A Quits vs Hires Beveridge Curve?

Stephen Gordon, on WCI, had a nice post exploring the JOLTS data coming out of the US (Another view of the sluggish US labour market). I wanted to follow up on Stephen's idea that in 2009 separations -- both quits and layoffs -- fell to match hires. I like this idea, particularly as it seems that quits are hires are strongly correlated. My base model is that quits is dependent upon hires, i.e., if there is lots of hiring going on in the job market people are more likely to quit.

In particular I wanted to look at the equivalent Beveridge curve for quits and hires. What I'm seeing is that not only did quits fall to match the drop in hires, but that the number of quits fell further than one would expect assuming a linear relationship between quits and hires.

In the chart above I've plotted Quits versus Hires over the Dec 2000 - August 2010 period, and I've split the data into two groups. The pre-crash period from Dec 2000 - Oct 1, 2008 is plotted in black and the Nov 1, 2008 - Aug 1, 2010 data is plotted in purple. I calculated the regression over pre-crash period, and plotted this regression in blue.

Assuming the linear regression is an accurate description of the relationship between hires and quits, the number of quits, post Nov 1, 2008 have fallen below what would be expected by the number of hires. Echoing what Stephen said, I'm not sure if this is good news, but it probably has helped keep the unemployment rate from going higher.