Skip to content

Economics as Junk Science: GDP and Historical Development

December 11, 2010

The title reflects the fact that sometimes Economists make tenuous claims that are scientifically weak.

This is an interesting video until you realize that it involves junk science.

 

Hans Rosling’s 200 Countries, 200 Years, 4 Minutes, – The Joy of Stats – BBC Four

Namely, comparing income across 200 years is meaningless. It makes much more sense to look at consumption choices. I’m willing to take a guess that the consumer price indexes used to weight the historical income don’t take into account whether the individual in 1810 can purchase penicillin, access to clean drinking water, or chemotherapy…. No wonder life expectancy is increasing.

Unfortunately, many economists fail to realize that GDP/money by itself is not a good indicator for well-being; indeed prices exist merely as relative quantities to one another within an economy of goods and services. But it matters what you can buy. As such prices are arguably useful in comparing year-to-year “wealth” where the type and prices of goods does not change drastically, but are useless for comparing consumption/utility across longer time periods where the nature of purchasable goods has changed.

So yes, it makes sense that things grow over time. It makes sense that technological progress increases economic productivity and life-expectancy. It does not mean that economic productivity and life-expectancy necessarily have a strong causal relationship, as the video seems to argue by juxtaposing them on the same chart. An econometrician would tell you that you need an instrumental variable. My guess is that time/technological advancement are the real causal link between the two.

Further to address the subtext of the video: there are problems with averaging country’s GDP to measure welfare, problems with the idea of convergence as there are increases in income inequality across countries, and problems with the arguments that on average for global citizens income inequality is decreasing.

Here is a good link: summarizing some of the conversation involving the difficulty of using GDP to measure welfare.

Update: Also it appears that Hans Rosling is displaying logged (exponentially growing) income on the bottom axis rather then linear income. This modification would be necessary only if implicit in his presentation is the argument that life expectancy is a linear function of logged income.

Advertisements
8 Comments leave one →
  1. retronomics permalink
    December 12, 2010 11:06 am

    It is true that these GDP figures can’t really depict the true progress of the last 200 years. If anything, technology and society have progressed much more than any GDP number can tell us.

    Still, there are some of your points I disagree with, or I don’t quite understand. The video claims that economic growth and health go hand in hand. Yet, you critizise economists for mixing up growth and well-being/happiness, which is a totally different thing. Well-being is not really measurable, but health is.

    Anyway, technological progress, which is indeed the driving factor behind both increases in health and in GDP, is also hardly measurable. GDP is probably the best available proxy for technology.

    Deriding this video as junk science is unjustified.

    By the way, prices are not just relative quantities. You forgot about demand. Prices reflect the marginal utility that people perceive of those goods and services.

    • December 12, 2010 4:07 pm

      I’m criticizing his use of weighted GDP as a historical measure period (for well-being/wealth/whatever). The health part is fine, but placing it in the context of GDP growth over a 200 year time-period is meaningless and possibly misleading.

      So I agree with what you are saying here:
      “It is true that these GDP figures can’t really depict the true progress of the last 200 years. If anything, technology and society have progressed much more than any GDP number can tell us.”

      I disagree with this:
      “GDP is probably the best available proxy for technology.”

      GDP is a measure of asset prices x economic activity and is not necessarily a good proxy for productivity/technology. As we have seen in the last decade in the U.S. you can have growing asset prices boost GDP without corresponding increases in productivity. The fact that the two both grow over time does not mean they are necessarily related.

      This gets back to my main point which is that the graph is misleading. The reason I call this “junk science” is because I am worried that people may link this spurious correlation with causality. How the information is presented here (logging income to show a linear relation between the two and then showing the same time trend for each variable), would lead both the average person or an economist to believe that implicit in the presentation is the argument that GDP growth and health exhibit a statistically and economically significant link.

      Some in the development field blindly equate wealth growth with welfare growth. That and the problems with using historical GDP as measurement are why I addressed welfare in the post. If you see Prof. Rosling’s lecture on TED it becomes clear that while he acknowledges the difference, he is still making a basic argument which favors blind economic growth without discussing the importance of the type/nature of economic growth relative to the country/region and within the population.

    • December 12, 2010 4:30 pm

      “Prices reflect the marginal utility that people perceive of those goods and services.”
      This is incorrect. The reason they are called marginal has to do with choices between goods not absolute measures of utility:

      “Marginalism explains choice with the hypothesis that people decide whether to effect any given change based on the marginal utility of that change, with rival alternatives being chosen based upon which has the greatest marginal utility.”
      http://en.wikipedia.org/wiki/Marginal_utility

      Prices are ordinal (they order preferences) and reflect relative consumption choices between existing goods; one dollar does not equal one “util”, hence the problems with GDP.

  2. retronomics permalink
    December 12, 2010 6:40 pm

    With all due respect, your claims are misleading your readers. GDP and productivity share a common long-term trend, which, as you certainly know, economitricians call “cointegration”. This is not only a statistical relationship, but it also makes sense in theory. The combined value of goods and services that can be produced in a given amount of time certainly depends on the productivity of labor and capital.

    And even if you insist on your opinion that GDP is not a good proxy, then I would like to hear from you what is a better proxy?

    I am not an expert on health economics but I presume that many important medical advances of the last two centuries would not have happened without productivity growth, and thus, without GDP growth. However, it is true that this causative relationship did not last long because it were rather simple improvements, at least viewed from todays perspective, that caused the largest gains in life expectancy, especially during the 19th century and early 20th century. Nowadays we can observe that even poor countries like Cuba can achieve relatively high life expectancy, though they certainly profited from copying the advancements of other more progressed countries.

    By the way, GDP is more than just asset prices. Or do you consider your haircut at the barber an “asset”?

    • December 12, 2010 9:03 pm

      You seem to have changed the subject. My post was not about productivity but rather about using historical GDP as an explanatory variable for life expectancy, and then inferring causality (economic significance)… in fact I think your second to last paragraph basically says you agree with me?
      But onto your other points:

      Cointegration does not imply causality, just that two variables move similarly over time… perhaps because of a third variable…

      As for the haircut comment I’m aware services constitute part of GDP, another reason why it is a poor measure for productivity. Further, the asset price comment was intended to reflect the fact that GDP measures output in prices (not units) . This means price fluctuations can affect output without changing quantities produced. Again, it is not an accurate measure of productivity. But we can further say that, going back to my point, that GDP is probably an even less accurate measure of medical research output, access to good health practices (U.S.), or any other things that effect human longevity.

      besides, some food for thought… maybe longer lives led to more GDP growth…

      If you want my opinion measuring productivity growth I would say net hours used to produce one unit of a given good… but this would not work on a macro-level, but I would be hesitant to use a measure which incorporates prices. If I had to use a price based measure it would be value added labor-productivity, because rises in input prices would be accounted for by rises in sale price which would minimize the effect of market fluctuations.

      If you are truly curious about measuring productivity here is a link:
      http://www.oecd.org/dataoecd/59/29/2352458.pdf
      it’s a bit long, but they have some general overview around pg 13.

      And regarding GDP and productivity, I’m not alone, these guys clearly argue that indeed there are other important variables at play:
      http://economia.uniparthenope.it/ise/sito/conferenze/miurVinci/Papers_Accepted/Marattin_Salotti_napoli.pdf

  3. L. D. (5random1 on YouTube) permalink
    December 14, 2010 11:42 am

    Life expectancy (longevity) is a very poor indicator of ‘health’ for any one individual, or for a nation as a whole. It may be a goal in health, but a very unreliable indicator.

    The temporal cost and continuing cost in lives, dollars, environmental changes of the two world wars and all wars since is not reflected in the dynamic graphic. Just one nuclear weapon today, intentional or accidental, can render this graphic meaningless.

    The ballooning environmental and ecological deficits from all human activities are not factored into the presentation.

    The health of the planet and the future potential for health are not even remotely considered. Hans Rosling pays lip service to “green technologies” at the conclusion, as if the Gulf oil spill or reliance on oil will easily vanish as some time in the near future.

    Nice presentation method for getting people, especially students from many disciplines, interested in analyzing problems, but technology + art + statistical methods do not always yield critical thinking skills or sound analysis.

    The short presentation (part of a larger program?) is disappointing when separated, presented on its own.

    • Bill permalink
      February 16, 2011 11:00 am

      He is simply making a point about a long-term trend in the development of nations. He is showing there is a geographic connection as well as an influence of type of government. He does not give an equation. If he did, maybe your criticisms would be more justified. It is not junk-science. He is graphing the GDP/Population or the per-capita income, is he not?
      Obviously technological development is involved but per-capita income is a proxy for this. He did not say it was perfect. Also, per-capita income tends to reflect the type of economy and social system the country has. I like it because it forces people to think about things over several generations, not just a few years, and then panic and want to impose some kind of a “solution”. It ties together a lot of things in a crude but simple fashion. Making the x-axis a log scale simply makes it more linear which is easier for non-scientists to interpret. If it was not log, it would be a curve that showed a relationship. You can not expect something this reductionist to be a perfect model for dozens of variables and the author did not try to do so.

      I just watched the video again and it has the additional variable of time and then he shows that parts of China that are rich and poor, etc. are similar to other countries that are rich and poor. So he is not just using life expectancy and saying the whole country is the same, which was one of the comments above. And if you watch it again, China and another of other countries have their life expectancies double before their wealth increases, so he is not saying there has to be a direct causation. So again, many of the comments above are misguided. His main point seems to be one of optimism about the future of humanity on a long-term scale. Maybe this is what bothers many of you.

      • February 16, 2011 4:04 pm

        Hey Bill,

        Thanks for voicing your opinion. You to jump around bit, addressing things that I feel I already adequately voiced my opinion on, and perhaps some of what you said is even a little self-contradictory.

        I did like your ending zinger however, so I’ll elaborate on it briefly.

        “His main point seems to be one of optimism about the future of humanity on a long-term scale. Maybe this is what bothers many of you.”

        I think that it is clear to any num-nuts that in the very long term historically things have been getting better for people (amazingly even before capitalism… people tend to forget how significant little discoveries and world-view changes originally were for getting to the point we are at now ). Certainly this is something positive. What I don’t like is when people take an overly simplistic view of this process (I’ve discussed this at length above) or when they take a linear view of history and progress. The benefits to humanity tend to be stochastic and things improve and worsen often in huge jumps. Look at Germany 70 years ago or the potential devastation of the Cold War for example. Thus positive trends can be erased very quickly and long-run correlations are spurious with such highly noisy irregular progress and such fat tails at the end of the distribution.

        On the positive side, things have generally trended positively, but that doesn’t mean we should blindly love the status quo and assume that it is the best possible outcome or has led to everything we see today and will not have potentially dire consequences. But yes obviously, if history is an indicator hopefully things will be getting even better for at least a portion of the World’s population.

Leave a Reply

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

WordPress.com Logo

You are commenting using your WordPress.com 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 )

Google+ photo

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

Connecting to %s

%d bloggers like this: