Monday, January 17, 2011

Having My Cake and Eating it Too

 Tim Duy and Andy Harless call me out for being critical of the the Fed's low interest rates in the early-to-mid 2000s and for being critical of the Fed for failing to stabilize total current dollar spending.  They say I cannot have it both ways.  They argue that in order to keep nominal spending stable in the early-to-mid 2000s, the Fed had to push the federal funds rate below its neutral rate level for an extended period.  Therefore, it is unfair for me to assign blame to the Fed for the credit and housing boom.  Scott Sumner and Bill Woolsey have also raised this question to me in the past.  So what do we make of it? I am being inconsistent?  

Though it may not convince everyone, there is a way to reconcile my two criticisms of the Fed.  The key to doing so is appropriately specifying the trend growth of nominal spending. I will define it here as the trend growth rate over the 1987-1998 period for several reasons.  First, its the part of the Greenspan period where there were no wide, unsustainable swings in economic activity.  Second, 1998 is when the Greenspan Fed for the first time significantly deviated from past practice by lowering the federal funds rates even though the economy was experiencing robust economic growth.  (Robert Hetzel argues in his book that the easing actually started in 1997 when the Fed failed to raise interest rates. See his chapter 16, Departing from the Standard Procedures.) The reasons for the easing were concerns that the global economic turmoil would spill over into the U.S. economy.  

With this trend, I now look to see how nominal spending has subsequently fared using two different measures in the figures below.  The first figure shows domestic total current dollar spending.  This is the narrower of the two measures of nominal spending since it leaves out foreign spending on the U.S. economy.  However, it is appropriate for looking at spending by U.S. residents: (Click on figure to enlarge.)



This figure reveals that there were two above-trend surges in nominal spending, one at the end of the 1990s and the other during the credit and housing boom.  The first nominal spending boom is consistent with the tech-bubble and the Fed's easing in the late 1990s.  Interestingly, the 2001 recession simply returns nominal spending to trend.  The second boom is the one in question and is much larger.  Note, that despite this boom nominal spending is now below trend and is increasingly moving away from it.  Thus, the Fed has failed to restore trend level nominal spending.   This failure amounts to monetary tightening.  QE2 is simply an imperfect attempt to bring the U.S. economy back to the trend.

The next figure shows all nominal spending on the U.S. economy, including that of foreigners.  This figure shows a similar picture though the nominal spending booms are little larger.  (Click on figure to enlarge.)



The monetary tightening is less pronounced in this figure, but it is still there.  Both figures imply there needs to be several periods of catch up nominal spending growth. Both figures also imply that the Fed allowed nominal spending to grow too fast in the early-to-mid 2000s.  Returning current dollar spending to trend would be much easier if the Fed would commit to an explicit nominal spending rule that would help shape expectations. 

7 comments:

  1. As you know, I use the second, (Final Sales of Domestic Product) and trend from 1984 to 2007 (and that is new in the last few months, before I always stopped in 3rd quarter 2008 before.) I show about 5.4 percent growth rate.

    What is the growth rate beween 1987 and 1998?

    ReplyDelete
  2. When I do it with NGDP (using 1997 to 1998 as the trend), I get NGDP consistently below trend after 2000. I guess the difference has to do with the increasing efficiency of inventories (and a shift toward industries that don't use inventories), leading to a decline in NGDP relative to final sales. Perhaps there's a philosophical argument to be had as to whether inventory investment should be included in the target series. (Offhand, I'd say that somebody is paying to produce those inventories, so they ought to be included, just as we include other forms of business investment.) In any case I can't see criticizing the Fed for being too loose on the basis of final sales when they appear to have been too tight on the basis of NGDP. If they had done what the model said and kept interest rates higher, my guess is that they would have undershot even by the final sales criterion.

    ReplyDelete
  3. My view is that the ideal series would include planned inventory investment and leave off unplanned inventory investment.

    The goal is to for firms to be able to plan (as I see it.) Including unplanned inventory investment (positive or negative) as spending is including failures.

    Ignore planned inventory investment. Suppose the target is 17 trillion, but final sales are 16.95 trillion. In my view, it is obvious that money was too tight, and the fact that firms produced 17 trillion expecting to sell it and that 50 billion of unsold output is counted as investment expenditure, hardly counts.

    Unplanned inventory investment represents error that needs to be avoided.

    Planned inventory investment? Well, you can treat it as a change in productivity rather than a use of output, but it is an area that could use more research.

    ReplyDelete
  4. David
    Check your trend. It cannot be correct. It says that from 1998 onwards dollar spendind (FSDP) was significantly above trend. Very counterintuitive!

    ReplyDelete
  5. Bill:

    The domestic demand growth rate is 4.8% and the aggregate growth rate is 5.7%.

    Joao:

    The figures does not show that nominal spending was above trend from 1998 on. It does show above trend growth in the late 1990s that returns to trend in the early 2000s and then goes above trend growth thereafter until the collapse. These two nominal spending booms are largely in line with the stock market boom in the late 1990s and the housing boom of the 2000s. This much is straightforward.

    As Bill and Andy note, though, one can question my determination of the dates used for the trend as well as the measure of nominal spending.

    ReplyDelete
  6. Excellent run of posts of late. Keep up the good work.

    ReplyDelete
  7. A good post very interesting

    ReplyDelete