Putting the LEI’s 19-Month Slide Into Perspective

The Conference Board’s US Leading Economic Index (LEI) fell again in October—the 19^th straight monthly decline—and headlines’ reaction mostly matched the past several months. LEI, they said, indicates a recession (still) looms. And hey, that could very well be. We see little to no sign of it in the vast majority of indicators, but we don’t dismiss the possibility. With that said, we also think there is another way to look at this: LEI is now down 19 months straight, and the US economy is not only growing, but GDP has accelerated this year, and stocks’ recovery from 2022’s bear market continues. We aren’t saying LEI is broken or useless or anything like that. But this is another powerful reminder that LEI isn’t a stock market forecasting tool, and even as an economic forecasting tool, it has some shortcomings worth noting.

Until now, LEI was one of the most reliable economic indicators on the planet. Because it aggregates 10 mostly forward-looking variables, in times not skewed by post-pandemic weirdness, it delivers a broad look at the earliest signals of economic direction. This includes a few sets of factory orders (today’s orders are tomorrow’s production), the yield curve spread (a proxy for future loan profitability and therefore a predictor of loan growth), The Conference Board’s proprietary measure of credit availability (aka the Leading Credit Index) and the average monthly S&P 500 index level. Generally speaking, if all of these look healthy, a recession is highly unlikely to be developing, which is why no recession before the COVID lockdowns ever began while LEI was high and rising. LEI isn’t perfect—it has some false signals during expansions and includes some less telling components (e.g., manufacturing work hours, weekly jobless claims, consumer sentiment)—but its record was about as good as they come.

Until now, with this very long misfire. We see a few reasons LEI isn’t working so well. One is the historically rare split between the yield curve and bank lending. Traditionally, the yield curve works as a predictor because it represents banks’ business models. Banks borrow at short rates and lend at long rates, making the gap between them a proxy for future loan profits. But that has broken down over the past year, with an inverted yield curve both detracting from LEI and accompanying continued loan growth. As we have discussed in more depth before, we think this is because the well-documented deposit glut enabled banks to keep getting cheap funding even as the Fed hiked rates. Meanwhile, higher long rates raised loan revenues. As a result, net interest margins finished Q2 a bit above prepandemic levels (the latest aggregate data available), keeping banks happy to lend—and inject more fuel into the economy.^[i]

The yield curve’s reduced relevance probably isn’t a permanent change or drawback for LEI. But the other recent main detractors are perhaps more worthy of some long-term consideration. One is the ISM New Orders Index, which has detracted for months on end. Core capital goods orders and manufacturing work hours have also dipped plenty of times this year. Together with factory orders for consumer goods, all four represent the manufacturing side of things—the “stuff” economy, if you will. So even where they are legitimately forward-looking, they ignore services.

We think this is LEI’s biggest shortcoming. Services, while more difficult to measure than widget production and sales, are over 70% of GDP.^[ii] Goods are just over 17%.^[iii] So LEI overemphasizes a minority sliver of US economic activity while all but ignoring the vast majority of it. Services are doing the heavy lifting this year, generating more than enough growth to offset manufacturing’s weakness. LEI correctly foresaw that weakness but couldn’t capture services’ contributions, delivering the macroeconomic miss.

This isn’t the first time LEI has faced such a quandary. As it happens, LEI has taken several forms since its creation in the mid-20^th century, evolving with the economy. When created, it emphasized commodity prices—leading indicators of construction and infrastructure development. Industrial building contracts also appeared for a time, as did supplier deliveries, before broader construction permits and new orders replaced them. In 2011, The Conference Board added the Leading Credit Index. Why? Because LEI didn’t do a great job foreseeing the recession that accompanied 2007 – 2009’s global financial crisis. It was mostly flat throughout 2006 and 2007, which dismayed the fine folks at The Conference Board. After careful study and research, they determined the LEI in use at that time wasn’t equipped to capture a recession preceded by a credit squeeze. So they added the Leading Credit Index and recomputed the whole shebang retroactively, showing that in hindsight the new LEI would have done a great job spotting 2008. (If you would like to know a little more of this history, we shamelessly recommend Fisher Investments founder and Executive Chairman Ken Fisher’s 2015 book, Beat the Crowd, which covers it more fully.)

So we can’t help but wonder, might they be noodling something similar now? In a year or two or three, might we see a report detailing the problems with using a manufacturing-heavy leading index in the digital services age? And might this lead to new economic indicators that focus on services and information?

We like data and new indicators, so this possibility is our silver lining to LEI’s ongoing travails. But it is also a speculative, future thing. So in the here and now, the important thing to bear in mind is this: Even if LEI were perfectly forecasting the economy right now, it wouldn’t forecast stocks. One, LEI includes stock prices, and stocks don’t predict stocks. Two, stocks move ahead of the economy and will typically decline well before a recession materializes or shows in LEI (since other components can mask stocks’ negative contributions). In both cases, leaning on LEI itself won’t give you an edge.

To us, LEI is most useful as part of a broader look at how economic fundamentals square with sentiment. If it sends telling signals that aren’t getting much notice—like, say, if it sends strong positive signals that get lost in a cloud of recession talk—that is quite useful. Or if, like now, it gets notice for confirming the world’s pessimistic biases, that can be a handy signal that recession talk lacks surprise power. But that is about the extent of it, in our view.