Why economics is the "dismal science"

We've all heard economics described as the "dismal science," yet it still qualifies for a Nobel prize. Many still grumble about the decision to tack on economics to the short list of true science Nobels, and while I don't know whether such complaints are justified, there is good reason to remain highly skeptical of the field's predictive powers in general. Take the just-released Tufts University study on the cost of climate change.

The main authors of the study, which was commissioned by the Natural Resources Defence Council, are economists from the Global Development and Environment Institute and Stockholm Environment Institute at Tufts. They get help from the Judge Business School at Cambridge University and Synapse Energy Economics, which is concerned primarily with energy economics. I'm sure they're all reputable folks.

They predict that a business as usual approach to fossil-fuel use will end up costing American civilization about $3.8 trillion dealing with the effects of a warming climate. This is not out of line with a whole slew of other studies, which tend to show it is far cheaper to mitigate global warming than trying to adapt to it. Of course, there are competing approaches from economists, notably those championed by non-economist Bjorn Lomborg, which use the sleigh-of-hand accounting technique know as discounting, to make the opposite case. Specifically, that it is cheaper to wait a few decades, until economies of scale, Moore's law and human ingenuity in general, has made dealing with climate change much cheaper than it would be today.

So who's right? The problem is, in economics, there's no way to tell. The problem is that all these economics forecasts are based on mathematical models that use the output of other mathematical simulations (climate models) as raw data. This means that the imprecision of one model is compounded by another.

The climate models are bad enough. They're the best we've got, but provide nowhere near the level of precision, either locally or temporally. that we need to generate useful predictions about just how the world will warm over the next century. This is why a bunch of climatologists recently called for a massive international investment in new supercomputer technology that would improve predictive capacity of the field.

One group says that what we really need is a "petaflop" computer that could model cloud cover (one of the biggest unknowns in climate circles) would cost $1 billion. "The system also would require 200 megawatts of electricity to operate, enough energy to power a small city of 100,000 residents."

At least the climate models are based on hard data (ice cores, paleoclimate records, etc.). Until we have something like the petaflop climate computer, any economics model is going to come with hopelessly wide margins of error. The Tufts study, is typical. It isn't quite up front about the limitations of the data they're using:

Global warming is already melting sea ice and glaciers that will contribute significantly to sea level rise. Sea level is expected to rise 23 inches in 2050 and 45 inches by 2100, with grave impacts expected for the Southeastern U.S.. By 2100, an estimated $360 billion per year will be spent on damaged or destroyed residential real estate in the United States as a result of the rising sea levels inundating low-lying coastal properties. The effects of climate change will also be felt in the form of more severe heat waves, hurricanes, droughts, and other erratic weather events--and in their impact on our economy's bottom line.

Global warming will change the nature of where Americans live. For example, this analysis found that if global warming continues unchecked, by 2100, New York City will feel like Las Vegas does today and San Francisco will have a climate comparable to that in New Orleans. In 2100, Boston will have average temperatures similar to those in Memphis, Tennessee today.

But we really have no good idea just what sea level will do over the next 80 years. We have no idea how much coastal property will require shoring up. The future trends in hurricanes in the North Atlantic are foggier then ever. Just this past month we've been told there may be fewer, although they may be stronger. No one really knows.

And to say without qualification that New York will feel like Las Vegas by 2100 is to assume far more certainty in climate models than the modelers have themselves.

Again, I think it's even sillier to argue that we know it will be cheaper in the future, but there's also no way to gauge the reliability of the Tufts results, either. All they do is take problematic climate forecasts and generate even more problematic number about the economy.

Given what the physical sciences tell us about the effects of climate change ;;;;; acidic oceans, more droughts and food shortages, declining biodiversity ;;;;; we should have ample reason to start slashing carbon emissions now, regardless of what economists tell us.

Tags

More like this

Sorry, but this post seems to have "soft science" bias written all over it. You've not really made any attempt to actually pick apart the methodology of the report, and I can't help but feel that this is just a lazy way of saying "pah, economics isn't a real science, it's just hypothetical models, it's meaningless". It says more about your own biases and misconceptions than about the wider debate.

No attempt to pick apart the methodology, except for saying that the inputs to their model are results of other models.

Garbage in = garbage out.

I can't imagine trying to justify some action because the cost could be $3 trillion or $10 trillion, maybe just a few hundred billion. Economics here can't add too much to the debate other than muddying the waters by ascribing questionable dollar values to actions. Whoo.

It is hard to imagine that New York in 2100 will feel the way Las Vegas does now. For one, Las Vegas is currently in a desert. Whereas New York is currently on the Atlantic coast, and by 2100 could be under water. A pretty silly comparison. How about Indianapolis?

Economics provides a way to aggregate and use the information provided by climatology and other disciplines in a...political context (not quite the word I'm looking for, but it'll do). Allaying climate change does have real significant costs, and there are bad ways to go about doing so, and better ways of doing so. While I'm sure some things do get questionable, it's hardly surprising; predicting the future in complex systems is never a simple or easy task.

Just proves that forecasting is difficult, especially the future.

By notthedroids (not verified) on 27 May 2008 #permalink

Pet peeve of mine: there is not and never has been a Nobel Prize in economics. There is the Swedish National Bank Prize in Economic Sciences in Memory of Alfred Nobel, a mouthful that has been well marketed as the "Nobel Prize in economics".

I can't believe the sheer ignorance on display here. Frankly, I'm disgusted that somebody paid to blog on science is so poorly educated on it.

@ChrisS: Blindly stating "garbage in = garbage out" isn't a valid criticism. Perhaps you'd like to give us a detailed breakdown on why what was fed in was garbage, and how exactly it led to garbage coming out the other end? Thought not.

You then say, ridiculously: "Economics here can't add too much to the debate other than muddying the waters by ascribing questionable dollar values to actions."

Yes, because I mean, hell, I can't think of any reason why attempting to put a dollar value on something would be a useful thing to do. I mean, why on earth would policy makers ever want to know how much something costs?

This entire criticism is based on the prejudice that economics is somehow junk science simply because predictions are very difficult. Never mind the fact that physicists used to make exactly the same claims about the life sciences. Economists are clearly idiots, and irrelevant, because they don't have a flawless knowledge of climatology.

Again we have this pathetic need of biologists and physicists to sneer at disciplines they don't feel are worthy of being called "true science". Frankly, I expect something a bit more intelligent from the Sciblings.

@Pareto & notthedroids: Nice comments, cheers. Predicting the future is difficult in any discipline, especially in complex systems.

One of the few specific things I remember about taking economics back in the 80's was one of the assumptions made in economics; all economic decision making is rational. This is of course not the case and that was made clear in the class. Better economics studies probably try to deal with this but it must be difficult to quantify irrational decision making.

Martin, Chris's point as I take it was a model is not a great source of date for another model because models are never correct (it would not be a model then). Using only (or mostly) data from a model to supply data for another model would just multiply any errors in the first data set.

This entire criticism is based on the prejudice that economics is somehow junk science simply because predictions are very difficult. Never mind the fact that physicists used to make exactly the same claims about the life sciences. Economists are clearly idiots, and irrelevant, because they don't have a flawless knowledge of climatology.

It's not a prejudice - it's a fact. Some biology is junk - little physics is junk - most of economics is junk. It's not because "predictions are so hard", but because predictions are impossible without adequate preliminary research, which has never been done in economics.

Doing economics today is like doing climatology in the nineteenth century. The best you are going to get is Arrhenius predicting CO2 linkage to temperature and a general concept of the risk of greenhouse effects. Going forward from that to posit highly specific numbers would have made him look like a fool (in much the same way that the vast majority of economists look foolish).

I can see why it is politically expedient to invent numbers; but I'll believe them when the field starts making forecasts that are better than random. Why not start with baseball card trading, or the inflation rate in Second Life? That last is half-serious.

Every elective class I've taken so far--sociology, psychology, and micro-economics--stressed how they are each hard sciences.

But on the supercomputer front.. The Top500.org site is predicting first petaflop computer on the next list due in June.

As to power, looks like numbers might be certainly less than 200MW. The #1 machine currently with half-petaflop capacity is contained in a building with total power capacity of 25MW.
https://www.llnl.gov/str/JanFeb05/Atkinson.html

It would be nice if they would use the most powerful supercomputers to simulate climate than to ensure that our nukes can still kill millions.

@Kevin: I don't disagree that models based on models are problematic. What I take massive issue with is somebody who claims to be a scientist dismissing an entire study without any reasonable analysis basis for no other reason. Also, this clearly isn't the point Chris wants to make - the point clearly is that he feels that economics is dismal, junk science and he wants to rant about it.

@Frog: Sorry, but you seem to know very little about this field. Predictions in economics are certainly not impossible, and there are many hard "econophysics" style models that successfully make predictions. Picking on one - or a few - bad studies and saying that somehow most of an entire field is "dismal" or "junk" is simply nonsense. The argument raised by you and Chris here is staggeringly similar to the creationist stance on biology - animals aren't rational, biology is very complex, there are many unanswered questions about what these complex systems do, therefore it's all junk.

It infuriates me to see this kind of junk argument on a supposedly quality blog.

The climate models are bad enough. They're the best we've got, but provide nowhere near the level of precision, either locally or temporally. that we need to generate useful predictions about just how the world will warm over the next century.

Define "useful". If they can say something like "the warming will most likely be around 3 degrees, and probably not much less than 2 or more than 5", that's much more useful than not being able to predict anything at all.

This is why a bunch of climatologists recently called for a massive international investment in new supercomputer technology that would improve predictive capacity of the field.

It's far from clear whether investing in a super-high resolution climate model is the way to improve decision making. The GCMs only give you a best guess, because you can't afford to run them very many times. It may be more fruitful to instead invest in less complicated models that you can run more times, in order to explore the parametric and structural uncertainties. I'd rather have a somewhat inaccurate estimate with reasonable error bars than a more accurate single point estimate with no quantification of the associated uncertainty.

At least the climate models are based on hard data (ice cores, paleoclimate records, etc.).

Economic models are based on data too.

Until we have something like the petaflop climate computer, any economics model is going to come with hopelessly wide margins of error.

Even a petaflop computer is not going to hugely reduce the uncertainty in key variables like climate sensitivity. (In fact, it probably won't give any indication of the uncertainty at all.) And the potential improvement in climate prediction is probably outweighed by uncertainties in the economics, such as in the discount rate to use. Yes, the margins of error are wide, but they're still much better than if we didn't know anything about the climate or economics.

But we really have no good idea just what sea level will do over the next 80 years.

Well, it's probably going to be more than 10 centimeters and less than a few meters. That alone tells us something.

The future trends in hurricanes in the North Atlantic are foggier then ever.

True, but since people need to prepare for the eventuality of hurricanes anyway, it's not clear how much that affects the decisions people make today.

Again, I think it's even sillier to argue that we know it will be cheaper in the future,

Silly? It's pretty plausible based on technological development and economic growth, until/unless things get so bad that the world economy crashes.

Given what the physical sciences tell us about the effects of climate change - acidic oceans, more droughts and food shortages, declining biodiversity - we should have ample reason to start slashing carbon emissions now, regardless of what economists tell us.

Duh, but that's not the point. We know that we have to reduce carbon emissions, but the question is how much and how fast is it economically feasible to attempt?

By Ambitwistor (not verified) on 28 May 2008 #permalink

Let me clarify: while it is reasonable to believe that mitigation will be cheaper in the future, it is less certain how much cheaper it will be. That's why you have the "delayer" arguments. The sensible strategy — as predicted by those horrible economic models — is to plan for some mitigation now, and either increase it or decrease it as we learn more.

By Ambitwistor (not verified) on 28 May 2008 #permalink

This post is linked in the scienceblogs banner so you'll get lots of comments.

As I understand it, the largest uncertainty in the IPCC forecasts is not from the climate modelling, but about what people will do - in other words, economics, which is essentially the science of the collective activity of large numbers of people.

If we want to generate outcomes at the low end of the forecasts, we will need to understand how proposed courses of action affect what people will do. So I don't see how we can avoid economic modelling.

If the state of economics as a science is as bad as you say (personally, I don't think it is), that is very unsatisfactory, but we have to try.

Martin,

Economics is a dismal science because their are to many variables and economics has an assumption that decisions are rational (a very common assumption in social science, try to quantify irrational decisions, they would not be irrational if one could quantify why they were made). Political Science, Anthropology, Sociology and most if not all social sciences are dismal. This does not mean that they are useless but that anyone doing research in these fields should be aware of the limitations. Also, Chris is correct, a model is a very poor source of date to feed into another model. First rule on models is that they are always incorrect, just like a model of a car a model of a system will not be that system. Thus although the data in one model may be useful if the model is good any errors (which will exit) will be multiplied if you use that data as the major data source for another model.

It would be great if we could create accurate models, then we could program a computer to tell us what to do. Think I read a science fiction book with that premise but it was fiction.

Also, policy made with bad predictions usually makes bad policy. Just because one can put a value on something does not mean that value is correct. Policy should be made more on cost, money can be found. Policy should be made on what is best for whatever community the policy is going to effect.

Policy should not be made on cost, money can be found.

Sorry for the error.

Shouldn't that be "petaflops"? "Flops" is an acronym, not a plural: FLoating Point Operations Per Second.

By Paul Murray (not verified) on 28 May 2008 #permalink

KevinC: I do not think you have studied economics or economists at all. Assumptions and model-building exist to help approximate the world, and nobody has ever said otherwise. Economists have become increasingly aware about the limitations of assumptions and models, though as you yourself ostensibly note, they are still useful. Why are you implying that this awareness does not exist? And do you know how accurate or inaccurate their models are?

Besides which, you've committed the most egregious of errors, considering only monetary values. The most basic idea in economics is that of opportunity cost, that the cost of something is what you give up to get it. Whence does money derive its value? Not by the mere fact that currency exists, or else we could create value just by printing more. Money is superficial; it is a medium of exchange. While it is easy to see that the government is gaining a citizen's $100 for a tax, it is not as easy to see what that citizen would have otherwise done with that $100, had they not been forced to pay the tax - that $100 which would have been spent on something productive, and continue along its path.

So let me ask you this: are you willing to save a life if it costs you five lives to do so?

I understand the complaint that all this modeling is full of uncertainties. But on the other hand, what alternative is there besides making the best-informed prediction we know how? You could say, well, just err on the side of caution, but what does that mean? There is a potential conflict between the desire to not disrupt people's lives today any more than necessary, and the desire to make sure that future catastrophes are averted. Which side of that dichotomy do you want to err on?

If you say that our models and ability to predict are imperfect and so we shouldn't base policy on them, then what should we base policy on?

This sounds exactly like the criticisms that global warming deniers used for years and years.

Moreover, discounting is not a 'sleight of hand' it's textbook economics and it doesn't take more than a few seconds of thought to realize that it's a factor that has to be taken into account. You might reasonably disagree with the application of discounting in this case but it's absurd to dismiss it out of hand.

Yes, but the problem with useing discounting at the usual rate (such as that by HM Treasury, etc), is that it will quite possibly be completely at odds with actual reality.

Even the most conservative predictions of climate changes effects include widespread drought in formerly highly productive agricultural areas (Australia is already having problems, the fall in the amount of snow-pack feed to the rivers of California will obviously have an effect on yields, and increasing flooding in parts of Asia will reduce the ability to produce staple crops);

increased costs in the production of energy;

increased costs of repair and renewal of infrastructure such as rail and roads, plus the costs of unstable weather patterns leading to problems such as flooding,

increased numbers of those suffering from heat related illness, and more widespread,

a possibility of movements of 'climate' refugees and the increased costs of security to prevent 'resource' wars.

None of these sound like business as usual. Its certainly true that commerce will continue, but all these factors reduce economic growth and increase uncertainty. Simply assuming the rate of discount will continue to be steady makes little sense. Policy should be to mitigate as many of these future costs now, because it will be far cheaper. Stern makes clear the old phrase 'prevention is better than cure'(and cheaper too).

Personally, I figured that economics was a dismal science when I took it, only to find that the three central factors of a free market did not actually seem to exist in reality.

I am a fresh-out-of-college economist who is working on a climate change research project - summarizing existing research and rewriting it in a more condensed and presentable format, at least that's the goal.

It did indeed strike me about the meaningfulness of many of the economic estimates, but I know that for the project I am working on the goal is to get a ballpark estimate so that policy makers are capable of making the best, most informed decisions they can given the information they have.

Yes, there is more than ample physical evidence to compel one to act but what many economists are trying to do (myself and those I work with included, to the best of my knowledge) is help prioritize spending. A certain amount of warming is inevitable at this point, and policy makers need help deciding which adaptation options are most efficient. They need to have expectations for what the damages will be in various sectors, how to plan budgets, and what problems to expect and of what magnitude.

While these estimates are most certainly uncertain at best and misleading at worst, the other option is to attempt nothing ... to tell policy makers "climate change is happening. reduce emissions". This is all well and good, but how are they supposed to know which ways of reducing emissions are most effective and efficient monetarily, or where it may be best not to adapt and what the adaptation options are.

Also, I'm not sure how typical that study is. I have yet to see a study that does not publish "ranges" and discuss variability by scenario, at the very least. One study I saw published the standard deviations on their precipitation estimate numbers and it made me wonder why they bothered to write anything at all. It's not really about being right, as far as I know, but about trying to help those making the decisions make the most informed decisions they can. I don't think that economists. estimates being uncertain invalidates the need for estimates or damages the credibility of those making the estimates.

I'm sure you're aware of the many criticisms leveled against Bjorn Lomborg's work - he has many economists working for him who are much more credible and erudite than I, so I'll restrain myself to this: the tools of the field are still relevant and useful, regardless of the uses some may put them toward.

the tools of the field are still relevant and useful, regardless of the misuses some may put them toward.

Fixed it for you.