Another example of the importance of starting dates

Back in October, Yglesias reported on how the Trump stock market rally wasn't all that impressive. Other countries' stock market indices had risen by more than ours had. To illustrate this, he included this graph comparing the US S&P 500 with Japan's Nikkei, Germany's DAX, and France's CAC indices:

It clearly shows that, while the S&P500 has risen considerably since Aug 2016, the rise is not as large as the gains experienced by other countries. He concluded:

That said, the fact that stock market enthusiasm over the past year has been worldwide with the United States lagging other key countries seems like a strong indication that Trump hasn’t done anything that’s particularly successful or exciting. ... For whatever reason, markets are up just about everywhere, not only in the United States. And markets generally seem to be up by more in countries with boring, competent-seeming leadership than they are in the United States.

I liked the graph and it is a good idea to think in terms of such counterfactuals, but I also had my doubts. Why normalize all the indices at the end of July, half a year before he became president and months before he won the election? In the middle of the semester I felt it was interesting enough to mention to my honors students while bringing up a concern or two, but I eventually forgot my desire to investigate further.

Well, time to investigate! Compare if you will the following graphs with three different normalizations: Yglesias' end-of-July 2016, the election in 2016, and the inauguration in 2017. I'm extending the data out to today, using weekly closing numbers.

The S&P's relative performance has improved since Yglesias wrote, so that even using his normalization the US is now modestly outperforming Europe and has been for the most part since the Tax Cut and Jobs Act. The Nikkei is still outperforming the S&P by a wide margin, but his conclusion would now have to be that the President's antics haven't harmed the US.

If we normalize just before the election, however, we see that the US is right on par with Japan and outperforming Europe by a wide margin.

And normalizing from the inauguration shows the US ahead of every other country with the most boring and competent seeming Germany performing worst.

So what's the takeaway? We need more crazy antics? I doubt it. My three lessons for today are about how little we know:

1) If you're going to do this kind of analysis, be forthright about why you are choosing your starting dates. Starting dates are everything. That's part of why statistics gets its reputation for lies: you can make the same numbers tell almost any story you want. If you want US stock returns to look as bad as possible, start at July 3, 2016 (really close to Yglesias' starting point) so that Japan is 20 points ahead of the US. If you want the US stock returns to look as strong as possible, go back nearly 3 years ago to June 28, 2015 and give Trump credit for growth that happened during Obama's term, putting the US 20 points ahead of Japan. Or maybe just pick a credible day from which we can justly and reasonably judge the President's performance.

2) Let's please remember on all sides that the stock market is not a great indicator of how the economy as a whole is doing. The correlation may even be negative in the very long run (http://www.businessinsider.com/equity-returns-and-gdp-per-capita-2014-2)

3) Let's not draw too many life lessons about how to run an economy and a presidency until all the data points are in. 

We need more stats, stat

Tabarrok: Diversified stock portfolios have earned 7% on average, but with a large standard deviation. Even if you follow a buy and hold strategy, almost 70% of the time (more than 2/3) you will get less than 7%. In fact over 30 years, nearly 10% of all portfolios will lose money. The average portfolio return includes a handful of enormously large winners, so that the median return is only 5.1%. As one comment on the post pointed out: "Have you ever heard someone says that when you are young you should invest in stocks because even though they are risky you will have time to cancel out the ups and downs? Sure, you have, even someone as smart as Burton Malkiel has made this argument. Alex’s post shows that this common argument is wrong."

Blattman - Science magazine is now requiring empirical work to pass a statistics editor fluent in your empirical methodology. He notes, and I agree wholeheartedly: "In particular, I think that a 21st century undergraduate degree in social science ought to require fluency in statistics. It’s such a fundamental part of science, medicine, social science, and even reading the newspaper."

Henderson comments on McArdle's post about the choice between a 15-year and 30-year mortgage: 

"If you expect higher inflation in the future and you expect that inflation to stay at that higher level, that argues for the 30-year mortgage over the 15-year mortgage because the 30-year mortgage leaves more principal for inflation to whittle away. Other than a handful of gold coins, I have few good inflation hedges. My fixed-interest-rate loan is one of them. The more slowly I pay it off, the longer I keep my inflation hedge."

Goldstien - Benford's Law tells us that the first digit in a lot of numbers is more likely to be a 1 than a 9. Goldstien tells us how to use that information to double-check the survey information we have gathered. A comment usefully adds: "There is also a Stata package called firstdigit that will run Benford's Law tests for you: http://ideas.repec.org/c/boc/bocode/s456842.html."

How do you know when the Fed is being successful?

The first question you should ask is: successful at what?

Sumner on the 1936 Fed:

This is one of the most chilling statements I have ever read.  The opening sentence is the sort of thing juvenile delinquents say to each other when their prank has gone horribly awry, and they are nervously working on a joint alibi.  An incredible effort at denial runs all through the piece.  First he admits that they raised reserve requirements because “some recession was desirable.”  Then he claims it was just a “coincidence in time” that the downturn followed the reserve requirement increase, even though the express purpose of the increase was to cause a “recession.”  Then he claims that if they reverse their decision it will look like the previous decision had caused the recession.  Then he said that a depression can’t be happening, because there is no good reason for a depression.  Well it was happening, unemployment rose to almost 20% in 1938.  In the end, they decided to stick with the high reserve requirements throughout the rest of 1937.

Sumner on the 1960's-70s Fed:

The Fed can always raise short term rates.  It’s true that this doesn’t always result in higher long term rates.  But that’s not a sign that monetary policy is ineffective, rather exactly the reverse.  The Fed raises short term rates to keep inflation down close to 2%.  If long term rates don’t budge very much, that’s a sign that monetary policy is credible.  In the 1960s and 1970s both long and short rates tended to rise together, hence increases in short rates weren’t enough to get ahead of the curve.  Real rates didn’t rise, as the Fed wasn’t following the Taylor Principle.  Because the “tight money” policies weren’t credible, inflation expectations rose.

Glasner on the September 2011 Fed:

In this environment small changes in expected inflation cause substantial movements into and out of assets, which is why movements in the S&P 500 have been dominated by changes in expected inflation.  And this unhealthy dependence will not be broken until either expected inflation or the expected yield on real assets increases substantially. ...

Operation Twist has almost certainly not been responsible for the rise in stock prices since it was implemented.

Why has the stock market been rising? I’m not sure, but most likely market pessimism about the sway of the inflation hawks on the FOMC was a bit overdone during the summer when the inflation expectations and the S&P 500 both were dropping rapidly. The mere fact that Chairman Bernanke was able to implement Operation Twist may have convinced the market that the three horseman of the apocalypse on the FOMC (Plosser, Kocherlakota, and Fisher) had not gained an absolute veto over monetary policy, so that the doomsday scenario the market may have been anticipating was less likely to be realized than had been feared.

Sumner on the August 2011 and January 2012 Fed:

My hunch is that I misjudged the Fed move back in August, when they promised low interest rates for the next two years.  That seemed pointless without making the promise condition on some sort of nominal growth target (GDP or inflation. Now there are indications that the Fed may do just that at the January meeting. ... It’s interesting that it took me so many months to have second thoughts about my negative verdict on the policy last August.  Equity investors seemed to need only about 30 minutes to figure this out (after the 2:15 announcement.)

And what should the Fed be doing? Here is Glasner arguing in favor of targeting nominal GDP over nominal wages - though both are highly correlated, targeting nominal GDP will avoid mistaking good supply shocks for contractions.