Don't confuse accounting with causal relationships
By Scott Sumner
Pierre Lemieux recently did a post explaining how people often misinterpret the GDP expenditure equation:
GDP = C + I + G + (X-M)
Many people, including even top trade officials in the Trump administration, wrongly assume that this equation demonstrates that imports subtract from GDP. That’s false, as the imported goods get added to C or I.
A commenter named Ahmed Fares makes a slightly more sophisticated argument:
If an American buys an American toaster for $40, US consumption, and by extension GDP, rises by $40. If an American buys a Chinese toaster for $40, US consumption rises by $40, which is then netted out of US GDP, leaving US GDP unchanged.
So that would mean imports have an indirect impact on GDP ($40 in this example).
Now I’m not forgetting that US imports put US dollars in the hands of foreigners, which can lead to US exports or a surplus on the US capital account, etc., but the first-order effects do seem to indicate that imports do have an indirect impact on GDP, which is what really matters.
In other words, the problem is not that imports caused US GDP to fall, when GDP actually remained the same, but rather that they didn’t allow US GDP to rise.
This common fallacy requires a bit more explanation. Let’s start with the assumption that the decision to consume a $40 toaster causes US GDP to rise by $40. Is that true? It may be, but it’s not a conclusion that in any way falls out of the accounting relationship. Recall that the national income accounting also implies that saving equal investment, at least ex post. So the decision to buy the $40 toaster also entails a decision to save $40 less than if one had not bought the toaster. If nothing else changes, then that reduces investment by $40.
Again, we are not yet considering macroeconomic factors such as the paradox of thrift or the relationship between saving and aggregate demand. We are still merely looking at accounting identities. And once again, the national income accounts show that, ex post, S = I.
One obvious counterargument is that the decision to buy a $40 imported toaster involves two negatives: less saving (implying less investment) and also the negative that is associated with imports in the national expenditure equation. There seems to be only one positive—a $40 rise in consumption. Does that save his argument?
Not quite, because we still have not worked through all of the accounting. There are actually two possibilities:
A. The foreign country takes the $40 spent on the imported toaster and buys US exports.
B. The foreign country runs a $40 larger current account surplus with the US.
In case A, there are two negatives and two positives. The first negative is that the decision to buy the imported toaster (instead of saving the money) reduces investment by $40. The second negative is that the imported toaster shows up as negative $40 in the trade balance. The two positives are the extra $40 in consumption and the extra $40 in exports.
In case B, there is one net negative and one net positive. The foreign country’s decision to run a trade surplus with the US causes US investment to be $40 higher than otherwise (for any given level of US saving.) This means that the normal decline in US investment that would occur when Americans reduce their saving does not occur, as foreign savings pour in to fill the gap. So the only negative is the $40 in imports, which shows up with a negative sign in the trade account. And of course the only positive is the $40 boost to consumption.
No matter what assumptions you make, the accounting relationships do not demonstrate any negative impact for imports.
What about if we look for broader macroeconomic causal factors, such as the paradox of thrift. Recall that this Keynesian idea suggests that an attempt to save more may end up simply reducing aggregate demand, thereby leading to no actual increase in realized saving or investment. Does that change the result? This turns out to be a very complex question:
1. If the Fed is targeting the money supply, then an attempt to save more may reduce V, and hence M*V (aggregated demand.) If the Fed is targeting interest rates, then when people try to save more the Fed will have to reduce the money supply in order to keep interest rates stable. Again, M*V falls.
More likely, the Fed is targeting something like inflation or NGDP. In that case, monetary policy offsets the impact of an increased propensity to save, leaving aggregate demand unchanged. So in practice, the issue that commenter Fares mentions is not much of a problem.
2. Now let’s assume that the Fed is not targeting inflation. Let’s say interest rates are stuck at zero and the Fed is allowing inflation to run below target, due to incompetence. Now it is at least possible that the decision to buy the foreign toaster will reduce aggregate demand. But even here the problem is not really trade, it’s saving.
Go back to the toaster example. There are two cases where AD might fall, and two where it would not:
Case A: If the seller of the toaster saves the $40, this tends to reduce AD. Importantly, this is true regardless of whether or not the toaster seller is domestic or foreign.
Case B: If the seller of the toaster doesn’t save the $40, then AD will not be reduced. A US seller would spend the money on some other goods, and the foreign seller of toasters would spend the money on US exports. (Here I’m looking at the aggregate behavior of the exporting country, not the decision of the individual seller.)
So any AD “problem” comes from too much saving, not the decision to buy imports rather than to buy local.
Bottom line, in order for the decision to buy the toaster to reduce US aggregate demand, you need both of the following to be true:
1. The foreign seller needs to be more likely to save the money than the domestic seller.
2. The central bank must respond passively, not targeting inflation or NGDP.
Even in the unlikely event that both conditions are met (and they are not currently met) you’ll still merely have impacted aggregate demand, not real GDP. To make the additional argument that this action would impact RGDP, you need a sticky wage/price AS/AD model. In that model, the short run AS curve is upward-sloping, and the long run curve is vertical. In that case, a trade deficit might possibly have a short run negative impact on GDP.
But the Trump administration officials making this argument are making a long run argument. They argue that the US has a secular problem of inadequate RGDP caused by trade deficits. This is one of those arguments that is false from almost every perspective imaginable:
1. The assumption about monetary policy is currently false. The Fed targets inflation at 2%.
2. AD has no long run effect on RGDP.
3. The empirical evidence also does not support the claim. The US and Australia have been running huge current account deficits, while the Eurozone and Japan have the world’s largest surpluses. And yet growth in US and Australian aggregate demand has been far more robust than in the Eurozone, and Japan has seen perhaps the most anemic growth in AD in the entire history of the modern world. There has been almost no increase in Japanese NGDP over the past 25 years.
Thus there is not one but two flaws in the theoretical underpinnings of the mercantilist argument, and on top of that the empirical evidence is completely inconsistent with their claims.
PS. One other question. President Trump says the condition of the US economy is now “great”. Things have never been better. If that’s true, then what is the purpose of launching a trade war right now? What problem is the trade war supposed to fix, even if we “win”?