Category Archives: Economics

Stimulus efficiency in a post-industrial economy

As we move into the middle portion of 2010, we’re hitting the fattest part of the ARRA stimulus allocations. While there was a lot of pre-passage wrangling on both sides of the aisle about efficiency and multipliers, there hasn’t been much talk about the stimulus money’s effect on unemployment in recent months. Not since the unemployment rate has been trending in a positive direction. That means it’s a great time to examine stimulus. lists 682,779 jobs as “created” or “saved” as a result of the stimulus package. ARRA started doling out money on Feb 17, 2009, and through March 31, 2010, $205.3bn has been laid out. This means that $205.3bn has puchased 682,779 jobs, at a cost of $300,686 per job.

That’s pretty horrible. Apparently it’s not quite as straightforward as this, though, thanks to some fancy hand-waving that magically brings this number down to ~$160,000 through methods that I’ve yet to see explained. Regardless of what you believe, it’s really expensive for Uncle Sam to create jobs using good old fashioned Keynesian stimulus.

Capital intensity

The neoclassical production function is y = (K,L); that is, output is a function of capital (K) and labor (L). Modern macroeconomics throws a lot more into this function to account for other factors, but using the older capital-labor tradeoff is good enough to illustrate my point. In general, there is a tradeoff between capital and labor. You could hire 1,000 men to shovel the streets in the winter, or you could hire one guy with a plow, and the guy with the plow will do more in less time. This is why nations tend towards industrialization.

The United States is a post-industrial economy. Watch a documentary on the Hoover Dam and look closely at the sheer number of men at work. In today’s economy, a large percentage of these men would be replaced with machinery designed to do a specific job.

That means that money is being spent on capital instead of labor because stimulus money is allocated for specific tasks — not to employ workers, which is merely a fringe benefit — and then these tasks are completed by either the public or private sector (or both) using whatever mix of capital and labor is appropriate for the job.

Capital flight

Capital has to come from somewhere. In terms of traditional stimulus spending on infrastructure, you’re mostly dealing with heavy equipment manufacturers, the majority of which are located overseas. In some cases they’re headquartered elsewhere, and in some cases, their manufacturing is located elsewhere. Sometimes both. When one or both of these conditions is met, profit, jobs, or both are shipped across borders into other nations. In this respect, stimulus money is being used to create or sustain jobs in other countries.

It is possible to mandate that firms using government money buy from American corporations — the Fly America Act is an example — but this isn’t possible or desirable in all cases. American heavy equipment manufacturers don’t make all of the machinery necessary to complete some of the larger infrastructure projects that are on the table.

Job destruction and obsolescence

Another problem with stimulus spending is that quite a bit of the funding is to improve efficiency, particularly in sectors like health care. Unfortunately, “improving efficiency” usually means shifting away from labor. In a sector like health care, capital takes the form of computers and software. Primarily the latter. Automating billing, cutting back on administrative personnel, decreasing overhead, getting rid of physical records. These are the things that software is great at; it is a substitute for labor.

Software is a unique example, too. Not only does it destroy jobs in the short run, but it doesn’t create new jobs very quickly, except in the software-based industries themselves, because it’s a virtual good: all you need is a computer, an Internet server, and the intellectual capital required to make it. This breaks the normal supply-demand models in some interesting ways, because supply is functionally unlimited, but price doesn’t go to zero, even after a firm’s fixed costs have been covered. This leads to some very high profit margins.

History is littered with examples of capital displacing entire workforces. Textile manufacture during the Industrial Revolution is the most prominent example in modern history, with huge numbers of people losing their jobs thanks to the electric loom. The upside is that because of this creative destruction, whole new industries are born, and more jobs are created than were destroyed.

This doesn’t lessen the pain in the short term, however. While the labor market can and does adjust for this creative destruction, this obsolescence is occurring so quickly that many of these displaced workers are unable to acquire the skills necessary to find new kinds of employment in the new economy that’s being created. In times past, the technological increase was on the order of 3% per generation (pre-Industrial Revolution), but now stands at about 3% per year. The rapidity of this kind of change is difficult for the labor market to absorb. This is one of the causes of a jobless recovery: technology increases wealth (GDP) without using more workers to get the job done thanks to software advances and other capital-based technologies.


If the goal is mass job creation as quickly as possible, then the government should employ fiscal stimulus as inefficiently as it can. Employ workers where the private sector would use machinery. Employ dozens of people with slide rules instead of one physicist with a computer. Build roads by hand instead of using machinery. In short, pretend we’re a pre-industrial civilization instead of a post-industrial one. This will get you the biggest bang for your buck in terms of rapid, low-cost employment. The work isn’t desirable, and people will jump ship back to the private sector when it starts creating jobs as a result of people spending their money.

In this case, inefficiency props up the labor market in the short term.

The downside to this is that it doesn’t create jobs in the long-term growth sectors like technology, health care, and education, which are physically and intellectually capital intensive. It does, however, get a lot of people employed very quickly.

The death penalty in civilized society

Today, Bob Herbert in the New York Times had a piece on Cameron Todd Willingham, a man executed by lethal injection for the arson murder of his children. Analysis published on August 17, 2009 by Dr Craig Beyler concluded:

in the end, the only (basis) for the determination of arson … is the burn patterns on the floor of the children’s bedroom, the hallway and the porch interpreted as accelerant spill. None of these determinations have any basis in modern fire science.


the state fire marshal who investigated the case and testified against Willingham “seems to be wholly without any realistic understanding of fires.” He said the marshal’s approach seemed to lack “rational reasoning” and he likened it to the practices “of mystics or psychics.”

These are some pretty damning statements.

Now, I am against the death penalty, though I wasn’t always, and I posted the article on Facebook as a textbook example of why I believe the death penalty should be abolished. It garnered a few comments, one of which I want to address in a more comprehensive way.

So let’s spend millions of dollars keeping inmates locked up for life, build new prisons, take care of their health problems, etc rather than run the very slim risk that there may be a mistake?

All eliminating the death penalty does is guarantee that people who commit lesser offenses such as manslaughter and rape will be released after the minimum required sentence (or less) due to overcrowding.

I’ve heard this, and similar arguments many times over the years. In the past, I’ve even said similar things myself. However, these arguments are spurious.

  1. It is more expensive to execute a prisoner than it is to lock one up for life. “Fixing the appeals process” is incompatible with lowering that cost, because the marginal cost of holding one more prisoner is quite low, and the appeals process is necessarily time- and labor-intensive. (More below.)
  2. The people who are likely to be released from prison in a real, honest-to-God prison reform are nonviolent criminals. (Drug users and the like) Of the 2.2 million prisoners, 1 million of them are non-violent.

That report, by the way, is from 1999… ten years ago. Many of these nonviolent offenders are in prison for drug-related crimes. Regardless of your opinion of whether or not drugs should be legalized, drug abuse is a medical problem, not a criminal justice problem. As such, it should be dealt with by the healthcare system, not the criminal justice system. Indeed, in any graduate-level mental health program, a significant portion of the curriculum is spent on substance abuse and its treatment. In fact, the federal government itself recognizes this, as it has quite a few openings for substance abuse experts.

There’s plenty of room to improve our criminal justice system from redesigning prisons to reforming the law itself to rethinking how we imprison people instead of just the “why”. Looking back at history offers some lessons as well (long-term imprisonment as a concept is only 200 years old (PPT)). Continuing to execute individuals, and justifying it using counterfactual ideas like “cost savings” doesn’t make sense. If we could reform our criminal justice system to the point where executing one more person made fiscal sense, then we will have made significant headway in criminal justice policy in general. And I would go so far as to argue if we ever get to that point, we won’t need to execute people at all because there won’t be any crimes committed that justify execution.

But of course, we’re nowhere near that point, so it’d be purely a game of “what-if”.

Some numbers:

  • 3,297 people on death row
  • 2,310,984 total prisoners
  • 0.1% of the prison population is on death row

You’re not going to save the system much (if any) money even if you executed all of these prisoners tomorrow.

Going further, the argument seems to be one of ideology: having an opinion and working backwards to justify it rather than letting the evidence guide one to a logical conclusion — even if it is a conclusion that one would otherwise not prefer. Speaking for myself, I have no problem with the concept of putting someone to death, though I could never do it myself. I’m apathetic about the death penalty as an institution, but what gets me going is the possibility of error. If there is a chance that an innocent person could be put to death, the whys and hows of execution are irrelevant. Life in prison is reversable. Execution is not; ergo we don’t kill people. Moral arguments over the right-ness or wrong-ness of killing someone cease to matter.

There are too many possible points of failure. Arguing that we should lower the costs of execution, by streamlining the appeals process while still doing “due justice” to satisfy some moral compunction reeks of trying to alter the world to conform to what one would prefer rather than making policy based on reality.

I have no patience for this kind of thinking.

Other links of interest:

Pigouvian taxes on soda

The Center for Science in the Public Interest — an organization I’ve wanted to work for — is pondering a soda tax. It’s not a new plan… I first had the idea in the fall of last year, though I didn’t write about it, and Ezra Klein has written about it twice. (1, 2) Quite a few people in a Kaiser Family Foundation/NPR poll (PDF) are in favor of it as well:

I don’t see a problem with it, really. It’s not a bad idea to take the negative externalities associated with a given activity and force the market to internalize them. I have a paper sitting on my hard drive that I’ve been meaning to adapt for the web for a while, but I’ll share a snippet here:

Money that would have greater marginal utility upstream is instead being spent downstream, because measuring the downstream effects are easier than measuring upstream effects. We spent an estimated $91.3 billion in diabetes care in 2002. How much money could be saved not just in diabetes but on healthcare in general if we combated the upstream problem of obesity instead of throwing money at its downstream consequences? Drastic expenditures on things like adverse cardiovascular events? In 1995, the economic cost of obesity was estimated at $99.2 billion. In 1997, an estimated 19.4% of the US population was obese (PDF). In 2007, that number jumped to 26.6% (PDF). Converting 1995 dollars to 2007 dollars, assuming that real per capita obesity costs did not increase over time (very unlikely) gives us this chart:

Obesity is probably the main public health concern in the United States these days. Communicable diseases have largely been eliminated through public health efforts, leaving lifestyle diseases as the main cause of morbidity in this country. If we cut our obesity rates in half, we could eliminate quite a lot of spending further down the line.

That adds up to about $240 billion a year spent on obesity care.

Pigouvian taxes work. By making the market account for the negative effects of its activity, it modifies consumer behavior — if not on an individual level, then certainly on a macro level. A 3 cent tax on soda doesn’t sound like much, but I’d put a significant chunk of change on it having a measurable effect on obesity. Of course it would only be one step in a real comprehensive public health strategy, but certainly not a bad step.

Before we go taxing soda, however, I would also humbly suggest that we get rid of the corn subsidies that make high fructose corn syrup so cheap, and eliminate sugar tariffs as well. Not that sugar is any healthier than HFCS from a public health standpoint, but it’s not generally good policy to layer a tax on an already-distorted market. If we stopped subsidizing corn production, we might not even need to have a soda tax… the increase in price might well take care of the over-consumption problem by itself.

Revisiting talent in the public and private sectors

At the end of November, I mused about public sector’s problem of attracting and retaining top-notch talent:

Ben Bernanke’s salary as chairman of the Fed is just over $191,000. Henry Paulson as CEO of Goldman Sachs made $16.4 million according to Forbes. […] I find it very hard to believe that compensation plays no role whatsoever in an individual’s choice of employment.

This all seems terribly obvious to anyone who thinks about it, and Nate Silver has commented on the same phenomenon with a real-world example:

In retrospect, it is clear that regulators did not have the human capital to keep up with the financial industry, and to understand it well enough to be able to exert effective regulation. Given the wage premia that we document, it was impossible for regulators to attract and retain highly-skilled financial workers, because they could not compete with private sector wages. Our approach therefore provides an explanation for regulatory failures.

That is, the excessive wages paid by Wall Street not only lure talent away from other parts of the private sector, but also from the public sector, where employees are subject to government wage controls. The very people who might be the most capable of enforcing regulations on the banks instead wind up working for them.

This is a very real problem. Some of the work that I did in my first job after college at KPMG involved valuing intellectual property in conjunction with international tax disputes. We had our economists, and the IRS had theirs. The thing was, however, that our economists were better than the IRS’s, because if someone at the IRS was any good, we’d hire them away and treble their salary. Part of a good regulatory reform plan, then, would be to increase the salaries paid to employees at institutions like the Fed, the Treasury, the IRS, and the FDIC.

So how do we solve the problem? Tripling an IRS economist’s salary probably means they’re making in excess of $400K/year, which is more than the President makes. Is it feasible to have regulators that make more than the President?

Nerdiest GPA requirements ever

BU’s GPA requirements for their Economics PhD students:

GPA Requirements
All courses must be passed with a grade of B- or higher. The core average must be pi (3.1416) and an overall grade point average (GPA) of 3.0 for the M.A.P.E. and the Ph.D. must be attained in all courses taken after enrollment in the Graduate School at Boston University. At the end of each academic year, a student’s course grade average is reviewed. Failure to maintain satisfactory standing results in recommendation for termination from the Ph.D. program.


Whither this data?

I’m trying to do some analysis on car accident statistics with respect to at-fault collisions for people driving vehicles with automatic transmissions vs those who drive stick. While the NHTSA has a surprisingly advanced query system set up that allows you to drill down to pretty specific results without much trouble, I can’t seem to find this data. I’ve gone through the raw data as well, and this information doesn’t appear to be collected.

I’m trying to find out whether transmission type has an effect on accident rates.

VIN numbers are collected, and this information is associated with a specific VIN, but is there a way to easily pull this information from somewhere? Hmm.

Children, social insurance, and birth rates

A new twist on children-as-social-insurance. Today’s Stone Soup:

The Sandwich generation

Sandwich generation indeed…

This reminded me of some fascinating information from the Gates Foundation. In his first annual letter, Bill wrote that better health is linked with smaller families, so as the overall health of an impoverished region gets better, overpopulation has a long-run tendency to decline, so Malthusian objections aren’t applicable. Women don’t have to have as many babies to ensure that some survive:

Better health, smaller families, Gates Foundation

The entire letter is worth reading (PDF). The Gates Foundation is doing some interesting work, and I really like their emphasis on measuring results. I look forward to the day that measurement is politically acceptable and expected in American society. Especially in places like public schools.