GDP is a funny number. The simple version is that a government statistics office takes every transaction in the economy, adds up the amounts, and spits out a total. In practice, it’s a bit more complicated.

Suppose Peter buys an apple for \$2. This transaction makes Peter happy – if he paid \$2 for the apple, he probably got at least \$2 of value from it. \$2 is added to GDP, and we can mostly assume higher GDP is better: a higher GDP implies more Peters are buying more apples.

The thing we care about, the thing we want to optimize for, is the apple, not the \$2. The \$2 is just two bits of paper, or a few numbers on a computer. You can’t eat paper; the end goal is sitting down and having a meal, not paper moving around. The paper is just an accounting device. We add up dollars because it’s simpler than adding up apples; apples are of different shapes, different types, different quality, and are of course only one good out of millions, while a dollar is the same as every other dollar (fungibility). The money is a little tag we attach to the apple to keep track of it, like tagging an elephant so you can follow it through the African jungle.

Now, suppose Peter likes his friend Paul, and gives him \$2 for lunch. This is a transaction, but it is not counted as part of GDP. There’s no apple; no goods or services are produced; money has just gone from one pocket into another. The economic term for these transactions is “transfer payments”. On a broader scale, if the government pays \$1 billion for a new road, this is part of GDP, because a real good or service is produced (the road construction). But if the government writes people \$1 billion in Social Security checks, this isn’t counted. It’s just moving money around from A to B; nothing is produced.

So far, all is well. However, it turns out many government programs are de facto transfer payments, even if they have some other nominal purpose. As a real example, suppose the government spends \$400 million on new tanks. The tanks are designed, built, tested, and shipped off to the Army, where they sit around in warehouses because the Army didn’t need them. Clearly, in such a case, the government isn’t really exchanging money for goods, since the ‘goods’ are useless.

If you ask a (relatively) honest politician why this happens, they’ll say it’s to create jobs. But people (mostly) don’t want jobs for the sake of having jobs. They want jobs for the money. Hence, what the government is really doing is transferring money to the politically influential, under the nominal guise of purchasing an (actually worthless) product. It’s as if Peter felt sorry for Paul, bought Paul’s plastic apple for \$2, and then promptly threw the fake apple away. The real goal is moving money, not the exchange of value. Call these transactions “quasi transfer payments”.

Quasi transfer payments artificially inflate GDP. Suppose that, in 2010, the government spends \$1 trillion on roads, \$1 trillion on tanks the Army doesn’t want, and \$1 trillion on Social Security. In official statistics, the government share of GDP will be counted as \$2 trillion – \$3 trillion total spending, minus \$1 trillion in transfer payments. But a fair accounting would also subtract the \$1 trillion in quasi transfer payments. Hence, the real number is \$1 trillion, not the official \$2 trillion.

This isn’t quite as bad as it sounds. What people are mostly concerned about with GDP isn’t the absolute number of dollars exchanged (the dollar is an arbitrary unit anyway), but how good things are relative to other times and places. If GDP is 5% higher this year, that’s good; if GDP is lower than some other country, that’s bad. If, say, 20% of the economy is quasi transfer payments, this doesn’t affect GDP growth numbers, because the same extra 20% is added to both this year and last year. An economy that grew at 2% will still grow at 2%.

However, quasi transfer payments do introduce some important biases:

– They cause overestimation of the benefits of government spending. Suppose the economy is \$10 trillion, 40% government, 60% private. Half of the government’s spending (\$2 trillion) is quasi transfer payments, so the ‘real’ economy is only \$8 trillion. Now, suppose all government spending goes up by 20%. The new GDP figure is still \$10 trillion – \$5 trillion government, \$5 trillion private – so it’s like nothing changed. However, quasi transfer payments are now \$2.5 trillion, not \$2 trillion, and so the ‘real’ economy is only \$7.5 trillion; things actually got worse.

– They motivate politicians to spend on useless projects. Consider a road from New York to Chicago, to be built next year for \$1 billion. This will add \$1 billion to the GDP. Now, suppose a senator from Nowhere, Ohio wants to build the road from New York to Nowhere instead. On paper, this will make no difference, it’s still \$1 billion added to GDP. However, in reality, the road is now a quasi transfer payment. If properly subtracted from GDP numbers, we’d discover that it makes the country \$1 billion poorer, compared to the original road to Chicago.