Cram and Jam

Insurance is probably the most complex industry as far as accounting goes.  Why?  When you sell the policy, you have a vague  idea of what the costs will be, and when those cash flows will occur.

That leaves room for a wide variety of games as far as the accounting goes.  Because hitting operating return on equity targets is often the “be all” and “end all” of management reporting, one of the holy grails was taking capital losses and turning them into operating income.  Net result on income is zero, but it looks like you are making a lot of returns off of operations.

At one company that I worked for, the new CEO want to great pains to declare how ethical the new CFO was.  I murmured to my boss, “Not ethical, but clever.”  He gave me a smile.  She had pulled that very trick, and if one reconciled the Statutory and GAAP accounting, the chicanery was obvious.

At AIG, my managers were quite concerned about what went above the line and below the line.  If an accounting item didn’t figure into net income my managers didn’t care about it, even if it diminished shareholders equity.

As an investor, this made me skeptical about income statements.  But if you don’t have an income statement, what do you do to estimate profitability?

Well, you could look at the change in tangible net worth due to common shareholders, and add back dividends, including the value of spinoffs, and net money spent on buybacks.  That is what a shareholder earns, in book value terms.  Back when I was an analyst of the insurance industry, there were companies run by value investors that would present their returns that way showing the the growth in fully converted book value over time.  In a sense , Berkshire Hathaway does that as well, but it doesn’t pay a dividend, so it is simply the increase in book value.

In the short run the market is influenced by net income due to common shareholders.  But there is a difference between the two measures of income, and I call the difference “cram.”  Cram is the amount of extra income reported through the income statements that does not makes its way through the balance sheet.

That said, I have another measure that I nickname “jam.”  Jam is the amount of money gained/lost from buying back stock.  In general, when companies buy back stock they dilute value for investors.  Better to retain and reinvest.

How do I know this?  I have been working on an accounting quality model, which is still a work in progress.  An aside, I have had my share of calls from consultants who tell me they have an earnings quality model that covers the whole market.  When they call me I ask them how they analyze financial companies.  I get the intelligent equivalent of a shrug.  The reason is that accruals on the financial statements of industrials and utilities are quite similar, but for financials, they are quite different.

Here are some of the results of my model on the S&P 100:

The data covers the last 4 3/4 fiscal years.  Why did I use fiscal years? Because data capture with companies is most complete at fiscal year ends, when they file their 10Ks.

What did I find?  In general, most companies lose money off of buybacks, whether it is 24% of cumulative net income, or 32% of final tangible net worth.  Individual company performance varies a great deal.  More surprising to me was that cram on average was only 1% of cumulative net income.  Maybe GAAP isn’t so bad on average after all.  But averages conceal a lot of variation — I would not want to own companies that lose a lot of money off of buybacks, or those that inflate net income versus growth in tangible book.

If buybacks ceased, companies might have a lot of slack assets on hand.  I know that companies keep themselves slim to avoid takeovers,  A large amount of slack assets invites others to come in and buy the assets to manage them.  Still, it seems that most buybacks waste the money of shareholders.  This seems to be another example of the agency problem, wheremanagers take an action that benefits them, but harms shareholders.

I would be negative on both cram and jam.  Good companies don’t report earnings in excess of what shareholders obtain, and they don’t buy back stock except when it is cheap.

Full disclosure: long ALL COP CVX ORCL PEP

13 Comments

  • tom brakke says:

    One more example of why you are the best investment blogger extant.

  • But What do I Know? says:

    This is great stuff, David. IMHO, “ethical” managers buy back stock for two reasons–because everyone does it, and because keeping the stock price up means they keep their jobs (that is, they won’t receive a hostile takeover offer). There are those managements who buy to increase/maintain the value of their own shares, but let’s ignore that for now.

    What is interesting to me is that companies are not motivated to get their best price for their stock during their buybacks–it seems to me to be more a matter of tactically allocating cash flow instead of strategically deciding when to “invest” in their own stock.

    These “non-economic” buyers distort stock prices with their purchases, which subjects the market to sudden dislocation when they stop or cut back (as many announced they would in 2008Q3). Just a theory I’ve been working on. . .

  • Bob_in_MA says:

    “Cram is the amount of extra income reported through the income statements that does not makes its way through the balance sheet.”

    David, in the chart are you listing cram as a negative? I mean, does the entry “AA (351)” mean Alcoa reported 351 on its income statements that did not make its way through the balance sheet? Or the opposite?

    Sorry if I’m being dense.

  • Bob, Negative cram is good — it means that tangible book value net of distributions is growing faster than net income. Gain on buyback/issuance we want to see positive — that means companies are intelligent about when they issue or buyback shares.

  • burt says:

    Great stuff! A lot of people have talked about this problem, but this is the clearest explanation I have seen. I would expect you could use this table as a long/short investor and do quite well.

  • najdorf says:

    Interesting model. Are MO and TWX’s cram numbers distorted by spinoffs though?

    It’s pretty amazing how bad the buyback numbers are, given that most companies had nice opportunities to buy back at good earnings yields in 2008/2009 and many had the cash on hand to do so. The fact that cash-cows like T, MMM and MSFT that have plenty of cash on the books and can issue decent-term debt under 5% (even in the midst of a crisis) failed to buyback mountains of stock at 10% (and growing) earnings yields tells you everything you need to know about the tendency of corporate management to be fearful when others are fearful, even when there’s no reason to be so in their actual business situation. Same thing overseas with NVS: management gets a gold star for making a $50 billion acquisition from a smart seller at 20x earnings while purposefully failing to buy back their own stock under 10x earnings.

    On the plus side, those who did buyback aggressively may ultimately be providing beneficial leverage to their shareholders, even if they didn’t maximize possible profits or make paper profits to date. For instance, I don’t have much of a problem with XOM buying shares in the 70s even though it shows a current loss. If KFT were run properly I wouldn’t mind it buying shares in the low 30s, but again there’s the acquisition problem…

  • najdorf — MO and TWX include the spinoffs, as do five other companies in the S&P 100. In earlier versions of the model I omitted goodwill and spinoffs, and the results were nutty the other way.

    That introduces a timing issue into the calculations. Spinoffs unlock value that was created much earlier. Some of that factors in through goodwill and net worth writedowns, but most through the value of the new company’s common stock.

    Now, if TWX went back prior to the AOL merger — the results would look different. Very different. ;)

  • Jay Weinstein says:

    I have followed stock buybacks for years and offer the following observations:

    1) You cannot believe how little managements know about the “value” of their own shares. Like any other investors, they get carried away when things are good and overly depressed when things are bad. I have never found insider buying or selling to be useful for that reason.

    2) As a result, they consistently buy high and sell low, as David’s chart shows. I asked quite a few managements early last year about buying shares at the depressed prices, and they were simply terrified to do it. Najdorf is spot on.

    3) Managements HATE to pay dividends–they use the “tax efficiency” argument to favor buybacks versus dividends, but the reality is, they don’t like to give money back to shareholders. It is a tacit gesture that they don’t know what better things to do with cash flow. They would rather make a stupid acquisition to grow their kingdoms.

    4) When you do find a management team that is buying shares in actively at a low price, you can make a lot of money with them. But it is rare.

  • Itamar Turner-Trauring says:

    Interesting! I’m going to try to recreate this from Fed flow of funds data. There’s some research using the FoF data that indicates dividends+buybacks yield (not including change in net worth, or adjusting for buyback loses) is much better predictor of returns that just dividend yield: http://www.ems.bbk.ac.uk/faculty/wright/pdf/robertsonFeb2007

    Any thoughts on how to bring in the impact of leverage?

  • chris g says:

    You are offbase concerning buybacks.

    (1) You assume that past reinvestment rates will be the same going forward. Management shouldn’t assume that and neither should an analyst.

    (2) I back tested companies that buy back stock and they noticeably outperform those that don’t.

    In a world of overcapacity, over-supplied money, over-supplied credit, unlimited everything, rapidly cheapening dollars … it’s nice to buy stock in a company that has fewer and fewer shares outstanding every single year. Fewer partners is good.

    I’m not popular at parties. My jokes suck. But one that I always laugh at is this: At the rate Intel and Walmart are buying back shares, there will be no shares for my daughter when she’s 30. How many $15 billion buybacks can a company do when their market cap is $200 billion before the share price goes up?

    I’m convinced that the reason my backtest for buybacks presented so much alpha was because a steady buyback presents a constant buyer. Walmart has bought, on average 51,000 shares per hour for the last 7 years, every single hour the market is open.

  • chris g says:

    Are you taking tangible book value from the actual unadjusted balance sheet? Do you understand that balance sheets, even if they say “tangible” or PPE, do not represent true economic value? Some managers mark assets down as much as possible, others do not.

    From what I can tell, this whole analysis is worthless. I would love to see a backtest. Are you better off as a portfolio manager with a 5 to 10 year time frame buying companies with cram/jam or buying companies that buyback shares and still have fat ROEs/low P/Es?

    disclosure: I never worked at AIG

  • Chris G — could you send me a copy of your backtest? From what I have seen, most academic research is mixed on buybacks — they are era-driven, looking smart in bull markets, and dumb in bear markets. What is interesting with my post is that the S&P 100 rose on average over the whole period, and firms lost money on buybacks.

    Of course I understand that tangible book doesn’t represent economic value. But it is a proxy, and one that is pretty powerful in the academic literature. Low P/B and P/TB stocks tend to outperform over time.

    So do stocks where the accounting is conservative. I have a post coming soon on some of the relatively simple tests investors can do to gauge how conservative the accounting is. Cram is one measure of accounting conservativeness, and one of the few that can be applied to financial companies. Most of the other tests don’t work on financial companies.

    Doing a backtest on Cram and Jam would be tough — I would be happy to do one, but I don’t have a database to do it with.

    As for Walmart and Intel — as more gets bought back the price will rise, maybe, making it harder to buy more back. But I would rather point you in the direction of Torchmark [TMK]. They are the poster child in the insurance in the industry for buybacks — having reduced their share count more than WMT and INTC — it has not improved the performance. It has weakened the balance sheet.

    Look, I am willing to hear a level-headed argument on this. I am not some nut married to my theories, so if you have something better, please send it on. And I mean this — thanks for taking the time to post, I appreciate it.

  • chris g says:

    Sorry for the late reply. No, I can’t send you my results and they’re really not academically up to snuff. I run market neutral portfolios from screens that I evolved from AAII and Magic Formula. I use AAII’s data sets to roughly back test my modifications to their screens. I consistenly find that buybacks are even better than freecashflow for predicting shareholder returns. I eliminate companies that are buying back at the expense of balance sheet. RHDonelly was one that bought back stock in the $50s and $60s then couldn’t make debt payments and was BK soon after… strangest thing I ever saw. I subjectively eliminate stuff like that. If I didn’t eliminate those companies my results wouldn’t be as good. High roa is important part of my screens and I subjectively find that buybacks don’t hinder future growth. Sure, it’s easy to say that if a company is getting 20% roa, it would have grown more had it applied more capital… but the world just doesn’t work that way.