Megan McArdle says med reform will pass, via bills exploiting CBO budget scoring errors: I now put the chances of a substantial health care bill passing at 75%, and the chances of the Democrats losing the house in 2010 at about 66%. … The real game changer is that the CBO is willing to score health care savings on the grounds that the bill contains automatic spending cuts.
But aren’t there substantial organized political groups dedicated to uncovering and promoting the best policies, no matter whose ox they gore? Apparently not.
Just thought of these people (admittedly in the UK):
I gather they have been working with the Conservatives and their Chairman was a Labour minister, which in the UK would suggest they are about as non-partisan as you could reasonably expect.
Well, companies certainly spend millions on lobbying, and influencing budgetary prediction markets could have a pretty high ROI if the prediction markets are small.
I don't know how big you think these prediction markets will be... my single data point is that a few hundred thousand dollars was enough to boost McCain's chance of winning the presidency by 10% on Intrade (see link), which makes me think they'd be pretty small.
So I'd think of it as lobbying spend, not speculation.
Brings up another interesting question - what would happen if the legislation in question never passed? If contracts were simply canceled, the cost to lobbying companies in that case would only be their cost of capital.
Some have suggested that they just don't know about prediction markets. Beyond that, I suppose they expect that, having made a prediction markets and gotten the expected results, their persuasive position is no better off. Since most of the world doesn't know about or trust prediction markets, they'd have to make essentially the same persuasive effort that they otherwise would. It's not clear that having a prediction market in their corner (so to speak) is more of an asset PR-wise than a liability. Indeed, the PAM affair suggests that it's a liability PR-wise.
Predictions markets still seem to be rarely used for making important decisions - and there is no reason to manipulate something which is not significant.
Yes you have speculators estimate the consequences of a bill conditional on it being passed; there is always a chance that the current bill will be passed. You can also estimate whatever bill is passed that doesn't vary from this one by too much. After the fact accounting should be a lot more reliable than before the fact CBO budget estimation. You look at the difference between estimates with a bill to estimates without a bill; factors that distort both these estimates equally don't matter. More here.
the cost to create prediction markets giving quality unbiased estimates of actual bill budgets would be small.
The cost of setting up a market is not the issue. How would you actually determine liquidating payouts? Prediction markets in sports and elections are easy: Security X pays of $1 if team or candidate X wins, 0 otherwise. It is not much harder with an index like the Dow Jones Industrial Average, where a security can pay off more money the higher the index is.
I don't see any workable way to do this for the budget estimate of legislation that hasn't passed yet. If the legislation doesn't pass, do you simply return everyone's money?
If the legislation does pass, it is certain to differ from whatever is described at the time people trade. The goal of the CBO is not to predict the cost of 'whatever bill that might get passed' but to predict the cost of a bill being considered, so that legislators can decide how to modify it to achieve budgetary goals. That means that you want people trading on estimates of a bill that will never pass (in its exact form), so how are how are you to calculate a liquidating payout?
Even if the legislation passes exactly as specified in the market, how do you calculate the payout? Do you wait 10 years? In that case, market prices are so distorted by risk considerations that it isn't a great prediction. Whose accounting do you rely on? Accounting for the cost of major legislation is a noisy process at best, and heavily politicized by both non-neutral allocations and subsequent legislative fiat.
If you have written on these issues, I'd love to see a link.
I don't know where you got the idea that insurance companies have such a massive fraction of available money for speculation. "Millions" is a tiny tiny faction.
I don't. My comment was actually intended more generally about prediction markets, not just their use on this problem. About gov't interventions in the economy, I don't think the particular numbers are important, even if the plan was revenue neutral or better, it would still be evil. To your more general point below, I am not worried about manipulation in prediction markets, simply the normal irrationality that would indoubtably be present. My comment was about the difficulty of making money off of corrections in the prediction market (that is, as bad as the stock market).
Good point. I'm only about 66% sure that I believe my own rhetoric at any given time.
Opinion polls show that most people are pretty skeptical of the rhetoric on both sides.
Here's an idea: everybody automatically gets a little bit of a tax break, say 1%, just by taking the time to register their opinions on a range of policy issues. But people can also opt to commit some money where their mouth is: they can lose that break or win an even bigger one (say, up to 5%) depending on whether they guess right about policy outcomes.
Would it be gambling? Well, so are state lotteries, which are basically just a tax on statistical stupidity and superstition (and those lotteries effectively levy that tax on those who can least afford it -- how regressive can you get?). Why not instead tax general policy stupidity among the electorate, while rewarding general policy wisdom? And why not appeal to those who are most willing to take deferred and indirect winnings, as opposed to those who buy lottery tickets, who presumably want to find out next week whether they won and to cash out immediately if they did?
"... the chances of the Democrats losing the house in 2010 ... about 66%"
The chances that McArdle would actually put some money down on that proposition? I'd say: approximately 0%.
InTrade currently puts the Dems keeping the House in 2010 at 62%. Trend seems upward at this point. I guess Megan's not a big fan of prediction markets.
I love the idea, but I'm afraid the major obstacle (as other commenters have hinted) is that it would be very hard to create prediction markets large enough that they couldn't be manipulated by those who had a high stake in the outcome of legislation. For example, Democrats would have you believe that insurance companies stand to lose billions in profits if healthcare legislation passes - no way would the industry would hesitate to spend millions dollars influencing the budget forecast of said legislation, were it in their power.
Not only would potential arbitrageurs be seriously outgunned by insurance companies (as Curt says above), but the long time horizon (when would the contract be closed out - 2019?) makes the arbitrage risky in other ways reduces the attractiveness of the return on an annualized basis.
Enabling deep-pocketed stakeholders to influence the budget forecasts for major legislation has the potential to be downright disastrous.
I have published many times on why manipulation seems not to be a big problem with prediction markets. Why not read what I've read before concluding I'm all wrong?
I don't see why prediction markets would work any better than stock markets, and as we all know "the market can stay irrational longer than you can stay solvent".
That requires rational expectations to work, and we know they don't from recent experiences like the housing bubble. It's hard to think of a situation much worse for confidently correcting an "error" as predictions are frequently about never-before-encountered situations about which anyone necessarily has profound uncertainty. In any case, who could muster the tens or hundreds of millions the insurance companies would be willing to spend to distort the market?
But aren’t there substantial organized political groups dedicated to uncovering and promoting the best policies, no matter whose ox they gore? Apparently not.
Just thought of these people (admittedly in the UK):
http://www.instituteforgove...
I gather they have been working with the Conservatives and their Chairman was a Labour minister, which in the UK would suggest they are about as non-partisan as you could reasonably expect.
Well, companies certainly spend millions on lobbying, and influencing budgetary prediction markets could have a pretty high ROI if the prediction markets are small.
I don't know how big you think these prediction markets will be... my single data point is that a few hundred thousand dollars was enough to boost McCain's chance of winning the presidency by 10% on Intrade (see link), which makes me think they'd be pretty small.
http://bit.ly/RLtAe
So I'd think of it as lobbying spend, not speculation.
Brings up another interesting question - what would happen if the legislation in question never passed? If contracts were simply canceled, the cost to lobbying companies in that case would only be their cost of capital.
Some have suggested that they just don't know about prediction markets. Beyond that, I suppose they expect that, having made a prediction markets and gotten the expected results, their persuasive position is no better off. Since most of the world doesn't know about or trust prediction markets, they'd have to make essentially the same persuasive effort that they otherwise would. It's not clear that having a prediction market in their corner (so to speak) is more of an asset PR-wise than a liability. Indeed, the PAM affair suggests that it's a liability PR-wise.
Predictions markets still seem to be rarely used for making important decisions - and there is no reason to manipulate something which is not significant.
Yes you have speculators estimate the consequences of a bill conditional on it being passed; there is always a chance that the current bill will be passed. You can also estimate whatever bill is passed that doesn't vary from this one by too much. After the fact accounting should be a lot more reliable than before the fact CBO budget estimation. You look at the difference between estimates with a bill to estimates without a bill; factors that distort both these estimates equally don't matter. More here.
the cost to create prediction markets giving quality unbiased estimates of actual bill budgets would be small.
The cost of setting up a market is not the issue. How would you actually determine liquidating payouts? Prediction markets in sports and elections are easy: Security X pays of $1 if team or candidate X wins, 0 otherwise. It is not much harder with an index like the Dow Jones Industrial Average, where a security can pay off more money the higher the index is.
I don't see any workable way to do this for the budget estimate of legislation that hasn't passed yet. If the legislation doesn't pass, do you simply return everyone's money?
If the legislation does pass, it is certain to differ from whatever is described at the time people trade. The goal of the CBO is not to predict the cost of 'whatever bill that might get passed' but to predict the cost of a bill being considered, so that legislators can decide how to modify it to achieve budgetary goals. That means that you want people trading on estimates of a bill that will never pass (in its exact form), so how are how are you to calculate a liquidating payout?
Even if the legislation passes exactly as specified in the market, how do you calculate the payout? Do you wait 10 years? In that case, market prices are so distorted by risk considerations that it isn't a great prediction. Whose accounting do you rely on? Accounting for the cost of major legislation is a noisy process at best, and heavily politicized by both non-neutral allocations and subsequent legislative fiat.
If you have written on these issues, I'd love to see a link.
I don't know where you got the idea that insurance companies have such a massive fraction of available money for speculation. "Millions" is a tiny tiny faction.
Any theories about why prediction markets aren't a human basic the way sports betting is?
I don't. My comment was actually intended more generally about prediction markets, not just their use on this problem. About gov't interventions in the economy, I don't think the particular numbers are important, even if the plan was revenue neutral or better, it would still be evil. To your more general point below, I am not worried about manipulation in prediction markets, simply the normal irrationality that would indoubtably be present. My comment was about the difficulty of making money off of corrections in the prediction market (that is, as bad as the stock market).
Good point. I'm only about 66% sure that I believe my own rhetoric at any given time.
Opinion polls show that most people are pretty skeptical of the rhetoric on both sides.
Here's an idea: everybody automatically gets a little bit of a tax break, say 1%, just by taking the time to register their opinions on a range of policy issues. But people can also opt to commit some money where their mouth is: they can lose that break or win an even bigger one (say, up to 5%) depending on whether they guess right about policy outcomes.
Would it be gambling? Well, so are state lotteries, which are basically just a tax on statistical stupidity and superstition (and those lotteries effectively levy that tax on those who can least afford it -- how regressive can you get?). Why not instead tax general policy stupidity among the electorate, while rewarding general policy wisdom? And why not appeal to those who are most willing to take deferred and indirect winnings, as opposed to those who buy lottery tickets, who presumably want to find out next week whether they won and to cash out immediately if they did?
"... the chances of the Democrats losing the house in 2010 ... about 66%"
The chances that McArdle would actually put some money down on that proposition? I'd say: approximately 0%.
InTrade currently puts the Dems keeping the House in 2010 at 62%. Trend seems upward at this point. I guess Megan's not a big fan of prediction markets.
I love the idea, but I'm afraid the major obstacle (as other commenters have hinted) is that it would be very hard to create prediction markets large enough that they couldn't be manipulated by those who had a high stake in the outcome of legislation. For example, Democrats would have you believe that insurance companies stand to lose billions in profits if healthcare legislation passes - no way would the industry would hesitate to spend millions dollars influencing the budget forecast of said legislation, were it in their power.
Not only would potential arbitrageurs be seriously outgunned by insurance companies (as Curt says above), but the long time horizon (when would the contract be closed out - 2019?) makes the arbitrage risky in other ways reduces the attractiveness of the return on an annualized basis.
Enabling deep-pocketed stakeholders to influence the budget forecasts for major legislation has the potential to be downright disastrous.
What leads you to think the CBO could estimate corporate returns better than the stock market does?
I have published many times on why manipulation seems not to be a big problem with prediction markets. Why not read what I've read before concluding I'm all wrong?
I don't see why prediction markets would work any better than stock markets, and as we all know "the market can stay irrational longer than you can stay solvent".
That requires rational expectations to work, and we know they don't from recent experiences like the housing bubble. It's hard to think of a situation much worse for confidently correcting an "error" as predictions are frequently about never-before-encountered situations about which anyone necessarily has profound uncertainty. In any case, who could muster the tens or hundreds of millions the insurance companies would be willing to spend to distort the market?