“Don´t reason from a chart change” – George Selgin´s question to Market Monetarists

In the past few days there´s been some back and forth among MM´s following George Selgin´s “A question to Market Monetarists”. In that post GS writes:

My third reason for hesitating to endorse proposals for doing more than merely sustaining the present 4-5 percent NGDP growth rate is the one I consider most important.  It is also one that has been gaining strength since 2009, to the point of now inclining me, not only to keep my own council when it comes to arguments for and against calls for more aggressive monetary expansion, but to join those opposing any such move.

My third reason stems from pondering the sort of nominal rigidities that would have to be at play to keep an economy in a state of persistent monetary shortage, with consequent unemployment, for several years following a temporary collapse of the level of NGDP, and despite the return of the NGDP growth rate to something like its long-run trend.

Apart from some die-hard New Classical economists, and the odd Rothbardian, everyone appreciates the difficulty of achieving such downward absolute cuts in nominal wage rates as may be called for to restore employment following an absolute decline in NGDP.  Most of us (myself included) will also readily agree that, if equilibrium money wage rates have been increasing at an annual rate of, say, 4 percent (as was approximately true of U.S. average earnings around 2006), then an unexpected decline in that growth rate to another still positive rate can also lead to unemployment.  But you don’t have to be a die-hard New Classicist or Rothbardian to also suppose that, so long as equilibrium money wage rates are rising, as they presumably are whenever there is a robust rate of NGDP growth, wage demands should eventually “catch down” to reality, with employees reducing their wage demands, and employers offering smaller raises, until full employment is reestablished.  The difficulty of achieving a reduction in the rate of wage increases ought, in short, to be considerably less than that of achieving absolute cuts.

U.S. NGDP was restored to its pre-crisis level over two years ago.  Since then both its actual and its forecast growth rate have been hovering relatively steadily around 5 percent, or about two percentage points below the pre-crisis rate.The growth rate of U.S. average hourly (money) earnings has, on the other hand, declined persistently and substantially from its boom-era peak of around 4 percent, to a rate of just 1.5 percent.**  At some point, surely, these adjustments should have sufficed to eliminate unemploymentin so far as such unemployment might be attributed to a mere lack of spending.

And so, my question to the MM theorists: If a substantial share of today’s high unemployment really is due to a lack of spending, what sort of wage-expectations pattern is informing this outcome?

Scott Sumner replied:

Let me admit right up front that the relatively slow adjustment of wages and prices is not 100% consistent with the natural rate model that I have in the back of my mind.  Krugman has made the same concession.  George was kind enough to send me a graph showing wage growth and NGDP growth over the last decade, which will help me to answer this question:

The unemployment rate peaked at 10% in 2009, and has since fallen to 8.2%.  Notice that a huge gap opened between NGDP and wages in 2009, and that gap has partially closed.  Just eyeballing the graph, it looks to me like perhaps 40% of the gap has closed.  And that’s pretty consistent with the fall in unemployment from 10% to 8.2%, if you assume 5.6% is the natural rate.  (I.e. a 1.8% drop is roughly 40% of the 4.4% drop needed.)

But George is right that there is another problem here; why haven’t wages adjusted more quickly?  Recall that in the 1921 recession (caused by severe deflation) wages fell quickly, and employment grew rapidly after wages adjusted.  What’s wrong with our modern labor market?  Or is the market fine, but I’m in error assuming the natural rate model explains recessions?

Meanwhile GS did another post: “…And my own attempt at an answer”:

Here (referring to the chart above) one can clearly see how, while NGDP plummeted, hourly wages kept right on increasing, albeit at an ever declining rate. Allowing for compounding, this difference sufficed to create a gap between wage and NGDP levels far exceeding its pre-bust counterpart, and large enough to have been only slightly reduced by subsequent, reasonably robust NGDP growth, notwithstanding the slowed growth of wages.

The puzzle is, of course, why wages have kept on rising at all, despite high unemployment. Had they stopped increasing altogether at the onset of the NGDP crunch, wages and total spending might have recovered their old relative positions about two years ago. That, presumably, would have been too much to hope for. But if it is unreasonable to expect wage inflation to stop on a dime, is it not equally perplexing that it should lunge ahead like an ocean liner might, despite having its engines put to a full stop?

But in an addendum he “backtracks”:

ADDENDUM (July 9, 2012): It turns out that the graph I concocted appears after all to have misled me (I confess I was apparently too willing to be convinced that there might be a case for further NGDP stimulus after all); playing around with it further (by using the same scale for both plots and letting 1/1/2005=100), I come up with another that actually reinforces the position I took in my original post:

In light of these further findings, I’m back to my original question: has the Fed, despite the still high level of unemployment, already done all that it ought to do in the way of monetary easing despite the still high level of unemployment?

The simple task of ‘rescaling’ a chart should not be enough to change one´s mind (or “mislead me”). If it does, there´s something missing in one´s reasoning.

Actually, the ‘rescaling’ could be even less ‘drastic’ than the one performed by Selgin and the ‘conclusion’ would be the same. For example, I picked his original chart and just rescaled the right axis to a narrower range (110 to 145 instead of 152). The result is that the wage-NGDP gap is much narrower at present.

In post today Bill Woolsey comments on Selgin:

To my reading, by the time the recovery started, Selgin thought it least bad to just accept the new baseline and then allow for slow, steady nominal income growth from there.   Perhaps at his preferred one percent trend.   (I may have been reading too much into his remarks.)     To me, this is the implication of growth rate targeting.

Nominal GDP growth rate targeting would be for nominal GDP to grow the targeted amount from wherever it happens to be.    While level targeting is often characterized in terms of catch up growth (or slowing,) I think this sort of definition makes growth rate targeting somehow fundamental.   To me, level targeting means that the target is a series of levels going out into the future.   Pick a future date, and there is a target level associated with that date.   You can look at the rate of increase in the target levels.  You can calculate a growth rate between some current value of nominal GDP and the targeted level at some future date.   But the target for any future date is the level–not some growth rate from the current value.

And ends with a ‘punch’:

Market Monetarists are well aware that it is possible that potential output has fallen about 13 percent below the trend of the Great Moderation.   (The CBO estimate of potential output is about 7 percent below the trend of the Great Moderation.)

And perhaps the equilibrium quantity of labor has fallen some huge amount too.   The supply of labor decreased along with the demand, or the supply of labor is very elastic.

But I don’t believe it.

As I said before, I think a traditional recovery in nominal expenditure, like that of the Reagan-Volcker years, would be appropriate.    And then we should move to a long run, noninflationary growth path for nominal GDP.

I think Bill gets it right. My take:

The following panel shows the behavior of NGDP and wages (the same wage definition contemplated by GS) over the 1964 – 2012 (first quarter) period. This was chosen simply because the data for wages start in 1964.

The panel is divided into four periods; 1964 – 79 which closely conforms to what is known as “The Great Inflation”; 1980 – 86 which I have named “The Volcker Transition”; 1987 – 07 “The Great Moderation” and 2008 – 12 which I call “The Bernanke Depression”.

Look at the left hand column of the chart – “Great Inflation” and “Great Moderation”. In both instances, nominal wages track nominal spending very closely. In the “GI” period there´s a rising trend (in logs) in NGDP & wages. In the “GM” period the trend is (log) linear, meaning growth in both NGDP and wages are “constant”.

And that pattern, wages evolving alongside with NGDP, is what we should expect, be it because inflation is rising but also in the case inflation is stable.

The right hand column shows situations where that pattern is broken. This is associated with periods of more drastic adjustments. In the 1980 – 86 period monetary policy was geared to bring inflation down. That required a change in the path of NGDP. Initially NGDP growth slowed markedly. Inflation was brought down and wage growth slowed too, albeit much less than the drop in NGDP growth (which remained positive all along).

After inflation came down the “Volcker Moment” materialized and NGDP ‘sprinted up’, just as Bill Woolsey thinks should be done at present.

But looking at the lower right hand side chart in the panel one could argue, like GS did in his ‘Addendum’, that the economy has experienced it´s “Volcker Moment”. After all, hasn´t NGDP surpassed wages?

It sure has, but GS was on the right track before being ‘tricked by his chart’ when he took into consideration the BIG drop in NGDP after mid-2008. And that´s the defining characteristic of “Bernanke´s Depression”, something unheard off since 1938!

In that case, reasoning in terms of growth rates leads you into all sorts of errors. As Bill Woolsey reminds us:

During the Great Moderation, nominal GDP stayed on a very stable growth path.   And now, it has shifted to a significantly lower growth path with a slightly lower growth rate.  Market Monetarists favor level targeting and so it is natural for us to insist that the Fed should have kept the economy on the trend growth path and reversed any deviation.  The only way to reverse any negative deviation is for nominal GDP to grow faster than trend.

And illustrates with MM´s by now traditional LEVEL chart:

And even if you assume a lower trend path, the economy is still deep inside “The Hole”.

The next panel shows the associated behavior of employment.

Punch line: Either GS will see University of Georgia wages being cut, level cut, or he should pray the Fed gooses up nominal spending.

24 thoughts on ““Don´t reason from a chart change” – George Selgin´s question to Market Monetarists

  1. Nunes delivers a powerful wrap-up and punch!

    I still wonder why Woolsey (a nice and smart guy) is so inclined to zero inflation. In the modern economy, such goals are utopian, and fantastically expensive to obtain.

    The goal of macroeconomic policy should be real growth, sustained. If that happens at 2 percent or 4 percent inflation, who cares? My guess is right now the highest real growth is consistent with about 4 percent inflation.

    I agree that inflation rates into double digits are alarming although in fact the USA prospered with such rates before.

  2. “To my reading, by the time the recovery started, Selgin thought it least bad to just accept the new baseline and then allow for slow, steady nominal income growth from there. Perhaps at his preferred one percent trend.”

    That is indeed reading more into my arguments than I meant to put in them. My claim was not necessarily that there was no need for further demand boosting “when the recovery started,” but rather that there may no longer be such a need today, that is, that perhaps by now the “gap” of which the graphs attempt to speak has already been sufficiently closed. Nor do I endorse the 1% alternative, or anything so markedly different from recent trends, except as a target to be approached gradually enough for expectations to go along with the adjustment.

    I also don’t see why even the “Sumner” trend should define the relevant demand “hole” in a world in which wage growth is itself capable of responding to some extent at least to slack demand and unemployment. In fact it has responded, by slowing down considerably. The question is, by how much has that slow-down reduced the hole? The various graphs represent more-or-less flawed attempts to get a handle on that crucial question.

    • George – Thanks for commenting…
      I´ll stick to your last paragraph, which comments on what I wrote (the other two refer to Bill Woosley´s comments that I linked to). I agree, the “Sumner trend” is one among several other possibilities. Sure that wage growth is capable to respond to some extent to slack demand. It has done so on other occasions.
      My main point (and the one you considered just before the “addendun” in your second post) is that “this time it´s different”. There was not just a slowdown in nominal demand growth. It TANKED. And I don´t believe we can go back to being a “1921 society” where nominal wages fall absolutely. That´s why I think the alternative, to get NGDP back to a higher level (research could surely help identify a ‘reasonable’ one), minimizing the pressure on sticky wages, is the better option.

  3. My take away from this debate is GS is talking growth rates and MM’s are talking levels – the gap between where NGDP would have been if it had not been allowed to plunge vs. just letting bygones be bygones and growing NGDP 4-5% since the plunge at the new level. Although I had little reason to doubt MM logic in the first place, the pictures are worth more than a thousand words and provide more precision to my understanding.

    On a different point, I am very interested in economics, but I have trouble with the moral hazard involved in the taking of positions that seem impossible to reconcile with our supposed humanity and the glibness in which many in the profession who should understand how the monetary ecosystem works take them. Perhaps I have lived a sheltered life over the last 4 decades because I never noticed such ‘c-ya wouldn’t want to be ya’ attitude taken enmass regarding monetary policy as I have witnessed of late. I feel like I went to bed one night in 2008 and woke up on another planet that is in the process of social and political collapse.

  4. The suggestion that I’m not taking level changes into account isn’t correct: I do have levels in mind. My reasoning is as follows: suppose we have two variables growing at approximately the same rate, and initially scaled so t\hat their paths coincide. Then suppose there’s a level drop in one of the two. Following the drop, the growth rate of the variable that dropped approaches it’s old value, while that of the other (un-dropped) series slows. Eventually the levels must again coincide. My questions are: (1) have they perhaps already coincided, and (2) if not, why not (or, why have economic forces not brought about a more substantial decline in the growth rate of the series that didn’t experience a level drop such as would be adequate to restore equilibrium)?

  5. George – That outcome -eventually the levels must coincide – is not true in this case. If you look at the NGDP&Trend chart above, NGDP is growing at a rate which is even slower than the previous one, so the LEVEL has been “permanently” reduced. If wages have only slowed, the levels will never coincide. Either the wage level drops or NGDP growth is sufficiently increased to put it at a higher level path.

    • Marcus, I confess I’m at a loss: that NGDP is growing more slowly than before does not prevent it from catching up to wage growth if the latter has decelerated even more. Any growth in a series raises the level of that series, and eventually raises it enough to more than made up for a one time drop. NDGP has in fact more than made up for the hole to which you refer, hourly compensation now grows at 1.5% to NGDP’s 4+%. The series paths must cross again at some point, as surely as a tree that growths 4 inches per year after having been trimmed by a foot, eventually becomes taller than one which, though never trimmed and originally of the same height as the first, grows less than 2 inches per year.

      • George, I made bad use of the word “never”. But maybe you´ll agree that the “catch-up time” is too long at the rate things are moving (and risks undesirable side-effects)! So I still think that an increase in the NGDP level target is the best solution.

      • Time cannot cure cutting a tree by a foot today, and then by an inch or more every week thereafter. That is the problem with monetary policy that lacks a credible floor and no plan for dealing with undershooting in an environment with all kinds of instability floating around the globe. It foists a near constant state of adjustment on markets and prices and makes all problems far more complex than they need to be. I have not been able to find any provable over all benefit to this, and considering legal obligations of the Federal Reserve, I believe it has the burden of proof unless we want to completely dispense with the rule of law.

  6. Pingback: George Selgin Asks a Question « Uneasy Money

  7. Marcus,

    I would be interested to know how you came up with your trend line to see how it fits on a longer range of GDP data going back to the beginning of your first panel, 1964. Even a graph showing your trend line going back to 1987 would be informative. It does not appear to me that your trend line would fit the GDP data very well but I could be wrong.

    I think your graph of the “Great Inflation” is actually proving George’s point (I hope I am characterizing his point correctly) that more rapid nominal GDP growth would not necessarily result in a reduction in the unemployment rate. Even if nominal wages were tracking nominal spending more closely (as your chart shows during the “Great Inflation”) that would not necessarily mean a further reduction in unemployment. Remember even with nominal wages and spending tracking closely together the annual unemployment rate was 8.5%, 7.7%, and 7.1% in 1975, 76, and 77. It even reached 9.0% in May of ’75. This was a time of rapid nominal GDP growth averaging between 9 and 11 percent, wages and spending tracking closely, and yet high unemployment.

      • Marcus,

        Thanks for the email. I have also run a trend using the log of nominal GDP from 1987 as you have done, except you have forecasted the 1998 the 2012 period. I have run the trend through 2012 and it is coming out quite a bit lower than your forecasted period. The log trend line starting in 1987 and ending in 2012 appears somewhat below what you have drawn as Sumner’s alternate path.

        Looking at the graph you have emailed me, the last time the NGDP was above your forecasted trend was around the first or second quarter of 2001. Even the period of 2003-2005, where you quote John Taylor in your email as saying the interest rates as being “too low for too long” the NGDP is below trend. I happen to agree with that quote too. But I see the Federal Reserve pushing the NGDP above the trend with those actions and creating an unsustainable bubble in economy that was prominently displayed in the housing market. So, I think your forecasted trend is too high. Of course, this will make the gap between the forecast trend and the actual GDP even larger than just the log trend line. Even using the log trend line from 1987, which I don’t necessarily subscribe too, make the shortfall around 8.6%. Using your forecast trend line the shortfall must be substantially higher.

        Well, again thank you for sending your email. It has given me a better idea how you arrived at your trend line.

  8. I agree Marcus, that the catch-up may take even longer than the several years that have already passed. But the question is, as you now seem to agree, how long it is actually taking, which can’t be answered a priori, by appealing to the distinction between level and rate changes.

    Dajeeps’ point takes neither the facts nor my analogy seriously: I mentioned cutting a tree only once, for good reason: the level of NGDP fell only once, though it fell by a large amount.

    • George, as you may know I´m a Brazilian who has had interest in the US economy for quite some time. So my perspective is that from an “outside observer”. In that condition I “state” that the “Great Moderation” is a refelection of very good monetary policy (something I think is completely independent of bubbles and other shenanigans). And the result of that MP was the most stable period in US econ history (and contrary to many, I don´t think “stability” is the cause of “speculation gone mad”). And that stability is embodied in nominal spending evolving along a stable level path. That´s why I insist that trying to regain a “satisfactory” level of spending is the key to getting back a “good economy”.

    • George, I don’t understand what you thought the point of rescaling the graph was. First of all, as Bill Woolsey points out (http://monetaryfreedom-billwoolsey.blogspot.com.au/2012/07/for-discussion-regarding-reflation.html), average hourly wages are supposed to be growing more slowly than NGDP, because of population growth and non-wage compensation. He shows that trend hourly wage growth has been 3% during the Great Moderation, whilst trend NGDP growth at 5.5%. So what we need to do is line up the growth paths of wages and NGDP in order to show the the combination of aggregate demand and aggregate supply growth paths which are consistent with full employment. Therefore, the first graph was correct, as it shows AS and AD moving in tandem to keep prices and output stable until “Bernanke’s Depression”. Now, we have seen the blue line for wage growth tapering down recently. The argument is that this represents a downward shift of the SRAS towards long run equilibrium, and so to that extent the “deficient” nominal spending (which as per Sumnerian convention I use interchangeably with aggregate demand, though I know the difference) matters less.

      So to evaluate whether or not NGDP has “closed the gap”, we both have to evaluate whether NGDP has recovered from the collapse off its growth path in 2008, as well as the extent that the gap has been closed “from the other end”, by a shifting down of wage rigidities or aggregate supply. That part of the analysis was correct. But as far as I can see it simply makes no sense for the argument to rescale the graph such that NGDP growth no longer lines up with wage growth during the full employment period, then point to a pre-2012 “intersection” of NGDP with wages and claim that the gap has been closed. As Marcus points out, appropriate rescaling and translating of the two graphs can get the “intersection” to appear anywhere. The point, which Marcus didn’t quite fully express, is that to evaluate whether AD has caught up with AS, or for that matter whether AS has caught down to AD, you have to measure the decline in wage growth from its growth path, then adjust the NGDP growth path by that amount, and see whether actual NGDP has caught up to that adjusted growth path. This is something like what Marcus shows in his graph, “NGDP & Trend”, which shows that even with favorable assumptions about supply-side adjustment to the lower spending, spending is still at an inadequate level for full employment. I.e. the data clearly suggests that AD currently intersects SRAS behind the LRAS. Now, as to why SRAS hasn’t shifted even further by now at this point, is a separate question, which I still don’t think Scott can adequately explain…

  9. Tom, I think you have to be more judicious. Why run the regression all the way to the present? We KNOW the trend path has dropped. So you get all sorts of distorted results like you do.
    If you run the regression up to 2007.IV the result is for practical purposes identical to running the regressio0n to 1997 and forecasting the future. This is saying, in effect, that during the next 10 years (up to 2007) NGDP remained close to trend

  10. Marcus,

    Yes, the trend path has dropped. But if we didn’t incorporate the drops in NGDP in our trend path the Federal Reserve would overheat the economy by shooting for too high a target NGDP level. I think when you forecast a trend line that puts all of the economic expansion of the 2000s below your forecasted trend then you might want to rethink your forecast. That’s my opinion. I am sure you disagree. But the NBER put the US expansion beginning in December of 2001 ending December 2007 and during that time actual NGDP does not once exceed your forecasted NGDP trend line.

    • Tom – Ha! But you only know the trend path has dropped after the fact (and due to bad MP!).
      During 2000-2005 NGDP first dropped below trend (01-03) and than rose towards trend (03-05), a fact that incites the too low for too long view of MP at the time. But it was the correct policy IF the objective was to bring the economy back to trend!

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.