Taxis, coffee, and another broken equation

Standard macroeconomic models are premised on the assumption that recessions occur because of “nominal rigidities” in wages and prices.

I’ve never found this convincing. Economists observe that quantities move by more than prices – and then conclude that prices don’t fluctuate sufficiently freely to allow markets to clear. I find myself wishing that prices fluctuated a lot more – if only to prove that we would still have recessions.

I am also suspicious of the prevailing methodology. Theorists typically start with clever, well-formed, market-clearing models and then try to tinker with them to create recessions, rather than attempt to describe what actually causes recessions and how they propagate. It’s a standard approach in economics: “This is how market-clearing works, now how does it break down?”. I prefer: “How does the world work? Ok, how do we model that.” Akerlof has the right approach to labour market theory.

Simon Wren-Lewis has written about his dissatisfaction with standard New Keynesian interpretations of labour demand. He cites interesting work by Pascal Michaillat which observes that jobs are rationed in recessions. Roger Farmer has also argued at length that the problem of recessions and unemployment is not wage and price flexibility.

I don’t like the standard description of labour supply, either. My hunch, in fact, is that wage flexibility can increase unemployment in recessions.

Some months ago, over coffee, Tony Yates and I discussed dissatisfaction with the premise that work has negative utility. It follows from the assumption that we have to be induced to work, otherwise we would choose “leisure”. I don’t think anyone who has thought about the standard microeconomic work/leisure trade-off finds it convincing. It’s rather obvious that people are hard-wired to work, much “leisure” involves effort, and most paid employment has good and bad aspects. What’s clear is that work is not like typical “goods and services” which we substitute in response to changing prices.

It is also important to be clear that “utility” in economics is not the same as “enjoyment”. “Positive utility” really means that people want to do something (or want to have something) beyond simply a response to incentives. That could be because of enjoyment, it could be a sense of vocational obligation, or simply a desire to work as an end-in-itself. Clearly, there is a broad set of motivations which encourage us to work.

Now it is true that people are not usually willing to work for nothing, but I am simple empiricist, and this is a blog. The very existence of the blogosphere proves that people will work for nothing.

So how should we approach the “supply of labour”? Many, if not most, people derive some positive utility from work. We also derive income, which we need. And we also obtain status from work (a motive George Cooper has astutely emphasised). These are three combined factors that cause us to work and influence our employment decisions, subject obviously to the availability of employment opportunities. There is also likely to be an optimal amount of work that we want to do, at any point in time. We don’t want to work to the point where either the quality of our work declines, or the quality of our non-work life declines. That’s what “balance” refers to.

Now these four considerations can be conflicting. The work we enjoy most may not deliver sufficient income, or status. We may need to work beyond “balance” to pay bills, or afford a house.

I think these factors are sufficiently general as to be deemed relevant to most, if not all, workers. Obviously, the sign and magnitude of each factor varies. Some people don’t like their jobs, and they want different jobs. Virtually everyone needs to work to generate income, and most target a minimum level of income – often determined with reference to a peer group. People derive varying degrees of status from work. Status is ranked, and unemployment typically has low status.

Is this a contentious description of how most people approach work? I don’t think it is. The problem for economic models is that this describes a parallel universe. But before we explain how labour supply is typically modelled let’s consider – with reference to these factors – how most people are likely to respond to a reduction in their hourly real wages. Let’s also assume that the decline in real wages is occurring across the labour market.

I am describing an environment similar to the last recession, where real wages are falling and people are losing their jobs. Whether or not you enjoy your work is not really relevant. It’s possible, that if you really hate your job and the real wage falls that you might decide to throw in the towel. But that implies a very low target minimum income – unemployment benefits. Most people in a recession probably enjoy work a lot less, because the threat of job loss hangs over them and their colleagues – that makes everyone miserable. Also, they’re in competition – only a fraction lose their jobs. Unemployment is low status, and implies a significant reduction in income. So people probably work harder and longer – they want to prove their value.

Despite working longer hours, this may easily coincide with significant declines in output (and by definition productivity). On a recent trip to Dublin, I had an enlightening series of discussions about the labour “market” with a number of cab drivers. They had been at the sharp end of what I am describing. During the recession hourly real wages for Dublin taxi drivers collapsed – perhaps by more than 30%. They all responded by working more hours. For obvious reasons, they were targeting a minimum level of income. They had bills and mortgages to pay. They also described a significant increase in the supply of workers from other professions: teachers and retired policemen worked as part-time cab drivers at the weekend on the lucrative runs from the airport, because they were trying to supplement reductions in wages and incomes.

Collectively, of course, this behaviour is somewhat self-defeating. Sharply falling hourly wages increases the supply of hours-worked, and causes hourly wages to fall further. But this is exactly what one should expect. People try to maintain their “standard of living”, and limit its decline.(There’s also a related measurement question as to what constitutes “hours worked”. Is a taxi driver working when he is waiting for customers, or only when he has a paying passenger? If sitting in queues waiting for customers is a measure of unemployment, it went up a lot as the supply of taxi drivers increased. So falling incomes increased unemployment.)

How well does macroeconomic theory map on to this reality? Not very well. Macroeconomic models assume that the labour supply curve is upward-sloping: higher wages cause an increase in the supply of labour and falling wages cause more people to “choose” leisure. What we are describing is the opposite: falling wages cause an increase in labour supply (a potentially big increase, in fact). Economic theory derives an upward sloping supply curve of labour from the assumption that wages induce us to substitute work (which has negative utility) for leisure. As we saw with the consumption Euler equation when any price changes there is a substitution and an income effect. So standard micro theory can easily incorporate a change the sign of the slope of the curve: if the income effect dominates. I am suggesting that when wages fall precipitously, the income effect does dominate. I also think that substitution effects are far weaker than assumed because work has positive utility.

So where does this leave the model of recessions that blames everything on nominal rigidities? It looks deeply unsatisfactory. Wage and price flexibility does not mean that all wages and prices fall by the same amount, if real wages decline, there may well be an increase in overall unemployment – even more so if jobs are being rationed.

Further notes

On the internet and working for free this by Tim O’Reilly is fascinating.

Even this caveated description of standard DSGE modelling of labour supply, which acknowledges the potential of income effects to the change the sign of labour supply curves, doesn’t consider the possibility that work has positive utility. It assumes that if wages rise a lot high income earners will reduce their supply of hours worked due to diminishing marginal utility of consumption. But if they derive utility from work, they are not working in order to consume more.