A simple model

So the US economy is not, in fact, falling off a cliff...just yet.  Although durable goods orders are notoriously noisy, yesterday's solid print should assuage at least a few fears about the recent soft tone to some of the data.  That core CPI managed to beat expectations was icing on the cake, as the dollar and US yields both jumped smartly after the releases.   As Hannibal Smith used to say, "I love it when a plan comes together!"

Of course, it wasn't all pony rides and lemonade.  Weekly claims were quite a bit higher than expected, albeit in the context of seemingly interminable snowy, nasty weather for much of the country.  Still, the decline in claims seems to have stalled a bit recently; although it's probably too early to read too much into it, it nevertheless remains a development worth watching.

Every so often, Macro Man likes to take a step back and look at data in its unadulterated form- no rates of change, moving averages, or anything like that, just the level of the underlying data.  When looking at the level of durable goods orders, ex-transport, you can see the recent soft patch; put into context, it doesn't yet look particularly abnormal or worrisome.  Certainly the loss of momentum is nowhere as severe as it was in 2012.

Perceptive readers will note that durables are basically at the peak level of the last cycle, whereas the SPX is comfortably above it.  How, then, do these orders figures compare with other data?   Macro Man assembled a chart showing nominal retail sales, the ex transport orders noted above, and industrial production, scaling each series to 100 as of February 1992.  As you can see, the relative fortunes of the 3 data sets vary starkly.

As the title of the chart suggests, these three series capture the changes in the economy quite well.  Consumers have continued to spend (note that retail sales are substantially above the level of the pre-crisis peak), but not, it would appear, on goods made in the US.  This is hardly a stunning revelation, of course, but it's nice to see it demonstrated so sharply in a single chart.

This got Macro Man to thinking.   Retail sales, like the SPX, are very comfortably above pre-crisis highs, IP is marginally so, and orders are basically at the level.   What would happen if we regressed these 3 series against the SPX?  The results are set out in the chart below.

As you can see, the relationship is generally quite good, with an r-squared of .88.   Note that to some extent that should be expected, insofar as the entire time period was in-sample and your author regressed levels rather than changes.  Nevertheless, it's interesting to note that the divergence between the current level of the SPX and that suggested by sales, orders, and IP is larger than  it has ever been over the past couple of decades.

To a large degree, of course, that is a function of QE/easy monetary policy/low bond yields/low discount rates.   One would naturally expect, ex ante, that stocks should look a touch rich to exclusively growth-based indicators given that global policymakers have begged investors to move out the risk curve for more than half a decade.

Naturally, equities also benefit (and suffer) from leverage, both on a corporate and investor level.  Indeed, generally speaking the amplitude of changes in stock prices dwarfs that of changes in the underlying data, so it may well be the case that a simple linear regression is a decidedly imperfect tool.

To mitigate the impact of leverage, Macro Man ran the same regression, but on the natural log of the SPX to smooth the swings in stocks relative to the economy.  Upon obtaining model output, it was a trivial matter to reconvert the data back into SPX terms.  The results of this study are below.

This model suggests a closer fit between the current level of stocks and that implied by growth data (and a closer overall fit, with an r-squared of 0.93), though it still views equities as nearly 10% too high (when using current stock prices.)  However, "mispricings" of this magnitude are fairly ordinary; indeed, the same model suggests that stocks were undervalued by a greater amount in the run-up to the crisis.  That very fact should tell you everything you need to know about relying exclusively on growth-based indicators that take no account of financial conditions.

So what have we learned?  A very simple growth-based model suggests that US stocks should indeed be at all time highs, albeit at levels lower than those currently prevailing.  Introducing a liquidity term would indubitably close that perceived valuation gap; indeed, in Macro Man's own, more robust equity model, liquidity factors explain almost all of the ex ante expected return of the SPX.

As long as liquidity remains ample, growth doesn't necessarily need to roar to justify current if not higher prices.   If and when the tap gets closed off, on the other hand, simple indicators like these can provide a rough and ready guide to how deep the correction might be if we were to revert to using growth rather than easy financial conditions as the primary valuation metric.