ROE, Retained Earnings, and Old Benchmarks
Board: Berkshire Hathaway
If we look back from about the beginning of time (Berkshire’s Buffett era) to the end of about 2003, we see that for about every three dollars reported (in earnings/ retained earnings), Berkshire had two dollars ‘unearned’ in unrealized securities gains. Sure these unrealized gains were in ‘other comprehensive’ earnings and not ‘retained earnings’ on the balance sheet, but the point is that there was a continual and significant ‘earning’ going on in this long-term equity holdings segment. Year-in, year-out it was about two for three, unrealized portfolio earnings vs reported earnings. Conveniently, the market generally recognized this, in either -- take your choice -- ‘look-through’ earnings-based valuations, or two-column balance-sheet-based analyses.
‘Two for three’ has been missing in action for awhile. Since 2003, that internal portfolio growth has slowed dramatically. 'Other comprehensive earnings' - those unrealized securities gains - have been essentially non-existent. In aggregate, these unrealized gains have been just a few $billion, and realized gains over the past five years have also been relatively meager.
Overall Berkshire has reported over $50B in total reported earnings over these past five years. Adding back $14B of (non-cash) depreciation over that time, and ignoring the $6B portion of those earnings generated by the realization of securities gains (these are just a shift from one pocket – or line on the brokerage statement - to another), that’s just under $60B in generated ‘new cash’ earnings.
Of that $60B, $27B was spent on internal capital expenditures, exclusive of acquisitions (or $25B of capex if we exclude BNSF’s 2010 $1.8B portion). Granted we picked up more than $20B in insurance float and issued $11B of shares over those years – providing much of the funding for the purchase of a major business. But our election to ‘reinvest’ almost half our total cash generated from operations back into internal uses (again, exclusive of acquisitions) has raised the ‘E’ hurdle in ROE in a magnitude that Berkshire hadn’t experienced in the past from internal investment, without also building the ‘R’ as we’ve come to expect from Berkshire investments.
Maybe we’ll someday see the kinds of double-digit returns on that capex we've come to expect from Berkshire's other investments in both businesses and securities. But at the rate we’re building capex vs the rate that MidAmerican and the other capex-heavy businesses are actually growing earnings (exclusive of the benefits of their acquisitions, which are another element of investment in those businesses), it may be tough. Right now it looks like it may be hopeful to expect, at best, high-single-digit returns on this new capex investment. These apparently meager returns on internal reinvestment, combined with the unprecedented flat spell in the long-term portfolio, makes for a tough deal right now in terms of ROE performance.
That previously long-running portfolio (‘other comprehensive income’) growth, by the way, was Buffett’s ‘ace in the hole’ for many years in his stated target of having the market cap growth exceed retained earnings over five year segments. The portfolio growth was of course outside retained earnings, but nonetheless recognized by the markets. When that performance element stalled, that Owners’ Manual commitment relative to retained earnings couldn’t be maintained - even on just a modest 'dollar for dollar' standard. The market seems reluctant to award us dollar for dollar on the $50B retained over these past five years, and we might conclude that the portion of reinvestment represented by that $27B investment in capex is at least partly responsible.
We’ve made our bed, so to speak, in getting into capital intensive businesses. While these may throw off impressive total earnings, it’s easy to lose sight of the ongoing capital requirements of these businesses. The available detail for an ROE (or even ROCE) analysis isn’t always as clear as we'd like it to be. But there’s no hiding from the resulting ROE results of this shift of the past decade.
Anyway, we may have to get away from the notion that the market will award us a dollar in value for each dollar in earnings that we retain. We've altered that objective. This shift in benchmark has nothing to do with accounting nuances (eg, classifications of earnings as comprehensive or retained, and such), and everything to do with the nature of capital intensive businesses.
Maybe the market isn’t so unobservant. And maybe as we complacently abandon long-standing principals we should at least be a little bit introspective. For years we didn’t think they were ill-conceived or unreasonable. Going back a few years, more than a few Berkshire followers probably would have reflexively claimed that it would be ill-advised to disparage them. Maybe now that these standards are beyond reach these same folks can again, without introspection, conclude that they were pretty dopey standards for Buffett to have included for so long. If we believe they were ill-conceived and ill-considered for so long, we can certainly give up these standards (as we have in this case) but we need to understand both the underlying reasons that we are now abandoning them, and the implications on our expected returns-on-investment.
We feel noble intentions should be checked periodically against results. We test the wisdom of retaining earnings by assessing whether retention, over time, delivers shareholders at least $1 of market value for each $1 retained. To date, this test has been met. We will continue to apply it on a five-year rolling basis. As our net worth grows, it is more difficult to use retained earnings wisely. We continue to pass the test, but the challenges of doing so have grown more difficult. If we reach the point that we can’t create extra value by retaining earnings, we will pay them out and let our shareholders deploy the funds.