D (3:05)

Regulated that.

C (3:08)

Led to a variety of disasters.

D (3:11)

And these were what was happening were.

C (3:12)

People were responding to strong incentives but Strong, badly designed incentives.

D (3:19)

So what I want to talk to.

C (3:21)

You about tonight is work that a number of people have done. Some of it is actually very well known, part of the standard canon, but I'll put a little different twist on it. But other parts of it are quite recent and have not appeared anywhere in print. But all of them are discussing are situations where muted incentives are the best incentives. And muted might mean absolutely flat. And then if you want, you can.

D (3:57)

Ask me at the end what I would have done about it or what.

C (3:59)

I would do about it now, and so on. So you've seen this already. So I'm an economist. I believe incentives work and they affect actions, they affect decisions. The basic economic model is fundamentally right, and in particular, stronger incentives have bigger effects. But that's true even when the strong incentives are ones that generate really bad behavior. So think about the subprime mortgages that were the start of all this mess. These were mortgages that were written for people with no income and no assets, people who had very low incomes, little or no savings. They borrowed 90% and more of the value of the house without the cash flow to support it. And the whole idea was that this would be fine if housing prices were to rise exponentially for the rest of human existence. Otherwise there was sure to become a day of reckoning. But the people in Southern California and Nevada and such places who were writing these mortgages were paid 8% of the face value of the mortgage for every mortgage they wrote. They weren't paid on its profitability. And apparently the people who were supposed to be monitoring them and watching over them were unconcerned about the quality of the mortgages, largely because of the possibility of selling the model. So they would write a bad mortgage and sell it on to Freddie Mac or Fannie Mae, or eventually to the investment banks who created these collateralized debt obligations. And so they weren't terribly worried. They should have had some reputational concern about whether they'd get a reputation for writing bad mortgages and they wouldn't be able to sell the future ones for as much. But that was clearly second and third order to them then the investment bank. So they created all these mortgages that were bound to fail at some point. There wouldn't be higher housing prices, they wouldn't have gotten equity into the house through the housing market, and they wouldn't be able to pay for it.

D (6:31)

Then the investment banks thought up collateralized.

C (6:34)

Debt obligations and what this involved. I hope I'm not telling you things all of you know intimately, but I Prayed I may be. What was involved in this was taking packages of mortgages from all over the US and putting them together and then creating new debt. So that gives you a flow of cash. And the expectation was that these cash flows from the different states would be pretty uncorrelated, be independent. Because there had never been huge nationwide movements in housing prices. Sometimes Texas real estate was in bad shape when the oil price was low, sometimes California, California was high. But there was a fair amount of independence across the states. Well, so the idea was to put them together and create a bunch of DEB instruments. And the first ones, the top of the line ones, would be collateralized by the payments on these mortgages. And the first tranche, first level ones, would get paid as long as there was any money. It would go to them until they'd been fully repaid. And then once they'd been repaid, the next level would get it and so on and so on. So the top level ones looked pretty safe.

D (7:57)

And the banks loved them.

C (7:59)

And the banks that created them loved them because they made lots of money and fees on selling them. And the bankers who were buying them loved them because they seemed to be paying a lot more interest than they should. And so they had very strong incentives. They're paid on profits on their trades and on the transactions they make. They loved buying these things. They had strong incentives to buy them. And they didn't have proper incentives to pay attention to how risky they were because the guys were buying these things were not being paid on the profits of the, the bank would earn over the 30 years of these mortgages, but rather on the immediate spreads. And so they had strong incentives. Somebody should have had incentives to inspect and see how risky these things really were. Because the second and third and fourth tranches were certainly very risky if the housing price ever fell. But nobody much did look up, in part because the rating agencies, Moody's and S and P and these people certified them all as aaa. Not just the top tranche, but many layers down into the system. And they did that because they were getting strong incentives to do so, because the banks who were issuing these things desperately wanted them to be rated highly. And S and P and Moody's were subject to moral hazard. They acted on their strong incentives in a totally inappropriate way. The other big.

D (9:40)

The big issue for at least Americans.

C (9:43)

Is that we have now sunk one out of every ten dollars of the personal and corporate income tax that the US government will collect this year has been given to AIG. That's a huge number. They talk about 170 billion. But I don't know many people who have a sense for how big a billion is. But what 170 billion is is one out of every ten dollars collected in the income tax system. So I personally a humble academic, I'm putting more than $10,000 and perhaps depending on how good a job my accountant's doing at home right now, maybe much more than that into aig. And what happened? AIG was the biggest insurance company in the world. It insures everything if houses, health care, real estate, all kinds of assets. But it had a little group of hotshot rocket scientist finance people here, largely in London, who were called the AIG Financial Products Group. And they came up with credit default swaps. What a credit default swap is, is an insurance contract that says if this particular debt is reneged on, isn't paid, we'll insure you against that event. And they wrote a little bit of. They wrote some insurance and charged fees for it. And they wrote, in fact just in CD CDS, they wrote $440 billion worth of them on a company that had a market value of 200 billion. So they wrote all this insurance. Now, insurance industry around the world is regulated. But thanks to particularly Hank Paulson and various other investment bankers, there was a push when these sorts of derivatives started being created that they not be regulated. And Hank Paulson, who later became U.S. secretary of the treasury and led the first attempts to clean up this mess, actually prevented there being any regulation. So there were floating around these trillions of dollars of insurance contracts with nothing to back them. They had gotten high ratings. They were viewed as safe because AIG was viewed as safe because there was this big insurance company with lots of steady cash flow. But nobody seemed to know that they had written this much of this stuff. And then when the CEOs started failing, Goldman and those people went to AIG and said we want money. We didn't get paid by the issuer of the collateralized den obligation to we're holding, we want money from you. Or we we didn't get paid on the mortgages we're holding, we want. That went into our CDOs. We want money from you. So at this point, of the 170 billion that's gone into AIG, most of it has been passed on to the big international investment banks, to Goldman Sachs, to UBS and so on. So all of this was because people had strong bad incentives. Badly designed, strong incentives are a recipe for disaster. So fine, the answer is design good incentives. Well, that's hard. That's sometimes impossibly hard. And in particular it can be impossibly hard to design strong good incentives. And the answer is then better to have weak incentives than strong incentives that generate bad behavior. So this is particularly true when we're talking about the incentives that are created within organizations. You notice all the things I was talking about were incentives for employees and managers. They weren't incentives that were in the market. What they were were largely an attempt to recreate market incentives within firms. And there's been a lot of talk over the last 20 years about bringing the firm inside the market and providing strong market like incentives. And many observers have noted and deplored that usually the incentives within organizations are weak one ones. But we've already seen those are maybe the best ones. And in fact, my colleague Ben Tolstrom has argued in an argument that's been made more precise by a young guy at Harvard named Derek van den Steen, the firm is really usefully viewed as an institution to permit giving weak incentives. Markets don't know how to give weak incentives incentives. Markets automatically give strong incentives. But if strong incentives are bad incentives.

D (15:11)

Then what you need is something an.

C (15:13)

Institution like the firm. Now, excuse me for a moment while I just do an obeisance to giving the theory. What I'm going to be talking about, I'm going to be talking about five instances where weak incentives or the right incentives, all of them are going to be in the context of agency theory, where most of incentive theory of economics is couched in these terms. So there's an agent, he, by convention the agent is he. And the principal on whose behalf he's supposed to take actions or make decisions is referred to as she. And there's some divergence in of their interests. And she cannot perfectly observe his actions. She can maybe observe some noisy signal of them, or maybe she can observe some elements of his performance, but she can't really understand what he's done for her. Or alternatively, she doesn't have the information to evaluate whether he's made the right decision. If she could, she probably would have made the decision herself. So she does have some sort of imperfect measures. And her job is to motivate him to act as best she can in her interest, whereas recognizing that whatever incentives she puts in place, he's going to do what's best for himself. So this is economics. There's little or no altruism in here. So the solution for her problem is to create incentives, perhaps using the imperfect measures of performance or of effort or quality of decision. Recognizing that he's going to do what suits him. And this may. And traditionally in most of the economics literature it does involve tying rewards to this measured performance. But the question is how much should the rewards vary with performance? Should a little small increase in performance be a big increase in rewards or quite small? So that's the strength of incentives question. And you could think of it being.

D (17:35)

Anywhere between I get paid the same no matter what my purchase perceived or.

C (17:40)

Measured performance, which is a base salary, or I get the full value of.

D (17:45)

Any extra performance that's generated, which is the case of what you get in the market. Basically in the market, if you own.

C (17:53)

Some asset and its value goes up because of something you've done, you get to keep that value. So that's full strong incentives. And that's apart from writing insurance code contracts around it. That's why market incentives are strong.

D (18:09)

I want to accentuate though there are lots of other options besides writing monetary contracts, explicit or implicit.

C (18:18)

So there are five cases I want to point to. How long do I have? About two more minutes.

A (18:25)

No, no, no, you've got plenty of time. Okay, 20 minutes, 25 minutes.

C (18:33)

The first one, well, you can read, read this and I'll go to the next page. So the first one is when measures are poor. The basic insight is that if you have good measures, they can be the basis for strong incentives. If you have poor measures and try and apply strong incentives to them, make strong incentives on them, you're going to get very bad results. Poor measures are of two sorts in the economics literature. The traditional agency theory that Holmstrom and Milgram and before them nearleaves put together. Poor measures are ones that have a lot of random noise in the measurement, a lot of random measurement error. So for example, a middle level manager, you're trying to gauge, well, gauge performance there. Performance meaning contribution to corporate profits, and you have little or no idea and there's no way you could get any idea of what that person's real contribution was. Any estimate you make is going to be hugely noisy, maybe unbiased, but it's very noisy. A more important and more recent modeling that's due largely to George Baker, but also builds on work by another course of work by Holmes, Drew and Milgram. Our poor measures are ones that don't reflect the ancient's contribution because they're biased. And for example, think about a transaction that involves immediate returns and long run returns. And suppose there's immediate returns are here and the long run returns are varied like that. Suppose all you can observe is the immediate returns. Well the agent, if you're going to pay him on immediate returns, he'd rather have this immediate return that's a little higher and this return that's negative than this and that because all you're paying them for is the variation here. So you've got a biased measure, a measure that isn't an accurate reflection of.

D (21:01)

The value being created for the principal. So either type of those poor measures.

C (21:07)

Lead to bad results. Incentives based on biased measures lead the agent to do what you pay him for. If you pay them for short term returns, that's what you're going to get. Independent of what happens in the long run there may be other factors like reputation and whatnot that mitigate that a bit. But basically you get what you pay for. And so he'll expend a lot of time and effort doing stuff that's not at all in the interest of the principle. On the other hand, if you use and so that says if the bias is great, use weak incentives.

D (21:43)

If there's a lot of noise, what.

C (21:46)

Happens is you're putting a lot of randomness into the person's pay. And most of us are not in a position to bear huge randomness in our pay. If we're going to have a lot of variation, we going to have to have a lot of base pay to be able to deal with that potential variation and to deal with those costs of risk barrier. And so if you've got bad measures, you should give weak incentives. And if there's some fixed cost to tie in pay to performance, give no.

D (22:17)

Incentives at all in the monetary side.

C (22:21)

Second point, and this is stuff that again Milgram, Holmstrom and Baker are largely responsible for. Suppose you want, as almost always the case, you want your agent to do more than one thing. Most jobs, there's more than one obligation in the job. For David and me were both expected to do teaching and research. Others of you have responsibilities, say to deliver current sales and to build customer relations, a whole variety of things.

D (23:05)

So in multitasking situations.

C (23:13)

Think about in the extreme, all the ancient cares about is how much time and effort he puts into the principal's interest. And the principal wants him to do two things. Now if that's the case, you have to pay him the same at the margin for performance on both dimensions because otherwise he'll just do the thing that's better paid. You know, if you'll pay me in effect $2 an hour for doing one job and $3 an hour for doing the other, and I don't care what I'm doing I'm going to spend all my time on the $3 an hour job. So that means that you have to give balanced incentives. And that's true even when it's not this extreme case. There's got to be balance in the incentives. Now it's quite possible you'd have very good measures for some things, but poor ones for others. So for example, in the building, in the sale, delivering current sales and building customer relations, or the other thing might be building current sales and bringing back knowledge from customers about their needs, you get pretty good measures on the sales part and on how hard and cleverly the salesperson worked at that job. But measuring how good a job they are doing in building long term relationships or at bringing back information from customers that they may not bring any back because there may be nothing to bring, you're going to have very bad noises measures there. So you can have biased measures that just don't capture what they're doing, or noisy ones like we talked about before. Or even more measures that are manipulable where it's possible for the agent to do something that affects them measure even though it doesn't do anything to affect the contribution that he's making to the principal's interest. So things like monkeying with the accounting numbers, if you're paid on the accounting numbers, you have every incentive to make your accounting decision so you look better even if they don't do anything to advance the shareholders interest. Or a famous example example that my colleague Paul Oyer documented of salespeople who were paid a bonus if they reached a quota and then nothing more after that. So the performance pay looked like they got no performance pay until they hit their quota. Then it jumped up and then it went like that again, flat. So it turns out if you got near the end of the period and you were a long way from making your quota, you'd stop selling and try and save the sales for next period when you might make it. And of course if you'd already made quota, you'd stop selling as well because there's no sense wasting the sales. You can use them next period. And so none of that is in the interest of the firm. The interest of the firm is having smooth sales, but that, that they're manipulable numbers. Well, the problem is if you've got poor measures on some of the tasks, you can't give strong incentives for the well measured ones and weak incentives for the poorly measured ones like you'd like to, because you have to give balanced incentives. So you either have to give strong balanced incentives, strong for both. That's just too costly. You end up giving weak incentives for both. So if you want your sales staff to be building a lot of relationships, bringing back a lot of information, what you have to do is not have them be operating just on commission, put them on salary, make them full time employees, tell them what to do. If on the other hand, all you care about is sales, then hire an outside decision distributor, let them pay them a high commission and send them up to make your sales numbers. So strong incentives are necessitated by multitasking when there's not everything has good measures. Third context is when cooperation is needed. And this is this issue of cooperation is one that, that I started thinking about with Jonathan Day here and Bent Holmstrom 1012 years ago, 14 now. But if you think about why we have organizations, we have organizations to coordinate and motivate people in the presence of spillovers, situations where what I do has an impact on what you do. And we take those. The market doesn't typically work so well when there are such externalities. And so we take those situations out of the market and sometimes we regulate them by the state doing them or the state overseeing them. Sometimes we put them inside organizations. The putting inside organizations doesn't obliterate the need for coordination on these complicated spillovers. And you can't. It's very hard often to measure them because sometimes cooperation will be nothing more than refraining from screwing the other guy over, not behaving too nastily within the organization. And it's really hard to measure how often the guy has not been a bastard. He's leaving on me, so I'm trying to liven it up a bit. He was supposed to leave Ermias. So cooperation is behavior that works to the advantage of others, others in the organization works the overall corporate interest. And it's often now, often this could be modeled as an action that's just hard to measure. And then you can use the multitasking framework and you can talk about initiative, doing your own job and putting cooperation helping other people. But there are other contexts where that modeling isn't quite right. For example, one that really starts with some work of Susan Athey's in mind, but recently has been done by. A couple of young German economists, Friedel, I believe, and Reith. So there's multiple agents. Think of them as division managers. And each of them has to take an action that builds his performance of his division. And he also has to make an investment decision. And the investment decisions may have spillovers under the other division, maybe it affects the reputation of the corporation or has some impact on sales or whatever. The only measure of either agent's performance is something like cash flow in his division. And cash flow in his decision is going to reflect what he's done, the investment decision he's made. And if there are any spillovers, the investment decision the other guy's made. So if there were no spillovers and the action, then what you'd have is a measure just of his cash flow would be a measure of his decision, his action, and could well be the basis for strong incentives. But if you give him strong incentives to pursue his interest, you're giving them, you're dissuading him from paying any attention to the other guy's interests to the extent that they're not perfectly correlated. So if I can make an investment that's good for me and bad for you, I'm just as happy to make that as to make one that's good for me and good for you. And if it's a little bit better for me, I'll take the one that hurts you. So the solution there, if they're important spillovers, is to reduce the weight, put on your own division's performance and make pay more on corporate performance or equivalently on the other guy's performance, to which you'll then be motivated to contribute to make better investment decisions because you're getting paid on both. So it's weaker incentives on your own performance. A former student of mine now at MIT named Gustavo Manzo has studied inducing people to experiment. And it turns out that weak incentives are often the secret. So here's the situation you should think of. There's some task that has to be done and there are at least two ways to do it. And one of them is the way we've been doing it all along. It's tried and true. We know what the chance of success is if we do it the old fashioned way. We know there's a point eight chance that it'll succeed. On the other hand, somebody suggested a new way of doing things. And this new way of doing things we don't know. We don't have experience with it. We don't know what its chances of paying off are. But we guess just up front that rather than point eight, they're 0.7. So if all we were to do is decide which way to do it and that's it, we'd go with the old fashioned way. But suppose we're going to be making, carrying out this task repeatedly. So what we could do is we could try the new way. And let's just say that the mathematics works out that well, it'll be the case that if the new way succeeds, you'll up your probability of its probability of success. And if it fails, you'll lower the probability of success that you leave it to hold. That's standard Bayesian updated. Suppose then that it's going to be if you try it and it succeeds, you then think it'll succeed in the Future with probability 0.9. So what could you do? You could, in a repeated context, it might be worthwhile to try the new way. And if it works, it then looks better than the old way and you continue with the new way. You experiment with it. If the experiment succeeds, you adopt the new way. On the other hand, if the new way fails, you think it's really very unlikely to be better than the old way and you go back to the old way. So that's exactly what an experiment is. So if you're our friend, the principal, you've got first of all, you have to motivate the agent to work in the first place, so you've got to keep him from slacking off. And then you have to decide whether you want them to experiment or stick with the tried and true method. Well, if you want to stick with the tried and true method, it's just exactly what you learned if you ever took agency theory. You just give them strong incentives for success and that's it. On the other hand, if you want to motivate experimentation, particularly if the agent would prefer to do the old way, if that's easier for him, if he doesn't get a thrill over trying something new, then you shouldn't reward him for early successes, or certainly not strongly because he's more likely to succeed. If he's done what you didn't want him to do and did the old fashioned way, you would expect remember you thought that was a 0.8 chance of success. If he did what you wanted him to do, it's only 0.7 chance of success. So if he succeeded in the first round, it suggests that he may not have done what you wanted him to do. So you shouldn't reward him too strongly for early successes. And in fact, you may have to reward him pay them more for an early 50 failure than an early success. If he really wants to do the old fashioned way and you really want him to experiment. So that means in the first period you give weaker incentives than you would have otherwise.

D (35:29)

The second period you give the last Period, you give strong incentives to motivate them overall, but in the first period.

C (35:36)

You give weak incentives. The last case is due to student mind at Harvard Business School named Eric Vandenstein. And Eric has spent the last several years looking at situations where there's open disagreement. And I believe that's the world that we live in. You and I may fundamentally disagree about what's the best way to do something, or whether this person is a good risk, or who the best candidate for a job is, or whether it's going to rain tomorrow, or whether Tottenham will win the next match as it has won the last two. David and I are bound by our love of Hotspurs. Nobody laughed. That's nice. That's because they're doing pretty well now. But after eight games they had two points anyway. So we live in a world of overt disagreement. And though most of economics for some.

D (36:41)

Reason has assumed away that overt disagreement.

C (36:44)

But Eric has explored its implications for management and design organizations. And so suppose that the principal and the agent fundamentally disagree about the right way to do something. And we can't just write a contract that says, I promise to do what you tell me to do. On the other hand, we'll both be able to, because it's going to be very hard to go into court and establish really that I did exactly what you wanted me to do. On the other hand, I can fire you if you don't do it. Now, if she gives the agent strong incentives for good performance, that's going to motivate him to do do what he wants, what he thinks is best, because he's trying to maximize the chance of.

D (37:33)

A good outcome, and therefore he will.

C (37:36)

Choose to do things the way he wants to do them. But that isn't what she wants. So the answer, in fact, is to give him weak incentives. And if you don't have to motivate him, if you just have to keep him from disobeying, you give him no incentives at all. You give him flat wages, and then he's indifferent as to what he does. He says, that stupid boss wants me to do it this way. Well, it's no skin off my nose. She's just hurting herself by insisting on it being done the wrong way. But fine, if you have to give him some incentives to motivate them to do anything not to goof off, then you give them very weak performance incentives. But what you do is you pay him a high wage and threaten to fire him if he misbehaves. So this is the efficiency wage ideas, and then There's a question of if there are assets around, who should own those assets. And economics has talked about asset ownership as a very strong incentive for device. But it turns out that out of this theory comes an argument that the principal should own the assets. They shouldn't be owned by the agent or by some third party, because what will happen if the agent disobeys the principal is called upon to fire him? If she has the assets, she's in a much better position to go out and hire a replacement than if he has the assets or if the bank owns the assets. So that gives her an incentive to follow through with her threat. And equally, if he hasn't got any assets, if he loses his job, he's.

D (39:23)

In a much weaker position.

C (39:25)

So he's more frightened by the threat of firing and that lowers the amount of extra pay you have to give him to keep him in line so you can spin a whole thing of the firm out of this idea. So to wrap up, market incentives automatically are strong. Strong incentives are dangerous.

D (39:51)

Bringing activities inside the firm allows provision of desirably weak incentives. And attempting to recreate market incentives inside.

C (40:00)

Firms is thus often a mistake.

D (40:03)

And it's that attempt to recreate market incentives inside the banking system that I.

C (40:09)

Laid as the cause of much of the crisis that we're suffering, genuinely suffering through now. Thank you for your attention. I'd be happy to take questions. John, thank you.

D (40:24)

We've got plenty of time for questions. So do we use a mic or not? Okay, we've got gentlemen here.

B (40:36)

Okay, thank you. Thank you for the talk.

C (40:38)

Thank you.

B (40:39)

I have a question. I think this slide is good for it. There is something I'm understanding and something I'm not understanding. If I think of an organization as state.

C (40:55)

The state.

B (40:55)

Yeah, like the government, like this society. Would this theory really be effective? Because if you say markets give strong incentives and strong incentives are not good.

C (41:15)

Well, sometimes they're good, sometimes they're good. If you've got good measures, they're good.

B (41:22)

Okay, because I'm just afraid of policy misunderstanding of what markets can do because markets have been functioning very well.

C (41:33)

There are lots of things that markets.

D (41:35)

Do much better than firms and much better than public bureaucracies.

B (41:41)

Yeah, that's the path I don't understand. Which could mislead policymaker. To think that you are advocating for some weak incentives would be like thinking of a centralized. If I'm thinking of our organization to be a society of government, then it would be like you're advocating for a centralized.

C (42:01)

No, no, I don't think so.

B (42:03)

No, I'm not saying. You're saying that. I'm saying that it could be. It could be seen like then. Well, the other part is if it is in the firm, then I don't understand how the workers. The market idea of incentive applies to the workers in the firm.

C (42:22)

Okay, so what you're asking, first of all, what I was trying to do was provide a counterbalance to 20 years or more of a general view that markets are great and firms are bad. And one of the reasons firms are bad is because they give weak incentives. And that's something that's been preached on both sides of the Atlantic for at least its last 20 years. And I've done my fair share of it. I am far from blameless in that position. But it's become clear a lot of these ideas that I was presenting are 15 years old, but they haven't gotten the play in public. What's gotten the play is the strong incentives are good, markets are good, organizations give weak incentives, marketizations are bad. Best thing to do would bring the market inside. So it remains the two. The market does a wonderful job on some things, but there are things where they're good measures. There are things where the people making the decisions bear a significant portion of the consequences of their decisions, where the people who are acting bear significant portion of the costs and benefits of their acts. And lots and lots of things are like that, but lots of things are. And what I was trying to do is highlight situations like that. Now, I don't think this necessarily. I haven't thought through your question about centralization. I haven't thought through it, but I don't think it's automatically the case that as soon as you step to weak incentives, you step to only one person or one small unit that presumably is somehow given good incentives making all the decisions. Far from it. I mean, I think a wonderful example is the Toyota production system. And in Toyota, certainly the flow of work and what they're going to make on any given day, and all that is decided centrally. But the workers on the line have a huge amount of autonomy and responsibility for deciding whether to stop the line if something's going wrong. In a typical US Or European auto plant, the worker stops the line, is in trouble. In a Toyota plant, if there's a problem and you don't stop the line, then you're in trouble. But the Toyota workers don't receive any individual insurance incentives. And indeed, the Toyota workers put huge amounts of time and effort into figuring out better ways to do their Jobs. And it's not just the Japanese, this is true in England and France and America and Canada. And they figure out a better way to do their job. They document it, they take video of it, they post it so the next shift can see how they're doing it. And they're free to do that. They don't call in the industrial engineers. So we've got people who have pretty weak performance incentives, but they're motivated. They're motivated by pride in their job, they're motivated by being trusted. They're motivated by a certain level of monitoring. And so one of my favorite examples is performance pay for teachers. And you know, there's an ongoing debate about that. And usually the notion of performance pay for teachers is we're going to measure the student's performance and pay the teachers on that. And what it invariably results in, at a minimum, is the teachers teaching to the tests. So they test reading and writing and arithmetic, and that's what the teachers teach. And everything else that we might have bought our school system to pay attention to, like ethical behavior and civic society, the interests of civic society and all that stuff gets shortchanged because you're paying strong incentives for one thing and not for the other part of the multitask. And it seems to me it'd be better to run a system where you had principals who were empowered or head teachers in this country. We're empowered to monitor the teachers and observe what they're doing and then recommend rewards on that basis. But you don't have these mechanical, high powered rewards sent in from the outside. Basically the idea should be if you can measure it well, do it in the market. If you can't, develop good measures for it, bring it inside the firm and use other mechanisms. Employee involvement, pride in the job, promise of promotions, subtler, more nuanced rewards to motivate people.

A (47:37)

Somebody at the top there, please.

E (47:40)

First of all, thanks very much for your lecture.

D (47:42)

Thank you.

E (47:43)

I'd like to ask the inevitable financial crisis application question, if you don't mind.

B (47:50)

So you said at the beginning of.

E (47:51)

The lecture, applying the discussion in advance to the financial crisis, that we had agency problems at work here and that these were, these created bad incentives. And I'm very used to bad being phrased in terms of the social optimum in the discussion that's going on in the political sphere these days. But here, did we have principals of the firms in question who essentially made mistakes in writing the contracts, or were there a couple of levels of principals and agents at work here, or principals with perhaps different incentives from the social optimum.

C (48:33)

Yeah, absolutely. The people who were writing those mortgages in Southern California had somewhat divergent interest from the managers of the branches. And they had divergent, somewhat different interests in turn than the credit risk management people. And at Bear Stearns or at Lehman Brothers, John Fuld held on to his Lehman Brothers shares and watched them go down the toilet. So he thought things were going right. So there was a lot of, it's not just simply moral hazard problems. There were mistakes, there were misapprehensions, and there was, I think, moral hazard by the, by the people who were doing risk management that it was much easier to build credit risk models using normal.

D (49:35)

Distributions that you could compute easily than.

C (49:39)

Using fatter tailed, more realistic.

D (49:44)

Distributions.

C (49:45)

And so many of these risks didn't look so risky if you used distributions where really bad events are very unlikely. But if really bad events are likely, then you've completely mispriced the risk. And if the people upstairs don't understand the models they think you're doing, everything's going fine. So there were, there were multiple layers of moral hazard. There were multiple layers of misapprehension and misunderstanding and, and then just plain short sightedness. The idea that housing prices could rise 10 and 20% a year out to the end of time. Now, what's his name, Prince from Citibank explained that nobody thought that it'd go on forever, but as long as it was going up, if you wanted to get paid, you had to stay in the game. And because the analysts were going to slaughter you if you didn't make money and everybody else did. Now it turns out that not all the big banks believe that.

D (50:52)

And a lot of them are worth.

C (50:53)

A lot more than Citibank is now. Whereas Citibank was the biggest bank in the world 10 years ago and one of the biggest two years ago.

D (51:06)

Gentlemen.

F (51:07)

Just him.

B (51:10)

Hi. I'm a student at LSE and I have a question regarding senior managers and how this theory fits into that. So can you explain a little bit on how this theory would explain, but the senior managers would actually perform to the best in the interest of the firm and also to their best capabilities, even though they have weak incentives.

C (51:31)

Okay, so I'm not sure I understood exactly what you asked me, but it seemed, I think you were asking me how should senior managers have been compensated and how would we know if they were doing a good job? Is that correct?

E (51:48)

Yeah.

B (51:48)

And in the context of weak incentives, if we do give them weak incentives instead of strong, then would they still perform to the best interest of the firm and also that maximum capability.

C (51:58)

So I think what has happened over the last 20 years is CEO compensation has risen from a matter of 20 times that of the average employee to a matter of 200 times that of the average employee. And most of that has been through incentive pay tied to stock price. And I think.

D (52:26)

That that essentially bought off the corporate leaders, the CEOs, and got them to agree to a lot of corporate restructuring that went on in the 80s and 90s. This is a point that's been made by Beg Tolstrom and Steve Kaplan. I think they are given more incentive pay than they need now. I think they can be given incentive pay, but it doesn't need to be anything like as lucrative as it is because, you know, it's a social embarrassment in the first instance. And it leads them to take risks that I don't think shareholders want taken because they don't face any downside. The very worst thing that happens to them is they get fired. And often if they're fired, they end up with golden parachutes. And you can justify those by encouraging risk taking and encouraging people to come into uncertain situations. But the payoffs, the strength incentives to top managers are too strong. Now the problem is what is a good measure? And we've had this from kind of a little bit old fashioned finance theory. We've had that the stock market is the best measure available and in fact is as good a measure as you can hope for. Well, it's pretty clear that the stock market is not an ideal predictor of future performance. And so we've got a problem. And it's a problem that was really very much bigger a year or two ago than I suspect it is now. Because one of the really big problems, this is what led to Enron and WorldCom and whatnot, is you have stock that the CEO knows is overpriced, that it should be worth less, and he and his cohort do everything possible to maintain the stock price and keep up the value of the stock so that they can cash in their dividend, their options. And it leads them to do things that ultimately destroy the firm. I think the best we're going to be able to do is to pay in terms of long term restricted stock. So we let the market do the best job it can of evaluating what they're doing. But they get paid on the upside and the downside. And if the stock goes down, their wealth goes down. It's not just they lose the value of unexercised options. And we require Them to hold it for significant periods so we can really see what the impact of their decisions is. So that's my best guess as to what to do with CEOs in the blue shirt just here in the front.

E (55:43)

Professor, you were referring a couple of times on the basic last 20 years where you put it, the rise of market like incentives. And probably one cannot say only that it's the business or the organizations have done that, but the academia has been probably at the forefront of it. What do you believe or what's your view? What is the reason why this movement has taken place? What is the mistake or shortcoming of the previous decades or centuries, what academia and business have been trying to correct?

D (56:23)

So it's certainly the case that the market is, I mean, Let me stop and think for a minute, but I'll answer that. The reason I think that the market has looked so attractive is, is that it has actually, until quite recently, done very, very well. If you look at the asset markets in terms of we were all always richer every year. And you know, my house that I bought for 235,000 was worth 1.8 million a year ago. And the savings that I had made out of my professor's salary were much more than that. So I thought markets were pretty neat. I certainly wasn't going to. It would have been a hard sell.

C (57:24)

To do anything different because as I.

D (57:26)

Say, much of this theory that I'm expounding has been around for quite a while. I think it was also popular because the people who were really banking out were opinion leaders, the CEOs. CEOs loved this because they were the biggest winners of all of this. And you know, their pay has gone.

C (57:50)

Up.

D (57:53)

10 times relative to the pay of the average worker. And I think we overdid it. We over overdid it. We pushed it too far. You know, if, if you go back, If you go back 200 years, there were very few corporations and mostly it was individual ownership or family ownership. Middle managers don't come along until 150 years ago. And they were certainly not paid any kind of performance pay. Bob Cratchit sat there scratching away and he got a goose on Christmas one year.

C (58:44)

And even if you go back 60.

D (58:46)

70 years, CEOs unless they had big ownership positions, didn't get a lot of performance pay. And they told a very different story.

C (58:55)

You go back to the 60s and.

D (58:57)

Read corporate reports and CEOs told their shareholders that what the corporation was there for was to serve customers and the public and the shareholders long term interest you go into this century and the purpose of the corporation is to generate shareholder wealth. And much of that was going to. A lot of it was going to CEOs, especially in banking, where we're talking about bonuses of billions of dollars being given out every year. So that it was popular because the people who got to people who made money out of it were in a position to implement it.

F (59:48)

John, you were talking about redesigning incentives for CEOs. Now that's going to be done by the board or by the compensation committee. And there's a whole pile of free rider problems and agency problems through shareholders, particularly if shareholders are holding trackers.

D (60:05)

Trackers.

F (60:05)

Tracker funds. Now, what's the where?

C (60:07)

Index funds.

F (60:08)

Index funds. Sorry. Transatlantic misunderstanding. So what are the incentives for boards or compensation committees to deal with this in the way that you were suggesting?

D (60:20)

Well, that's a difficult problem because.

C (60:26)

I.

D (60:27)

Don'T have a lot of experience on boards. But as a board member, you have a number of jobs. One of them is to pick the CEO. One of them is to get rid of the CEO when. When appropriate. One is to decide on CEO compensation, and the other is to advise and mentor and help the CEO. And to do the last of those requires a closeness and you naturally build up friendship relations. And doing the pay piece then becomes particularly difficult. It's not just free rider. It's really unpleasant to tell the boss he's not doing well. But I think a general resolution which is going to be hard to measure, hard to bring off, that we're not going to pay you huge, we're going to pay you in restricted stock that you can't sell. We're not going to give you nearly as much as we're not giving you any raises for many years. And that's going to be very hard to bring off because there'll be other people who continue to pay like that and their boards will be happy to do it and it'll be problematic. And I think the spread of American capitalism models in the last decades has made that a problem around the world. It wouldn't have been a big problem in Germany, even in German banks 10 years ago I think it would be now. It certainly would be a problem bringing it into the British banks, though maybe Royal bank of Scotland is a good place to start. That's another way to move. You know, you can break windows of Mercedes Benzes and that's another form of rewards for performance.

F (62:37)

In Venice, they used to execute the heads of failed banks.

D (62:44)

Chuck Prince, that would be a good one.

B (62:53)

What they actually used to do with them was to put them into sacks with snakes, tie up the sacks and throw it in the river.

C (63:01)

Well, let's be imaginative.

B (63:04)

I'm entirely serious.

C (63:06)

I do.

B (63:08)

But two questions, one of which is what is the heritage of the weak incentives theory? You've talked about ideas which have been developed relatively recently, but how does this relate, for instance, to the work of Oliver Williamson and Koza? And it seems to me that this actually has quite a long heritage. Could you try and put it into that context? The other issue which arises with respect to weak incentives, particularly as you have described them, is that they seem to promote opportunities for grave intransparency, so that they also provide opportunities for favoritism and inefficiencies which are associated with that. And so whilst you have looked at some of the positive sides of weak incentives and quoted Toyota like examples, I think we could also find some negative examples.

D (64:06)

I'm absolutely sure you're right right on that last point. It's hard to design good incentives. If it was easy, we wouldn't have the problems we have in organizations. In terms of your history of thought.

C (64:19)

Question.

D (64:21)

It'S a. Coase, certainly, when he first looked at the firm and tried to explain what's done inside firms, talked explicitly about taking activity activities out of the market and putting them under the authority of a hierarchy. And there isn't a lot of talk in Coase, or for that matter in much of Williamson, about early Williamson, about the incentives that are given inside firms.

C (64:55)

By 85, Williamson's talking about that, but in 75 he's not. But Williamson, 75 was very much a.

D (65:05)

View of the firm as an authority structure, was a central part of it.

C (65:09)

And weak incentives get talked about in.

D (65:12)

85, but they're viewed sort of as.

C (65:17)

A cost of the firm, a drawback of the firm.

D (65:22)

And there's a lot of talk about why is it that they give weak incentives, not that they're desirable, but somehow.

C (65:29)

It'S difficult to give anything else. So he talks about problems of commitment and things like that that prevent strong.

D (65:39)

Prevent weak incentives.

C (65:41)

Then what happened was authority kind of fell out of the economist's purview. Oliver Hart and, and.

D (65:55)

Sandy Grossman and.

C (65:56)

John Moore came up with these formal.

D (65:59)

Models of asset ownership and called that the firm.

C (66:03)

And what happens is there's bargaining between the principal and the agent or the.

D (66:10)

Asset owner and the other person, and.

C (66:14)

There'S no sense of authority.

D (66:16)

The firm is just a commonly owned set of assets.

C (66:19)

And economists, Paul Krugman said a lot.

D (66:24)

Of things and some of them are right.

C (66:26)

And one of them that he said is that it doesn't matter how good an idea is in economics, if it's not put in a formal model, it's.

D (66:38)

Not going to get any traction.

C (66:40)

And there was no formal models of authority.

D (66:44)

There were form.

C (66:45)

You know, there were very sensible discussions.

D (66:47)

But there were no formal models.

C (66:49)

And there were sensible discussions saying that authority wasn't anything.

D (66:54)

It was illusionary, that there was no.

C (66:57)

Difference between an employer, an employee, and a customer and a butcher. You know, I pay you if you.

D (67:03)

Do what I want you to do.

C (67:05)

And if you don't, I fire you.

D (67:07)

That's true whether you're firing you butcher or your employee.

C (67:11)

It's only within the last years that Eric Vandenstein in particular has brought authority back into the firm.

D (67:19)

And that's the last piece that I was talking about. The open disagreement portion gives you formal model, the first formal models that I'm aware of, where authority, in the sense.

C (67:30)

That I tell you what to do with the reasonable expectation that you'll do.

D (67:35)

It, even though it may not be.

C (67:37)

What you think is best to do.

D (67:38)

And even though you have free will.

C (67:40)

To choose not to do it, that's only just come into economics now. It isn't in print yet, and it will be in the AER soon. But it's really just frontline stuff. And that's why there seems to be this disjointness from the old theory. One hit.

D (68:02)

Got time for two more?

A (68:03)

One and then one up.

G (68:06)

Thanks very much, John. I'm trying to be specific and provocative at the same time. I'm trying to be specific in the sense that I use your model of five parameters and take investment banking. And I guess what I can. And that's the provocative part, I guess. I guess what I can derive from that is you do need. And there is a reason in your model, according to your model, for strong, you know, for strong incentives in investment banking. And if you go through that and you say, you know, poor measures, I'd argue you don't find poor measures in investment banking. It's pretty clear what people are there for, what they're looking for, and it's making money for the shareholder. So, you know, you can rule out that one more multitasking. I'd argue there are probably fewer industries in the world which are less streamlined to what they are aiming to achieve. So they're not really multitasking entities. Cooperation is needed. I agree upon that. You find that in the institution. But there are tons of ways, and I've seen that for many years where you can incentivize people that are not, not just looking for the bottom line, but also building the bench, getting juniors in training others, et cetera, PP's. So that can be tackled and are from what I can judge, at least pretty well tackled in the past. Experimentation is desired. Again, I'd argue there's hardly probably any other industry where there's more experimentation and innovation than in the financial services industry. And that can be shown by the growth of the industry, et cetera. So I'm trying to go through those five parameters and I'm actually coming out of the fact, you know, there is in your model a very strong proof that you would have strong incentives and you would need them in investment banking. Is that something you would agree upon or not? And I'm not talking CEOs specifically because I would agree the CEO question is a different thing. That's a binary bet and you would always take the super high risk.

D (70:02)

So if it's true that there's good.

C (70:05)

Measures, then fine, as.

D (70:08)

And if multitasking isn't a big deal, in fact, multitasking, as you say, I.

C (70:12)

Mean sales and trading people, I don't.

D (70:15)

Think there's much multitasking goes on. On the other hand, on the capital markets, the MA is probably a lot.

C (70:25)

Of multitasking because you've got a lot.

D (70:27)

Of juniors you're working with and different.

C (70:29)

Parts of the business, wealth management may be in between or whatever.

D (70:33)

But if you've really got good measures and, and you've got people who are willing to bear risks, fine, fine, give them strong. And if you think they need the.

C (70:41)

Incentives to perform, the other question is.

D (70:46)

Whether they really need 100 million a year to perform. And you know you're not. If you're giving them 100 million when you could get the job done for 10, you're not really taking care of the shareholders interest, except to the extent that you're caught in some sort of game where the other guys are willing to pay 11 and you end up with these traders absorbing a huge portion of the rents. And that may in fact be what goes on. I don't know, it's not absolutely clear to me how good the measures are because you know, several of the banks, Bear Stearns, Lehman Brothers, got in significant trouble and so somewhere something was going wrong. And whether it was simply misunderstanding the risks that were being taken or whether it was because people were driven to take the wrong risks because the right measures weren't in place, that's something I don't know. But if the right measures are there, use them, give strong incentives.

A (71:54)

None in the blue.

D (71:55)

Pull over. I'm afraid this will have to be the last one.

E (72:03)

Thank you very much for your presentation. I wanted to focus on the experimentation question or experimentation point you made. It made a lot of sense what you said in the context of Google where employees have a particular amount of time reserved for the development of new projects. But within the financial industry I found some of the discussion about what's gone wrong with the banking crisis has been driven by experimentation in the sense of. I mean a lot of times the products that caused the trouble are referred to as innovative financial instruments. And so it seemed that having strong incentives led to the development of things like CDOs and other products that may not have been good in the sense of their ability to wreak havoc on the system, but they were a very interesting and innovative way of distributing risk more broadly through the financial system. And even though they were over subscribed, they will likely continue in some form or another. So I was just wondering if you could comment about that tension.

D (73:04)

So absolutely none of this could have happened without the innovation that going on in financial services. If we were living in 1975 banking context, none of this could ever have happened. But. And there have been strong incentives that have driven innovation in capital markets and particularly development of derivatives. But those are, those were strong bad incentives because there wasn't put in place the necessary context of regulation, of oversight, of properly functioning rating agencies.

C (73:51)

And perhaps.

D (73:52)

Of adequate risk management models that meant that the incentives that might have been just fine incentives in a context with proper monitoring and proper controls were bad incentives. Incentives aren't just the pay, they're also the monitoring and the whole managerial context that they're put into. And that's what was missing there.

C (74:24)

Okay, good.

A (74:29)

John got married on Monday. So it's a hell of a way to spend your honeymoon, eh?

D (74:39)

It's a pleasure for me, but not for my wife.

C (74:41)

I suspect.

A (74:46)

One thing that sort of comes over from this is we've had, I mean one way of putting it rather bluntly is we've had 20 years of Jensen and Murphy ruling the Harvard Business Review 1990 with any luck of John Rice's upper husband had 10 years of Roberts ruling. So we'll wait and see. But I'll finish up by setting John on a challenge. He's right. Him and me both support spurs. Let's see if we can devise an incentive system which would actually cause them.

C (75:09)

To be perhaps even second or third.

A (75:11)

Maybe first in the Premier League.

B (75:13)

There's a challenge.

A (75:14)

John, thank you very much.

C (75:15)

Thank you. Thank you all for coming and for your attention and your questions.