From the Huffington Post, I see that “non-partisan” Brookings has published a research piece showing that stocks will fall 10 to 15 percent if Donald Trump is elected president of the United States.
I put Brookings’ non-partisanship in scare quotes there because their tendency to be moderate and empirical means these days that they are partisan Democrat. It is not that Brookings has moved, but that the GOP has gone bonkers. This is an excellent time to be “partisan,” rather than a weak-kneed professional centrist, of the ilk the NYT runs an affirmative action program for on their op-ed page. Brookings should embrace its partisanship, like a German Social Democrat in the 1930s might.
The Huff Post article is really accessible and mostly captures a tweet storm from Justin Wolfers with some really cool, very-high-frequency charts showing how risk asset prices have moved in line with changing probabilities of a Clinton or Trump presidency. If you want the original paper, here is the link to it.
This is the event-study approach to analyzing the effects of policy innovation or political shock on financial markets. And economists really need to stop doing this, because these studies are based on the false premise that financial markets are sufficiently rational and noise-free to allow short-term movements in them to be a proxy of likely enduring effects. Really, guys: stop doing this. It is foolish.
Here is the money paragraph from the Brookings paper setting out their “findings.”
The magnitudes involved are large. The S&P 500 futures rose by 0.71 percent during the debate window in response to a 6 percent increase in Clinton’s victory probability. This implies that market participants believe that the S&P 500 will be worth 12 percent more under a President Clinton. Movements in other U.S. stock indices tell a similar story. A 12 percent difference is large both in absolute terms and relative to how previous political shocks have moved the market.
I can see why they would want to draw such a grand conclusion. Thanks to some contacts at Predictwise and apparently in markets, they gained access to some BIG DATA that allowed them a unique insight into these short-term movements. And it is indeed fun reading. As a scribbler myself, I assume it was also fun writing too. Very cool. Bit envious.
But the conclusions they draw are way too strong for the method they use. And the false confidence with which they present their findings is irksome because it discredits economics. Hence I wish guys would stop doing this!
On the other hand, macro has not covered itself in glory in recent years, Krugman aside, so maybe the first step to recovery is for macro to admit it has a problem. Maybe the discrediting is not a bad thing? Maybe I need to give that some thought.
I remember just before the Brexit vote that Larry Summers trotted out this same approach to claim that a Leave result would hit the FTSE really hard:
Second, markets are likely to suffer extraordinary volatility in the wake of Brexit. A Black Friday could follow referendum Thursday. It is likely that foreign investors in British stocks would lose 15 percent off the bat, adding together market declines and currency losses. This is a judgement supported by the gyrations in markets induced by relatively small fluctuations in the perceived chance of Brexit and by the very high prices commanded by out of the money options. The truth is that even with all the regulatory changes that have been put in place we do not know for sure how the financial system will respond. A return of systemic risk as large losses lead to cascading liquidations cannot be ruled out. At a time when central banks have far less ammunition than they did in 2008, the consequences could be grave.
I whined about that in real time, making three points:
First, the event study approach that he cited implied a decline more along the lines of 40% than 15%.
Second, event studies are incredibly unreliable in this context. In the event, the FTSE went up. Sure it went up for a reason. Unless you are a modern physicist, everything has a cause, in this case perhaps the BoE’s reaction. But were markets unaware of the existence of the BoE? Please.
Third, why must we measure political events through their effects on markets? Brexit can be dumb regardless of how equities do. (I would now add sterling, which seems coincidentally right, but is not itself a problem.)
Incidentally, this same nonsense approach is why academic economists puffed themselves up and proclaimed with great gravitas and pseudo-objectivity that the “evidence” (heaven forbid any presence of priors) implied that QE was worth 200 or 300 or whatever basis points off the 10-year Treasury yield. I don’t know the exact number, because I stopped caring a while ago, it being so ridiculous.
Of course, when QE ended, bond yields went down. Oops. I know, counterfactual, not a controlled experiment, blah blah blah. There is no end to the ex-post rationalizations for having got the direction wrong. Again, we have a case of event studies not working. Can we accept that before moving on to counting angels?
Returning to Trump, my opinion does not matter but I will give it anyway. The problem with the Brookings piece is not that it is too partisan, but that it tries too hard not to be. I think a Trump presidency could be an extinction level event, where equities might stop trading outright, perhaps on their highs. Who knows?
Why be pseudo-centrist about it and pretend just to be objectively assessing the evidence, which was probably chosen ex post to fit the priors anyway? Would that “study” have been done if markets were indifferent?
In my view, which I fully admit is speculative, if Trump wins we will have much more to worry about than a 10 to 15% decline of the stock market.
Indeed, I am hoping stocks go down 10 to 15% after Hillary wins so I can buy some to fund the retirement, which has maybe already begun? Geeze. Ouch.
Right now, the returns look pretty uninspiring. But they are hardly the main thing confronting the country.