Colleagues outside of Economics often marvel at the coordinated nature of the Economics job market. The job market is so efficient, that the profession no longer wastes resources by having everyone read each candidate’s job market paper. That task is assigned to one person (Tyler Cowen) who reports back to the rest of us. In case you missed the report, here it is

Economics is not alone in having a coordinated job market. Philosophy has one, but it has begun to show signs of unraveling. The ability to interview via Skype, for example, has reduced the value in the eyes of many, for a preliminary interview at their annual meeting. In response, the American Philosophy Association posted the following statement regarding the job market calendar:

For tenure-track/continuing positions advertised in the second half of the calendar year, we recommend an application deadline of November 1 or later. It is further recommended that positions be advertised at least 30 days prior to the application deadline to ensure that candidates have ample time to apply.

In normal circumstances a prospective employee should have at least two weeks for consideration of a written offer from the hiring institution, and responses to offers of a position whose duties begin in the succeeding fall should not be required before February 1.

Its natural to wonder if the Economics market is not far behind. Skype interviews are already taking place. The current set up requires a department to evaluate and select candidates  for preliminary interviews within a month (roughly the middle of November to mid December) which is hardly conducive to mature reflection (and argument).

I don’t often go to empirical talks, but when I do, I fall asleep. Recently, while so engaged, I dreamt of the replicability crisis’ in Economics (see Chang and Li (2015)). The penultimate line of their abstract is the following bleak assessment:

Because we are able to replicate less than half of the papers in our sample even with help from the authors, we assert that economics research is usually not replicable.’

Eager to help my empirical colleagues snatch victory from the jaws of defeat, I did what all theorists do. Build a model. Here it is.

The journal editor is the principal and the agent is an author. Agent has a paper characterized by two numbers ${(v, p)}$. The first is the value of the findings in the paper assuming they are replicable. The second is the probability that the findings are indeed replicable. The expected benefit of the paper is ${pv}$. Assume that ${v}$ is common knowledge but ${p}$ is the private information of agent. The probability that agent is of type ${(v,p)}$ is ${\pi(v,p)}$.

Given a paper, the principal can at a cost ${K}$ inspect the paper. With probability ${p}$ the inspection process will replicate the findings of the paper. Principal proposes an incentive compatible direct mechanism. Agent reports their type, ${(v, p)}$. Let ${a(v, p)}$ denote the interim probability that agent’s paper is provisionally accepted. Let ${c(v, p)}$ be the interim probability of agent’s paper not being inspected given it has been provisionally accepted. If a provisionally accepted paper is not inspected, it is published. If a paper subject to inspection is successfully replicated, the paper is published. Otherwise it is rejected and, per custom, the outcome is kept private. Agent cares only about the paper being accepted. Hence, agent cares only about

$\displaystyle a(v, p)c(v,p) + a(v, p)(1-c(v,p))p.$

The principal cares about replicability of papers and suffers a penalty of ${R > K}$ for publishing a paper that is not replicable. Principal also cares about the cost of inspection. Therefore she maximizes

$\displaystyle \sum_{v,p}\pi(v,p)[pv - (1-p)c(v,p)R]a(v,p) - K \sum_{v,p}\pi(v,p)a(v,p)(1-c(v,p))$

$\displaystyle = \sum_{v,p}\pi(v,p)[pv-K]a(v,p) + \sum_{v,p}\pi(v,p)a(v,p)c(v,p)[K - (1-p)R].$

The incentive compatibility constraint is
$\displaystyle a(v, p)c(v,p) + a(v, p)(1-c(v,p))p \geq a(v, p')c(v,p') + a(v, p')(1-c(v,p'))p.$

Recall, an agent cannot lie about the value component of the type.
We cannot screen on ${p}$, so all that matters is the distribution of ${p}$ conditional on ${v}$. Let ${p_v = E(p|v)}$. For a given ${v}$ there are only 3 possibilities: accept always, reject always, inspect and accept. The first possibility has an expected payoff of

$\displaystyle vp_v - (1-p_v) R = (v+R) p_v - R$

for the principal. The second possibility has value zero. The third has value ${ vp_v -K }$.
The principal prefers to accept immediately over inspection if

$\displaystyle (v+R) p_v - R > vp_v - K \Rightarrow p_v > (R-K)/R.$

The principal will prefer inspection to rejection if ${ vp_v \geq K}$. The principal prefers to accept rather than reject depends if ${p_v \geq R/(v+R).}$
Under a suitable condition on ${p_v}$ as a function of ${v}$, the optimal mechanism can be characterized by two cutoffs ${\tau_2 > \tau_1}$. Choose ${\tau_2}$ to be the smallest ${v}$ such that

$\displaystyle p_v \geq \max( (R/v+R), ((R-K)/R) ).$

Choose ${\tau_1}$ to be the largest ${v}$ such that ${p_v \leq \min (K/v, R/v+R)}$.
A paper with ${v \geq \tau_2}$ will be accepted without inspection. A paper with ${v \leq \tau_1}$ will be rejected. A paper with ${v \in (\tau_1, \tau_2)}$ will be provisionally accepted and then inspected.

For empiricists the advice would be to should shoot for high ${v}$ and damn the ${p}$!

More seriously, the model points out that even a journal that cares about replicability and bears the cost of verifying this will publish papers that have a low probability of being replicable. Hence, the presence of published papers that are not replicable is not, by itself, a sign of something rotten in Denmark.

One could improve outcomes by making authors bear the costs of a paper not being replicated. This points to a larger question. Replication is costly. How should the cost of replication be apportioned? In my model, the journal bore the entire cost. One could pass it on to the authors but this may have the effect of discouraging empirical research. One could rely on third parties (voluntary, like civic associations, or professionals supported by subscription). Or, one could rely on competing partisan groups pursuing their agendas to keep the claims of each side in check. The last seems at odds with the romantic ideal of disinterested scientists but could be efficient. The risk is partisan capture of journals which would shut down cross-checking.

When analyzing a mechanism it is convenient to assume that it is direct. The revelation principle allows one to argue that this restriction is without loss of generality. Yet, there are cases where one prefers to implement the indirect version of a mechanism rather than its direct counterpart. The clock version of the English ascending auction and the sealed bid second price auction are the most well known example (one hopes not the only). There are few (i.e. I could not immediately recall any) theorems that uniquely characterize a particular indirect mechanism. It would be nice to have more. What might such a characterization depend upon?

1) Direct mechanisms require that agents report their types. A concern for privacy could be used to kill’ off a direct mechanism. However, one would first have to rule out the use of trusted third parties (either human or computers implementing cryptographic protocols).

2) Indirect mechanism can sometimes be thought of as an extensive form game and one might look for refinements of solution concepts for extensive form games that have no counterpart in the direct version of the mechanism. The notion of obviously dominant strategy-proof that appears here is an example. However, indirect mechanisms may introduce equilibria, absent in the direct counterpart, that are compelling for the agents but unattractive for the designers purposes.

3) One feature of observed indirect mechanisms is that they use simple message spaces, but compensate by using multiple rounds of communication. Thus a constraint on message spaces would be needed in a characterization but coupled with a constraint on the rounds of communication.

From Kris Shaw, a TA in for my ECON 101 class, I learnt that the band Van Halen once required that brown M&M’s not darken their dressing room door. Why? Maybe it was a lark. Perhaps, a member of the band (or two) could not resist chuckling over the idea of a minor factotum appointed to the task of sorting the M&Ms. When minor factotum is asked what they did that day, the response was bound to elicit guffaws. However, minor factotum might have made it a point to not wash their hands before sorting the M&Ms. Then, who would be laughing harder?

A copy of the M&M rider can be found here. Along with van Halen’s explanation of why the rider was included:

……the group has said the M&M provision was included to make sure that promoters had actually read its lengthy rider. If brown M&M’s were in the backstage candy bowl, Van Halen surmised that more important aspects of a performance–lighting, staging, security, ticketing–may have been botched by an inattentive promoter.

So the rider helps screen, apparently, whether the promotor pays attention to detail. I think the explanation problematic. It suggests that it is hard to monitor effort expended by promoter on important things like staging for example. So, monitor something completely irrelevant. The strategic promoter should shirk on the staging and expend effort on the M&Ms.

Duppe and Weintraub date the birth of Economic Theory,  at June 1949. It was the year in which Koopmans organized the Cowles Commission Activity Analysis Conference. It is also counted as conference Zero of the Mathematical Programming Symposium. I mention this because the connections between Economic Theory and Mathematical Programming and Operations Research had, at one time been very strong. The conference, for example, was conceived of by Tjalling Koopmans, Harold Kuhn, George Dantzig, Albert Tucker, Oskar Morgenstern, and Wassily Leontief with the support of the Rand corporation.

One of the last remaining links to this period who straddled, like a Colossus, both Economic Theory and Operations Research, Herbert Eli Scarf, passed away on November 15th, 2015.

Scarf came to Economics and Operations Research by way of Princeton’s mathematics department. Among his classmates was Gomory of the cutting plane method Milnor of topology fame and Shapley. Subsequently, he went on to  Rand ( Dantzig, Bellman, Ford & Fulkerson). While there he met Samuel Karlin and Kenneth Arrow who introduced him to inventory theory. It was in this subject that Scarf made the first of many important contributions: the optimality of (S, s) polices. He would go on to establish equivalence of the core and competitive equilibrium (jointly with Debreu), identify a sufficient condition for non-emptiness of the core of a NTU game (now known as Scarf’s Lemma), anticipated the application of Groebner basis in integer programming (neighborhood systems) and of course his magnificent Computation of Economic Equilibria’.

Exegi monumentum aere perennnius regalique situ pyramidum altius, quod non imber edax, non Aquilo impotens possit diruere aut innumerabilis annorum series et fuga temporum. Non omnis moriar…….

I have finished a monument more lasting than bronze and higher than the royal structure of the pyramids, which neither the destructive rain, nor wild North wind is able to destroy, nor the countless series of years and flight of ages. I will not wholly die………….

A Markov Decision Problem (MDP) is a model for sequential decision making, in which the underlying state of nature evolves in a stationary way. An MDP is given by a set of states S, an initial state s(0) in S, a set of available actions A(s) for each state s in S, and, for each state s in S and available actions a in A(s), a payoff r(s,a) and a probability distribution q(. | s,a) on S.
The process starts at the initial state s(0). At every stage n, the current state s(n) is known, the decision maker chooses an action a(n) in A(s(n)), receives the stage payoff r(s(n),a(n)), and a new state s(n+1) is chosen according to q(. | s(n),a(n)) and is told to the decision maker. The decision maker’s goal is to maximize the discounted sum of his stage payoffs:

sum-over-n-from-0-to-infty of λ-to-the-power-n times r(s(n),a(n)).

The value of the MDP, that is, the maximum that the decision maker can obtain, depends on the discount factor λ. Denote by v(λ) the value function of the MDP. Which functions can be obtained as the value function of some MDP with finite sets of states and actions?

From now on I restrict the discussion to MDP’s with finite sets of states and actions. Blackwell (1965) proved that for every discount factor λ the decision maker has a pure stationary optimal strategy. It is easy to see that the payoff that corresponds to a pure stationary optimal strategy is the solution of a set of equations, which are linear in λ, and whose coefficients are determined by the payoff function r and the transition function q. It follows that for every pure stationary strategy σ, the corresponding payoff function g(λ ; σ,s) is a rational function of λ. Since there are finitely many pure stationary strategies, we deduce that the value function is the maximum of finitely many rational functions.

Can we obtain any maximum of rational functions as the value function of some MDP? The answer is negative. For example, since the set of states and actions are finite, the payoff function r is bounded, say, by M. In particular, the payoff function of any strategy is bounded by M/(1-λ). In particular, any rational function whose denominator has a root inside the unit ball of the complex plane, or that has a root on the unit ball of the complex plane that has multiplicity larger than 1, cannot be the value function of an MDP.

Is that the only restriction that we have? The answer is still negative. It is not difficult to see that the roots of the denominator of the payoff function of a pure stationary strategy are the inverse of the eigenvalues of the transition matrix, which by a known result in matrix theory must be unit roots, that is, for any root ω of the denominator (which is a complex number) there is an integer k such that ω-to-the-power-k is equal to 1. Thus, a rational function whose denominator has a root that lies on the unit ball of the complex plane and is not a unit root cannot be the value function of an MDP.

Is that all? Yes. Let F be the set of all rational functions f : [0,1) → R that satisfy the following property: any root of the denominator either (a) lies outside the unit ball of the complex plane, or (b) lies on the unit ball of the complex plane, has multiplicity 1, and is a unit root. Let V be the set of all functions that are the maximum of finitely many functions in F. A function v is the value function of some MDP if and only if it is in V.

In this post I outlined one direction of the proof. Anyone who is interested in reading the construction that proves the other direction is referred to this paper.

You shouldn’t swing a dead cat, but if you did, you’d hit an economist doing data. Wolfers wrote:

“…...modern microeconomists are more likely to spend their days knee-deep in large-scale data sets describing the real-world decisions made by millions of people, and less likely to be mired in Greek-letter abstractions.”

Knee-deep usually goes with shit, while mired with bog. I’ll pick bog over shit, but suspect that that was not Wolfers’ intent.

The recent paper by Chang and Li about the difficulty of replicating empirical papers  does rather take the wind out of the empirical sails. One cannot help but wonder about the replicability of replicability studies. No doubt, a paper on the subject will be forthcoming.

Noah Smith on his blog wrote:

So the supply of both good and mediocre empirics has increased, but only the supply of mediocre theory has increased. And demand for good papers – in the form of top-journal publications – is basically constant. The natural result is that empirical papers are crowding out theory papers.

Even if one accepts the last sentence, the first can only be conjecture.  One might very well think that the supply of mediocre empirical papers is caused entirely by an increase in the supply of mediocre theory papers whose deficiencies are  glossed over with a patina of empirics. Interestingly, when reviewers could find nothing nice to say about Piketty’s theories they praised his data instead. Its like praising the author of a false theorem by saying while the proof is wrong, it is long.

The whole business has the feel of  tulip mania. Empirical papers as abundant as weeds. Analytics startups as plentiful as hedge funds. Analytics degree programs spreading like herpes. Positively Gradgrindian.

“THOMAS GRADGRIND, sir. A man of realities. A man of facts and calculations. A man who proceeds upon the principle that two and two are four, and nothing over, and who is not to be talked into allowing for anything over.”

In empirical econ classes around the world I imagine (because I’ve never been in one) Gradgrindian figures laying down the law:

“Facts alone are wanted in life. Plant nothing else, and root out everything else. You can only form the minds of reasoning animals upon Facts: nothing else will ever be of any service to them.”

I have nothing against facts. I am quite partial to some.  But, they do not speak for themselves without an underlying theory.

Chu Kin Chan, an undergraduate student from the Chinese University of Hong Kong, has collected the placement statistics of the top 10 PhD programs in Economics from the last 4 years. You can find the report here. In it you will find the definition of top 10 as well as which placements counted’. Given that not all PhD’s in economics who get academic positions do so in Economics departments, you can expect some judgement is required in deciding if a placements counts as a top 10′ or `top 20′.

The results are similar to findings in other disciplines (the report refers to some of these). The top 10 departments place 5 times as many students in the top 20 departments as do those ranked 11 through 20. If you score a top 10 placement as +1, any other academic placement as a 0 and a non-academic placement as a -1, and then compute an average score per school, only one school gets a positive average score: MIT.

Chan also compares ranking of departments  by placement with a ranking  based on a measure of scholarly impact proposed by Glen Ellison. What is interesting is that departments that are very close to each other in the scholarly impact rating can differ quite a lot in terms of placement outcomes.

Read in tandem with the Card & Della Vigna study on falling acceptance rates in top journals and the recent Baghestanian & Popov piece on alma mater effects makes me glad not to be young again!

In a previous post I wrote on my experience as a consultant to participants in auctions. I was interested to hear the other side of the coin: how do the ones who set the rule of the auction perceive the competitive situation they are in charge of. To answer this question I met Dorit Levy Tyller, a well known Israeli advocate who has decades of auctions in her professional past.
According to Ms. Levy Tyller, the issue that bothers her most is coordination among the bidders. In a small country like Israel, in which everybody knows everybody else and have friends who know the rest, participants do their best to talk, exchange information, and dissuade others from increasing their bid. When the bidders are all present at the same hall, the auction turns into an oriental bazaar with a lot of psychological pressure on participants.
To overcome this difficulty, Ms. Levy Tyller assigns each participant to a different room and asks the participants to arrive to their designated rooms at different times. During the auction she moves with her team from one room to the next, informing each participant of the current highest bid and asking them whether they increase their bid. This is a slow process that requires the participants’ trust in the auctioneer, a trust that she gained with the dozens of auctions she had already organized.
What are the issues that affect the participants’ behavior? According to Ms. Levy Tyller, the expectation to win the auction and the tension that builds along the process causes the bidders to increase their bids. Pressure from other participants, on the other hand, hinders price increase.
Most winners are the calculated and level-headed participants. Anxious bidders who make plenty of noise usually quit before the end. Moreover, those who come prepared and know well the status of the auctioned item, tend to win more often.
At the end of our conversation Ms. Levy Tyller admitted that I was the first game theory consultant she ever met. I take it as a good sign: the utility function of game theorists puts higher weight to research and teaching than to consulting jobs. I am glad to be part of this group.

Trump’s rise in the republican polls puzzles many. It shouldn’t. He is the Putin that some republicans have longed for. Here is a sampling:

Bush II:

I looked the man in the eye. I found him to be very straight forward and trustworthy and we had a very good dialogue.

Mike Rogers, GOP chairman of the House Intelligence Committee:

Putin is playing chess while Obama is playing marbles.

Sarah Palin:

Look it, people are looking at Putin as one who wrestles bears and drills for oil. They look at our president as one who wears mom jeans and equivocates and bloviates.

Rudolph Giuliani:

But he makes a decision and he executes it, quickly. Then everybody reacts. That’s what you call a leader.

If you think the comparison to Putin far fetched, here is Putin:

For the first time in the past 200–300 years, it (Russia) is facing the real threat of slipping down to the second, and possibly even third, rank of world states.

Now,  compare with Trump’s slogan to make America great again.