07 Mar 2018

Fun With Graphs! Murphy vs. Erdmann

Economics, Market Monetarism, Scott Sumner 9 Comments

I don’t want to put too much stock in this post. My main point is to demonstrate that you can tell just about any story you want, if you play with FRED long enough. This is why Russ Roberts asks people when they present their slides showing p-values: “How many regressions did you run before you got this result?” (He talks about that in this interview.)

Anyway, in the comments of a recent post here, Kevin Erdmann rushed to defend the honor of Scott Sumner. Erdmann challenged my view (which is actually close to the conventional wisdom) of what happened with the housing crash. Among other points, he presented the following chart and commentary:

[Kevin Erdmann talking about the above:] “The funny thing is, if you look at Phoenix – the quintessential bubble city – construction and non-construction employment also both start to drop at the same time in mid 2006. And, in Phoenix, where you might think the drop in construction should have led the collapse, the drop in non-construction employment was actually a little sharper than the drop in construction employment.”

You can stare at Kevin’s chart for a bit and see if you follow his argument. (If you see Elvis, you’ve been staring too long.) Make sure you look at the units he uses.

OK, great. Now I’ll take the same variables Kevin chose, but I’ll pick different units. That gives me this picture:

I would argue that my picture corresponds pretty closely to the conventional wisdom on this one. Construction employment (expressed as an index, since after all there are way more total jobs than simply construction jobs) spiked and tanked far more severely in Phoenix than general employment. Contrary to Scott Sumner’s interpretation, you can clearly see that the collapse in construction employment slowed the rate of growth in total employment and “pulled it down”–it wasn’t that these construction workers were easily absorbed into other sectors, so long as the Bernank kept NGDP humming. (That was Scott’s theory, when he was looking at the stats on new housing starts.)

Before closing, one thing I’ll admit: I had expected the national unemployment rate to lag the Phoenix one, and that isn’t apparently correct. (It’s not on this chart I’m showing you–I wanted to just use Kevin’s variables–but I added it on my other browser tab to see.) So that might be a feather in Kevin’s cap.

Last point: Kevin says in his comment, “Yet, something caused all of that job growth and migration [in Phoenix] to suddenly break down in mid 2006.”

This observation is way more consistent with the Austrian story, than with Sumner’s story. According to Sumner, everything was fine until a self-fulfilling prophecy occurred in 2008, when the markets suddenly expected the Fed to fall asleep at the wheel and let NGDP growth collapse. The Fed said “Yep that is indeed what we’re going to do,” and then we had the worst recession since the 1930s, for no other reason than that the Fed stopped paying attention to TIPS yields. There was no “real” “structural” problem in the economy that would have required a recession. (Of course I’m somewhat caricaturing Scott’s position to make a point, but this isn’t made up.)

In contrast, the Austrian story says that the Fed slashed rates after the dot-com crash, brought them down to 1% by June 2003, held them there for a year, then started raising rates from June 2004 onward. So yeah, I agree that “tight money” caused the crash, but not in the way Scott means.

Note also that our disagreement on this is similar to how Krugman thought he blew me up (story here) by pointing out that central banks can influence the business cycle. Right, that’s key to the Austrian theory. Monetary disturbances cause malinvestments–setting in motion an unsustainable inflationary boom–which then necessitate a bust for “real” reasons. The Austrian theory is neither purely demand nor purely real.

9 Responses to “Fun With Graphs! Murphy vs. Erdmann”

  1. Kevin Erdmann says:

    Scott and I differ in our view of the timeline. I think the indicators suggest marginally recessionary conditions as early as late 2006, whereas he focuses more on the sharp downturn in 2008 in terms of calling monetary policy distinctly too tight. So he would probably not put as much emphasis on the correlated downturn in employment growth, GDP growth, etc in late 2006 as I would.

    • Andrew_FL says:

      This is because you cannot name a single time and place in the entire history of money when you don’t believe money was too tight.

      • Kevin Erdmann says:

        I don’t think money was too tight in 2002-2005, or throughout most of the 80s or 90s, or obviously the 70s, or the 50s and 60s, I suppose, if I’m thinking about it. But other than that, your comment was very cutting and insightful.

        • Andrew_FL says:

          I’m glad to hear you’ve changed your views since this is in contradiction to everything you’ve said in the past.

        • Kevin Erdmann says:

          Honestly, you must have confused me with someone else.

  2. David R Henderson says:

    How can the Fed slash interest rates? You seem to think that the Fed has a lot of power over interest rates. I don’t think there’s much evidence for that. Maybe the Fed can affect short-term rates by 25 basis points, but I would be surprised if it’s much more than that.

  3. E. Harding says:

    Construction employment in that graph starts declining in April 06, but unemployment continues declining rapidly and only bottoms out in Aug 2007.

  4. Kevin Erdmann says:

    A couple of comments on the last graph and the idea of coordination.

    1) Here are the employment measures from the last chart in the post, to get a sense of scale:
    https://fred.stlouisfed.org/graph/?g=iUiE

    2) Phoenix is a migration town. In business cycles, workers don’t initially resort into new sectors. They resort into and out of Phoenix. In 2004 and 2005, about 35,000 households migrated into Phoenix each year, net of outmigrants. Then, that declined linearly until it was down to 7,000 in 2008.

    Many had been moving from California in order to lower their housing costs. California had plenty of jobs, but a lack of homes. In 2006, when job growth leveled out, the lack of jobs became more important than the lack of homes, so they stopped moving out of LA.

    This was a peculiar economic event, which I think it may be difficult to place into broad macro models. At the core of the event is a political barrier to housing which prevents capital from entering housing markets where it could earn above-market returns. It wasn’t like other downturns.

    I don’t claim to have a universal explanation for all macro events over all time. I just have the explanation for this one.

  5. Ty Fyter says:

    Speaking of Russ, has he interviewed you about any of your books (i.e. Choice or HPBRW)? If so then I missed it somehow….

Leave a Reply