By Jamie McGeever

ORLANDO, Florida The longer the heavily touted U.S. recession fails to materialize, the more doubt is cast over the relevance and usefulness of leading economic indicators that have accurately predicted every downturn for decades.

The economy grew much faster than expected in the first quarter of the year, unemployment is ultra-low, job growth remains solid, and inflation is decelerating fast.

Recession calls are getting pushed back - early next year is the latest new dawn - and the 'soft landing' narrative is regaining traction.

Federal Reserve Chair Jerome Powell believes avoiding a recession is more likely than not, which would be remarkable after the central bank's most aggressive interest rate-hiking campaign in 40 years.

"It's possible that this time really is different," he said in May.

But tempting as it is to buy into that - leading indicators have been flashing red for months, as yet to no avail - we are probably not at that stage just yet.

These signals include an inverted yield curve, plunging consumer confidence, weak industrial production and slowing bank credit. The Conference Board rolls 10 of these components into its Leading Economic Indicator (LEI) index.

The lead time between the LEI index falling from its cycle peak and the start of all but one of the last eight recessions has ranged from nine months, in the early 1970s, to 22 months, before the 2007-09 Great Financial Crisis.

The average lag time is 14 months, according to Eric Basmajian, founder of economics research firm EPB Research, who excludes 1981 from his calculations because the LEI index did not make a cycle peak then.

The lag time now is 17 months, and counting. In other words, if recession doesn't hit by the year end it will mark the longest ever lead time - new territory, or a broken model?

But Basmajian cautions against dismissing the signals that have worked for decades - they are still within historical ranges, and their track record still stands up to scrutiny.

"Most analysts have no choice but to have their initial bias gravitate to the mean or median range of these leading indicators," he said. "But we are certainly exceeding the mean and average of these historical outcomes."

LONG AND VARIABLE LAGS

One of the most reliable recession indicators is the spread between three-month and 10-year U.S. bond yields. An inverted curve - longer-dated yields falling below short-term borrowing costs - has always preceded recession, as well as giving a couple of head fakes in 1998 and 1971.

The average lead time between inversion and recession going back more than half a century is 11 months. The current lag is nine months, approaching the average but still well short of the peak 17-month lag before the GFC.

Again, if the economy isn't in recession by the end of the year, this time it really is different.

The post-pandemic economy is certainly not following the pre-pandemic playbook.

Trillions of dollars of fiscal and monetary stimulus, the lockdown and re-opening, and inflation pressures sparked by global supply chain disruption and war in Ukraine have distorted all aspects of the economy. Perhaps permanently.

Patterns of consumer savings and spending, corporate hiring and firing, and business activity in the last few years were not in pre-pandemic models. It stands to reason their signals were off.

Forecasting models based on a quarter century of the 'Great Moderation' - falling market and macro volatility - have been exposed in the post-pandemic world. The Phillips Curve, the once orthodox theory explaining the relationship between inflation and employment, has lost its luster.

But economists and analysts should be cut some slack. How else were they meant to interpret collapses in bank lending, consumer confidence, industrial production and yield curves consistent with almost every recession in the past 50 years?

Milton Friedman's "long and variable lags" may just be a little longer and a little more variable. When it comes, recession will not be official until the National Bureau of Economic Research calls it, and that could be backdated by some time.

What's more, it's prone to endless revision anyhow.

In that light, some historical perspective is worth noting. In November 1973, the first month of a recession that lasted until March 1975, a net 313,000 new jobs were created. Non-farm payrolls did not decline for another nine months.

And in December 1969, the onset of a year-long recession, the unemployment rate was 3.5%, then - and still - one of the lowest on record.

Of course, employment is a lagging indicator, and so more likely to send misleading signals. The signals sent by leading indicators recently have been pretty clear - it just remains to be seen whether they will be accurate.

(The opinions expressed here are those of the author, a columnist for Reuters.)

(By Jamie McGeever; Editing by Jan Harvey)