Would there be an assumption of a global optimum if women had invented economics?

I had yet another long conversation with a female friend about life and its struggles. I’ve had ten thousand such conversations previously. We spoke about the difficulties of raising children (I don’t have children, but she does).

We spoke of the task of getting children to school. I recalled that when I was a child, as early as Kindergarten, I walked to school without parental supervision. My elementary school was a bit more than a kilometer away. It was considered an important step towards adulthood that 5-year-olds should find their way to school on their own. But of course, I didn’t walk alone, I walked with all of the other elementary school kids who were leaving their homes and walking to school at the same time. We had the protection of a herd. The suburb where my parents lived was built in 1964, and they bought the house that year, and for the next 20 years everyone in the suburb was having kids, so it was a neighborhood with uniform demographics. We all walked to school together.

I’ve read that in places like Japan and Finland and much of rural South America and Africa it is still entirely normal for children to walk to school on their own. But it is no longer normal in the neighborhood where I grew up. In the modern suburbs of New Jersey, if a parent told their 5-year-old that they had to walk to school on their own, the parent would promptly be arrested for child endangerment. It just doesn’t happen any more.

Why? Some of that is because of the perception of danger. Crime in my hometown is very low, it’s always been very low, but somehow parents felt it was safer a few decades ago, and now there is the universal feeling of danger. Some of that is because of the changing demographics. I should emphasize, when I was kid, the neighborhood had blacks and hispanics and whites, but they were all uniformly middle class. Every family owned its own home, it was unthinkable that anyone would rent a home. Nowadays the neighborhood is much more heterogeneous: some people rent, other people own their homes, there are some young couples with kids, also some single parents, also lots of older couples whose children have grown. I assume it is the heterogeneous nature of the neighborhood that causes parents to view it as unsafe. When I was a child, every single family in the suburb had children whose age was within a few years of mine, so everyone had an incentive to ensure the streets were appropriate for small children. Nowadays, only a minority of families are raising children. The absence of child-raising families changes the culture in ways that are uncomfortable for child-raising families.

So what is like to raise children now? Is it easier now to raise children than it was 40 years ago? The economy has grown, so we have a nominal notion that everything should be easier — if there is more wealth, then surely that wealth solves some problems? But in fact, some things are much harder. Raising children is much harder. The children have to be supervised, and that puts demands on parent’s time. When I was a child, me and my friends would disappear into the woods early on a Saturday morning and we would reappear in time for dinner. Nowadays, parents are nervous about going 9 or 10 hours without knowing where their children are.

And of course, some of these changes have been underway for a long time. My great-grandmother raised 16 children, and got them all good educations. That was in the Austro-Hungarian Empire, in the late 1800s and early 1900s. It would be nearly impossible to raise 16 children now, and give them good educations, unless one was extremely affluent. My great-grandparents were not especially well-off.

Even when real wages go up, it doesn’t mean anyone’s standard of living has gone up. An unstable job can be a terrible emotional drain — one is always worried that one will soon be unemployed. Especially if one has young children, a stable job is often more important than a high-paying job. Patricia Resnick sums this up nicely, comparing 1980 to now:

In certain ways, things are even harder for women now, just because of the advent of all the tech stuff, which didn’t exist in 1980. Anyone would kill for a nine-to-five job now. We’re all expected to be reachable and ready to work, 24 hours a day. Women are expected to work and parent, and they’re expected to be available. That’s changed for the worse, and I would want to look at that.

People are struggling. You could squeak by back then on a secretarial job—you could raise a family. People have to do crazy things now. People are bringing their kids to work when they’re sick because they can’t afford to take care of them. Many more people are one or two paychecks away from losing everything. Those pressures are much stronger.

One of my pet peeves are people who try to reduce a vector to a scalar, with no admission of the information loss. (If you don’t know the terms “vector” and “scalar”, please assume that a “scalar” simply means “number” and “vector” means “a bunch of numbers that only have meaning when they are grouped together). For some reason, this has become accepted practice when it comes to estimating our standard of living. This drives me absolutely crazy.

Suppose the economy grows 5% but violent crime goes up 10%. Has the standard of living increased?

Suppose the price of gasoline declines 30% but people stop going to church. Has the standard of living increased?

Suppose the price of cell phone plans decline 90% but the people who speak your language become a minority, and people who speak a different language become the majority. Has your standard of living increased?

Suppose your wages increase 10% but the amount of air pollution increases 200%. Has your standard of living increased?

Suppose your wages increase 20% but the government changes the law to make it easier for companies to fire workers, so you face a greater risk of losing your job. Has your standard of living increased?

In every case, we are talking about a vector that can not be reduced to a scalar without information loss.

Note that some of the dimensions have an obvious good and bad, but others don’t. Less air pollution is good, and more air pollution is bad. That much is clear. But what about more people moving to your neighborhood who speak a different language? It is a chance for you to learn a new language, which is probably good, but it might lead to frustrating moments when you go to a store to buy something and no one can understand you — this is a dimension that has no obvious good or bad gradient, so things can only get different, not better or worse. But certainly it has some effect on our standard of living. The same is true of people going to church — there is no obvious good or bad, but it certainly changes one standard of living. When society goes from 70% people attending church to 20% people attending church, the culture changes in ways that make some things easier and some things more difficult. But it would be impossible to say that society was better in some absolute sense, or worse off.

I am making two separate points in this essay, and I worry that I will confuse some readers. One point is about the statistics we need to have a conversation, as a nation, about our standard of living. The other is about the much more complex set of factors that we track as we try to keep track of whether our life is going well.

Let me talk first about the the statistics we use to make sense of our national situation. These will necessarily be a simplified set of numbers. But it should be more than one number. Our lives are complex, and full of contradictions; we should make some effort to discuss this honestly. As a nation, we could probably agree on a vector with 20 or 30 numbers which could track our national standard of living. We could track unemployment, underemployment, labor participation, air pollution, water pollution, crime, civic involvement (any kind of NGO participation, including church attendance), voter participation, public perception of corruption, and a dozen others. It wouldn’t be that hard. All of these statistics already exist. They only need to be published as a vector, and talked about as a vector. This would increase the intellectual accuracy of our public discussions. As I said, it drives me crazy that it is currently considered coherent to say things such as “Even though crime increased 3%, our standard of living hit a new high point.”

But that simplified set of numbers is only for national conversations. On a personal level, we tend to intuitively consider thousands of different factors when we consider the standard of living of our own lives.

For some reason, women seem to sense this more readily than men. My friend and I discussed life options and the many tradeoffs people face:

Is it better to have children or retain the autonomy of being childless?

Is it better to fall in love or avoid the risk of being hurt?

It is better to work for money or devote all of one’s time to one’s children?

It is better to continue one’s current pregnancy, or get an abortion and try again later when one’s life is more stable?

It has been mathematically proven that the efficiency of a utility function declines as the search space increases. That is, the utility function becomes more likely to get stuck at a local optimum, rather than the global optimum, as the space of possibilities grow, and given an infinite search space, a utility function has a no chance of finding the global optimum. Given an infinite search space, there can only be local optimums. (This has lead to a new research focus on novelty functions, which I find fascinating, but which I won’t discuss here.)

Modern economics was invented by men and has largely been advanced by men. As an intellectual project, it got started (mostly) in the 1700s, in an era without computers. Because of the lack of computers, much of economics focused on equilibrium models, because this is something a person can do in their heads. The math is relatively simple. A classic example is supply and demand. If a village has x bakeries and y demand and bread costs z, how will the price of bread change when a new bakery opens? This much one can work out on a piece of paper.

Modern computers allow new approaches, including “agent-based simulations” and “permanent disequilibrium.” These are great advances, since we all know that on some level the notion of global equilibriums is a lie. The economy is never actually at equilibrium, it is always in motion. But the new approaches have made slow progress in changing the canon of economics. Global equilibriums are still the norm, and everything else the exception. (Also, when discussing the national situation, sometimes the old approaches are pragmatic. Maybe permanent disequilibrium is more accurate, but one pays a high price, in terms of complexity, to try to mine insights from that approach. After the crisis of 2008, Paul Krugman used simple equilibrium models to offer insights about what the government should do. Simple equilibrium models are useful when options are limited, and the government certainly had limited options. The situation is different when we consider our own lives, exactly because we have, in some sense, an infinite set of personal choices. We also understand our preferences at a fine-grained level that can never be reconciled to a national conversation. We are not an aggregate.)

Economics is a fascinating intellectual tradition and I’ve spent a lot of time studying it because it appeals to the rational side of my brain. I love the idea that one can model much of the world with simple rules. But I constantly remind myself that the model is not reality, because mistaking the map for the territory is one of the greatest intellectual sins that we can commit.

After I was done talking with my friend, I wondered what economics would look like if it had been created by women, and built up over the centuries by women. I’m only half-joking when I say the whole tradition would be been built without the assumption of some global optimum. If you start with the assumption that there is no global optimum, then you chase after a different question: what sort of life do you get chaining together one set of local optimums, versus another? This approach requires some arbitrary filter (though not any more arbitrary than assuming a global equilibrium that we know doesn’t actually exist). Suppose we agree that 12 options are a reasonable number of options to consider. The local optimum is not necessarily a great choice by itself, what makes it interesting is what local optimums might be near it later on. How does one chain together life choices over the course of 80 or 90 years? There is no “best” just “different.” (Or, again, there are a few dimensions which have a definite “good” pole and “bad” pole, but there are other dimensions where the poles are different, but not obviously better or worse.)

What sort of economics do we get when we reject the notion of “better”? What sort of economics do we get when assume there can only be “different”? What sort of intellectual tradition do we build, when we acknowledge that some choices take us to a new place, but not necessarily a better place than a different choice would have taken us?

Talking about the limits to the efficiency of a utility function can be thought of as another way of discussing The Curse Of Dimensionality. How many dimensions does life have? Some people have a mania for shrinking life down to a small model. I assume this protects them from anxiety and depression. Life seems manageable if you can convince yourself that there are only 5 or 6 dimensions along which your life needs to be judged. Perhaps depressed people are simply honest people. An unmasked confrontation with life reminds us that it is, for better and for worse, richly dimensional. [1]

This feels much closer to how women, and sensible men, talk about their lives than what standard economics preaches. Life is a series of local optimums that need to be chained together over time. There is no global optimum. There certainly isn’t anything resembling equilibrium. There is never a point where we get to rest, unaffected by what went before and what will occur later. There is never a moment when things stop moving.

Despite my recent conversation with my friend, I don’t mean to put this entirely in terms of gender. As I said, I am half-joking when I say everything would be different if this intellectual tradition had been built by women. There is a grain of truth there, though it is also true that in the era before modern computers, the simplicity of equilibrium models meant they necessarily attracted a lot of attention. (And yes, sometimes the math was actually very complex. But humans could still do it, unlike an “agent based simulation” which tracks the actions of 10 million agents — such a thing can only be done by a computer.)

Honesty demands that we stop reducing a vector to a scalar without admitting the amount of information lost. Honesty demands that we stop pretending our lives contain equilibriums when we all know that everything is in permanent disequilibrium. Honesty demands that recognize that our lives consist of a series of local optimums chained together, rather than any global optimum. It is some kind of secular religion, to believe there is some nirvana, some global optimum, that we can all achieve if only we are rational enough to maximize our utility. Common sense, and lived experience, reminds us everyday that real life is nothing like that soothing model.

*[1] — I wanted to make a swipe at rationalists, but the swiple might be distracting, so I will confine it to this footnote. When I read forums where rationalists congratulate each other on being so rational, what I see is people (mostly men) building very simple models of their lives. They then solve the model rationally, and that justifies their feeling of being rational. But they often fail to examine the process by which they choose the variables for their model. Many of these variables have to be chosen entirely on the basis of preference, and preference tends to be non-rational — not irrational, but there is often no criteria for judging other than pure innate preference. Examples might include what kind of food do you enjoy, what kind of sex do you like, what kinks do you have, what body types do you prefer to have sex with, how much do you need to take risks, how easily traumatized are you, do you wish to have children, and so much else. Once they’ve picked out a few variables, by which they judge their lives, they might make decisions with all the rationality of someone solving a system of linear equations, and at this phase of their decision making process they are right to call themselves rational, but the overall process can not be described as rational since it rest on the non-rational roots of personal innate preference.