Max R. P. Grossmann

One of the foundational ideas of modern economics is the idea of marginalism. Early economists were much surprised that water is able to fetch much lower prices on the market than diamonds, even though diamonds are clearly a gimmick, while water is necessary to sustain life.

The marginalists were the first to resolve this paradox. Simply put, prices reflect not some general “worth” or “value” of any good. Rather, prices reflect marginal value. That is, the willingness-to-pay of a person who can scarcely decide whether to consume another unit of the good.

It just so happens that in most situations, an additional unit of water is not that valuable, whereas an additional diamond is. The subjective judgments of many people are combined to form prices that reflect this tradeoff. The main attraction of marginalism is that it is well-defined in borderline cases. It applies to all individuals everywhere. If you think about someone in the desert who is barely surviving, that person will be willing to pay much for another unit of water and very little for another diamond. It just so happens that this is not the typical case reflected on the typical market. Typically, water is plentiful. Thus, water is cheaper than diamonds.

Thinking at the margin is a very simple but powerful technique to improve one's economic reasoning. It can be applied to less economic situations as well.

A well-known controversy relates to “push and pull factors” in migration, and especially to the question of whether aspects such as welfare payments and other state-provided benefits cause migration to countries that offer them. Empirical studies on this issue disagree. The general consensus appears to be that the effect of such benefits is small but positive or non-existent.

How can we think about this issue in marginal terms? I believe there are three levels of analysis we can take.

The first level relates to individual migration decisions. In other words: Leaving all else unchanged, is it plausible that any individual human being on earth who would have immigrated to country X had it offered benefits of €0 will now not migrate to X if benefits increase to €1? I am confident to assert that no such individual exists. In other words, it is not sensible to assume that benefits can represent a “bad” that causes migration away from country X.

Note that we leave all else unchanged—that is, we abstract away from the simultaneous implementation of benefits and some other policy that may be less attractive. For example, a country may increase payments while decreasing the level of freedom. Clearly, this analysis would be much more difficult. We leave all other factors that may influence the migration decision fixed. If we then slightly increase the amount of benefits, we conclude that it is unlikely that anyone (!) would switch away from country X because of that causal change.

The second level relates to collective migration decisions. If we trust the first level of analysis, the sum of nonnegative changes is itself a nonnegative change. In other words, the first level would compile into what is an increase in migration to country X, or at least no change. Indeed, it would be easy to argue that the small change from €0 to €1 really does not motivate anyone. However, real-life changes are typically much larger than just the marginal euro. But let's suppose that the first level somehow breaks apart—perhaps at least one individual does indeed find the increase of benefits to be repellant. On a collective level, we would have to ask whether, on net, the number of people who switch away from X is likely or unlikely to exceed the number of people who would have migrated to another country and who now switch to X.

Reasoning from a collective level is always much more difficult to do. I believe it should not be done. There is no “collective” of migrants; they are distinct human individuals with their own goals and motivations. Economic analysis should always proceed on an individual level. However, it would be convincing to me to say that even if some people don't like more benefits, that number is smaller than the number of people who do like them or at least do not switch away from country X. In that case, the second level would also conclude that, on net, a change in benefits increases migration to X, or at least leaves it constant.

The third level relates to the statistical measurement of the effect of state benefits on migration. As we mentioned above, a marginal euro does not move many people. Even if we trust that no-one dislikes the tiny positive increase in benefits, there may only be a tiny portion of people who are moved by it. In frequentist statistics—the dominant paradigm—a statistical test can only assert the existence of a difference, not the non-existence of a non-difference. Absence of evidence is not evidence of absence. Any empirical measurement of the causal effect of an increase in benefits on migration has to confront power issues. That is, it may be true that there is some change in migration when benefits change, but the statistical test may be unable to detect it! Even if we ignore the possibility that false effects are detected, the other type of error—that a true effect is not detected—is alive and well. This is true for any empirical study, even the most rigorous and well-constructed study. If we consider that some studies on “pull factors” have found small effects, it is not unlikely to presume that if the true effect truly is small, less well-powered (e.g., smaller) studies find no effect. Or, rather, that they are “unable to reject the null hypothesis that no effect exists.” For small effects, this can be very likely. Empiricism is not a panacea.

In summary, there is much reason to suspect that pull factors matter in a nonnegative sense. It may be a very small effect, though. That is an empirical question. But, a priori, we would certainly assume that a nonnegative effect exists.

Marginal thinking gives us directional hypotheses on causal relationships. It's important to understand the assumption of leaving all else unchanged. If it is for some reason impossible to leave all else unchanged, our conclusions can be meaningless. For example, trade policies often involve large “package deals” where not only the tariff for wheat is reduced, but indeed the import of chickens is banned. While the reduction of the wheat tariff is likely to make consumers better off—reducing prices for wheat—the ban of imported chickens may make consumers worse off. In that case, the causal relationship should not be analyzed in terms of only the wheat tariff, but rather in terms of the whole package. This can be profoundly challenging; the conclusion may be much less clear.

Similarly, a decrease in VAT does seem to make consumers buy more. However, what the decrease in VAT and concomitant increase in demand means for the supply of goods depends on the market structure. Such questions require further analysis. Generally, decreases in VAT do increase consumption, though.

These ideas highlight something further. Economic theory is a priori reasoning. The role of theory is to form expectations about causes and effects. Economic thinking is clear thinking.

A theorist, however, needs to be aware of their own cognitive limitations. The theorist necessarily relies on their own intuition in building mental models of the economy. Above, I used my own enjoyment of money and benefits to assume that others have similar preferences. There is no method besides introspection by which I could have made that inference.

If we were to analyze the effects of minimum wage legislation, it would be easy to conclude that an increase in the minimum wage decreases employment because firms substitute away from labor. And indeed, that is true if the market for labor is fully competitive. However, if there is only a single employer or very few employers, then minimum wage legislation can actually move the labor market closer to the fully competitive equilibrium, increasing employment. This is because the “monopsony” (only one company employs everyone) or “oligopsony” (few companies employ everyone) use little labor to produce their output. While minimum wage legislation reduces the industry's profits, under these lower profits it can now be profit-maximizing to employ more labor! In that case, our intuition can fail because reality is richer than it.