Genetics and the increase in obesity

In a discussion on the rise of mental health issues at Core Economics, Paul Frijters touches on the increase in obesity over the last 50 years.

One can basically out of hand reject the excuses most individuals give for their problems as being the reason. The rate of increase rules out any reasonable role for genetics. The fact that the poor suffer more from obesity, whilst it is cheaper to eat less and whilst food has always been cheap for the rich, rules out any obvious effect of the lower price of food or the availability of fast-food. The sustained increase over a long time rules out any story depending on some major current crisis. Like it or loath it, but it is clear that one must look at ‘cultural factors’ to have a hope of understanding what is going on.

A big hint comes from cross-national differences amongst rich countries, where things like wealth and food affordability don’t differ much. As you can see here, the Anglo-Saxon countries, and then particularly the US, stands out. Whilst a third of adults in the US are now obese (with about 25% of Australian adults), only 4% of Koreans and Japanese are such, and in the more egalitarian Northern European countries (Sweden, Norway, Holland) rates are below 10%. The same holds for Italy and France, though rates in those countries too are quite a bit up from what they were 50 years ago. So your one major clue is that there are major unexplained differences over countries.

I’ve heard this “it can’t be genetics” argument from a few people recently. And in some respects it is right. Clearly, the genes in the population have not changed substantially over the last 50 years. However, to dismiss genetics in trying to understand obesity is ignoring an important piece of the puzzle.

First is the high heritability of obesity – both before and during the increase in obesity of the last 50 years – usually measured in the range of 55 to 85 per cent. This level of heritability exceeds that measured for most behavioural traits (although it is in the realm of heritability for intelligence). This suggests that both before and after the increase in obesity, genetics plays a substantial role in who is obese. As the environment has changed (such as the “cultural factors” that Frijters alludes to), people of different genotypes have responded in different ways.

Another pointer to the role of genetics is in the high levels of obesity and obesity-related diseases in some populations, particularly indigenous populations with a limited history of agriculture. Pacific islanders, with almost no history of grain based agriculture, have the highest rates of obesity in the world. American Indian and Alaskan Natives (as a group) have the second highest level of obesity in the United States of any major ethnic group (behind people of Pacific Island origin). In Australia, aboriginals and Torres Strait Islanders are twice as likely to be obese as Caucasians. The costs of obesity, such as Type 2 diabetes, also tend to higher for these groups.

Frijters’s point on the response of the rich and the poor to food prices also has a hint of genetic factors. These groups are not the same. The rich and poor differ in average levels of IQ, willpower, conscientiousness and a host of other traits with a genetic component (even if you don’t believe there is a genetic component, you still should not treat them as the same). So when we ask why obesity is higher among poor people (or less educated people), we should ask what role those traits and their underlying genetic influences play. When faced with the same choices, they are likely to select different options.

These traits might also be relevant for cross-country comparison. Should we be surprised that East Asian countries where populations have higher measured IQ, lower rates of time preference and higher savings rates also have lower rates of obesity? (Obviously, on a cross-national basis, this is not the complete explanation. For example, East Asians in the United States have higher levels of obesity than their counterparts still in East Asia, although they are obese at rates lower than Caucasians and other ethnic groups).

We should also not be too quick to dismiss price. It is not only absolute price that matters, but also relative price of different food types. As argued by Rob Brooks, Steve Simpson and David Raubenheimer, simple carbohydrates have never been cheaper relative to protein. If you are price sensitive, you may shift consumption towards simple carbohydrates. As someone who tends to avoid simple carbohydrates, I can also attest that a large part of the relative price of food is the search effort in finding a low carbohydrate option.

The reason this matters also has an evolutionary basis. Eating food is not a simple “eat calories and feel full” process. Different foods create different responses in appetite. Brooks and his colleagues base their argument on the protein leverage hypothesis, which is a hypothesis that humans have a stronger propensity to regulate protein intake than they do for other non-protein calories. Humans eat until we satisfy out basic daily protein need. If the food we are eating has low protein content, we need to eat more before hitting that satiation point. These extra calories are what make someone obese. Trends in carbohydrate, protein and fat consumption in the United States over the last 40 years offer support for this argument.

Arguments such as the protein leverage hypothesis also have interesting implications for any arguments about the willpower of the obese. Someone eating a diet high in simple calories would need more willpower to constrain their calorie intake than someone on a high protein diet.

While Frijters points to the cross-national differences as a major clue to why obesity has increased, the above suggests that within country and cross-population differences will also be useful. The cultural changes that have resulted in the increase in obesity play out in different ways depending on who the person is. Genetics is clearly not the only factor that should be examined – look at the Anglosphere compared to Northern European countries – but any cultural explanation will need to accord with the evidence that the cultural changes do not affect all people equally.

12 thoughts on “Genetics and the increase in obesity

  1. Would you be happy with the slightly modified version? The change in obesity is not due to a change in genetics. Naturally it is also 100% genetic, if only because you can die from a genetic decease before you can get obese.

    The whole nature/nature debate makes no sense to me. Taken to the extreme, every trait is 100% due to genetic and 100% due to environment. Only after you specify the variability in the genes and in the environment can you get x% genetics and 100-x% environment.

    If you need willpower to stay lean, there is already something wrong. No animal in its natural environment needs willpower to stay healthy. Only humans and pets.

  2. Hi Jason,

    When I say ‘its not genetics’ I mainly mean that the genetic variation over the population hasnt changed and hence that the change in obesity rates has to be driven by a change in something else. By the same token, current obesity is not genetic in that people who tell me they are obese for genetic reasons usually have a vast majority of their ancestors who were not obese: even for those with the ‘wrong genes’ other factors are at play. You seem to agree with both of those, so we are probably squabbling over semantics here.

    Where we of course agree is that there is genetic variation over a population and that this will be useful in understanding the cross-sectional pattern. The Pacific Islanders are a good case. But how does genetic variation help in explaining the change? And indeed, how certain are you that its the genes of the native Americans and Aborigines and Innuit that matter for their relatively higher obesity rates in current times (following low rates previously, I presume)? These are, of course, also the poorer less advantaged groups who for very different reasons could face more temptations and less ‘cultural protection’ from those temptations? What genetic markers do you actually have in mind when talking about obesity in those groups?

    Let’s then look at your protein story: the protein hypothesis would have as a clear hypothesis that people who on average eat more meat are less obese, because they are more quickly full and hence stop eating other things. Its policy prescription would be to get kids to start each meal with solid meat, or better still a protein stick, so as to make them feel full. How plausible is either prediction? I seem to recall protein intakes have gone up over time in Western countries, so whilst it may well be true that in the cross-section the even bigger meat eaters are less obese (rich people eating steaks), the protein hypothesis sees to violate the time-series evidence, no?

    1. Hi Paul,

      I don’t think we’re too far apart on the first couple of paragraphs. The point of the post was that genetics are still relevant to the analysis even if there isn’t population genetic change. A good test of any cultural explanation for the increase will be that it is consistent with the cross-sectional pattern due to genetics.

      As for why I’m certain on the indigenous genes, I’m not, but I’m reasonably confident. There have been a few published genome-wide association studies that have looked for genes linking obesity, alcoholism and diabetes in indigenous populations. Although they usually turn something up, I don’t trust a lot of them them as they tend to be underpowered. The argument I consider strongest is the growing evidence for large evolutionary change in populations that have been exposed to agriculture – including adaptations to starch and gluten. We wouldn’t expect to see those adaptations in populations without that same exposure.

      I’m not the only one leaning this way, with increasing research effort in indigenous health focusing on genetic factors (e.g. see here and here). Some serious benefits could accrue through gaining a better understanding of the genetic underpinning to diseases in indigenous groups. Regardless, with the drop in prices of genomic sequencing, we’re going to start to see some strong evidence accrue on these points over the next few years and it will quickly render my theorising out of date.

      The evidence on the protein leverage hypothesis is looking OK. The article I linked to above (here it is again) looked at the period 1971 to 2006 and estimated that carbohydrate consumption was up by 21% in men and 23% in women, fat intake down 2% in men and up 5% in women, and protein intake up 2% in men and up 5% in women. This is broadly consistent with most other studies, and this time series is consistent with the idea that people eat until they get a certain level of protein. Most of the increase in calories has come from carbohydrates (and simple carbohydrates at that). The cross-section evidence from this study that increased energy from protein was associated with a decrease in total energy intake is also consistent with the hypothesis. I should say, however, that I don’t think the protein leverage hypothesis is the full picture here, and I expect other factors such as insulin sensitivity are probably also important.

      This story also fits with the pattern of meat consumption – massive increases in meat consumption between 1930 and 1970 (accompanied by a reduction in complex carbohydrate consumption from whole grains) and a moderate increase in meat consumption since 1970 (a balance of increasing chicken and decreasing beef consumption, and accompanied by a large increase in lower protein simple carbohydrate consumption). The reduction in per person meat consumption of the last few years in the United States has them almost back down to the meat consumption levels of 1970.

      1. Hi Jason,

        What at the end of the day are you in the above actually explaining? There is no story here about changes and no genetic markers that differ by population.

        Take the protein story, You keep insisting it is relevant and explains things. What does it explain though? If you look at protein uptake across countries (see eg. here: then you see that the uptake in the US is double that in Latin America and much of South-East Asia, but that its also high in Europe. As you say, it increased a lot since WWII and then plateaud in the US. How can one then possibly run the story that people eat until they hit a protein ceiling? Are people in Latin America and South-East Asia then continuously hungry? What about Americans in 1950, were they continuously hungry? If not, why have many of them hit upon the habit of eating until they feel stuffed? Similarly, do you really believe that whole populations would start to lose weight if you gave them protein drinks and protein sticks early in the morning? When thinking of the increase over time and the cross-sectional variation across countries, the protein story is just another distraction, at least in terms of explanations of that change. It might down the line help to develop diet pills but that is another topic and, of course, the number of ‘wonder diets’ that have come and failed to make a dent in the obesity epidemic is rather large, including the ‘meat binge’ diet.

        I guess the fundamental difference in our approach is that you continuously look for explanations that sound genetic and physical (‘insulin sensitivity’). I try to get my head round the relevant social interactions defining culture and economics, and more naturally look for explanations there. In some sense this is a matter of levels and solution strategies: even those looking for mental and cultural traits to explain things can always ask the follow up question where those traits have evolved from, and those always looking for the physical mechanisms can wonder about the interaction with mental processes. But that doesn’t mean that the odds of coming up with something useful are the same.

        Its like trying to explain why so many ships carry oil from the middle east to the industrial centers. One can try and focus on the nuts and bolts of the ships interacting with the minds of the sailors on those ships, or one can describe patters of global comparative advantage and development. I know the approach I would favour!

      2. Hi Paul,

        What I’m saying is that part of the explanation for cross-population differences in obesity is likely to be genetic. Even if every population experienced exactly the same “cultural change”, they would not exhibit the same change in levels of obesity. For example, why do Korean adoptees into the United States have levels of obesity that reflect their biological, not their adopted family? Questions such as these could be useful in honing your cultural explanation. Of course, as you note, cultural change is also partly endogenous to the genetics.

        You suggestion that there are no genetic markers by population goes too far. My previous point was about indigenous obesity, but there are plenty of markers that differ by historical diet across major populations (for example, the amylase gene). In the same way that only a small fraction of the heritable variation in height has been explained by specific genes, the genes associated with diet and obesity discovered by researchers to date explain a small part of the variation we see there. However, that says more about the state of the science than about what we are likely to find over the next few years for both height and diet.

        Protein consumption increased less than meat consumption (although you are right that it once went up). But people in developed countries became much taller (in the increase in height due to cultural factors?). I expect we’ll see similar increases in height in many of these lower protein countries as their protein and calorie intake increases. It’s an interesting question about whether they are more hungry, but I’m not convinced they are currently eating the optimal diet or sufficient protein (how much of the Flynn effect is diet?). However, as I also noted in my last comment, I’m not overly attached to the protein leverage hypothesis. Other hypotheses about insulin sensitivity etc. are probably important. Regardless of which one the evidence ultimately leans toward, this protein/carb balance probably affects BMI, disease propensity, appetite and the level of willpower required to eat less.

        Ultimately, I am not advocating an approach to this question. I am simply suggesting that ignoring genetics is throwing out useful information that your cultural explanation will have to be consistent with.

  3. I think the reason the Protein Leverage Hypothesis is appealing is because it’s simple idea and is easily quantifiable. But there is so much more to each of the macros (not just protein) than simply the volume the body requires (eg. insulin). The Protein Leverage Hypothesis assumes all protein sources are equal, and all you need to do is eat the “right amount”. Unfortunately, you can still be hungry and consume enough protein.

    Bodybuilders consume a large amount of protein. When they are preparing for a competition, they reduce carbohydrates significantly, and also reduce fats. Their protein intake remains very high. Despite getting more than enough protein, they are constantly hungry. Their energy levels drop as they get closer to a competition, and it is only through sheer will power that they can continue.

    If you think this only relates to a specific population like bodybuilders, I suggest your try the following:
    Eat NOTHING but your “required” protein daily intake each day (eg. from 300-400g of chicken breast) and do this for a week. No carbs, no additional fats (aside from the small amount in the chicken). I’d wager you’ll be quite hungry and low on energy by the end of the week, despite consuming an adequate amount of protein.

    1. Rabbit starvation also points to the need for more than just protein. But you are creating a caricature of the hypothesis. The question that needs to be addressed is what regulates appetite in people eating the types of diets eaten, first, in the environment of evolutionary adaptedness (which may be highly variable and population specific) and second, in today’s modern environments.

  4. The whole nature/nature debate makes no sense to me. Taken to the extreme, every trait is 100% due to genetic and 100% due to environment. Only after you specify the variability in the genes and in the environment can you get x% genetics and 100-x% environment.

    1. Heritability calculations are based on measurements of genetic variability, either directly through DNA testing or indirectly through examining differences between fraternal and identical twins. With the assumption that everything not genetic is environmental, that is all you need.

Comments welcome

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

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

Google photo

You are commenting using your Google 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