Wednesday, 19 February 2014

Mitchell and Wray’s flawed ideas on the Job Guarantee.

Bill Mitchell and Randy Wray are producing a new text book: Modern Monetary Theory and Practice”. Chapter 18 deals with what they call “Job Guarantee”. JG is just a name a very simple idea which has been around for centuries (not that I’m condemning the idea for its age or simplicity). It’s an idea you’ll have heard dozens of times, namely that there are an infinite number of relatively simple jobs which government could create for the unemployed: environmental clean ups, helping pensioners with their gardens, etc. Pay, of course, consists of the unemployment benefit that the unemployed are getting anyway, or a wage of about that level. Weakness in the authors’ treatment of this subject are as follows.
1. The authors engage in a practice common amongst academics, namely using pseudo technical langue where such language is not necessary.
2. An example of this is the song and dance they make about their claim that the unemployed can be compared to a “buffer stock”. Now while there are obvious similarities between the “stock” of unemployed and buffer stocks of wheat, crude oil, etc, there are important ways in which the analogy does not work. That’s why about 95% of economists, quite rightly, do not bother with the buffer stock analogy.
3. As just mentioned, the JG idea is as old as the stars and has been tried numerous times and then abandoned. But the authors don’t make a single reference to this fact. Anyone with a grain of common sense knows that if something has been tried before, it’s a good idea to look at those attempts to see what can be learned. Indeed, there is plenty of empirical evidence as to what works and what doesn’t when it comes to JG type schemes. The evidence does not seem to be of much interest to the authors.
4. JG employees can be allocated to “specially set up employers” as was the case with the WPA, or to EXISTING employers. There is no analysis of which of those two options is the better. Indeed, the second option has been the one adopted in the UK over the last ten years at least. You wonder whether Mitchell and Wray know what is going on in the world around them.

It’s been tried before.
As almost anyone over about 50 years of age knows, the JG idea has been implemented and then abandoned dozens of times since WWII in the West. For example in Britain alone we had the “Job Creation Scheme” and the “Community Programme” about 40 years ago, and more recently the “New Deal”, the “Future Jobs Fund” and currently there is a variation on the JG idea up and running: the “Work Programme”.
However, Mitchell and Wray don’t seem to be aware that the idea has been tried before: dozens if not hundreds of times. And even more hilarious is the fact that Mitchell seems to think he invented the JG idea. At least in section 13.2 of this work (1) he says, “The JG proposal was conceived independently by Mitchell (1998) and Mosler (1997-98)”.
In fact the idea is a good 2,500 years old! It was implemented in Ancient Athens. Plutarch says “Pericles undertook vast projects of buildings and designs of work it being his desire and design that the undisciplined and mechanic multitude….should not go without their share of public salaries, and yet should not have them for sitting still and doing nothing”. (I got that from “Unemployment in History” by J.A.Garraty, Ch2, p.13 ).
Given that Ch.18 is a good 20,000 words in length, you’d think a few sentences on the history of the JG idea might be in order. But more than that: there is VERY STRONG case for looking IN DETAIL at previous attempts to implement any idea: else you’re liable to end up re-inventing the wheel and repeating others’ mistakes.
But doubt less Mitchell and Wray are perfectly well aware of the history. As to why they keep quiet about it, that’s a puzzle. Perhaps it’s so as to make their ideas look more original. Or perhaps publicising the fact that the idea has been tried and abandoned dozens of times would make it look as though the idea is useless (which I don’t think it is).
In short, I’m not opposed to the JG idea, but hopefully Mitchell and Wray will have no influence on HOW the idea is implemented.

The details.
Anyway, let’s now deal with some of the specific weaknesses in Mitchell and Wray’s Ch.18. The chapter is divided into four sections, which I’ll take in order (though some ideas or proposals in Ch.18 are spread over different sections, thus not all the ideas in Ch.18 are not taken in STRICT order.)

Section 18.1  - Buffer stocks.
On the first page of Ch.18 the authors set out two basically different ways of controlling inflation: 1, a “buffer stock” of unemployed people, and 2, a “buffer stock” of people doing JG. And a central characteristic of No.2 is allegedly that “government exploits the fiscal power embodied in a fiat-currency issuing system…”.
Now quite what the fiat currency versus commodity currency debate has to with JG is a mystery. The authors don’t explain. At least the word “fiat” does not appear anywhere in the rest of the chapter. Plus there is no reason JG would not be feasible in gold standard regime, i.e. a non-fiat regime. Indeed, various countries in the 1930s DID implement JG schemes while on the gold standard. But never mind: the phrase “exploiting the fiscal power embodied in a fiat-currency issuing system” sounds good, doesn’t it? And that’s just the first bit of pseudo technical waffle in Ch.18.

Buffer stocks: the non-insight of the century.
As to “buffer stocks”, the phrase appears about fifty times in the Ch.8, and Mitchell and Wray’s buffer stock idea is presented as some sort of important insight into the workings of labour markets. In fact it’s the non-insight of the century, and for the following reasons.
The big idea is that the unemployed (and/or JG employees) act in a similar way to buffer stocks of physical commodities like crude oil or wheat. That is for example, when the price of a physical commodity rises, some of the buffer stock can be sold onto the market, and that ameliorates the price rise. And likewise, given rising aggregate demand, the price of labour would tend to rise, were it not for a “buffer stock” of unemployed who apply for jobs, or who, as it were, can be sold onto the market, to meet employers’ need for more labour.
Now there must be a good ten thousand economists over the last century who have cottoned onto the fact that unemployment ameliorates inflation, and thus that the unemployed are a buffer stock of a sort.  But none have seen a need to use the buffer stock analogy. Nor do I.
And the reason is that while the analogy holds in some ways, it does not hold in others. For example, giving rising aggregate demand it’s pretty obvious that a decent supply of skilled labour from the ranks of the unemployed will ameliorate wage increases.
However, given DECLINING demand, wages don’t fall significantly: but that’s not purely or even mainly because of the “buffer stock”. It’s because of something that Keynes pointed to long ago, namely that “wages are sticky downwards”. That is, if you try cutting nominal wages, you tend to get riots or strikes (e.g. the year-long miners’ strike in Britain in 1926).
Indeed, in the 1800s, long before the “unemployment benefit buffer stock”, wages did not fall precipitously in recessions.

Every grocery store has “buffer stocks”.
Mitchell and Wray will perhaps be amazed, if they walk into their local supermarket, to see large quantities of tinned fruit and other stuff which is not at any particular instant being bought. That is, shops, supermarkets and indeed almost every business holds “stocks” of goods so as to deal with temporary upsurges in demand. And those businesses do not bother with the fancy phrase “buffer stock”. They simply use the word “stock”.

Each unemployed person is different.
Another flaw in the buffer stock idea is that while each unit of a physical buffer stock (like crude oil or wheat) is more or less identical, that is certainly not the case with the unemployed. That is, each unemployed person is different. And that’s of particular relevance when unemployment is relatively low, and for the following reasons.
When a physical buffer stock is on the low side, the sale of some of the stock onto the market certainly has an effect. That is, the sale reduces prices or ameliorates price increases.
In contrast, when UNEMPLOYMENT is on the low side, the QUALITY of labour available from the ranks of the unemployed is relatively low. That is, specific types of labour or skills that employers want tend to be hard to find or impossible to find amongst the unemployed. Thus the availability of that labour to employers is largely useless so far as ameliorating increases in the price of labour goes. Indeed, that explains why inflation kicks in in a serious way BEFORE unemployment drops to zero!!!!!!!
To summarise, the buffer stock analogy is a very poor one. It’s one huge red herring, which, to repeat, explains why about 95% of economists don’t bother with it. It’s just pseudo technical waffle.

Section 18.2 – Inflation.
On page 4, the authors claim that Western countries became obsessed about inflation after about 1975, and unemployment rose in consequence. Actually inflation was higher in the three decades AFTER 1975 than in the three decades BEFORE 1975 (at least in the UK). Thus the latter claim by Mitchell and Wray is very debatable. It would be more accurate to say the inflation / unemployment relationship deteriorated after 1975.
The reality is that in the decades if not centuries before 1975, there was plenty of concern about inflation.
E.g. Keynes himself who died in 1946) certainly did not advocate aggregate demand raising policies with a total disregard for inflation.
Likewise, Beveridge who died in 1963) was concerned about inflation.

Section 18.3 – Inflation targeting.
This section starts, “Under inflation targeting monetary policy regimes…”. And that’s the start of a bizarre use of the phrase “inflation targeting”. The phrase is used as a synonym for a regime where unemployment rather JG is used to control inflation  - or “target inflation” if you like. Noticed the self-contradiction? It’s that when JG is used to control inflation, there’s just as much “inflation targeting” as when the unemployed are used to control inflation. But there’s no need for me to point to the self-contradiction: Mitchell and Wray contradict themselves very eloquently on p.14 (section 18.4).
That is on p.14, the authors introduce a new idea called “NAIBER” (Non Accelerating Inflation Buffer Employment Ratio). And that’s simply the equivalent of NAIRU in a regime where all the unemployed do JG jobs. That is, NAIBER is the minimum feasible number of JG people before inflation kicks in in a serious way. So under a regime where there are no unemployed because everyone is on JG work, INFLATION IS STILL TARGETED!!!! The very term NAIBER says so: the “I” stands for what?  It stands for INFLATION!!!
Moreover, on page 2 the authors themselves say in reference to the two types of buffer stock (the unemployed vis a vis JG people), “The two approaches to inflation control both introduce so-called inflation anchors.” Well quite! Inflation is “targeted” in both cases!
In short, the idea that a system under which just the unemployed are used to control inflation is an “inflation targeting” regime, while a system under which all the unemployed are replace by JG people is not an “inflation targeting” regime is complete nonsense. In both cases, government minimises the number of unemployed or JG people in as far as that’s possible without sparking off excess inflation.

Under a Mitchell and Wray’s JG scheme the pay for JG work is the same as for regular minimum wage work, which in turn is not vastly different to what some people get on unemployment benefits (particularly if they have kids). So you might be wondering how Mitchell and Wray reach their assumption that ALL the unemployed will do JG work, rather than claim benefits. After all, if the pay you get on unemployment benefit is not much different to what you get on minimum wage work, there’s not much incentive to go out to work, especially given that living on benefits gives you a life of leisure.
Well the answer is that their system is one huge workfare scheme (see p.12). That is, they say “We assume for the moment that the Job Guarantee policy does not offer an unemployment benefit and that most displaced workers will prefer a JG position over wait unemployment. These assumptions serve to simplify the analysis and relaxing them does not alter the basic dynamics of the system.”
Now I don’t oppose a workfare element in JG systems, but I object to muddled thinking. In particular, the claim that relaxing the workfare assumption “does not alter the basic dynamics of the system” is a joke.
That is, there is a HUGE DIFFERENCE between a JG system that is purely voluntary and one that is compulsory. And pseudo technical waffle like “alter the dynamics of the system” are not an answer to that point.
Moreover, if there is really so little difference between a voluntary and compulsory JG system, why make it compulsory? Abandoning the compulsion means you can avoid being accused of advocating workfare.

Section 18.4 – no unemployment before 1970??
This section starts with the bizarre claim that “Between 1945 and the mid-1970s, western governments realised that with deficit spending supplementing private demand, they could ensure that all workers who wanted to work could find jobs.”
So between 1945 and mid-1970s there was no unemployment? Or put another way, all unemployment was voluntary (i.e. not unemployment at all)?

The JG wage.
This section starts with the odd claim that “While it is preferable to avoid disturbing the private sector wage structure when the JG is introduced, a case can be made to offer the JG wage at a level higher than the existing private minimum if it is thought that productivity is too low in the economy.”
Well wait a moment. If the existing minimum wage is $X/wk and JG wage is $X+Y/wk and the, then $X+Y/wk becomes the de facto minimum wage. That is, people will tend to move from regular min wage work to JG work. But what d’yer know: at the end of the previous paragraph, the authors say “Since the JG wage is open to everyone, it will functionally become the national minimum wage.” Yes, quite: just another example of muddled thinking.

Anyway, having established that pay for JG work should be the same as the national minimum wage, the authors then claim that means JG work does not compete for labour with regular work.
As they put it, “main principle of a buffer stock scheme like the JG is straightforward it buys off the bottom (at zero bid) and cannot put pressure on prices that are above this floor..”
That is plain nonsense: if someone starts growing apples and offers them at the same price as existing apple growers, the new entrant is bound to attract SOME CUSTOM from existing apple growers.
Moreover, in that people are motivated JUST BY pay, and given that Mitchell and Wray’s JG system offers higher pay than that obtainable on unemployment benefit, their system certainly will “put pressure on” the price of labour to a greater extent than an “unemployment benefit only” regime. That is, if you can get $A/wk on benefits and you are then offered $(A+B) on JG, that reduces the RELATIVE ATTRACTIONS of regular work, doesn’t it? In short, Mitchell and Wray’s claim that their system does not compete with regular low paid work just isn't true.
However, it would be possible to have a JG scheme where there was no competition at all with regular low paid work: if JG people were paid the same as on benefits (and forced by the workfare sanction to do the relevant work) that would involve no competition. But even that is not the only condition required for a JG scheme to be a net creator of jobs. Another condition (at least on the face of it) is that the output of JG schemes is given away rather than sold. Many JG advocates, Mitchell and Wray included, seem to implicitly go along with that point, though most of them (M&W included) don’t spell it out EXPLICITLY. That is M&W advocate a JG system which produces public sector stuff, rather than stuff that is sold.
Thus “hiring off the bottom” is not the only condition required for a JG system to work. To repeat, a second condition would seem to be that JG output is given away rather than sold because any sales would require an increase in demand, and assuming unemployment is at the minimum that is commensurate with avoiding excess inflation, an increase in demand is not possible. (At least, that second condition WOULD SEEM to be necessary. Actually it’s more complicated than that, as I’ll explain below.)

Agricultural buffer stocks.
On p.13 there is a box containing about 1,000 words and entitled “Advanced Material”. Gosh: something intellectually challenging? In fact all the 1,000 words do is explain in detail how agricultural buffer stocks work presumably in case anyone doesn’t understand what a buffer stock is.
Strikes me the buffer stock idea can be explained in about 60 words. Here goes: a buffer stock is a stock of some commodity often owned by government, which is maintained with a view to moderating the price volatility of the commodity concerned. That is done by selling some of the buffer stock onto the market when prices are thought to be too high, and buying from the market when prices are thought to be too low. Everyone understand that? It’s not difficult is it?
The end of “Advanced material” box we’re told, “The JG policy is an example of storage for use where the “reserve is established to meet a future need which experience has taught us is likely to develop”.
Well hang on… there is no “reserve” that is “established to meet a future need..”. The reserve is there ANYWAY!!!!! I.e. even if there’s no JG scheme (as indeed the authors themselves point out) the UNEMPLOYED are a buffer stock of “stored material”. So JG is not an example of a buffer stock which is “established to meet a future need…”.
All of which confirms my above point that the whole buffer stock idea is a waste of ink and paper.

Extra demand and inflation.
Given that Mitchell and Wray’s JG system involves higher pay than is obtainable on benefits, plus the fact that JG schemes involve the purchase of materials, capital equipment, etc the effect of introducing a JG system is to raise aggregate demand. And assuming demand is at the maximum that is commensurate with avoiding excess inflation, obviously the introduction of JG would be inflationary. So how do the authors deal with that? Well they make the following claim (p.18).
“Rising demand per se does not necessarily invoke inflationary pressures, because by definition the extra liquidity is satisfying a net savings desire by the private domestic sector.”
Now that sentence is complete nonsense and for the following reasons. If before the introduction of JG the deficit is such that it exactly satisfies savings desires and brings a level of employment that is as high as is possible without excess inflation, then extra demand stemming from JG (or indeed extra demand stemming from any other source) is guaranteed to be inflationary. As to unmet savings desires, there aren’t any (on the above assumptions). Or put another way, there is no reason to assume that just because a “demand increasing” JG system is introduced, that therefor savings desires rise.
But…. if by an chance savings desires DO RISE, then that swallows up or negates the increase in demand!!
However the authors are actually right to say that increased demand stemming from a JG system might not be inflationary, but they’ve got the reasons wrong. That is, as intimated above, their savings desire argument does not hold. However, the workfare element (and perhaps other elements) in their JG system has a NAIRU reducing effect, or inflation reducing effect if you like, and that would obviously to a greater or lesser extent deal with extra demand stemming from a JG system.

Demand constrained economies.
In the next paragraph, and with a view to supporting their claim that the extra demand stemming from JG won’t exacerbate inflation, the authors claim: “Additionally, in demand constrained economies, firms are likely to increase capacity utilisation to meet the higher sales volumes rather than risk losing market share by increasing prices.”
Well of course, but in “demand constrained economies” JG is not the best solution to unemployment (not that I’m opposed to a JG scheme where there is inadequate demand). The best solution is a straight rise in demand!!! So that point of Mitchell and Wray’s is of little relevance.
Put that another way, given a demand constrained economy, the demand increasing characteristics of JG are nothing to shout about because there is a better way of raising demand: standard monetary or fiscal stimulus. Or put it a third way, JG really comes into its own where demand cannot be raised by conventional policies because of the threat of inflation. Or put it a fourth way, really useful role for JG is so to speak to “cheat NAIRU”: that is, reduce unemployment to below the NAIRU level.

Hysteresis, at least when the word is used in reference to unemployment, refers to the fact that unemployment generates more unemployment because unemployment involves loss of skills.
In the section entitled “Would The NAIBER Be Higher Than The NAIRU?”, Mitchell and Wray make much of the skill retaining or skill enhancing characteristics of JG.
Unfortunately the EMPIRICAL EVIDENCE (which does not seem to interest Mitchell and Wray very much) is that JG type schemes do not maintain employability nearly as well as schemes under which those concerned are subsidised into work with existing employers (private sector employers in particular). For some of the evidence on this, see here and here.

Other matters.
I’ll now make some GENERAL POINTS about Mitchell and Wray’s Ch.18: points not related to SPECIFIC passages in their work. And let’s start with the fact that they ignore EXISTING JG schemes. That omission is a bit odd: to illustrate, people who write about cars normally say a fair amount about cars currently in existence or currently being produced.
Anyway, there’s a glaring difference between Mitchell and Wray’s JG proposals and JG schemes currently or recently in operation (at least in the UK). That is schemes currently and recently in operation have allocated JG employees to EXISTING EMPLOYERS rather than to SPECIALLY SET UP employers as was the case under the WPA in America in the 1930s, and as is envisaged by Mitchell and Wray. The UK’s Work Programme and Future Jobs Fund were two examples. But Mitchell and Wray given no reasons as to why the option they favour is better than the “existing employer” option. Mitchell and Wray are almost in a dream world which has limited relevance to the world around them.
And as a result of their failure to address the “existing employer versus specially set up employer” question, there is no analysis of the question as to whether to allocate JG employees to just public sector employers and private sector ones as well (as per the UK’s Work Programme).

Coming up in a day or two: an analysis of the “existing employer versus specially set up employer” question.


  1. Good post.
    M & W 's draft textbook needs substantial polishing to be a an effective textbook for students.

    Plutarch's account of Pericles' "vast projects of buildings and designs of work" can be found in pages 49-50 of:
    According to Plutarch, Pericles had vast public works programme designed in part to provide employment and incomes.
    Is this the same as a workfare scheme?
    And in the absence of any guarantees, is this a JG/ELR scheme?

    1. Hi KK,

      Obviously it's impossible to know exactly what Pericles's system consisted of. But from the short quote I gave above, it certainly seems to have JG written all over it, and with an element of workfare thrown in: i.e. "your benefit is conditional on doing some work".

  2. Thank you for this post, I really enjoyed it. As a whole I am a fan of Wray and Mitchell's work, but I agree that their JG proposal needs work. A quick question. Isn't the widely held belief that Keynes saw wage stickiness as the primary cause for unemployment incorrect?

    1. I agree with Lars Syll’s post and the point that Keynes did not attribute unemployment to the failure of wages to fall in a recession. But I’m not clear what that has to do with my above article. I do argue in an article a few days later that JG people should be allocated to existing employers at a subsidised rate or for free, and that arguably conflicts with Keynes. Article is here:

      My answer to that is that Keynes disputed the idea that a fall in ALL WAGES would achieve anything much, and I agree with that. My point is very different. I’m saying that the lower unemployment is, the less suited will the unemployed be to vacancies, so how’s about subsidising them into work for which they aren’t brilliantly well suited till something better turns up. I.e. Keynes was concerned with the workforce as a whole, whereas I’m just considering what happens at the margin.

  3. Thank you for this post, I really enjoyed it. As a whole, I am fan of Wray and Mitchell, but I agree that their JG proposal needs work. A quick question. Isn't the widely held belief that Keynes saw wage stickiness as the primary reason for unemployment incorrect?


Post a comment.