Tom Pepinsky

Category: Research

  • Regression Estimates the Conditional-Variance-Weighted ZZZZzzzzzzz….

    As part of my Comparative Methods course, I assigned the book Counterfactuals and Causal Inference by (Cornell sociologist) Stephen Morgan and Christopher Winship. That means—do students know this?—that I had read it again myself before we discuss it next Wednesday. On my long plane ride back from Japan, I read the entire book, which is full of delicious quotes like “the OLS estimate represents the conditional-variance-weighted estimate of the underlying causal effects of individuals, δi, where the weights are a function of the conditional variance of D” and “It turns out that the Wald estimator is indeed consistent for δ in this scenario, but only when is δ considered an invariant structural effect.” I slept pretty well.

    But I must emphasize, this book is really useful, and should be read more widely by political scientists than it is. I want the students to understand how to relate the potential outcomes framework to their own research, but my greatest personal interest in the book lies in its treatment of OLS regression, which is really complete. I must confess that I was unaware how difficult it is to interpret OLS regression coefficients within the potential outcomes framework, even with a binary treatment and the “proper” covariates and no selection or endogeneity. We almost never estimate the effect of X on Y. (And this is not just that there isn’t one causal effect out there waiting to be estimated, my point goes deeper.) What we do instead is calculate statistics that if we are lucky have some causal interpretation, one that we usually cannot state with much precision.

    This leads to a question. Is there any paper out there that describes simple multivariate OLS with continuous treatments in terms of the potential outcomes framework? I mean really simple: Y = a + b1X1 + b2X2, where both X1 and X2 are continuous and there are no issues of selection, endogeneity, etc. to worry about. What precise causal effect does b1 estimate in this model? Morgan and Winship do this for the cases where X1 is binary, and suggest ways to generalize this to many-valued treatments, but I’m interested in how to match the way that we talk about b1 in most observational work using regressions (“holding X2 constant, a one-unit increase in X1 is associated with a b1 increase in Y”) with the potential outcomes way of discussing the same (“the something-something-something treatment effect of X1 is b1“).

  • Talking Crises

    A short note from Japan, where I am visiting for what is officially the shortest trans-Pacific work trip of my life (38 hours and 55 minutes, assuming my flight tomorrow night leaves on time). I’m here to talk about crises, and specifically, the effects of the Global Economic Crisis in Southeast Asia and what they tell us about regional political economy.

    “Wait,” you say, “the GEC didn’t have any effect on Southeast Asia!” Yes, but that’s the point: Southeast Asia was hammered by the Asian Financial Crisis in the late 1990s, but escaped this one relatively unscathed. Sure, growth rates slowed in Malaysia and turned negative in Singapore, but Indonesia cruised along quite nicely. The Philippines and Thailand haven’t done as well, but their troubles stem from domestic political uncertainty and long-term structural issues in their political economies rather than some form of international financial contagion. It’s often observed that the last crisis was about financial contagion, whereas the exposure in Asia today is through trade, not financial, channels. That’s true, but it’s not just a statement of fact, it’s a phenomenon to be explained.

    The non-crises in Southeast Asia demand explanation, so I’m here to offer my own. Briefly, my argument is not that Southeast Asia shaped up over the past decade, but rather that changes in foreign perceptions about Southeast Asia’s political economy after the last crisis lowered the ex ante vulnerability to external financial shocks in the region. For example, exchange rates are no longer misaligned (in part because the last crisis fixed that problem right up), nor is there a sense of unbridled optimism in the West about the growth prospects of emerging Southeast Asia (the same applies).

    Another way to think about this is that having had a crisis in 1998 was a convenient way to ward off having another one in 2008. There is a triumphalist sense among some observers that the fact that the GEC didn’t spread to SE Asia means that local governments have fixed their big political and economic problems, but I don’t think that much real progress has been made on most important issues. It’s not an accident that I title the paper “99 Problems (But A Crisis Ain’t One).”

    Of course, this isn’t as neat as I’d like it to be. One question that bugs me is why the banks and investment houses in Southeast Asia did not jump on the MBS-CDO train the way that we saw in the US and Europe. (I first posted about this in 2010, comparing Singapore and Iceland.) This is puzzling, given means (smart, globally-aware bankers), motive (greed), and opportunity (open KA, plenty of people selling these things). In fact, we do know that one or two important bigshots found themselves heavily exposed in 2008, but the rot wasn’t deep enough to spread to the entire economy. I can’t pretend to know the answer to why this didn’t happen, but I don’t think that enough people in the policy world realize how important it is to pay attention when bad things don’t happen, and to explain these non-crises. So when I lecture to a bunch of Japanese policy scholars tomorrow (or is it today? who knows) on the (non-)effects of the GEC, I’m going to make this point as clearly as I can.