Recent Posts

Wednesday, January 22, 2020

Job Guarantee as a Policy Variable

The Job Guarantee is the most natural implementation of the concept of having the central government act as a price setter. By making an open bid for labour at a fixed price, an effective minimum wage is created in the economy, and it will eliminate almost all involuntary unemployment. This discussion will not cover the tricky question of implementation details, but will instead discuss how this fits in with the Monetary Monopoly model.

(Note that this is the final article in a sequence, the previous article discusses the theoretical concept of a policy variable.)


The concept of a Job Guarantee is straightforward: the central government offers everyone a job at a fixed wage – the Job Guarantee wage. This wage would act like a minimum wage, and so if one wanted to imagine that if the programme was already implemented, the wage offered would presumably be somewhat lower than existing minimum wages in order to keep relative wage structures intact.

(It should be noted that the most likely group to implement the Job Guarantee would push for a higher wage, and force a flattening of the wage structure. These are the sort of details that are interesting, but tangential to my discussion here.)

The obvious question is: what do the workers do? This is a big question, but for our purposes here, it should be noted that they are not being hired to run a business that makes a profit. They are being paid to provide services, but those services are not meant to act as a direct competition to the private sector. Although payment and high-level administration is centralised by the national government, the implementation is meant to be decentralised. One representative example would be for the government to pay people to provide required labour for registered charities.

Note that there would be a permanent staff to manage the programme, being paid standard salaries for bureaucrats in the central government. Those wages are expected to be higher than the Job Guarantee wage.

Although involuntary unemployment might technically be nearly eliminated, that does not mean that everyone would be either working as desired in the governmental or private sector or in the Job Guarantee. Provisions might be made for parents with home care, people with incapacities, students, etc. Also, one might expect that highly-paid workers “in between jobs” might prefer to live off savings rather than work for the Job Guarantee. Those individuals may or may not be considered “unemployed,” but that status is voluntary – they chose to not take a position with the Job Guarantee.
(Unfortunately, some individuals might be ejected from the programme, as they are disrupting the ability of others to work. Whether or not they are considered “unemployed” is another semantic issue.)

I will also note that I am ignoring one theoretical possibility herein for simplicity: the case of an extremely low Job Guarantee wage. If a government offered a Job Guarantee paying 1/100 of the prevailing minimum wage, it is entirely likely that nobody would show up to take part. For all intents and purposes, the programme does not really exist in such a case.

The Job Guarantee Wage

The wage paid in the Job Guarantee is a set price for provided services, with no limit on the quantity. This matches the Monetary Monopoly model. A big difference is that the Job Guarantee is not the only form of government expenditures. However, it was clear that the model was incomplete, and would need augmentation with more dynamics.

If we look at the labour market, if we want to have interesting dynamics, we see that there are three wages of interest.

  1. The Job Guarantee wage, which is a policy variable.
  2. The average private sector wage.
  3. The average permanent governmental employee wage.

The third category – permanent governmental employee wage – is often not thought about. Since the government needs to provide certain services, it cannot be too far detached from private sector wages. (In the real-world, just wage rates are not enough to compare positions, benefits matter, particularly for government positions. From the perspective of a simplified economic model, we will assume that benefits are just embedded in wages.) At the same time, not all positions are filled at all times. For example, imagine that there was a buzz of activity and wages in the private sector have a upward bump. The government could leave its wages steady, and just let the number of employees drop. To a certain extent, that looks like a monopolist fixing the price, and letting the quantity adjust. Given that wages are quite often set in multi-year contracts, this is not completely unrealistic. The issue is how far the wage differential can be allowed to build up; there is presumably a minimum quantity of employees that are required.

We will put aside the permanent governmental employees, and then look at relationship between the private sector wage and the Job Guarantee wage.

The usual story is that private sector wages would be set as a markup over the Job Guarantee wage (at least for the lower wage percentiles; one might imagine that the Job Guarantee wage would not bear much relationship to Wall Street lawyer salaries). The justification for the markup story is straightforward. At the lowest end of the wage scale, employers would need to pay a bit more than the Job Guarantee wage to bring in workers (unless they can dangle the prospect of a promotion, or so forth). So the Job Guarantee wage acts as floor for wages in the private sector. Meanwhile, so long as the pool of Job Guarantee labour is non-empty, why pay a higher price than the usual “reservation wage”? As a result, the markup of the lowest wages over the Job Guarantee wage will tend to be steady.

Positions with more skill requirements would command a greater wage, but we expect that wages would remain in a spectrum. If we look at the average across wage quantiles, the result is that the average wage markup would vary with the cycle.

If the markup is in fact fairly steady, nominal wages would track the trajectory of the Job Guarantee wage in the long term. As a result, the Job Guarantee is a policy lever that could allow for control of the path of nominal wages. If we assumed that the consumer price index has a steady relationship versus average wages, this would also cover the usual definition of inflation. Correspondingly, (long-term) inflation control can be achieved without any input from the central bank.


The theoretical argument is plausible, and one could build it into a mathematical model. However, I think the mathematical models might over-sell the inflation control aspect of the Job Guarantee, at least in the short term.

The main concern I see is the rather extended level of wage inequality. A significant portion of wages right now are detached completely from the minimum wage. As a result the markup of average wages might be large enough that the variation of the markup would swamp the effect of the path of the Job Guarantee.

Another concern is that the Job Guarantee pool could “effectively empty:” the only remaining workers in the Job Guarantee pool are workers that nobody in the private sector is interested in. In many cases, this lack of interest would be for reasons that are highly unfortunate. However, it might happen for economic reasons as well. In a country like Canada with a dispersed population, people in the Job Guarantee programme might be living in areas where no private sector employee is going to set up operations. Telling people “to get on their bikes”[i] to get a job is a politically difficult suggestion.

If the Job Guarantee labour pool is effectively empty, its wage should lose its relationship with the private sector wage. That said, one must ponder that implication. Society would have reached 100% voluntary employment (modulo concerns noted earlier about left-behind workers). Although the post-war welfare state in its heyday achieved high employment numbers, that would probably be better. Although interesting to contemplate, the business cycle exists, and I do not see such an outcome as being plausible.

If we made the somewhat questionable choice of arguing that national output corresponds to national welfare, a 100% employment rate is the optimal outcome. Even if inflation does something crazy – so what? Output is maximised, so obviously inflation is not lowering output. (Please note that I think that equating some objective function to maximising “welfare” is poppycock – political economy matters. Furthermore, one may note environmental and resource use concerns, but note that we can use other policy levers to employ people to something useful on that front.)

As such, worrying about everyone being employed is somewhat bizarre. Nevertheless, the solution is straightforward. A progressive income tax system acts as a claw to drag average post-tax incomes towards the Job Guarantee wage, and it increases its drag as the wage markup increases.

Job Guarantee versus NAIRU

MMT proponents like making an argument that runs something like this: they want a system that keeps the price level stable by guaranteeing a job to all that seek one. Conversely, the mainstream approach to stabilise the price level is to guarantee a certain amount of unemployment (close to will-o’-the-wisp NAIRU).

Presumably, mainstream economists would not like that characterisation.

Technical appendix: The Three Wage System and Utility

The three-way wage split I outlined causes a certain amount of grief for standard neoclassical models. The benchmark treatment is that households have to allocate between leisure and work hours. Working generates income that allows consumption – which increases utility – but not working (leisure time) also generates some utility. The equilibrium solution finds the optimal allocation. These leisure hours in the model are mapped to unemployment, raising the ire of heterodox authors.
However, that basic utility function runs into trouble when facing the three-wage system. If we assume that all jobs are filled in by households splitting duties, there are four buckets:

  1. Leisure.
  2. Working for the Job Guarantee.
  3. Working for a permanent position for the government.
  4. Working in the private sector.

We would assume that wage rates would increase as we went down the list. However, options 2-4 are all jobs, and thus do not generate “leisure utility.” Meanwhile, why would anyone voluntarily take a position at a lower wage? The optimal solution is obviously allocating all working hours to the highest (private sector) wage.

Although a Job Guarantee has not been put into place, it still seems that such an outcome seems implausible in the real world. The economy would continue on as before in the absence of the Job Guarantee, and people would voluntarily choose to be unemployed. Although some high-earners might sit out unemployed rather than take a Job Guarantee wage, it seems implausible that the entire pool of unemployed workers would do so.

The only way to generate plausible-looking numbers is to throw fudge factors into the utility function. These fudge factors would create the possibility of the representative household splitting time between jobs with different wage scales. However, this kind of blows a hole in the rhetoric around dynamic stochastic general equilibrium models. The “deep structural parameters” that were previously estimated would have to be thrown out the window in response to a policy change. If a neoclassical model of a proposed Job Guarantee were to be built, utility functions and parameters would have to made up out of whole cloth, which is not how neoclassical economics is typically marketed.

Heterodox economists would enjoy pointing out that the ideological edifice would be in disarray. Unemployment would no longer be “optimal leisure allocation,” it would be morphed into a factor that forces workers to spend time in the Job Guarantee. Households are free to maximise their utility, but oops, some fudge factor forces them to spend time in the Job Guarantee.


[i] Famously said by British Conservative Politician Norman Tebbit. Telling people to hop on bikes made at least a bit of sense in England but would be met by confused looks by inhabitants of towns that are hundreds of kilometers from the nearest town of any size.

(c) Brian Romanchuk 2020


  1. Wouldn't unemployment benefits effectively act as a "job guarantee." The government is essentially paying you while you re-train yourself, network, and do other activities which constitute a job-search.

    My fear with a JG is that we replace the work of networking/re-training etc with something akin to a low-productivity government make-work program (which would presumably also have a dozen ways to dodge). We may make the unemployed less-employable by switching from unemployment benefits to a JG.

    1. Feel free to look through the MMT literature (I’m not an expert), but the story is the opposite. Paying people to do nothing causes them to drift away from the work force. The Job Guarantee is a job, and meant to be useful. Charities can use bodies for heavy lifting.

      Training would presumably be an option, but training people for non-existent jobs is pointless. Tightening up the labour market forces the private sector to get its act together, and go back to training itself.

      The Job Guarantee also provides a centralised point for job searches. Any big employer with mixed skill needs would just keep a contact at the JG on a speed dial.

      It’s a safe bet that many highly paid professionals will be snooty and not take a JG position. They are entirely free to do a job search on their own dime.

    2. I think it would greatly depend on the individual. Some are much better off using their unemployed time to do a job search (retraining, networking, interviewing, etc). Others less so. I do not trust the government to make that distinction (and can't even imaging what they would base it on).

      I'd be more inclined to accept the notion of a JG if classic unemployment benefits still existed. Then individuals could decide how best to spend their time between jobs.

    3. People are free to not go into the programme. “Training” has been a neoliberal fetish since the early 1990s. In the real world, people are massively underemployed, and many are overqualified. It’s just crony capitalism bailing out a handful of loser firms that are unwilling to invest in training.

      The JG wants to get people out of there, not keep them. Certainly there should be structures to support job searching and interviewing while “on the job”. Given that the administration of the pool is centralised (the actual work places are decentralised), job searching is way easier for blue collar workers. It’s way better than having people sitting on the couch all day. Yeah, white collar workers will want to network in the right places. That’s their choice.

      I’d keep unemployment insurance, on the argument that the free market parties will relentlessly attack the JG programme. There’s no guarantee of survival.

  2. Let's use your Monetary Monopoly Model to see what happens the to money supply by using the JG program in today's economy.

    First some background:

    1. Government today is paying workers more than it possible could if taxes alone financed government expenditures. This leads to an increase in the money supply held in private hands.

    2. A JG program would fit into present government expenditures but could be separated if desired.

    3. Assume that fiat money introduced into the economy is all introduced by paying government workers.

    The method of separating will be to pay workers with marked currency. This currency would be identical (except for the mark) with present currency. We would limit the amount of new (marked) currency issued to the amount needed to make payment that exceed tax-take. Assume this amount to be 20% of every government pay check.

    The initial introduction of the marked bills would be an impulse event. The very first exchange could be recorded as

    Privately held (Marked) money supply (M(0) = Pg(0)Qg(0) - T(0)

    where "Pg(0)Qg(0) - T(0)" is a description of the net (marked) transaction.

    As the time passed and payment continued to exceed tax collection, the amount of marked currency in private ownership would increase as (marked) money supply increased. If taxes were paid using both new and old currency, new (marked) currency would eventually displace old (unmarked) currency.

    We could use this system to track the effect of the JG program. Simply pay JG workers using a mark different from other government workers. Tax collection would settle (eventually) into a government selected ratio of JG to other currency.

    I am at a loss to predict how this would impact wages of private sector workers. Private sector workers are defined as being the recipients of money spent by government workers.

    1. You need a model that is built to handle the added complexity. The Monetary Monopoly model dynamics would be just a part of a larger model.

    2. Yes, the Monetary Monopoly model is just a start but it is a beginning reference point.

      Using the base I suggested, decide if JG will represent an increase in annual government money supply increases. That gives us one parameter.

      I suspect a JG program will increase government demand for unskilled labor. The military is (probably) the biggest present government demand for unskilled labor. With the JG program in place, it will probably be harder to recruit military enlistees. That gives us a second parameter.

      With just those two possible parameters, we have the makings of government created hyperinflation.

  3. This comment has been removed by the author.

  4. Brian says in relation to JG people, “… should be noted that they are not being hired to run a business that makes a profit. They are being paid to provide services, but those services are not meant to act as a direct competition to the private sector.”

    Actually there have been JG type schemes where those concerned ARE TAKEN ON by the private sector. There was one such scheme in Switzerland. See:

    One advantage of private sector JG is that the private sector is better at employing the relatively unskilled than the public sector. Second, the evidence from Switzerland, is that private sector JG improves the employability of those concerned more than public sector JG.

  5. Is there any merit to considering a minimum wage for the private sector that exceeds the JG wage? (i.e. a spread over the JG wage as a policy variable)

    1. In North America, minimum wages are set at the sub-sovereign level (I think the US has a Federal one, but it was last changed decades ago?) and how they are set is a political mess. So there’s implementation issues there.

      If a JG existed, I’d argue that the only role of the minimum wage is to stop employers from taking advantage of workers. You wouldn’t want a huge wedge between the lowest private sector wages and the JG wage - tiering will drive the average wage higher than the JG wage, by a varying amount over time.

  6. "The Job Guarantee is the most natural implementation of the concept of having the central government act as a price setter."

    If JG wages are set to enable workers a standard of living, where on a scale of luxury should the JG worker live?

    1. The JG is an effective minimum wage. The relationship of that wage to the average wage in the economy is a political economy question, and where it ends up will depend on institutional factors.

      If it increases, it might reduce income inequality, and so the average wage rises proportionally less. But if the relative wage structure is unchanged, all that has happened is that average wages rise proportionally. This will feed into consumer price inflation. This largely proportional shift would be guaranteed to happen if we multiplied the JG wage by a large number (100x say), it is impossible for reduced inequality to offset a 100x multiplication of the lowest wages.


Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.