U.S. Unemployment

I am tired of Bloomberg News quoting single numbers and percentages in their articles on unemployment, it gives me an incomplete picture. So I decided to graph the relevant series myself. I sourced the data from the St.Louis Fed and started graphing.


The gray areas are the recessions as reported by the FED. The current recession has apparently ended, but even the FED has the latest date as a placeholder only.

As you can see Roubini and Bernanke are "right". This recovery is jobless. Never before has the duration of unemployment been so long. Basically, people who lost their jobs are still unemployed. The "official" unemployment is at 10%, but as you can see in the second graph the labour force has decreased in the last year, which is strange since the labour force has a general uptrend. Also the participation rate is down, which means that the FED is hiding unemployment behind their definition of "people who have not looked for a job in the past 4 weeks are not part of the labour force".

I created the graph using Python with matplotlib. For a while now I've been using the two to prototype studies at work. Matplotlib is great, and I can see that it's constantly improving. A year ago it could handle plotting only a day or two of high-frequency data, now I can plot a week. The widgets (sliders, radio buttons, etc) included in the package make prototyping extremely easy. I am also learning more and more about Python and I like what I am seeing, generally.

It was fairly easy to put together the graph, but I found out that another package which I like R with quantmod can do the work as well. Choices, choices... maybe I'll end up combining Python with R :)

Now, on to graphing more of the economy ...