,
14 tweets,
5 min read
Read on Twitter
1/ Oh, no, no, no. Here's what people overlook about buybacks when they add them to dividends. The 3.7% SPX revenue growth in recent decades is index-level and already fully captures the impact of buybacks (via the SPX divisor). Adding buyback "yield" is sheer double-counting.
2/ Given that structural real GDP growth, absent endless further declines in the unemployment rate, is now down to 1.6%, the expectation of roughly 3.7% nominal growth remains about right for revenues, and also for earnings, unless record profit margins were to expand forever... 
3/ So now you've got roughly 3.7% nominal growth, plus a dividend yield of about 2%, which yes, does give you an expected return of about 5.7%, but then, that's only if valuations - particularly objects that behave like the price/revenue multiples, never, ever retreat...
4/ Note that these valuation measures are far better correlated with subsequent market returns, easily bettering price/forward op earnings & the Fed Model. Why? Cause stocks aren't claims to next year's earnings. Your fundamental has to be representative of decades of cash-flows.
5/ Indeed, one of the acid-tests of a useful valuation measure is whether it produces valuations essentially proportional to the long-term discounted stream of actual cash flows delivered to investors over time, so that changes in valuation imply variations in expected returns.
6/ Presently, the most reliable valuation measures we identify are ~ 2.6x the historical norms we associate with average subsequent market returns. In 07-09, those norms were slightly breached. The 00-02 decline got within ~25% (the highest valuation for a bear low on record).
7/ So now some unpleasant arithmetic. 1/2.6 - 1 = -62%, which is roughly the decline that one would expect if the completion of this cycle draws reliable valuation measures to their long-term norms (and not below). That's not a worst-case scenario, just a run-of-the-mill outcome.
8/ Moreover, despite the notion that "it always comes back" - which is always popular when the market is near hypervalued record highs, let's do a little bit of long-term return arithmetic. Assume growth bumps up to 4% annually, with valuations now 2.6 x norm, and dividend 2%...
9/ First, assume that reliable valuation measures merely touch their historical norms (which they've breached in most cycles except 2000-2002). If that happens 10 years from now, the annual SPX total return estimate would be (1.04)(1/2.6)^(1/10)+.02-1 = -3.5% annually...
10/ What about a 20-year horizon? In that case, you've got an estimated annual SPX total return of (1.04)(1/2.6)^(1/20)+.02-1 = 1.1%. So despite very low Treasury bond yields, stocks are still likely to lag bonds over that horizon...
11/ Now, it may seem preposterous that stocks could possibly lag T-bills over a horizon of well over a decade, but unfortunately, even a return to normal 6% peak-to-peak growth rates would have a difficult time overcoming any valuation drag: (1.06)(1/2.6)^(1/20)+.02-1= just 3.1%.
12/ … and see, that return to durable valuation norms is why stocks gave up the entire benefit of two successive bubbles and ultimately lagged T-bills for the entire period between 1995 and 2009. That's forgettable only because we're at the peak of a third valuation bubble.
13/ So yes, if valuations are near durable valuation norms, it's appropriate to estimate long-term equity market returns by adding expected growth to the dividend yield. But if you're using per-share values, buybacks are already embedded in the growth rate. You can't count twice.
14/ But if valuations are nowhere near durable valuation norms, nearly all of the excess is routinely wiped away over time. Those transient returns are driven by shorter-term swings investor psychology toward speculation or risk-aversion (see "internals" in other threads).
