At one of his seminars, a student asked Professor Mises, “Why aren’t all businessmen in favor of capitalism?” “That very question,” Mises answered, “is Marxist.” Mises’s response shocked me at the time. It took me some time to realize what he meant. The questioner assumed, as had Karl Marx, that businessmen had a special group or “class” interest in capitalism that other people didn’t.
“Capitalism,” Mises went on, “benefits everyone — consumers, the masses. It doesn’t benefit only businessmen. As a matter of fact, under capitalism some businessmen suffer losses. A businessman’s position on the market is never secure; the door is always open to competitors who may challenge his position and deprive him of profits. Yet it is this very competition under capitalism that assures consumers that businessmen will do their best to furnish them, the consumers, with the goods and services they want.”
Compare this to current political economic models. Any representative agent model and most interest group models are based on identifying “classes” people. Though we shun Marx’s terminology, that’s the effect. Capital owners and laborers; high-skilled and low-skilled; urban consumers and rural farmers; net-food-buyer farmers and net-food-seller farmers…
Any time we decide our grouping is too broad for the purposes of the moment, we only subdivide it into smaller classes and assume agreement between members of the smaller group. It’s not “the poor”, but “the newly poor, the transient poor, and the chronic poor.” Still not enough? Fine, the chronic poor are really the minorities, the disabled, the elderly, and the low-education female-headed rural households, and their interests all agree with each other. Unless we need to separate low-education female-headed rural wheat farmers from low-education female-headed rural landless households.
Groups of people are useful constructs. They help organize our thinking. However, it is easy to become sloppy and over-generalize. Are there costs to considering groups? Yes. There are also benefits. One of the real policy problems is that we can never be certain we really have gotten the precisely right degree of generalization, where the costs of failing to acknowledge the distinctiveness of a subgroup or individual’s situation outweigh the computational burdens of modeling yet more layers.
One answer is that this is why policy experimentation is important. Start with something tractable, find out the flaws, and adapt to the most relevant groups. Repeat.One answer is that this is why policies should be done at the lowest level of centralization possible (even … at the individual or household level? Gasp!). The further away we get from the individuals we are trying to understand, the more we lose in understanding.