27 Nov 2017

Why Economic Models are Bullshit (Part II)

Hello readers!

Previously, I wrote a post entitled “Why Economic Models are Bullshit (Part I)”. Therein, I covered one of the problematic areas of macroeconomics: namely, that students are badly taught the subject. But that does not tell us enough about the other side of the coin—the reason why macroeconomic models are, by themselves, problematic (aka “bullshit”). In this post, I explain just this.

Before I begin, and in case you are wondering: yes, I am progressing with Fallen Love. I am about halfway through the revision process; I will post an update later on. For the time being, I am extremely busy both with the book and with my university studies. Consider this my final update for the month.

Anyway, onto the topic of today’s post...

The Follies of Macroeconomic Models

I am not the first to criticise an economic model, and in particular, I am not the first to criticise the discipline of economics as a whole. Some critics speak from a position of ignorance; they sometimes make good points, but cannot articulate their criticisms beyond relatively vague generalities. (A few examples: economic models don’t work because you can’t put people in an equation. Or, economics is not a science. Both hold a grain of truth, but are not extended upon beyond platitudes.)

Some critics, however, are economists. Thomas Piketty and Ha-Joon Chang are good examples of the latter, though the venerable Steve Keen is my personal favourite among the rebels. If you have had read these economists, you may detect some of their criticisms among my own; though I quite fancy that I am original. Anyway, vanity is a vice, so allow me to get to the meat of the arguments...

Problem No. 1: economic models are not dynamic. To clarify, by “dynamic” I mean that the models do not explicitly refer to time—either graphically, mathematically, or even in argument. Some economists (especially those who have never done a proper science, like physics or chemistry) seem to think that economic models are dynamic because... time is in there somewhere.

Of course time is present in these models, in some way—they wouldn’t make any sense otherwise. (They make little enough sense as it is!) The problem is that this relationship is dreadfully ambiguous; there is absolutely no clarity about what happens when, and this leads to a number of conceptual errors and oversights.

IS-LM graph

The above graph is a clumsily drawn example, representing an IS–LM model of a small open economy, with e (nominal exchange rate) on the vertical axis and Y (national income) on the horizontal axis. The two sloping lines are IS-curves, while the vertical line is an LM curve. The shift of the IS curve outwards represents a ceteris paribus fall in taxes.

The moral of this story is that, in a small open economy with perfect capital mobility, fiscal policy doesn’t work: you can’t change national income with fiscal policy measures. Even if we assume this is true (there are economists who do not agree with this assessment), the problem is that the graph is extremely obtuse.

There is a whole time dynamic involved here. First, a fall in taxes leads to an increase in Y, which in turn leads to an increase in r (interest rate) as the money supply is fixed. The increase in r leads to a situation where r>r* (the domestic interest rate is higher than the international); this leads to an influx of capital, which in turn drives up the exchange rate e. The appreciation of e leads to a fall in NX, which brings Y back down to the initial Y.

None of that is shown on the graph. We only see a new equilibrium point at e2 and Y. Great explanation there!

Problem No2: economic models confuse cause and effect. If you look at the IS–LM graph aforementioned, you might be forgiven for thinking that somehow Y (national income) affects e (the nominal exchange rate). In the sciences, we put the independent variable on the x-axis, and the dependent variable on the y-axis.

To peruse one of many examples from physics:

Force-extension graph

The story here is pretty straightforward: you apply a force, and the material stretches in a particular way dependent on its material properties.

Occasionally in physics, some graphs don’t follow this convention, usually for reasons of convenience.

The problem is that nearly all economics models have it backwards: they put the independent variable (the causation, the mover) on the Y-axis, and the independent variable (the observed change) on the X-axis. This small change makes economic graphs unnecessarily confusing. In the IS–LM model, e affects Y because e affects NX and Y is dependent on NX; however, there is no clear relationship going the other way round.

In formal logic notation,

(x → y) ≠ (x ↔ y)

This says that (x implies y) is not the same as (x and y imply each other). Or to put it in more comprehensible terms: if I sleep through my alarm I will be late; but if I am late, that doesn’t necessarily mean I slept through my alarm (I could have been stuck in traffic!)

Problem No3: economic models make overly idealised assumptions. This is a big one. Economists say that the art of economic modelling is choosing good assumptions; but if so, economists must be terrible at their job.

Let’s look at the previous model I showed: the IS–LM model under conditions of a small open economy with perfect capital mobility. You may now observe that, actually, well—capital isn’t perfectly mobile. You can’t do a runner with a house. What’s more, houses take time to sell (again, dynamic systems!) and the resale value is not always high (risk element). In many parts of the world, there are restrictions on foreigners buying houses.

Because the assumption of perfect capital mobility is wrong, the aforementioned conclusion is wrong as well. Fiscal policy does have an effect on national income—just look at the UK under austerity. It is thought by many economists that Osborne’s economic policy cost the UK a lot of lost income growth. The Sterling did not depreciate and net exports remained pretty dismal (the former stayed high and the latter stayed negative).

A more useful assumption would have been: assume that some capital assets are mobile while others are not. Determine the share of mobile-assets for the economy you are looking at. This way, you get a much better grasp for what’s actually going on.

Problem No4: vagueness. This is a problem that I have rarely seen mentioned, perhaps because it is of a slightly more philosophical nature. Essentially, what I have noticed in economic models is that they can be quite unclear as to what a concept or variable is referring to.

Take the example of r*, which represents the going interest rate across the globe. Or even just r, which represents the going interest rate in a national economy. My question is: which interest rate does it represent, exactly? Investments have many rates of return. We all know that some investors make a fortune on the stock market; others make a loss. Bonds have different returns based on their maturity period.

If we just take a weighted mean of all these different interest rates, we risk missing some important constituent details.

If we look at the globe, we... observe that there are many interest rates, across both private sector and government investments. Even if we confine ourselves to only government bonds, we see that there are large discrepancies based on the countries’ riskiness (Argentina or South Africa have higher interest rates on their bonds than Germany or the US).

At this point, economists just thought: “Aha! We can model interests rates as being r* + P, where P is the risk premium.”

Except it’s not that simple; the concept of risk premium is itself vague. How do you quantify a risk premium? No one knows. Investors make investment decisions based on their perception of that risk, but the risk itself is uncertain; the interest rate we observe is just the expression of a social belief, not some neat numerical correction.

To put it in philosophical language, the ontological status of the risk premium (and numerous other macroeconomic concepts) is misunderstood. And the consequences are not just philosophical; they can lead to a number of conceptual errors with serious policymaking implications. One prominent example is in neoliberal economics, and its belief in the divine importance of the market price.

In a debate about rent prices in London, the neoliberal economists might say: “All these social housing schemes are nonsense. Why should the state interfere and distort the housing market price?” The use of the word distort is very important—it suggests that the market price is almost like a physical quantity, a reality that should not be meddled with. In reality, of course, the housing prices of London are really just a reflection of the (deluded) expectations of property owners on future prices, among other things.

Problem No5: the role of risk, uncertainty, and expectations. This is another area of economics that is under active research, and in which we are starting to see improvements. I’ve decided not to go detail here; the topic is quite technical, and anyway, I’m doing research on it right now. Perhaps I will cover it in a future post. Until then, I will (again) recommend reading the venerable Steve Keen, along with various other economists such as Frank Knight and Gunnar Myrdal.

Concluding Remarks

What I have written ultimately only scratches the surface; there are much more fundamental questions to be asked about macroeconomics and its ability to accurately model and predict real world economies. Nevertheless, I think the five key problems I have highlighted constitute a good set of methodological problems with macroeconomics—and they are problems that can be feasibly solved.

My conclusion for students, policymakers, and other economists is this: presently, economic models are pretty rubbish. They are in urgent need of improvement—or else economists will find themselves stuck in the credibility crisis they are now in. But better models will demand the work of newer, wiser, and better educated thinkers.

In other words, we need a twin revolution; a revolution in the way we teach economics, to attract stronger students from a wider variety of fields, and a revolution in the way we do economics. Will the field rise up to this challenge? Perhaps. People like Steve Keen give me hope. On the other hand: there are a lot of economists who prefer to keep their head in the sand. What can I say? I hope they die quickly.

17 Nov 2017

Why Economic Models are Bullshit (Part I)

Hello dear readers!

Previously, I wrote on a number of topics, chiefly among them: my exams, and Fallen Love, my upcoming novel. Alas the former has prevented me from working on the latter; Fallen Love will probably not be finished until January, as I stated. Still, with my exams finally over, I can get back to working on it.

You may be wondering as to the title of this post. Your guess would be correct—this post is indeed a brief argumentative essay (read: rant) about economic models, on which I have spent the last week of my life revising for. I am taking both micro and macroeconomics, but this post will mainly be about macro; I will get onto why in a moment.

A Pedagogical Disaster

The simplest reason for my particular hatred of macroeconomic models has to do with teaching. That’s the simple reason, but the more complicated reason has to do with content (though the two are, of course, tied together).

To put it simply: the teaching has been disastrous. More than half of our class failed the first exam—this is in a selective university, mind you, with many of the student body having attained excellent grades in secondary school. One reason was the teacher. We had two teachers, and the first was quite dire.

“That’s one bad apple,” you say. “There are bad teachers in the world. That doesn’t mean macroeconomics is bullshit.”

This fact alone does not prove my point—except that this is not a single, isolated phenomena. Economics students across the world routinely struggle with their courses, complaining that they do not really understand it; that indeed, “it”—macroeconomic models—don’t make any sense. One bad teacher is one thing. But can the entire pedagogical structure of economic teaching be at fault?

I would argue yes. The most common complaint I’ve heard in my university is that (and I am paraphrasing only slightly) “I draw the graphs, but I don’t know what it means or why.” There are a few reasons for this. To begin with: concepts. Macroeconomic concepts are strongly under-explained. The course introduces things like “inflation”, “GDP”, “unemployment” and (my personal favourite) “money”—but these macroeconomic concepts differ significantly from the prima facie conception that students begin the course with.

A case in point: a number of students conflated the AS–AD model with the supply-and-demand model from micro economics. They even sound similar—one is “aggregate” supply and demand, the other just vanilla supply and demand.

AS–AD graph

Microeconomic supply-and-demand graph

Although they look extremely similar, they aren’t the same. The microeconomic model has P (prices) on the vertical axis and Q (quantity) on the horizontal axis—this arrangement is problematic, but I’ll get to that. Anyway, the AS–AD model has P (price levels) and Y (real GDP, output) on the respective axes. These are different concepts. Price levels are a measure of weighted, generalised prices across a macroeconomy (usually they are calculated in the form of the CPI)—they’re not the same thing as the price in a market. Y, representing real GDP, is sometimes called output, leading students to conflate it with quantity output.

Money is the worst, however. Students have no idea what money actually is (in fact a lot of economists don’t understand what money is, but students are even worse). In a macroeconomic context, money doesn’t just mean the euros in your pocket; it represents a wide range of things, from liquid assets held in bank accounts (M1) to savings accounts (M2) to more nebulous concepts of money that are too technical to go into here.

This is also why students struggle with the IS–LM model, which rests on a complicated set of assumptions about money and what money does in an economy.

Anyway, onto the next pedagogical error: mechanistic teaching and oversimplification. Our teachers presented all of these models as a series of mechanical steps, expressed in equally mechanical equations. “What happens if taxes increase under the classical model?” (Some curves shift.) “What happens if labour supply increases under the AD-AS model in the short-run and long-run?” (A complicated mess.) “What happens in the Mundell-Flemming model if, under a fixed exchange rate condition...” (I give up.)

There was very little explanation of why all these things happened. Why would a government want to increase taxes anyway? Why does the model look at these variables? What explanatory power do these models have, and what assumptions do they make?

These are all key questions that remained unanswered. This leads me onto the third pedagogic mistake: not teaching history. These models did not fall out of the sky. They were developed by economists—in a particular time and place, in a particular intellectual climate, and in a particular historical context. It’s difficult to understand these models, much less criticise them or apply them, without this precious context.

Yet, even without all these mistakes of pedagogy, there are more fundamental reasons why macroeconomic models are difficult for the students to comprehend. To repeat the title of this post: macroeconomic models are bullshit.

Conclusion Part I

I realise that you are probably tired of reading this, dear reader, so I will save my juicy critique of macroeconomics for the next post (titled “Why Macroeconomics is Bullshit, Part II”). For the time being, I will let you ponder the parlous state of economics teaching in our schools and universities.

Until then, make sure to check out Fallen Love in case you haven’t already.