Sunday, April 20, 2014

Beware CAPE, it could be your undoing

The blogosphere has been slowly shifting from worrying about the tepid nature of the current recovery to biting its nails over the timing of the next downturn. Feeding its fears is Robert Shiller's cyclically-adjusted price earnings (CAPE) ratio, the elevated nature of which would seem to indicate that the fun can't go on (see chart below). I think the the CAPE is a crappy measure for measuring valuations and should be largely ignored.

The general idea behind CAPE is that there exists a long-term average price earnings ratio to which stock markets will eventually revert. In the 1970s and early 80s, markets were undervalued on an earnings basis relative to their 16.5x average, so purchases made sense. Now they are overvalued relative to their historical average, so sales would be appropriate.

I have two explanations for why CAPE is a crappy measure for determining the over or undervaluation of equity markets. These are both "money" reasons, meaning that they have a monetary basis. I discussed the first last year in Beyond Buffett: Liquidity Adjusted Equity Valuation. I'll briefly summarize my points from that post before launching my second swipe at CAPE.

In brief, CAPE ignores the changing moneyness of stocks, or their liquidity. Stocks provide owners with a flow of earnings, but they also throw off a non-pecuniary flow of liquidity services. These non-pecuniary services stem from an investor's expectation that they will be able to easily liquidate those shares in secondary markets. For example, should your roof start leaking, your IBM shares can be quickly sold, the proceeds used to hire a contractor to patch the leak. The more liquid the stock, the more easily it can be dispatched to resolve the various unexpected problems that arise in life. Since these uncertainty-shielding services are valuable, people will pay a premium to enjoy them, a liquidity premium developing. Illiquid stocks may take longer to sell, and therefore provide a smaller flow of uncertainty-shielding services, commanding commensurately smaller liquidity premia.

Anyone who uses CAPE as a model is implicitly assuming that investors only purchase a stock so that they can own its expected flow of earnings, not its flow of liquidity services. Put differently, the CAPE model sets the liquidity premium on stock to zero. Thus a user of CAPE will attribute any rise in the CAPE above its long term average to changes in investors' willingness to pay more for each dollar of earnings. But if we bring liquidity into the picture, a rise in CAPE above its long-term average could just as easily be the result of a technological improvements to stock market liquidity. If the typical S&P 500 stock is more liquid than it was a decade ago, then people will pay more to own the liquidity return associated with stock, and the price of the S&P 500 will rise independent of earnings. This doesn't mean that a stock is expensive. It only means that stock "does" more things for the investor than before, and trades at a deservedly higher price.

A strict interpretation of CAPE says that we are currently so far above the market's long term valuation range that we need a market crash to bring things back into line. But if we adopt a liquidity-adjusted view, the idea that there exists a long term average to which price earnings ratios need to fall is silly. Stocks today are not your grandfather's stocks. They have become evermore cash-like and will probably continue to evolve in that direction, a progressively larger liquidity premia over the decades arising as a result. If so, observed price earnings ratios are not destined to revert to mean, but have attained a new and justifiably higher plateau, and will continue to hit higher plateaus in the future.

The second reason I don't like CAPE is its failure to properly account for inflation. Shiller uses earnings as his denominator, but during inflationary periods like the late 1960s, 70, and early 80s, earnings were a terrible measure of the true financial health of a company. Inflation, combined with historical cost accounting, has the effect of creating "phantom" earnings. These phantom earnings are mere artifacts of accounting rules, yet firms have to pay very real taxes on these earnings. As inflation mounts, a firm's artificially accelerating tax bill robs them of the cash they need to fund operations and new projects.

What follows is a short explanation of this effect, but if you prefer a longer one, try my old post A stock portfolio is a bad hedge against inflation.

If inflation doubles, the standard view is that stock is a great inflation hedge since a firm's revenues and costs immediately adjust upwards, the real value of the bottom line staying unchanged. However, historical cost accounting impedes a fluid 1:1 inflation adjustment. First, the cost of goods sold line on a firm's income statement doesn't rise in line with inflation. Inventories are accounted for on a first in, first out basis, which means the prices used to compute costs of goods sold are stale prices, as yet unadjusted for the ravages of inflation. Secondly, the depreciation line item doesn't rise with inflation. Machinery and other equipment are depreciated based on the item's historical (and therefore stale) purchase price, not on the basis of the good's current inflated price.

Since neither cost of goods sold nor depreciation rise during the early stages of an inflation, the firm announces higher real pre-tax profits. However, the rosy picture provided by the accountants obscures the fact that the firm's true economic position has not changed one bit. As inflation accelerates, the effect on the firm's cash flow is neutral. The net quantity of cash flowing into the firm from its clients less the cash flowing out of it to suppliers rise together. If the firm must pay 20% more cash to purchase inventory, that rise is completely compensated by 20% more in cash receipts from clients.

While the firm's net cash inflows from clients less outflows to suppliers remain constant during the inflation, the firm will find itself incurring larger cash outflows due to taxes. Based on the firm's growing accounting profits, the firm faces a higher tax bill than it did prior to the inflation. The growing quantities of cash that leak away to the government as inflation accelerates mean that less cash is available to pay suppliers, expand operations, or add to dividends. Inflation has made shareholders worse off. Taken to the extreme, a wild inflation will force a firm to pay an increasingly large portion of its wealth to the tax man, eventually resulting in the firm's bankruptcy.

So in periods like the late 1960s, 70s, and early 80s, it made absolutely no sense to value companies on the basis of their earnings. This makes a mockery of the CAPE parable. According to CAPE, investors in the 1970s irrationally bid stock down to very low prices relative to earnings. A smart investor should have picked up shares at these low levels in anticipation of a reversion to the long term CAPE ratio, a bet that would eventually payoff in the 1980s bull market.

In reality, since such a large portion of the earnings during that era were phantom, or non-existent earnings, investors placed a large value discount on them, or only purchased stock at very low price to earnings ratios. Stocks weren't being undervalued, they were being properly valued as the terrible inflation hedges that they were.

Contra the CAPE parable, the 1980s bull market was not the eventual payoff to the patient few who invested in low PE stocks. Rather, the bull market has Paul Volker to thank for, as it was his inflation-reducing policies that saved stocks from their own Achilles' heel; the adverse mixture of rising prices and historical cost accounting. Terrible inflation hedges they may be, when the reverse happens—low and falling inflation—stocks become stellar investments, as the 1980s would bear out. As inflation withered, phantom profits disappeared and firms were no longer forced to pay undeservedly high taxes. The large discounts that had been applied to earnings during the inflationary period were steadily removed.

So the CAPE fails because it ignores two monetary phenomena. It does not properly adjust for liquidity, nor does it account for the illusory profits that are created during inflationary periods. Until the quants figure out how to create a CAPE measure that corrects for these monetary effects, throw the CAPE in the toilet.


  1. All good points, but isn't it somewhat safe to assume mean reversion in inflation, given the dual mandate?

    A random thought, I've read the Summers paper from a while back but forgotten much about it. Is there any paper that examines stocks by deciles of asset turnover/inventory turnover/cash conversion cycle and relates that to inflation? I imagine higher turnover/cash conversion should mean they're a better inflation hedge.

    1. Higher turnover/cash conversion certainly results in a better inflation hedge. There's probably a paper somewhere on this.

      Sure, inflation can be mean reverting. But you still need a proper valuation metric that adjusts for inflation effects so that you can analyze for under/overvaluation over the entire inflation-deflation-inflation cycle.

  2. Nomad, nice reference. Speaking of which... any thoughts on Cap 2? I thought Marvel Studios really hit it out of the park.

    Given the tax and information distortions caused by using historical costs, why hasn't constant purchasing power accounting caught on more?

    1. Haven't had a chance to watch it yet.

      Why hasn't constant purchasing power accounting caught on? Probably because central banks have been pretty good at keeping inflation low and constant, except for the 70s, so the need hasn't arisen.

    2. Random thoughts.

      1. It's true that inflation has generally been low in the US. But historical cost is still being used in Argentina (it's optional I believe, but no companies have adopted it). I suppose managers like being able to post phantom earnings, but wouldn't the tax issues you cited lead industry groups to favor CPPA?

      2. Even though inflation has been low in the US, it still hasn't been zero. This can make comparisons of financial statements (and analysis of what went right or wrong) from different time periods difficult. Also, no accounting system I'm aware of adjusts for booms and recessions. Being able to tell how a firm performed relative to the overall economy seems important (i.e. did the company do well or just ride the wave of a booming economy?) Perhaps accounting in Shiller Trills or something similar might be useful.

      To use a sports analogy: is a .300 batting average good or great? Answer: depends on the year. In '68, when the American League avg was .230, it would have nearly led the AL (Yaz won with .301). In '94, when the AL avg was .273, it wouldn't have made the top 25. Sabermetrics nerds correct for this by using advanced metrics that measure a player's performance relative to the league average (weighted runs created plus, etc). If fans can smooth out "baseball stat illusion," can't/shouldn't multinationals correct "money illusion?"

      Today, firms are trying to wring out every last bit of efficiency by making better decisions. Even if the benefits of correcting for inflation and/or NGDP swings are small, it would still allow for more precision in financial reporting.

  3. Since money is neutral in the long run, corporate earning's yield (inverse of P/E ratio) should be correlated to interest rates. The relationship to inflation is indirect.

    1. That's the Fed model, it's also silly.

  4. Great article JP. You're definitely making me think way too hard on a Monday morning...

    I get (because of you) that the improved earnings on the front end are just from "shifting the goal posts," but how do you account for the "shifting of the goal posts" that also occurs from the depreciation (or appreciation in deflation) of cash? Or put differently: if the relative value of the US dollar decreases why would being long stock better than being long anything else?

    1. Sorry Dave, I don't quite understand your point. Give it another shot.

  5. JP Koning, JW Mason has a new post where he makes the case that share buybacks function as a way to make stock more liquid. I'd be really interest to know your take on this as I thought you would be the person to know what proportion of stock liquidity typically comes from that.

  6. JP, O/T: I asked Rowe if the logic of his "Temp vs Permanent money mult" post could apply to his "Commercial banks can create an excess supply of money" post, and he thought it could: that is if people don't believe the excess supply of commercial bank money is permanent, then it may have a reduced effect: (e.g. not raise the average price level)
    What would influence people's perception of whether it was permanent or not? The blogs of Nick Rowe, David Glasner and JP Koning? Lol

  7. How about it's track record of actually forecasting 10 yr returns? As Hussman and others show, it seems pretty good