates of return, I really do wonder how true that is.

    I like somrh's statement.
    "Focusing on firms with higher quality asset (whether we look at low debt or high debt) should (I hypothesize) give greater returns."

    A great company with no debt can earn awesome returns, while crap company with high debt may earn crap returns as it continues to struggle.

    Probably getting off track to where this discussion is at, but I don't think I've come across many high debt companies capable of maintain high returns. The returns are usually cyclical or short lived.
  • imageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageimageI've been pulling data on this for the last couple days.  I'm using the spreadsheet to do it and getting the financial data into a database type format (sorry Jae if I break your quota).  It takes a while to pull this manually.  I've gotten about 500 companies done so far this week. 

    I'm working my way down by market cap.  So I started with XOM and am working down as I type this.  I'm in the $7B market caps now. 

    Once I have all the financial data in place, I'd be curious how you'd guys want to see it.  What am I comparing? 

    The difficulty is that some companies only go back so far.  I can get 10 yrs on some companies but only 1 or 2 or 5 on others.  Quite a few are coming the new big dogs on the street (FB for example)

    I also don't want to have survivorship bias in this which may be unavoidable.  It works both ways though.  Bad companies are dead and good companies may have been acquired by others in the meantime.  That may mean we are missing quite a few outliers.  That is part of why I've only looked at large companies.

    My hypothesis is mainly around the point that companies that borrow do not typically invest in assets that are of equal quality to what they would if they didn't borrow.  Thus, the gain you get from any leverage is eaten away by interest expense.  This would hold with theory the NPV mindset that you invest in anything with a positive NPV.  However, the failure in this thought process is that the company is adding additional risk to their investments that isn't captured in the NPV.  Basically, if they can borrow the capital, its all good.  I disagree with this fundamentally.  When companies have debt, they may operate sub optimally because they have to take a worse deal just to make sure the interest expense is covered.  I think this happens much more often than we realize.

    I think this is a major failure in Modern Finance and the whole concept of WACC.  I'm curious how you guys would like to see the data compared.

    I also want to know what companies should be taken out.  Banks? Trusts? LP's? REITS?

    I think there is some goofiness to thier financial that may mess up the research. 
  • somrhsomrh
    Posts: 984
    @gammastyle

    I would think removing financials and REITs would probably be a good idea. Most trusts are going to be lower quality assets (most are selling commodities).

    In terms of metrics, some unlevered return on capital metric should be looked at. This way you can actually directly compare the high and low debt ones. Maybe something like EBIT to Assets would work. There are plenty of different options of course.

    Regarding your hypothesis, I guess I'm not sure how we would test it from the alternative I suggested.

    So let's suppose that, in aggregate, higher debt companies tend to have lower quality assets. But what caused what?

    Your hypothesis, if I understand it, is that the lower cost of debt encourages companies to take on lower quality assets so long as the spread between the return on asset and the cost of debt exists. Another way to look at your hypothesis (perhaps an extension) is that since the high debt causes the equity to be riskier, cost of equity will be higher so expanding by issuing equity won't be as profitable.

    Earlier I suggested two possible hypotheses which suggest the causation works the other way around. One is that companies with higher quality assets will generate more cash flows which can be reinvested so they won't need to take on as much debt to finance operations.

    I also suggested that companies with higher quality assets will have more cash flows to pay down debts faster than companies with lower quality assets.

    So which is it and how do we test it? I guess I'm not sure there. I'm guessing we'd probably need to collect some more information (what information? I have no idea).

    I'm guessing reality is probably a mix of all of those scenarios and I think it will probably depend on a case by case basis as I think each situation will be slightly different.

    The interesting question to me is that if we just focus on high quality assets (say by looking at high EBIT/Assets), will high debt be vindicated? Because, frankly, I'm to the point where low quality assets aren't going to be worth the trouble, regardless of the how much debt. To that point, a power point I linked to from McKinsey:

    Thinking About ROIC and Growth

    One of the most interesting charts is on slide 8 which shows that 43% of the low ROIC companies still had low ROIC 10 years later. Contrast that with 50% of the high ROIC company's that have high ROIC 10 years later.

    So if you buy a low ROIC company (regardless of how cheap and how low the debt), how are you going to do if it's still a low ROIC company 10 years from now?

    On the other hand, if ROIC > 20%, unless their interest charges are somewhere near 20%, it should be enhancing returns.
  • somrhsomrh
    Posts: 984
    @JaeJun

    The cyclical thing is an interesting hypothesis as well. One of my suggested hypotheses was that some take on debt due to financial distress (e.g. the bottom of cycle hurting them) and they may need to take on distress debt.

    Of course there are plenty of relatively stable businesses with high debt (REITs, utilities, etc). All low return on capital of course :)
  • JaeJunJaeJun
    Posts: 2,503
    @gammastyle

    You're alright with the data. Well within the limits. The email I sent out last time was related to some people grabbing anywhere from 500-800 stocks a day for a couple of weeks.

    Let's see what you come up with though.
    Post edited by JaeJun at 2013-09-20 09:44:36
  • I've got a bunch of data and I'm churning it into information.
    The preliminary data is very surprising. I'll post the spreadsheet once I've got it prettied up (it's a mess). I'll need a second set of eye on it too.
  • somrhsomrh
    Posts: 984

    And I'm not sure where I got the data from but here's the data.
    OK, I finally found the article from where this data came:

    The Distribution of Common Financial Ratios by Rating and Industry For North American Non-Financial Corporations: July 2006

    Say that 10 times fast.
  • JaeJunJaeJun
    Posts: 2,503
    Gammastyle sent me his paper to review. I think he'll post it here soon, but he also gave me permission to post it on the blog. Lots of work went into it for sure. Like everything else, it's open to interpretation, but it's a great effort trying to answer the original  question posed in this thread.
  • imageimageimageimageimageimageimageI'm going to let Jae set it up as a blog post whenever he's ready.  I was interested in the results, but as Jae said, it is open to interpretation. 
    Your thoughts are always appreciated. 
    Also, I don't want anyone to think that I am advocating this as the be all end all of picking investments.  There are tons of companies that run counter to this data.  I just found it to be a great place to start in finding companies I am successful at valuing.
    There are many ways to skin the cat.  This is just the style I have been successful using.imageimage
    I hope it sparks some discussion.

  • somrhsomrh
    Posts: 984
    Looking forward to it.
  • Not sure if spam or not above.

    So I've been looking at building my portfolio around this thesis.  I've found you can build a pretty well rounded portfolio from debt free companies.  It isn't a strategy that forces you into a sector.

    So far, I've got the following in my portfolio:
    $ACN - Accenture (Consultants/Outsourcing, very stable growth last 10 years)
    $RGR - Strum Ruger (Gun maker that has been consistently growing)
    $MYGN - Myriad Genetics (Do genetic testing for diseases, this is going to be SOP soon)
    $VIFL - Food Technology Services (See Video I did)

    There are quite a few I'm looking at, but the price is a little steep right now.  On the list are:
    $CBOE - CBOE Holdings (The options exchange, take about selling shovels in a gold rush)
    $V - Visa (get a cut of every transaction)
    $CHRW - CH Robinson (shuffle papers for shippers)
    $QCOM - Qualcomm (leading technology company, into all sorts of things)
    $RL - Ralph Lauren (solid brand)
    $DLB - Dolby Labs (Patents galore and are into all types of media)
    $ATVI - Activision Blizzard (Game maker in a growing online gaming space, large following for new products)
    $GOOG - Google (When your name is now a verb, you have power in the market)
    $TNH - Terra Nitrogen (Huge dividend, growing population means we need more farm production)
    $MORN - Morningstar (Brand name, trusted data source.  @JaeJun: How high are your switching costs?)
    $IPAR - Inter Parfums (Perfume maker with nice cash flow, cheap)
    $EXPD - Expeditors International (Another paper shuffler for shipping)
    $PETS - PetMeds (Niche market and people are more and more concerned about pet health)
    $HIBB - Hibbert Sports (Sporting good store, nice financials)
    $BBBY - Bed Bath and Beyond (See video I did)

    Some of these are in the buy range, but I'm trying to spread out my position entries because I fear the overall market. 

    The ones I see as fair value/overpriced are my targets in the next downturn. 

    Anyone have any thoughts on any of these?  There isn't anything fancy here.



    image
  • somrhsomrh
    Posts: 984
    @gammaystyle

    Yeah, the above looks like spam.

    A couple comments.

    CBOE - I think there was an investment thesis a couple years ago at Ira Sohn.

    CHRW - I actually started researching this one a little while ago. It is a bit pricey but it's a pretty solid company. Dividend + buyback yield is about 5.5% so it's probably not too bad. Low capital investment requirements and solid cash flows. It might be worth starting its own thread.

    QCOM - I see this one come up all the time and never get around to looking at it.

    DLB - Jae has some posts on this one.

    GOOG - It's pretty pricey. There's also class structure to factor in. I've been tempted a few times. I wonder if that $1000 mark would a buy-in point for some people. Maybe you could write puts on it.

    PETS - same comment as QCOM

    HIBB - There was a post or thread on this somewhere around here.
  • JaeJunJaeJun
    Posts: 2,503
    RGR: the whole gun ban thing took this stock up big time. Now that it's settled down, how do you see future growth in the gun industry?It's an overloooked and hated indusry so there is always opportunity to buy again too.

    V: always tempted. But is the current price "fair" or is it a "premium". What about MA and AMEX?

    DLB: yup written about it a couple of times. No longer hold though.

    ATVI: well aware of this company because I like their games. Haven't played any of them for a long time, but I know how hardcore the following is for things like world of warcraft and diablo.

    MORN: the switching cost isn't extroadinarily high. It will hurt and it will take up time for institutions and businesses like mine to switch, but very possible.
    If I could get better deals from other reliable sources, I would do it.

    The main moat comes with aggregating and standardizing financial data. The financial data business is extremely difficult to do well. I see a lot of smaller companies trying to take advantage of free data and SEC filings, but they won't work.

    IPAR: had a similar position several years back. Didn't like the business though. Too many fads and tacky names. Every celebrity has a perfume it seems.
    Anyone want my Paris Hilton fragrance :)

    PETS: it's been on the value list for a long time. Don't have much info on this.

    HIBB: written about it a couple of times. Awesome retailer. Waiting for this to come down. Maybe HIBB will be my first successful retail investment if it comes down in price.

    Nearly read it as BBRY which is a yuck.

    BBBY is a good retailer. They squeeze vendors like crazy, but they make money from it.




Welcome Value Investor

Signing up is very quick and easy or log in with Facebook, Twitter or Google to start discussing in the stocks forum!

Login with Facebook Sign In with OpenID Sign In with Twitter

In this Discussion

Top Posters