Your model is wrong – and it kills people

A common response to a homeless person asking for money is “get a job.”

When people say this they reveal some assumptions about the way they believe the world works. Some of the assumptions revealed by “get a job” are

  1. there are enough jobs for everyone
  2. everyone is physically and mentally capable of the available waged labor
  3. available waged labor pays enough to maintain housing

These assumptions are easy to disprove. Telling someone to “get a job” is a moral judgement based on false beliefs.

Jacobin had a great article several years ago on the moral philosophy of economics that I’ve been thinking about lately.

When the findings, predictions, and prescriptions of economists are in the news, they’re often presented as if they’re natural facts about the word. Maybe someone has made a calculation error somewhere, but there are rules and equations and pretty much anyone would have come up with the answer they did.

But economics has more in common with the “get a job” example than a calculation of soil-water moisture or how far you can drive on a tank of gas.

Economists are making moral judgements based on a model of how the world works that may or may not be true. The very models used to come up with those numbers, even to ask the questions, represent philosophies and moral judgements that you may disagree with – or even be able to disprove.

When I run the numbers on my budget, I’m building a model of sorts and that model makes declarations about the world – small ones like “I have a job,” “I buy a lot of cheese,” “I get paid on the 15th,” and big ones like “wage labor exists,” “people can charge me rent,” “I can exchange my paycheck for goods.” All of these things are true right now in the world, but all of them can be changed.

At some point in recorded human history, all of those statements weren’t just not true, but they didn’t even make sense. The world where a household budget makes sense is one that is real, but it is also one that we constructed. And we could, collectively, change it if we wanted to.

There’s a aphorism that “all models are wrong, but some are useful.” In evolutionary biology, the Hardy-Weinberg principle states that “if nothing changes, nothing changes”, or more formally:

allele and genotype frequencies in a population will remain constant from generation to generation in the absence of other evolutionary influences. These influences include genetic drift, mate choice, assortative mating, natural selection, sexual selection, mutation, gene flow, meiotic drive, genetic hitchhiking, population bottleneck, founder effect and inbreeding.

We can approximate the conditions under which Hardy-Weinberg equilibrium occurs, but it’s mostly useful as a baseline – something to measure against. When an allele isn’t in HW equilibrium, we know that some evolutionary force is acting on the population. It’s simple and very fruitful, occurs in the real world, and is based on assumptions that are physically possible (or at least approximately physically possible – e.g. infinite populations are impossible, but populations so large that they behave like infinite populations are very possible).

At first blush, HW equilibrium sounds kind of like the economist’s “ideal competitive benchmark” model. But the ideal competitive benchmark is presented as a goal, can’t ever be produced, and if it were it would be terrible.

All models in social science are unrealistic. But the “ideal competitive benchmark” (the Arrow-Debreu world and its family of general equilibrium models) is not just unrealistic. It depicts a world that is neither possible nor imaginable— and yet it is also undesirable. Here are some of its assumptions: All markets must be perfectly competitive (whereas most of ours are not); if such a world existed, the requirement of perfect competition would rule out any division of labor or long-run economic growth.

There must be an infinite number of futures markets— one for every good in existence, delivered at every future date, for the rest of time. And yet, in the model, time doesn’t really exist: all economic decisions for all of human history were made in an auction at the beginning of the world.

Moreover, far from being harmonious, this theoretical world has been discovered to be chaotic— perpetually in random motion, never actually arriving at any of its “optimal” configurations except by accident. This finding alone nullifies the very meaning of the theory.

For a model to be useful, it has to tell us something about the world. The ideal competitive model doesn’t seem very useful, yet it’s been used to make policy decisions that affect us all.

But economics is a big field, with lots of models and assumptions underlying different schools of thought that overlap and conflict. Consider economists discussing full employment instead of an arrogant businessperson telling a poor person to get a job –

On the surface, Keynes’ critique of neoclassical economics (which he called “classical”) was much more limited in scope than Marx’s. His fundamental innovation was the theory of effective demand: the idea that employment is set by total spending, so that the market system has no automatic tendency to settle on full employment. Keynes himself was keen to stress that the General Theory was radical only on that particular point, and that once the state intervened to assure full employment, “the [neo]classical theory comes into its own again.”

Yet in order to reach that conclusion, Keynes had to challenge conventional economic theory on fundamental points, which lent themselves to more radical readings— and brought them into contact with Marx. Neoclassicals had held that full employment was ensured by the workings of the market, that the wage functioned like any other price, rising and falling to align the supply and demand for labor (at least eventually, or once wage rigidities and other imperfections were swept away).

If you’re a neoclassical economist, you think that full employment can be achieved and the way to do that is to get rid of things like collective bargaining and minimum wages. Full employment doesn’t mean to an economist what it does to you and I – and our official policy for decades was to increase unemployment because some economists believed it would keep inflation low.

But Keynes established that the wage was not like any other price— it constituted not just the employer’s cost but the bulk of society’s income, out of which spending and demand for goods was generated, so there was nothing preventing a persistent equilibrium of substantial unemployment.

Rather than depending on the wage, the level of employment depended on effective demand. This, in turn, danced to the tune of investment, so that employment today depends on firms’ expectations of profitability in the future— expectations held more or less confidently, but always fallible. Keynes saw human beings as coping with fundamental uncertainty about the future. This could leave market outcomes wild and unpredictable, so that free-market price flexibility might lead not to harmonious equilibrium but to chaotic results.

And then the article starts talking about income distribution and puts it smack dab at the heart of economic theory – despite it being ignored or treated as just an outcome, not an input.

By the same logic, [Keynes] rejected the neoclassical notion that workers bargain over their real wage— that is, over units of consumption. Lacking knowledge of how much goods will cost in the future, workers can evaluate only their relative wage; and that brings the question of income distribution into the heart of economic theory.

All of this opened the way to what neoclassical economists resist most militantly: indeterminacy, with all its radical implications.

Unlike the neoclassical vision, in which income distribution is fated by existing technologies, preferences, and endowments, in this vision it is a process of active conflict. The incomes of different groups, rather than smoothly adjusting to shifts in supply and demand, tend to be the baseline around which the rest of the economic system adjusts. The income distribution is treated as an evolutionary process, shaped by norms and institutions inherited from the past, which change as a result of extra-economic events— that is, history, politics, institutions, and struggle.

The neoclassical vision of income distribution rests on two very shaky assumptions. First, unlike Ricardo and the other classicals, it simply assumes that firms are able to respond to changes in the prices of different factors— the different kinds of labor and capital— by freely adjusting the various proportions in which they’re used. Without that assumption, labor may literally have no marginal product, and the same would go for any other factor, or any particular type of labor. In that case, the Marxian or Ricardian conclusion would hold: the wage would be whatever workers could wrest for themselves, and profit would be whatever was left over.

Of course, it’s fine to build a simplified model with unrealistic assumptions and then see what happens when the assumptions are varied. But at some point, it seems, mainstream economists largely forgot that this assumption of “differentiable production functions” was a simplification— let alone one of questionable realism. [emphasis mine] As a result, in today’s economics literature it’s almost never questioned, and textbooks don’t even alert students to the issue. Yet as a general supposition about how production works, it is, of course, unrealistic: What are you supposed to do if your labor consists of ditch-diggers and your capital consists of shovels? How exactly do you vary your proportions of labor and capital?

In a series of works over the past two decades, Michael Mandler, a University of London general-equilibrium theorist with impeccable neoclassical credentials, has shown that once economic decisions are pictured as being made sequentially, as in real life, ownership patterns turn out to evolve through time in highly specific ways— and they systematically gravitate toward precisely the kinds of patterns that generate indeterminacy of factor prices.

As a result, the central problem with marginal productivity theory that John Hicks recognized in the 1930s has never gone away: without the arbitrary assumption of freely differentiable production functions, wages and profits are not fixed by technologies and tastes. They are set by “something else”— something outside the competitive model.

In my field, we’re often encouraged to write our ideas and hypothesis as equations. We don’t do this to claim they’re natural laws or anything, but because they’re tools for investigating those ideas and the assumptions that go into them. We emphasize that just because something is true in the small world of our model doesn’t mean it’s true in the real, big world, that just because something in our model does hold in the real, big world doesn’t mean it always holds.

Our assumptions and simplifications matter. My budget is a good model of my household spending and how much I can save, but that model becomes a bad model if the government introduces a Universal Basic Income, or my sister gets sick and I have to fly across the country regularly to take care of her, or we have a socialist revolution, or climate change causes food prices to spike.

Economists seem to have confused the small world of their models with reality and it’s hurt us all. Models are tools that help us understand and interact with the world. They are never perfect representations. If we don’t acknowledge their limitations when we use them, models aren’t just wrong – they’re dangerous.