Analyzing a Politically Correct Education/Economic Study

John Taylor’s blog Economics One points to an interesting OECD paperThe High Cost of Low Educational Performance, by Prof. Rick Hanushek and Prof. Ludger Woessmann.  They make projections based on the relationship between performance on international math and science PISA tests and economic growth.

The estimated coefficient on cognitive skills implies that an increase of one standard deviation in performance (i.e., 100 on the PISA scale) would yield an annual growth rate that is 1.74 percentage points higher.              

Their analysis leads to the following politically correct chart:

Except in the case of the lowest performing countries, this approach is very misguided.  While allowing a sizable fraction to achieve basic literacy is necessary, the marginal worker in an advanced economy isn’t going to help much once the network effect of a literate society is established.  By aiming to put a minimum under the PISA scores, they are changing the distribution of the variable that was found to be significant in the past. Scientific and technological progress will more often come from the inventions created by the high achievers on the far right of the distribution than the more sizable group on the left side, so just cutting off the leftward distribution changes the nature of what the PISA test scores were measuring in the first place.  Even if the share of literate population was more significant in their regression than the share of high achieving students*, the relationship between literate students and high achievers can be assumed to be relatively constant up until programs such as No Child Left Behind in the US encourage schools to favor creating the former over the latter.

The following charts from a BOE speech on Inflation and the Service Sector by Timothy Besley highlights what is wrong with trying to increase only the minimum PISA performance to raise economic growth. 

This is a chart of components of the UK service sector.  While the UK’s economy is geared more towards high end services than high end manufacturing, the principle remains the same. The low end services remain a sizable part of the economy. 

The employees at the bottom rung of the service sector are low productivity due to the nature of their industry. More investment in education the left side of the distribution might cut off some of the really bad results, but it will not result in a vastly more productive economy.  Focusing on manipulating the education scores upwards by targeting what looks like the lowest hanging fruit changes the nature of the underlying relationship.  Considering that in the US there is already an order of magnitude more money spent on special needs programs than on gifted programs, the low hanging fruit in education might just be on helping skew the educational score distribution further to the right.



* These regressions can be skewed by some cultures that excel at test scores at the expense of creativity or by looking at too broad a measure of high achieving students to accurately identify the countries with the highest proportion of superstar students. The really high achieving may also be more likely to change their country of residence, benefiting the countries that attract their human capital instead of the countries that produce them in the first place.

Don't be stupid and misinformed, chocolate is way better than vanilla

Michael Kinsley, commenting on disagreement in political discourse:

It's a free country, and people can believe whatever they want. If evidence or reason persuades them that some opinion they hold is wrong, they are free to change it. So at any given moment, we all believe that our own beliefs are correct and anyone who disagrees with us has some explaining to do. Furthermore, if I believe that evidence and reason support my own views, then I also must believe that they do not support the views of those who disagree with me.  

Three possible answers are that they are misinformed, they are thinking poorly, or they are blinded by self-interest.   

This completely misses the idea of differing values, which is among the most common reasons for political disagreement.  Jonathan Haidt has done some interesting work on this issue, which is highlighted in his 2008 TED talk.  He found five foundations of morality across cultures.

  1. Harm/care
  2. Fairness/reciprocity
  3. Ingroup/loyalty
  4. Authority/respect
  5. Purity/sanctity

Liberals care much more about Harm and Fairness a lot more than Authority, Ingroup and Purity while conservatives care about all 5 channels equally.  These are fundamental differences and bipartisanship or even political discourse in general is very difficult as long as people pretend that everyone wants the same thing and the only disagreement is on how to get there.  Politicians playing to the median voter can pretend that they care equally about every moral value, but responsible commenters should be less biased.  At the very least, pretending that the only possible way their values differ from others is that they are less selfish is not constructive discourse. 

An example of another major value premise difference is time preference, where two parties can reasonably disagree on the proper discount rate.  For this premise, it can be a little bit funny watching the debate shift, because liberals discount the future benefits to economic growth but don’t discount the damage to the environment while conservatives do the opposite.  If pressed, liberals might say they don’t believe that their regulatory or redistributive policies will slow economic growth while conservatives against environmental policies believe that environmental protection done incorrectly hampers growth and as long as there is economic growth their descendents will be rich enough to deal with any issues that come up. 

Of course, it is also true that voters from both sides of the political spectrum are drastically misinformed.

Hat tip: Arnold Kling

Living Rent Free

The housing crisis, crash and subsequent pressure to forestall foreclosures keeps people in their homes even though they aren’t making payments.  The average time to foreclosure has drastically increased during the recent recession, rising from three to six months to over a year in some cases.  This has had an interesting side effect: A lot of the marginal consumers are saving quite a bit of money on rent.  There are about 7.8 million mortgages that are delinquent or foreclosed. If average monthly payments are around $700 then these people are in aggregate potentially saving 65 billion dollars a year. This works out to 0.65% of 2009 personal consumption expenditures.  This helps explain how consumption has held up even in the face of declining credit, as the most marginal consumers who are cut off from credit are the same ones that are potentially living rent free.

Macro Mood Hysteresis

Bryan Caplan has some questions about how to apply mood to macro:

What would a full-blown mood theory of macro fluctuations look like?  Ideally it would begin at the micro level - with the individual psychology of traumatic events.  What exactly scares people, and how long do they stay scared?  Then we'd move to the social psychology of fear - how do we respond to other people's fear, and how does "social proof" affect individuals' emotional recovery? 


Empirically grounded mood theories will probably imply that fluctuations are (slowly) self-correcting even in the presence of total nominal rigidity.  The large literature on hedonic adaptation finds that even after blood-curdling experiences, normal people don't remain miserable/ fearful/ angry forever.  After six months or a year, people come to terms with what happened.  It's almost like they get bored of feeling afraid. 

Instead of focusing on hedonic adaptation, the interplay between confidence, catastrophe and engagement could be more enlightening.  One model that I have found useful is from Optimism & Pessimism: Implications for Theory, Research and Practice.  In chapter two, page 46, C.S. Carver and M. F. Scheier present a figure similar to the one below.

Optimists have higher confidence and will more often act in the face of adversity while pessimists will be less likely to put in effort if they think they are not going to succeed.  The change in effort from optimism to pessimism isn’t linear. After a certain level of negative feedback the optimist gives in to doubt and despair and stops trying and at some level of positive feedback the pessimist will decide that they have a chance and put in more effort.  This switch occurs in an area called the region of hysteresis, called such because the level of engagement depends on recent history.

During booms, the feedback investors get is generally positive and confidence is high*.  Minor downturns are generally shrugged off by the investment community and are generally seen as good buying opportunities.  The bust brings most investors down through the region of hysteresis as investors give up and decides to sell assets to protect their capital.  The reason the financial recovery is so rapid is partially due to the fact that many investors are jumping up through the region of hysteresis in the positive direction.  One reason the markets remain fragile today is that when it comes to systemic confidence, many people are still in or near their region of hysteresis where a large enough negative surprise can undo a lot of the recently regained confidence in the system.

*This is partially because the financial community consists of more optimists than pessimists, but that is a topic for another post.

An interesting technique for dealing with academic bias

For people trying to find unbiased answers, there are sometimes ways to get around the politically correct research mentality in academia.  When the variable is a politically sensitive issue, the papers published studying the issue itself may be biased.  They only have to data mine for the right control variables in order to prove their politically correct point.  However, when an academic is trying to prove unrelated points they will often control for these variables in their studies.  Looking at the average of many of these coefficients should give less biased information than a literature survey of the published studies that directly analyze the issue.

I ran across this phenomenon when I was looking into age and productivity statistics. Of the studies looking at age and productivity, many had mixed results.  This is partially because crystallized intelligence is not always worse than fluid intelligence and partially because age discrimination and the government provision of additional work training for older workers is a political issue.  Looking at studies where firm productivity was analyzed in a different context provided more straightforward results: when the proportion of older workers was controlled for it had a significant negative sign.

Topics where this approach could yield fruit:
  • IQ and/or heritability of traits
  • Protected classes such as women, minorities or the elderly
  • The negative effects of tobacco, alcohol or pollution
  • Inequality
  • Any other politically contentious or politically correct issue
Hat tip to Eliezer of Less Wrong, who attributed it to Robin Hanson.

Entitlements in the workplace

Research has shown that for workers over the age of 55 a workers’ productivity is often lower than their wages.

That makes this chart somewhat worrisome:

Data Source: BLS, Author’s calculation

Considering the current demographic picture, the percent of the civilian labor force over 55 is only going to increase. The current society wide implicit and explicit seniority based pay system is not sustainable.  Things will have to break somewhere, and it is more likely to happen in the private sector than in the unionized public sector.

Academics as Bricks in a Wall

I realized early on that the academy and the literary world alike -- and I don't think there really is a distinction between the two -- are always dominated by fools, knaves, charlatans and bureaucrats. And that being the case, any human being, male or female, of whatever status, who has a voice of her or his own, is not going to be liked.

-Harold Bloom

An interesting post over at NCBI ROFL highlights an editorial that looks into the question “Why are modern scientists so dull?”  The theory is basically that the academic filtering process selects people who have moderately high IQs and are high in Conscientious and Agreeableness. Conscientiousness is also preferred to high IQ, as fine attention to detail in projects and tests is driven more by hard work than by brilliance.  Agreeableness helps future professors work their way through the bureaucratic politics of modern academia.  This process weeds out Psychoticism, which is correlated to creativity.  The really high IQ people are also weeded out, as they will not necessarily have high conscientiousness or agreeableness.  This may be true, but if this was always the case then things aren’t getting worse, it is just another instance of someone recognizing a problem in their own time and thinking that it has gotten worse due to a romanticism of the past.  So in order to find out the truth of things, we must analyze what has changed.

1. The increasing importance of the grant process and money driving research:
  • Scientists who can get grants are rewarded over brilliant scientists.
  • Due to the incentives of grant givers, there will be more type leads to more type II errors in which brilliant scientists are denied grants because the grant committee is afraid of being taken in by a charlatan.
  • Specific grants lead to less open ended research, though this effect would have to be measured.
  • Researchers generally don’t get tenure and control of their own research agenda until their fluid intelligence is on the wane.
  • Decreased willingness to criticize high status professors who review the grants in specific fields.
    2. Increased specialization: Overall specialization is a necessary factor in expanding human knowledge. It does have a few side effects:
      3. Political correctness and ethical issues and regulation are getting in the way of research more than before.  There are many ways that this occurs:
      • The reaction of Berkeley’s campus to the research of Arthur Jensen as described by Victor Niederhoffer is one example.  The sociology and psychology professors who were protesting his work could not have read his research because the only two copies of the journal were checked out by genetics professors; they opposed it based on ideology alone.  Other supposedly academic bodies have been found promoting false ideas for ideological instead of scientific motivations.
      • Due to ethical restrictions, Philip Zimbardo’s prison experiments and Milgram’s experiment could never be duplicated today. A lot of interesting psychological research is off limits, perhaps for good reason.
      • These ethics restrictions have been taken to logical extremes, where experiments testing whether making doctors and nurses use checklists improved health outcomes were stopped by the HHS because patient permission was not obtained.
      • When I was at UCSD, my professors would tell me that when the FDA started regulating medical devices progress on these devices slowed to a crawl. Research became much more burdensome and expensive after it was regulated.

        One theory is that while a lot of these distortions started in the 1950’s they had a positive impact before they had a negative impact. The positive impact came from throwing money at an academia with a sizable fraction of geniuses who were doing brilliant work in their fields.  The negative impact of this money was lagged by quite a few years as the structure academia pushed out more and more of the creative types.

        In terms of how to build a better University (Antiversity?) starting from scratch and a lot of money: The problem of picking creative geniuses instead of bureaucrats while avoiding the charlatans is potentially solvable. The problem of increased specialization could be solved partially if polymaths were the right type of people to solve the first problem.  If those two issues were solved, it is likely that ideological blinders are a function of the bureaucratic mindset and this issue would also fall by the wayside.  Finally, the joint issue of ethics and regulation do not seem solvable in the jurisdiction of the developed world, so perhaps they’d have to go somewhere else to address these issues.  Potentially solvable does not mean easy, so this issue deserves a lot of thought.

        Major Choices

        When thinking about the future of the country, many people say “Think of the children!” but fewer people wonder what those children are thinking.  Looking at the chosen majors of today’s graduates is one way to measure what America’s children are thinking.  Even better, their revealed preference is probably more trustworthy than surveys.

        Source: US Department of Education, National Center for Education Statistics, Digest of Education Statistics

        The biggest thing that stands out is the increased focus on the practical.

        1. Business majors are up while social science and history majors are down.
        2. Engineering technology, which prepares graduates for a job right out of school, is up while engineering majors are down. 
        3. Students have decided that they need majors for jobs that used to be done by high school graduates as they are getting more degrees in transportation, securities and protective services as well as materials moving as well as parks, recreation, leisure and fitness studies. 
        4. Education majors have shrunk, but this is probably due to cultural changes that allow women more choices when it comes to their field of study. 
        5. Students also see the increased demand for health services in their future and have gone into biology, psychology and health professions and related clinical sciences at increasing rates. 
        6. Computer science majors have increased with the emergence of personal computers. It should be noted that this major actually peaked at 4.5% of college graduates in 2003-04 with the class of students who picked their major at the height of the internet bubble.  The recent drop off in CS of 1.5% of all students choosing other majors almost exactly matches the additional 1.4% of college students choosing to major in health professions and related clinical sciences between the 2003-04 year and the 2006-07 year.

        For those worried about a decline in the quantitative focus of students, the shortfall of quantitative majors is only 1.25% of the student population compared to 1970, and this number completely disappears if it is assumed that half the double majors are doing something quantitative or a fraction of the business majors are learning about statistics.  Of course, the college graduate situation only looks good when comparing the United States to a past version of the United States.  What is worrisome about the education picture to many people is the declining relative position of the United States compared to other counties and that will be covered in a future post.

        Aging and Saving

        The Consumer Expenditure Survey by the BLS just released an interesting article highlighting their 30 years as a continuous survey.  I decided that it could be interesting to see what their data combined with Census demographic projections imply about the future trend in savings. One way to do this is to break down the savings rate by age and project it forward.

        Years: 2004-2008

        All consumer units

        Under 25 years

        25-34 years

        35-44 years

        45-54 years

        55-64 years

        65-74 years

        75 years and older

        Average Savings Rate









        As the savings rate obviously wasn’t 17.5% over the past few years, it should be noted that the CES implied measure of savings is rather different than the personal savings rate data calculated by the BEA.

        In 2004, the CES changed the way it calculated income for incomplete survey responders, so the savings bias is calculated as an average of 2004 through 2008, which is about 15%.

        The analysis assumes that the savings rate of each group will remain constant, and combines CES data with the US Census projections with constant net immigration to predict the future path of savings rate. The census cohorts are adjusted by the relative amount of consumer units because they suffer the same discrepancies as household data, with fewer young consumer units per capita.  This savings rate is adjusted downward by about 15% to be comparable to the BEA’s personal savings rate data.

        The results are interesting in their mildness. The savings rate is only expected to drop a little over 1% due to demographic factors.  Of course, this study is biased for many reasons.

        1. The accuracy of the using the aggregate consumer expenditure data without adjustment is questionable.
        2. Social security and other retirement benefits are assumed to exist at levels comparable to the past few years. This biases the savings rate of the old cohorts higher compared to a likely scenario in which government and pensions have solvency issues due to the increasing dependency ratio. 
        3. The educational and racial composition of the projections is held constant when the savings behavior of these cohorts is actually rather variable.
        4. Wealth effects are completely ignored.

        Recalculation and Healthcare

        Arnold Kling has many interesting posts on recalculation.  His theory about this recent recession is that it is driven by a need to figure out what is happening next.

        Imagine a central planner who decides to radically change plans. He has a huge recalculation to make in order to figure out where to allocate labor and capital. He says to some people, "Wait a minute. I am thinking. Some of you just have to stand idle while I figure this out."

        The market economy is like that central planner. We are undergoing a Great Recalculation

         An interesting application of this theory is that the economy will recover more slowly if is more difficult than normal for a business to invest in what is happening next within the United States.  The CBO’s 2009 long term budge outlook has an interesting chart that is not good news for most US companies focused on domestic sales.

        Source: Long Term Budget Outlook, CBO p 32.

        Their forecast is that the only spending that will increase on a per capita basis is healthcare spending. This forecast was partially made by forecasting the current excess cost growth in healthcare into the future, assuming that these costs grow less quickly past 2020 (less than half the healthcare spending is Medicare and Medicaid).  There are definitely going to be changing markets and increased productivity within the “all other spending” category, but without healthcare this is a pie that is only really growing at the rate of population growth.

        This is an increase in the size of the economy where profits are nonexistent or seen as evil.  The nature of healthcare makes it very hard to earn money in the sector. The supply of labor is restricted by licensing and thus the returns to labor are high relative to the return on capital.  In less service oriented healthcare areas the FDA approval process that doesn’t shield companies from liability increases the costs and risks to investors. The uncertainty surrounding healthcare regulation makes investing in the sector even less unfriendly than it is already.

        If the recalculation story is true and the CBO’s projections on this matter can be trusted, the United States’ market economy might remain confused and below potential for some time to come.