Follow by Email

Monday, June 18, 2018

Krugman: Thinking About a Trade War (Very Wonkish) [feedly]

Thinking About a Trade War (Very Wonkish)

Until fairly recently, I didn't really think we were going to have a trade war. What I thought would happen, instead, was a bit of kabuki: America's major trading partners would make cosmetic concessions – perhaps with some lucrative payoffs to Trump businesses on the side – that would let Trump proclaim a "win", and trade would go on much as before.

The reason I expected this relatively benign outcome wasn't that Trump would get or take good advice. It was, instead, the expectation that big money would talk: corporations have invested trillions based on the assumption that an open world trading system, permitting value-added chains that sprawl across national borders, was going to be a permanent fixture of the environment. A trade war would disrupt all these investments, stranding a lot of capital, and I thought big business would get either manage to get that message through to Trump or at least get it through to Republicans in Congress, who would act to limit his room for maneuver.

But these political considerations look a lot less compelling now than they did a few months ago. With Gary Cohn gone, it's not clear that big business has any real pipeline into the White House (OK, polluters have an open line to Scott Pruitt, and predatory lenders a line to Mick Mulvaney, but these aren't the groups who will stand up against trade war.) And Congressional Republicans, terrified of the Trump base, have proved unwilling to take a stand on anything, even if big money is at stake.

Meanwhile, trade decisions are being made at Trump's whim, without input from anyone who knows anything about trade economics (Peter Navarro thinks he understands the economics, which is even worse.)


Oh, and Trump's version of diplomacy – not just trade actions, but the systematic praise of brutal dictators and disdain for democratic leaders – has created a very angry world out there. Nobody out there wants to give Trump even the appearance of a win, and elected leaders would be punished by their voters if they did.

So a serious trade war now looks very possible, and it's time to think about what it might mean.

There are, I think, three main questions:

1. How high might tariffs go?

2. How much would this reduce world trade?

3. How costly would the trade war be?

These are all slightly tricky questions, as I'll explain. But there's a pretty good case that an all-out trade war could mean tariffs in the 30-60 percent range; that this would lead to a very large reduction in trade, maybe 70 percent; but that the overall cost to the world economy would be smaller than I think many people imagine, maybe a 2-3% reduction in world GDP.

This last calculation, however, doesn't take account of the disruptive effects of deglobalization: some people would actually gain, but a lot of people, very much including large groups and many communities in the U.S., would take big hits, especially in the short-to-medium run.

How high might tariffs go?

What do we mean by a trade war? In the current context, we mean a situation in which the world's economies, taking their lead from the U.S., abandon the rules and agreements that currently constrain their tariffs and start setting tariffs unilaterally in their perceived self-interest.



The problematic word, of course, is "perceived". On trade, as on other subjects, Trump's perceptions generally don't seem very tethered to reality. And to be fair, the economic policy performance of other major players, notably the EU, hasn't been noteworthy for the triumph of clear economic thinking.

Still, there have been a number of attempts to model trade war over the years, relying on one of two approaches. The first is to imagine that governments actually do maximize national income, or perhaps an objective function that gives extra weight to well-organized interest groups. The second appeals to the historical experience of the world before international trade agreements became the norm. Fortunately, these approaches suggest similar tariff levels.

On the first approach: any country large enough that it can affect world prices of the goods it exports, the goods it imports, or both, has an "optimal tariff" greater than zero. The reason is that by limiting its trade, such a country can improve its terms of trade: the price of its exports relative to the price of its imports. This raises real income, other things equal. And the optimal tariff trades off the costs of reduced trade – e.g., the cost of producing goods domestically when they could be purchased more cheaply abroad – against the gains from improved terms of trade.

The problem is that if everyone does this, you get the costs of reduced trade without the benefit of improved terms of trade, because other countries are doing unto you the same thing you're trying to do unto them. So you end up in a situation of "optimal tariff warfare", which is actually more like an arms race than a shooting war, in the sense that there's (usually) no victor and no resolution, just a lot of wasted resources.

So how do you estimate the effects of optimal tariff warfare? You need a "computable general equilibrium" model of world trade – something that shows how production and trade flows depend on tariff rates, calibrated to match the actual data. Then you have to find an equilibrium (a Nash equilibrium, for readers of "A Beautiful Mind") in which each country is charging its optimal tariff given what everyone else is doing.

There are many assumptions and imputations involved, with the results depending a lot on how easily goods from one country can be substituted for goods from another – a parameter that's hard to estimate. Still, there are several fairly recent efforts: Ossa finds that a trade war would, under his favored assumptions, lead to tariffs of nearly 60 percent, while Nicita et al, using slightly different assumptions, estimate a rise in tariffs of 32 percentage points from current levels.


If this seems too abstract, we can look at history: the Smoot-Hawley tariff, the last great protectionist move by the U.S. before we created the system of trade agreements, pushed tariffs up to around 45 percent on "dutiable" imports (tariffs were so high that most imports involved goods that for whatever reason faced no tariff at all.) You may wonder why I don't say 59 percent, which was the peak reached in 1932; but that was sort of an accident, even higher than protectionists wanted, the result of deflation that pushed up the rate of protection for those goods whose tariffs were specified in dollars per unit rather than percentages.

So both history and quantitative models suggest that a trade war would lead to quite high tariffs, with rates of more than 40 percent quite likely.

How much would trade decline?

For any given tariff rate, the amount of trade reduction depends on the elasticity of import demand – the percentage fall in imports for every one percent rise in their price. Such elasticities are hard to estimate, because we don't get many natural experiments. (Fluctuations in exchange rates change import prices, but those only give us an idea of short-run effects, and everyone believes that long-run elasticities are much bigger.)

As best I can read the literature, consensus estimates for the elasticity of import demand are something like 3 or 4, but there's not much certainty here.

If we really believe in optimal tariff warfare, however, the effects of the war on trade volumes are surprisingly insensitive to the precise value of the elasticity. Why? Because the optimal tariff also depends on the elasticity. If foreigners can easily substitute away from your goods, the optimal tariff is fairly low; if they can't, it's high. So high elasticities mean low tariffs, low elasticities mean high tariffs, and the decline in trade is similar. (See this little write-up.)

My back of the envelope calculations suggest that we might be looking at around a 70 percent fall in trade for a wide range of cases. I'd be happy to be corrected by trade modelers if that's wrong.

But if that's right, we're talking about a really big rollback of world trade. Figure 1 shows world trade (exports plus imports) as a share of world GDP back to 1950; a 70 percent reduction would bring us roughly back to 1950s levels. If Trump is really taking us into a trade war, the global economy is going to get a lot less global.

Figure 1CreditWorld BankHow big are the costs?

There has historically been a lot of hype about the evils of protectionism – Smoot-Hawley caused the Great Depression, and all that. It's also tempting to assume that because the Trumpist argument for trade war is so stupid, Trump trade policies must be totally disastrous.

But I've always ended up being really sorry when I let my political feelings override what my economic analysis says. And simple trade models, while they do say that trade wars are bad, don't say that they're catastrophic.

To do this right, we should use one of those computable general equilibrium models I mentioned above. These suggest substantial but not huge losses – 2 or 3 percent of GDP. What I'd like to do is offer a bit of intuition about why those losses aren't bigger, then explain why a trade war would nonetheless be highly disruptive.

To do this, I'll exploit a dirty little secret of trade theory: while ultimately stories about trade have to be general equilibrium, that is, they must make sure that you've kept track of all markets simultaneously, trade policy analysis using partial equilibrium – ordinary supply and demand – usually gets you more or less the right answer.

So let's think about the demand for imports as if it were an ordinary demand curve, with the costs of a tariff coming in the form of lost consumer surplus (Figure 2). Those who remember their Econ 101 will know that the costs of a market distortion normally take the form of a rough triangle (rough because the demand curve doesn't have to be a straight line, but that's a fairly minor detail.) That's because the first unit of imports lost has approximately zero cost, because people are indifferent at the margin between that unit and a domestic product, but the last unit lost imposes a cost equal to the tariff rate, because that's how much more people would have been willing to pay for the import. And the average cost of reduced imports is halfway between these values.
Figure 2

So when a tariff drives up the price of imports to consumers, leading them to buy fewer imported goods, the welfare loss will be roughly
Loss = fall in imports * ½ tariff rate

Now, the U.S. currently spends 15 percent of GDP on imports. Suppose we end up with a trade-war tariff of 40 percent, and (as I've been suggesting) a 70 percent decline in trade. Then the welfare loss is 20% * 0.7*15, or 2.1% of GDP.

That's not a small number, but it's not that huge either: at the bottom of the Great Recession, CBO estimates that we were operating 6 percent below potential GDP. Of course that loss was temporary, while a trade war might be forever.

But these net welfare costs may miss the real point, which is disruption.


The U.S. currently exports about 12 percent of GDP. Not all of that is domestic value added, because some components are imported. But there's still a lot of the economy, maybe 9 or 10 percent, engaged in production for foreign markets. And if we have the kind of trade war I've been envisaging, something like 70 percent of that part of the economy – say, 9 or 10 million workers – will have to start doing something else. And there would be a multiplier effect on many communities now built around export industries, which would lose service jobs too.

This is just the flip side of the "China shock" story: even if you believe that the rapid growth of Chinese exports didn't cost the U.S. jobs on net, it changed the composition and location of employment, producing a lot of losers along the way. And the "Trump shock" that would come from a trade war would be an order of magnitude bigger.

You can see hints of what might be to come in what's already happening. So far we've had only small skirmishes in what might be the looming trade war, but the effects don't seem trivial to soybean farmers already facing sharp price cuts and steel users already facing much higher costs. If the trade war happens, expect to see many, many more stories like this.

O.K., there's no certainty that any of this will happen. In fact, I still find it hard to believe that we're really going to go down this path. But I also don't have any plausible stories about what's going to make Trump stop, or induce other big players to give in to his demands.

 -- via my feedly newsfeed