Wednesday, November 1, 2017

Monetary Regime Change Update

I recently made the case that we got a monetary regime change in 2008 that explains the stubbornly low inflation since that time:
A monetary regime change has occurred that has lowered the growth rate and growth path of nominal demand. Since the recovery started in 2009Q3, NGDP growth has averaged 3.4 percent. This is below the 5.4 percent of 1990-2007 period (blue line in the figure below) or a 5.7 percent for the entire Great Moderation period of 1985-2007. Macroeconomic policy has dialed back the trend growth of nominal spending by 2 percentage points. That is a relatively large decline. This first development can be seen in the figure below.

The figure above also speaks to the second part of this regime change: aggregate demand growth was not allowed to bounce back at a higher growth rate during the recovery like it has in past recessions. Historically, Fed policy allowed aggregate demand to run a bit hot after a recession before settling it back down to its trend growth rate.  This kept the growth path of NGDP stable. You can see this if the figure above by noting how the growth rate (black line) typically would temporarily go above the trend (red line) after a recession.  
Had macroeconomic policy allowed aggregate demand growth to follow its typical bounce-back pattern after a recession, we would have seen something like the blue line in the figure. This line is a dynamic forecast from a simple autoregressive model based on the Great Moderation period. This naive forecast shows one would have expected NGDP growth to have reached as much as 8 percent during the recovery before settling back down to its average. Instead we barely got over 3 percent growth. This is why NGDP has never caught back up to its pre-crisis trend path.  
Again, these two developments are, in my view, the real story behind the drop in trend inflation. And to be clear, I think both the Fed's unwillingness to allow temporary overshooting and the safe asset shortage problem have contributed to it. 
I went on to say this is the monetary regime change no one asked for. It is also one that many observers seem to miss in their analysis of Fed policy since the crisis.  

Well, I was on twitter discussing long-term treasury yields and monetary policy with Tim Duy and Joel Wertheimer. I decided to whip up some charts comparing 1-year head NGDP forecasts from the Philadelphia Fed's Survey of Professional Forecasters against 10-year treasury yields. The results, in my view, are consistent with the claim that there was a monetary regime change in 2008. 

First, here is the chart for the 1980:Q1-2007:Q4 period. It shows a fairly strong relationship between expected NGDP growth and long-term treasury yields:

The next figure shows the relationship for the period since 2008:Q1. Say goodbye to that relationship:

Just to be robust, I took out the outliers in the above chart (which are for the periods 2008:Q4-2009:Q2) and got the following chart:

So something has changed in the relationship between expected NGDP growth and long-term treasury yields. The monetary regime change story outlined above coincides closely with this breakdown. Here is one way to connect these two developments. Before 2008 the Fed allowed its tightening to follow the pace of recovery, whereas afterwards the Fed has tended to get ahead of the recovery in its desired and actual rate hikes. If so, the Fed's tightening post-2008 would have slowed down the recovery and lowered the expected future path of short-term interest rates. This overreacting by the Fed would have kept long-term interest rates from rising with expected rises in NGDP growth. This type of behavior would also be consistent with a monetary regime change where only low rates of NGDP growth and inflation are tolerated. This is what appears to have happened in 2008. 


  1. I did some analysis on this. I used Q3 1981 to Q4 2007 because I wanted to use some additional variables. There is a strong relationship between 10yr treasury (end of period) and expected NGDP from SPF. However, if you break it up to RGDP and PGDP, RGDP is not significant. The whole thing comes from PGDP. Using just two variables, the best I found was the expected 3 month yield and the expected unemployment rate. The R^2 over this period was like 92% compared to (a still very good) 88.5% for just PGDP and a more modest 74% for NGDP.

    So now if we look over your later period (with the exclusion), I get an R^2 of basically zero regressing 10year yield on NGDP. It's still basically zero when you split RGDP and PGDP. However, using the two variables above (expected 3 month yield in 1 year and expected unemployment in 1 year), the R^2 is around 50%. While it's significantly below where it was before, it's still a lot better than the NGDP forecasts.

    I also figured that it might be interesting to split the expected 3month yield into a real rates component and a expected inflation component. Things don't change all that in the first period. However, in the second period, the expected inflation component is no longer significant.

    I also considered moving the date to consider only the period where the current 3-month yields were near zero (Q3 2009 to present). Your time period actually included a period where yields were declining, where things weren't exactly different. In this period, it's only the expected inflation that describes 10-year yields. Not the expected unemployment rate (b/c it declines a lot in the beginning and not totally representative of a whole cycle) or expected real component of 3 month yields. However, the coefficient on expected inflation is negative! At a very significant level. This is mostly because of the 2009-2010 period when inflation expectations were low and interest rates were still high. One way to look at it is that it took some time for the market to adjust to the new normal.

  2. Is this really a new paradigm? Didn't 10 year treasury yields stay way below NGDP growth in the 1930 to 1965 timeframe?

  3. I recommend to David and other Macro Market Musings readers to read this working paper:

  4. I believe that they are trying to keep the yield curve flatter this time around. This reflects their belief that the very steep yield curve circa 2003 encouraged a dangerous buildup of short-term debt that then exploded as the curve inverted in 2007.

    I don't see anything fundamentally undesirable about a flatter yield curve. Historically they have indicated economic stagnation only because an excess of short-term debt was allowed to build up beforehand. The issue is that this strategy necessarily forces them to tighten on the short-end as GDP growth accelerates, which necessarily will slow down growth in the short-term.

  5. Well, I was right all along. I said the Fed was too tight.

    Not only that, we could ponder if the Fed had really gone great guns, and shot for a robust recovery, ala Arthur Burns, 1976
    through 1979. Burns obtained a 20% expansion in GDP in four years.

    Central bankers actually believe that holding inflation under 2%, as opposed to 3%, is important, no matter how much real output is lost.