Thursday, July 27, 2017

The Obesity Code and Economists as General Practitioners

"The past generation, like several generations before it, has indeed been one of greater and greater specialization…This advance has not been attained without cost. The price has been the loss of minds, or the neglect to develop minds, trained to cope with the complex problems of today in the comprehensive, overall manner called for by such problems.”
The above quote may sound like a recent criticism of economics, but it actually comes from a 1936 article, "The Need for `Generalists,'" by A. G. Black, Chief of the Bureau of Agricultural Economics, in the Journal of Farm Economics (p. 657). Just 12 years prior, John Maynard Keynes' penned his much-quoted description of economists for a 1924 obituary of Alfred Marshall:
“The study of economics does not seem to require any specialized gifts of an unusually high order. Is it not, intellectually regarded, a very easy subject compared with the higher branches of philosophy or pure science? An easy subject at which few excel! The paradox finds its explanation, perhaps, in that the master-economist must possess a rare combination of gifts. He must be mathematician, historian, statesman, philosopher—in some degree. He must understand symbols and speak in words. He must contemplate the particular in terms of the general and touch abstract and concrete in the same flight of thought. He must study the present in the light of the past for the purposes of the future. No part of man’s nature or his institutions must lie entirely outside his regard” (p. 321-322).
While Keynes celebrated economist-as-generalist, Black's complaint about the trend of overspecialization, coupled with excessive "mathiness" and insularity, continues. This often-fair criticism frequently comes from within the profession--because who loves thinking about economists more than economists? But at the same time, there is a trend in the opposite direction, a trend towards taking an increasing scope of nature and institutions into regard. Nowhere is this more obvious than among the blogging, tweeting, punditing economists with whom I associate.

The other day, for example, Miles Kimball wrote a bunch of tweets about the causes of obesity. When I liked one of his tweets (because it suggested cheese is not bad for you, and how could I not like that?) he asked if I would read "The Obesity Code" by Jason Fung, which he blogged about earlier, and respond to the evidence.

I was quick to agree. Only later did I pause to consider the irony that I feel more confident in my ability to evaluate scholarship from a medical field in which I have zero training than I often feel when asked to review scholarship or policies in my field of supposed expertise, monetary economics. Am I being a master-economist à la Keynes, or simply irresponsible?

Kenneth Arrow's 1963 "Uncertainty and the Welfare Economics of Healthcare" is credited with the birth of health economics. In this article, Arrow notes that he is concerned only with the market and non-market institutions of medical services, and not health itself. But since then, health economics has broadened in scope, and incorporates the study of actual health outcomes (like obesity), to both acclaim and criticism. See my earlier post about economics research on depression, or consider the extremely polarized ratings of economist Emily Oster's book on pregnancy (which I like very much).

Jason Fung himself is not an obesity researcher, but rather a physician who specializes in end-stage kidney disease requiring dialysis. The foreward to his book remarks, "His credentials do not obviously explain why he should author a book titled The Obesity Code," before going on to justify his decision. So, while duly aware of my limited credentials, I feel willing to at least comment on the book and point out many of its parallels with macroeconomic research.

The Trouble with Accounting Identities (and Counting Calories)

Fung's book takes issue with the dominant "Calories In/Calories Out" paradigm for weight loss. This idea-- that to lose weight, you need to consume fewer calories than you burn-- is based on the First Law of Thermodynamics: energy can neither be created nor destroyed in an isolated system. Fung obviously doesn't dispute the Law, but he disputes its application to weight loss, premised on "Assumption 1: Calories In and Calories Out are independent of each other."

Fung argues that in response to a reduction in Calories In, the body will reduce Calories Out, citing a number of studies in which underfed participants started burning far fewer calories per day, becoming cold and unable to concentrate as the body expended fewer resources on heating itself and on brain functioning.

In economics, an accounting identity is an equality that by definition or construction must be true. Every introductory macro course teaches national income accounting: GDP=C+I+G+NX. What happens, we may ask, if government spending (G) increases by $1000? Most students would probably guess that GDP would also increase by $1000, but this is relying on a ceteris parabis assumption. If consumption (C), investment (I), and net exports (NX) stay the same when G rises, then yes, GDP must rise by $1000. But if C, I, or NX is not independent of G, the response of GDP could very well be quite different. For example, in the extreme case that the government spending completely crowds out private investment, so I falls by $1000 (with no change in C or NX), then GDP will not change at all.

The First Law of Thermodynamics is also an accounting identity. It is true that if Calories In exceed Calories Out, we will gain weight, and vice versa. But it is not true that reducing Calories In leaves Calories Out unchanged. And according to Fung, Calories Out may respond so strongly to Calories In, almost one-for-one, that sustained weight loss will not occur.

Proximate and Ultimate Causes

A caloric deficit (Calories In < Calories Out) is a proximate cause of weight loss, and a caloric surplus a proximate cause of weight gain. But proximate causes are not useful for policy prescriptions. Think again about the GDP=C+I+G+NX accounting identity. This tells us that we can increase GDP by increasing consumption. Great! But how can we increase consumption? We need to know the deeper determinants of the components of GDP. Telling a patient to increase her caloric deficit to lose weight is as practical as advising a government to boost consumption to achieve higher GDP, and neither effort is likely to be very sustainable. So most of Fung's book is devoted to exploring what he claims are the ultimate causes of body weight and the policy implications that follow.

Set Points 

One of the most important concepts in Fung's book is the "set point" for body weight. This is the weight at which the body "wants" be be; efforts to sustain weight loss are unlikely to persist, as the body returns to its set weight by reducing basal energy expenditure.

An important question is what determines the set point. A second set of questions surrounds what happens away from the set point. In other words, what are the mechanisms by which homeostasis occurs? In macro models, too, we may focus on finding the equilibrium and on the off-equilibrium dynamics. The very idea that the body has a set point is controversial, or at least counterintuitive, as is the existence of certain set points in macro, especially the natural rate of unemployment.

The set point, according to Fung, is all about insulin. Reducing insulin levels and insulin resistance allows fat burning (lipolysis) so the body has plenty of energy coming in without the need to lower basal metabolism; this is the only way to reduce the set point. The whole premise of his hormonal obesity theory rests on this. It is the starting point for his explanation of why obesity occurs and what to do about it. Obesity occurs when the body's set point gradually increases over time, as insulin levels and insulin resistance rise in a "vicious cycle." So we need to understand the mechanisms behind the rise and the dynamics of this cycle.

Mechanisms and Models

Fung goes into great detail about the workings of insulin and related hormones and their interactions with glucose and fructose. This background all aims to support his proposals about the causes of obesity. The causes are multifactorial, but mostly have to do with the composition and timing of the modern diet (what and when we eat). Culprits include refined and processed foods, added sugar, and emphasis on low fat/high carb diets, high cortisol levels from stress and sleep deprivation, and frequent snacking. Fung cites dozens of empirical studies, some observational and others controlled trials, to support his hormonal obesity theory.

Here I am not entirely sure how closely to draw a parallel to economics. Macroeconomists also rely on models that lay out a series of mechanisms, and use (mostly) observational and (rarely) experimental data to test them, and like epidemiological researchers face challenges of endogeneity and omitted variable bias. But are biological mechanisms inherently different than economic ones because they are more observable, stable, and falsifiable? My intuition says yes, but I don't know enough about medical and biological research to be sure. Fung does not discuss the research behind scientists' knowledge of how hormones work, but only the research on health and weight outcomes associated with various nutritional strategies and drugs.

At the beginning of the book, Fung announces his refusal to even consider animal studies. This somewhat surprises me, as I thought that finding a result consistently across species could strengthen our conclusions, and mechanisms are likely to be similar, but he seems to view animal studies as totally uninformative for humans. If that is true, then why do we use animals in medical research at all?

Persistence Creates Resistance

So how does the body's set point rise enough that a person becomes obese? Fung claims that the Calories In/Calories Out model neglects the time-dependence of obesity, noting that it is much easier for a person who has been overweight for only a short while to lose weight. If someone has been overweight a long time, it is much harder, because they have developed insulin resistance. Insulin levels normally rise and fall over the course of the day, not normally causing any problem. But persistently high levels of insulin, a hormonal imbalance, result in insulin resistance, leading to yet higher levels of insulin, and yet greater insulin resistance (and weight gain). Fung uses the cycle of antibiotic resistance as an analogy for the development of insulin resistance:
Exposure causes resistance...As we use an antibiotic more and more, organisms resistance to it are naturally selected to survive and reproduce. Evenually, these resistance organisms dominate, and the antibiotic becomes useless (p. 110).
He also uses the example of drug resistance: a cocaine user needs ever greater doses. "Drugs cause drug resistance" (p. 111). Macroeconomics provides its own metaphors. In the early conception of the Phillips Curve, it was believed that the inverse relationship between unemployment and inflation could be exploited by policymakers. Just as a doctor who wants to cure a bacterial infection may prescribe antibiotics, a policymaker who wants lower unemployment must just tolerate a little higher inflation. But the trouble with following such a policy is that as that higher inflation persists, people's expectations adapt. They come to expect higher inflation in the future, and that expectation is self-fulfilling, so it takes yet higher inflation to keep unemployment at or below its "set point."

Institutional Interest and Influence

How a field evaluates evidence and puts it into practice--and even what research is funded and publicized--depends on the powerful institutions in the field and their vested interests, even if the interest is merely in saving face. According to Fung, the American Heart Association (AHA), snack food companies, and doctors repeatedly ignored evidence against the low-fat low-carb diet and the Calories In/Calories Out model to make money or save face. His criticisms of the AHA, in particular, are reminiscent of those against the IMF for the policies it has imposed through the conditions of its loans.

Of course, the critics themselves may have biases or vested interests. Fung himself quite likely neglected to mention a number of studies that did not fit his theory. In an effort to sell books and promote his website and reputation, he very likely is oversimplifying and projecting more-than-warranted confidence. So how do I evaluate the book overall, and will I follow its recommendations for myself and my family?

First, while the book's title emphasizes obesity, it doesn't seem to be written only for readers who are or are becoming obese. It is not clear whether the recommendations presented in this book are useful for people who are already maintaining a healthy weight, but he certainly never suggests otherwise. And for a book so focused on hormones, he makes shockingly little distinction between male and female dietary needs and responses. Since I am breastfeeding twins, and am a still-active former college athlete, my hormonal balance and dietary needs must be far from average, and I'm not looking to lose weight. He also doesn't make much distinction between the needs of adults and kids (like my toddler).

Still, despite the fact that he presents himself as destroying the conventional wisdom on weight loss, most of his advice is unlikely to be controversial: eat whole foods, reduce added sugar, don't fear healthy fats. Before reading this book, I already knew I should try to do that, though sometimes chose not to. I was especially focused on nutrition during my twin pregnancy, and most of the advice was basically equivalent. After reading this book, I'm slightly more motivated, as I have somewhat more evidence as to why it is beneficial, and I still don't see how it could hurt. Other advice is less supported, but at least not likely to be harmful: avoid artificial sweeteners (even Stevia), eat vinegar.

His support of fasting and snack avoidance, and his views on insulin provision to diabetics, seem the least supported and most likely to be harmful. He says that fasting and skipping snacks and breakfast provides recurrent periods of very low insulin levels, reducing insulin resistance, but I don't see any concrete evidence of the length of time you must wait in between eating to reap benefits. He cites ancient Greeks, like Hippocrates of Kos, and a few case studies, as "evidence" of the benefits of fasting. Maybe it is my own proclivity for "grazing," and my observations of my two-year-old when we skip a snack, that makes me skeptical. This may work for some, but I'm in no rush to try it.


Friday, July 7, 2017

New Publication: Measuring Uncertainty Based on Rounding

For the next few weeks, you can download my new paper in the Journal of Monetary Economics for free here. The title is "Measuring uncertainty based on rounding: New method and application to inflation expectations." It became available online the same day that my twins were born (!!) but was much longer in the making, as it was my job market paper at Berkeley.

Here is the abstract:
The literature on cognition and communication documents that people use round numbers to convey uncertainty. This paper introduces a method of quantifying the uncertainty associated with round responses in pre-existing survey data. I construct micro-level and time series measures of inflation uncertainty since 1978. Inflation uncertainty is countercyclical and correlated with inflation disagreement, volatility, and the Economic Policy Uncertainty index. Inflation uncertainty is lowest among high-income consumers, college graduates, males, and stock market investors. More uncertain consumers are more reluctant to spend on durables, cars, and homes. Round responses are common on many surveys, suggesting numerous applications of this method.