Performing Consumption Smoothing

Performing Finance

I think about performativity and economics often. It seems a little grad school bullsh**ty, but in finance it can be a killer. Does Black-Scholes work because we all assume it works? Here’s a practical example: You believe that there is ‘momentum’ in stock prices – that stocks that go today up will go up more tomorrow, and stocks that go down today will go down more tomorrow. So you at your hedge fund – or more accurately, a computer you program – buys stocks that are going up, and sell stocks that are going down. Since there is more demand for stocks on the rise, the price increases, creating momentum! The same effect happens for those on the way down. So by acting on this model, you create the conditions for it becoming true. This may have happened to the quants before, and as trading becomes more algorithmic and computer based, it may happen again. (Don’t worry, I’m working on the prospectus for my Judith Butler Hedge Fund right now.)

If the momentum trading goes over your head, replace stocks with housing. Was housing in your neighborhood valuable because everyone assumed it was valuable? Did believing housing would always rise create the conditions for a massive rise in housing prices? Thinking about whether this is an amusing sideshow or a central problem for financial markets should be in the back of any critical thinking quants’ mind.

Consumption Smoothing

I bring it up because I saw the most fascinating thing on The Atlantic’s business blog. By Connor Clarke, who I really enjoy reading, Let College Students Get Into Debt:

But why on earth don’t we want college students to take on debt?…

The second problem is more specific: if the the point of credit-based consumption is to bring lifetime consumption more in line with lifetime income — as I believe it is — then college students more than anyone else should be getting into debt. They have the largest gap between between desired present consumption and expected future income. That is a gap society can and should try to regulate, but it also one society should try to close.

Professor Richard Serlin attacked this from the financial planning side. (By the way, I just discovered blog from internet reactions to Connor’s piece, and it is highly recommended.) He wants to do it in 5-parts, and I encourage it, because student debt is a major problem for my generation, and how to frame the debate is important. I want to add my part to it.

One important economic history point of the 20th century was that many people thought the United States would go into another Great Depression immediately after WWII. We didn’t. In fact, as a result of many factors including perhaps a Fordist social contract and Keynesian Planning, we generated a large middle-class centered around consumerism (this lack of an immediate second Great Depression is one reason economists have the prestige they do). This middle-class was brand new, and economists were not sure how’d they behave. Would they spend all their new money immediately? Would they spend the minimum and save a lot? They didn’t have the tools for this kind of modeling in the 1940s.

Then along came Franco Modigliani and Milton Friedman who said that consumers will spend as if they are smoothing their lifetime income. This was a guess, a model to try and explain how people behave. I’m assuming Connor came across the idea of consumption smoothing from an economics class or from his familiarity with economic research. What was once a way to guess how people would behave in light of economic decisions now informs people on how to behave in light of how to make economic decisions. Connor is acting, and advising others to act, as if this theory was true. This is why I don’t think of economics as a science like physics is a science. Gravity did not change the day we wrote down the equation for gravity, and has not changed since.

And what is really cool is how away the theories can get. I read Connor as saying students at age 20 should smooth consumption by borrowing against age 30. Perhaps even age 50? Here’s Milton Friedman saying (1957) “The permanent income component is not to be regarded as expected lifetime earnings… It is to be interpreted as the mean income at any age regarded as permanent by the consumer unit in question, which in turn depends on its horizon and foresightedness.” That’s a way of saying at age 20, you should smooth against your college income, the income around 18-22, the income you know directly and are comfortable with the horizon, not the income around ages 18-50.

Consumption in Practice

Now back to income smoothing. We know that people do not perfectly smooth their incomes (Larry Summers, from that paper – “it appears that consumption smoothing takes place over periods of several years not several decades”). Here’s a graph:

Now here is a good question. If you are an economic researcher, what do you do? Do you assume the theory is wrong for lifetime income, and that consumption is more associated with short-term horizons or perhaps sociological phenomenon like having a child, or buying a house, or seeing the car your co-workers and neighbors drive? Or do you assume it is right, it is just that we are missing something?

Against Student Debt

Two quick thoughts in “the theory is right but we are just missing something” vein that are relevant to whether students should consumption smooth. This goes beyond credit cards, to student debt more generally. One is that imperfect knowledge is a big problem. At age 18 you have no idea what you are going to make in the future. You do not say “At age 56 I’ll makes $56,652 discretionary income.” (Wouldn’t being able to predict that be the lamest superpower? And I’m a big fan of lame superpowers.)

Here are two computer simulations of an individual, taken from the excellent Learning Your Earning paper (as is the graph above – thanks Econ Dave for the recommendation!). The income process is the same for both, but in one he knows his income perfectly, in the other he doesn’t update his learning about his income (he experiences his income as transitory, instead of permanent, so doesn’t update his expectations of earnings). Think of it as a dial of uncertainty between perfect and no-knowledge for a person. In these cases, you can see income smoothing only happens with perfect (“full”) knowledge.

So uncertainty of what you’ll do, how much money you make and how reliable it will be should be a major hedge against taking on student debt. Indeed, taking on too much student debt precludes certain life possibilities, forcing you to emphasize certain types of labor over others, where you are now targeting your income against your consumption, when the model is supposed to predict the other way around! (I think this is a major problem with how lawyers are educated in the United States, where working-class law students have to take firm jobs to pay down debt instead of work that may be more class-aligned.)

The other explanation for this effect is that young people save a lot, more than is optimal under perfect knowledge of how your life unfolds, because they want to have a precautionary savings fund early to ‘self-insure’ against the volatility of their income streams. This is the result of something like Consumption over the Life Cycle, where “we find that consumer behaviour changes strikingly over the life-cycle…Young consumers behave as buffer-stock agents.” Optimally, we would want young consumers to save a ton, the better to insure against the ups-and-downs of life, not spend as if they were a 50-year old.

I think that gets into a bigger problem with phrasing college as a consumption smoothing problem, instead of a good we invest in as a country like clean air, water and roads, or private college debt as a rational choice, instead of a corrupt nasty little GOP-insider industry that should be smashed like a cockroach. But that’s for another time!

This entry was posted in Uncategorized. Bookmark the permalink.

8 Responses to Performing Consumption Smoothing

  1. Jeff says:

    “Gravity did not change the day we wrote down the equation for gravity, and has not changed since.”

    Not to pick nits, but yes it did. To name a few developments in gravity since Galileo’s “Ooh, things fall towards the center of the Earth!” experiment: Gravity wells, singularities, time-space fabric, warping of light by gravity, gravitons, dark matter.


  2. Mike says:

    But Gravity hasn’t changed! Our equations and understanding of it have changed, but gravitons and dark matter are the same now as they were 10,000 years ago. (Unless you want to get into the deep end of the construction of science, etc.)

  3. pebird says:

    Great, so governments, hedge funds and banks can’t figure out risk over a 3-year horizon with all the experts, systems and geeks in the world. It makes perfect sense that college students all by themselves (alright, they have an income-smoothing iPhone app) will figure out their personal risk profile over a 10 year horizon. Big surprise after seeing the tuition bill they try to save a little more than Larry Summers optimal model.

    What do these people smoke – it wasn’t there when I went to college?

  4. Terrry Ivanauskas says:

    And it is just sadly funny that these are the people who say that it is ok to bailout banks and automakers, but it is not ok to put taxpayers’ money in health care and education. It is ok to use money from everybody to keep alive financial institutions, but it is not ok to use money from everybody to improve health care and education. Because it is a serious threat to the society the financial system collapsing, but apparently things like this are not. Oh, yeah, sorry, I forgot, people who had their bonuses guaranteed by bailouts don’t go to Kings Hospital. I seriously don’t understand the values these days.

  5. Amol Shelat says:

    Wonderful post. It’s quite rare to find Judith Butler and hedge fund mentioned in the same sentence!

  6. jult52 says:

    Could you edit that penultimate sentence? It isn’t clear to my eyes.

  7. John the Cheap says:

    This was an interesting post. What you are describing in the first bit are positive feedback loops — in other words, the sign of the signal being fed back to the input is the same as the sign of the input. (This is what makes PA systems howl.) Any system with (overall) positive feedback is inherently unstable; there will (almost) always be some condition that makes the loop gain >1 resulting in a runaway situation. The fact that adaptive humans are involved makes this pretty much a certainty.

    “The other explanation for this effect is that young people save a lot…”

    I suspect the evidence is against this explanation.

    BTW, Larry Kotlikoff is selling a personal finance modelling tool that is heavily oriented toward consumption smoothing. I bought a copy a few years ago to play with, but I didn’t pay the annual sub so I don’t know what the updated version is like. However, the one I had suffered from a lack of transparency in the results (I had a hard time interpreting what it was telling me) and an inability to readily model certain scenarios (for example, getting at the safe withdrawal rate given one’s initial retirement account balance and investment choices).

  8. Pingback: Goolsbee on Supply-Side and Subprime. « Rortybomb

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s