More on immigration - of the intergalactic kind

Gabriel links to a webpage which has tools for teaching economics. Some of it's good, some goofy, and pretty much a lot of it of the kind that prompts some people to cry that "neoclassical brainwashing" is going on (essentially because you're teaching kids about scarcity and there are theories of economics out there which are not based on the idea of scarcity which are not getting equal treatment). Anyway, I did think this one was funny:



Concept: Opportunity cost, comparative advantage What's wrong with this picture? It's nice of Superman to rescue a kitten, but has he considered the opportunity cost of doing so? Rescuing kittens is so easy that children often do it. With all the accidents, crimes, and natural disasters that occur in the world, surely he could spend his time more productively. The concept of comparative advantage suggests that Superman should focus on tasks that others can't do well, like stopping runaway trains or transporting nuclear weapons into deep space so they can detonate safely.

(isn't it just like an economist to ruin all the fun? typical!)

But it makes you wonder. The guy can fly so fast around the earth that he can make it rotate backwards. Surely then he would have the time to, I don't know, build everyone in the world a car. And a house. And saw (sew?) like 500 t-shirts for everybody in the world. And plant a bunch of sugar cane. And write some script. And answer some telephone calls in India. And mow my lawn. Then he get can get back to chasing those goofy looking villains.

And why doesn't the Fraternal Order of Police ever complain that Superman is taking away their jobs and lowering their wages? Surely as much crime as that guy fought and defeated there must've been huge layoffs in the law enforcment/crime fighting industry and who's ever left is making a pittance. You know that if this stuff was for real people'd be organizing and demanding that Superman get deported to wherever he came from.

Now of course in the world of Middle Earth it's perfectly understandable why someone like Gandalf wouldn't use his superpowers to increase output. He'd know, being wise, that he was living in a Malthusian world and an increase in output would just result in more hobbits with no change in hobbit standard of living. And the last thing he'd want to do is lower the hobbit mortality rate since that would actually reduce hobbit income per capita.

Yeah, Dani Rodrik's right

That was the first thought that occured to me when I woke up this morning. What can I say in my defense? Sometimes I'm slow and I have to sleep on it (and of course I'm not sneaky). Basically, the first-order/second-order impacts of tariff reduction on efficiency and producer surplus means that it is always possible to draw your supply and demand curves so that the efficiency gain from tariff reduction is arbitrarily small. Not zero, but as small as you want it. On the other hand it is NOT possible to draw your supply (and demand, but this doesn't play a role) curve in a way so that the change in producer surplus is small:




You can draw the above curves to make the blue triangles really small. But, given a change in prices due to change in tariff, you cannot make that red rectangle go away.

Of course if the efficiency gains are small and positive and loss to producers is large than this means that the change in consumer surplus has to be large. This is another way of saying that most of the benefits of tariff reduction are diffused over the millions of people who purchase a given good, while the costs of tariff reduction are concentrated within the particular industry - the group of producers, both workers and capital owners.

I also vaguely recall some floatin', shrinkin', shriekin' Harberger triangles from my dreams.

How much of a jerk do you have to be to oppose immigration?

Update: Don't Trip Up does a similar calculation for Britain and EE

Both Alex Tabarrok and Dani Rodrik have come out in favor of immigration into US on the basis that the relevant "moral community" one should consider is the world and not just the US natives. It might be the case that immigration from Mexico into US lowers the wages of the unskilled workers here (the extent of this effect is subject to some controversy, see the previous post on Ottaviano and Peri). However, the increase in the migrants' wages is so large that support for immigration is still justified.

This kind of argument provokes the expected response from the expected folks, roughly along the lines that we should care more about native workers - the citizens - then the migrants - the non-citizens. Ok. But how much more? Let's put on our annoying-economist hat and consider the question; if you consider a foreign national to be only 1/2 a human being (alright, alright, only 1/2 as "important") as a native citizen, are you justified in opposing immigration? After all, it takes a real jerk to argue that foreign people's welfare should not count at all. Suppose the foreigners are only 1/10th as important? Surely, if natives' welfare counts for ten times as much as that of foreigners, we would be justified in banning immigration since it may adversely affect the wages of the unskilled in US? Well, let's see...

Suppose we transfer one person from Mexico to United States (illegally or otherwise). As a result his wages increase compared to what he was making in Mexico. Let us also suppose that as a result of this transfer the wages of some unskilled worker in US fall. Furthermore we will ignore the aggregate gains from immigration that occur and which all economists, including Borjas admit exist. We do this to make our job harder, not easier.

How much do you have to weight the native's welfare relative to that of the Mexican immigrant in order to oppose moving this migrant into US?

Well, in the standard framework we have ourselves a Social Welfare Function - in this case a utiliterian one with unequal weights attached to the welfare of natives and immigrants. The key is actually that there's decreasing marginal utility of wealth. Let be the weight attached to the well being of the native. So the weight attached to the well being of the immigrant is . Let
be the wage rate of the native worker before migration and
be the wage rate of the native worker after migration.
Here we're assuming that

Similiarly let be the wage rate of the immigrant worker before migration (in Mexico) and be the wage rate of the immigrant worker after migration (in US).

Of course


Let utility be a CES function of wage/income:



Then welfare before migration is given by:




and after migration:



Setting these equal and solving for lambda tells us how much you have to weight the well being of native workers in order to be indifferent between migration and no migration. If your lambda is less than this value (and that's up to you) then you should support the migration, if your lambda is greater then go ahead an oppose the migration. We get:



To calculate the "jerk factor" lambda we have to get some estimates of how much wages change for both migrants and natives.

Let's get crazy and accept the Borjas results, which say that in the long run the migration will depress a native unskilled worker's wages by 5% (note this isn't exact translation of the Borjas findings but close enough). Of course these kinds of estimates are in the aggregate but it makes things simpler here (and doesn't do much injustice to reality) to assume that there's some one particular native worker who bears the full brunt of moving a Mexican worker into US.

In addition, let's use the findings from here that a migrant's wages increase from $2.56 to $9.34 after migration. In fact to make things simple, suppose that once the Mexican worker migrates, the wages of the unskilled native are equal to $9.34 (this assumes there's no premium for being a native, for speaking English, etc. Again, this makes my job harder, not easier). So



which simplifies the above lambda equation to:




Finally, if the migration decreases the natives' wage rate by 5% this means that originally his wage rate was (1.05)*9.34=9.81. So



This still depends on theta which measures the extent of diminishing marginal utility of income. There's various estimates for this. Theta=1 is nice and easy to use because then the utlity function is logarithmic, simplifying many things. On the other hand some macro studies and the equity premium suggest much higher values for theta. At any rate, here's a table of lambda against various values of theta and the implications for the jerk factor - how much more you have to weight natives' welfare than that of potential migrants (click to enlarge):



So, for example, with the implausibly low value of theta=1/2 one would have to consider the welfare of the native worker to count about 20 as much as that of the potential migrant. With logarithmic utility (theta=1) each native worker counts about 26 and a half times as much as a migrant. With a (what I consider more plausible) theta=2 you need 55 and some migrants to make up one native.

Here's the same in a graph:


Clearly one doesn't need to be a rootless cosmopolitan to reject these kinds of weights. One only need not be a jerk.








-- What about the immigrants crowding our schools? Using our health care system? Living on welfare? What about crime? Well. First, most of that is bunk. Most studies which look into the amount of tax money that (illegal) immigrants put into the collective kitty find that it's much more than what they take out in terms of benefits - one obvious reason is that illegal immigrants usually pay a lot in payroll taxes but never collect Social Security or medicare. Likewise the statistics for crime are ambigous to say the least. Illegal immigrants, by and large, are afraid of getting deported which means that they try to stay out of trouble as much as they can. However, even if these things were true in essence this would be a change in the post-migration wage rate (adjusted for externalities) of the native. Given the change in the wages of the migrant, this is not going to affect the results that much.

-- This post has already been edited for civility. What incivility remains pretty much belongs here.

The Economics of The Wire

*Possible spoilers below*

I'm really surprised that no one's has mentioned the Economics of the HBO show "The Wire" (though there have been discussions as to whether a show embraces a libertarian or a socialist ethos). The series is bursting with economic concepts, and it mostly gets them right which I guess is somewhat surprising. First, because it's TV and TV usually does not get that stuff right, and second because of the political leanings of its creator. But nope, they get it pretty much spot on.

In fact, you could probably base an entire course in Industrial Organization on episodes of the Wire. Stringer Bell is actually taking an economics class in the show, in order to figure out how to boost the profits of his drug organization (gets an A- on his paper).

Like I said, there's a lot there but the part which I find really interesting is the various forms of competition that do or can take place between the various drug gangs.

First you got the obvious competition over turf - who controls the corners where drugs are sold. This immediately screams "Hotelling location model!". While this kind of competition can be, and occasionally does become very deadly, most of the time on the show there's a sort of a semi-stable equilibrium. Proposition Joe controls the drug corners of East Baltimore and the Barksdale Crew controls West Baltimore, and aside for the annual basketball game they stay out of each other's way, like two firms in a linear city which move to the corners. There are two 'shocks' which upset this balance. One is the entry of the third firm, Marlo. The other is the cooperative solution reached by Prop Joe and Stringer Bell which involves a trade off of turf for quality product.

Which brings us to the second type of competition which is in quality of the product. Prop Joe has the quality drugs which beat out that of his rivals. After Avon's arrest the Barksdales are hit with disruptions in their supply chain and are forced to put out very diluted drugs. At this point the difference becomes great enough so that "there's a steady flow of fiends from West to East". The difference in the price/quality ratios is greater than the transportation cost for the consumers, the drug fiends, and Prop Joe ends up 'stealing' most of Barksdales' demand. Stringer Bell responds with some ad hoc solutions which rely on the bounded rationality of his customers - he re-brands low quality drugs under a new name - but he's well aware that this is only a temporary measure. Stringer knows that this trick can only work for awhile and that ultimately even drug fiends are rational consumers (the other trick is to engage in some 'make-pretend' competition among different brands which are both controlled by the same supplier). As a result he is forced to share some of Barksdales best turf and location with East Baltimore.

And this is where the rare instance of true price competition appears. Like firms in real world, the drug organizations seem to try to avoid price competition at all costs. They would rather shoot it out over who owns what corner than have two drug crews under cutting each other on price. In a way it makes sense. When Stringer forces Bodie to let some of Prop Joe's people into the towers, the rival crews start playing Bertrand very quickly. There is a scene where Bubbles, who used to buy from Bodie, switches to Cheese's drug due to lower prices. Bodie runs up and offers Bubbles a buy one get one free deal to which Cheese responds with a price cut of his own. One can think of this situation as an instance in the Hotelling game where the two firms are located too close together and as a result price competition completely erodes the profit margins.

There is of course a lot more. The double marginalization problem of a downstream and upstream monopolists. Brands as imperfect signals of quality. Barriers to entry, pecuniary and lethal. Contestable markets. The stock of reputation as value in itself (Omar!). The incentives that police officers face ("That's what you get for giving a fuck when it's not your turn to give a fuck"). Rational addiction. Political economy and public choice theory. Outsourcing of certain tasks to better suited outsiders (obviously Brother Muzone took some local Baltimore enforcer's job!). Asymmetric information. The optimal time to defect in a repeated game (why did Avon and Stringer set up each other when they did? And why both at the same time?). The sustainability of a cooperative/collusive arrangement (Prop Joe's drug Co-op). The economics and politics of Unions (including that of labor substituting technological change). Law and economics. And of course elasticities:
"What you're thinking is that we have an inelastic product here. But what we have here is an elastic product"

How much inequality would Plato tolerate?

This is a version of a problem I gave to my students on their last midterm.

Part I
(Ha! I'm like the television series. Endless suspense with no resolution.)

UPDATE: Gabriel has some really nice graphs on this. And no, I'm not a Platonist. If anything, I'm more of an Epicurean.

In his Laws Plato states that no man should own more than five times the land of any other man. Of course in this statement Plato is pretty much ignoring the incomes/wages of common laborers and slaves. But let's do some quick, rough exercises to see how much income inequality, in terms of the Gini coefficient, this can generate. Are the economies of the Platonic ideal comparable to present day economies? Was the ancient world more equal to the one today, albeit poor? Or was it just a bunch of wanna-be King Arthurs opressing the autonoumous collectives with the violence inherent in the system? What kind of assumptions matter?

Since we're talking about the pre-modern world let's make the Classical assumption and do the analysis in terms of "classes" of people. We can change the number of classes and the proportion of each in population to see how inequality measure varies. Part of the point of the exercise is that the Gini coefficient hides a lot of information about the income distribution.

First, let's suppose that there's just two classes in the economy. The income of a lower class person is Y_n and that of a higher class person is Y_x. In fact let



Let the proportion of lower class people in the economy be p.

The Lorenz curve and the Gini in this case are illustrated below:



Then a quick calculation ("quick calculation" means that I'm too lazy to try to write the math in blogger's version of Latex. Nothing more nor less. See the misuse of the word "trivial" in mathematics) of the Gini index of inequality gives us:




Obviously if every body's poor then p=1 and the Gini is 0 and likewise, if every body's rich then p=0 and Gini is also 0 since we have perfect equality in both cases.

Since we're interested in the maximum amount of inequality that Plato would tolerate we can maximize this with respect to d, the fraction of rich folks' income that the poor receive. After some messy algebra and the like we get




Then we plug that back into our Gini and we get (after some more algebra):



Now here's where subtleties of "What did the Master really mean?" come into play. If Plato simply meant that no person should have more than 5 times the income than another person than p is simply 1/5. Plugging that into the G_max gives us a Gini of about .38. This is somewhere between the present day Gini for Ireland and Israel. In other words Plato would be mostly ok with the kind of inequality we find in the world today.

However, I'm pretty sure that the richest person in Ireland or Israel today has way way way way more than five times the income of the poorest person in those countries. This is another way of saying that a lot of the action is in the middle of the income distribution. To put it yet another way, beware of inequality measure which compare the top 1% to the bottom 1%, or the top decile to the bottom quintile, or the top deci-trie-ile to the bottom quatri-platy-pus-ile etc.

But wait! What about diminishing returns to land, which are a standard assumption of the pre-modern, industrial economic analysis? Under some (essentially, wrong) assumptions this means that if the ratio of land ownership is




then




where a measures the extent of diminishing returns to land. If you got yourself an aggregate production function for this pre-modern economy then this a will actually correspond to the income share of land in total income. As it happens we do have some measures of this variable. Hansen and Prescott use a =.25 in their calibration of a growth model. More generally Clark looks at the share of land rents in output and finds it at between .25 in medieval England to somewhere a bit below .5 in parts of 17th century China. The Chinese data actually probably overstate a because they're derived from farm-level data (this is important for what follows below), so let's say .4 as an upper bound. In this case the max-income to the min-income ratio would be between 1.5 and 1.9. Plugging these into our G-max we get that it is between .1 and .15 - a low level of inequality not found in today's world! Could it be that Plato, or the world he was speaking of, was a radical (authoritarian) egalitarian?

Well, no. Like I said, the above has some very wrong implicit assumptions. What it really assumes is that each landowner works her own land and has the same level of productivity and uses no laborers or slaves. In other words, this is a model of just one class and of inequality variation within that class. What we need to do is to put the peasants - a separate class, back into the analysis.

But this means 3 classes at which point the analysis gets more complicated. We could assume however that each landlord doubles as both a worker and a land owner. Making another extreme assumption that the market for labor is competitive...

(I'd actually guess that this was more true in Plato's time than in medieval Europe - Greek system of slavery essentially amounted to a slave paying a fixed payment to his owner and then for all intents and purposes pretty much being left alone - and even that kept getting reformed with slaves' and poor people's debts being cancelled - anyway this essentially worked like a lump sum tax, a pure transfer not affecting any marginal conditions. On the other hand the feudal system extracted a PERCENTAGE of output hence affected marginal conditions and incentives driving a wedge between marginal products and wages. I vaguely recall reading Joan Robinson somewhere making the point that essentially this should be the basis for a theory of exploitation rather than the whole Marxist "surplus labor", "labor theory of value" approach.)

... so that we can keep the analysis contained to one class but allowing for labor input. Welp, in that case land owners live off their land rents but they also hire labor to work the land (some of the labor could be their own). But this means they will hire labor to the point where the marginal product of labor equals the wage rate. Given a total supply of the fixed input labor this means that a landlord's rents will be strictly proportional to the amount of labor she owns. If there are constant returns to scale in both factors - land and labor - then this means that a poor landlord's income is the same fraction of a rich landlord's income as his land is. Another words, we're back to



and the .38 Gini coefficient we got above.

Some notes:

No, I don't really torture my students like this. The problem I gave'em was a lot more stylized and straight forward. A good number of them picked up on the basic point.


The (see below) that was above and now that you see below essentially depends on the market structure of the economy. The assumption that each landowner only works his own land and hires no labor means that essentially there is no labor market. However if there is SOME kind of a labor marker (and it need not be competitive) then this means that the land ratios of the landlords will equalize (in imperfect markets approximate) the income ratios. This is how we go from "ratio of incomes" = "some root of land ownerships" to "ratio of incomes" = "ratio of land ownership"

The part II will contain the analysis with three classes - these could be {peasants, small landowners, large landowners}, or if you wanna compare pre-modern world inequality with modern inequality then {peasants, capitalists, landowners} - the point of the latter being that inequality actually decreased with the Industrial Revolution. Even though "capitalists" gained, "workers-used-to-be-peasants" gained much more. The folks who got really screwed by the technological changes of the IR were the aristocratic landowners. Which gives you hope for this world. Karma and all that. And even honest Marxist oughta sympathize.

My spell checker is homotheticophobic

This post from Crooked Timber on the danger of rellying to heavily on the spell checker (aside from being a horrible grammarian, I'm a horrible speller as well, if anyone noticed) is a bit outdated (things move fast in the blogosphere) but my last post, where I mention "homothetic preferances" made me think of it. Basically I gots lots of friends from grad school who aren't native English speaker and as a consequence I've done a lot of proof readings of their economics papers. Some of the papers I've proof read would talk a lot about "homoerotic preferances", and I always wondered how that happened. Now I know. "Hamitic preferances" and "hermetic preferences" just isn't as funny.

Yes, "preferances" should be spelled that way. When in doubt, blame the crazy structure of the English language which seems never able to make up it's mind. Actually blame the Battle of Hastings. I've always felt sorry for poor ol' Harold Godwinson. Harold Hardrada too.

The spell checker also doesn't know what to do with White-corrected standard errors.

You know Rawls'd be all about taxin'im some height!

He'd be all over it!


This is part I. Part II is coming soon (I'm pulling a Robert here).

Here I just wanna show with a simple example that the Mankiw and Weinzierl results on optimal height tax apply to the Rawlsian social welfare function, in fact, to any egaliterian social welfare function, in fact^2, to any social function (the last part of the argument involves some hand waving).

First let's see what exactly is going on in the Mirrlees/Vickrey framework that Mankiw and Weinzierl use - i.e. open up the hood and see how the engine works. I'm going to keep things as simple as I can (and it's still messy). The basic point is that what drives the result is not the choice of the social welfare function but the incentive compatibility constraints, or in other words, the Revelation Principle.

Suppose, for simplicity, that there's only two people in the economy (they could be "types"), high-ability and low-ability. High ability folks receive high wages, wH, and low ability folks receive low wages, wL. Let's pick these equal to wL=1 and wH=2. Also for simplicity suppose there's only two labor effort levels a person can choose LL=1 or LH=2. So income is w(i)L(i) which is:
LL: 1 for a low ability person choosing low labor
LH: 2 for a low ability person choosing high labor
HL: 2 for a high ability person choosing low labor
HH: 4 for a high ability person choosing high labor.

Let's make utility really simple too:




Note that this implies the following utility levels without transfers:
LL: 1*1-(1/2)*1^2=1/2
LH: 1*2-(1/2)*(2^2)=2-2=0
so a low ability person would choose low labor effort
HL: 1*2-(1/2)*(1^2)=1/4
HH: 2*2-(1/2)*(2^2)=2
so a hi ability person would choose hi labor effort.

The social planner wishes to maximize some function of the two person's utility levels: SWF(u(1),u(2)). The problem is that she does not observe w(i) (which is correlated with ability) only labor effort L(i) and perhaps some other characteristic like height that is also correlated with w(i). First let's just look at the situation where height is either unobservable or known to be uncorrelated with w(i), hence taxes and transfers - the instruments at the social planner's disposal - can only be conditional on observed labor effort.

The key are the "incentive compatibility" constraints. Basically, the Revelation Principle says that in the constrained-optimum the principal (in this case the Social Planner) can implement, all agents have an incentive to truthfully reveal their "type" - whether they're high or low ability. In other words, to do the best she can do, the social planner should pick tax/transfers so that no low ability agent would pretend to be high ability (by picking L(i)=2) and no high ability agent would pretend to be of low ability (by picking L(i)=1).

Letting TL and TH be the tax/transfers, this implies the following IC constraints:

u(low ability with low labor effort) + TL > u(low ability with hi labor effort) + TH
u(hi ability with hi labor effort) + TH > u(hi ability with low labor effort) + TL



or
1/2+TL > TH or TL > TH - 1/2 (call this IC 2)
and


or
2+TH > 3/2+TL (call this IC 1)

We also have the budget constraint that says that TL+TH=0

Here's a picture of the set of {TL,TH} which satisfies the IC constraints and the budget constraint. The thick red line represents the feasible tax rates:



Which point would the planner choose? Well, it depends on the exact form of the social welfare function, or in other words, her preferances. An egaliterian planner would like to choose TL and TH so as to make the two utilities equal:



or

solving for TL as function of TH we get the line of utility equality:


But note that this line lies OUTSIDE the feasible set:




What can the planner do? Well, she's gonna try to reach the highest indifference curve she can given all them constraints. In fact that line of total equality is also the "Engel curve" - or the expansion path - for the planners' indifference curves if they're homothetic (it don't change nothing if they ain't, I believe). So the egaliterian social planner will pick the most egaliterian allocation available, given that it has to be one where no one wants to pretend they're someone else. The full graph, in all its glory, shows it all. This one's for the kids:



The "45 degree" lines represent the preferences of a Rawlsian social planner. The more curved ones represent an egaliterian social planner that isn't quite Rawlsian. In this particular case, the social planner wants to implement the tax/transfers {3/4,-3/4}. But this would violate the IC constraints and hi ability people would rather supply low labor effort, pretend to be low ability and get the 3/4 subsidy. So the best the planner can do is to implement the still-egaliterian-but-less-so tax/transfers of {1/4,-1/4}.

What is this concept called "free trade" and why should I care?

Dani Rodrik links and agrees that "little gain from free trade" means "little pain from free trade" (in terms of dislocation costs, losses to the loosers, etc.). Basically if the Dead Weight Loss/Harberger triangles are small, then so is the change in producer surplus (unless supply is way way way more elastic than demand). Cool. I have a few more thoughts though, too long to post as a comment.

First, I'm actually of the opinion that a lot of things that worries the public in the developed countries, and, to a significant extent, economists as well, doesn't matter that much. The welfare gains from free trade are probably small. So are the welfare gains from decreasing the volatility of output, of lowering inflation another percentage point, of cutting that deficit a bit, of increasing savings somewhat, of regulating monopolies better, increases in inequality don't matter that much (if at all), poverty rates have been constant (and of course the poverty line in US is well above the average world income, even adjusted for purchasing power). Minimum wages hardly matter. Of course, all those things together could add up, so you can still defend the principles with a clear conscience but... when you take the global perspective, the conclusion "Developed countries: problem essentially solved. Let's move on" invariably comes to mind.

Having said that, I'd like to note a few things:
1. Most protectionist measures these days aren't tariffs, but rather "non-tariff barriers" such as quotas, subsidies, VERs, and anti-dumping duties. As tariffs have fallen since GATT's inception, other forms of protection have risen (basically the other forms of protection are harder to detect/remove/get rid off - there's a literature on the political economy of this phenomenon that explains it, but I'm unfamiliar). And these other forms of protection, except for subsidies, are far more distortionary than tariffs. For example, quotas have essentially the same effect as tariffs, except the home country gets no revenue. Including these in the welfare/pain calculation would in all likelihood increase the numbers significantly. I don't know if you'd get the 3.5% of GDP from the study Dani quoted, but I'm sure it's much much more than the .25% you get with the simple formula.

2. What's small to us, in a developed country, may be big for folks in other, much poorer, countries. If you take the global perspective as Alex Tabarrok advocates, you can still argue for less protection from the US even though for us it doesn't matter much. (As an aside, looking at it from this perspective, I don't have much of a problem with subsidies, European or American, agriculture or whatever, since these hurt the home country but help the poor foreign country). Even if you think that foreigners shouldn't count as much as natives, assigning them the weight of 0 in the social welfare function is just plain nasty. And it probably doesn't take much of a relative weight to get those benefits to matter significantly.

3. There's of course other arguments for trade. Gabriel's instrumentalist argument - trade restrictions impinging on personal freedom (of the natives!). Possibility of increasing returns. The gain from variety. Technology transfer (though obviously I doubt this applies to North South trade for the North). Etc.

Finally,

4. As I've said before, you really don't have to work hard to convince non-economists that trade has distributional effects. They believe it. They think these are huge! They think trade is always and everywhere a race to the bottom! Just look at the commentators on all the main blogs that discuss trade (and gold stars to Mark Thoma for kickin' some butt on free trade in his posts and his own comments recently). Why exert resources on something that most people are already (blindingly) convinced off. If our job is to educate then bias-correction is surely a part of that. And the bias runs the other way. It's nothing about barbarians... well, actually it is. But just because there are barbarians on both sides of the issue, the symmetry is only qualitative and not quantitative.

A conjecture

This was prompted by Karl Smith's (what I think is mistaken) objection to the Mankiw and Weinzierl Optimal Height Tax, which I came out in favor off earlier.

Conjecture:
Mankiw and Weinzierl results generalize to any social welfare function which satisfies the Dalton-Pigou principle (basically places value on equality of incomes). It also generalizes to a Rawlsian social welfare function in expectations (might have to assume risk aversion on the part of the social planner, I'm just conjecturin' here. Problem here is that Rawls is a 'corner-case').

In other words, if you think the M&W results are enough to throw utilitarianism out the window, then they're enough to throw any kind of social welfare function which places value on equality out the window. Utilitarianism is just easy to pick on.

Of course some people *cough, Gabriel*, would be happy with that.

There you go. 20$ on the sidewalk, kids. Of course I could be wrong. I'll check once I'm done grading midterms.


An update, or thinking about it a bit more:
Actually under the Rawlsian criteria the redistribution should become even more extreme. So the tall people should really suffer, though the short rich might too. It would be a classic example of "soak the tall" policy. Basically there you want to get the {short, poor} people (in expectations, the worst off), up to the level of the second-worse off. These would be either the {tall, poor} or the {short, rich}, depending on how labor supply elasticities and diminishing returns to consumption play out. Either way, you basically wanna take enough money away from the {tall, rich} (the ones most likely to be best off), and give it to {short, poor} to get them up to that level.

And of course this whole subject matter is just begging for this. (Take it up with Randy)

Update on the update, or now I speak heresy:
The M&W result is essentially trivial. Its interest lies in the fact that they picked an instrument, height, which makes the result seem controversial and, to some, counterintuitive. In fact, for ANY social welfare function, egaliterian or not - that is, as long as there is some complete, transitive, ranking of social outcomes - having more information is better than having less when trying to maximize that particular function. For example, suppose there's only two possible effort levels (HI and LO) and two possible intrinsic ability levels (hi and lo). If the taxes and transfers are limited to being based solely on observed incomes then there's only four possible tax/transfers (y-very lo, y-lo, y-sorta hi, y-hi) . However with height (ta' or sho') it's possible to have eight different possible tax/transfers (y-very lo and sho', y-very lo and ta', y-lo and sho', y-lo and ta', y-sorta hi and sho'... etc). Or in other words, the social planner always has the option of ignoring height and just taxing income. Hence, including height cannot lead her to implement an allocation which makes society (i.e. a complete transitive ordering) stricly worse off then ignoring it. The fact that she can do this is actually the result of the Revelation Principle, not any particular social welfare function that is being chosen, Utiliterian, Rawlsian, or whatever. The subtlety is in how the incentive-compatibility constraints change here, but I'm pretty sure everything goes through. So I think my original conjecture was wrong. It was too modest.
Of course this is still conjecturin'^2, so I could be wrong.

Assumptions

Robert has a post on assumptions in economics which, despite the title is not irrelevant. He starts of by saying:

You may sometimes read (neoclassical) "Economic theory shows" p.

I know of no academic work that makes these kinds of statements (and if it does exist it should be rightly condemned). The usual phrase is something like "OUR results SUGGEST" p. Of course the phrase "Economic theory shows p" is sometimes, perhaps often, used in Internet polemics or what have you, usually by people who have less than half a clue of what they're talking about. Economic theory, broadly taken, shows or suggests many different conclusions, some of them in opposition to each other. As Robert points out it's all in the assumptions.

(which, as an aside, why I'm more in favor of proofs rather than numerical examples. Using numerical examples is like doing a magic trick. Using a proof is like explaining the inner workings of the trick.)

But his post has this ... implicit ... tone ... suggestion - that some people (perhaps those that are consistently placed in parentheses) choose their assumptions based on the results that these will generate, and hence end up with unrealistic assumptions. At least that's the implied accusation in Keen, if I remember correctly. But that's ad hominem foolishness. We'd all like to choose more realistic assumptions, properly speaking. The thing is that there is a trade off between realism and other aspects of modeling. More so, one is usually forced to choose between different sets of unrealistic assumptions rather than between more and less realistic ones.

The usual example I use here is that of adaptive vs. rational expectations.

Adaptive expectations mean that you expect some variable x to be in the future what it is today. I.e. Ex(t+1)=x(t). This is a very simple forecasting rule, and one which will lead you to make consistent mistakes. It's good for forecasting the sunrise, maybe even the stock market, but horrible at forecasting anything that depends in semi-predictable ways on other variables that you observe. Like inflation. So assuming adaptive expectations in your model is equivalent to assuming that people are really really really stupid, even when it comes to their pocketbook. This is clearly an unrealistic assumption.

On the other hand assuming Rational expectations means that you're assuming that the people in your model know the model (what Gabriel calls "model-consistent expectations") and use the model to form their predictions. As a result while they make random mistakes in their forecasts they make no systematic mistakes. It also means that people use all available information in their forecasts and that this information is sufficient to make a good enough forecasts (the world is non-chaotic, otherwise you can't make accurate forecasts anyway, almost by definition). Basically Rational expectations assumes very sophisticated behavior on the part of the agents (plus some stuff) and their knowledge of the relevant parameters. Clearly another unrealistic assumption.

The problem is, if you're gonna model expectations, anything in between these two extremes is very very hard, if not impossible to model. There is a literature on learning and bounded rationality which is somewhat relevant and gets at some of the difficulty but usually it involves some unrealistic assumptions of its own, is irrelevant for the topic in question, or you get inconclusive results, or, in the best case scenario, it's just a royal pain to solve for anything.

Basically, in economic modelling, there's a resource constraint with regard to assumptions, realism and practicability and suggesting that "the assumptions should be more realistic" only makes sense if a more realistic assumption is in the feasible set. Otherwise it's like telling a poor person that she'd be better of if she drove a Ferrari. Of course, technological progress, intellectual discovery and innovation in thought pushes the relevant possibilities frontier outward and overtime hopefully allows us to make both more realistic and more practical assumptions in our models.

In another comment Robert calls my posts/comments "quirky". That's fine, I guess I plead guilty. Personally I find some of his post/comments to be hard-to-understand-on-purpose (there's a fancy word for that) and pretend-I-don't-understand-what-you-mean (there's a fancy word for that too). I consider these, like quirkiness, to be venial rather than mortal sins. And you need some venial sinnin' for life to be fun.

Hmm, maybe for my next birthday or Xmas

No, not the book review, but the actual three volumes of "Capital Theory" edited by Bliss, Cohen and Hartcourt, retailing at 395 pounds, or 675 dollars

(implying an exchange rate of 1.71 dollars per pound, as compared to the roughly 2 bucks per pound of official exchange rate. More weirdness: one price for a used copy on Amazon is 73$ - maybe that's just for one volume (still, not bad for 1/3 of total) - and then ranging up to 813$, a significant mark up over Amazon's own price. Where's the arbitrage?)

Anyway, the book review is really good too.

Here's some choice excerpts from the review:

Blaug’s ‘Final judgement’ [1975] accuses the post-Keynesians of having a theory without measurement, as “they are totally incapable of producing testable predictions” (p. 203), and of adopting a method of analysis deliberately framed to deny simultaneous determination of economic variables. (Footnote: See Petri 1999 for a thorough discussion.) In Blaug’s view, neoclassical overwhelming superiority lies in its flexibility and manipulability as well as from the fact that one cannot get along from the concepts of maximisation, equilibrium and substitution that are at the roots of mainstream economics.

In another words
effects

and:
For the neoclassicals – who were obliged to admit the possibility of such an event – reswitching remains a mathematical possibility (a curiosum) with no relevance for real economies...post-Keynesians, on the contrary... maintain – (that reswitching) is a theoretical problem, not an empirical one, and the mere fact of its logical possibility destroys the explanation of income distribution in terms of marginal productivities and scarcity of factors.

and finally, a full paragraph:
While understanding the weight of such an argument, I remain sensitive to the empirical side of the scientific method. Now, on the question at stake, we have two kinds of evidence. The first one consists of computing the probability that reswitching actually happens on the basis of the Monte Carlo method The result was that the likelihood is rather small but positive. The other kind of evidence is strictly empirical: the wage-profit rates curve (w-r curve) are estimated for real economies (Ochoa 1989; Da Silva 1991; Han and Schefold 2006). Ochoa and Da Silva calculated the w-r curves respectively for the US and Brazil, obtaining a surprisingly quasi-linear relation. (D’Ippolito 1989). (Footnote: The quasi-linear relation is surprising because the mathematical form of the profit-wage relation is a polynomial of very high degree (the degree depending on the number of industries of the input-output tables considered), which in principle should show a rather irregular shape.) This could suggest that, in practice, reswitching does not occur because straight lines cannot intercept more than once. However, the quasi-linearity of these w-r relations is not really proof that reswitching is unlikely, because the correct experiment consists in comparing the w-r curves that would result from the alternative techniques available at a given moment (“the book of blueprints”), and not just by considering the w-r curve resulting from the choice that has been made for a given year. Han and Schefold overcome this objection estimating the w-r envelopes for 32 input-output tables for nine OECD countries for 1979-1986. Reswitching appeared in one case, and in 3.65% of tested cases there was a “perverse” substitution of labour, i.e. capital-labour substitution that contradicts the neoclassical parable.

Let's round that last number up and concede that in 4% of cases raising the minimum wage might increase employment.