Cato Online Forum

Decision Markets as Meta-Policy

By Robin Hanson
November 2014

We economists know of many promising ways to grow the economy. Some ways, like more immigration or wasting less on medicine and the military, could give big but mostly one-time boosts. Other ways could keep boosting growth for a long time, by increasing our rates of innovation. For example, we might reduce excess regulation that cuts our flexibility in land use, occupation choice, venture financing, and product design. Or we might redesign intellectual property, or coordinate better globally to promote innovation.

All of these options, however, are dangerous, as badly implemented versions could do net harm. After all, the world is very complicated, requiring good policy to attend to a great many messy details. So what we really need, more than suggestions for good policy directions, are very specific policy proposals that well address all the messy political, economic, and administrative details.

Unfortunately, one more often sees writings like those in this collection, with high-status impressive folks offering articulate and thought-provoking but not very specific commentary. Yes, this has value, but the main problems happen elsewhere, in the dark corners where political and policy sausage gets made. So what would actually help growth the most is to improve this policy process.

That is, even more than better growth policy, we need better meta-policy, i.e. policy on how to pick policy. And it seems to me that our biggest meta-policy failure is the failure to aggregate information. On specific policy proposals, we fail to collect what everyone does or could know into visible consensus estimates about their likely consequences. So when ultimate decision makers choose to support or oppose specific proposals, they are often quite badly informed about their consequences. Far fewer bad policies would be approved if everyone knew that their consequences were bad.

I think we can do better by using decision markets. Decision markets are speculative financial markets on the consequences of specific policies. For example, to see if a particular bill to subsidize research would actually increase long-term growth, we might create two speculative markets. In one market, speculators would estimate real GDP in twenty years conditional on passing this bill (and on keeping it passed). In the other market, speculators would estimate real GDP in twenty years conditional on never passing this bill (or anything like it). (Markets would trade assets that later pay inflation-adjusted cash in proportion to future GDP, with trades called-off when conditions are unmet).

The difference in the prices between these two markets would estimate the effect of this research subsidy bill on future GDP. Since speculative markets are the best mechanism we know for robustly aggregating information, this estimate should on average be at least as accurate as those from competing mechanisms. This estimate would also be hard for interested parties to manipulate. And it would let informed parties profit by correcting for observed biases. Yes, since one will never observe both outcomes (bill passed vs. not passed), one would never know if the effect estimate was correct. Even so, speculators would have good incentives to estimate well, which would make this estimate a good basis for choosing policy.

One could imagine using markets like these in a purely advisory capacity. In this case, decision makers would be free to use or ignore market advice as they wish. But alas I expect the ultimate decision makers in democracies, i.e., voters, would pay them too little attention. And if voters ignored the markets, representatives and agency heads would mostly ignore them as well. So I’d prefer to have decision markets directly choose policy.

And so now we come to my specific proposed meta-policy. I call it “futarchy,” and it has four parts. First, I propose that our national legislatures pass bills to define national welfare, and fund and authorize an agency to collect statistics to measure this numerical quantity after the fact. National welfare could initially be defined as a discounted sum of future real GDP, but afterward the legislatures could pass bills to measure and include other important factors like leisure, happiness, artistic achievement, etc.

Second, I propose we create an open bounty system for proposing policies to increase national welfare. For example, anyone who pays one million dollars might convert their specific bill to be approved or not on a specific day into an official proposal. This proposer might win ten million dollars if his or her proposal is officially adopted. (I offer specific dollar figures here for concreteness, but I expect a wide range of figures would work out okay).

Third, I propose that we create two open speculative decision markets for each official proposal, to estimate national welfare given that we do or do not adopt this proposal. These markets would have low transaction fees, match best offers into trades, and produce a history of trading prices. They would be open to all who could make good on trade offers. If over the decision day the average if-adopted price is higher than the average if-not-adopted price (plus average bid-ask spread), then the proposal is officially adopted. An adopted proposal becomes a new law of the land.

Fourth, I suggest a few minor adjustments. To avoid damage from temporary errors in national welfare definition, we could create two more markets per proposal, this time using national welfare as it will be defined one year later. Disapproval by the future-welfare market estimate would veto a proposal. We could also subsidize these speculative markets to encourage more trading, though efforts to manipulate the decisions should give plenty of trading activity. Also, we could let this system become recursive, approving other policies like it but with less restrictive rules.

So there it is, my specific proposed meta-policy, intended to help us to better aggregate information on the consequences of policy proposals. If adopted, it would help us to more often choose growth-promoting policies, at least when we are willing to pay the short-term costs required to get long-term growth. And I expect this to happen often enough to greatly increase economic growth rates.

Of course my proposal does have one huge and perhaps fatal flaw: maybe voters don’t care much about policy effectiveness. Maybe voters instead want the high status meta-policy that all the other cool nations have. Or maybe voters prefer to hypocritically say they want some things, while electing representatives who understand that they really want other things. Or maybe voters just love the group-bonding ecstasy of crushing opponents via not-particularly-rational us-them tribal chants. All of these are reasons why voters might not want to change to a system that produces better policy outcomes.

So, I think I can identify a policy proposal that, if adopted, would greatly increase growth rates, as well as many other good policy outcomes. It would do this by consistently adopting specific policy proposals that the best informed people think will actually achieve good outcomes. I also think that once people had lived under this system for a while, they’d be pretty happy with it. But even so, I’m not sure how to get voters today to want to adopt it. You can lead voters to good meta-policy waters, but you can’t make them drink from it.


The opinions expressed here are solely those of the author and do not necessarily reflect the views of the Cato Institute. This essay was prepared as part of a special Cato online forum on reviving economic growth.

Robin Hanson is an associate professor of economics at George Mason University.