A firm that uses prediction markets to get info from its employees to forecast sales or deadline success is in essence saying: Don’t just give us cheap talk for a flat fee; signal your confidence by risking a loss if you are wrong, in return for a gain if you are right.
I work for a company similiar to the one described by Salary Bloke. We spend a lot of time on risk identification and management, we even have "experts" in risk, but I could not convince them to try prediction markets, despite many attempts. It's depressing, but you have to remember that a large public company doesn't adopt new processes because they will improve its business, usually they adopt new processes because they are advised to adopt them by opinion formers (eg management consultants, stuff printed in influential newspapers and magazines etc), which is a subtle but crucial distinction. So I would suggest Robin would be more effective if he persuades Mckinsey and co rather than approaching companies directly.
Bob, the reason we'd want to track the time employees put into a prediction market is that people tend to care about their marginal income. If employees had a chance to make hundreds of dollars a month on a football pool but they thought their regular work wouldn't get reviewed much, would they spend 40 hours or 80 hours a month on the football pool instead of their regular work? Sure.
An employee on straight salary might put most of his effort into getting his job done. An employee on salary plus gambling winnings might skimp his effort on everything that doesn't show so he can devote more time to analysing the rules of the gambling game. Hidden costs are a problem.
But then, employee evaluation is always a problem. A junior programmer told me about his experience -- he was designing an install program, and he spent tens of hours working around a bug in the commercial install-program-creator software. A second bug later caused the company a lot of tech-support problems. But what management noticed was that he'd put in splash screens that showed several of the company's logos and photos of the company headquarters. Five minutes of mindless work paid off far more than the hard stuff, in the short run.
So anyway, companies might want better estimates. And they might be willing to pay more for them. But it's hard when they can't tell how much the estimates cost and can't tell how disruptive they are to the company's regular business.
Robin, I believe you. Perhaps I am simply working for the wrong company; however I don't think employer is untypical of large corporations and I am intrigued to think about what makes *some* companies different.
I would imagine that companies that would set up internal prediction markets might be risk-taking companies, familiar with developing models - including models with lots of math - and with making hard financial decisions based on them, might employ some some young, very bright and very clever people both at the innovative sharp edge, and in positions of management, and might make the bulk of their money from new ideas.
But oddly that profile describes my company quite well.... but our *culture* is million miles short of an internal prediction market... so why is that? and what makes other companies more receptive?
Robin, I think this is a cute idea, but it's hard to think of a company that would take it up. I work for a company that - on the face of it - would seem ideal: large, global, corporate(you would have heard of it) active in financial markets, with plenty of internal measures that we forecast.....
....but let employees participate in a prediction market? We don't allow our employees blogs, wikis, or instant messaging...even on the intranet.. I think the idea of - let's face it slightly anarchic and unpredictable thing like a prediction market is way, way off.
You shouldn't understimate the internal consertiveness of even the most externally innovative companies.
I think you'd be better off asking the manager to pick some parameters that would improve the firm's bottom life if they were more accurate, then negotiate a fee based on a percentage of the improvement. No improvement, no fee.
Tracking staff time to set your fee is wading into a quagmire. Sure, you could measure how long staff spend logged into your prediction markets, but does that really measure how much time they are devoting to improving the accuracy of their estimates? They might be working longer hours, they might be shifting effort from other tasks, they might be thinking extra hard on their smoke breaks. Who knows?
Most managers don't (or shouldn't!) micro-manage how staff spend their time arriving at estimates. You pay them a salary; you expect accuracy. If they can be more accurate using prediction markets, great! How they spend their time is immaterial.
Drinking Our Own Kool-Aid. I think that for me to take a a mutual fund seriously that would have to take some percent of the gain above the S&P index and fully compensate me for the difference if they under perform the index.
Hamish, there are a great many things of which I have failed to convince my colleagues. And, alas, they almost never give me reasons for their rejections.
What about *really* drinking your own kool aid? Have you convinced your colleagues within the economics faculty to adopt prediction markets? What about within the university? If not why not? Wouldn't it be easier to start there? :)
I Suspect any company simple relyin on guess work for its future. This can lead to some dangerous sitautiuon in future
my company are having a downer on consultants, as well!
I work for a company similiar to the one described by Salary Bloke. We spend a lot of time on risk identification and management, we even have "experts" in risk, but I could not convince them to try prediction markets, despite many attempts. It's depressing, but you have to remember that a large public company doesn't adopt new processes because they will improve its business, usually they adopt new processes because they are advised to adopt them by opinion formers (eg management consultants, stuff printed in influential newspapers and magazines etc), which is a subtle but crucial distinction. So I would suggest Robin would be more effective if he persuades Mckinsey and co rather than approaching companies directly.
Bob, the reason we'd want to track the time employees put into a prediction market is that people tend to care about their marginal income. If employees had a chance to make hundreds of dollars a month on a football pool but they thought their regular work wouldn't get reviewed much, would they spend 40 hours or 80 hours a month on the football pool instead of their regular work? Sure.
An employee on straight salary might put most of his effort into getting his job done. An employee on salary plus gambling winnings might skimp his effort on everything that doesn't show so he can devote more time to analysing the rules of the gambling game. Hidden costs are a problem.
But then, employee evaluation is always a problem. A junior programmer told me about his experience -- he was designing an install program, and he spent tens of hours working around a bug in the commercial install-program-creator software. A second bug later caused the company a lot of tech-support problems. But what management noticed was that he'd put in splash screens that showed several of the company's logos and photos of the company headquarters. Five minutes of mindless work paid off far more than the hard stuff, in the short run.
So anyway, companies might want better estimates. And they might be willing to pay more for them. But it's hard when they can't tell how much the estimates cost and can't tell how disruptive they are to the company's regular business.
Robin, I believe you. Perhaps I am simply working for the wrong company; however I don't think employer is untypical of large corporations and I am intrigued to think about what makes *some* companies different.
I would imagine that companies that would set up internal prediction markets might be risk-taking companies, familiar with developing models - including models with lots of math - and with making hard financial decisions based on them, might employ some some young, very bright and very clever people both at the innovative sharp edge, and in positions of management, and might make the bulk of their money from new ideas.
But oddly that profile describes my company quite well.... but our *culture* is million miles short of an internal prediction market... so why is that? and what makes other companies more receptive?
Salary, dozens of companies already are taking it up; it is a very large world out there. Come visit sometime. :)
Robin, I think this is a cute idea, but it's hard to think of a company that would take it up. I work for a company that - on the face of it - would seem ideal: large, global, corporate(you would have heard of it) active in financial markets, with plenty of internal measures that we forecast.....
....but let employees participate in a prediction market? We don't allow our employees blogs, wikis, or instant messaging...even on the intranet.. I think the idea of - let's face it slightly anarchic and unpredictable thing like a prediction market is way, way off.
You shouldn't understimate the internal consertiveness of even the most externally innovative companies.
I think you'd be better off asking the manager to pick some parameters that would improve the firm's bottom life if they were more accurate, then negotiate a fee based on a percentage of the improvement. No improvement, no fee.
Tracking staff time to set your fee is wading into a quagmire. Sure, you could measure how long staff spend logged into your prediction markets, but does that really measure how much time they are devoting to improving the accuracy of their estimates? They might be working longer hours, they might be shifting effort from other tasks, they might be thinking extra hard on their smoke breaks. Who knows?
Most managers don't (or shouldn't!) micro-manage how staff spend their time arriving at estimates. You pay them a salary; you expect accuracy. If they can be more accurate using prediction markets, great! How they spend their time is immaterial.
Drinking Our Own Kool-Aid. I think that for me to take a a mutual fund seriously that would have to take some percent of the gain above the S&P index and fully compensate me for the difference if they under perform the index.
Hamish, there are a great many things of which I have failed to convince my colleagues. And, alas, they almost never give me reasons for their rejections.
What about *really* drinking your own kool aid? Have you convinced your colleagues within the economics faculty to adopt prediction markets? What about within the university? If not why not? Wouldn't it be easier to start there? :)