Summary.

"Summary of summary": It is argued below that temporary subsidised work for the unemployed can indeed be created in the public sector, but that there is no reason to confine such work to this sector: this form of work can also be created in the private sector. The argument runs briefly as follows.

1. Creating employer of last resort (ELR) work in the public sector is easy in principle. For example just tell the unemployed that their unemployment benefit is henceforth conditional on their keeping their neighbourhood free of litter. Those refusing the work would not be counted as unemployed. Hey presto: unemployment has vanished.

2. The above is the ludicrously unproductive work. However productivity can be improved by getting better “temporary unskilled labour” to “permanent skilled labour and capital equipment” ratios by allocating the unemployed labour to existing public sector employers.

3. Unfortunately the latter are not good at employing unskilled labour. So can the labour be allocated to private sector employers as well?

4. A problem would appear to be the increase in demand required to create work for the labour, and the inflation that would result. However it is not really excess demand as such that causes inflation: it is excess demand for skilled labour that is the problem. And demand for skilled labour is no more likely to be exacerbated by creating ELR work in the private than in the public sector.

5. The skilled labour problem problem can be solved by pricing the temporary unskilled labour at a level that does not induce employers to seek additional skilled labour when taking on the temporary unskilled labour.

6. The employment subsidy that results from taking the above theoretical points into consideration turns out to be something very similar to a totally free labour market, but with government ensuring an adequate take home pay for those on low wage, temporary, and not desperately productive jobs.

7. The above points can be put yet another way. As Keynes pointed out, as employment rises, the marginal net revenue product of labour drops. The point at which the net revenue product of labour reaches the minimum wage or union wage is NAIRU, and a further rise in employment is supposedly not possible. However, the cost of marginal labour to the employer is below the latter wage because of an appropriate subsidy, then the “NAIRU obstruction” can be by-passed. This "Keynes - NAIRU" point is probably not obvious from the above summary, but it is clarified below.


Origins of this article.

The article below is a shortned version of "Workfare: a Marginal Employment Subsidy for Private and Public Sectors" (1991) by the same author. The latter is about 18,000 words and the article below is about 2,500.


Introduction.

The idea that government should act as Employer of Last Resort (ELR) is as old as the stars. Perhaps the earliest recorded instance was 2,500 years ago in ancient Athens. Unemployed Athenians got some sort of dole or hand out. It occurred to Pericles (leading Athenian statesman) that instead of offering people money for nothing, those concerned could repair the walls of the city.

Between roughly 1600 and 1900 we had workhouses in Europe and North America , which were ELR of a sort. Then came the 1930s depression with a host of ELR schemes of which the probably the biggest was the Work Progress Administration (WPA) in the US.

It is argued below that ELR in the form of schemes set up specially to provide work for the unemployed (like WPA) are condemned to inefficiency. However examining the reasons for this inefficiency leads to something better: temporary subsidised work with existing employers public and private.


The role for ELR

There is no case for trying to use ELR to deal with demand deficient unemployment. The best way to deal with this form of unemployment is to raise demand (though as will be shown later, ELR can help a bit with demand deficient unemployment). The limit to using demand to deal with unemployment is posed by inflation, and the level of unemployment at which inflation becomes a problem is normally called NAIRU (Non Accelerating Inflaltion Rate of Unemployment).In short, the role for ELR is to reduce NAIRU, if possible.


A simple WPA scheme.

Setting up an ultra simple WPA scheme is easy: just pass a law saying that the unemployed henceforth have a “job” which consists of their walking round their neighbourhood keeping it free of litter. Pay for the sake of simplicity will equal unemployment benefit (UB). Those refusing the work would be deemed to have turned down work, and would be classed as voluntarily unemployed. Perhaps they should get a reduced level of social security. Hey presto: unemployment has vanished.

Incidentally the phrase “unemployment benefit” was the official name for the social security that the unemployed used to get in the UK. It has actually been renamed, but I’ll stick with UB.

The above simple ELR system might seem silly, but the point of setting it out is just to demonstrate that there are no big problems in principle in setting up ELR, and in particular no big budgetary problems, at least to the extent that UB is somewhere near the minimum wage. As to countries which have no UB (or to the extent that UB is less than the minimum wage) there is still not much of a problem: the sums required can always come out of tax. This arguably makes life on ELR more attractive than unemployment, which reduces job search efforts, which in turn would to some extent create ELR work at the expense of normal work. But never mind. Such a system is certainly feasible.

But the above simple ELR system does have problems, which will doubtless be obvious. First the work is ludicrously unproductive. Second, why would people turn up for a boring unskilled job in exchange for a very small increase in income? The answer to this is that this is not a valid criticism of the above simple ELR scheme because exactly the same problem applies to all ELR schemes, and indeed to all forms of low paid work. That is, in many developed countries UB is often more or less equal to the minimum wage (at least in the case of those with several children).

Another apparent problem is that UB for a proportion of the unemployed is much less than the minimum weekly wage (assuming a full 40 or so hours per week). But this is actually no problem. These people could work part time. Indeed, this makes some sense in that this gives them time to look for more productive work (not that job searching at the same time as working is too difficult: the majority of job changers find their new job while still working at their old job).

Having shown that setting up an ELR scheme easy in principle, the next question is whether it can be improved upon. An obvious first step by way of improvement is WPA.


The flaws in WPA.

The first problem with WPA type schemes is that they inevitably suffer very high labour turnover. This is because the unemployed turn over far faster than employees in normal employment, and assuming the unemployed spend no longer on WPA than on the dole, then turnover of WPA employees will be far higher than for a normal employer. Of course turnover on WPA doesn’t have to be as fast as it is for the unemployed, but if it is not, then labour supply to the normal jobs market will be constrained, which is inflationary, which in turn causes more unemployment (hardly the object of the exercise).

The second problem is that inflation arises near full employment precisely because employers cannot find the skills they need amongst the unemployed. I.e. the unemployed tend to be unskilled. And even where an unemployed person is skilled, the reason for their unemployment is likely to be an excess of their skill in their neighbourhood. Thus skilled people are quite likely to work in an unskilled capacity in WPA type schemes. Thus the ratio of skilled to unskilled on WPA is going to be very poor compared to a normal employer.

A possible objection here is that some WPA projects in the 1930s were reasonably productive. This point is irrelevant, because WPA operated during a period of massive demand deficient unemployment, the 1930s. And in this scenario it is easy, as pointed out above, to find any necessary skills amongst the unemployed. Plus in this scenario labour turnover will be lower on WPA than in a full employment scenario.

Conclusion: ELR will be more efficient if those concerned are allocated to existing public sector employers because ratios of skilled to unskilled labour will be better, plus average labour turnover for each firm or employer will be lower.


Public sector employers don’t like the unskilled.

The next problem is that public sector employers are not good at employing the unskilled relative to the private sector. Just how much use can your local public sector hospital, school or tax office make of temporary unskilled labour? Not much, probably. And in any case, why not make use of the private sector if it is possible to do so?

So can the private sector take a proportion of ELR employees? The obvious problem here is that demand has to be increased to make this happen, and demand cannot be increased – unless NAIRU can be lowered.

So, are there any reasons for supposing private sector ELR lowers NAIRU? The answer is yes. And the first reason is what might be called the workfare element in ELR. In the above ultra simple ELR scheme, the unemployed are faced with a choice: pick up litter or get less benefit. This clearly encourages job searching. (Incidentally, the letters "ELR" at the start of this article refer to a very different beast to that referred to from now onwards: i.e. temporary subsidised work with existing employers in both public and private sectors. However, I'll stick with "ELR").


The marginal product of labour.

There is a second reason for supposing private sector ELR reduces NAIRU. It has to do with a point made by Keynes about the marginal product of labour. Its an obvious point, and indeed Keynes is not the only one to make it. The point essentially derives from the point made above about the relatively poor quality of labour available from the dole queue at NAIRU. The point is as follows.

Given rising demand and falling unemployment, the suitability of each succeeding person hired for whatever job they take will tend to decline. That is, the output of those hired will decline as unemployment falls (i.e. the "marginal product of labour" declines, to use economics jargon). This decline will continue until employers find that they get better value for money from poaching other employers’ labour than taking labour from the dole queue. Of course there is not a fixed level of unemployment at which employers suddenly give up on the dole queue and start poaching other employers' labour (and raising the wages of their own labour with a view to stopping other employers poaching their own labour). Nevertheless, a point comes, as unemployment falls, where the bidding up of the price of labour (or giving into union demands more readily) reaches the point where the inflationary consequences are serious.

Now assuming an economy is at NAIRU and unemployed and relatively unskilled labour can be made available to employers at a subsidised rate, then employers will be more likely to take on such labour instead of poaching. Thus NAIRU ought to decline.

There is, however, an apparent problem: what’s to stop employers using the system to subsidise employees who are perfectly productive and who the employer would have employed anyway? Well, there are various anti fraud measures, and combinations of different measures, and I wont go into a detailed discussion as to which is best. They are as follows.


Fraud prevention.

1. Limit the time ELR people stay with a given employer. Employers don’t want to lose their more valuable employees, thus they will tend not to put the latter onto ELR. (Another desirable rule is not to allow ELR employees who have reached the end of their term with a given employer to return to the employer concerned, otherwise employers would just put valuable employees onto the ELR subsidy for short periods.)

2. Limit ELR people to a small percentage of each employer’s workforce.

3. As already pointed out, the pay of ELR people has to limited so as to give them an inducement to find more productive work. Valuable employees tend to get decent pay, and they will not take kindly to being put onto the sort of minimum wage levels of pay that would be involved with ELR.

4. Another obvious fraud is to hire relatively productive employees on ELR and boost their pay by means of "cash from under the counter". However, this fraud is widespread anyway: it is hardly unheard of for people on UB to get "cash" jobs on the side. Thus ELR would probably not result in an expansion in this particular fraud.


The marginal product of labour (again).

The point made above about the workfare element in ELR reducing NAIRU is obvious: probably half the general population can see this, never mind economists. However the marginal product of labour point is more subtle, and I have not come across anyone else who has made this point.

It is pretty obvious that there is something to be said for subsidising the unskilled or others with labour market problems. The marginal product element in the system not only caters for these people, but also facilitates the employment of relatively skilled members of the dole queue. This might seem a waste on the grounds that the relatively skilled will probably get unsubsidised jobs quite easily. There at two answers to this criticism, and as follows.

First, most employers are not falling over themselves to take on totally unskilled employees. What most employers need is a mix of highly skilled, semi skilled and unskilled. Thus if the number of semi skilled employees with a given employer can be raised, this will probably make it easier to take additional totally unskilled employees.

Second, the idea that "it is a waste of taxpayers' money to subsidise the skilled because they will find employment anyway" is to confuse movements of money with real economic effects. So long as subsidising the employment of the skilled raises real national income, such a policy cannot be criticised. Though, of course, it is always possible that the administration costs of any ELR system are so high that the these costs outweigh the real output of ELR employees, in which case the above critism might be valid.

The only real world ELR system I have come across which appeared to have taken on the above "subsidise the skilled" point was one that operated in Switzerland a few years ago. The Swiss system subsidised skilled people into temporary work, and not only that, but allowed skilled people to use the system in this way soon after becoming unemployed (2.). The evidence seems to be that this produces benefits (7). If the above reasoning is correct, the Swiss have got their heads screwed on. (Given the huge number of multinationals owed and run from Switzerland, it has long been obvious that "the Swiss have got their heads screwed on"!).


The central flaw in public sector ELR.

The above points about the marginal product of labour are no more than reasons for thinking ELR would work to some extent in the private sector. These reasons do not address what is probably the fundamental reason put by the advocates of public sector ELR for preferring the latter to private sector ELR. This is that expanding the public sector ELR does not involve an increase in demand, and thus this expansion cannot be inflationary.

The first reason for being at least suspicious of this argument is that there has been an astronomic expansion in the public sector in most countries over the last century, yet we seem to have had no obvious improvement in NAIRU ! However, the central flaw in the public sector ELR argument has to do with the root cause of inflation, and it is thus.

Inflation can be “cost push”, and there is such a thing as a inflationary cost-price spiral. But as far as ELR is concerned, the worry is inflation caused by the increased demand needed to bring private sector ELR jobs into being. Now the root cause of demand lead inflation is excess demand for specific types of labour (usually skilled labour). But if the public sector is given hoards of relatively unskilled ELR people to employ, it is quite likely that this sector will require additional skilled labour (and/or more “skilled permanent” labour) to supervise the relatively unskilled ELR labour!

Indeed, it is more likely that the public sector will have this “inflationary” problem than the private sector, because the latter is better at employing unskilled labour!

Conclusion: there is no reason for any artificial preference to be given to the public sector under ELR.



Using ELR to deal with demand deficient unemployment.

Unemployment rises and falls, and there are clearly longish periods (two years or more) when unemployment is above NAIRU. So would there be a case for relaxing one or more of the above rules during a recession? It is hard to see why not. Lets say ELR labour is supplied to employers for free, and lets say demand in money terms is constant. The availability of free ELR employee ought to enable employers to cut unit costs and thus produce and sell more stuff for a given demand in money terms. I.e. the result would be a rise in output in real terms and a rise in employment. So ELR can reduce demand deficient unemployment, but as already stated, it is not the best tool for the job.


Should training be mixed with ELR? NO !

To be cynical, anyone wanting to appear socially concerned can advocate training. Its an easy way to appear virtuous. There are in fact some simple theoretical reasons for thinking that training and ELR should not be mixed.

Training courses necessarily last a specific period of time. This is because efficient education and training involves one teacher and several pupils. The pupils cannot all start their course at different dates, and/or do a course lasting a different time, and/or finish the course on a different date. Thus any worthwhile course lasts a specific period. But the whole essence of effective ELR is that those concerned are available at a moment’s notice for a job which is more productive than ELR. This is liable to mean quitting the course before finishing it.

In short ELR and training are incompatible. This is not to say we don’t need more training. Nor is it to rule out training which is independent of ELR and normal employment (e.g. evening classes).

A related but separate point is that the training associated with ELR seems to provide poor value for money (1, 2, 4, 5 & 7).


ELR imitates the market.

UB and minimum wage rules are not free market phenomena. In a free and perfectly functioning market, there is no unemployment, thus any move towards a better functioning market should reduce unemployment (though in the case of UB and minimum wages some would of course argue that there is a social price to pay for abandoning these two). In a perfect market the unemployed, to a greater extent than where UB and minimum wages operate, would get themselves temporary and relatively unproductive jobs pending the appearance of something better. And the latter is exactly what ELR is. (The evidence seems to be that the above is actually what happens in relatively free labour markets (3)).

As to the idea that temporary jobs might not be worthwhile, the evidence is that peoples' subsequent employment experience is improved by taking unsubsidised temporary work (6).


Other marginal subsidies.

Economically literate readers will doubtless have noticed that the subsidy advocated here is a marginal subsidy. (For those not acquainted with this term, see the blog entitled “Marginal employment subsidies don’t work” – link is at the top right.)

There might seem to be a contradiction here in that the paragraphs above advocate a marginal subsidy, while the above mentioned blog claims that marginal employment subsidies do not work. The explanation is that the “Marginal employment subsidies don’t work” blog criticises what might be called the simplest or most basic marginal employment subsidy. In contrast the paragraphs above advocate a subsidy which hopefully deals with the weaknesses in the “simple, basic marginal subsidy”.


Conclusion.

The arguments for projects or "employers" specially set up to take on the unemployed are feeble. In contrast,there is a good case for subsidising the unemployed into work with existing employers public and private. Pay in these jobs must be limited and those concerned must be about as available for normal or unsubsidised jobs as when unemployed. Various schemes in Europe over the last two decades have taken more or less the above form, e.g. the New Deal in the UK. Thus above arguments provide some support for these real world schemes. But these real world systems need several changes in order to fully comply with the above ideas.


References.

Note: There are a large number of references at "Workfare: A Marginal...." to back the points made above. A few references appear below to works published after 1991 (date of publication of "Workfare: A Marginal....".

1. "Not Working" by Adam Bogdanor, (2004), Policy Exchange, London.
2. Does subsidised temporary employment get the unemployed back to work? An econometric analysis of two different schemes, Michael Gerfin, Michael Lechner and Heidi Steiger.
3. Countries with successful employment policy... by Heinz Werner, Institut fur Arbeitsmarkt und Berufsforschung, Nurnberg, Germany. See p. 21.
4. Does Active Labour Market Policy Work? - Lessons from the Swedish Experience, Calmfors, L., Forslund, A. and Hemstrom, M. (2002)
5. “Microeconometric Evaluation of the Active Labour Market Policy in Switzerland", Gerfin, M. and Lechner, M. (2002), The Economic Journal, 112 (482), pp 854-893.
6. Temporary Jobs: Who Gets Them, What Are They Worth, And Do They Lead Anywhere? Booth, A. L., Francesconi, M. and Frank, J. (2000) Institute for Labour Research, University of Essex.
7. The Employment Effects of Active Social Policy, Bolvig, I., Jensen, P. and Rosholm, M., (2003), Discussion Paper 736 Institute for the Study of Labour (IZA), Bonn. Discussion Paper 00/54