There was one post on Saturday, and one yesterday. Saturday’s post was “More on the Short-Run Macroeconomics of Brexit:”
On Thursday I weighed in on the short-term effects of Brexit, questioning the near-universal premise among economists that it will be a major negative shock to demand. I wasn’t trying to be counterintuitive for the sake of sounding clever, let alone trying to defend the Brexiteers. I was just genuinely puzzled about where this consensus came from, given that nothing in standard macroeconomic models says that a policy bad for the economy’s long-run supply side necessarily hurts its short-run demand side. And I worried that the apparent consensus among economists was in some sense political rather than analytical: free trade good, breaking from Europe bad, therefore you don’t have to be careful in your arguments about exactly why it’s bad.
I’ve received several thoughtful responses from economists I respect, all making a particular argument about the effects of Brexit-induced uncertainty. It goes like this: right now, firms don’t know how closely Britain will be tied to Europe, so it makes sense for them to postpone investments until the situation clarifies.
This is an interesting and defensible argument — basically, that the unclear shape of Brexit creates an option value to waiting. But I have three questions about it.
First, is this really the argument underlying all of these dire post-Brexit forecasts? My guess is that very few people reading news reports, or even briefing papers, about Brexit are hearing this; what they’re getting is much closer to the notion that uncertainty = increased probability of bad things. That is, this argument is a lot more nuanced and subtle than anything I previously heard in this discussion.
Second, doesn’t this argument imply a later investment boom once the uncertainty is resolved in either direction? That is, once Prime Minster Farage and President Le Pen have engineered the demise of the EU, there’s no reason to wait, and all the pent-up investment comes roaring back, right? But I haven’t heard anyone arguing that the contractionary effect of Brexit will be followed by a compensating boom once things settle down.
Third, doesn’t this argument suggest essentially the same effects from any policy negotiation whose end result isn’t known? Why don’t we say that the possibilities of TPP or TTIP are contractionary, because firms have an incentive to postpone investment decisions until they know whether these agreements actually happen? Somehow, though I’ve never heard anyone argue for the depressing effects of pending trade liberalization.
Again, I’m not trying to defend Brexit. But I worry that the urge to condemn it has led to a lowering of intellectual standards.
And let me also say that the narrative of disaster is coloring some (not all) financial reporting. On the whole, the market reaction looks pretty muted to me. It’s not just stocks: European bond spreads are about where they were a month ago. True, globally, rates are considerably lower; but we aren’t seeing the kind of financial disruption so widely predicted. Yet headlines about turmoil are everywhere.
Could I be wrong about all of this? Of course! But everyone really should ask where the consensus about Brexit macro is coming from.
Yesterday’s post was “Trade and Jobs: A Note:”
Given Trump’s economic policy speech – well, “policy” speech – it seems time to brush up on trade and jobs, an issue I used to write a lot about. The big story in the academia/policy space lately has been the work of Autor et al, who in two papers have estimated large losses from Chinese import penetration. I basically agree with this conclusion, at least when we’re talking about manufacturing employment. But I’m troubled by some conceptual issues, which I think are important for interpreting the results.
Maybe the best way to explain all of this is to start by talking about how I would do this – in fact, the way I’ve been doing it on the backs of various envelopes over the years. I would begin by posing a counterfactual: what would U.S. employment look like if we had pursued policies such as Trump tariffs that prevented the large trade deficits in manufacturing we actually have?
I’d start by arguing that a balanced expansion of imports and imports would have, to a first approximation, no effect on manufacturing value added, and an effect on employment only to the extent that import-competing industry is more labor-intensive than exports. Leave that on one side. Then what matters is the manufacturing trade deficit, which according to the WTO was approximately $600 billion in 2014.
How much manufacturing did that deficit displace? It doesn’t all come out of manufacturing value added, because a fair bit of a dollar spent on manufactured goods eventually shows up in purchases of non-manufactured imports. I need to do this more carefully (on deadline right now), but a rough number would be 60 percent for manufactured content; so we’re talking about $360 billion.
Then, employment: value added per worker in manufacturing is approximately $175,000. So this should translate into a bit over 2 million jobs.
OK, what about the effect on overall employment? In general, you can’t answer that with a similar computation, because it all depends on offsetting policies. If monetary and fiscal policy are used to achieve a target level of employment – as they generally were prior to the 2008 crisis – then a first cut at the impact on overall employment is zero. That is, trade deficits meant 2 million fewer manufacturing jobs and 2 million more in the service sector.
Since 2008, of course, we’ve been in a liquidity trap, with the Fed either unable or unwilling to hit its targets and fiscal policy paralyzed by ideology, so trade deficits are in practice a major drag on overall employment. But this is a bit different from the usual “trade costs jobs” argument.
So, how big a deal is displacement of 2 million manufacturing jobs? Not trivial, to say the least. But if you want to place it in the context of deindustrialization: we’re talking about 1.5 percent of the work force. So absent the trade deficit – that is, again, imagining some policy that prevents deficits from emerging – we would have roughly 11.5 percent of the work force in manufacturing, rather than the actual 10. Compare this with the realities of the past: more than 20 percent in manufacturing in the late 1970s, more than 25 percent in the 1960s.
OK, now for Autor and various co-authors, who do something very different: a bottom-up approach. They start with an empirical analysis, using cross-section data, of the impact of the China shock on employment, wages, and so on at the regional level – which is perfectly fine, and in fact beautiful work.
But what they do next is to apply the implied coefficient from this analysis to the aggregate effects of the China shock. And that’s much more dubious – especially when, in the second paper, they purport to estimate the effects on overall employment. In general, you can’t do that: applying estimates of partial regional effects to the overall aggregate exposes you to huge possible fallacies of composition.
And in this case the crucial issue is monetary and fiscal response. Up through 2007 we basically had a Fed which raised rates whenever it thought the economy was overheating; in the absence of the China shock it would have raised rates sooner and faster, so you just can’t use the results of the cross-section regression – which doesn’t reflect monetary policy, which was the same for everyone – to predict how things would have turned out.
Now, it so happens that my alternative procedure yields results for manufacturing alone that aren’t too much out of line with those papers. But this should be seen as jobs shifted out of manufacturing to other sectors, not total job loss, at least pre-crisis.