Links to Consider, 3/21

Emily Oster on observational studies; Paul Dobransky on boundaries; Jean Twenge on academic pressure; Bob Eisenbeis explains an odd feature of the Credit Suisse resolution;

Emily Oster writes,

in typical data you cannot observe and adjust for all differences. You do not see everything about people. Sometimes this is simply because our variables are rough: we see whether someone has a family income above or below the poverty line, but not any more details, and those details are important. There are also characteristics we almost never capture in data, like How much do you like exercise? or How healthy are your partner’s behaviors? or even Where is the closest farmers’ market? 

For both of these reasons, in nearly all examples, we worry about residual confounding. That’s the concern that there are still other important differences across groups that might drive the results. Most papers list this possibility in their “limitations” section. 

We all agree that this is a concern. Where we differ is in how much of a limitation we believe it to be. In my view, in these contexts (and in many others), residual confounding is so significant a factor that it is hopeless to try to learn causality from this type of observational data. 

Hear, hear. Brendan Case and Ying Chen recently wrote in the WSJ that marriage makes women happier.

In most cases, we were able to control for the nurses’ well-being and health in 1989, before any of them had gotten married, as well as for a host of other relevant factors, such as age, race and socioeconomic status. This helped us to rule out the possibility that, for example, happiness predicted marriage rather than being predicted by it

No it didn’t. This approach of “control for” was discredited fifty years ago. It is what Oster does not trust. Modern economists look for something that looks more like experimental data. I don’t know how you would do this to answer this question. It might very well be true that women are happier married than unmarried, but this kind of study does not provide evidence for that.

Paul Dobranksy writes,

Narcissism is, then, today’s euphemism for “immaturity” (or low character), while “emotional intelligence” is today’s euphemism for “maturity” (or high character.)

Pointer from Rob Henderson. Dobransky’s essay is difficult to summarize. Other excerpts:

It’s [i.e., projection] scolding, but with the added dimension of accusing others of something they’re not and instead what we ourselves are trying to dispense with.

Projection is also what is at work in any pejoratives—heaped on those we disagree with but who are trying to offer us something of themselves, if not actual, hard-won wisdom—such as “mansplaining” or “nagging” or “lecturing at me.”

The unembodied experience of electronic communication and social media, not physically spoken in person, strongly encourages the most “primitive ego defenses” to flourish to the neglect of the “mature ego defenses” that are characterized by solid boundaries.

I have said that smart phones dissolve the distinction between our intimate world (friends, family, co-workers) and the remote world (celebrities, criminals, politicians). They all show up on the same screen, with celebrities pretending to be our friends and friends pretending to be celebrities.

Does this relate to Dobransky’s psychological analysis? We’ll see as he develops his thesis.

I have seen a couple of references to a critique of Jonathan Haidt that says that mental health problems among young people preceded social media, and instead the cause of mental health problems is that young people are under too much pressure to succeed, particularly in school. I did not link to the critique, which was heavily based on anecdotes and citations of experts/authorities claiming that academic pressure is/was a big factor, but without any statistical support. If I find the critique again (is Derek Thompson in the Atlantic the source?), I will link to it. In any case, Jean M. Twenge has a rebuttal, posted on Haidt’s substack.

The good thing about an argument like this is we can test it. Three questions come to mind based on these recent pieces and observations:

1. Do Gen Z teens spend more time on homework than previous generations at the same age? Did homework time increase concurrently with teen depression?

2. Do Gen Z teens report feeling more academic pressure than previous generations at the same age? Did feelings of academic pressure increase concurrently with teen depression?

3. Are higher-achieving teens — those fighting for good grades and aiming for college — more likely to be depressed than lower-achieving teens? And has depression increased more sharply among high-achieving teens compared to less academically oriented teens?

The Monitoring the Future study of U.S. teens collects a nationally representative sample in schools every year and has data that can answer all three of those questions. The answers don’t look good for the academic pressure hypothesis.

Bob Eisenbeis wrote a few years ago,

There are three basic forms of CoCos. One type, commonly called “going‐​concern” or “bail‐​in” CoCos, convert existing debt into common equity when a specific conversion event, or “trigger,” occurs.⁷ Going‐​concern CoCos do not result in the injection of new funds into a troubled institution. Instead, they simply convert existing debt instruments on a firm’s balance sheet into common equity, thereby increasing its capital, facilitating deleveraging, and restoring capital adequacy.

The second form, “write‐​down” or “bail‐​out” CoCos, are debt instruments whose values are written down when a trigger is breached.⁸ Write‐​down CoCos simultaneously reduce a firm’s assets and its liabilities, thereby facilitating the firm’s reorganization at the point of nonviability. The resulting bailout can be permanent, temporary, partial, or total. Because write‐​down CoCos reward stockholders at the expense of the CoCo investors, they reverse the traditional rules of seniority, placing shareholders’ interests ahead of those of a particular set of debt holders.

This helps clear up some confusion. The “bail-in” version is the one I remember being proposed by center-right economists after the S&L Crisis as a way to protect taxpayers from bank failures. The “write-down” version is what Credit Suisse had, which seems to have caught off guard many observers, perhaps including some investors who bought the bonds.

You were caught off guard if you wonder how the bondholders could get zeroed out before the shareholders. Well, that’s the way the Swiss designed it.

Substacks references above:

@

Academic Pressure Cannot Explain the Mental Illness Epidemic

@

Sentiment and Facts, Academic Cheating, Paying on Dates

@

Personal Boundaries: The Only Factor that Makes Social Media "Toxic”

@

Why I Look at Data Differently