Often times, non-economists attribute much more confusion to the ideas of Keynesian economics than is really the case. Take for example the idea of smoothing out business cycles. Anyone who's ever tried to keep to an exercise regime understands the concept intuitively. The fancy graph way of making the case looks like this:

The idea here is simple. The black line is the boom and bust cycle that characterizes un-managed markets. The red line is what happens when the government or private actors step in to manage the economy to smooth out these business cycles. Note that while at any one point, the black line may show a much higher rate of growth than the red over the long run, it ends higher.

Returning to the exercise metaphor. While you can can have an exercise regime that involves exercise binges followed by food binges, in the end slow and steady is going to when the day. Because each time you switch from pigging out to being an exercise freak there's a lot more effort involved than if you had just stuck to the program. The preference is for a stable program, and that's what any economic stimulus package should aim for too. But, that's not what's happening.

Where's the money going?

At the end of last week, USA Today ran a piece about how the $ 4 billion of stimulus money currently disbursed has been spent.

There's a cool flash graphic if you follow the link. I've recreated their graphic with a little clearer color and coding scheme here. Note the scale.

To put some actual numbers to this, we can look at the top 10 and bottom 10 states in terms of stimulus money per cap. And it's pretty amazing.

That's right. Idaho got $245.63 per capita in stimulus funds even though unemployment in that state's unemployment rate is 1.6% below the national rate.

Compare this to the long suffering state of Michigan, which received the princely sum of $0.21 per capita in stimulus funds, while the unemployment rate is 4% above the national rate. What the hell is happening here?

What kind of stimulus is this?

Let's return for a moment to that graph I used earlier about business cycles. This time, I've shown how those ups and downs are smoothed out. The arrows show the general idea.

Think of this is terms of employment. When employment drops, the government should step in to spend money to keep employment steady. The idea is to create confidence. Half the problem of the current recession isn't the people who are out of work. It's the people still on the job, but who've stopped spending cash because they are afraid of losing their job. By working to keep employment steady, the government limits the contagion from falling sectors to less affected sectors.

In this recession, it's construction and manufacturing that are losing jobs and these losses have been concentrated on the West Coast (Construction), the Midwest (Manufacturing), and the South (both).

The beautiful thing is that by pumping money into these sectors and areas early, the contagion effect can be limited.

For example, keeping an auto worker on the job in Michigan means that workers at suppliers in Kentucky stay on the job too. As do the waitstaff at the diner that Kentucky worker goes to. And the suppliers that diner in Kentucky orders may keep employees and a food processing plant in Georgia on the job.

Put the worker auto worker in Michigan out of work, and it's not only him or her that's going to lose, it's going to be everyone that money touches.

Drop a little cash to keep that one auto worker on the job, and you keep a whole other group of people who are linked to that auto worker on the job too.

The impact of the money spent is many, many times $1 to $1. $1 spent in manufacturing may allow workers elsewhere in the economy to take home $20-$30. But, stimulus spending has to be targeted at falling employment to make that happen. This has not happened.

No Stimulus at All

One of the statistical tools available to people looking at information of this sort is linear regression. It allows us to figure how much a given increase in one thing will lead to an increase (or decrease) in another thing.

For example, if the stimulus is being targeted towards keeping employment steady, then presumably this means that in as a rule where there has been a drop in unemployment there will be more stimulus money spent. The greater the drop in employment, the bigger the amount of stimulus cash.

I've created a database that includes these numbers, and when I run a regression to look for a relationship between, there is little or none. R-squared is the measure we use to look for relationships between things. It runs from 0 to 1, and the higher the number the stronger the relationship.

When I ran a regression using the employment drop as the driver behind stimulus spending, my adjusted R-squared was -0.018.

Which is actually pretty cool, because I've never gotten a negative R-squared before.

It means there is literally less than no relationship between the two.

In other words, stimulus spending isn't being targeted to states where employment drops have occurred. It's worse than random. It's as though someone were (ever so slightly) targeting stimulus funds to those states that need it the least.

Seriously, I've seen fucked up statistics, but this one truly takes the cake. It's like getting heads up in a coin toss 25 times in a row. It's bizarre.

How would this be done right?

It's still early in the game. So there's time to fix this situation. Of the more than $700 billion included in the stimulus package, only $3.941 billion has been released. On other words less than 1% of the total.

From the start of the recession till the end of March, US employment had fallen by 4,926,100. Which means that if the money had been distributed on an equal basis per job lost, that for support for each job loss would be $800.

This is not how things look right now.

How do things look now?

Below, I've produced a graph which shows the amount of stimulus money disbursed per job lost in each state.

The shocking lack of direction in the disbursement of stimulus funds continues.

4 states (Alaska, Wyoming, North Dakota, Louisiana) actually have gained jobs since the start of the recession and managed to still take in stimulus funds.

Among the other 46 states that have lost jobs, the amount disbursed per job loss varies greatly from a high of $16,131.50 in Washington to 3 states that while losing jobs having received no stimulus funds.

I've created a graph below ranking the states for stimulus funds received per job loss. I've excluded states that gained jobs and those that have received no stimulus funds. The list is pretty amazing. Remember that if the stimulus funds were disbursed according to job losses, that each job loss would lead to a state getting $800 in stimulus money. The column on the right shows the actual amount of stimulus per job loss received by a state as a percentage of its "fair share."

Conclusion

Like I said above, less than 1% of the stimulus funds have been disbursed. So there is time for change. It may be that this first set of disbursements is a fluke. I hope so, but I'm not the trusting type.

It's more than simple fairness that should concern us here.

Remember that the reason that stimulus packages are a useful policy tool is because they allow us to smooth out business cycles.

The problem here is that the money spent here isn't accomplishing that.

Which means not only that those currently out of work are likely to remain so, but also that many more are likely to join them.

Call me a cynic, but I see a distinct anti-manufacturing bias in the way that these funds have been distributed. Those states that have not gotten less than 10% of their fair share are overwhelmingly dependent upon manufacturing.

More specifically, of the six auto states (Michigan, Ohio, Indiana, Kentucky, Tennessee, and Mississippi) only one (Tennessee) has received it's "fair share."

Beyond that, Ohio leads the other 5 with 11% of it's "fair share" of stimulus funds.

The distribution of stimulus funds is never going to be entirely equal across states, but it should be more equal.

Just as there's something distasteful about a CEO making 500 times a line workers pay, it doesn't seem right that Washington and Idaho are getting over $10,000 per job loss while 18 states are getting less than $100 per job loss.

In order to work, stimulus spending must work to stabilize the employment situation. This isn't happening now.

In order for it to occur, there needs to be more attention to the way stimulus money is being spent. Only then can we know if the effort is going to be effective.