Unemployment is not the same thing as a labor surplus. This way of talking about it bugs me to no end, and I've laid out my position in more detail before (here, here, here, and here). Jonathan is making the same points in this post. The key line is:
"If you use a supply and demand graph to show the labor market, then this interpretation is a bad one. You have a supply of labor that exists before the fall in aggregate demand. This supply of labor does not shrink when aggregate demand falls — the number of people willing and able to provide their labor remains the same (or, may even grow, if new laborers are willing and able to enter the market). If we assume full employment at the original level of demand, then the bracket (on the above graph) showing the change in the quantity of labor employed represents the number of unemployed in a depressed market."
Being unemployed means that you are willing and able to work, but not working. In other words, you are on the supply curve, but you are not to the left of the equilibrium point. Basic supply and demand pictures make somewhat more sense if you have a search model attached (then you have some sense that those to the right of the equilibrium point are "searching" and in the process of lowering their reservation wage, so there's some context to this non-employment status), but usually people don't bother presenting it that way in casual or even some formal discussions. Instead, they talk as if unemployment is a labor surplus and so they automatically zero in on high wages as the cause of unemployment.
I also like that Jonathan gets into a discussion about the marginal productivity of labor and its relation to unemployment. He writes:
"During a depression, you have a certain number of workers who are now unemployed (see that supply and demand graph). These workers are able and willing to work, but firms are not looking to buy extra units of labor. Why? Because the costs of employing that labor are higher than the benefits.
Does this not fly in the face of the Austrian claim that in a free market everyone will find employment? Yes and no. Like I said, I think that this disproves the notion that perfect wage flexibility will end all involuntary unemployment. The derivation of the marginal product of labor is based on “everything else being equal.” Austrians are not counting on everything else being equal. Since wants are essentially limitless (as long as people are acting), then there is always opportunity for employment. But, capital and labor are complimentary, and thus for there to be a greater quantity of people employed there must be an increase in the quantity of employed capital — i.e. an increase in marginal productivity.
What this tells me is, that in the study of depressions the most important factor is not necessarily labor, but capital."
First, I want to push back on this "wants are essentially limitless" line. Let's grant that they are. Still, that doesn't quite solve the puzzle because when we are talking about demand, we are talking about both willingness and ability to pay. Wants may be limitless, but if your ability to pay is not limitless that insatiability is not going to guarantee a fully employed economy.
Second, I like that he's thinking about the marginal productivity of labor. But I would put it slightly differently. "Labor is not a commodity", after all. When a firm hires a worker, they are looking at that very similarly to the way they look at an investment. They are taking their expected stream of wage payments to the worker and comparing it to their expected stream of benefits from employing that worker. We have a tendency to treat labor as a spot market, but it's not. It's much more like a capital market. (In fact it's more complicated because workers can shirk - machines can't. Workers can break the contract and leave - machines can't. Workers can bargain - machines can't, etc. etc.). So the time dimension comes into the employment decision in the same way that it comes into the capital decision.
Now I'm going to turn Keynesian on the quesiton. Why do we think depressions happen? We think they happen because firms compare the marginal efficiency of capital to the interest rate, and the equilibrium level of capital at that interest rate is not enough to achieve full employment. I would argue (Keynes did not, to my knowledge) that firms also consider a marginal efficiency of labor and also compare that to the interest rate. Labor is unemployed for the same reason that capital is, in other words. The future stream of earnings expected from employing an additional unit of labor is not sufficient to justify employing that labor.
Keynesians say it's because of insufficient demand and uncertainty about future demand. This particular Keynesian would say that because labor is not a commodity, and because ever since Mincer we know that it acts an awful lot like capital, labor is unemployed for much the same reason that investment levels are low.