Monday, February 27, 2017

NAIRU and other connections between inflation and employment

The topic of the hour is NAIRU (Non-Accelerating [1] Inflation Rate of Unemployment). Simon Wren-Lewis put forward the mainstream defense of the concept before I left on vacation. In general, the idea is plausible from a rational agent viewpoint: if labor is plentiful, then there is little pressure on wages. At some level of labor scarcity, the bargaining positions of prospective employees will be better leading (so the theory goes) to expectations of higher wages for the same level of productivity. With those higher wages chasing scarce output, prices could be expected to increase. Of course other "stories" are possible, but the basic idea is that unemployment below a certain level leads to higher inflation expecations, leading to higher inflation.

I think Simon accidentally gives us the real rationale:
But few of these attempts to trash the NAIRU answer a very simple and obvious question - how else do we link the real economy to inflation?
That is to say, it seems to me to be a post hoc narrative developed out of the original desire to link the real economy and inflation (i.e. monetary policy). The problem is that we have two completely unobservable variables interacting: NAIRU and inflation expectations. This should raise eyebrows to say the least.

But as I said: NAIRU appears to be an attempt to understand theory, not data. What does the data look like?

To that end, I thought I'd introduce a new "model": the dynamic equilibrium picture of the price level. I've previously shown how prices in information equilibrium models should also manifest dynamic equilibria subjected to shocks (just like ratios of variables in information equilibrium with each other such as the unemployment rate). Applying the same procedure as in the previous link, we obtain a pretty good first order description of the core PCE price level and inflation:

There's a single transition centered around 1978.7; all of the other shocks are dwarfed by this one. However, this one major shock doesn't really match up with any shocks to unemployment:

What this broad shock does line up with is the non-equilibrium process of women entering the workforce:

This view of the data is exactly the same demographic story of 1970s inflation as the "quantity theory of labor" (as well as Steve Randy Waldman's account in terms of baby boomers and women together) [2].

Let's return to our discussion of NAIRU above. The data more plausibly connects the employment-population ratio and inflation than unemployment. In fact, it does not appear (from the data, at least) that the unemployment rate dynamic equilibrium technically has to stop its downward path -- it could proceed towards zero independent of the inflation rate:

This is of course very unlikely because a recession would almost certainly intervene (happening randomly with a time constant of about 8 years). This is a silly counterfactual, but the main point is that unemployment and inflation do not appear to be connected. Based on the dynamic equilibrium view of the existing data, unemployment could head to zero and inflation would remain constant.

[I would like to add that at some low unemployment rate we might run into a "noise floor" of people changing jobs for non-macroeconomic reasons, as I mention in the presentation where this graph originally appears (28 Feb 2017)]

Therefore, while there is a plausible connection between employment and inflation (changes in the employment-population ratio), it's not the story told by NAIRU (inflation and unemployment).

PS See also this Twitter thread.


Update 27 Feb 2017

If you squint, you might be able to convince yourself of some residual (anti-)correlation between the unemployment rate and the difference between the dynamic equilibrium model and the data:

However, it is not statistically significant. Even if it did exist, the primary component of 70s inflation would still have nothing to do with the unemployment rate.

Update 27 Feb 2017, the second

Added the code to the GitHub dynamic equilibrium repository.



[1] Some Wynne Godley fans seem to think this term is some kind of misnomer, saying "prices" are accelerating. I don't understand it. First off, Godley has explicitly used the inflation acceleration terminology to mean inflation is increasing or decreasing. This is also perfectly in alignment with the terminology in physics: price is a position, inflation is a velocity. We say "velocity accelerates" (for a rate of change of velocity), not "position accelerates".

[2] Now even though women entering the workforce appears to be the best candidate for an explanation of 1970s inflation, that doesn't necessarily nail down the mechanism behind it. Women entering the workforce could have produced a big aggregate demand shock (sort of an AD-AS model explanation). In thinking about the mechanism, I also came up with a sexism theory of inflation: women entering the workforce made men demand to be paid more than women, driving up wages and prices. That is to say the connection between inflation and the employment-population ratio could be entirely sociological (i.e. different in different societies), not economic (i.e. the same for any demographic transition).


  1. Karl Whelan concedes it's a misnomer here:

    1. From link:

      ... then inflation will be increasing, but not accelerating. The price level is what will be accelerating.

      That is also incorrect. Again, position doesn't accelerate, velocity does.

      My Latin is rusty (actually, terrible), but celeritas (related to the root) is speed. It's not locus (position). Ad celero (to speed [up]), not ad loco (to place). The latter becomes the English "allocate" which is entirely different but can be thought of as moving to position.

      The derivative of position (velocity) accelerates (speeds up, gets bigger); likewise the derivative of price (inflation) accelerates (speeds up, gets bigger).

      In any case, the economic definition of acceleration is borrowed from physics (second derivative of price analogous to second derivative of position), so the physics usage should be the correct one.


    2. "position doesn't accelerate, velocity does."

      Well, velocity doesn't accelerate either! It's a particle or a thing that accelerates.

      But there's a difference from the physical world. This is because in physical systems, there's a thing which has position, velocity, acceleration etc. But here it's the variable itself. So price rising, price accelerating etc.

    3. That still wouldn't make sense of the usage because price is the observable function of time p(t) that corresponds to position S(t) in this analogy. As S(t) is a property of some object, p(t) is the property of some object -- like a granola bar.

      Saying "the granola bar accelerates" to mean that the price of the granola bar is more than just linearly increasing is a strange phrase indeed.

      In any case, the "accelerating velocity" construction is good enough for government work (e.g. here -- first example in Google search).

      In any case, as I said above: I do not get this. Say I cede that price accelerates. What have we accomplished? We're still talking about same thing: d^2p(t)dt^2. What have we learned? We change the acronym to NAPLU. How is this not a superficial change?

    4. "This is because in physical systems, there's a thing which has position, velocity, acceleration etc. But here it's the variable itself. So price rising, price accelerating etc."
      I am just pedant and i do not have any content issue, but even though neoclassic economics is a pseudo science when it is borrowing metaphors from hard science it should respect the logic and rules.
      Price can be seen as a property, an height and its variation along time can be described using derivatives.

  2. I'm thrilled to learn there is no connection between unemployment and inflation. As I've been saying to NAIRU bashers for 20 years and as Simon Wren-Lewis recently said: in that case why don't we have a massive increase in demand and reduce unemployment to near zero?

    NAIRU bashers invariably go quiet when asked that question. And the reason is stark staring obvious: they know perfectly well that THERE IS a relationship between inflation and unemployment!!!

    1. There does not appear to be an empirical relationship between inflation and the unemployment rate in the data (see above); there could potentially be a relationship between inflation and the employment-population ratio (the latter isn't ruled out by empirical data).

    2. Then why don't you advocate a massive increase in demand? Think of all those poor unemployed people for God's sake: millions of them would get jobs. Don't you care about them? I know why you don't advocate a big increase in demand, but you won't admit to it. You know perfectly well inflation would go thru the roof.

    3. Again, inflation does not appear to have an empirical link to unemployment in the data (see above), your data-free assertions notwithstanding.

      It is plausible that inflation could have a link with the employment population ratio.

      And I would advocate a massive increase in demand. The data seems to show the only way to do this is through increasing population and/or capital (here) or increasing the employment-population ratio (per above).

      However, the unemployment rate will continue to fluctuate based on exogenous shocks regardless of the path of demand (at least as can be seen in the data).

  3. I can't imagine why a NAURU basher wouldn't have a response to that. We don't address the demand deficiency because that would lead to wage increases and a greater share of productivity going to labor.

    The NAIRU is a weapon of mass destruction on the battlefield of class warfare, a tool of systemic wage suppression. Any person of integrity should know that we don't need the central bank to maintain a "targeted minimum level of unemployment" to "protect" us from inflation. That anyone thinks that is an acceptable policy tool merely serves to illustrate the manifest moral bankruptcy of the established paradigm.

    If inflation is an important enough peril to "protect" us from, then surely we can find a more suitable protection than using the unemployed as human shields. That no one has done so in decades should speak clearly enough as to the priorities involved here - the welfare of the working class is to be sacrificed to protect price stability on the basis of nonexistent evidence.

    1. Where's the evidence that labor's share rises in booms? Looks to me from this chart that, if anything, labour's share peaks in recessions.

      I fully accept that using unemployment to control inflation is cruel. If you have any better and workable suggestions you'll get several Nobel Prizes. JG is one option, I've read and written a vast amount about that. My impression is that the most vociferous advocates of the idea are clueless. They're just virtue signalling loudmouths

    2. "I fully accept that using unemployment to control inflation is cruel. If you have any better and workable suggestions you'll get several Nobel Prizes."

      The point is that the imaginary theory has not much to do with real world, leaving aside that the unemployed would stay voluntarily unemployed, it is the "rational" inflation expectation that would operate and induce the raising of monetary prices by capitalists or producers, which would in turn discourage the labor offer. Nairu just adds a "coherent" non sense to a non sense tale.
      It is not workers requesting higher nominal wages or more probably workers offered higher nominal wages for leaving other employers to cause inflation directly.

    3. A better, more workable suggestion would be to not target inflation at all if that's the only way to do it. Prior to Milton Friedman and this NAIRU garbage wages *consistently* grew faster than inflation. Ever since NAIRU was instituted, growth in the real wage stopped altogether even as productivity and profits continued to grow. The scaremongering about inflation might have served *someone's* interests, but those people are in the minority.

    4. David,

      I don't know how old you are, but inflation was really considered problematic by ordinary people in the US in the 1970s and 80s. For one thing, it pushes up nominal interest rates making mortgages more expensive.

      Now that goes for 10% inflation (although in reality it's more complicated); we could probably get along just fine with a 4% inflation target rather than 2%.

      But then, per the data analysis above, the high inflation may have had little to do with government policy and was instead about social change (women entering the workforce). It's not so much that low inflation benefits someone as we've reached a new *social* equilibrium level of the employment-population ratio.

      This view turns the whole monetary policy debates in macroeconomics in the 1960s and 70s into something of a farce: the Milton Friedmans and Paul Volckers working diligently on theory and practice for controlling something over which they had no control.

      I actually find it kind of funny.

    5. I would say that the OPEC oil embargo should receive a good portion of the blame for 70s inflation, along with the adjustment from leaving the gold standard. Low unemployment was certainly not to blame, as it spiked to 6% around '70, briefly dipped below 5% then did not again fall below 6% until the late 80s.

      The IMF has recently stated that it would be better to run 10% inflation than to allow a 1% drop in GDP. Sounds reasonable, since no one can demonstrate that the anti-inflation fetish has helped the economy or wages.

    6. The oil shocks appear to be very narrow fluctuations on top of the trend due to demographics:

      And in the analysis above, they appear to add only a few percentage points to the demographic trend in 1973 and 1979.

  4. "But as I said: NAIRU appears to be an attempt to understand theory, not data."

    That is the recursive attitude in a pseudo science like neoclassic economic theory, which is essentially an ideology and class conflict tool.

    When the rate of accumulation is high for a series of reasons including a vibrant effective demand, it is likely that workers could gain power for getting higher wages in particular when they are well organized: wage increases can be translated into price increases specially if production has reached full capacity. In that case any wage increase would represent inflation, and, assuming a control of hyperinflation, only more mechanization and technological progress would be the solution.

    The nairu is an inconsistent fabrication which forces a relation and causality between only two variables, leaving aside most important factors related to accumulative process and effective demand. Its political contribution is to help to keep low wages and low effective demand in presence of constant unemployment, hence it includes even a racist component.
    Finally, it should be mentioned that in a high oligopolistic and monopolistic contest like the contemporary one price increase probably would traduce a different kind of power.