In today’s episode of Politically Dicey But Important Topics Of Research …

The newspaper article summarising the research: http://www.guardian.co.uk/science/2010/jun/30/disease-rife-countries-low-iqs

People who live in countries where disease is rife may have lower IQs because they have to divert energy away from brain development to fight infections, scientists in the US claim.

The controversial idea might help explain why national IQ scores differ around the world, and are lower in some warmer countries where debilitating parasites such as malaria are widespread, they say.

Researchers behind the theory claim the impact of disease on IQ scores has been under-appreciated, and believe it ranks alongside education and wealth as a major factor that influences cognitive ability.

[…]

The actual research article: http://rspb.royalsocietypublishing.org/content/early/2010/06/29/rspb.2010.0973.full?sid=f65fe5b5-b8d4-4e62-82ee-60c7bd44e3d3

Abstract

In this study, we hypothesize that the worldwide distribution of cognitive ability is determined in part by variation in the intensity of infectious diseases. From an energetics standpoint, a developing human will have difficulty building a brain and fighting off infectious diseases at the same time, as both are very metabolically costly tasks. Using three measures of average national intelligence quotient (IQ), we found that the zero-order correlation between average IQ and parasite stress ranges from r = ?0.76 to r = ?0.82 (p < 0.0001). These correlations are robust worldwide, as well as within five of six world regions. Infectious disease remains the most powerful predictor of average national IQ when temperature, distance from Africa, gross domestic product per capita and several measures of education are controlled for. These findings suggest that the Flynn effect may be caused in part by the decrease in the intensity of infectious diseases as nations develop.

For reference, the Flynn effect:  http://en.wikipedia.org/wiki/Flynn_effect

The Flynn effect describes an increase in the average intelligence quotient (IQ) test scores over generations (IQ gains over time). Similar improvements have been reported for other cognitions such as semantic and episodic memory.[1]  The effect has been observed in most parts of the world at different rates.

The Flynn effect is named for James R. Flynn, who did much to document it and promote awareness of its implications. The term itself was coined by the authors of The Bell Curve.[2]

The effect’s increase has been continuous and approximately linear from the earliest years of testing to the present. There are numerous explanations to the Flynn effect and also some criticism. There is currently a discussion if the Flynn effect has ended in some developed nations since the mid 1990s.

Political comic strips around the Mississippi Bubble of the 1710s

I wish that I had time to read this paper by David Levy and Sandra Peart.

It’s about political comics (cartoons) drawn to depict John Law and the Mississippi Bubble of the early 1700s.  It also speaks to subtlely different meanings of the words “alchemy” and “occult” than we are used to today. Here is an early paragraph in the paper:

Non-transparency induces a hierarchy of knowledge. The most extreme form of that sort of hierarchy might be called the cult of expertise in which expertise is said to be accompanied by godlike powers, the ability to unbind scarcity of matter and time. The earliest debates over hierarchy focused on whether such claims are credible or not.

Here is the abstract:

Economists have occasionally noticed the appearance of economists in cartoons produced for public amusement during crises. Yet the message behind such images has been less than fully appreciated. This paper provides evidence of such inattention in the context of the eighteenth century speculation known as the Mississippi Bubble. A cartoon in The Great Mirror of Folly imagines John Law in a cart that flies through the air drawn by a pair of beasts, reportedly chickens. The cart is not drawn by chickens, however, but by a Biblical beast whose forefather spoke to Eve about the consequences of eating from the tree of the knowledge. The religious image signifies the danger associated with knowledge. The paper thus demonstrates how images of the Mississippi Bubble focused on the hierarchy of knowledge induced by non-transparency. Many of the images show madness caused by alchemy, the hidden or “occult.”

Hat tip: Tyler Cowen.

Epistemology in the social sciences (economics included)

I’m not sure how I came across it, but Daniel Little has a post summarising a 2006 article by Gabriel Abend:  “Styles of Sociological Thought: Sociologies, Epistemologies, and the Mexican and U.S. Quests for Truth“.  Daniel writes:

Abend attempts to take the measure of a particularly profound form of difference that might be postulated within the domain of world sociology: the idea that different national traditions of sociology may embody different epistemological frameworks that make their results genuinely incommensurable.

[…]

Consider this tabulation of results on the question of the role of evidence and theory taken by the two sets of articles:

[…]

Here is a striking tabulation of epistemic differences between the two samples:

Abend believes that these basic epistemological differences between U.S. and Mexican sociology imply a fundamental incommensurability of results:

To consider the epistemological thesis, let us pose the following thought experiment. Suppose a Mexican sociologist claims p and a U.S. sociologist claims not-p.  Carnap’s or Popper’s epistemology would have the empirical world arbitrate between these two theoretical claims. But, as we have seen, sociologists in Mexico and the United States hold different stances regarding what a theory should be, what an explanation should look like, what rules of inference and standards of proof should be stipulated, what role evidence should play, and so on. The empirical world could only adjudicate the dispute if an agreement on these epistemological presuppositions could be reached (and there are good reasons to expect that in such a situation neither side would be willing to give up its epistemology). Furthermore, it seems to me that my thought experiment to some degree misses the point. For it imagines a situation in which a Mexican sociologist claims p and a U.S. sociologist claims not-p, failing to realize that that would only be possible if the problem were articulated in similar terms. However, we have seen that Mexican and U.S. sociologies also differ in how problems are articulated—rather than p and not-p, one should probably speak of p and q.  I believe that Mexican and U.S. sociologies are perceptually and semantically incommensurable as well. (27)

Though Abend’s analysis is comparative, I find his analysis of the epistemological assumptions underlying the U.S. cases to be highly insightful all by itself.  In just a few pages he captures what seem to me to be the core epistemological assumptions of the conduct of sociological research in the U.S.  These include:

  • the assumption of “general regular reality” (the assumption that social phenomena are “really” governed by underlying regularities)
  • deductivism
  • epistemic objectivity
  • a preference for quantification and abstract vocabulary
  • separation of fact and value; value neutrality

There is a clear (?) parallel dispute in the study of economics as well, made all the more complicated by the allegations leveled at economics as a discipline as a result of the global financial crisis.

Changing the typesetting margins in Scientific Workplace

At least half of the LSE economics department uses Scientific Workplace, but an absurdly large fraction of all PDFs they produce have two-inch margins so they end up wasting half the page.

I finally got sufficiently annoyed to discover how to change it:

  1. Open a SW tex file
  2. Under the ‘Typeset’ menu, choose ‘Options and Packages…’
  3. Under the ‘Packages’ tab, add the ‘geometry’ package
  4. Under the ‘Typeset’ menu, choose ‘Preamble…’
  5. Add a line at the end specifying the margins.

For example:

\geometry{left=1in,right=1in,top=1in,bottom=1in}

Units of measurement available are listed on the webpage where I got this:  http://www.mackichan.com/index.html?techtalk/370.htm

Approximating a demand function with shocks to the elasticity of demand

Entirely for my own reference …

A demand function commonly used in macroeconomics is the following, derived from a Dixit-Stiglitz aggregator and exhibiting a constant own-price elasticity of demand ($$\gamma$$):

$$!Q_{it}=\left(\frac{P_{it}}{P_{t}}\right)^{-\gamma}Q_{t}$$

A demand-side shock can then be modelled as a change in the elasticity of demand:

$$!Q_{it}=\left(\frac{P_{it}}{P_{t}}\right)^{-\gamma D_{t}}Q_{t}$$

Where $$\ln\left(D_{t}\right)$$ is, say, Normally distributed and plausibly autocorrelated.  We can rewrite this as a function of (natural) log deviations from long-run trends:

$$!Q_{it}=\overline{Q_{t}}e^{q_{t}-\gamma e^{d_{t}}\left(p_{it}-p_{t}\right)}$$

Where:

  • Variables with a bar above them are long-run trends:  $$\overline{X_{it}}$$
  • Lower-case variables are natural log deviations from their long run trends (so that for small deviations, they may be thought of as the percentage difference from trend):  $$x_{it}=\ln\left(X_{it}\right)-\ln\left(\overline{X_{it}}\right)$$
  • The long-run trend of all prices is to equal the aggregate price:  $$\overline{P_{it}}=\overline{P_{t}}$$
  • The long-run trend of $$D_{t}$$ is unity

We’ll construct a quadratic approximation around $$q_{t}=p_{it}=p_{t}=d_{t}=0$$ but, first, a table of partial derivatives for a more general function:

Function Value at $$x=y=z=0$$
$$f\left(x,y,z\right)=ae^{x+bye^{z}}$$ $$a$$
$$f_{x}\left(x,y,z\right)=ae^{x+bye^{z}}$$ $$a$$
$$f_{y}\left(x,y,z\right)=abe^{x+bye^{z}+z}$$ $$ab$$
$$f_{z}\left(x,y,z\right)=abye^{x+bye^{z}+z}$$ $$0$$
$$f_{xx}\left(x,y,z\right)=ae^{x+bye^{z}}$$ $$a$$
$$f_{yy}\left(x,y,z\right)=ab^{2}e^{x+bye^{z}+2z}$$ $$ab^{2}$$
$$f_{zz}\left(x,y,z\right)=abye^{x+bye^{z}+z}+ab^{2}y^{2}e^{x+bye^{z}+2z}$$ $$0$$
$$f_{xy}\left(x,y,z\right)=abe^{x+bye^{z}+z}$$ $$ab$$
$$f_{xz}\left(x,y,z\right)=abye^{x+bye^{z}+z}$$ $$0$$
$$f_{yz}\left(x,y,z\right)=abe^{x+bye^{z}+z}+ab^{2}ye^{x+bye^{z}+2z}$$ $$ab$$

So that in the vicinity of $$x=y=z=0$$, the function $$f\left(x,y,z\right)$$ is approximated by:

$$!f\left(x,y,z\right)\simeq a + a\left(x+by\right) + a\left[\frac{1}{2}\left(x+by\right)^{2}+byz\right]$$

From which we can infer that:

$$!Q_{it}\simeq \overline{Q_{t}}\left[1+\left(q_{t}-\gamma\left(p_{it}-p_{t}\right)\right) + \frac{1}{2}\left(q_{t}-\gamma\left(p_{it}-p_{t}\right)\right)^{2}-\gamma\left(p_{it}-p_{t}\right)d_{t}\right]$$

If introduced to a profit function, the first-order components ($$q_{t}-\gamma\left(p_{it}-p_{t}\right)$$) would vanish as individual prices will be optimal in the long run.

Update (20 Jan 2010): Added the half in each of the last equations.

Not raising the minimum wage with inflation will make your country fat

Via Greg Mankiw, here is a new working paper by David O. Meltzer and Zhuo Chen: “The Impact of Minimum Wage Rates on Body Weight in the United States“. The abstract:

Growing consumption of increasingly less expensive food, and especially “fast food”, has been cited as a potential cause of increasing rate of obesity in the United States over the past several decades. Because the real minimum wage in the United States has declined by as much as half over 1968-2007 and because minimum wage labor is a major contributor to the cost of food away from home we hypothesized that changes in the minimum wage would be associated with changes in bodyweight over this period. To examine this, we use data from the Behavioral Risk Factor Surveillance System from 1984-2006 to test whether variation in the real minimum wage was associated with changes in body mass index (BMI). We also examine whether this association varied by gender, education and income, and used quantile regression to test whether the association varied over the BMI distribution. We also estimate the fraction of the increase in BMI since 1970 attributable to minimum wage declines. We find that a $1 decrease in the real minimum wage was associated with a 0.06 increase in BMI. This relationship was significant across gender and income groups and largest among the highest percentiles of the BMI distribution. Real minimum wage decreases can explain 10% of the change in BMI since 1970. We conclude that the declining real minimum wage rates has contributed to the increasing rate of overweight and obesity in the United States. Studies to clarify the mechanism by which minimum wages may affect obesity might help determine appropriate policy responses.

Emphasis is mine.  There is an obvious candidate for the mechanism:

  1. Minimum wages, in real terms, have been falling in the USA over the last 40 years.
  2. Minimum-wage labour is a significant proportion of the cost of “food away from home” (often, but not just including, fast-food).
  3. Therefore the real cost of producing “food away from home” has fallen.
  4. Therefore the relative price of “food away from home” has fallen.
  5. Therefore people eat “food away from home” more frequently and “food at home” less frequently.
  6. Typical “food away from home” has, at the least, more calories than “food at home”.
  7. Therefore, holding the amount of exercise constant,  obesity rates increased.

Update: The magnitude of the effect for items 2) – 7) will probably be greater for fast-food versus regular restaurant food, because minimum-wage labour will almost certainly comprise a larger fraction of costs for a fast-food outlet than it will for a fancy restaurant.

The likelihood-ratio threshold is the shadow price of statistical power

Cosma Shalizi, an associate professor in statistics at Carnegie Mellon University, gives an interpretation of the likelihood-ratio threshold in an LR test: It’s the shadow price of statistical power:

[…]

Suppose we know the probability density of the noise p and that of the signal is q. The Neyman-Pearson lemma, as many though not all schoolchildren know, says that then, among all tests off a given size s, the one with the smallest miss probability, or highest power, has the form “say ‘signal’ if q(x)/p(x) > t(s), otherwise say ‘noise’,” and that the threshold t varies inversely with s. The quantity q(x)/p(x) is the likelihood ratio; the Neyman-Pearson lemma says that to maximize power, we should say “signal” if its sufficiently more likely than noise.

The likelihood ratio indicates how different the two distributions — the two hypotheses — are at x, the data-point we observed. It makes sense that the outcome of the hypothesis test should depend on this sort of discrepancy between the hypotheses. But why the ratio, rather than, say, the difference q(x) – p(x), or a signed squared difference, etc.? Can we make this intuitive?

Start with the fact that we have an optimization problem under a constraint. Call the region where we proclaim “signal” R. We want to maximize its probability when we are seeing a signal, Q(R), while constraining the false-alarm probability, P(R) = s. Lagrange tells us that the way to do this is to minimize Q(R) – t[P(R) – s] over R and t jointly. So far the usual story; the next turn is usually “as you remember from the calculus of variations…”

Rather than actually doing math, let’s think like economists. Picking the set R gives us a certain benefit, in the form of the power Q(R), and a cost, tP(R). (The ts term is the same for all R.) Economists, of course, tell us to equate marginal costs and benefits. What is the marginal benefit of expanding R to include a small neighborhood around the point x? Just, by the definition of “probability density”, q(x). The marginal cost is likewise tp(x). We should include x in R if q(x) > tp(x), or q(x)/p(x) > t. The boundary of R is where marginal benefit equals marginal cost, and that is why we need the likelihood ratio and not the likelihood difference, or anything else. (Except for a monotone transformation of the ratio, e.g. the log ratio.) The likelihood ratio threshold t is, in fact, the shadow price of statistical power.

It seems sensible to me.

Who has more information, the Central Bank or the Private Sector?

A friend pointed me to this paper:

Svensson, Lars E. O. and Michael Woodford. “Indicator Variables For Optimal Policy,” Journal of Monetary Economics, 2003, v50(3,Apr), 691-720.

You can get the NBER working paper (w8255) here.  The abstract:

The optimal weights on indicators in models with partial information about the state of the economy and forward-looking variables are derived and interpreted, both for equilibria under discretion and under commitment. The private sector is assumed to have information about the state of the economy that the policymaker does not possess. Certainty-equivalence is shown to apply, in the sense that optimal policy reactions to optimally estimated states of the economy are independent of the degree of uncertainty. The usual separation principle does not hold, since the estimation of the state of the economy is not independent of optimization and is in general quite complex. We present a general characterization of optimal filtering and control in settings of this kind, and discuss an application of our methods to the problem of the optimal use of ‘real-time’ macroeconomic data in the conduct of monetary policy. [Emphasis added by John Barrdear]

The sentence I’ve highlighted is interesting.  As written in the abstract, it’s probably true.  Here’s a paragraph from page two that expands the thought:

One may or may not believe that central banks typically possess less information about the state of the economy than does the private sector. However, there is at least one important argument for the appeal of this assumption. This is that it is the only case in which it is intellectually coherent to assume a common information set for all members of the private sector, so that the model’s equations can be expressed in terms of aggregative equations that refer to only a single “private sector information set,” while at the same time these model equations are treated as structural, and hence invariant under the alternative policies that are considered in the central bank’s optimization problem. It does not make sense that any state variables should matter for the determination of economically relevant quantities (that is, relevant to the central bank’s objectives), if they are not known to anyone in the private sector. But if all private agents are to have a common information set, they must then have full information about the relevant state variables. It does not follow from this reasoning, of course, that it is more accurate to assume that all private agents have superior information to that of the central bank; it follows only that this case is one in which the complications resulting from partial information are especially tractable. The development of methods for characterizing optimal policy when di fferent private agents have di fferent information sets remains an important topic for further research.

Here’s my attempt as paraphrasing Svensson and Woodford in point form:

  1. The real economy is the sum of private agents (plus the government, but ignore that)
  2. Complete information is thus, by definition, knowledge of every individual agent
  3. If we assume that everybody knows about themselves (at least), then the union of all private information sets must equal complete information
  4. The Central Bank observes only a sample of private agents
  5. That is, the Central Bank information set is a subset of the union of all private information sets. The Central Bank’s information cannot be greater than the union of all private information sets.
  6. One strategy in simplifying the Central Bank’s problem is to assume that private agents are symmetric in information (i.e. they have a common information set).  In that case, we’d say that the Central Bank cannot have more information than the representative private sector agent. [See note 1 below]
  7. Important future research will involve relaxing the assumption in (f) and instead allowing asymmetric information across different private agents.  In that world, the Central Bank might have more information than any given private agent, but still less than the union of all private agents.

Svensson and Woodford then go on to consider a world where the Central Bank’s information set is smaller than (i.e. is a subset of) the Private Sector’s common information set.

But that doesn’t really make sense to me.

If private agents share a common information set, it seems silly to suppose that the Central Bank has less information than the Private Sector, for the simple reason that the mechanism of creating the common information set – commonly observable prices that are sufficient statistics of private signals – is also available to the Central Bank.

In that situation, it seems more plausible to me to argue that the CB has more information than the Private Sector, provided that their staff aren’t quietly acting on the information on the side.  It also would result in observed history:  the Private Sector pays ridiculous amounts of attention to every word uttered by the Central Bank (because the Central Bank has the one private signal that isn’t assimilated into the price).

Note 1: To arrive at all private agents sharing a common information set, you require something like the EMH (in fact, I can’t think how you could get there without the EMH).  A common information set emerges from a commonly observable sufficient statistic of all private information.  Prices are that statistic.

    Be careful interpreting Lagrangian multipliers

    Last year I wrote up a derivation of the New Keynesian Phillips Curve using Calvo pricing.  At the start of it, I provided the standard pathway from the Dixit-Stiglitz aggregator for consumption to the constant own-price elasticity individual demand function.  Let me reproduce it here:

    There is a constant and common elasticity of substitution between each good: $$\varepsilon>1$$.  We aggregate across the different consumptions goods:

    $$!C=\left(\int_{0}^{1}C\left(i\right)^{\frac{\varepsilon-1}{\varepsilon}}di\right)^{\frac{\varepsilon}{\varepsilon-1}}$$

    $$P\left(i\right)$$ is the price of good i, so the total expenditure on consumption is $$\int_{0}^{1}P\left(i\right)C\left(i\right)di$$

    A representative consumer seeks to minimise their expenditure subject to achieving at least $$C$$ units of aggregate consumption. Using the Lagrange multiplier method:

    $$!L=\int_{0}^{1}P\left(i\right)C\left(i\right)di-\lambda\left(\left(\int_{0}^{1}C\left(i\right)^{\frac{\varepsilon-1}{\varepsilon}}di\right)^{\frac{\varepsilon}{\varepsilon-1}}-C\right)$$

    The first-order conditions are that, for every intermediate good, the first derivative of $$L$$ with respect to $$C\left(i\right)$$ must equal zero. This implies that:

    $$!P\left(i\right)=\lambda C\left(i\right)^{\frac{-1}{\varepsilon}}\left(\int_{0}^{1}C\left(j\right)^{\frac{\varepsilon-1}{\varepsilon}}dj\right)^{\frac{1}{\varepsilon-1}}$$

    Substituting back in our definition of aggregate consumption, replacing $$\lambda$$ with $$P$$ (since $$\lambda$$ represents the cost of buying an extra unit of the aggregate good $$C$$) and rearranging, we end up with the demand curve for each intermediate good:

    $$!C\left(i\right)=\left(\frac{P\left(i\right)}{P}\right)^{-\varepsilon}C$$

    If that Lagrangian looks odd to you, or if you’re asking where the utility function’s gone, you’re not alone.  It’s obviously just the dual problems of consumer theory – the fact that it doesn’t matter if you maximise utility subject to a budget constraint or minimise expenditure subject to a minimum level of utility – but what I want to focus on is the resulting interpretation of the lagrangian multipliers.

    Let’s rephrase the problem as maximising utility, with utility a generic function of aggregate consumption, $$U\left(C\right)$$.  The Lagrangian is then:

    $$!L=U\left(\left(\int_{0}^{1}C\left(i\right)^{\frac{\varepsilon-1}{\varepsilon}}di\right)^{\frac{\varepsilon}{\varepsilon-1}}\right)+\mu\left(M-\int_{0}^{1}P\left(i\right)C\left(i\right)di\right)$$

    The first-order conditions are:

    $$!U’\left(C\right)\left(\int_{0}^{1}C\left(j\right)^{\frac{\varepsilon-1}{\varepsilon}}dj\right)^{\frac{1}{\varepsilon-1}}C\left(i\right)^{\frac{-1}{\varepsilon}}=\mu P\left(i\right)$$

    Rearranging and substituting back in the definition for $$C$$ then gives us:

    $$!C\left(i\right)=\left(P\left(i\right)\frac{\mu}{U’\left(C\right)}\right)^{-\varepsilon}C$$

    In the first approach, $$\lambda$$ represents the cost of buying an extra unit of the aggregate good $$C$$, which is the definition of the aggregate price level.  In the second approach, $$\mu$$ represents the cost of buying an extra unit of income, which is not the same thing.  Comparing the two results, we can see that:

    $$!\lambda=P=\frac{U’\left(C\right)}{\mu}$$

    Which should cause you to raise an eyebrow.  Why aren’t the two multipliers just the inverses of each other?  Aren’t they meant to be?  Yes, they are, but only when the two problems are equivalent.  These two problems are slightly different.

    In the first one, to be equivalent, the term in the lagrangian would need to be $$V – U\left(\left(\int_{0}^{1}C\left(i\right)^{\frac{\varepsilon-1}{\varepsilon}}di\right)^{\frac{\varepsilon}{\varepsilon-1}}\right)$$, which would give us Hicksian demands as a function of utility level ($$V$$).  But since we assumed that utility is only a function of aggregate consumption, then in order to pin down a level of utility, it’s sufficient to pin down a level of aggregate consumption; and that is useful to us because a) a level of utility doesn’t mean much to us as macroeconomists but a level of aggregate consumption does and b) it means that we can recognise the lagrange multiplier as the aggregate price level.

    Which, when you think about it, makes perfect sense.  Extra income must be adjusted by the marginal value of the extra consumption it affords in order to arrive at the price that the (representative) consumer would be willing to pay for that consumption.

    In other words:  be careful when interpreting your Lagrangian multipliers.