Skeptical about Economic Models? Don't Trust Climate Models Either, Then.

I read an interesting article at The Market Mogul today entitled, "The Problem with Economic Models," by Aarondeep Hothi.  I reproduce the essay below.

The Problem with Economic Models
Aarondeep Hothi

Economic models are everywhere. They're used both implicitly and explicitly by politicians, economists, journalists and even the general public. These models somehow manage to wrap entire concepts in beautifully presented mathematics and graphs before presenting clear, concise conclusions. They provide a framework for digesting things such as how wages are determined, the effects of a living wage, how countries can grow after exhausting the gains from capital accumulation, the effects of certain policy decisions and many, many more key economic concepts. What could possibly go wrong when applying them to the real world? 

The Assumptions

Economic models become problematic when we attempt to take these models and their conclusions as facts and fall into the trap of thinking that things in the real world work as these models suggest. 

An economic model is defined as an interpretation and not a representation of reality. It is nothing more than a simplified interpretation of reality through the use of assumptions, most of which don't hold in the real world. These assumptions are used to set up an economic reality - a scenario in which aims or questions can be addressed and answered. Once these assumptions have been established, some methodology will be applied to try and derive analytical conclusions. These conclusions are often interpreted as facts about how things work in the real world, leading to confusion about why the models don't hold in reality.

If the assumptions of these economic models don't hold in the real world, then why would their conclusions? To be clear, all economic models have assumptions. Every economic model has assumptions that may not hold in the real world; hence neither do the conclusions of these models. Some models and their assumptions have been stated below, and it is rather simple to identify those that may be unrealistic and may not actually hold in the real world

Mundell-Fleming Model (Imperfect capital mobility model, the most realistic)
1. An open economy with constraints in physical movements and information
2. The time period is short term (capital fixed, equilibrium reached within a year)
3. Three markets: Goods Market, Money Market and Foreign Exchange market
4. The country in question is small
5. Domestic price level must be fixed
6. Balance of payments consists of a current account and a capital account
7. Exchange rate expectations are static
8. Imports are competitive
Solow-Swan model
1. Closed and private economy
2. One sector producing a homogenous good
3. Saving is a proportional function of income
4. The labour force growth rate is constant (it is equal to the population growth rate)
5. There is substitution between factors
6. There are constant returns to scale
7. There is no independent investment function, investment equals saving always
The Phillips Curve
1. Short to medium term model
2. Prices and Wages are slow to adjust to changes in aggregate demand
3. There is an unemployment rate that doesn't accelerate inflation (NAIRU)
4. Workers are only concerned with their nominal wages
5. Workers do not hold inflation expectations

So what does this all mean?

Final Word

Economic models should be used as a loose tool or framework for examining issues. These models are to be used not believed. Combined with real world testing using historical data, these models can be used to explain, in part, some phenomena. They should be used as part of an explanation, but shouldn't be relied on solely. Reality checks with real world data are an absolute must. Unfortunately, it seems as though these models are often interpreted as how things work in the real world and that leads to a whole host of problems. It's not just the average member of the public that falls into this trap either.


Here is a related article that I just wrote:

The Problem with Climate Models
Kerk Phillips

Climate models are everywhere. They're used both implicitly and explicitly by politicians, weathermen, journalists and even the general public. These models somehow manage to wrap entire concepts in beautifully presented mathematics and graphs before presenting clear, concise conclusions. They provide a framework for digesting things such as how global temperatures are determined, the effects of changing weather patterns, how sea levels can change after the melting of polar ice, the effects of certain policy decisions and many, many more key ecological concepts. What could possibly go wrong when applying them to the real world? 

The Assumptions

Climate models become problematic when we attempt to take these models and their conclusions as facts and fall into the trap of thinking that things in the real world work as these models suggest. 

A climate model is defined as an interpretation and not a representation of reality. It is nothing more than a simplified interpretation of reality through the use of assumptions, most of which don't hold in the real world. These assumptions are used to set up an ecological reality - a scenario in which aims or questions can be addressed and answered. Once these assumptions have been established, some methodology will be applied to try and derive analytical conclusions. These conclusions are often interpreted as facts about how things work in the real world, leading to confusion about why the models don't hold in reality.

If the assumptions of these climate models don't hold in the real world, then why would their conclusions? To be clear, all climate models have assumptions. Every climate model has assumptions that may not hold in the real world; hence neither do the conclusions of these models. Some models and their assumptions have been stated below, and it is rather simple to identify those that may be unrealistic and may not actually hold in the real world.

Box Models (simplified versions of complex systems)
1. Within a given box, the concentration of any chemical species is uniform.
2. The boxes are assumed to be mixed homogeneously.
3. Box properties (e.g. their volume) do not change with time.
Zero-dimensional models
1. All changes in temperature are driven by solar radiation.
2. Cloud cover and albedo are taken as given or fixed.
3. Whole planet is the same with a single temperature.
Radiative-convective models
1. Upwelling and downwelling radiative transfer only
2. Atmospheric layers absorb and emit infrared radiation in fixed ratios
3. Upward transport of heat only.
Higher dimensional versions of the above
1. These may have more than one temperature, but still generate only zonal averages.

So what does this all mean?

Final Word

Climate models should be used as a loose tool or framework for examining issues. These models are to be used not believed. Combined with real world testing using historical data, these models can be used to explain, in part, some phenomena. They should be used as part of an explanation, but shouldn't be relied on solely. Reality checks with real world data are an absolute must. Unfortunately, it seems as though these models are often interpreted as how things work in the real world and that leads to a whole host of problems. It's not just the average member of the public that falls into this trap either.


This didn't take much work, as you can see.  Change a few words and spend 10 minutes on Wikipedia.  I don't mean to pick on Hothi because his view is quite commonplace and he's certainly not the first to have expressed it.

My point is not that climate models are not to be believed.  Rather that one should place equal credence on economic and climate modeling.  They are both built on mathematical foundations that abstract from reality.

You may argue that I have chosen the simplest climate models and, of course, these simple models have unrealistic assumptions.  Large-scale climate models are much more complex.

I would reply that Mundell-Flemming, Solow-Swan, and the Phillips curve are also terribly simplistic models and, of course, they have unrealistic assumptions.  Large-scale economic models are much more complex.

But aren't large-scale economic models built using the same basic structure and similar assumptions as the simple models? Of course, models are just that, small imitations of the real thing.  Large-scale climate models embed the assumptions from box models, zero-dimensional models and radiative correction.

If you are skeptical about economic modeling based on the arguments laid out by the original article, you should be equally skeptical of climate models.