Killer Creek has launched! 

I've launched a consulting practice, focused on pricing, promotion effectiveness, new product development, ... really, you should check us out and see what we can do!


Some links:

Killer Creek home


Also, we will be putting lots more content at the new blog site here:

The blog at Killer Creek


Finally, link-in with me and get updates on the latest news from Killer Creek

My LinkedIn


Thanks for stopping by,

Bryan

Information when you don't know what the heck the client wants

Long-ish post on one of Tirole's sub-cases -- yes, I'm back to information transmission.  I just can't quit it!

So, the simple Tirole model I've been talking about makes a couple of analytical cuts -- one on whether the Sender of information (instructor, analyst, whoever) knows the value of the information to the Receiver (student, marketing manager, etc.).  For this post, let's look at the case where S doesn't know R's valuation of the information.

This actually happens quite a lot, in my experience.  Foe example, beyond the immediate grade, plenty of economics students in Econ 101 don't really attach any value to the course.  Or, people who are new at a job might not know the true value of a proposal because they don't understand the web of marketing initiatives currently being implemented: maybe they don't know the opportunity cost of their idea.  Point is, this happens plenty.

So, the Sender has done some analysis on a project that has an uncertain payoff for the Receiver.  So, he is going to present it and see what happens.

Now, the Receiver will know perfectly well how valuable the project is going to be -- IF he makes the investment in time and effort to understand the proposal.  But making that investment is a choice variable in Tirole's model and the decision on the level of effort put in is a function of the Receiver's ex ante belief about the quality of the proposal.  Proposals that have a high value to the Receiver get implemented -- that's what makes them "good."

See how quickly things get complicated?  Even in a simple model?

Tirole measures this ex ante belief as a parameter of congruence, alpha.  Big alpha just means the Receiver figures the idea/analysis is probably pretty good: you came from the right school, look smart, speak French, whatever.  

Small alpha?  Mismatched type fonts, southern accent, northern accent, French accent, who knows?

Point is, when we don't know the real payoffs to the client, Receiver, we look a two cases.  First case: Receiver thinks we have a bad idea, probably.  Second case: Receiver thinks we probably have a good idea. 

So, suppose this alpha happens to be low.  (By low, we mean some critical value that causes the behavior to flip.  Turns out, there is a critical value that causes behavior to flip: that value is derived via first-order conditions of expected utility maximization, etc.)  But if it's low, then both Sender and Receiver make investments in communicating / understanding the results that maximize their respective utilities, subject to this low value.  So, there is some marginal condition that described how hard the two work to communicate.  Communication efforts are made, information might get understood, projects might get implemented, we might all be surprised.  Pretty standard, expected marginal cost equals expected marginal benefit stuff.

It could be that the alpha is sufficiently low that the equilibrium effort level is zero -- R really assumes S's ideas are terrible.  All this is pretty much what you'd expect.  So, cool.

Or not cool.  If the Receiver has very low expectations, or if he has very high costs of understanding the analysis, then communication can breakdown.  With very poor results.  

Very poor.

The second-worst meeting I was ever involved in ended with the Sender telling the Receiver something like, "Look, understanding this analysis is just part of the standard 21st century marketing skill set.  Surely they taught those at whichever business school you went to."

Wasn't me who said it.  But, no, things did not get better after that for the analytics team.

Now suppose that alpha is high; R has an ex ante belief that the project is going to have a good payoff for him and that he will, therefore, implement it.  Now things get interesting.

They get interesting because what happens next depends on the sort of oversight R is supposed to exercise over the proposal.  If the oversight is what Tirole calls "executive," then R must really, truly understand the proposal in order to implement it.  He's got to implement it himself, under his own direction.

So, if R is a student, he is going to have to take a test on the subject.  If R is a marketing manager, maybe he is going to have to understand the pricing analysis in order to implement it on a case-by-case basis.  Point is, the efforts made to ensure communication are exactly the same for high ex ante beliefs as they are for low ex ante beliefs about the project congruence.

So, prior beliefs don't matter here.

So, the effort in understanding the analysis will be calculated the same way as it was when the Receiver assumed Sender's idea was bad.  It's just that, in this case, we have higher equilibrium levels of investment ('cause everybody assumes it is a good idea) and a different motivation (R is on the hook for implementation, not just trying to figure out if it is a good idea or not).

Where they do matter is when the oversight is simply "supervisory."  By which, Tirole means that the Receiver can simply approve a project and it gets implemented.  If the analysis simply said, "Change the price from X to Y," the marketing guys wouldn't have to understand the analysis in order to make the change.  They would just have to believe the analysis.

With the ex ante belief about alpha being high and the oversight to be exercised being supervisory, communication will actually break down.  

!!??

Nobody will make efforts to communicate because

1.  The Receiver already figures the proposal is pretty good (that's what a high alpha means, after all) and will rubber stamp the idea, and

2.  The Sender is smart enough to take "yes" for an answer. 

So, in this case, the Sender has real authority when it comes to projects that get implemented.  The Receiver becomes a rubber stamp for whatever ideas the Sender cooks up.  In this case there isn't any real communication at all, maybe just a cheesy PowerPoint presentation with a couple un-serious area charts with arrows and hand-waving predictions.  

But it's all good.  Receiver gets to not invest in costly understanding but still gets projects that are probably really good.  Sender gets effective control of project selection.  What's not to like?

So, in the case where S does not know R's payoff, we expect two sorts of outcome.  First, it might be that both sides make an effort to communicate and understand the analysis.  That's because either

1)  the Receiver is going to have to implement the idea himself or else it is because

2)  the Receiver doesn't think the analysis is going to be very good (but thinks it is worth finding out if he is right or not).

The second outcome is where communication breaks down and nobody makes an effort to communicate. That is because either

3)  the Receiver thinks the idea is going to be plenty good and, since he doesn't have to understand it to make it work, he might as well rubber stamp whatever idea come his way, or else

4)  the Receiver has a very low guess about the quality of the idea (and the guess is so low that it isn't worth it to bother finding out more).

 

Shadow Banking in Spain

The Wall Street Journal had a really interesting piece yesterday about the rise of "time banks" in Spain.  The idea being that, in an economy as thoroughly broken as the Spanish economy, many unemployed attempt to use these intermediaries to facilitate a barter economy.

So, a bunch of people join together and offer to work on various odd jobs for each other, banking the time and using those collected hours to request services from others.  Thus, Silvia Martin

...has relied on other time-bank members to give her lifts around town for her odd jobs and errands, as well as to help with house repairs.  In return, she has cared for members' elderly relatives, organized children's parties and even hauled boxes for a member moving to a new house

Of course, the unemployment rate in Spain for 25-34 year-olds is an absurdly high 27%, while the rate for 16-24 year-olds is an even more absurdly, dangerously, high 53%.  At least they aren't in a recession!

But it is interesting to observe that people are willing to enter these very small economies rather than keep attempting to make the regular economy work.  At some level, they must be making the decision that trading without specialization (all hours are valued equally) and among a very small geographical group are costs worth paying in order to begin making some trades for labor.  Amazing, really.

But, as you might expect, the idea of a time bank, where the unit of exchange is measured in hours, is just a very small step away from turning it into a real bank which uses its own currency.

Which is exactly what has happened in some cases. 

One bank launched something called an "eco," a currency they just made up.  Turns out, not only have dozens of local businesses decided to accept it as a basis for trade, so have two town governments!  

There are some ways in which this is a very positive story: people working to solve problems the government is unwilling to solve.  On the other hand, this represents groups of people choosing to work in a parallel economy with a non-convertible currency (and non-convertible even in, like, the next town over).  That just gives you an idea of how desperate things are in Spain. 

Hard information and cheap talk

It is pretty typical for economists to view information as either “hard” or else “soft.” The difference being the ability of the person receiving the information to verify the claim made. Soft information is unverifiable and is believed only when the person receiving it figures his interests and the interests of the person sending it are aligned.

So, when my wife suggests we meet for lunch at our favorite place in Carytown, I just go ahead and believe her. Probably, she really does want to meet for lunch instead of wanting to design an elaborate ruse to waste time. Soft information can be useful in cases like this.  This is cheap talk -- unverifiable and untrustworthy information, unless the interests of all the parties are sufficiently aligned.


Hard information is verifiable. Receivers of hard information are more likely to accept it as true even when the sender’s interests are not perfectly aligned with the receiver’s interests because false statements could be exposed (though, of course, the receiver might still verify statements whenever interests are very poorly aligned). UPS used to tell a story about a good employee who does not get a promotion and a great employee who does. Turns out, the difference was that when the good employee was asked a question, he returned with a verifiable answer to it; when the great employee was asked the same question, he returned with a boatload of verifiable information about it: shipment s left at time t, weighing w pounds and driven by driver d, etc., etc.

Of course, information is often neither hard nor soft, but somewhere in-between. Lots of information requires an effort by the person receiving it to turn it from soft information into hard information. A great example is a mathematical proof you might come across in a paper. Most of us, even if we are good at math, would have to make an effort to confirm that, in fact, for every epsilon there really is a delta such that…

Otherwise, and the way I usually do it, I just go ahead and believe the guy writing the proof. Here, information that could be verified is simply left as soft information and accepted based on the idea that the writer is probably not trying to get me to believe something about Proposition 26 that isn’t true.

But it is also generally true that soft information can be turned into verifiable statements when both the sender and the receiver make investments (time, effort, education, whatever). So, the “hardness” of information is variable and, more than that, a variable the listener can control. And that makes things more complicated.

Communicating Sophisticated Analytics

I've got a problem with our client group...


Marketing analytics must include, finally, communication with other parts of the firm. As anybody who performs sophisticated analytics can attest, communication is not necessarily the easiest part of the process, not least because it involves a strategic interaction between the analysts (Senders) and the client groups (Receivers). This strategic interaction is filled with places where even the best analytics can wind up being discarded. So, we want to avoid that. How?

The communication effort should be built to manage the incentive alignment concerns of the recipients of the analysts and also take into account the fact that there is a moral hazard in this sort of communication.

One intuitive example of incentive alignment is when new analysts show up and begin looking at a process that has been active for a while. Newbies tend to find all sorts of places where the process seems sub-optimal to them and make recommendations on how to improve the process. It is very difficult for others to evaluate these recommendations because the incentive for new employees is to over-state the degree to which old processes need to be fixed. Finding problems puts them on the map, so they tend to find more than they should.


Well, we aren't newbies.  So what's the real problem?

But the moral hazard is the real problem. 


Side explanation: moral hazard

Moral hazard is the name economists use to talk about the problem that creeps up whenever costs are borne privately but benefits are shared: people tend to not want to invest in those costs.  So, to use a typical example, car insurance is subject to moral hazard: we'd all benefit if I drove carefully, but since my personal cost of an accident is low, I might drive faster than I ought.  Similar thing with communication: we both benefit if a good project gets implemented, but I might want to skimp on the cost of understanding the analysis that lets us understand the project.

It’s like this: getting knowledge from the head of the researcher into the head of the group that needs it requires investments from both sides. The analyst needs to make a good and effective presentation (which might very well involve multiple presentations, background conversations, multiple presentations given to multiple groups, etc.), but the group receiving the analysis needs to make investments, too. Understanding sophisticated analysis isn’t easy, especially if it uses unfamiliar techniques.

So, the payoff to the communication depends on the effort the other team puts in to understanding the analysis. And that creates room for moral hazard.

So, what do you do? Dewatripont and Tirole (2005) have a very interesting paper on this problem, which develops a model for this sort of communication. Like a lot of papers which deal with cooperative solutions, they wind up with a fairly large set of sub-cases, depending on the receiver’s assumptions about the degree to which his interests are aligned with the sender’s, the type of oversight (supervisory or executive) the receiver will ultimately have to exercise over the recommendation, the level of certainty the receiver has about the sender’s payoff associated with the project, etc.


Sounds complex

It gets dicey, for sure. The key insights are that a decrease in either party’s stake in the project will lessen the total communication effort. If there are communication “set-up costs” (new analytical techniques being introduced to the discussion, for example, that not everyone is familiar with), then we can see sudden and discontinuous breakdowns in communication. Senders of information should invest in positive cues about their credibility; that way, Receivers will be more likely to invest in the joint communication.

These cues come from sources you might expect: people who know the history of the Sender, people who have done some investigation of the analytical claims, etc. The goal of obtaining these cues is to convince the client to engage and evaluate the analysis. Once that happens, the analyst’s job is made much easier.



So,  need other people to endorse the analysis?  Where have I heard that before?

OK. Sounds obvious, I know. But it is a mistake I’ve made in the past and I’d bet it’s one others have made as well. Client can’t / won’t engage the analysis because he doesn’t have the skills to do so and doesn’t want to just rubber-stamp my project. So what do you do? If you are naïve, you ask him again, only using more flattery this time. Won’t work. What you need is some borrowed credibility – that’s the key.  Without it, you simply can't expect to win -- communication is broken and can't be fixed.

Borrowed Money

Some interesting time series charts from the Federal Reserve in St. Louis I've been looking at recently.  They involve consumer revolving credit.  In the aftermath of the Great Recession, Americans are re-balancing their accounts and, for the first time in forever, reducing their revolving debt levels.  Check this:

-- Per-capita revolving debt in 2005 constant dollars (shaded regions represent recessions)

We can see the big increase beginning in the 1990s, with the development of considerably more sophisticated credit models.  These models allowed credit card companies to separate the unsecured credit market into correct risk categories.  Because of this, we should probably interpret this big increase as both an increase in credit demand as well as in increase in credit supply: the equilibrium where we are able to separate credit riskiness probably has lots more loans being offered than the equilibrium where we have to pool the risky households with the safe ones.

But since the recession, we see decreases in these borrowing amounts.  It is important to note that these are separate from the loans that have been written off or otherwise defaulted; similarly significant changes also live in time series that look at cohorts of stable accounts that have not defaulted.

There are several interesting possible explanations for this shift in consumer behavior.  The two that I find most interesting are the changes in government spending and the lack of change in personal disposable income.  A couple more charts should make the point.


Here is the Federal spending as a percent of GDP.  Note that, during the Clinton years, this percent was decreasing ("The era of big government is over.") and consumer borrowing was increasing rapidly.  Meanwhile, as government spending has increased significantly, consumers perhaps grasp that all that spending has to be repaid by somebody and so you'd better start saving for the time when taxes increase to pay that debt off.

So, this has some plausible affect on the level of consumer borrowing.  The other -- and to my mind more interesting -- time series is the constant dollar household disposable income, per capita.  Check this nightmare out:

Here is the per-capita disposable (that is, after tax) income series since 1991.  This measure has stalled entirely since 2006.

This gives some support to the feeling that consumers perceive that their lifetime income profile might not be as good as they used to think.  We can match this picture up with all sorts of poll data that suggests people think the next generation won't be as well off, or that the country is on the wrong track, or that consumers are gravely concerned about the economy.  This chart provides a pretty good explanation.

In an environment where real disposable incomes are increasing, real wealth is increasing rapidly (we don't need a picture of housing prices!) and credit supply is expanding (because the companies can tell good risks from bad ones), it shouldn't surprise us that consumers want to pull lots of future consumption into right now.  Our borrowing should have increased in that situation (maybe not as much as it actually did, but still..).

I suspect that people interpret the decrease in wealth associated with the housing bust, the significant increase in government debt, and the six years of no income increases as strong signals that the future income will be small enough that they don't want to pull as much of it into today.  

This is a praticular risk for firms and individuals who buy consumer debt.  And there are forward-looking indicators that should be of keen interest to them.

Mortgage Defaults

Came across an interesting paper on mortgate defaults the other day while I was, actually, searching for some models of credit card default.

The authors, Campbell and Cocco, develop a microeconomic model of this important behavior and come up with some interesting results.

In particular, they make a clean connection between negative equity and borrowing constraints when it comes to default. Negative equity increases the default risk, of course, but the degree of negative equity that triggers a default depends on the degree to which the household is constrained in the credit markets. pofoundly constrained households can default at low levels of negative equity.

The other interesting result is the breakdown of the particular risks of the various types of home financing. For example, interest-only mortgages have the highest risk of default waves. Even so, there are ranges of risk events for which they have some clear advantages over other forms of financing. For one, these borrowers are less likely to face borrowing constraints than other types of borrowers. This benefit is generally overwhelmed by the equity risk they face -- interest-only homes are the most exposed to decreases in home value.

The paper is pretty deep and will take several readings to get a handle on the arguments. But the idea of approaching the question with a dynamic micro-model is significant.