As if you needed more.
In this post we’ll stick to the descriptive parts of these notes and leave aside the prescriptive, as the potential efficacy of various monetary loosening measures has been (and will continue to be) debated ad nauseum pretty much everywhere, including on this blog.
First up is an analysis from Goldman about increasing labour market hysteresis, the notion that long-term unemployment has begun to coagulate and thereby raised the level of structural unemployment.
Here’s a quick primer on Goldman’s model (emphasis ours in all cases, with paragraphs broken up to make it more digestible):
Using data for 20 OECD countries from 1985 until 2007 this model identifies three key determinants of changes in structural unemployment.
First, adverse shocks—proxied by a decline in real GDP growth below trend—raise structural unemployment.
Second, structural unemployment rises with less “flexible” labor market institutions. In particular, higher unemployment benefits, higher labor union density and lower public spending on labor market programs raise structural unemployment.
Third, adverse shocks interact with the institutional variables in the model. For example, a given decline in GDP growth raises structural unemployment by more when unemployment benefits are high.
All the usual caveats for such models and their underlying assumptions apply, and we’re not here to relitigate the structural vs cyclical unemployment debate. (We think the distinction between the two is somewhat blurrier than most ardent advocates of either position are willing to admit.)
The point here is simply to highlight again the idea the high share of long-term unemployed (42.9 per cent of total unemployed have been jobless for more than 26 weeks, against the historical average of 15 per cent) continues to do increasingly lasting damage.
At the end of last year, Goldman’s model spat out a rise in structural unemployment of half a percentage point, climbing to 5.5 per cent. And now…
Today we extend this model to allow explicitly for labor market “hysteresis.” We do so by including a lag of the unemployment gap (the difference between actual and structural unemployment) in our explanation of the change in structural unemployment. This effect turns out to be highly statistically significant. We also find evidence that this effect depends on the “flexibility” of the labor market. For example, today’s unemployed workers are less likely to become long-term unemployed when active labor market programs — which provide retraining and job search support — are in place.
Taken together, our results suggest that the structural unemployment rate tends to rise by 6 basis points (bp) for each percentage point by which the actual unemployment rate exceeded the structural one in the previous year. Specifically, this updated model suggests that the structural unemployment rate has risen further this year, from around 5½% last year to around 5¾%.
Barring an improvement, here is Goldman’s forecast for what will happen:
First, we consider a “trend growth” scenario in which we use our current 2% GDP growth forecast for 2012 and assume trend growth (2.5%) thereafter. In this scenario, the unemployment rate drifts up to 9.4% at the end of 2012 (as we currently forecast) and then remains stuck at this level thereafter. In this scenario, the structural unemployment rate rises from the current 5¾% to 6½% in 2015 and 7¼% in 2020 as significant “hysteresis” takes place. Second, we consider a “strong growth” scenario in which we assume an optimistic 4% GDP growth rate from 2012 onwards. In this scenario, the structural unemployment rate peaks at around 6% and then declines to 5¼% in 2020.
Second is an interesting spot by from Barclays Capital, which found a correlation between historically high market volatility (measured by the Vix) and weak employment growth:
Historically, job creation generally increases as uncertainty declines, and vice versa. Clearly, however, many factors influence payroll growth — so to estimate the relationship between payroll growth and uncertainty more precisely, we use an econometric model that also takes into account a range of economic factors. According to estimates from this model, a 10pt rise in the VIX index on a m/m basis lowers private nonfarm payroll growth the following month by about 75k. These estimates suggest that if uncertainty remains elevated at its current level of around 30, net private job creation proceed at a pace of around 90k per month. If, instead, the VIX receded to levels consistent with last June (e.g., about 20), the model indicates that private job growth would proceed at a pace of about 165k.
The Vix is now at about 34 and has been above 30 since early August.
Barcap add that this depressing relationship is strengthened when business confidence is low (as measured by an NFIB optimism index of more than 115) and weaker when confidence is high (above 115).
Guess where we are now.
Just something to chew on as you prep for the start of the two-day Fed meeting tomorrow.
Related links:
Breaking on FedWire – FT Alphaville
Hysteresis begins – Krugman