Anatomy of Recessions

There is a widespread belief that recessions are periods when the economy “stops creating jobs.” Television anchors say it. Politicians repeat it. Even some economists speak as if the flow of job creation dries up during downturns, leaving workers stranded on a barren shore. The BDS data tells a fundamentally different story. In Q1 2007 — the last quarter before the Great Financial Crisis began to bite — the American economy created 7.8 million jobs at a rate of 6.9%. In Q1 2009 — the worst quarter of the crisis — it created 5.9 million at a rate of 5.4%. That is a decline, certainly, but it is a decline from a very large number to a slightly less large number. The economy did not stop creating jobs. It kept creating them at a pace that would have seemed robust in any other context.

What changed was destruction. In Q1 2007, the economy destroyed 7.2 million jobs at a rate of 6.4%. By Q1 2009, it was destroying 8.6 million at 7.8%. The loss rate increased by 1.4 percentage points while the gain rate decreased by 1.5 points. In absolute terms, the economy lost 1.4 million more jobs per quarter than it had before the crisis, while creating 1.9 million fewer. Both sides contributed to the net deterioration, but the increase in destruction was the sharper, more sudden change. The gain rate declined gradually over six quarters, like a tide going out. The loss rate surged in two quarters, like a wave crashing.

This asymmetry has a simple explanation rooted in the mechanics of business decision-making. Hiring is forward-looking. A company hires because it expects future demand, and expectations adjust gradually. Even during a recession, some companies are growing, some sectors are expanding, and some regions are gaining population. The flow of new hiring never stops because the economy is never uniformly bad. Firing, by contrast, is reactive. A company lays off workers because revenue has already fallen, because orders have already been cancelled, because the bank has already called the loan. The decision to fire is usually made under duress, often quickly, and often in large batches. This is why the loss rate can spike suddenly — thousands of companies making the same reactive decision in the same quarter — while the gain rate drifts down more gently.

The BDS data captures this asymmetry with quarterly precision. From Q1 2007 to Q4 2008, the gain rate fell from 6.9% to 6.0% — a slow, steady decline of 0.15 points per quarter. Over the same period, the loss rate rose from 6.4% to 7.7% — an accelerating climb that went from barely noticeable to catastrophic in the final two quarters of 2008, when Lehman Brothers collapsed, AIG was bailed out, and the credit markets froze. The loss rate’s spike in Q4 2008 (+1.2 points from Q3 2007’s level) was the single largest quarterly deterioration in the BDS record. It reflected a moment when thousands of companies, unable to borrow, unable to refinance, and facing evaporating demand, all decided to lay off workers simultaneously.

The GFC did not hit all industries equally. Construction was ground zero. Its net creation rate went from +0.6% in Q1 2007 to −6.6% in Q1 2009 — the steepest collapse of any sector. In a single quarter, the construction industry was destroying jobs at nearly twice the rate it was creating them: a gain rate of 8.8% against a loss rate of 15.4%. The arithmetic is staggering. One in six construction jobs was lost in a single quarter through contractions and closings, while the industry continued to create new jobs at a near-normal pace. The workers pouring foundations for a new hospital in Houston were being hired even as the workers framing houses in Las Vegas were being laid off. The industry was simultaneously creating and destroying at a furious pace, and the destruction was winning.

Natural resources and mining posted a net rate of −6.4%, driven by the collapse in oil prices from $145 a barrel in July 2008 to $33 in February 2009. Manufacturing, already in secular decline, saw its loss rate nearly double from 4.2% to 7.7%, yielding a net of −5.3%. The Big Three automakers — General Motors, Chrysler, and Ford — were all on the brink of bankruptcy. GM and Chrysler would eventually file, while Ford survived only by mortgaging every asset it owned, including the blue oval logo itself. The manufacturing sector as a whole shed 876,000 jobs in a single quarter.

Professional and business services, a bellwether for white-collar employment, went from +0.5% net to −3.1%. This sector includes the temporary staffing agencies that are often the first to cut workers in a downturn, because temp contracts can be cancelled overnight. In the GFC, companies that had been using temps as a flexible buffer suddenly cancelled those contracts en masse. Manpower, Kelly Services, and Robert Half — the big staffing firms — all reported revenue declines of 20-30% in 2009. The BDS captures these cuts not as a decline in the professional services sector’s gain rate (which fell only from 7.6% to 6.0%), but as a spike in its loss rate (from 7.1% to 9.1%).

Two sectors were essentially recession-proof. Education and health services posted a positive net creation rate of +0.3% even at the trough of the crisis. Hospitals do not close during recessions. Schools do not lay off teachers (at least not immediately). The sector’s loss rate barely moved, from 4.1% to 4.1%. Utilities were similarly insulated, posting a net of +0.2%. People do not stop using electricity when the stock market crashes. The split between cyclical and defensive sectors was as stark during the GFC as at any point in the BDS record, with construction at −6.6% and healthcare at +0.3% — a 6.9-point gap between the best and worst sectors.

If recessions are defined by a sudden spike in destruction, recoveries are defined by a gradual normalization of that destruction back to baseline levels. The GFC created six consecutive quarters of net negative job creation, from Q3 2008 through Q4 2009. During those six quarters, the economy destroyed a cumulative 4.7 million more jobs than it created. The recovery that followed was not symmetrical. It took until Q2 2010 for the net rate to turn positive, and the recovery was driven almost entirely by the loss rate returning to normal — not by any surge in the gain rate.

The gain rate at the GFC trough (5.4% in Q1 2009) recovered to 6.0% by Q1 2010 and then essentially plateaued. It never returned to its pre-crisis level of 6.9%. The loss rate, meanwhile, fell from 7.8% in Q1 2009 to 6.2% by Q1 2010, and eventually settled at 5.6% by 2012. The recovery, in other words, was not a story of the economy learning to create more jobs. It was a story of the economy learning to stop destroying them. The elevated loss rate was the disease; its normalization was the cure. The gain rate remained permanently lower, contributing to the sense that the post-GFC recovery was “jobless” or slow — an impression that was technically correct but misleadingly attributed to weak hiring rather than to the structural decline in dynamism that Episode 2 documented.

The pattern shows up clearly in the absolute job numbers. In Q1 2007, the economy created 7.8 million jobs and destroyed 7.2 million, for a net gain of 584,000. By Q1 2012, it was creating 7.1 million and destroying 6.1 million, for a net gain of 948,000. The net gain in 2012 was larger than in 2007, but only because the loss rate had fallen faster than the gain rate. The economy was recovering not because companies were hiring with renewed optimism, but because they had finished laying off the workers they no longer needed. The distinction matters for policy: if the binding constraint is weak hiring, the appropriate response is demand stimulus. If the binding constraint is elevated firing, the appropriate response might be different — credit easing, foreclosure prevention, bankruptcy reform.

The closing rate tells this story even more clearly. The share of job losses coming from establishment closings (as opposed to contractions) spiked from 1.2% to 1.4% during the crisis — a seemingly modest increase that, applied to a labor force of 110 million, translated to 220,000 additional jobs lost to permanent closures per quarter. These are not jobs that return when the economy recovers. A restaurant that closes, a construction firm that dissolves, a small manufacturer that shutters — these establishments are gone, and the jobs go with them. The recovery can only replace those jobs through new openings, which is a slower and more uncertain process than the expansion of existing firms.