Monday, May 24, 2010

More Aggregate Demand Weakness?

The wailing and gnashing of teeth over deflation has begun anew. With the CPI showing a decline in April, folks like Paul Krugman and Greg Ip are concerned about a Japanese-style deflation emerging in the United States. Other observers like Tim Duy and Scott Sumner are similarly concerned as they fear U.S. aggregate demand is going to get increasingly weak. I do not mean to be a contrarian here, but I am having a hard time seeing how these concerns are justified. Yes, the U.S. economy has been plagued by a weak recovery, but the image one gets from this discussion is that we are standing on the precipice of another collapse in spending. If so, the data sure don't show it. Take, for example, monthly retail sales. It grew at a year-on-year rate of almost 10% in April as can be seen in the figure below. The figure also shows domestic demand through 2010:Q1. Given the strong correlation between these two series, it seems likely that the strong growth in current retail sales will also be seen in domestic demand once the numbers are released.(Click on figure to enlarge.)

Of course, the above figure only shows current values. It is possible that the current growth in demand may not be sustained. But that is not what the consensus forecast says. Below is a figure constructed from the Philadelphia Fed's quartely Survey of Professional Forecasters. This figure shows the recently released average forecast for nominal GDP (NGDP) for the period 2010:Q2-2011:Q2. No sign of an expected aggregate demand collapse here. (Click on figure to enlarge.)

In short, these figures shows that both current and expected spending are growing. It may be not be growing as fast as we want, but it is growing and there is no sign of an imminent collapse. Now if aggregate demand is growing and is expected to grow how is it possible to have inflation falling? The simple answer is that aggregate supply must be growing as well. This understanding finds support in the following figure which shows the year-on-year growth rate of labor productivity. Notice the large spike in the productivity growth rate starting in 2009:Q4 (Click on figure to enlarge.)

As I noted in my previous post, this spike in the productivity growth rate may be why the unemployment rate has been remained so high. If, in fact, this productivity surge is the cause of the downward price pressures then here is another case that illustrates why it is important to distinguish between deflationary pressures arising from a negative aggregate demand shock versus those coming from positive aggregate supply shocks. I have written about the importance of this issue before, especially with regards to the 2003 deflation scare.


  1. I wonder what forecasts of future nominal GDP looked like in the summer of 2008 ? Have you run those numbers by any chance, using data available to people at that juncture ?
    And I suppose any number of economists of the more leftish persuasion would argue that aggregate demand is prima facie weak when you have 10% unemployment...almost by definition dont you think ?

  2. ECB:

    I suspect the forecasts then were way off. Here, however, the issue is not forecast accuracy but whether there is any evidence that expectations of future AD have declined. If so, then the Krugman-Ip story makes sense. If not, then we have to look elsewhere--positive AS shocks.

    Yes, the high unemployment is treated by many as prima facie evidence of weak AD. As I note in my last post, that is simply an assumption, a contestable one.

  3. In retrospect I think it's safe to say there was a swift decline in AD starting in July 2008. The financial indicators showing this were:
    1) Declining commodities prices
    2) Declining T-Bond yields and TIPS spreads
    3) Declining equities
    4) Rising dollar
    Anyone watching those indicators probably could have forecasted a decline in AD by early September.

    This time we have all the same signs. And the market rumblings are continuing. This morning the Dow dropped nearly 300 points on opening, the German and US 10 year bonds have fallen in yield to 2.56% and 3.11% respectively, oil is down to $67 etc. It's hard not to share Krugman's, Sumner's and Duy's concern.

    Mark A. Sadowski

  4. I think Scott Sumner would agree with Mark Sadowski above. However, there is an inherent problem with using the "declining commodity price" signal as a proxy for AD. In the summer of '08, oil prices declined from an unprecedented high of $147. Is the roughly twenty percent decline from $147 to $120 evidence of deflation? What about a decline to $85? Given that the latter figure was still above the historical average, it is hard to get too excited about the deflationary implications of $85 oil.

    So using the first derivative of commodity prices as a guage of AD expectations is difficult. Sumner proposes NGDP level targeting as a better alternative, yet the problem is similar: assuming that we had overshot NGDP targets in the summer of '08 (given high inflation expectations), then the initial decline to target would hardly have been reason for the Fed to act. Does one lower rates as we decline to target, or wait until we overshoot the target? Sumner, to my knowledge, has not outlined the mechanics, and yet from the mechanics flow the effectiveness of policy.

    The S&P 500 is a decent proxy for NGDP expectations, and it is falling, yet still far above the levels where deflation was the overriding concern. Assuming NGDP futures followed the same path, should the Fed be resuming asset purchases that it began when the S&P was at 700-800 now, or wait until the S&P is back at that level? How would the Fed know, in advance, if the correction in the S&P is merely temporary (like Febuary's), or more lasting and severe? How should it take account of gold prices and 10-yr TIPS spreads, both of which are much higher than the fall of '08?

    The bottom line is that I have some difficulty differentiating between targeting the S&P and expected NGDP levels. In the context of the S&P, it is much easier to imagine that the policy would be difficult to implement and have significant unintended consequences.

  5. Mark and David:

    I just did another post acknowledging my evidence only goes through April. Current market conditions are less promising for AD. Wouldn't it be nice to have nominal GDP futures market about now?

  6. How is it that high unemployment is due to productivity shocks? We had very high TFP growth in the 1960s with comparatively low unemployment. Again, the 1990s were the age of high IT-driven productivity, and by 1999 unemployment was down to 3.9%. How come productivity shocks suddenly cause high structural unemployment in the 2000s? Is it because the skill set of workers was complementary to technology shocks in the 60s and 90s, but now it isn't, or perhaps the housing crisis has just gummed up labor mobility ?