Monthly Archives: August 2015

An interesting argument on why we should colonize space

Via Tim Urban:

Humans have never experienced a mass extinction event, and if one happened, there’s a reasonable chance it would end the human race—either because the event itself would kill us (like a collision with a large enough asteroid), or the effects of an event would (like something that decimates the food supply or dramatically changes the temperature or atmospheric composition)…

Let’s imagine the Earth is a hard drive, and each species on Earth, including our own, is a Microsoft Excel document on the hard drive filled with trillions of rows of data…

  • in August of 2014, the hard drive was loaded up with Excel documents (i.e. the origin of animals). Since then, new Excel docs have been continually created and others have developed an error message and stopped opening (i.e gone extinct).
  • Since August 2014, the hard drive has crashed five times—i.e. extinction events—in November 2014, in December 2014, in March 2015, April 2015, and July 2015. Each time the hard drive crashed, it rebooted a few hours later, but after rebooting, about 70% of the Excel docs were no longer there. Except the March 2015 crash, which erased 95% of the documents.
  • Now it’s mid-August 2015, and the homo sapiens Excel doc was created about two hours ago.

Now—if you owned a hard drive with an extraordinarily important Excel doc on it, and you knew that the hard drive pretty reliably tended to crash every month or two, with the last crash happening five weeks ago—what’s the very obvious thing you’d do?

You’d copy the document onto a second hard drive.

That’s why Elon Musk wants to put a million people on Mars.

Information vs. Attention

Herbert A, Simon via Timothy Taylor:

in an information-rich world, the wealth of information means a dearth of something else: a scarcity of whatever it is that information consumes. What information consumes is rather obvious: it consumes the attention of its recipients. Hence, a wealth of information creates a poverty of attention and a  need to allocate that attention efficiently among the overabundance of information resources that might consume it…

Striking the right balance is something I struggle with.


Daniel Nazer on the importance of venue reform to combat patent trolls

Daniel Nazer:

the Eastern District of Texas… An astonishing 1,387 patent cases were filed there in the first half of 2015. This was 44.4% of all patent cases nationwide. And almost all of this growth is fueled by patent trolls

It is just one of 94 federal district courts… If patent cases were distributed evenly among the federal district courts, each one would have received about 33 cases so far this year – a far cry from the 1,387 filings in the Eastern District of Texas.

The Eastern District of Texas was not always so popular. In 1999, only fourteen patent cases were filed there. By 2003, the number of filings had grown to fifty-five. Ten years later, in 2013, it was 1,495.

This massive rise in litigation followed the appointment of Judge T. John Ward in 1999, and his drive to create local patent rules. Judge Ward’s rules, while similar to patent rules in other federal districts, had some additional plaintiff-friendly features such as a compressed discovery schedule and a short timeline to trial. This so-called “rocket docket” attracted patent plaintiffs eager to use the compressed schedule to pressure defendants to settle. For those cases that went to trial, the district got a reputation for huge patent verdicts. As one commentator explained, the Eastern District’s “speed, large damage awards, outstanding win-rates, likelihood of getting to trial, and plaintiff-friendly local rules suddenly made [it] the venue of choice for patent plaintiffs.”…

It’s time for Congress to act. Although the Federal Circuit has overruled some of the Eastern District of Texas’ most egregious venue decisions, it has failed to bring basic fairness to where patent cases are litigated. We need new legislation to clarify that patent cases belong in forums with a real connection to the dispute.

Fortunately, Congress is looking at the problem. Representative Darrell Issa recently offered an amendment (PDF) to the Innovation Act that would tighten venue standards in patent cases. On June 11, the House Judiciary Committee approved the amendment.

Libby Nelson explains the recent Fed study on the Bennett Hypothesis

Libby Nelson has a good round-up of the recent NY Fed study on the Bennett Hypothesis (the notion that federal financial aid leads to colleges raising tuition). My own take on the study is here.

Her post concludes with a skeptical take from Donald Heller of Michigan State University, who tries to dismiss the Bennett Hypothesis with an analogy:

“Nobody has argued that the availability of car loans has pushed up the price of cars… And that’s the argument they want to make about federal loans and the price of college.”

Heller is right that subsidizing car loans won’t lead to car manufacturers increasing the price of cars. But he is wrong in thinking this teaches us anything about the Bennett Hypothesis in higher education. As I explain here:

We would not expect bread producers to simply spend more making bread until they had captured the entire subsidy, so why does this happen in higher education?

The key difference between higher education and the bread industry is the nature of competition: In higher education, colleges essentially compete in a zero-sum game for relative standing. Due to the lack of measures of output and outcomes, colleges cannot compete on quality, and instead compete based on reputation/prestige/excellence. Essentially, they use high quality inputs as proxies for quality because there is no way to demonstrate high quality directly. Since high quality inputs are costly, and colleges are playing a zero-sum game of relative position, there is no limit to what college[s] will spend in the pursuit of excellence. Thus, they will spend as much as they can, meaning that revenues drive costs (an implication of Bowen’s Rule).

In contrast, bread producers compete based on value (roughly defined as quality divided by price, both components of which are observable to consumers). Costly improvements to bread making will only be undertaken if they improve quality enough to compensate for the increase in costs (and therefore, in the long run, price). Bread makers seek to make the bread with the highest value, they do not seek to spend as much as possible in the pursuit of the highest quality bread regardless of cost. In other words, there is no Bowen’s Rule for bread making. And because there is no Bowen’s Rule for bread making, there is no need to worry that subsidizing bread will lead to higher bread prices and capture of the subsidy by bread makers…

Matthew Ladner on the importance of comparable information on schools

Matthew Ladner:

Greatschools provides an invaluable public service by providing a more rigorous ranking system. In addition, Greatschools also provides a source for student, parent and teacher reviews of schools. Mike McShane cited research showing that parents rate this reviews highly in their decision-making (see page three). This makes it all the better that Greatschools gets more traffic than the state website- it has more of the information they desire…

Greatschools however must make use of public data, and therein lies a growing threat to this invaluable parental choice resource.

It is with great sadness that news has reached my ears that an Arizona Congressman plans to offer an amendment to the ESEA reauthorization bill that would fundamentally undermine Greatschool’s ability to provide comparable data between schools for consideration by parents. The amendment would create a parental opt-out of testing. This would create what seems like a fatal blow to data comparability between schools, and a rather powerful perverse incentive for schools to nudge their weaker students to “opt-out.” Parents and researchers would have little to no way of knowing the rates of strategic “opting out” and thus compatibility between schools would be utterly compromised…

Yes you can argue the federal government should have no role at all in financing schools and/or bossing them around. So why should you care? Notice however that the option of getting the feds out of the schooling business will not be seriously considered anywhere in the process. The federal government will still be sending money, will still be bossing schools around, but just would not get data reliable enough for Greatschools to even pull an Arizona-style rescue manoeuvre, or to consider student learning gains in making more rational retention and tenure decisions. In other words, federal policy will have been entirely captured by adult interests- funding will continue to flow, but transparency will suffer a train wreck.

Evidence on educational signaling from China

Alex Eble and Feng Hu have a new paper on the role of signaling in education. From the abstract:

Wages are positively correlated with years of schooling. This correlation is largely driven by two mechanisms: signaling and skill acquisition. We exploit a policy change in China to evaluate their relative importance. The policy, rolled out from 1980 to 2005, extended primary school by one year… If the primary mechanism behind schooling returns is signaling, we would expect little change in the distribution of credentials in the population, but a large increase in schooling. If skill acquisition dominates, we should see no change in length of schooling but a change in credentials. Our results are consistent with the signaling story… We estimate that this policy, while redistributive, generates a likely net loss of at least tens of billions of dollars, reallocating nearly one trillion person-hours from the labor market to schooling with meager overall returns.

Assumptions and math in economics

Arnold Kling describes how assumptions and math interact in economics:

What math can do is rigorously connect assumptions with conclusions…

economics differs from physics in that physicists usually can undertake direct tests of their assumptions, while economists generally cannot…

when we prove that assumptions a, b, and c together imply outcome X, and we instead observe outcome Y, we have no way of independently testing which of assumptions a, b, and c is untrue in the real world. Because there are so many plausible assumptions available to economists, this means that real world does not constrain economic models nearly as much as it does in physics.

Assumptions persist in economics as they get copied from paper to paper. That is, because of a combination of convenience and path dependence, not because of empirical verification…

New support for the Bennett Hypothesis

A new study by David O. Lucca, Taylor Nadauld, and Karen Shen at the Federal Reserve of New York finds evidence supporting the Bennett Hypothesis (the notion that federal aid enables colleges to increase tuition more than they otherwise would):

we find that each additional Pell Grant dollar to an institution leads to a roughly 55 cent increase in sticker price tuition. For subsidized loans, we find a somewhat larger passthrough effect of about 70 percent. We also find a loading of tuition on unsubsidized loans of 30 percent. All of these effects are highly significant and are consistent with the Bennett Hypothesis.

They study recent changes in aid award maximums, which is a very clever design. The changes in aid theoretically applied to all colleges, but, due to the different types of students at different colleges, plausibly affected each college differently to determine the impact the changes in aid availability had on tuition. In other words, a college with a bunch of students borrowing the old maximum amount has a different exposure to a change in the maximum loan than a college where few students were borrowing at the old maximum.

This is yet one more piece of evidence that care needs to be exercised when designing and altering financial aid programs. As Josh Mitchell notes,

Conservatives have long held that generous federal-aid policies inflate higher-education costs… Now, more mainstream economists and academics are adopting that view, or at least some variation…

In the seven years before Grad PLUS, college tuitions were rising faster than grad-school costs. In the seven years after, the reverse occurred.

Robert Kelchen also offers some insightful comments, including

An even more interesting finding from the Fed paper is that shareholders in for-profit colleges responded favorably to the passage of legislation that increased federal financial aid amounts. They concluded that across three pieces of legislation, the cumulative increase in stock prices was about 10% above what would have been expected without an increase.

The Fed paper spurred a couple of additional thoughts.

1. In the past, I’ve argued that aid programs targeted to low-income students are in less danger from the Bennett Hypothesis, but the Fed study found that increases in unsubsidized loan amounts led to a smaller increase in tuition than increases in subsidized or even Pell grants.

2. This is a good illustration of why the Dept of Ed needs to do a better job of making data more widely available. A key portion of the analysis relied on access to data from the National Postsecondary Student Aid Survey (NPSAS). But use of this data is restricted to select users. For example, I wouldn’t be able to retrace their steps to confirm their analysis because I don’t have access to NPSAS data. Other federal offices do a better job making data widely available while respecting privacy concerns (e.g., the Census bureau’s Public Use Microdata Sample (PUMS)).

3, A straightforward policy recommendation that would kill the Bennett Hypothesis would be to eliminate the Cost of Attendance (CoA) that is used to determine aid amounts and replace it with the median cost of college. See here for more on this idea.