Using United States data from 1948 to 2004, I find that there are substantial fluctuations in unemployed workers’ job finding probability at business cycle frequencies, while employed workers’ separation probability is comparative acyclic. This is particularly true in the last two decades, during which period the separation probability has steadily declined despite two spikes in the unemployment rate. In other words, virtually all of the increase in unemployment and decrease in employment during the 1991 and 2001 recessions was a consequence of a reduction in the job finding probability. If one wants to understand fluctuations in unemployment, one must understand fluctuations in the transition rate from unemployment to employment, the ‘outs of unemployment’. This conclusion is in direct opposition to the conventional wisdom, built around research by Darby, Haltiwanger, and Plant (1985) and (1986), Blanchard and Diamond (1990), and Davis and Haltiwanger (1990) and (1992), that recessions are periods characterized primarily by high job loss rates.
I’ve highlighted Shimer’s work before (Robert Hall is also a player in this). It strikes me as important and probably correct.
READER COMMENTS
dave smith
Sep 21 2007 at 10:50am
This is important and interesting since it goes against what nearly everyone thinks.
If this turns out to be true, it should shake things up considerably.
spencer
Sep 21 2007 at 1:34pm
If you look at the new employment dynamics data what you find is that job gains fell sharply in only one year, 2000. But the job loses rose gradually from 1997 to 2001.
This data just seems to further confuse the situation.
Will
Sep 21 2007 at 2:37pm
Does this make Farmer’s search based new “old Keynesian” economics more parsimonious than sticky priced based “new Keynesian” economics?
Matt
Sep 21 2007 at 3:31pm
Sounds like business picks up surplus workers during the upswing start and slowly releases them over a longer cycle. Why?
caveat bettor
Sep 21 2007 at 3:57pm
It is easier to hire employees than to fire them, in emotional, legal, and business process dimensions.
Ray Gunter
Sep 21 2007 at 9:05pm
I would be interested to see how these spikes in unemployment coincide with factors that an industry has no direct control over, such as shifts in foreign and domestic policy and war. History shows us that these outside forces can be either a boom or a bust for a particular industry or market depending on how much pressure they exert upon an industry to produce. My guess is that one would find that many of the spikes are not randomly occurring incidents but are instead the industries response to an outside. It is also probable that many of these spikes are the results of past factors that have slowly been eroding any given industry over a number of years, for example the current loss of jobs that we have seen in recent years in the American automotive market is ultimately a response to long existing forces. Entry of smaller more gas efficient foreign cars into the American market in the 1970’s has been placing steady pressure on U.S manufacturers to produce more efficient vehicles in order to compete, however, domestic and foreign policy was geared to protect U.S manufacturers from direct competition from these foreign competitors. With the breakdown of industry tariffs over the past 20 years we have seen non-American manufacturers grab a substantial share of the market leaving U.S manufacturers with little choice but to employ massive layoffs.
Mark T
Sep 23 2007 at 6:36am
I am a partner in a fairly large prof svcs firm so I will give a brief summary of what I have seen us do the past 20 years since service sector employment is an increasing proportion of employment.
We don’t do layoffs. That would be damaging for our brand in recruiting, which is strong. We have ridden out downcycles through ordinary attrition, minimizing salary increases and bonuses, modest reductions in new hiring, freezes in hiring above the entry level, freezes in hiring support staff and temp help, modest deferrals in admitting new partners and modest reductions in our personal income. Regardless of the cycle, in any given year, we tell 5% or so of the firm to find new employment for performance reasons and another 10-15% will leave for other jobs, lifestyle, etc. so it is pretty easy to get costs down through ordinary attrition and freezing new hiring. So I think our practices are consistent with the new research because we don’t increase layoffs but reduce new hiring and the biggest percentage cutback is in new hires above entry level. We do this for three reasons – one, the competitive advantage of our brand in recruiting must be protected, as I have mentioned. Two, there is greater efficiency in keeping employees who know what to do in our firm versus churning them to match the cycle and incurring recurring training and monitoring burdens. And three, while there is a risk in good times or bad that anyone looking for work above an entry level position has been an underperformer or a problem-causer in his or her current or prior firm, in good times we might take that risk but in bad times we are less inclined to. Although we state this as not being willing to take the risk, I am unsure whether this is because we really think the risk premium that attaches to hiring decisions increases in downcycles or we think the present value of the expected return diminishes due to decreased expectations of revenue generation in the earliest years.
On the branding point, I believe that people entering the job market for the first time are very unfamiliar about the specific characteristics of firms, and uncertain about the world in general. Therefore they are more likely to focus on risk and give great weight to negative information and especially any that suggests one employer is more risky than another. Moreover, media and gossip both tend to publicize negative surprises more than other information.
While anecdotal, I hope this is a contribution. I do suggest some thought be given to possible differences between service sector and other sectors.
Matt
Sep 23 2007 at 10:49am
What Mark T describes is what I expected, raw recruits are a commodity when unemployed and partially converted into a capital investment when hired.
So, the tendency is to properly hire in the shorter business cycle and retire in the longer term depreciation cycle.
When you leave a job early, you retain your skills and salary, but your go back onto the short term hiring cycle.
This research has important implications, the author is worth reading, and Arnold is right if he considers this a good thing.
Comments are closed.