Friday, 3 September 2010

Match-fixing and the Market for Lemons

As you most probably know, this week has been a bad one for cricket, what with the spot-fixing controversy and all (see Cricinfo's full coverage here if you have been blissfully aware so far). The news was especially sad because one of the players accused of spot-fixing is Mohammad Amir, easily my (and everybody else's) pick for the emerging player this year. Anyhow, to deal with the shock and the pain, I figured I'd try and come up with a flippant pseudo-economics-based post on what can be done about match-fixing etc. While I was still formulating my ideas, Cricinfo's excellent Surfer blog pointed me to this article by Malcolm Knox on Back Page Lead, which sets up a pretty nice segue into some of what I had to say:
If bookmakers are stupid enough to take spot bets that are fixed, and players are corruptible, then the result will be that the bookmakers will be stung often enough to refuse taking such bets. If the Pakistan players are corrupt all or most of the time, the market would have become a sham and would have ceased to exist. The fact that the market does exist tells us one thing: most of the time, the players are trying their hardest. When they are not, they are choosing their moments selectively. Otherwise there would be no bookies left to fool.
Now before I get into what I had to say proper, there's something that Malcolm doesn't get exactly right (which was also pointed out by one of the commenters on his blog): bookmakers don't usually get on the other side of a bet. They're supposed to set up a market by setting the odds of a particular result and finding two entities who are willing to take either side of a bet, with the bookie usually earning a decent fee from both, and the winner of the bet taking the money. (Note:In a way this is not unlike an investment bank helping to set up a securitization deal by putting together say a bunch of mortgage-backed securities from one set of lenders and getting a rating agency to assign a rating, like a set of odds, that define how risky the resulting CDO's tranches are, and then selling said CDO tranches to some other chump and taking a hefty fee in the process, thus getting a fixed payoff while leaving the buyer to face any risks involved in the deal. Of course, the last few years saw the i-bankers believe their own spiel and holding on to said CDOs, eventually bankrupting their parent companies and more. Sadly, bookmakers seem to be more aware of the risks involved in their bets than i-bankers.) (Note: the previous note was drafted just to show that I've recently read Michael Lewis' 'The Big Short' and now feel like dissing a few i-bankers).
So anyhow, what Knox should have been worrying about is not the bookmakers but the punters who are willing to take the other side of the bet for a spread-bet, even though there was the possibility of fixing. My own guess is that most punters don't take up just one side of one bet - they too would make a string of bets to hedge against losses.

Now one of the factors that would actually encourage punters to think that the bets are fair and they can easily hedge, would be the belief that most players are honest, and the bad ones are weeded out. A life ban for a player who cheated, in this case, could plausibly encourage more betting! How? Here's my pseudo-econ explanation:
Way back in 1970, George Akerlof came up with a seminal paper on asymmetrical information called 'The Market for Lemons' (wiki link). The idea there was that in a market where the sellers of a product -specifically, used cars - knew more about the condition of their cars than buyers. Since buyers were unsure of the quality, the average price they would offer would be lower than the price of a well-preserved used car would be. Now this would in turn mean that the sellers who actually have good used cars would not want to sell at a lower price, which leaves only the sellers of badly-maintained cars willing to sell. This in turn would reinforce the buyers' belief that all used cars are bad (lemons), and would drive their price lower, and so on and so forth in a vicious circle. Now in cricket, we have the opposite situation. Life bans for cheaters would signal that those who are left are quite probably honest which means that the events that are being bet upon aren't fixed and are instead decided by a combination of honest effort and chance. Here's one point where perhaps i-bankers win back a point: if the punters had Phd-toting quants assisting them, they might have more rationally looked at the events of the past and calculated a probability of any given player being dishonest and factored that in when making a bet. However, since most punters are also (probably) die-hard fans, they would confidently (and somewhat irrationally) assume that all players not yet caught are completely honest.

Which brings me to a discussion of the players themselves, and what might be a way of setting punishments for cheats. Knox rightly points out that the bent players don't always cheat, but rather, would pick the moment when they can let their standards slip. We could probably make a conjecture of what the decision-making process in this case would be. Since players stand to gain both money and reputation (which can help if they want to cheat later), they (at least, not the smarter ones) would not cheat in the bigger marquee events - the World Cup, say, or the big Test series. The best occasions are in inconsequential tournaments and matches (think Sharjah and other cricketing backwaters) where the spotlight isn't very much on the players and fans may be more forgiving of a 'loss of focus'. Admittedly, those who got caught were those who cheated at more marquee events (Cronje's Test, this England-Pakistan series), which only goes to prove that they were either a little too greedy or too naive.
So what does this mean? Well maybe instead of an all-or-nothing approach - a life ban or a clean chit, how about handing out graded punishments? Perhaps we could ban players from certain forms of the game for a given period - no World Cups and Tests, but they could be allowed to play first-class cricket and T20 and tournaments-sponsored-by-cell-phone-companies, perhaps. That way, the players with tarnished reputations know that if they want to keep going at quite possibly the only job they know, they will have to play twice as hard and honestly. At the same time, the average punter will know that there are more players who have been dishonest in the past involved, they might actually think twice about taking on a bet that sounds too good to be true. Which might in turn bring down the amount of betting.

On then to another important aspect of cricket: the fans. Would they want to watch games that might involve players with dodgy reputations? Probably not. Maybe (hopefully), we would we see a flight to quality, with people opting instead to see stuff like Test match cricket for the spectacle. Rather than fill up the calendar with hundreds of meaningless ODIs, we could have more series that are eagerly awaited (and monitored) and which actually linger on in our collective memories. More discerning viewers might also mean that broadcasters might have to improve the quality of their programming - more Michael Holding, less Navjot Sidhu, for instance.

So all in all, win-win then. We could make something really positive out of all this, if we just give it some thought.

Yeah right, who am I kidding. This sucks.

UPDATE: K very astutely pointed out in the comments that I hadn't explained the link between match-fixing and the market for lemons too well. I've elaborated a little further in the comments. Going by what I've said, it struck me that the ICC and various administrators would be analogous to used-car salesmen. Which seems about right.


  1. "Life bans for cheaters would signal that those who are left are quite probably honest which means that the events that are being bet upon aren't fixed and are instead decided by a combination of honest effort and chance."

    how does this link to the lemons problem?

    actually, the point of the lemons article is somewhat missed by a lot of people. This is that what akerlof was really putting forth the following argument:

    "by my (rather clever) analysis, a second hand market shouldn't exist.

    yet it does. why/how?"

    the suggestion from him was to look into the institution underlying the second hand market for cars, to see how they solved this information problem.

    of course, economists, with their love for cleverness over wisdom, marveled at the method of proof. Not the more interesting, and deeper, suggestion of figuring out how the real world works..

    i'm not making this up; a co-author of akerlof took an intro micro class for us and told us the above.

  2. Point noted. Think of it in terms of this - the reason why buyers in the first place are unsure of the appropriate price is because they have seen a few instances of lemons on the market but don't know how many such lemons there are. But that gets resolved somewhat by the fact that besides a lot of individual sellers, there are used car dealers that aggregate the used cars and provide some sort of guarantee on the quality of the cars they sell - the buyers would reason that for dealers to survive in the market, they need to weed out at least the worst lemons. That way they get more customers and those customers are willing to pay a higher price given higher confidence in the quality of the vehicle. Of course this confidence may be misplaced since now the dealer may be able to use his brand to slip in a few lemons. This may be some of what Akerlof may be getting at.
    Applying the same logic to cricket, weeding out some cheats would make fans more willing to trust the remaining players.

    Good thing I didn't explain this in the post- at least this way I get a comment :-)

  3. Incidentally, if I remember the Akerlof paper correctly, he didn't set out too many assumptions on whether all buyers were homogenous or what the exact decision-making method they used to arrive at their price was, which allows for people to understand the broad logic more easily. If he was writing in the late 80's perhaps he would have had to be more explicitly mathematical, which in turn would have required simplifying assumptions and a further departure from reality. And then maybe economists would have missed the wisdom even more completely and ended up fighting over whether the model specifications are true and if under the EMH there could ever be such uncertainty since a market-determined price would contain all information needed

  4. well actually the first draft of the paper had some pretty heavy math in it; akerlof's advisor urged him to reduce this because he feared most people would not understand it or misunderstand it (I'll come back to this).

    The result? A first attempt at submission was rejected with the main problem being that the idea was "obvious".

    The problem with having a lot of mathematics is that most people will not, or cannot, take the time to really understand the idea behind it. The problem with having too little is that you risk being dismissed as being "obvious".

    Incidentally, it was while Akerlof was in ISI, Delhi that he got the main insight into the issue: (from the Nobel web site)

    "Here Akerlof introduces the first formal analysis of markets with the informational problem known as adverse selection. He analyses a market for a good where the seller has more information than the buyer regarding the quality of the product. This is exemplified by the market for used cars; "a lemon" – a colloquialism for a defective old car – is now a well-known metaphor in economists' theoretical vocabulary. Akerlof shows that hypothetically, the information problem can either cause an entire market to collapse or contract it into an adverse selection of low-quality products.

    Akerlof also pointed to the prevalence and importance of similar information asymmetries, especially in developing economies. One of his illustrative examples of adverse selection is drawn from credit markets in India in the 1960s, where local lenders charged interest rates that were twice as high as the rates in large cities. However, a middleman who borrows money in town and then lends it in the countryside, but does not know the borrowers' creditworthiness, risks attracting borrowers with poor repayment prospects, thereby becoming liable to heavy losses. Other examples in Akerlof's article include difficulties for the elderly to acquire individual health insurance and discrimination of minorities on the labor market.

    A key insight in his "lemons paper" is that economic agents may have strong incentives to offset the adverse effects of information problems on market efficiency. Akerlof argues that many market institutions may be regarded as emerging from attempts to resolve problems due to asymmetric information. One such example is guarantees from car dealers; others include brands, chain stores, franchising and different types of contracts.