Update 5/15 by Eric: We send memos too. Here is a fuller version of our concerns.


The Western Climate Initiative—the multi-state, multi-province pact designing a cap-and-trade system for the Western U.S. and, now, large swaths of Canada as well—has contracted out a bunch of economic modeling, in an attempt to get a handle on how various cap-and-trade policies might play out in the real world.

That’s a good thing.  Entering blindly into a cap and trade system would be dopey. 

Smart Kid 140 IstockThat said, your friendly neighborhood research nerds are a little worried about how the economic analysis is being done.  It’s not that we have specific complaints about the model itself.  It’s the opposite—we can’t possibly complain, because the company that the WCI chose to do the modeling won’t tell anyone how their model works.  It’s a proprietary, closed system, you see, so there’s absolutely no way to tell what inherent biases the model may have. 

That’s obviously a substantive problem:  without a lot of eyes scrutinizing the model, flaws can go undetected.  But—far worse—it’s a perception problem.  No matter how the model results turn out, people can point to the fact that the model is closed to challenge or outright reject its conclusions.

We even wrote a little letter to the WCI about these issues.

  • Our work is made possible by the generosity of people like you!

    Thanks to Grant Nishio for supporting a sustainable Cascadia.

  • We’ve got some deeper and more general concerns about the economic modeling, too.  In particular (Geek Alert!) the type of economic model that’s being used here, called a “computable general equilibrium” model, or CGE model, has more than its share of problems and critics, both on mathematical and practical grounds.  For example, there seems to be a surprising lack of research about whether CGE models have much predictive power at all:  nobody seems to have compared the results of CGE modeling exercises done a decade ago, say, with the actual performance of the economy that the models were attempting to predict.  Seems like the sort of thing one would want to know, if you’re using a model to predict the future.

    So for these reasons (among others) I think the WCI should treat modeling results simply as suggestive and potentially informative—rather than as a rock-solid, perfectly reliable indicator of which policies are most worth pursuing.