The GDP Adjustment

A week-plus ago I posted some data compiled by Bianco Research analyst James Bianco indicating that our $8.4 trillion-and-counting bailout dwarfs just about every huge government project you can think of, combined. Reader domoarrigato argued that re-casting those numbers as a percentage of their contemporaneous GDP might be more illuminating, and then he went ahead and did the math himself. After sending the informatics to our top experts, we are now prepared to publish them in a hopefully easy-on-the-eyes chart. Here goes:

Speaking of bailout graphics, try this more colorful one from Pro Publica–a bubble-chart of post-1970 bailouts, adjusted for 2008 dollars (though not as a percentage of GDP!).

A tangential topic for discussion: What do you all think of the argument that percent-of-GDP is the real way one should ponder stuff like the size of various government programs, or defense spending, or the whole state apparatus itself? I've always thought it extremely helpful for personal context (thanks again, domoarrigato!), and extremely dangerous for the purposes of deciding how to spend.

Here's why: The biggest chunk of GDP, and certainly the most dynamic, is the stuff produced by the private sector. Pegging a public-sector program to a private-sector number basically rewards inefficient non-innovators with the innovators' gains. Put another way, if it cost 4 shekels a year to adequately defend a country with a 100-shekel economy (let's say that 77 of those 100 shekels were produced by the private sector), why on earth should we increase the defense budget to 8 shekels when (as inevitably happens) the profit-seeking privates double their money? I understand that labor and materials can become more expensive in a growing economy, thus adding costs to guvmint operations, but essentially this is about making the booming size of government look rational, just because the private sector is (or was) booming as well. Where am I wrong here?

Comments

bubba | December 23, 2008, 11:56am | #

The GDP correction handles a number of factors, one of which is population.

It's also relevant to your example of defense, because you're not just defending a territory, you're defending an economy, and your enemies are becoming more capable as well.

And your soldiers are becoming more valuable.

So, you do want to spend more on gadgetry, etc, because it's more important to defend that factory, and it's more important to keep that solider alive.

Yes, it's possible that GDP allows for scope/budget creep, but a simple per capita adjustment doesn't account for the increasing value of the things we're defending.

Another consideration is that as our standard of living goes up, we might rationally choose a *larger* government as a percentage of GDP, to the extent that it supplies safety, security and comfort. After all, the after-tax income is growing as well, so we can have our cake and eat it to.

At least, that's how we're voting.

Warty | December 23, 2008, 12:13pm | #

Expressing the numbers as a percentage of the average family's income might be helpful, as well.

robc | December 23, 2008, 12:20pm | #

Warty,

Which average: mean, median or mode?

Nigel Watt | December 23, 2008, 12:25pm | #

Why is NASA pegged to 1984?

Warty | December 23, 2008, 12:41pm | #

robc: I don't mean to be mean, but you should be aware that the meaning of average is mean.

TrickyVic | December 23, 2008, 12:41pm | #

What's the benefit of comparing money you spend to money that not yours to spend? Best I can tell it makes government spending look small.

Episiarch | December 23, 2008, 12:47pm | #

Good job, robc, you just put Warty into his mean dick mode.

Warty | December 23, 2008, 12:56pm | #

Wait, does the mode even make any sense when talking about incomes? I haughtily assert not.

Alan | December 23, 2008, 1:00pm | #

Failure on the "easy-on-the-eyes" part - bad scaling and pixelating make it hard to read. Does anyone else have this problem? Possible to re-post in a different format?

robert capozzi | December 23, 2008, 1:04pm | #

percentage of GDP works well. It normalizes for a variety of abherrations, and can be plotted over time.

Team America | December 23, 2008, 1:46pm | #

@Warty, robc and everybody else who hasn't about the central-limit theorem.

Income-distribution is asymptotically log-normal.

Therefore, mean, mode and median depend on two parameters, m and s.

mean=e^(m+s^2/2)
median=e^m
mode=e^(m-s^2)

It follows that mean>median>mode and the logarithms are linear in m and s^2.

Now, libertarians know that redistribution (goal: make incomes more equal by making s^2 smaller) but is also costly (effect: makes m smaller) and is therefore really stupid (we will be equal as soon as everybody has nothing).

So, libertarians suspect that m/s^2 is in fact constant.

From which follows that the ratios of means, modes, and medians are identical.

So, would the innumerate please stop fighting?

Cab | December 23, 2008, 1:58pm | #

why on earth should we increase the defense budget to 8 shekels when (as inevitably happens) the profit-seeking privates double their money?

Would you spend the same amount on a safe if you were going to put $1,000,000 in it as opposed to $500,000?

The more you have to protect, the more you spend on protection. Kinda what bubba said.

Chisailor | December 23, 2008, 2:02pm | #

You are wrong in the sense that, in some area, spending required is dependent on population and/or GDP. National defence isn't a good example of this - the national borders are fairly static, though with greater overseas ownership by americans who support (bribe) politicians, there is some scope creep here. A better example would be the cost of social security - the output required to maintain the same benefit per person is going to increase in some reasonably proportionate ratio to the number or people retiring and/or getting old and injured. This would be something that a GDP percentage ratio would be appropriate for. As a counter-example, I would not expect to see a proportionate increase per GDP in spending on embassy costs for instance - there may be some increase, but is should not track closely with GDP.

Invsible Finger | December 23, 2008, 2:38pm | #

The GDP is a bullshit number, lotsa double and triple counting of middleman transactions that account for the entire transaction instead of only the value-added part.

Invisible Finger | December 23, 2008, 2:40pm | #

And meanwhile the tomatoes I grow in my garden and eat myself don't count in GDP at all.

phlogiston | December 23, 2008, 3:01pm | #

HTML table please? That image is very hard to read, plus you can't copy-n-paste the numbers.

The Angry Optimist | December 23, 2008, 3:03pm | #

It's also relevant to your example of defense, because you're not just defending a territory, you're defending an economy, and your enemies are becoming more capable as well.

Ah ha. Thanks, bubba!

Some backstory: I remember arguing with my economics professors back in the day that national defense no longer makes sense as a quintessential example of a "public good", because we would defend say, a private tanker that was attacked by another nation in somewhere far from United States borders, like the South Indian Ocean. Because there is an additional cost-per-defendee in terms of overseas residents and merchants.

Another example: if tomorrow, some nation overran our embassy and some private American enclave and slaughtered Americans, you can bet that our "national defense" mechanism would be brought to full force, even though we would not technically be defending the nation.

blaupanzer | December 23, 2008, 3:18pm | #

Defense spending should be based upon threat, which goes way up when we deploy into overseas combat. The best argument to forward a proper understanding is: If the GDP goes down, should we lower defense spending accordingly? Based on a threat-independent percentage of GDP, woe is us if the war tools we paid for have to be used. "We can't be at war; we can't afford a war."

squarooticus | December 23, 2008, 3:40pm | #

The GDP is a bullshit number, lotsa double and triple counting of middleman transactions that account for the entire transaction instead of only the value-added part.

This is the important insight. Basing an opinion of whether the government is spending too much on a number the government itself pulls out of its ass shows very little insight.

Team America | December 23, 2008, 3:47pm | #

@Matt Welch

Defense spending is a problem of game theory, because if no country spent money on weapons, there would be no need for a defense budget at all, which is the pareto-optimal strategy.

So, why is there a defense budget, at all?

The first reason is that if no other country has weapons, it is rational for your country to invest in weapons and undertake raids.

So, the above optimum is a strategic saddle-point for all involved players (i.e. it is unstable, because you'll 'slide' off).

Therefore your assumption that the cost of defending a country is constant is not applicable.

But there is a second reason: countries themselves are made up of entities with opposing strategies.

In the same way a physician doesn't want to cure his patients, but doesn't want to kill them either (who'd pay him?), a rational soldier loves enemies, but doesn't want to fight.

So, in is natural for any soldier to be in collusion with the enemy soldier: no peace, but no fighting, either!

Now, one way for two rational countries to get rid of their parasitic military, is to start a war, and to make sure that no side achieves an advantage.

The reason why this isn't happening, is that countries (i.e. tribes of primates) are simply not rational.

In fact, a significant majority of humans believes that winning a war is an objective and that casualties are to be avoided.

:-)

robc | December 23, 2008, 3:48pm | #

Warty,

robc: I don't mean to be mean, but you should be aware that the meaning of average is mean.

Way too late and I need to do some math on Team America's post, but I just wanted to point out how wrong you are. Average, in mathematics, refers to a number of different measures, including mean, median, and mode. In common usage, mean is generally what is being referred to, but not always. Median is an average. As is mode.

And, no, mode doesnt make any sense wrt income really.

rose | December 23, 2008, 3:54pm | #

To Matt Welch's question -- yes, that seems right. A business would spend on any one input just as long as it produced more revenue than it cost; there's no reason why defense, or roads, or bailouts, should be a constant fraction of GDP. Your food expenses aren't a constant fraction of your income.

I wonder a little bit about the figures, though. It isn't actually known how much the bailout will cost. The 700 billion dollar figure, for example, is a worst-case scenario maximum; in theory, it's supposed to be paid back. I actually saw Alan Blinder say (two months ago) that he thought it was possible the taxpayers might come out ahead in the end. Optimistic, yes, and maybe colored by the fact that he worked in Washington. But it does make me distrust this back-of-the-envelope shocker.

Kolohe | December 23, 2008, 3:54pm | #

GDP is useful, even if the absolute value is methodologically flawed, because you can still make apples to apples comparisons of relative values.

kevrob | December 23, 2008, 4:02pm | #

I absorbed the meme of Govt. Spending as a % of GDP from Milton Friedman. The context was that Unca Miltie warned us not to get het up over the annual deficit, or the debt ceiling, or whatever. A budget that was balanced with spending at 40% of GDP would be worse for freedom than one with spending @ 30% of GDP, with a deficit. [It is best to think of Total Govt. Spending - fed, state and local, even if only the Federal deficit can be monetized.]

The real problem with all this spending is that at some point, the debt will be monetized, and the $ will go in the crapper. Good thing I don't have many of them, huh?

Kevin

Team America | December 23, 2008, 4:06pm | #

@robc
And, no, mode doesnt make any sense wrt income really.


Sure it does. It's the answer to the question:
"What do most people earn?"

For example, at McDonalds the mode will most likely be minimum wage.

Kevin Nowell | December 23, 2008, 5:01pm | #

Does anyone have any idea what the best estimates are for what percent of the bailout will be paid out or what percent loss USGOV will take? Is there anywhere I can bet on that?

jtuf | December 23, 2008, 6:29pm | #

Kolohe | December 23, 2008, 3:54pm | #

GDP is useful, even if the absolute value is methodologically flawed, because you can still make apples to apples comparisons of relative values.

Agreed. Measuring government spending relative to GDP gives a easy to calculate way to view trends. It's not to realistic, so just take it with a grain of salt. Ideally, you would calculate the "reasonable" costs of government by totaling the market value of all the services government should provide. Then you could compare the current budget to this reasonable cost. This method could work well in libertarian circles, because we more or less agree on what the government should be doing. In the mainstream, this method would lead to endless debate over what the role of government is and how much each service should cost.

jtuf | December 23, 2008, 6:32pm | #

In my view, the right size of government depends on what it is doing, not how much it costs. The War on Drugs would still be a waste at $5. On the other hand, if the country is actually under threat, it's worth any price to save it.

bubba | December 24, 2008, 11:47am | #

"What do most people earn?"

The mode doesn't tell you that. It tells you the single most common wage.

If 10 people earn $5/hr, and 1100 people each earn a different wage above $20/hr (20.01, 20.02, 20.03, 20.04 ... 30.99, 31.00) the mode will still be $5/hr.