Choose Your Recession-Forecasting Framework Carefully

Not all recession forecasts are created equal. Meanwhile, US recession forecasts are everywhere. From the lips of famous economists to casual observers of the macro scene, warnings of economic contraction are sprouting like primrose and tulips. The challenge is that the projections don’t come with warning labels, making it difficult to interpret the validity, relevance and timeliness of the forecasts on a case-by-case basis.

The worst thing you can do is to assume that all recession forecasts are equivalent. If you consume ten business-cycle reports and find that seven are bearish, it’s tempting to assume that there’s a 70% chance that a contraction is near. Tempting, but almost certainly wrong, since you’re probably comparing apples and oranges, each with a dramatically different methodology, if any.

You hear economist A chatting up the TV host at XYZ network, warning that a downturn is a “high-risk” probability in the year’s second half. You then turn to the financial pages of the ABC Journal and read a similar analysis from the director of economic studies at the John Bill Mary school of public policy. You round out your research with a quick read of your favorite macro blogger, who sees trouble ahead.

The main takeaway: a new recession is likely, or so it seems. But quite often with these macro travel episodes there’s little analytical context. That’s a problem because context is important for assessing the validity, relevance and timeliness of a given forecast/evaluation. The context may be there, but it’s not available, thereby putting a premium on understanding the conditions and assumptions that created the forecast/analysis.

Enter The US Business Cycle Risk Report (US-BCRR), your editor’s weekly newsletter that offers a solution by cutting through the noise and focusing on a quantitative-based review of the evolving state of real-time recession risk. The crucial difference: US-BCRR analyzes macro risk in a particular way, providing a consistent methodology that’s comparable through time. As a result, US-BCRR provides the necessary context that’s usually missing in the firehose of recession commentary in the wider world.

Critically, US-BCRR focuses on a specific type of business-cycle analytics for a relatively narrow time window, namely: estimating a high-confidence probability that a new contraction has started recently, is currently in progress, or is about to start in the next one to two months via a broad set of economic and financial indicators.

The focus is distinct from most of the recession forecasts/evaluations that swirl about from the usual suspects. But this methodology is chosen for a reason: it’s relatively reliable for assessing recession risk.

The logic here is that 1) uncertainty spikes when looking ahead by more a couple of months; and 2) relying on a handful of indicators (or, even worse, just one) sharply raises the risk of noise.

US-BCRR minimizes these risks by focusing on a specific type of recession-risk analysis that targets the sweet spot of high-confidence estimates that look ahead as far as possible with minimal reduction in signal quality. That’s a high bar and it requires, among other things, reviewing a broad set of indicators. It also requires managing our expectations in terms of what’s possible, and what’s not, in the art/science of high-probability analytics for the business cycle.

As one example, here’s how this week’s US-BCRR edition evaluates recession risk at the moment, based on the latest published numbers to date for one of several business-cycle indices routinely updated in the newsletter. In this case, our proprietary ETI and EMI data sets are highlighted (for details, see this sample issue of the newsletter).

For the moment, the probability that the US economy is contracting is virtually nil. Looking ahead through May, via forward estimates of ETI and EMI, points to an ongoing but gradual slowdown in growth that still leaves the expansion intact.

None of this is to dismiss the possibility that at some point an NBER-defined recession may start (or not). It’s easy to predict a downturn is fate, based on cherry-picking indicators and seeing trouble six months or year ahead. But such efforts usually come with a low signal-to-noise ratio, largely because recession forecasting beyond a couple of months is highly uncertain. Unfortunately, that caveat is usually MIA. Caveat emptor!