Employment-Population Ratio - a scarier way to look at unemployment

The Bureau of labor Statistics compiles a stat called the Employment-Population Ratio that is the number of people working of the available workforce, defined as people 16 and over. I am sure some of those folks are not working because they wouldn't otherwise, but in the current climate many have no choice. As has been reported repeatedly, the 9.2% unemployment underestimates the number of people unemployed because after a long enough time people just leave the workforce and are not counted in the number. The chart below shows the current ratio is 58.2%, that's the lowest Employment-Population ratio since 1984.

A longer time series shows the history since 1948.

It's been lower in the past, but that is a generation ago. I personally would like Congress and the President to start working on jobs now.

(Unstable Isotope at Delaware Liberal got me thinking about this)

Number unemployed vs. unemployment rate

While collecting the unemployment data I also ran some reports on the time that people are unemployed. The data is available showing up to 5 weeks, 5 to 14 weeks, 15 to 26 weeks and longer than 26 weeks. One of the signs of the depth of this recession is not just the number of people unemployed but also the length of time they have been unemployed. The charts below show the number of people unemployed for a given number of weeks in a bar chart which builds the categories on top of each other. I have overlaid the rate of unemployment which corresponds to the right axis. (click for larger)

The chart below just shows the faction in the various unemployment categories adding up to 100%, instead of the actual number. (click for larger)

If I have time, I want to overlay dates of unemployment extensions and see if there is any correlation. I still don't think that would be completely relevant because it is obvious that if an extension was passed then the time people receive unemployment insurance will be longer. Will I be able to see whether this in fact means they stayed unemployed longer or do extensions occur when Congress sees that people are already unemployed longer and the extension is to address that issue?

(Once again the data is available in this spreadsheet on Google Docs. - How Long unemployed vs rate unemployed )

Recreating Laffer's unemployment benefits payout vs. unemployment rate chart but not his ridiculous conclusions

Howard points to an excellent refutation of the ridiculous claims by Laffer that unemployment benefits drive unemployment. The chart from the opinion page clearly shows a lag of 10 months to a year of the peak in benefits payouts vs. unemployment rate peaks. Perhaps it is unemployment which drives unemployment benefits.

One thing that drives me nuts on these charts is when the presenter doesn't cite the data. I would make it a requirement that for any charts like these that you must publish your work and the data you used, but as a minimum the presenter should cite the references. Laffer cites as his source Laffer associates, but I guarantee you they didn't generate the data. References for the data are given below. I have recreated the chart below (click for large size).

The unemployment rate is the straight data from the Bureau of Labor Statistics, seasonally adjusted, while the benefits paid is the average weeks on unemployment multiplied by the average weekly benefits per person and is from the Dept of Labor source below, but the key is that it needed to be adjusted to 2010 dollars. Perhaps Mr. Laffer could have included that in his description.

Once I recreated the chart and data, it is simple to see that the peaks in the benefits paid occur from 9 to 13 months after the peak in unemployment rate. Even the most egregious examples of correlation being mistaken for causation have the cause preceding the effect, unlike here. I am sure someone even more mathematically inclined can run some sort of time series correlation with the data, now that I have provided it.

Data sources and links follow:

Unemployment statistics from the Bureau of Labor Statistics. Use this form to enter in your choices and you can get an excel file of the data. I only changed the Labor force status to Unemployed or to Unemployed rate and checked the monthly data and seasonally adjusted boxes to generate my files for actual number unemployed and rate of unemployment. You may need to adjust the years on the next page. Not used but you can get the duration of unemployment benefits using this page and checking the appropriate boxes.

Unemployment claim amounts paid can be found under the on the program statistics page, which leads to this form from the United States Department of Labor. You can uses this calculator to get the yearly adjustments for inflation. I actually used the tables from here for CF adjusted to 2007, and the calculator to adjust 2007 to 2010 (1.05).

Data to create the charts above is in this google docs spreadsheet - Unemployment statistics 1971 to 2010.