Do most buybacks destroy value?

Good morning. Two short emails landed almost consecutively in my inbox yesterday morning, both from investment pros. Email one: “US CPI inflation seasonally adjusted, month over month, three month moving average is -2 per cent . . . I think there is a serious risk of [Fed] overtightening.” Email two: “Rates to seven — any [Fed] pause a mistake. Soft landing is a fool’s goal.” Disagreement makes a market, and beating the market is hard. Email us: [email protected] & [email protected]. 

Buybacks revisited

Yesterday’s argument — or rather accusation, given its lack of proof — that most buybacks destroy value prompted a lot of mail, both approving and disapproving. The gist of the claim was that (a) companies buy back more of their shares when times are good and their stock is expensive, and (b) buybacks only create value when the shares are bought back at or below fair value, and therefore (c) most buybacks probably destroy value. Conceptually a tidy argument, perhaps, but (to mangle a famous line of Kant’s) concepts without data are blind.

Before turning to readers’ letters, then, let’s look at some data, none of it decisive, some of it suggestive. First, using data provided by the great Howard Silverblatt of S&P Indices, here is a chart of the year-over-year change in dollars spent on buybacks by companies in the S&P 500 index, plotted against the year-over-year change in the level of the index:

You’ll note that buybacks are more volatile than the index, and that peak repurchase activity tends to correspond with market highs (2000, 2005-7, 2021; the big percentage jump in 2011 is an artefact of the very depressed repurchases the year before) and buybacks fall sharply when the market falls most (2001, 2009, early 2020). So, it seems that buybacks are procyclical and therefore at risk of destroying value, if you assume that the market, at its peaks, tends to be overvalued.

Next is a somewhat less incriminating picture, which charts the percentage of the total market value of the S&P 500 index bought back each quarter against the price/book value ratio of the index — that is, buyback intensity versus the valuation of the market. I used price/book value because it creates a somewhat smoother chart, but using price/earnings produces something quite similar:

You will notice that buyback intensity is sometimes high when valuations are moderate — from 2010 to 2014, for example. That’s good! But you will also notice buybacks intensity rising in recent years as valuations pick up — look at 2018-2019 for example, and 2021. That’s not so good!

Several readers wrote to point out, with approval, the buyback policy of UK retailer Next PLC. The stock has traded sideways since 2015, but in the decade or two before that it had a staggering run, which the readers put down in part to a disciplined buyback policy. At Next, a buyback must hit an “equivalent rate of return” threshold of 8 per cent. This simply means that a buyback’s impact on earnings per share must be at least equivalent to an investment in an asset with an 8 per cent yield.

Having such a fixed threshold makes sense to me, and Next’s argument for it, in its 2013 annual report, is worth reading. I have no idea if Next is a good or well-run business otherwise (my friends over at Lex seem to like it, though).

I got several letters pointing to studies showing a correlation between buybacks and long-term market outperformance. My general worry about these studies, though, is that they might mistake correlation for causation. Companies that do steady buybacks tend to be companies that generate consistent and rising free cash flow — the money has to come from somewhere. Consistent and rising cash flow definitely underpins long-term stock performance (I mean, if it doesn’t, everything is chaos and nothing matters). But a cash-flowy company may outperform over time despite an inefficient, wasteful buyback policy.

And I still just don’t see how procyclical buybacks can be, for the market as a whole, a great source of value. Of course that doesn’t imply that a Next or a Berkshire Hathaway can’t create oodles of value with a well-run buyback program.

Finally, a very interesting email came from Charles Cara, head of quant strategy at Absolute Strategy Research, who proposed an argument about buybacks that was new to me, based on the influence of passive index investors on the market:

[Passive] investors are not price sensitive — they do not need to rebalance their portfolios when a stock rallies or decline — BUT they are share count sensitive. When a company buys back its shares, its weight in an index will fall by the same proportion (although there might be a delay), and so a passive investor has to sell to keep in line with the index. So share buybacks create forced selling by passive investors. If over half the market is passive, or at least closely benchmarked to a market cap weighted index, then over half of the benefit of share buybacks will be reversed, blunting the efficiency of buybacks.

Moreover, passive investors then complete their rebalance by buying the stocks that have not done buybacks or issued shares, and so lift the prices of the rest of the market. So passive investors create a leakage of the impact of buybacks into the wider market . . . Maybe this why indices such as the Nasdaq Buyback achievers has failed to outperform the market over the past 15 years despite spending ~$5 trillion on repurchases.

Here is a chart of the Nasdaq US Buyback Achievers performance relative to the Nasdaq as a whole. The Achievers index includes companies that have bought back at least 5 per cent of their shares over the previous 12 months:

Curious to hear if readers like Cara’s argument.

US stocks: where my bargains at?

The S&P 500 is not horribly expensive. Not like 2021, anyway. The S&P rests at 17.5 times forward earnings, a bit above the long-run average. In normal times, you might call such a valuation fair. But with higher-for-longer rates, a clear earnings contraction and potential recession hanging over equities, fair isn’t fair; it’s expensive.

We made this point last month by looking at the equity risk premium, the additional reward stock investors get relative to risk-free Treasuries. No matter how you measure it, the ERP is at something like a decade low — partly reflecting the fact you can get 4 per cent on a 10-year Treasury while taking no risk! Equities’ risk-adjusted pay-off looks paltry by comparison.

We were struck by a note out yesterday from Morgan Stanley’s Mike Wilson, who a week ago declared the ERP had entered a “death zone” for equities, building on this low-ERP point.

Wilson notes that the ERP (calculated as forward earnings yield less the 10-year Treasury yield) might be kept artificially low by a handful of expensive megacap growth stocks. Remember that a company’s earnings yield is just the inverse of its P/E ratio, and the megacaps aren’t particularly cheap on a P/E basis. Microsoft, for example, trades at 25 times forward earnings, while Nvidia trades at 50. So to pave over any lumpiness at the top, he calculates an equal-weighted ERP. Yet equities’ pay-off still looks wanting:

Wilson’s simple calculation of the ERP is common among practitioners, though we’re aware academics have all sorts of more sophisticated methods. But since those approaches point to the same low-ERP conclusion, keeping the maths simple works for us.

Moreover, he finds that this low-ERP story spans

. . . virtually all sectors except for Energy. ERPs are at their lowest level since the Financial Crisis for Tech, Industrials, and Materials. They are under the 2nd percentile for Consumer Discretionary, Health Care, Staples, Comm Services, Utilities, and Financials.

This sector breakdown broadly holds under traditional valuation metrics, too. Wilson looks at forward P/E and forward P/sales ratios, comparing current levels to their post-financial crisis medians. Few look cheap. Here’s his table:

By these measures, we count four areas (excluding energy, which is in a world of its own) that could be argued are cheap: telecoms, banks, transports and real estate.

The latter three categories are all cyclicals, which probably should go on sale when recession risk is elevated (the bigger question is why other cyclicals aren’t following suit).

But consider telecoms. This is a capital-intensive, commoditised, oligopolistic industry with just three notable companies — AT&T, Verizon and T-Mobile, all of which are low-beta, defensive names. There are the makings of a value play here, including depressed valuations (10 times forward earnings) and good dividend yields (AT&T and Verizon offer 6-7 per cent). But not an easy one; debt loads are heavy and competition for market share is fierce. Irene Tunkel at BCA Research figures that telecoms will get more compelling the closer we get to a recession.

Step back, though, and these few pockets of value look tenuous. Individual names can of course still do well for idiosyncratic reasons. But on the industry level, the simple fact seems to be that US stocks are expensive right across the market. Markets may have priced in higher-for-longer rates, but they haven’t drawn the line to recession just yet. (Ethan Wu)

One good read

Henry Kissinger, Eric Schmidt and MIT’s computer science dean think AI chatbots are “poised to generate a new form of human consciousness” on par with the Enlightenment.