Friday, September 7, 2012

Why Are Metropolitan Unemployment Rates So Divergent?


If you take a look at the United States unemployment data by metropolitan area, you'll notice that unemployment rates vary widely.  While the average unemployment rate across the United States for August 2012 was 8.1 percent, the July metropolitan rate ranged from 2.5 percent in Bismark, North Dakota to 31.2 percent in Yuma, Arizona

A total of 67 out of 372 metropolitan areas in the Bureau of Labor Statistics monthly data base had unemployment rates of 10 percent or greater, representing 18 percent of the total as shown here:  


In contrast, 46 metropolitan areas had unemployment rates of 6 percent or less, representing 12 percent of the total as shown here:


Why the dichotomy?  A recent article by Jordan Rappaport, a senior economist at the Federal Reserve Bank of Kansas City attempts to answer the question.

Mr. Rappaport's research looks back at unemployment data from 2007, the peak of the business cycle before the Great Recession, and reveals the following distribution:



Looking over the longer term from 1990 to 2011, the difference between the 95th percentile (dark blue line) and 5th percentile (black line) metropolitan unemployment rates was always at least 4.5 percentage points and in some years, as shown on this graph, it was twice as high:


Generally, high unemployment rates in metropolitan areas tend to persist over time.  As shown on this graph, metropolitan areas with high unemployment in the period between 1990 and 1999 tended to have high unemployment in the period between 2000 and 2007:


Looking at the U.S. job market as a whole, generally, one can conclude that as national job growth accelerates, national unemployment declines.  For metropolitan unemployment statistics, the same correlation does not hold.  This has been the problem facing unemployed workers since the end of the Great Recession.  As a result of worker migration from one metropolitan job market to another, long-term metropolitan unemployment rates do not correlate with long-term metropolitan job creation.  In the period between 1990 and 2000, metro employment growth accounts for less than 1 percent of the variation in metro unemployment rates and in the period between 2000 and 2007, metro employment growth accounts for only 4 percent of the variation.

Worker migration is a key part of the increases in metro employment and variations in metro unemployment rates.  Between 1990 and 2000, for every 100 jobs added by metro areas, there was an average inflow of 104 workers.  Between 2000 and 2007, this number rose to 105.  Nearly all new jobs were filled by workers migrating in rather than by workers already living there.  The influx of new workers is also accompanied by changes in labor force participation; from 1990 to 2000, the excess of the inflow of workers over the number of new jobs was offset by a modest decrease in labor force participation.  Between 2000 and 2007, labor force participation increased; this, combined with an excess of migrant workers, led to a higher unemployment rate.  In fact, over this period, for every 100 jobs created, a median of 16 additional unemployed workers resulted in higher unemployment rates.  In the cases where  metro areas noted a drop in net employment, laborer migration outflow was minimal.  In the period from 1990 to 2000, out of 100 jobs lost, only 26 workers migrated out.  In the period from 2000 to 2007, labor outflow was almost zero, meaning that nearly all workers who lost their jobs remained where they were.    Putting all of these observations together, we see that metro areas that experienced employment increases typically saw no decrease in unemployment due to an inflow of workers.  Areas that saw employment decreases, saw increases in unemployment because laid-off workers did not migrate to other metro areas.  This is key to what may be the employment market problem since the end of the Great Recession.

Certain metropolitan areas have intrinsic advantages that others do not that explains part of the differences in metropolitan unemployment rates.  These include exogenous characteristics that are not caused by unemployment; for instance, weather, proximity to coastlines and how close the metro area is to mineral and energy deposits.  As well, positive aspects that are related to unemployment may include low taxes, high quality government services, crime rates, restaurants, sports teams, pollution, ample cultural opportunities, access to transportation infrastructure and easy access to developable land.  Of the exogenous characteristics, weather has the strongest connection to the unemployment rate.  Alone, weather accounts for between 36 and 40 percent of the variation in metropolitan unemployment rates, more than proximity to a coastline.  Households and firms are more likely to locate in areas with a reasonable climate wherever possible, unless proximity to mineral and energy deposits is key.  Just ask people living in North Dakota during January and February!

Why are unemployed workers reluctant to move to metropolitan areas with lower unemployment rates?  High moving costs definitely contribute to worker immobility; the total monetary cost of a city-to-city move including commissions, legal fees, mortgage costs and moving expenses is estimated to be in the range of $40,000 or more, a significant portion of a median household's income.  On top of that, the intangible costs of leaving friends, family, schools and health professionals may throw up a barrier to migration.  Laid-off workers who face low moving costs may be willing to move for small gains in employment prospects, dampening the rise in unemployment in their original metro area.  Laid off workers who face high moving costs may prefer to stay put, resulting in elevated unemployment levels in their metropolitan area, further widening the unemployment rate between one metropolitan area and another.

It is intriguing to see how widely variable American unemployment rates are from one metropolitan area to another.  Despite the fact that the economy is now over three years into the "recovery phase" and that national unemployment is down from its peak, there are many metropolitan areas that are suffering from U-3 unemployment levels that are well above the rates seen in downturns prior to the Great Recession.  The fact that high unemployment levels tend to "follow" a community over a period of decades is of concern and does not particularly bode well for the next recession, particularly since many of these metropolitan areas have not seen their unemployment rates fall to anything approximating their pre-Great Recession levels.  


No comments:

Post a Comment