How lumpy is the labour market? Several recent posts prompt me to ask.

Here's what I mean. Think of two possible extreme mental models of the labour market. At one extreme, the market is flat, and workers are easily substitutable between jobs. At the other extreme, the market is lumpy with each job and each worker being different, and so workers are not substitutable. These markets differ in some important ways.

First, if the market is flat, then John Quiggin is right to say that the internet should have reduced unemployment because it greatly reduces the cost of searching for jobs or workers and so facilitates more matches between them.

However, if the market is lumpy, this isn't necessarily the case. If you're looking for very particular worker (or job) then we're in Nick's world. Where there's lots of heterogeneity, he says, "there might not be any suitable matches on the market right now, and one or other side might choose to wait until a better match appears on the market." His analogy to the car market is good: there are thousands of used cars, but the market for a precise type of Mazda MX6 near Ottawa could be thin.

Secondly, in a lumpy market, we might see shifts in the Beveridge curve as mismatches betweenworkers and vacancies increase and decrease. In a flat market, by contrast, we'd see shifts along the curve in response to changes in aggregate demand.

Thirdly, in a lumpy market wages and profits depend upon bargaining over how to divide the surplus that results from the match between specific workers and jobs. The division of this surplus will depend upon individual bargaining power. This means there might be many employers who have some monopsony power by virtue of being (near-) sole offerers of some specific jobs. And this justifies minimum wage laws or the exercise of union power - because as Mike says, in the presence of monopsonies, higher wages don't necessarily destry jobs.

Fourthly, in a lumpy market efficiency requires investment in job-specific skills, because each job is different. In such a world, Jackart's advice to workers - if you don't like pay and conditions, leave - becomes inefficient as leaving would break efficient matches.

What's more, workers with an eye on the exit would be deterred from investing in job-specific skills. Take, for example, me at the Investors Chronicle. If I feared for my job, I'd write lots of articles on economics of the sort that would appeal to future employers: pieces of macroeconomic futurology that praised or blamed Osborne. But the IC wants something more specific - pieces that are helpful for individual investors. Job security encourages me to do this, and to develop job-specific skills to the benefit of my employer.

In this context, trades unions and employment protection laws aren't labour market frictions, but are in fact productivity-enhancing, as they encourage job-specific investments.

Finally, in a lumpy market, workers are imperfect substitutes for each other: for example, Aditya Chakraborrty and I are both economics writers, but because of job specificity were are hardly competitors. This means that an excess supply or demand for workers will have little impact on wages. This is one reason why immigration does not depress wages, except at the very bottom end of the market - it's because immigrants aren't substitutes for native workers.

So, is the labour market flat or lumpy? The answer of course is: a bit of both. On the one hand, the Beveridge curve does slope down, which alone is evidence for John's claim for the superiority of simple aggregate demand models over search models and for flatness over lumpiness. But on the other, the relationship between vacancies and unemployment isn't a 100% fit; unemployment now is higher that the number of vacancies would suggest. That's evidence that, as Noah says, aggregate demand isn't the whole story. And of course, casual empiricism tells us that workers aren't perfectly substitutable; I'm better at writing for the IC than the Guardian, and better at both than heart surgery.

My point here is twofold. First, simple mental models don't map perfectly onto a messy world - which is why simple arguments against immigration, minimum wages or employment protection laws fail.

Secondly, the degree of lumpiness in the labour market can change all the time depending upon the nature of technical change and creative destruction. In this sense, the old warning about economic models - that the map isn't the terrain - actually understates the case: the terrain can change whilst the map doesn't.