Why falling unemployment may not be making voters happy

The unemployment rate dropped in November to 8.6 percent from 9.0 percent in October and from 9.8 percent a year ago. This is clearly welcome news. However, the underlying dynamics of the drop-off in unemployment this last month and over the last year are disappointing and have clear implications for policy and for politicians.

The issue is a decline in labor force participation, a topic that both Jared Bernstein and Ezra Klein have picked up on. To be blunt, among groups with high voter turnout rates, the fall in unemployment has been driven by people leaving the labor force and not because of job gains: this applies to those 25 and older who have a high school credential, some college, or a college degree or further education. In contrast, job gains were responsible for falling unemployment among lighter voting groups: young people (ages 16-24) and the 8.0 percent of the labor force that lacks a high school credential. The only exception to this breakdown is that job gains lowered the unemployment rate of those 55 and older (but only 40 percent of this group is in the labor force). Among women, unemployment has fallen very little (0.3 percent) while employment has fallen as well, indicating that job growth has not driven their modest unemployment gains. Men, in contrast, have seen a large drop in unemployment (1.2 percent) but modest growth in employment, indicating a shrinking labor force as the major explanation.

Overall, the dynamics in the labor market do not point to people generally feeling happier or more prosperous because a great deal of the falloff in unemployment is not because people are earning money in newly found employment, but because people are no longer in the labor market. There are some analysts who point to demographic changes (e.g., the population aging) as a reason to expect labor force participation to not return to prior levels: however, such longer-term trends are not salient in explaining the trend over the last year because such demographic shifts occur gradually.

This morning’s news prompted me to do a bit of analysis on how much of the drop in unemployment over the last year is due to greater employment and how much is due to the shrinkage of the labor force. It is not easy to produce a clean decomposition, but simply displaying the trends in the unemployment rate, the employment rate (the share of the population employed), and the labor force participation rate (the share of the population in the labor force, meaning they are either employed or unemployed) certainly helps. The table below presents the data for key demographic groups along with the shares of the labor force of each group. The data are for the most recent three months compared to the comparable months a year ago (avoiding the volatility of one month’s data).

Labor Force Share* Unemployment rate Labor force/population Employment/population
Sept.-Nov. Sept.-Nov. Sept.-Nov.
2010 2011 Change 2010 2011 Change 2010 2011 Change
All 100.0% 9.7 8.9 -0.8 64.6 64.1 -0.4 58.3 58.4 0.1
Education, 25 years and older
Less Than High School 8% 15.5 13.7 -1.8 46.8 47.0 0.2 39.5 40.5 1.0
High School 24% 10.0 9.4 -0.7 61.5 60.5 -1.1 55.4 54.8 -0.5
Some College 24% 8.8 8.1 -0.7 70.1 69.0 -1.1 64.0 63.4 -0.6
College Degree or More 31% 4.8 4.3 -0.4 76.4 76.0 -0.4 72.8 72.7 -0.1
By Age
16-24 14% 18.3 17.0 -1.3 55.2 55.5 0.3 45.1 46.1 1.0
25-54 66% 8.6 7.9 -0.7 82.0 81.4 -0.6 74.9 74.9 0.0
55+ 20% 7.2 6.7 -0.5 40.2 40.4 0.2 37.3 37.7 0.4
By Race/Ethnicity
White 81.0% 8.8 7.9 -0.9 64.9 64.5 -0.5 59.2 59.4 0.2
Black 11.6% 15.9 15.5 -0.4 62.1 61.7 -0.4 52.3 52.1 -0.2
Hispanic 15.1% 12.8 11.4 -1.4 67.3 66.7 -0.6 58.7 59.1 0.4
By Gender
Male 53.5% 10.4 9.3 -1.2 71.0 70.5 -0.5 63.6 63.9 0.3
Female 46.5% 8.8 8.5 -0.3 58.5 58.1 -0.4 53.4 53.2 -0.2
* Labor Force in November 2011. Shares by race/ethnicity sum to greater than 100% because Hispanics can be of any race.


The top line tells a clear story that unemployment fell by 0.8 percentage points but the share of the population employed rose by just 0.1 percentage point. The share of the population in the labor force fell by 0.4 percentage points. This tells you that in the aggregate it was not greater employment driving the drop in unemployment.

The breakdowns by education level are pretty startling though. The unemployment rate fell for every educational group but was far greater for those lacking a high school credential, the group with the highest rate of unemployment. This is not surprising since those with the least education see their unemployment rise more in downturns and fall more in recoveries. What is a bit of news, though, is that over the last year the employment rate did not grow for any group except those without a high school degree and it fell by roughly one-half a percentage point for the middle groups who have a high school degree or some college experience (comprising 48 percent of the labor force). Moreover, labor force participation declines (people giving up looking for work or failing to return after having dropped out) completely explain the drop in unemployment among those with a high school degree or further education.

This dynamic is confirmed when looking at the age breakdown. Unemployment fell 0.7 percent among prime-age workers but there was NO growth in the employment rate in this group. Both employment and labor force participation grew among those who were young and those who were 55 or older (I’m NOT going to call those folks old!).

Among blacks, there was shrinkage in employment and labor force participation and the smallest fall in unemployment. The drop in Hispanic unemployment was due to both job gains and labor force shrinkage. Whites saw a larger decline in unemployment but employment growth played a minor part of this improvement.

Finally, these data suggest that women should not be expected to feel very happy about the last year. Their unemployment did fall modestly (0.3 percent) but that was in spite of erosion in women’s rate of employment (down 0.2 percent) and because of labor force shrinkage (down 0.4 percentage points). Men’s unemployment started much higher (at 10.3 percent) than that of women and fell much more (1.2 percent). Still, men’s employment rate grew modestly (up 0.3 percent) while their labor force attachment shrank by 0.5 percent.

The bottom line is that the fall in unemployment looks good but it is very likely leaving a broad swath of the electorate not feeling very good.


  • Conrad Ornstein

    And where have all the jobs gone?
    ‎’U.S.-based multinational corporations added 1.5 million workers to their payrolls in Asia and the Pacific region during the 2000s, and 477,500 workers in Latin America, while cutting payrolls at home by 864,000, the Commerce Department reported.’Wall Street Journal, Nov. 22, 2011

    • JWW

      Excellent point…  a great example of this is to look no further than the darling of the current administration, GE. 

  • Bynary Fission

    This is a great analysis, but are you using the household data to calculate the growth in employment, or the payrolls survey? If you are using the payrolls survey then your stats won’t be accurate, because unemployment is calculated from the household survey, which has increased 1.3 million in the past 4 months, and accounts for much of the gain.

    From what I know, some of the drop in unemployment is genuine – remember that the household survey rose by 278,000 this month which accounted for fully half of the drop in unemployment. Furthermore, since the sharp gains in the household survey in August the labor force has still grown by over 600,000 in that time including November’s drop.

    Just my two cents. The labor force hasn’t grown relative to November 2010 (it’s the same as from that time), so it’s feasible that given the major gains in the household survey unemployment would drop considerably given the sharp rise in household employment.

    As a final note, I noticed on your chart that the unemployment in Nov 2010 was 9.7% versus 8.6% now. If you account for the 0.4 drop in the labor force participation rate, that only bring it up to 9.0%. I believe that discrepancy is due to strong payroll growth not picked up in the survey as well as a stagnant pool of participants in the labor force (healthy growth relative to historical trends would have resulted in a much smaller drop, naturally)

  • John Dennis Chassr

    I wish you would look more closely at what is happening at the state and local level.  The GAO 2010 report on their fiscal pressures is a little frightening.  If we could get rid of the anti-tax sentiment at the national level, it would be time to talk about revenue sharing–perhaps a national sales tax on internet sales completely redistributed to the states using the old revenue sharing formual

  • Rsfenwick2

    Thanks Larry.  Another example why the official unemployment rate’s explanatory power is limited at best.  I wish more people understood this.

    • JWW

      I agree that I wish more people understood this matter.  However, we are in an environment where the entities responsible for reporting accurate news is not doing their jobs – Rather, the news too often is ‘agenda bias’d’ and not objective, which unfortunately results in only part of the story to be told, and only the part of the story which supports the media source and their agenda.