Tuesday, 17 March 2009

There is a type of employment subsidy that has long had great appeal. It is the idea that employment can be raised by rewarding each employer in proportion to the additional employees taken on as compared to some starting date. This form of subsidy was actually implemented in the UK for a short while two or three decades ago under the name “Small Firms Employment Subsidy”. The big idea, of course, is to raise employment overall.

In late 2009 something of this sort was proposed by Obama under the guise of "
tax credits for new jobs" or similar. And in Sept 2011 he proposed the same idea again.

This idea has a long history. E.g. see
here, here, and here. But none of the latter papers / articles seems to have tumbled to the basic flaw in the idea.

An introduction to subsidies, marginal and “not marginal”.

All subsidies can be divided into two sorts. First there are marginal subsidies. As pointed out above, these involve subsidising just the additional units that get produced or employed as a result of the subsidy. As to the second type of subsidy, there is unfortunately no one widely accepted name, but the phrases “average subsidy” and “intra-marginal subsidy” are sometimes used. I’ll use the latter, i.e. “intra-marginal subsidy”. This second type involves subsidising all units, e.g. all apples, all youths, all steel production, etc.

The big attraction of marginal subsidies (if you can get away with it) is that they cost less. E.g. if someone wants to subsidise apples by enough to raise apple sales by 10%, then subsidising just the “marginal apples” would be 1/10th the cost of subsidising all apples.

I’m not sure whether a marginal apple subsidy could be made to work, but certainly there is no problem – at least initially - with a marginal EMPLOYMENT subsidy. That’s because if a firm employs X people at one point in time, and the firm employs X + Y some time later, then the firm has obviously expanded its workforce. The problem comes a few months after implementing the subsidy as is explained below.

The flaw in MES.

Explaining the flaw in MES requires a basic understanding of what goes wrong with economies as demand and employment rises and, at some point, inflation kicks in (result: no more demand is possible – unless you want hyperinflation).

Let’s start with a hypothetical economy where unemployment is far too high. In this situation, raising employment is easy: just raise demand. This can be done, for example by printing more money and dishing it out; the money gets spent, which in turn employs people.

At first there is no problem. Employers require various types of labour to increase production (skilled, semi-skilled, etc) and for the most part, whatever employers want, labour-wise, they can find amongst the unemployed. But as the total number of unemployed shrinks, the variety of skills available on each local labour market also shrinks.

The point comes where employers begin to satisfy their need for more labour by
poaching it from other employers, rather than take such labour from the dole queue (plus see here and here). Of course there is no one level of unemployment at which “poaching” suddenly kicks in. But the rise of poaching as unemployment falls has been studied and measured. And these studies confirm what is little more than common sense, namely that as unemployment falls, poaching rises.

This poaching is not necessarily conscious on the part of employers. The point is that the more desperate employers are for the right kind of labour, the more attractive their job offers and remuneration offers become. This inevitably attracts labour from other firms.

In addition to poaching, employers resort, as unemployment falls, to raising wages in their own firm, so as to discourage other employers from poaching. But this is clearly a zero sum game. The only net result is rising pay (in money terms) for the country as a whole. And the further unemployed falls, the faster the wage rises become and the worse inflation gets.

Indicentally and as regards the marginal employment subsidy announced by Obama in the US in September 2011, the first reason this subsidy is pointless is that it achieves nothing that cannot be achieved by a straight rise in demand: that is, given the very high unemployment levels in the US in 2011, inflation is unlikely to stem from the labour market.

Let's now return to the hypothetical economy where demand has risen as far as it can, and employers are engaged in poaching labour to such an extent that inflation kicks in.

There is clearly a trade off between inflation and unemployment, and economists normally call the minimum level of unemployment compatible with acceptable inflation NAIRU (which stands for Non Accelerating Inflation Rate of Unemployment).

If unemployment is above NAIRU, there is no case for any employment subsidy: unemployment can be lowered simply by raising demand. I.e. the big question is whether MES can lower NAIRU. Well, the answer is it can, for a while. The way it does so is as follows. Let’s assume MES in introduced on 1st Jan.

Employers who previously rejected dole queue labour because of the latter’s unsuitability will be tempted to take on such labour because the subsidy compensates them for the unsuitability. I.e. MES certainly lowers NAIRU initially. Indeed it lowers it till the remaining unemployed are so unsuitable, that even the subsidy does not persuade employers to take on this labour: i.e. inflation will rear its ugly head again if demand is raised further. But at least NAIRU has been lowered a bit.

Incidental note on “suitable” versus “skilled”.

Most economists when discussing labour shortages refer to “skilled” labour shortages. I will use words suitable and unsuitable instead because the latter words more accurately describe the nature of the problem. To illustrate, the supply of skilled labour available from the ranks of the unemployed certainly diminishes as employment rises, but it’s never just skilled labour that employers are looking for. They look for semi-skilled and unskilled as well. And even the so called unskilled (or 95% of them) have some sort of skill or experience to bring to a job, which employers almost invariably enquire about at the job interview stage.

Moreover, employers when they look for skilled labour are not looking for “skilled labour” in the simplest and most banal sense of the phrase. For example if an employer has a vacancy for a fully qualified plumber (a “skilled” vacancy) a fully qualified electrician is certainly skilled, but the latter will not be suited to the former job. In sort, the word which captures what employers are after is suitable rather than skill.

Returning to the subsidy.....

Returning to the subsidy, to keep things simple, lets assume that the additional employment created by the subsidy happens almost instantaneously, or at least within the first week of the subsidy. Employers take on additional labour, and the subsidy compensates them for the unsuitability of new employees. No problem. The subsidy works.

But a problem gradually emerges which is as follows. Labour markets are in constant flux: employees retire, fall ill or quit voluntarily. Plus employers requirements labour-wise change over time. Six months or a year after implementing the subsidy, employers (in exactly the same way as before the subsidy was implemented) will be looking every month for an assortment of different types of labour to replace those who have retired, quit, fallen ill etc.

But there is a big diffeence between the initial week of the subsidy and the situation a few months down the road is that in the former case the employees taken on were all relatively unsuitable, and employers were compensated for this unsuitableity. In contrast, a few months down the road, employers are not just on the look out for relatively unsuitable labour: they are on the look out for all types of labour, including the highly skilled and the "very suitable" (becaue a proportion of the latter will have retired, quit voluntarily, etc). And these latter types of labour just aren’t there in the dole queue in sufficient numbers. (Remember NAIRU: there is a minimum feasible amount of unemployment for a given level of demand).

The latter point can be put another way. On implementing the subsidy, it is obvious to employers (and everyone else come to that) which employees are being subsidised. In contrast, a few months later, the subsidy simply becomes a monthly cheque that comes from government which is not obviously applicable to any particular group of employees. Most importantly, the subsidy is not even necessarily applicable to the employees who were initially taken on as a result of the subsidy because not all of them will still be unsuitable. For example, some will have gained firm specific skills and will have become perfectly viable without the subsidy. In addition, most employer’s labour requirements will have changed over the months, and some employees who were perfectly suited to their work are allocated to different jobs within their firm, jobs to which they are not so suited. Some of these latter employees in the absence of the subsidy and rise in demand would have lost their jobs and would have been available to help other employers fill vacancies. Unfortunately they are not available.


Marginal employment subsidies are pointless given very high unemployment: a simple straightforward increase in demand is better. Second, given lower unemployment or unemployment which is at NAIRU, a marginal employment subsidy will work (i.e. reduce NAIRU) but only for a few months.

And finally, is there a solution to the latter “works only for a few months” problem? The answer is “yes”: see here.


Note: I made a few alterations to this article in Sept 2011 in reaction to Obama's promotion of the marginal employment subsidy idea, but the basic argument remains exactly the same.