Archive for graphs

What The July Unemployement Rate Means (And Doesn't Mean)

Today the unemployment rate for July 2009 was released by the Bureau of Labor Statistics. The rate dropped from 9.5% in June to 9.4% in July.

Before I explain why this might not be as awesome as it looks, let me just say “hooray!” for what seems like a slowing in the rise of the unemployment rate. I am ecstatic to see that the economy is not accelerating downward.

Stupid Moralizing (skip if you don’t care)

Some people seem to almost be cheering the decline of the economy for political purposes. Before last November, those people were mostly liberals. After January, those people were mostly conservatives. It is an activity I find creepy and slimy.

The decline of the economy means people losing their jobs, losing businesses that they’ve spent years trying to painstakingly build. This can be devastating on every level, personal and professional. The pain it brings is almost unspeakable. When someone cheers or hopes for a decline in the economy simply so that their political team can come out ahead, they reveal themselves to be without the basic human emotion of sympathy.

I don’t give a crap who is in office… I prefer to have a reduction in human misery if possible.

End of Stupid Moralizing

So… now that I’ve gotten all self-righteous and morally irritating, let’s talk about the numbers. (If you get bored by this discussion, feel free to skip to “The Point” at the bottom)

The unemployment rate is… well, it’s exactly what it says it is: a rate, a percentage based on two numbers. Your average non-economic American might think that the two numbers are as simple as “people employed vs. people unemployed”. Under this definition, you might think that a lower unemployment rate means that there are more jobs.

Sadly, you would be wrong.

The numbers actually start with the US population*. From that number, we take out children under 16, prisoners, those in mental institutions, those who require nursing care and the military and we get the “civilian non-institutional population”. From that number, we take out those who, for whatever reason have not tried to find work for 4 weeks. This is important because you don’t want to count housewives and high school seniors in the unemployment numbers. Remember that, because it’s going to be important in a second.

That brings us to the “civilian labor force”, which consists of the employed and the unemployed. It is from the civilian labor force that we calculate the unemployment rate. Therefore, there is a good way and a bad way to reduce the unemployment rate.

  1. Increase the employment number (good)
  2. Decrease the number of people in the labor force (bad)

The reason decreasing the number of people in the labor force is bad is that it means that people are extracting themselves from the labor force by:

  1. getting arrested in alarmingly huge numbers (unlikely)
  2. joining the army in alarmingly huge numbers (unlikely)
  3. getting younger in alarmingly huge numbers (that would be awesome)
  4. deciding that they’re just not going to look for a job anymore

And among the people in section 4, there are several options:

  1. deciding to stay home due to a lifestyle change (staying home with the kids)
  2. going to school to train for a new job
  3. retirement
  4. despair

So, let’s cut to the chase. We’re not seeing new jobs. The employment number in July continued to decline (though at a much slower rate than it did in June). What we saw instead was a decrease in the labor force. More and more people are just not looking for jobs anymore.

The Point

On the surface the unemployment rate going down seems good, but when you dig into the numbers, we can see that it has nothing to do with an increase in the number of jobs and everything to do with the fact that the labor force is shrinking.

Is this good or bad? I tend to think bad, but the economy also tends to be really complex, so I could be misreading something or I could be just plain old ignorant. I’m not an economist, so I won’t make a pronouncement on that issue. All I can do is show the numbers and wonder what the hell is going on.


@D_B_Inman on Twitter pointed out that I was looking at the unadjusted numbers in my analysis and that the unemployment rate is based on the adjusted numbers. When taking that into account, my charts and extra analysis are strikingly ignorant. This is actually comforting, because it means things aren’t as out-of-whack as I thought they were. I’ve adjusted my “Point” accordingly.

* The Bureau of Labor Statistics lays this all out in more detail, if you want to check it out for yourself.

** There is something a little weird in this because the historical data at BLS doesn’t match up with their current press releases. According to their historical data, we saw an increase in the labor force in the last couple months. But according to their historical data, the current unemployment rate is 9.7%, which is not the number being currently reported. I took the numbers from their current press release and I substituted them into the historical data, since I’m assuming that their current press release is more accurate. If you think I’m wrong, please let me know why.

President Obama, I Fixed Your Graph For You

I’ve been pretty quiet recently because 1) I’m on vacation and 2) I’m trying to wrap my head around the health care issue before I talk about it at length.

But today I saw something on that bothered me:
Combined PPO

Here’s the thing, Mr. President. There is such a thing as visual lying. That is when you show a graph and you show the numbers but the two things are not in any way related to one another.

That is the problem here. If someone looks at this graph, they see that the sky is falling because the bars have increased so dramatically. On the left, your team has represented a 30% increase with a graphic that shows a 966% increase. On the right, your team has represented a 63% increase with a graphic that shows a 308% increase.

And are the two sets bars related in any way? You might think so, given that they show up next to each other and are supposed to measure the same thing. But from a data perspective, they are not even remotely close to being right.

It is possible to use graphs and numbers in such a way that is honest. That’s an important part of transparency. So, I fixed your graph for you.

You’re welcome.

UPDATE: In the comments section, James quickly identified the problem… the graph starts the y-axis at 1000 instead of 0. I double checked and it looks like he is spot on. Thanks!

With that in mind, the graph is more of a rookie mistake than a conscious attempt to deceive. I’ve edited my post to reflect that (I left my original comments in so everyone can see what a smart-ass I tend to be).

The Obama Stimulus: Predictions vs. Reality


In this video, I take a look at the economic predictions that President Obama made in February regarding the stimulus plan and how those predictions are corresponding to reality.

The answer is: Not well.

But first, some references.

OK… now into the math. The chart that everyone is using does not have a corresponding table with hard number (at least no table that I could find), so I had to guess-timate what they were predicting the unemployment rates would be in May. I assumed that, because their graph divergence began immediately after the Q1, 2009 line, that that line represented the beginning of Q1 2009 (as opposed to the middle). So I estimated that the May would be just a shade before Q3, which is about the same place that Geoff put his May data.

Based on that, I estimated the points on the line like so:

Unemployment Rate Unemployed Population
Predicted Unemployment without the Stimulus 8.7% 13,492,000
Predicted Unemployment with the Stimulus 7.9% 12,251,000
Actual Unemployment with the Stimulus 9.4% 14,511,000

Now… here is the problem. In order to make our data symmetrical, we would have to have another row… a row called “Actual Unemployment without the Stimulus”. This, of course, is a row we cannot have because we sadly live in a space-time of collapsed quantum possibilities. We can never know what that row would hold.

This is where I start getting a little less analytical and a little more irritated. The president’s predictions have been shown to be completely off the mark… almost laughably so. And yet he acts as if he alone knows what would have happened if we hadn’t passed the stimulus because he keeps making statements like “we’ve saved 150,000 jobs“.

It is clear that, if he is referring to the chart we were presented with above, such a claim is absurd. What the president is doing is ignoring the fact that his predictions in the past were horribly inaccurate and simply moving ahead with new predictions. The big difference is that his new predictions can’t be judged against any set of objective reality. He is pitting the actual universe in which the stimulus bill passed against the imaginary universe in which it did not pass. Not surprisingly, the imaginary universe is worse that the real universe and the result is that the President is a hero for saving us from that imaginary universe.

I am not a very anti-Obama person. Predicting the future is tricky business and I think his team should get some leeway on this.


Their predictions were not just kinda wrong. They were horrifically, disasterously wrong. If President Obama is going to use statistics and charts to push nearly $800 billion in spending, I think we should be able to expect his numbers to at least kinda match the reality that comes out of his policies.

At the very least, I’d like to know how his team got those numbers. More importantly, I’d like to know how they have changed their method of prediction. President Obama is fond of saying that we tried tax cuts and they didn’t work, so we should try something else. In that same vein, his team tried predicting the effect of the stimulus and that didn’t work. So I would like to know if they are using the same failed methods they used before or if they are doing something different.

A Little Recession Perspective

Hat tip from Greg Mankiw for the pointer to Donald Marron’s post on keeping the current economic downturn in perspective.

Marron states:

The economy would have to fall much, much further than economic forecasters expect for the losses to come anywhere near those experienced in the Great Depression.

And then we get to see this graph, which is a fantastic way of trying to put things in perspective.


I try to keep this graph in my head whenever people talk about “the worst economy since the Great Depression” and especially when they talk about how crisis requires the government to spend enormous amounts of money.

The difference between this economy and the economy of the Great Depression is nearly the difference between a day trip from New York City to Washington DC and a cross country trip from New York City to LA.

AP Apparently Dislikes Accurately Representing Abortion Violence

I love the way Freddie introduced a post on abortion late last year. He titled it “you know what we don’t talk about enough? Abortion“. 

I kind of feel the same way about the level of discussion going on with it. I probably would never have mentioned it at all on this blog if it hadn’t been for the incident on Sunday in which a man shot and killed Dr. George Tiller “one of the nation’s few providers of late-term abortions”.

In the fourth paragraph of the AP article, I came across this line:

“But the doctor’s violent death was the latest in a string of shootings and bombings over two decades directed against abortion clinics doctors and staff.” 

After reading that, I decided to look into the statistics of abortion violence with a view toward perhaps creating a visualization about it.

Sadly, there are few things more skewed than abortion violence statistics. I found this pdf on “Abortion Violence and Disruption Statistics” done by the National Abortion Federation and it is mainly propaganda dressed in numbers. But it looks like their numbers on shootings and bombings are verified by legal authorities, so I assume they are pretty accurate. 

Let’s use those statistics to deal with the “string of shootings and bombings over two decades” that the AP talks about. (In order to give the AP the benefit of the doubt, let’s assume that all the “Attemped Murders” of abortion clinic staff involved shooting of some kind. )

According to the NAF document above, this is that the “string of shootings and bombings” looked like over the last 15 years:


Did you know that this is the first abortion related murder since 1998?

I didn’t.

I was under the impression from the AP that abortion killings were like school shootings… the kind of thing that we tragically see on an ongoing basis. (I thought about a graph comparing school violence to abortion violence, but it seemed kind of apples-to-organges to compare sociopathic, psychotic and suicidal teenagers to politically motivated terrorists.) 

Given the actual data, the characterization of this incident as “the latest in a string of shootings and bombings” is deeply dishonest. It embeds into people’s minds the idea that this is a very common tragedy, like school shootings, hurricanes or gang-related violence. In fact, until I looked at the data very recently, I was under exactly that impression. 

It would be much more accurate to say something along the lines of:

This incident has shattered an eight year lull in anti-abortion related shootings, an activity that spiked to record levels in the 90’s.

UPDATE: Upon re-reading my post I realized that it sounded very dry and unfeeling… very matter-of-fact… when I talked about the recent murder. I hope no one got the impression that I’m wholly unphased by this crime. Nothing could be further from the truth. I hope that the fact that referred to crimes of violence against abortion clinics and the staff as acts of terrorism would indicate how I feel about the topic.

It’s Tough Making Predictions…

This graph has been going around a good deal in the last week. (Source)


Basically, the light blue line is the unemployement rate the Obama administration predicted would happen if we didn’t pass the stimulus bill back in . The dark blue line is the unemployment rate the Obama administration predicted would happen if we did pass the stimulus bill. (Here’s the raw document.) And the red triangles are the actual unemployment rate as it has panned out. Not only are they worse than the Obama adminstration expected, they’re worse than what they expected even if we didn’t pass the stimulus bill.

I think it is fair to say that the stimulus bill has not been as stimulating as they told us it would be.  Now, it could certainly be the case that the unemployment rate would be even higher than this if we hadn’t passed the stimulus bill, but that is about as non-falsifiable a statement as you can get. 

(UPDATE: The author of this graph explains why he thinks there has been little effect … we’ve spent almost none of the stimulus money yet. I’m trying to figure out where he’s getting his data because I don’t see any infrastructure projects on there. I’m certain that there is infrastructure spending going on right now because there is a stimulus project not 3 miles from my house causing daily traffic jams.

UPDATE 2: Here’s the best I could find on stimulus money currently being spent.)

 I don’t really feel like dogpiling on the adminstration on this particular issue, so I want to hit a broader topic here… the administration’s use of numbers. This graph tells us some simple things that are scary and a complex thing that is scarier. 

The simple thing it tells us is that the Obama administration was completely unable to predict the economic conditions four months into the future. They thought we would be at about 8.0% unemployment if the stimulus bill passed and at 8.5% unemployment if we sat on our hands.

As it turns out, we passed the stimulus bill and we’re at 8.9%. The easy lesson is that they didn’t get that one right. But, as Robert Strom Petersen said, “It’s tough making predictions, especially about the future.” And I probably couldn’t have done any better.

But no one is hanging the weight of hundreds of billions of dollars around my neck, which makes it more OK that I can’t project the future economic conditions. It seems fair to demand a slightly higher level of predictive accuracy from an administration that is using their predictions to push trillion dollar policies. 

The complex thing that this graph tells us is that the Obama administration is comfortable using graphs that don’t really have a basis in reality in order to bolster support for  their decisions. Graphs make us think that something is scientific and studied and therefore more reliable. But reliability is something that has to be earned. The team that put this graph together should be questioned on what they got wrong and what they would do next time to get it right.

Basically, the next time the president uses projected figures to push his policies, I would like to see someone ask the following question:

“Mr President, the last number predictions you threw at us turned out to be pretty far off the mark. What assurances do we have that these new numbers are accurate?”