# heavytailed

## May 13, 2013

### On Economics, ENCODE, and blogging (Monday Morning Blogroll)

Filed under: Uncategorized — heavytailed @ 4:29 pm

The Econowonkoblogosphere is a wonderful source of distraction, entertainment, and insight. I’m not sure something equivalent exists in any other field, where a group of independent thinkers routinely dissect not only scientific issues, but policy implications as well. Belatedly we find posts suggesting that sovereign bond spreads were not measures of soundness, but in fact reflective of market panic (seriously this is 3 months old at this point). One conclusion to draw is that the ECB did the right thing in stepping in as a lender of last resort. Across the pond, there is more refinement on the inflation expectations side-channel which describes one monetarist view of the Fed’s treasury-buying activities, and the dismay it is causing macro hedges who, DeLong suspects, may view the Fed as a “normal” market participant rather than as a central bank charged to push down both inflation and unemployment. Vox (linked by Krugman) provides a historical case-study of the expectations side-channel by examining the Bank of England’s policies in the 1930s. At the same time, a three months-old speech given by Jeremy Stein raises some potential risks of quantitative easing: market overheat driven by reaching for yield, earning a call for critiques of the reach-for-yield hypothesis  from Brad DeLong. Mike Konczal answers the call, suggesting that this behavior is not necessarily caused by low interest rates, that the empirical evidence for it is also evidence for what FT Alphaville calls “regulatory arbitrage,” and suggests perhaps more regulation on top of QE is a solution. While speaking on a different topic, Paul Krugman suggests that if QE is to succeed, it is in spite of people calling for the Fed to be more responsible, as the belief that the Reserve will alter its policy undercuts the expectations side-channel through which it works.

(It’s a little bit unclear to me what “reaching for yield” means in the context of high inflation expectations. The argument seems to be that in the context of low interest rates, where yields on debt are low, entities with fixed servicing costs, like banks, or promised minimum rates of return, are forced to take on riskier debt to cover those costs or promised return. However, if the effect of QE is to raise inflation expectations, couldn’t that entity snatch up some equity?)

Okay. So I’m preeminently unqualified to comment on any of these topics. However what I can comment on is the difference between this kind of economic blogging and what exists in the genetics world. You see, these economic thinkers have an idea to wrestle with (or, more aptly, against), and marshal theoretical arguments and data to support what is ultimately a policy position. Any of these blogs are an entrypoint into a vast exchange of ideas, something that doesn’t really happen in the genetics blogosphere. There are tons of independent directions, but rarely do a group of people talk to each other through blogs. One might imagine a number of reasons, from the relative proportion of personality “types” in each field to the differential publicity that each receives. The real reasons I think are twofold: first (and alarmingly) blog-type commentary, opinions, and tone seem to have a life as peer-reviewed publications (this has a serious cooling effect on the exchange of ideas); and second there’s really not a lot to disagree about. The whole ENCODE hubub is semantic quibbling over what is meant by the word “function”, and does not tease out something in any way fundamental about genetics. So while we can quibble about experiments and parameters, we don’t ever bother about the underlying model (maybe about how to decompose it). In other words

$y \sim f(G,E)$

(phenotype is some function of genes and environment) is incontrovertible. Ask about the variance of $y$, and then expand $f$ in a Taylor series and you get the fundamental identity of statistical genetics

$\sigma_y = \sigma_E + \sigma_A + \sigma_D + \sigma_{AE} + \sigma_{DE} + \sigma_{AA} + \dots$

In some sense this is because we’ve “black boxed” genetics into a rather unassailable general form. Not that it makes it easy to accurately measure someone’s risk of, say, cancer, but there aren’t really cases where my $\sigma_A$ is different from your $\sigma_A$, (while it’s not uncommon for economic identities like $I=S$ to vary by context).

TL;DR blogs like mine are boring (I try not to read it, myself), so I’m linking out to other things