When I was in London recently, I was repeatedly assailed with the idea that the…
Comments made last week by the former Clinton, Obama and now Biden economist Lawrence Summers contesting whether it was sensible for the US government to provide a $US2,000 once-off, means-tested payments was met with widespread derision and ridicule from progressive commentators. There were Tweets about eviction rates, bankruptcy rates, poverty rates, and more asserting that the widespread social problems in the US clearly meant that Summers was wrong and a monster parading as a progressive voice in the US debate. I didn’t see one response that really addressed the points Summers was making. They were mostly addressing a different point. In fact, the Summers statement makes for an excellent educational case study in how to conduct macroeconomic reasoning and how we need to carefully distinguish macro considerations from distributional considerations, even though the two are inextricably linked, a link that mainstream macroeconomics has long ignored. So while Summers might have been correct on the macro issues (we will see) he certainly wasn’t voicing progressive concern about the distributional issues and should not be part of the in-coming Administration. This is Part 1 of a two-part analysis. In Part 2 we will do some sums. In this part, we will build the conceptual base.
Two levels of analysis involved
Laurence Summers was interviewed by Bloomberg (December 24, 2020) – $2,000 Stimulus Checks Don’t Make Sense, Says Larry Summers.
I wrote about Summers in this blog post – Being shamed and disgraced is not enough (December 18, 2009).
My assessment of him has only deteriorated since.
He will occupy a senior economics position in the incoming US administration.
Here is the – (edited) transcript of the interview from Bloomberg:
… I don’t think the two thousand dollar checks make much sense. The real issue is going to be sustaining this expansion. Think about it. The nine hundred and eight stimulus bill probably would pay out two hundred two hundred and fifty billion dollars a month for the next three months.
The level of compensation is running about 30 billion dollars a month below what we would have expected it would. GDP is running about 70 billion dollars a month below what we would have expected it would. So in a way that’s quite unprecedented. We have stimulus already much more than filling out a hole.
And given that lots of a hole is from the fact not that people don’t want to spend but that they can’t spend because they can’t take a flight and they can’t go to a restaurant.
I don’t necessarily think that the priority should be on promoting consumer spending beyond where we are now. So I’m not even sure that I’m so enthusiastic about the six hundred dollar checks. And I think taking them to two thousand dollars would actually be a pretty serious mistake that would risk a temporary overheat.
I would like to see more assistance to state and local governments. I would like to see more money put into testing more money put into accelerating of vaccines.
But gosh David I think it would be a real mistake to be going to two thousand dollars. And I have to say that when you see the two extremes agreeing you can almost be certain that something crazy is in the air. And so when I see a correlation of Josh and Bernie Sanders and Donald Trump getting in behind that idea I think that’s time to run for cover.
And I know that many of my fellow Keynesians who believe in fiscal stimulus will likely be in favor of this. But sometimes there can be too much too poorly designed of even a basically good idea. And that’s my reaction to two thousand dollar stimulus …
Almost as soon as the interview was made public, progressives went wild.
They pointed out the harm done by COVID to workers and their families, the impending rent eviction maelstrom, the rising mortgage defaults, and all sorts of other social pathologies that are threatening social stability in the US and have been increasing over the last several decades.
The problem is that these responses didn’t really address the issue that Summers was making.
The critics were largely constructing their claims in terms of Summers being a heartless so-and-so who didn’t have a progressive bone in his body and was just reverting to form (of the type I discussed in the blog post linked to above).
I have sympathy with that assessment of the man.
But in doing, the critics failed to contest the actual point he was making – a macroeconomic point.
There were thus two levels of analysis at work:
1. A macro analysis of output gaps (Summers).
2. A distributional assessment of how existing resources and income is distributed in the US (the critics)
Summers might be correct on (1) but sadly lacking on (2).
We need to carefully separate the arguments.
Lawrence Summers was making a macroeconomic point that simply is this.
1. We define potential output where all productive resources are fully employed and no further output can be produced
2. An output gap forms when the current level of spending in the economy drives output below the potential level. This is an indication of how much extra spending is required, taking into account the multiplier effets that follow spending injections, to move the economy back to potential.
3. Driving spending growth beyond that full employment capacity will probably introduce demand-pull inflationary pressures as firms exhaust their capacity to produce and respond to the increasing spending demand by pushing up prices.
So Summers believed that the $US900 stimulus alone would threaten that limit.
Adding another $US1,400 to the cash handout would, in his view, push the economy past the limit.
That assertion is empirically testable (see below) although the frameworks we use to conduct that sort of analysis is ridden with conceptual and measurement problems which means that we are doing art rather than science.
But within the notion of ‘ball park’ estimates, we can make some statements that bear on whether Summers was correct in his analysis.
However, and this is the important point – there is no guarantee that at full capacity, the well-being of the people, the fortunes of the least privileged, the scope and quality of government services, etc – will be at any levels considered desirable.
At the extreme, a nation could be producing all sort of military equipment and all the workers and productive capital fully employed, but with the majority living in abject poverty.
The output gap might be zero – but the society is devastated by poverty and disease.
So within that spending room, there is a debate that to be had about what initiatives are best – so Summers favoured health care and assistance to sub-federal governments rather than a $US2,000 cash handout.
I found that part of his assessment disclosed his own preferences, which progressives might find rather offensive, but doesn’t make him wrong.
So here are some deeper considerations.
The NAIRU redux
The following blog posts cover past writing on the NAIRU:
1. The NAIRU/Output gap scam reprise (February 27, 2019).
2. The NAIRU/Output gap scam (February 26, 2019).
3. No coherent evidence of a rising US NAIRU (December 10, 2013).
4. Why we have to learn about the NAIRU (and reject it) (November 19, 2013).
5. Why did unemployment and inflation fall in the 1990s? (October 3, 2013).
6. NAIRU mantra prevents good macroeconomic policy (November 19, 2010).
7. The dreaded NAIRU is still about! (April 6, 2009).
I also have written books, PhD theses and many Op Eds about this topic.
Here is a summary of the Non-NAIRU facts which taken together provide strong evidence against the dynamics implied by the NAIRU approach:
- Unemployment rates exhibit high degrees of persistence to shocks.
- The dynamics of the unemployment rate exhibit sharp asymmetries over the economic cycle. The unemployment rate rises quickly and sharply when overall spending (demand) contracts but persists and falls slowly when expansion occurs.
- Inflation dynamics do not seem to accord with those specified in the NAIRU hypothesis.
- The constant NAIRU concept (the earliest form of the assertion) was abandoned and replaced by so-called Time-varying-NAIRUs, which became just another ad hoc fudge to try to get the concept to fit the data.
- All NAIRU estimates have large standard errors, which make them all but meaningless for policy analysis. The majority of econometric models developed to estimate the NAIRU are misspecified and deliver very inaccurate estimates of the NAIRU. Most of the research output confidently asserted that the NAIRU had changed over time but very few authors dared to publish the confidence intervals around their point estimates.
- Estimates of steady-state unemployment rates are cyclically-sensitive (hysteretic) and thus the previously eschewed use of fiscal and monetary policy to attenuate the rise in unemployment has no conceptual foundation.
- There is no clear correlation between changes in the inflation rate and the level of unemployment, such that inflation rises and falls at many different unemployment rates without any systematic relationship evident.
- The use of univariate filters (Hodrick-Prescott filters) with no economic content and Kalman Filters with little or no economic content has rendered the NAIRU concept relatively arbitrary. Kalman Filter estimates are extremely sensitive to underlying assumptions about the variance components in the measurement and state equations. Small signal to noise ratio changes can have major impacts on the measurement of the NAIRU. Spline estimation is similarly arbitrary in the choice of knots and the order of the polynomials.
- In the end, the NAIRU estimates are just some smoothed trend of the actual unemployment rate and provide no additional informational content.
Consider the following graph, which uses data from the March-quarter 1950 to the December-quarter 2020 (the last observation is the average for October and November 2020).
The blue line is the is the gap between the actual unemployment rate (UR) and the CBO estimate of the NAIRU (long-term) while the grey bars are the change in the inflation rate.
We would expect, if the NAIRU framework provided any predictive capacity, that when the gap was positive, that is, the actual unemployment rate is above the NAIRU, that the inflation rate would be falling and vice versa.
There might be some lags in this relationship but we should be able to detect a fairly systematic relationship over time.
There is no such relationship.
Another way of seeing this more clearly is to consult the next graph which uses the same data but with the horizontal axis depicting the gap between the actual unemployment rate (UR) and the CBO estimate of the NAIRU (long-term) and the vertical axis taking the value of 1 if the inflation rate is falling and zero if it is rising.
We should expect the zeros to be mostly to the right of the zero line on the horizontal axis and the ones to be to the left of the zero line.
The distribution of the observations is nothing like that.
The output gap redux
I started with the NAIRU because it has been widely used to define full capacity utilisation. If the economy is running an unemployment equal to the estimated NAIRU then mainstream economists conclude that the economy is at full capacity.
Of-course, they kept changing their estimates of the NAIRU which were in turn accompanied by huge standard errors. These error bands in the estimates meant their calculated NAIRUs might vary between 3 and 13 per cent in some studies which made the concept useless for policy purposes.
But they still persist in using it because it carries the ideological weight – the neo-liberal attack on government intervention.
But the point here is that the NAIRU estimates are tied in with estimates of the Output Gap, which is the difference between potential and actual GDP at any point in time.
As I note in this blog – Structural deficits and automatic stabilisers (November 29, 2009) – the problem is that the estimates of output gaps are extremely sensitive to the methodology employed.
It is clear that the typical methods used to estimate the unobservable Potential GDP reflect ideological conceptions of the macroeconomy, which are problematic when confronted with the empirical reality.
For example, on Page 3 of the US Congressional Budget Office document – Measuring the Effects of the Business Cycle on the Federal Budget – we read:
… different estimates of potential GDP will produce different estimates of the size of the cyclically adjusted deficit or surplus …
The CBO is representative in the way they seek to estimate Potential GDP.
They explain their methodology in this document.
Effectively, the estimated NAIRU is front and centre, so Potential GDP becomes the level of GDP where the unemployment rate equals some estimated NAIRU.
It becomes a self-serving circularity.
Potential GDP is not to be taken as being the output achieved when there is full employment.
Rather, it is the output that would be forthcoming at the unobservable NAIRU. If the estimates of the NAIRU are flawed, then so will the result output gap measures.
The problem is that policy makers make assessments of their current fiscal position based on these artificial, (assumed) cyclically invariant benchmarks.
And, because the estimates of the NAIRU are typically ‘inflated’ (well above what the true full employment unemployment rates are), the conclusion is always that the current discretionary fiscal policy stance is too expansionary (because their methods understate the cyclical component).
Which then means that in recessions, the output gap estimates will be too small, and the extra net government spending that is advocated will subsequently also be too small.
As we saw during the GFC, the more extreme misuse of these under-estimated output gaps provokes the austerity bias – cutting discretionary spending or increasing taxes (mostly the former) when, the reality of the situation is usually indicating the opposite is required.
New Keynesians talk of ‘deficit biases’, when in fact their framework leads to ‘austerity biases’ and elevated levels of mass unemployment and resulting income (production) losses persisting for long periods.
The first problem then is assessing what the output gap actually is.
In Part 2, some arithmetic will be forthcoming.
That is enough for today!
(c) Copyright 2020 William Mitchell. All Rights Reserved.