Follow the money

To absolutely nobody’s surprise, towns in Massachusetts with higher median household income do significantly better than towns with lower median household income. You can see a nice little scatter plot here for HH income and percent proficiency (average of math, English and science for a school district).

Other than the hard cap at 100 (no matter how well funded a school is, it’s not going to have more than 100 percent of its students passing); its a really damn good correlation.

Proficient Scatterplot


Here’s the same data but with advanced:


Advanced Scatterplot

It’s even more clear (also, the hard cap doesn’t really come into effect). A full 65% of the variance in advanced test scores can be explained by median household income.

So whats going on here? Well, the obvious idea is that these schools have more funding than poor school districts; this leads to better schools. Is that true?

In Massachusetts, the school system with the least per pupil expenditures is East Bridgewater. Yet they are above average in the MCAS (the state testing regime) for proficiency in English, Math and Science. The highest is Cambridge, yet they are below average in English and Science. Well, those are just anecdotes; so lets talk data:

I ran regressions for percent of students who are proficient in English, Math and Science and who are advanced in English Math and Science, all against per pupil expenditures.

The result, of the six regressions, none of the test scores were positively and significantly correlated with expenditures. Two, (percent of students who are advanced at English and percent of students advanced at math), were negatively correlated at p < 1%. Percent of Students advanced in Science was negatively correlated with expenditures at p <10%. The other three were not significant, but all the signs were negative.

Here’s the scatterplot for  for the percent of students advanced at English (with a trend line):

Eng Adv Scatterplot

The answer is no, it is not explained by (only) funding. If anything, funding is negatively correlated with test scores (ok, this is a cheat, I’m doing two single-linear regression at the same time and comparing them which is kind of a statistical bad thing. I did run a some multiple linear regressions, which don’t change much, although the the sign on funding flips to positive but not significant for advanced; the sign for advanced is still negative but also not significant).

This is weird at first glance it appears that rich towns have better schools, but that this has nothing to do with funding.

One theory is reverse causation; that is some school systems are randomly good, which attracts well off people. After all, school systems are probably the number one concern for for families in choosing which town to live in. There’s a ton of truth in this, I imagine. But i highly doubt its the whole story.

The most likely theory is that, for whatever reason, children of well off families are more likely to do well then children of less well off families. Whether that thing is due to culture, genetics, or other economic reasons (better access to healthcare as an example), I don’t know, but I think this is very important.
So here’s the aside where I’ll talk about how I wanted to finish this. I wanted to make points about culture, about moral reasoning, about economic aid and welfare. And I’m realizing that I’m just not able to do those topics justice, at least right here. So I’ll close with one thought.

The above analysis should scare you. Because, basically, the rich are getting richer. And its not because we just poor resources in to the rich districts and starve resources from poor districts. If it was we’d almost have reason to celebrate. All we’d have to do to fix schools is just pass a bill and move money around; but its just not that simple.


(all stats regarding education from


Housing Crisis

Lets imagine 4 groups of people: responsible rich people, responsible poor people, irresponsible rich people, irresponsible poor people.

Everybody wants a house. Banks won’t lend to poor people regardless of how responsible they are, but they’ll lend to any rich people, regardless of whether they are responsible or not (and the rich will not default even if they’re irresponsible).

The government sees many advantages to home ownership, so it decides to get responsible poor people to buy houses.

Since banks won’t lend to them, the government makes it much easier for poor people to borrow.

Who takes advantage of this? Responsible poor people make a decision about how they can pay off debts, and some of them are helped by the lower lending standards, so some portion of responsible poor take out loans. The irresponsible poor take advantage of the program no matter what, meaning they buy lots of houses.  This drives up the price of housing a lot. After a short time, housing prices rise so much that the responsible poor cannot afford housing anymore, so they stop buying houses. The irresponsible poor can’t afford it either but buy anyway. After a time, the irresponsible poor start to default on their loans, have their houses foreclosed, housing collapsed, yada yada yada, 2008 financial crisis.

A few things about this model. First, obviously there was a lot more going on in the 2008 financial crisis than this simple model. Second, it doesn’t say anything about whether the rich are more or less responsible than the poor; only that they have more money. There needs to be a certain percent of the poor who are irresponsible (or at least bad with money); I don’t know what that percent needs to be or what it was, but as long as there is a high enough percent of irresponsible people in your target group, a program like this is bound to fail.

Way back in the Bush years, there was a lot of discussion about the benefits of an “ownership society.” Basically, they saw that people who owned homes were much better off than those who didn’t. Obviously this screams correlation is not causation, but lets ignore that for a second. Let’s assume that home ownership really does magically make people more responsible. Does my model still work?

Yes. Even if, once a homeowner moves in, he moves from irresponsible poor to responsible poor, its already too late. He’s already made the decision and cannot easily back out.

Again, in our little model here, we basically hurt everyone, the rich have to pay more for houses, so they’re hurt a little bit; the responsible poor are probably effected the least, they couldn’t afford housing before and can’t afford it now, and the irresponsible poor now have debt and bankruptcy and foreclosure to think about. Also, the country is thrown into a recession; which is good for no-one. Its also pretty clear that the poor are hurt more than the rich are by this.

All of this is a really long winded way of saying that programs can backfire. Over my lifetime, the gap between the rich and the poor has widened. And I think that most or all of the programs we have designed to fix it do more harm than good.  I’m going to attempt to relaunch this blog while tackling this phenomena.

Bad Arguments vs Bad Faith Arguments

There is a bit of an argument between Scott Sumner and Jason Smith, its a couple months old because I saw it and filed it away somewhere to be used later because I think it illustrates a point. The post is an argument about the sequestration and macroeconomics; read it if you want to.

I’m not really going to comment on the post itself, but on a single sentence of it, which reads “Now that Sumner is on his way to Mercatus, I can only assume it will get worse.”

Mercatus, for those of you who don’t know, is “world’s premier university source for market-oriented ideas—bridging the gap between academic ideas and real-world problems.” Its a libertarian/market oriented think tank located at George Mason University, and is funded in part by none other than the Koch brothers.

So, if I parse the argument from Jason Smith correctly, its that Scott Sumner is dishonest in his opinions, but going to the Koch funded Mercatus center will only make it worse; which on a quick read kind of makes sense, but if you examine it closely, it really doesn’t.

A bad argument is one that is wrong. If I believe that the moon is a hologram and argued so because stuff like this happens, then I would be making a bad argument. If I didn’t believe that the moon is a hologram, but argued that it was anyway, I’d be making a bad faith argument. Bad arguments can be harmful, but the problem is that its impossible to tell a bad argument from a good argument until after you’ve examined it. They’re a necessary evil, after all, many “bad” arguments can turn into good ones after years of examination.

A bad faith argument is different, and they can be more harmful. The Koch brothers, for instance, own large interests in logging companies, one can make a case that they would therefore profit from denying global warming, and that the profit motive, not their actual belief, is partially driving the debate.

Now, why might this be a bigger problem than other arguments? Countering a bad faith argument with evidence will never work, as he was is arguing in bad faith will only ever change when the underlying incentives for arguing in bad faith change (in the global warming case above, the profit motive, but there can be other causes). Whereas if people were simply arguing because of what they believe, then as soon as the weight of evidence changes, people will change their mind. (ding ding ding ding! most naive thing I’ve ever written on this blog award goes to… the previous sentence!). The one arguing badly in good faith can still present ideas which may be missed otherwise or prove a testing ground for good theories, while the arguer in bad faith will twist arguments and attempt to confuse rather than illuminate ideas.

Looking at the original statement in this lens, its pretty clear that Jason Smith is arguing that Scott Sumner will be arguing in bad faith (if he isn’t already). That because not only his opinions but also his economic well being depend on agreeing with market based though, he will compromise himself and his scholarship.

But this can’t be right; if it were it’d essentially be saying that studying with people you argee with or taking a job to advocate a position is inherently bad faith. A logical extension would mean that any environmental activist would be arguing in bad faith so long as they worked for a environmental organization, which in effect would by saying that we can’t trust the sierra club when the speak about preserving the environment, because they only care about preserving the environment. The same principle holds true for the Mercatus center, should we not trust them with regards to market based solutions, because they only care about market based solutions?

One could argue that its the Koch influence which is the problem, and that Mercatus is only a mouth-piece for Koch industries. (I doubt that it is, but at least that’s a plausible story). This may hold true for Mercatus issues in general, but not in macroeconomics. The Kochs may lose if people decide to stop cutting down trees, but nobody wins because of a recession. If the Kochs are backing Scott Sumner for their own selfish reasons, well that just means they think NGDP targeting will prevent recessions/cause growth, and if that’s true that everybody should be for it. If its false, everybody, including Koch brothers, should be against it.

Of course there are differences in opinion on controversial issues, (that’s pretty much a tautology if I’ve ever seen one), and maybe the Kochs are right about NGDP targeting (or maybe they just fell that the Mercatus center is doing good work in general and are happy to help fund it), maybe they’re wrong. But regardless, so long as neither they nor Dr. Sumner are arguing because of a hidden incentive then they’re simply part of the debate. We can’t commit to evaluating only good arguments, because we can’t tell if an argument is good or not until after we evaluate it.

Why I’m not a Keynesian

After world war II, the decline in government spending was enormous. Government spending as a percent of GDP went from over 45% to around 15%, which threw the US into a terrible recession, where GDP fell by over 12% in 8 short months. To put this in perspective, the “Great Recession” which lasted from 2007 to 2009, and which was the second worse recessions in the drop of GDP since WWII, only saw GDP decline by 4.3%.

This recession saw unemployment skyrocket to 5.2%. The average person was hurt by this, in 1946, while the GDP in 1946 fell by 11.6%, personal consumption grew by a paltry 12.4%, to put that in perspective it was the 85th worse year since 1930 (source:

So what actually happened? We see that the decline in government defense spending after World War II caused the measured GDP to fall a lot, but what this meant is that the government stopped building tanks, aircraft carriers, rifles and ammunition. This meant that the US “produced” a lot less in 1946 than it did in 1945 or 1944, but that production was for military purposes only. While this created a recession, it was a recession on paper only. If you haven’t noticed, that 85th worse year for personal consumption growth? Out of a period of 85 years, (ie, the best).

The decline in one sector of GDP (government spending) was great enough to cause GDP to significantly retract (worse year for GDP since the great depression), but it did not make other sectors of GDP fall. In fact, not only did personal consumption have its best year eve (or since 1930 when the BEA statistics start), so did Gross Private Domestic Investment (same source as above). Unemployment did increase, but not to a very high level (we’ve seen a steady fall in unemployment for 6 years, but we’re still above the 5.2% immediate post war peak). Basically, the idea that the end of the war would throw the US back into depression was wrong, the only fall in GDP was only a reduction in anti-Nazi spending. Its a bit like saying that the best measure of a person’s economic well being is his consumption spending. And that in year 2, when he was healthy, he spent significantly less than in year 1, when he was fighting cancer. It was good that he was able to spend money fighting cancer in year 1 no doubt, but it was better that he didn’t have cancer in year 2.

If we’re to tell the story more, the drop in government defense spending allowed resources to be dedicated to other areas, such as business investment, new home construction, and creation of consumer goods. But this this the opposite of what is supposed to happen according to Keynesian economics. The fall in government spending is supposed to depress incomes in general, as all the soldiers return home and the tank builders are laid off, the total demand for goods and services wil fall, causing GDP to fall, not just in terms of fewer tanks and rifles, but everything else as well. This didn’t happen.

Now, this may not be the best case. You can argue that the proper Keynesian government level was still much lower in the war than actual government spending, or that the special circumstances dominate in immediate aftermath of the war.

More recently, we saw two major changes in government spending in the past 8 years. The first was the stimulus bill of 2009, which was supposed to jolt us to recovery and keep unemployment below 8%. The second was the sequestration, which, due to the Republican preference for lower government spending and political fighting about the debt limit, caused government spending to be reduced (and taxes to raise, both of which are considered anti-stimulative effects by Keynesians). Let’s look at how this effected the economy:

GDP Growth 2009 to present

Unemployment 2009 to present

It looks to me like government spending had little to no effect on GDP growth, and unemployment looks totally unaffected. It went up during the recession and a little bit afterwards, and has fallen consistently since, regardless of what the government tries to do.

This is my biggest complaint against Keynesian economics, or I guess fiscal stimulus to be specific. We had an $831 billion government program, and nothing to show for it. The Keynesian economist can make the case that things would have been worse without the stimulus or better without the sequester, that’s certainly a possibility and I don’t want to dismiss it. But I do think that the burden of proof should be on the people proposing trillion dollar expenditures. And I don’t think that the burden of proof was at all met. This is the strongest case I have against fiscal stimuli, they have enormous costs and very unclear benefits.

None of this is a counter to the idea that we shouldn’t have government spending.  There are many great things that government has provided, including roads, schools, police.  To return to the World War II example, we basically prevented a madman from conquering the world, which is a worthy a goal as there can be.  But what I am arguing against is the idea that government spending by itself typically stimulates the economy; the idea that we’re better off having the government hire people to do useless things.

There may be instances when this is true, that simply borrowing and spending money by the government can counteract a bad economic environment.  But as of right now, I’m convinced that it’s probably not true generally (ie, that any bad economic environment can be improved by the government spending more money), and that we can’t distinguish the instances where government spending helps from when it hurts or has minimal effect.

Insider Trading is Stealing

Over at econlib, Charles Hooper argues that insider trading doesn’t really hurt anyone. Insider trading instead is a victimless crime, and is in fact beneficial.

His argument goes like the following:

In the market for any given stock, there are “uninformed buyers” and “uniformed sellers.” That is, people who buy and sell stock but without inside information. In his example he uses the case of something very good happening to a company soon. The number of uninformed buyers (or technically the shares which uniformed buyers buy) will equal the number of uniformed sellers (with the same technicality).

The uniformed seller is hurt because he does not reap the rewards of the good news, but remember this is before any insider trading occurs, so it can’t be said he is hurt by insider trading.

Then, and insider comes along, buys a number of shares, and profits from the good news, making a large sum of money after the news is made public. Mr. Hooper asks who is hurt from this? Certainly not the uniformed seller, because he was going to sell anyway. The person most hurt is the uniformed buyer, who loses on the opportunity. Yet the uniformed buyer has no real claim to the potential benefits, after all, the buyers can’t complain that the “Insider snatched Uninformed Buyer’s dumb-luck windfall”.

There is one fatal flaw in the argument: Mr. Hooper seems to believe that any increase in the number of buyers will be offset by an equal decrease in buyers. But there are two ways that an increase in buys can effect the market, it can “crowd out” other buyers, or it can cause more people to sell, most likely some mix of the two. For various reasons, I think that the effect would be more sellers rather than fewer other buyers, but all we need to establish is that some of the effect of buying shares is encouraging others to sell them.

Now, we see the problem. The insider, an employee and agent of the shareholder, is transacting with the shareholder, using the knowledge that they gained only through being an employee against the shareholder. That is simply wrong.

Let’s clarify with an example: lets say that three friends pool their money to form an oil exploration company, hoping to find the next big well. One of them runs the company (being an engineer and all), and gets paid a salary, but the three friends split all profits evenly. They go a long time without finding any oil, and all three are discouraged. Then, one day, it becomes very evident that they are about to strike oil; but only to the one running the company. Instead of telling his co-investors about it, he instead offers to buy up their shares.

We can argue about ethics and why this is or isn’t ethical; but those discussions go nowhere. Instead, lets talk about economics. Each outside investor knows that, whenever they transact with insider, they can expect to lose. If they sell their shares to the insider, well they can bet that they’re on the verge of hitting oil. And if the insider offers to sell his shares, then they can be certain that the wells are dry (or that the expected find won’t pan out). Basically, the outsiders have one rule if they don’t want to get screwed: never transact with the insider. This of course hurts the insider, why shouldn’t he be allowed to sell to his co-investors if he needs the money, or buy if he has a windfall he wants to invest? In fact, we can see that the insider is worse off in this example than the outsiders (assuming they have jobs which pay the same), they can transact with each other (buying to invest, selling shares in order to consume), assuming their counterpart wants to do the opposite. But the insider can never transact, nobody will every buy from him or sell to him.

How do we fix this? Simple, make it a law that says that the insider can’t transact based on knowledge that he has that the outsiders don’t; the outsiders will now be willing to sell, knowing that they are protected if the insider has material nonpublic information. Thus, the insider now has the same privileges as the outsiders, and (in our simplified example), everybody is better off (they all have more potential trading partners) and nobody is worse off – clearly a pareto improvement.

But lets imagine a twist to the scenario. Instead of buying the shares directly from the outsiders, the insider sets up a shell company to purchase the shares from the outsiders, the outsiders have no way of knowing that the company is really owned by the insider. Is this ethical? I certainly don’t think so, all the shareholders put their trust in the insider that he would act in their best interests; keeping all potential upside is certainly not within their interests.

Lets use a third example. Same company, same three friends. They strike oil in a well which is expected to produce 100 barrels a day. Instead, it produces 120 barrels a day. The insider decides to take the extra 20 barrels and sell them himself (whether he does this buy selling them through the company and then embezzling the money or physically diverting the 20 barrels a day is unimportant). Is this action ethical? No – it is stealing. But it’s really the same action as above – using insider information (in this case the number of barrels being produced) in a manner against the best interests of the shareholder. How is this any different from insider trading – finding the extra 20 barrels a day, and then offering a price to the other shareholders based off of 100 barrels a day. In fact, it can be shown that the economic impact is exactly the same- the outsiders get the net present value of the sale of 100 bpd split evenly while the insider gets the net present value of the 20 bpd entirely to himself.

Now, you may argue that this doesn’t apply to the situation of the market as a whole; after all transactions on the stock market are much more anonymous. But this doesn’t make it better, it makes it worse. There is (typically) no way of knowing who you are buying shares from; so the insider who buys shares using material nonpublic information would be as if the insider in our above example set up a shell company to buy shares from the other investors. They would sell without ever even knowing that they were selling to the business partner. I can’t see how any of this is ethical, it is currently illegal and should remain so.

Of Flowers and Man

Imagine you are a visitor to a primitive society. And lets say that, for whatever reason, you are treated as something of a magician, or wise man, not quite a God, but just because you’re an outsider you are treated very specially and definitely listened to.

Now lets say that the religion, or superstition if you will, of the people involves sacrifices to the weather god or gods (or goddesses or whatever, its not important), in the hopes of bringing favorable weather. Let us also suppose that the great divide between the society is two camps, one of which believes that humans should be sacrificed, the other that flowers should be sacrificed. There’s quite a bit of rhetorical fighting between the two sides as they argue about which is correct.

Furthermore, lets say that the position in favor of human sacrifice is generally considered more correct (the society is divided about 60-35-5 for Human, Flower, and everybody else factions respectively), and (in what is of course a totally unlikely development), the society, although quite primitive in their understanding of weather, has developed advanced statistical methods, and the standard regression seems to re-inforce the human sacrifice position. For the sake of argument (of course, granting that this whole example, if not whole blog, is for nothing but the sake of argument, as nothing like this is obviously some sort of analogy that I hope is going somewhere), the popular regression compares rainy days per time period with human sacrifices in the same period, and that the formula is something like:

Number of Rainy Days = Constant + A (number of sacrifices) + B (Season)

Where season is a dummy variable on whether its the rainy season.   A has the right (in this case positive) sign on it, and is statistically significant, but only at the 10% level. (I know there’s probably a lot of ways you can attack this regression and there are any number of scenarios which are non-causal but may lead to this relationship, for instance, perhaps the society is more likely to sacrifice when rain seems likely, such as on cloudy days, but I’m already spending way too much time on analyzing an argument that doesn’t actually exist so I’m not going to put any more effort into trying to make the mathematics behind an imaginary regression analysis more sound).

The human sacrificers don’t like human sacrifice per se (ie, they have no attachment to it outside of the fact that they think it brings rain), but they argue that the small number of lives lost due to the sacrifice are a tiny cost compared the large number of lives saved by the better weather.

The question posed to you as an outsider is, which faction do you support, those who would sacrifice humans, or those who would sacrifice flowers? Even though, from the viewpoint of the people involved, the human sacrificers have the better case, I would absolutely argue for the flower faction, not because I think they are more likely to be correct (in fact I think that neither action would have any effect on the weather), but because the costs of flower sacrificing are much smaller.

The point is that we often view arguments and debates on which side we find to be more likely to be right, without taking the entirety of the decision making process. We can do similar things. If we think that there is a 51% chance that our flight will crash, we absolutely will not get on it. But if we revise our estimates to think that there is only a 49% chance that our flight will crash, well, we still won’t get on it. I don’t know what the tipping point is, but its probably something well below a single percent (humans, myself included, are very bad at interpreting very low probabilities, so we’d probably do something like put the odds at 1 in a thousand justify getting on the plane, when in reality they shouldn’t, but I’m on a tangent).

The original argument that I was thinking of regarding the flowers vs the human sacrifices is of Keynsian Economics vs Monetarism (or market monetarism, or Chicago school, or whatever); that is should we, at the zero interest rate bound, use government deficit spending Fed Open Market purchases to stimulate the economy? My guess is that they’re probably both equally wrong, but the Keynsian process is more costly, so I tend to support the less costly option. If in fact, you do believe that Keynsianism works, than by all means support it, but the question you should be asking isn’t “what economic theory is most likely to be correct,” the question should be “What is the probability that Keysnianism is effective” and then have some sort of bar from which you would support it. The question of “what is most likely to be correct” is a perhaps useful for academics for determining what to study, but not for which policy to support.

(this post is an analogy: I’m not trying to actually compare Keynsian economics to human sacrifice).