It’s a Dirty Rotten Job but Someone’s Got to do it

Robert Waldmann

Below I ask why the argument that the stimulus must crowd out private spending because output can’t be higher than aggregate supply hasn’t been made (or made more often if I missed it). So now I will try to make that argument. One rule is that I must concede that there are unemployed workers and idle capacity and, in particular, that there are more unemployed workers and idle capacity than before. Second, I must assume that we are in a liquidity trap. I was working on an argument, but Nick Rowe, in comments, suggested that this is really simple. After the jump I give my effort at an argument then reply to Nick Rowe’s comment.

OK how can there be increased involuntary unemployment and a liquidity trap yet fiscal stimulus has no effect on output. I will start with models of involuntary unemployment. Some such models are based on nominal rigidities. We know that in these models the stimulus will cause increased output.

How about efficiency wage models ? Well start with the best known such model the moral hazard model. In the model workers are tempted to shirk (pretend to work, work carelessly or pilfer) and firms can’t monitor them perfectly. This means that the cost of being caught shirking and fired must be positive. This can only happen if workers wages are above their reservation wage. In the work/shirk model unemployment is not only possible, it is necessary to make the workers work.
the no shirking incentive compatibility condition gives a relationship between the real wage, the average disutility of working an hour (hours per worker are constant in the oldest model) and the unemployment rate. High unemployment makes it costly to lose a job. A high real gain per year of having a job makes it costly to lose a job. Consider the unemployment rate required for a given real wage. It increases if there is a high rate of layoffs, job separations for reasons other than being caught shirking.

The financial crisis has increased the probability of layoffs. This could, in theory, happen with a financial crisis even if aggregate demand isn’t a problem (which can be solved by a stimulus plan). The reason is that banks can’t tide firms over periods of poor cash flow (due to idiosyncratic shocks not aggregate demand) because the banks are broke. So there is a much higher risk that firms will go bankrupt or be forced to shut down production and liquidate inventories.

In the model this means a higher unemployment rate is needed for given real wages.

In the oldest (Shapiro Stiglitz) model employment is a standard function of real wages. So far reduced employment requires a higher real wage which isn’t noticeable in the data so far. With the financial crisis added in, employment is lower for a given real wage as firms are trying to liquidate inventories and replace them with inventories of cash (to replace lines of credit etc at banks that might fail any second). The result is an ambiguous impact on the real wage and reduced employment.

In the model this can happen even if, for example, investment is automatically and magically aggregate supply minus consumption minus government consumption minus net exports. Thus it can happen even if there is no problem with aggregate demand.

Aggregate supply in the model depends on, among other things, the exogenous separation rate. One can solve out for the real wage, the unemployment rate and GNP as a function of tastes, technology and, in the extended model, the balance sheets of banks.

It has nothing to do with the price level (no nominal rigidities). This means that for all existing models of aggregate demand there will have to be 100% crowding out. It has to happen somehow.

So, I think, it can be done. The model would require absurd parameters to fit the data, but it is a standard model. It happens to be one of the very many models which are anathema to fresh water economists.

OK now Nick Rowe. He pointed out below that modern DSGE models have imperfect competition in labor and product markets, so markets don’t clear. I’d say it is still implied that firms won’t sell any more at given prices (increased demand will cause them to increase their price for given wages and other prices) and workers won’t work more (perhaps they’d each like to but are restrained by unions which uhm you know Nick represent 5% of private sector workers in the USA). So the crazy market clearing claim that firms won’t sell more and workers won’t work more at current wages and prices still hold. Firms would sell more at current prices *if* demand increased and there were price controls so they couldn’t hope to raise their prices. So I’m not sure how plausible a DSGE model with imperfect competition but no nominal rigidities is. I mean it has firms setting higher relative prices if they increase output and that sounds a lot like refusing to meet demand at the current price.

I also don’t see how you can fit the pattern of consumption (down) and hours worked (down) without a sharp decline in the real wage (not in the data) unless leisure is an inferior good (so people work more hours after winning the lottery). With perfect competition, this is a statement about a simple first order condition. With imperfect competition in labor markets and fixed markups of wages on the marginal disutility of labor ditto. Why would that markup increase ? Now this is a claim that DSGE models with or without perfect competition sure don’t fit the data. It is an old claim (I heard Greg Mankiw argue it in the 80s). I haven’t heard a convincing counter argument.