Are Economic Indicators Broken?

Notes On A Weird Cycle

This economic cycle has been really weird.

We know this because people can’t quite agree on when it started.

Some people say we remain in the expansion that has been in play since the 2008/09 global financial crisis. That was the last time we had an endogenous economic shock which caused a big contraction in GDP.

This contrasts with the Covid pandemic, which was an exogenous shock, with a rapid recovery, and less creative destruction.

The fiscal response to the pandemic was Keynesian, on steroids. The US instituted a major pickup in fiscal transfers to households (stimulus cheques), which meant that workers who retained their jobs had loads of cash to spend while those that lost their jobs could get by. In contrast, Europe focussed more on furlough schemes, meaning that firms could maintain their workforces while they shut down as governments effectively covered their payrolls.

The effect was similar. Workers could chill at home, supported by government transfers. The US setup meant that workers had cash but could still end up needing to hunt for a new job when things reopened, whereas the European setup meant that workers were almost guaranteed to retain their job on the reopening.

This meant that there was some creative destruction in the US job market, but overall creative destruction was limited in both the US and Europe by robust fiscal responses (which massively supported demand).

Creative destruction has traditionally been key to the economic cycle, with firm and job closures/losses freeing up resources (capital and workers) for more productive enterprise. This informs part of the argument for claims that we remain in the post 2008/09 cycle.

Without creative destruction and with a rapid rebound for the economy, was 2020 really the end of the cycle?

Are Economic Indicators Broken?

We’ve also had all sorts of weird going on in the aftermath of the pandemic. The global economy has been continuously rocked by supply shocks, while fiscal policy has been unusually supportive across DMs. The supply side is being rocked while the demand side is constantly supported by pro-cyclical fiscal policy.

Recent years have also seen traditional forward looking indicators leaving economists flatfooted. These have sounded the recessionary alarm throughout the period, but have each proven to be a poor predictor of what is going on.

This leaves us with the question of why indicators that have previously consistently been good at predicting recession have largely missed since 2020.

Two caveats before we get going.

First, some economists, believe that the NBER will call a recession in 2024 (they often judge a recession around 18 months after the fact), so some of these indicators may have been successful in the recession that they were forecasting. However, if there was a recession in 2024, it will no doubt have been incredibly shallow and the recovery very quick. For our purposes, I think we can park this.

Second, there may be a recession in 2025. As we have written about several times, Trump is doing some mad shit. This could cause the economy to contract; however, there is broad consensus that the economy had successfully emerged from the highly inflationary post pandemic period without a recession. The economic indicators we are talking about were not flashing red because they were really good at guessing that insane US policy was around the corner. If we have a recession due to Trump, it will not mark success for these indicators.

So we’re gonna run down a few indicators. We care about why they normally work, their track record, and whether they are going to be useful in the future. Our focus is on the US, but we can replicate this across many DMs.

The Indicators

Yield Curve Inversion

The idea: Bond yields provide compensation for creditors which is in line with the average of expectations of the policy rate over the maturity of the instrument, plus some additional yield to account for the uncertainty entailed in lending capital for a long period of time (the term premium). This means that the yield curve is normally upward sloping and the yield curve slope (the yield of a longer maturity bond less that of a short maturity bond, e.g. 10s-2s) is normally positive. The yield curve slope often turns negative at the end of the cycle when inflation and central bank policy rates are often peaking, and expectations are that they will fall over the coming years.

The track record: Yield curve slope is well regarded as an indicator that doesn’t miss. Between 1976 and 2019, the yield curve has headed negative five times, and within 18 months, the economy has consistently been in recession. Considering 10s-2s, the US yield curve inverted in July 2022, triggering calls that the economy was heading down the shitter. However, three years later, we are yet to see the recession it has been calling.

Is it still useful? A ridiculous amount has changed in global bond markets over recent years, so it is quite hard to point to a simple explanation. Long end yields have faced considerable downward pressure due to QE and structural demand for safe haven assets like US treasuries (for repo, liability matching, or for CB FX reserves), while neutral interest rates have been in structural downtrend and central banks have focussed more on forward guidance, anchoring the long end of the yield curve. However, for me the most important factor remains the nature of the economic shock. Since the 1990s, we have been in a world where the supply side has supported NICE (non-inflationary continuously expansion) growth, where economic expansions have been dominated by demand. An economic cycle dominated by demand will see inflation and interest rates moving in line with economic activity, whereas an economic cycle dominated by supply will see interest rates moving with inflation but not necessarily economic activity. This means you can have a situation where near term expectations of the policy rate can be higher than long, to account for supply driven inflation, without that meaning there are expectations that the central bank will be easing due to low growth.

The Stock Market

The idea: The stock market is thought to begin to decline before a recession is notable in the economic data, with moves in equities containing information on expectations of corporate earnings, bankruptcies, and the general economic environment. The stock market is often expected to bottom before other economic indicators, beginning to rise through the worst of the real world outcomes it predicted.

The track record: The stock market’s forecasting of recessions over recent decades is at best patchy. Equities declined notably in the run up to the recession in the mid 70s, and the recession in the early 2000s. However, they only began declining once recessions had began in both contractions of the double dip recession in the early 80s, the early 90s recession, and the GFC. 2022 saw a huge drawdown in equities, which was not followed by a recession.

Is it still useful? I’m not sure it was ever useful. This is a coincident indicator at best, with some ability to predict the GDP releases which will confirm a recession (e.g. the stock market will decline in a quarter when GDP declines, but you will only find out about the GDP decline some time after that quarter). 2022 appears to have been the result of those big fiscal transfers in 2020 and 2021, which – coupled with easy monetary policy - pushed loads of cash into speculative assets like stocks (and NFTs and SPACs and crypto and probably some other bullshit). Soaring prices in 2020 and 2021 were not necessarily based on economic optimism, so why should falls in the same stock market lead to economic despair? The sharp decline itself could have led to some adverse economic outcomes, perhaps, but we had ChatGPT come out at the end of 2022, kicking off the AI hype cycle and soaring equity valuations. As is evident in 2025, equity markets – at least in the US – remain divorced from economic reality, so it is unclear how they can be a good forecaster of turning points in the economy.

Consumer Confidence

The idea: As we explored last week, households are at the centre of the economy, so a read on how households are feeling can help determine demand throughout the economy. If consumer sentiment falls, that may indicate falling final demand, and a coming recession.

The track record: Consumer confidence, as measured by the UMich survey, has fallen into every recession since the series began in 1978. There is no consistent level to watch given the series shifts over time. Consumer confidence plummeted during 2022, which was not followed by recession.

Is it still useful? Consumer confidence has been a consistent indicator of recession, providing a decent leading indicator. However, there have been clear issues since 2022. To me, it appears that consumers are really bad at interpreting inflation. Consumers are asked how confident they feel and their appetite to make purchases, and will consider how bad inflation is, and reply negatively. But, they neglect that their real wages are holding up OK/recovering, and they actually end up sustaining activity in a way that is inconsistent with their survey responses. This indicator might be useful during a period of stable inflation, but useless when inflation is high.

New Orders For Durable Goods

The idea: Durable goods orders are incredibly cyclical, with households and businesses buying more of these when money is aplenty but cutting back on this area quickly when times are hard. When durable goods orders fall, we can take this as a proxy for the outlook for the consumer.

The track record: Durable goods orders declined well before the recession in the early 2000s, but only declined once the economy was in recession in 2008. There has been no notable decline in durable goods orders since the pandemic.

Is it still useful? Durable goods orders seem in part trying to proxy consumer confidence, which draws the question as to why we don’t just use consumer confidence here. Isolating ourselves to durable goods orders neglects the dominant non-manufacturing side of the economy, and leaves us vulnerable to distortions. For instance, durable goods orders have picked up sharply in 2025 as tariff front running began, and were on a broader pickup since the easing of post-pandemic supply chain pressures.

Initial Jobless Claims

The idea: As times get tough for firms, they begin to lay off workers, driving up initial claims.

The track record: Initial claims have a strong track record of rising significantly before/during US recessions, having done so for every recession since 1967. However, there are clear outliers in the series where a single month comes in unexpectedly high. That said, the four week moving average series may be a better series to use. There are false positives which distort the indicator.

Is it still useful? The labour market remains key to the economic outlook and initial claims are a consistent leading indicator of the labour market. However, there are some weaknesses in this series, for instance, short run volatility in the reads, while the threshold to use for a significant read may change over time. Finally, this series may mask trends, for instance whether the claims are due to job cuts in the public sector which are not indicative of private sector weakness, while job losses in higher paying industries may not lead workers to make unemployment claims to the same extent as lower paying.

Building Permits

The idea: Construction is one of the most sensitive sectors of the economy to the economic cycle, with activity often slowing in the sector in the run up to downturns. Historically, unemployment during US recessions is concentrated in construction. Building permits give a forward indicator of desire to build, so are indicative of the outlook for a sector which tends to lead the economy into contraction.

The track record: Building permits have declined into every single recession since 1960; however, there are false flags. For instance, there was a major fall in permits in 1966, which was not followed by recession. Permits fell fairly sharply in 2022, before stagnating.

Is it still useful? It is unclear to the extent that building permits might be impacted by the nature of the economic cycle, in a similar manner to the yield curve. If we are in a demand driven cycle, then permits could be a great read on coming activity. But in a supply driven cycle, where factors such as interest rates and higher input costs are weighing on propensity to build, the usefulness of the indicator may decline. Further, permits may be impacted by idiosyncratic construction factors that do not impact the rest of the economy, like trends in family formation, changes in planning rules (although in the US these tend to be driven by local rather than federal decisions), or mortgage financing changes.

The Sahm Rule

The idea: Unlike our other indicators, which are forward looking, the Sahm Rule is intended as a coincident indicator of recession. The Sahm Rule signals a recession should the three month moving average of the unemployment rate move 0.5pp above its low over the last 12 months. This translates to a sharp rise in unemployment signals a recession.

The track record: The Sahm Rule is a consistent coincident/lagging indicator of turning points in the economy, with the rule consistently triggered early in a recession. The Sahm rule was triggered during Q3 2024, with the three month average unemployment rate peaking 0.57pp above its 12 month lows in August.

Is it still useful? This rule was created by then Fed economist Claudia Sahm (who has a great Substack) in 2019, which does raise the prospect of some overfitting. Why 0.5pp? Why the 12 month low? Why the three month moving average? It’s good to have a simple rule as to when we should be concerned with how rapidly the unemployment rate is rising, but this is not a fool proof recession indicator. Sahm herself has been clear on this following her own rule being triggered in 2024, citing post pandemic labour market shifts distorting the read.

OECD Composite Leading Indicator

The idea: The OECD’s CLIs are intended to identify turning points in the economy. They are composites of several underlying indicators, which we have largely spoken about already. Given their breadth, and less exposure to idiosyncrasies which can distort the constituent indicators, they should give relatively few false signals. For the US, the OECD CLI is made up of the below indicators:

The track record: With the CLI, a read below 100 is consistent with growth lower than the long-term trend, and we might start to become concerned when the reads push below 99.5, and reads below 99.0 might be considered recessionary. Unlike the Sahm Rule, these are not predefined levels. Between 1955 and 2019, the 99 threshold was crossed 12 times, with 9 of these coinciding with or being followed by recession. The US CLI fell to 98.93 in February 2023, which I’m not sure counts as broken fully.

Is it still useful? The CLI is our most comprehensive measure of turning points, and it’s good to have a broad indicator which should average out to some extent the recent distortions to the underlying indicators. The big issue is just where to define our threshold for a recession signal. At 99, through 1955 to 2019, we had zero recessions missed (type II errors) but three false recessions were identified (type I errors). However, the 99 threshold was generally only breached once recession was well underway. We could raise the threshold, and this indicator could be more leading, but it would be at the expense of more false positives. This is a clear weakness, but the broadness of this indicator and its availability across many economies (see table below, from the OECD) is a huge plus.

In conclusion, when observing these indicators, we need to consider idiosyncratic factors which may impact some indicators, reasons why survey respondents may have distorted responses (e.g. elevated inflation), whether we are in a demand or supply dominated economic cycle, or major changes in monetary policy. Each of our indicators have issues, meaning we cannot use them blindly. But if we use them in conjunction with each other, considering our caveats, they can be powerful predictors of turning points in the economy.

JB Macro is my blog, where I splurge out my brain. I’m building a following for my passion, writing about economics and markets, and it would be really great to have you on board. Please consider pressing the subscribe button below (it’s free!!). Thank you, James.

This newsletter is for informational purposes only. It does not constitute investment advice or an offer to invest. The views expressed herein are the opinions of JB Macro exclusively. Readers should conduct their own research and consult with professional advisors before making any investment decisions.