I started thinking about Say's Law, which is usually stated as "Production creates its own demand." IIUC, the idea is that if you produce something, you now have an asset that you can trade with somebody else for his/her assets, thus creating a demand for those other assets exactly equal to the asset you produced. The more production, the more assets, so the more demand.
One possible problem with this is that the desire for something isn't in the same place as the assets to pay for it. Suppose you produce something that lots of people want, but most of the people who want it are producing very little -- say, they're unemployed or underemployed -- so they don't have a lot of assets to trade for what you produce. Perhaps they can't even afford to buy it at your cost, which is the lowest price at which you're likely to want to sell it. So you stop producing it. Rather than production creating its own demand, we have a sort of contrapositive: the lack of demand creates a lack of production. But presumably the reason you were producing that thing is that you're GOOD at producing it, so by switching your production to something else, you've made the whole economy less efficient.
A related problem: suppose you produce something, but you DON'T have the asset you just produced -- your employer does. Your employer now has a bunch of buying power, but isn't interested in buying the same things you want, so there is effectively no demand for the things you want, which means they won't be produced. Indeed, your employer may be not interested in buying anything at all, which means all the assets you produced are just sitting in your employer's bank account, not creating any demand at all.
How can this happen? Ask anybody who works at McDonald's or Wal-Mart.
What is a worker worth?
One answer says "you get paid what you're worth -- indeed, what you're worth is DEFINED (in a market economy) by what you can get through negotiation."
Another answer says "you're worth what you produce," i.e. the value added by your labor.
These answers don't necessarily match one another very closely. In fact, if you work for a for-profit company, what workers get paid must be LESS than the value-added that they produce, or there would be nothing left to show as profit. (For simplicity, I'll assume that all of a company's value-added -- its revenues minus its non-labor spending -- is created through labor, not through (say) buying something and holding onto it until it appreciates and can be sold for a profit.) But that's not true for all workers, only for the AVERAGE. Specifically, the total wages of all the workers in the company must be less than the total value-added of all the workers in the company, in order for there to be a profit. How much less depends on the worker's bargaining position: if the worker is hard to replace and knows it, (s)he can negotiate a wage that's pretty close to his/her value-added, while a worker who's easy to replace is in a weaker bargaining position and will probably get paid considerably less than his/her value-added.
There are even circumstances in which a worker can negotiate a wage that's HIGHER than his/her value-added. This clearly isn't in the best interest of the company, but if the worker in question happens to be in upper management, able to effectively set his/her own salary, it's in the worker's best interest to set that salary as high as possible, regardless of the best interest of the company. Naturally, the manager in question doesn't want the company to go bankrupt, so in order to raise his/her own salary, (s)he needs to keep the AVERAGE salary below the AVERAGE value-added by lowering everybody else's salary. But you can't do that if your workers are in a strong negotiating position, so for your own best interest (if not necessarily the company's), you need to put them in a weak negotiating position: bust the union and keep overall unemployment rates high.
I was involved with the Co-ops And Enterprises Board at my graduate school (UC San Diego). UCSD had a thriving bunch of student-run, on-campus businesses, many of which were organized as egalitarian co-ops. At one point there was a discussion of how to set salaries for co-op workers, and the University administrators on the board said "we can't let people set their own salaries: there's an obvious conflict of interest, people will set them too high and the co-op will go bankrupt." The Co-op people replied "There's no problem with people setting their own salaries, because they know that if everybody in the co-op gets paid too much, the co-op will go bankrupt. The conflict of interest arises when one person can set not only his/her own salary but other people's as well, as happens at the University bookstore: the manager can set his own salary high and the check-out clerks' low. In a co-op, no one person's salary will be much higher than everybody else's, so this isn't a problem for a co-op." The University administrators somehow didn't see the logic in this.
Anyway, the problem arises, predictably, whenever the people making salary decisions for a whole company are a small subset of the paid employees of that company. If the decisions are made by a large subset, those people's salaries can't be too much higher than anybody else's, while if the decisions are made by people who aren't paid employees at all, there's no conflict of interest and the decision-makers can act in the best interest of the company. Yet there IS a skill to management, managers DO provide a useful service to a company, so they deserve to get paid for that service. The question is how to pay fairly for their management services without allowing their management position in the company to enable them to set their own salaries higher than their value-added. You could try some sort of "peer" salary-setting -- I don't set my own salary, but I get to participate in setting the salaries of all the other managers at my rank -- but as long as the set of managers is a fairly small subset of all the workers, this still encourages them as a group to set their own salaries irrationally high and everybody else's low.
It's late: I'd better get to bed.