Measurement without Theory on the Empire State Manufacturing Survey

Corrections examines today's Empire State Manufacturing Survey:  the release contains 88 variables:, current and future estimates of New York manufacturers for: business conditions, orders, future shipments, delivery time, inventories, unfulfilled orders, prices paid, prices received  number of employees, and average employee workweek, along with future capital expenditures and future technology spending.

Corrections takes each series, normalizes it, HP filters it, multiplies it by -1 if it correlates negatively with the average of all series together (so variables that are high when conditions are bad flip sign), and finally sorts the series vertically based on the value in July 2009.  We then graph the series in a heat map:  red when the normalized, filtered, and signed variable is "high" and blue when it is "cold."

The result is blind economics (click to enlarge), without theory, seeing if these variables are telling us anything related to one another, suggesting they're measuring something.

University of Michigan Sentiment Survey

The University of Michigan's sentiment survey doesn't seem to be incredibly stable.  Instead, it seems incredibly noisy for a number thats used so frequently in the press for month to month comparisons (click to enlarge):

Impulse Response Functions and Data

Every once in a while, you create a graph without really massaging the data that looks very similar to a calibrated model.  Below, a "shock" and return to trend in initial jobless claims (click to enlarge).  Note that the data is in percent change from a year ago.

Give Peaceful Resistance a Chance

New York Times opinion "Give Peaceful Resistance a Chance" (March 9th, 2011) offers insight into part of an economist's raison d'être. The article by Erica Chenoweth, a Professor of Government at Wesleyan University, misunderstands heterogeneity in revolutionary starting conditions. Her claim:

Unfortunately for the Libyan rebels, research shows that nonviolent resistance is much more likely to produce results, while violent resistance runs a greater risk of backfiring.

To back up this statement, she claims that:

Indeed, a study I recently conducted with Maria J. Stephan, now a strategic planner at the State Department, compared the outcomes of hundreds of violent insurgencies with those of major nonviolent resistance campaigns from 1900 to 2006; we found that over 50 percent of the nonviolent movements succeeded, compared with about 25 percent of the violent insurgencies.

Why is this bad research? Because not all revolutions have the same starting conditions. Imagine a world in which there exist dissidents who optimally choose between armed rebellion and peaceful protesting. These dissidents know the government they face and the probability of success of a given venture. Then the observed probability of failure for armed revolution may be higher than of peaceful protest even when both were always chosen optimally. To understand why Chenoweth and Stephan get the results they do, observe the following conjecture in figure form (click to enlarge).

A rational rebel will take the upper envelope of these two decisions. If this is the case, then we will see violent revolutions under oppressive regimes that generally fail, and peaceful protest under non-oppressive regimes that generally succeed. There will be no Ghandi's under Hitlers, Stalins, and Mussolini's. Ghandis can only be successful under men who are inherently peaceful when others are peaceful with them, such as Churchill. Similarly, it may often be difficult for Malcolm X's or The Weathermen to succeed in vastly less oppressive societies, where hoses and dogs rather than machine guns and concentration camps are used to control protest. They are unable to gain the sympathies of the populace. Martin Luther Kings are better suited for the United States than the Malcolm Xs or Bill Ayers of the world.

It is no coincidence that Bill Ayers changed from attempting to bring down the United States by armed rebellion to bringing it down from within--he reoptimized.

The conclusions Chenoweth attempts to draw are faulty due to sample selection.  Armed rebellion may be more likely to succeed conditional on the situation but less likely to succeed when looking at the unconditional success rate.

Why California should just say no to Prop. 19

LA Times OpEd "Why California should just say no to Prop. 19" (August 25th, 2010) deceives readers.

A 2004 meta-analysis published in the journal Drug and Alcohol Review of studies conducted in several localities showed that between 4% and 14% of drivers who sustained injuries or died in traffic accidents tested positive for delta-9-tetrahydrocannabinol, or THC, the active ingredient in marijuana. Because marijuana negatively affects drivers' judgment, motor skills and reaction time, it stands to reason that legalizing marijuana would lead to more accidents and fatalities involving drivers under its influence.

The meta-analysis the article mentions is called "A review of drug use and driving: epidemiology, impairment, risk factors and risk perceptions" in the Drug and Alcohol Review Volume 23, Issue 3 by Erin Kelly, Shane Darke and Joanne Ross, and is available here (gated). The analysis notes explicitly that, "There is inconsistent evidence regarding the impairing effects of cannabis in field studies." It lists four field studies and notes that three of them indicate no significant impact of cannabis consumption on driving. The article concludes, "the relationship between between THC and street driving performance is equivocal." Any reader inclined towards believing what the data is telling them, rather than what they wish they could find, would conclude not that "legalizing marijuana would lead to more accidents and fatalities involving drivers under its influence", as the LA Times piece has, but rather that we shouldn't expect legalizing marijuana to change the number of accidents and fatalities. The OpEd's conclusion is ridiculous and unsupported by the evidence they themselves cite.

In addition the the egregious error discussed above, the article also provides readers with a foolish analysis of negative externalities.

The current healthcare and criminal justice costs associated with alcohol and tobacco far surpass the tax revenue they generate, and very little of the taxes collected on these substances is contributed to offsetting their substantial social and health costs. For every dollar society collects in taxes on alcohol, for example, we end up spending eight more in social costs. That is hardly a recipe for fiscal health.

This analysis implies that the pleasure people get from drinking is worthless. Drinkers and non-drinkers alike deserve to have their utility taken into account when analyzing the costs and benefits of a liquor tax. Basic economic theory would tell us that social surplus is maximized when the marginal social cost of drinking equals the marginal social benefit of drinking. If we are counting drinkers as members of society, then it follows that some level of drinking is optimal. The figure below demonstrates how a pigouvian tax allows us to arrive at the optimal social level of alcohol consumption (click here to enlarge).

In equilibrium, the level of the tax just measures the difference between marginal private cost (cost of drinking to drinkers) and marginal social cost (cost of drinking to society, including alcohol-related externalities). Taxes on goods with externalities are meant to regulate consumption to socially optimal levels. As in the figure above, it is possible to draw social and private cost curves that yield the result discussed in the article--tax revenue is one eighth the area between social cost and private cost curves. The red shaded area represents the tax revenue and the blue shaded area represents the difference between social and private cost. This is a socially optimal (but not necessarily revenue maximizing) tax.

Handguns shouldn't be a household staple

Chicago Tribune article "Handguns shouldn't be a household staple" (July 9th, 2010) argues that handguns in homes make suicide easier for young people:

A gunshot is quick and irreversible. About 90 percent of suicide attempts with a firearm are fatal, compared with 2 percent of drug overdoses and 3 percent of attempts by cutting. Adolescents who experience a moment of despair and act impulsively with easy access to a gun almost always have a deadly result. Adolescents also are likely to survive most other methods of suicide, and nine out of 10 who attempt suicide and survive will not die by suicide at a later date.

Of course, people choose different weapons when they want to die with certainty. Those who take drugs or cut their wrists may not substitute into a handgun even if one is available because they seek attention or risk, not certain death. The proper comparison for the author to make would be whether suicide rates rise as soon as gun laws change in a state. A study by doctor David C. Stolinsky published in the Medical Sentinel notes that

The U.S. suicide rate has fluctuated between 10 and 17 for a century, with peaks in 1908 and 1932, and shows no relation to gun laws or gun availability.

Further research that exploits within-state variation is necessary to fully understand the impact of gun availability on suicide rates. Nonetheless, the Chicago Tribune's article attempts to make fact from what is quite possibly fiction, using irrelevant statistics to convince readers of a largely untested hypothesis.

For a New Generation, an Elusive American Dream

New York Times article "American Dream is Elusive for New Generation" (July 6th, 2010) neglects forces of supply when discussing the labor market facing recent college graduates. Referring to the choices of one recent college graduate, the article notes:

Over the last five months, only one job materialized. After several interviews, the Hanover Insurance Group in nearby Worcester offered to hire him as an associate claims adjuster, at $40,000 a year. But even before the formal offer, Mr. Nicholson had decided not to take the job.

Rather than waste early years in dead-end work, he reasoned, he would hold out for a corporate position that would draw on his college training and put him, as he sees it, on the bottom rungs of a career ladder.

Interestingly, a recession resulting in a decrease in the number of jobs available to skilled workers (a decrease in the demand for skilled labor) may not be the only thing keeping college graduates from "good" work. In fact, the supply of college graduates has continued to increase over time. The graph below plots BLS records of the number of college graduates in the labor force (both employed and unemployed) over time (click here to enlarge). Such competition for work drives down wages.

In addition, we should expect the number of college graduates to increase during a recession simply because fewer job opportunities make investment in future productivity more attractive. Potential labor force members will instead spend their time in school so that when they are able to find work, they will be paid even more for their labor. The article then cites work by a Yale economist meant to worry readers that a low payoff to education now will persist into the future:

In a recent study, she found that those who graduated from college during the severe early ’80s recession earned up to 30 percent less in their first three years than new graduates who landed their first jobs in a strong economy. Even 15 years later, their annual pay was 8 to 10 percent less.

Of course, those who choose to enter the labor force during a recession are those for whom additional human capital accumulation will likely not pay off. Specifically, if everyone who does not take their first job during a recession chooses to invest in a two year law or business school degree, taking their first job after the recession passes, why would we expect them to have the same annual pay fifteen years later? The author notes this in her paper:

I also find that cohorts who graduate in worse national economies are in lower level occupations and have slightly higher educational attainment.

In fact, the author argues that such high effects persist when we assume that students cannot shift their graduation and educational attainment decision in response to economic fluctuations. Far weaker effects than those reported in the article persist when people are allowed to endogenously choose their education--OLS results suggest a long run 1.5% wage loss. Though the New York Times article may be correct in noting a long-run wage differential, it completely mis-interprets a theoretical exercise (instrumenting for year of graduation using year of birth) to make a point.

Autistic Teen Picks First Two NCAA Rounds Perfectly

NBC News report "Autistic Teen Picks First Two NCAA Rounds Perfectly" (March 25th, 2010) appears to report a non-event. It notes that an autistic individual picked the first two brackets perfectly. Corrections suggests that is expected to happen about 4 times every year.

In fact, he picked every game through the first two rounds correctly. The odds of anybody doing that? One in 13,460,000, according to BookofOdds.com. It's easier to win the lottery. Twice.

According to a survey, about 58 million people will fill out a bracket (more brackets may be filled out, as the average per participant is presumably greater than or equal to one). If the tournament is based purely on chance, we expect a one-in-thirteen-million event to happen four times, on average, each trial of 58 million. Indeed, the NCAA tournament is not composed of random events, and we should likely expect this to happen more than four times, not less.

Finally, while it's easier to win the lottery, according to NBC News, Corrections notes that the winning of the lottery is an everyday event.

Tyler Perry's Crack Mothers

New York Times editorial "Tyler Perry's Crack Mothers" (February 26th, 2010) confuses level of drug rates with per capita crack consumption. Specifically, it claims that depictions of black women as crack-addicted are improper because the total number of blacks admitted to clinics for addiction is now less than the total number of whites.

Furthermore, data from the Substance Abuse and Mental Health Services Administration revealed that of the total admissions to treatment services for crack use, blacks outpaced whites in 1996, but whites outpaced blacks in 2005 for those under 30 years old.

There are six whites in the U.S. for every black. If blacks used crack cocaine less than whites, as the Times appears to attempt to suggest, one should see black admissions to clinics being one sixth that of whites, rather than approximately the same.

This piece of evidence that the Times gives is simply ludicrous. If movies are to be ethnically fair, then they would show (approximately) six times as many black crack addicts as white.

One might add that a priori we expect that whites, who are on average weathier, are over-represented in the data the Times provides.