A lot of attention has been paid recently to the role of increasing payments to top executives (CEO's usually) in generating inequality. Public media is full of claims that CEO's are being paid "too much", particularly in the US. The Economic Policy Institute informs us that CEO's compensation in the 1960s was around 20 times that of average worker compensation, but grew to 60 by 1990 and is now almost 300. But is this happening at the within-firm level (i.e. the gap between CEO and workers in the same firm is increasing) or at the between-firm level (i.e. the gap in average payment - including CEOs - between firms is increasing)?
I would suggest that most of the public media proposes that it is within-firm inequality what is driving inequality. Even Piketty (2013, p.315) suggests that the "primary reason for increased income inequality in recent decades is the rise of the supermanager." Nevertheless, the most recent (and serious) evidence goes in the opposite direction. Working research from people in Minnesota, Stanford and the Social Security Administration has merged tax data for employees and firms in the US between 1982 and 2012 (note this covers most of the increase suggested in the first paragraph). They basically find that within-firm inequality has remained mostly flat over the past three decades.
A first look at the data is provided by splitting the population of workers in groups according to their own income. Then, for each group they calculate the mean individual income (blue), the mean of average wages at their firms (red) and the difference between the two (green). Figure 1 shows that income of individuals at the top of the distribution has grown a lot more than those at the bottom since 1982. This upward line indicates that inequality has increased. However, the fact that the Individuals (blue) line is mirrored by the Firms (red) one suggests that most of the increase in these individuals' incomes has also happened to lower paid individuals in their firm. The average income in firms of top paid individuals has increased at a similar speed than the earnings of the top paid individuals. Hence, within-inequality (or CEO's excessive payments) may not be the main driver of inequality.
Figure 1: Decomposing income growth (1982 - 2012).
Their results extend to when they focus on individuals at the top 1%. More interestingly, the result is not explained by increasing dispersion between industries (e.g. Health vs Car industry). Even within sectors, most of the rising inequality is explained by firms' dispersion. You may be thinking now about geographical differences, like California having grown more than Alaska? Well, no, that does not seem to be driving their results.
All in all, as the authors summarize it, highest-paid individuals now work at higher-paying firms, but are not higher paid relative to those firms. What explains this findings? Two ideas come to mind. First, firms may now be more specialized, leading to higher concentration of skills which could lead to bigger differences in the average workers characteristics. Secondly, growing firm productivity differentials may be generating increased differences in wages (which could generate even more sorting). However, patience is needed to find out as their research is still going...
Last time I wrote about some examples of Early Childhood Investment (ECI), actions that if done early enough in children's lives might have persistent and significant impacts on their adult outcomes (education, income, crime...). A recent study by one of my favorite economists - Raj Chetty - and coauthors provides another excellent example on how some unexpected policies could be understood as ECI.
The Moving to Opportunity (MTO) experiment offered randomly selected families living in high-poverty housing projects vouchers to move to lower-poverty neighborhoods in the 1990s. Previous studies had found that even though this tended to improve health and safety of the families, it did not have significant impacts on the earnings and employment of adults and older youth. Hence, this was taken as evidence that neighborhood environments were not really important for economic success. However, only recently have researchers become able to study the effect on those children who moved at younger ages. And results are quite different.
Chetty and Hendren had previously used evidence from movers to suggest that gains were larger for children who are younger when they move: every year spent in a better area during childhood increases earnings in adulthood. The figure below shows the percentage gain from moving to a better area by the age at which the child moves. For example, children who move at age 9 have outcomes that are about 50% between the outcomes of children who grow up permanently in the origin and destination areas.
And how big are the differences of being born and raised in one area versus another? This is particularly hard to answer because people are not born in random places. Being born and raised in the Upper West Side of NYC is probably associated with coming from a high income, high educated family. But the authors did their best to distinguish effects from each other (once again by using movers), and found that the neighborhood impacts on earnings can be over 30% (moving from bottom to top counties). In case you are thinking about moving, Table 1 shows the bottom and top 10 places in the US.
Table 1: Percentage Gains/Losses Relative to National Average.
With this evidence in hand, they went on to try to confirm their findings using the MTO experiment. The evidence from this experiment was more particular (only poor people were the target) but was also more robust to estimation problems. As mentioned before, people are not born in neither do they move to random locations. They choose were they live or move to. And this brings selection problems during estimation. The MTO experiment provided people who actually were randomly given the opportunity to move (though they were still partially able to choose the destination). And results were confirmed.
Even though the MTO experiment suggested that better neighborhoods had almost no effects on adult earnings of children who moved after the age of 13, Chetty and co-authors found that those who moved to better neighborhoods before the age of 13 had on average a 31% increase in their income. Moreover, it also increased their college attendance. Individuals with Section 8 vouchers were given the freedom to choose where to move to, while those with experimental vouchers were forced to choose places with poverty rates below 10%. As impacts were smaller for those with Section 8 vouchers, actively encouraging families to move to lower-poverty areas (and not just anywhere they want) seems to also be important for these type of policies.
Moving a child out of public housing at age of 8 would increase her total lifetime earnings by $302,000. But how expensive is this policy? With an average family of two children, the authors estimate that it amounts to a gain (net of extra costs) of almost $100,000 per child. Moreover, given the gain in tax income paid by these children in the future (extra $400 per year in the mid-twenties), results suggest that the additional tax revenue may be enough to finance the difference between public housing and moving vouchers costs. Whether this policy is still worth it in simple gains-costs terms for those who are not in public housing remains to be seen. However, I believe benefits other than earnings, like health, education or crime, should be considered. Moreover, the effect on future generations (the children of those moved) should also be taken into account as children of higher educated and higher income parents tend to do better in life.
Based on an article by Chetty, Hendren and Katz.
It is well known that children's socioeconomic background shapes their outcomes in school, both intellectually and behaviorally. However, most policies focus on improving schools, taking those disadvantages as given and simply acknowledging them as an excuse for inadequate performance. But another alternative is to actually improve these early background conditions. And, at least as I interpret them, most such policies could be called "Early Childhood Investment." For the sake of clarity, let's go over some examples of early social disadvantages and policies that are being (or could be) used to reduce them.*
Background: Home Intellectual Environment
Literacy activities at home - like reading aloud, telling stories or doing art - predict better social skills, fewer teacher-reported behavioral problems and higher literacy rate. Table 1 takes race (white vs black) as the splitting source for children's background. It is seen that white adults spend 36% more time reading to young children than black parents. Raikes et al (2006) show that toddlers of low-income mothers who read to them daily have better vocabulary and comprehension at 24 months. Table 1 also reports that white parents interact over 200% more with their children in activities that develop their critical thinking and creative skills (like puzzle solving, arts or simply talking).
Table 1: Home intellectual environment (children age < 5).
Another standard proxy for intellectual environment is the number of books at home (though it is self-reported by parents, so it is not as objective as one would like). Table 2 shows that the better the socioeconomic status [SES] the higher the number of books. Similarly, white children live in households with 150% more books than black children. Behavioral skills of entering kindergartens in the top and bottom SES differ by 10-23 percentile points, and gaps do not disappear over time: over 30 ppt lower high-school completion rates and 15 ppt higher arrest rates are observed.
Table 2: Books at home, by race and socieconomic status.
By age 6, white children are reported to have spent over 1300 more hours engaged in conversation with adults than black children. Interaction with parents is one way children's choice-making and stances toward authority are developed. Upper-class parents typically give fewer orders, letting children develop their critical thinking and choice-making. On the other hand, lower-class parents expect more deference to authority. Similar differences are found for recreational activities between affluent and low-income families. Lower class children typically have more unstructured leisure time. Middle class children are more used to structured leisure, which is thought to help them operate later in controlled environments like classrooms.
Early Childhood Investment: Policies
Improving schools is certainly useful. However, coordinating school improvements with other community services that reduce these early disadvantages seems a very promising way to go. Here are some examples:
A clear pattern of these policies is that they do not simply target children. They also aim at teaching parents to get better involved in the development of their children. Taking into account that children spend most of their time at home (rather than at school), this makes a lot of sense. Moreover, it is also noticeable how early these programs start. Some begin even before children are born, and there is some consensus that the earlier the start the better the results.
Obviously no background influence is completely definitive. Some children with "bad backgrounds" may do better than some with "good" ones. We all know stories of some person who started from nothing and became a big success. But socioeconomic background does influence opportunities (or distributions of outcomes) and, as a summary statistic, it is seen that average outcomes of children tend to go hand in hand with their backgrounds. Complementing improvements at schools with reduction in backgrounds' inequalities, through some of the policies mentioned above, may be the way to go.
* I focus here on home intellectual environment characteristics. Other issues like wealth, housing, neighborhood, malnutrition or health provision may be equally important. Moreover, all these characteristics interact (household with positive intellectual environment tend to have high wealth and live in good neighborhoods for example) making precise estimates on the impact of each background element not possible, so numbers in this post should be read with caution (although measures are taken to reduce these issues).
Based on a recent Economic Policy Institute article.
Early childhood, when brain plasticity and neurogenesis are very high, is a key period for the development of skills (both cognitive ans psychosocial). Differences in this early investment might be the seeds of inequalities observed later in life across countries and within countries. More than 200 million children under the age of 5 are at risk of not reaching their full potential, the majority of them living in extreme poverty. Starting from a disadvantaged point, Figure 1 shows that remediation later is less effective and more expensive than early interventions.
Figure 1: Returns to investment by age.
Source: Heckman (Economic Inquiry, 2008)
An experimental study in Jamaica, recently published in Science, has provided more rigorous evidence on the benefits of early childhood investment. Growth-stunted children - reliable indication of severe economic disadvantage in developing countries - between the ages of 9 and 24 months participated in a 2-year randomized controlled trial. Usually we cannot compare the benefits of parents who themselves decide to promote their children's situation because of selection issues. For example, parents who invest in education are probably different in many characteristics from those who do not (maybe more motivated or caring for example). Hence, we would be comparing both the increased investment as well as parents "quality." To avoid this issue, the Jamaican study randomly assigned children to one of four groups: 1) psychosocial stimulation (explained below); 2) nutritional supplementation (1 kg milk-based formula per week); 3) both psychosocial stimulation and nutritional supplementation; and 4) a control group that did not receive treatment. Another group of non-stunted children was surveyed to provide a external comparison group, so that the question "Were these children able to catch-up?" were possible to answer.
Groups 1 and 3 participated in a program that consisted of two years of weekly one-hour sessions at home focused on improving parent and children interaction, with innovative activities to develop child cognitive, language and psychosocial skills. They were also encouraged to practice the activities and games learned during the visit beyond the home visitation time. The particularly interesting element of this project was the length of time for which children were studied. Participants were interviewed 20 years later and evaluated on a number of economic indicators. While stunted growth is largely due to a lack of nutrition, the nutrition intervention alone didn't affect later adult economic outcomes. The nutrition-only group showed no long-term effect on any measured outcome, but this may be due to the sharing of the nutritional supplementation within the family. On the other hand, psychosocial stimulation seems to have brought substantial long-term benefits.
In addition to improving direct parent-child interaction during the 2 years of the program, it seems that it also promoted greater parental investments later in life with an improved home environment, which probably improved later outcomes. Years of education increased by an average of 0.6 years by age 22, while college attendance increased three-fold. Even though they were also more likely to be in school (and hence working below their potential), their income was between 40 and 60% higher among those who received early psychosocial stimulation. Early intervention also reduced involvement in violent crimes.
Finally, could they catch up to remediate their initial disadvantages? Yes. The differences in income between children who were disadvantaged but received early stimulation and those who were not disadvantaged was zero (or not significantly different from zero). Early improvements probably encouraged the families to seek greater education and employment opportunities. 22% of the families in the treatment group emigrated to countries with great opportunities for upward mobility, compared with only 12% of the control group families.
These results are remarkable. Returns to early intervention are close to 10%, much higher and more persistent than other alternatives later on in life. So why are policies concerned with inequality so focused on re-distribution later on in life? Maybe instead we should try to pre-distribution during the earliest stages of life.
What are inequalities around the world associated with? Some typical answers include: gender and race discrimination, education or family background. However, Branko Milanovic's study of global inequality shows that over 2/3 is actually explained by country of origin. Differences in the qualities (read GDP per capita levels) of countries are so massive that they translate to big differences in opportunities for individuals. This citizenship premium is what leads to thousands of people every year to spend all their savings, risk their lives and venture across oceans with the slim hope of reaching a better place. (The recent case of almost one thousand migrants dead crossing the Mediterranean with illegal traffickers is a clear example)
Up until the First World War traveling between countries was simpler, without the formalities of passports or strict border controls. Open borders was the norm. The terrible period until the Second World War led to stricter controls for all flows between countries, including investment and trade as well as people. Nevertheless, after 1945 the world has managed to recover some freedom in the flow of the first two, but free migration of individuals has not been recovered. This was argued by Lant Pritchett at the recent Harvard IDC conference to have lead to massive gaps in the wages of observably equivalent (e.g. same nationality, age, sex, education) labor around the world. Branko Milanovic's research could be read as people born in "bad" countries are not able to access proper education so they are doomed to low income, no matter where they live. However, Lant argues this is not true as differences in productivities (even for low skill workers) between countries are big enough to compensate for that education. Together with Micheal Clemens and Claudio Montenegro, they were able to compare the wages of people with same characteristics in their country of origin and the US (say Peruvian-born, Peruvian-educated 35 year old with nine years of schooling living in Peru versus living in the US).
Table 1: Wages of observably equivalent workers at country of origin and US.
Table 1 shows that the apparently same workers make in average five times as much in the US as in his home country, which amounts to over 15 thousand dollars a year (even after adjusting for price differences). A well trained economist would argue that selection could be a problem: the individuals that decide to migrate could have more ambition or drive, making them more productive even if they had stayed at home. This could bias the results upward, but the authors suggest that controlling for that would amount to (at most) dividing those results by a factor of 1.4. Even taking selection into account, letting a person move from his home country to the US would amount to an increase of around four times in his income.
These results are impressive. In order to get a sense of their magnitude it is useful to compare them with other suggested measures to reduce poverty in developing countries. How long would a person have to work in the US in order to equal the benefits from a lifetime of other programs or interventions? Table 2 shows that a Bangladeshi would need less than a month in the US to be able to make as much extra money as if she gained the net present value of a lifetime of access to micro-credits (one of the supposedly most successful interventions available, pioneered by Nobel Peace Prize winner Muhmmad Yunus). Similarly, an Indonesian worker would have to work half a year in the US to equal a lifetime of gains from the anti-sweatshop movement.
Table 2: Gains from international movement of workers relative to other policies applied at home country.
Nowadays we focus on the recent "globalization" trend of opening trade and investment opportunities across the world. However, comparing with the world we had before 1914 we clearly live in a world of "nationalization", where flows of individuals across borders are prevented by coercion. Changes at the margin of labor mobility seem capable of leading to impressive results in reducing global poverty and inequality. The question that remains open is whether major changes in labor mobility would lead to similar results: It could be the case that allowing for just one extra migrant leads to impressive results because there are 10 employers looking for workers, but open borders could lead to thousands of migrants which could outnumber the positions required by those employers. But the magnitude of the benefits highlighted in Pritchett's work seems to suggest that, at least for the migrants, the benefits may be massive anyways.
What about the workers from the recipient countries? That's an open question, but there is no clear evidence (from academic research) that they would be damaged. Maybe a possibility would be to start by allowing migrants for positions not desired by Americans (e.g. elderly or child care) or imposing an extra immigrant-labor-tax for the first years in the US that could be used to compensate national workers. This would introduce needed workers or tax revenue benefits for the recipient countries on top of allowing people from poor countries to improve their lives and avoid the many deaths due to illegal trafficking of migrants.
Most of us think about inequality within the borders of countries. But why shall we do so? Another very interesting perspective is offered when we go beyond those arbitrary limits and explore inequality across all individuals in the world. Moreover, as the world gets more and more integrated, this dimension of inequality becomes increasingly relevant. This perspective is the one I got the opportunity to participate in at Harvard's 2015 IDC conference with Branko Milanovic. I will try to summarize those ideas here.
There are many ways to look at global inequality. The simplest forms refer to comparing GDP per capita between countries. This alternative abstracts from the fact that Liechtenstein is almost 35 thousand times smaller than China. Weighting by population may be a solution for that but even then it would miss the fact that within each of those countries there is also inequality. Most people do not earn exactly their own countries average income, so we should also include this. A few people in China earn a lot, while lots of them make a lot less than China's average income. However, taking this into account requires having household surveys in at least 120 countries to cover at least 90% of the world's population and 95% of world's income. Fortunately, this has been explored by Branko Milanovic and for the years after 1980 we can get a decent estimate of this global inequality. A typical way to look at inequality is through the Gini coefficient, where a higher Gini means more inequality. As Figure 1 shows, global inequality is much higher than even the one from within very unequal countries like Brazil. (All figures adjust for different levels of prices between countries.)
Figure 1: Global Gini compared to selected countries Ginis.
To get a sense of what a Gini of about 0.7 means we can picture splitting the world's income in two: the first half is kept by 8% of the world, while the other half is given to the bottom 92%. A serious level of inequality. And this is true even when Africa is under-represented in the sample.
How has globalization affected global inequality? Figure 2 explores this by showing the percentage change in real income for each percentile of the global income distribution, between 1988 and 2008. The largest gains are observed at the top 1% as well as among emerging middle class. As expected, 200 million Chinese, 90 million Indians and 30 million people each from Indonesia, Brazil and Egypt are found between the 50th and 60th percentiles, the largest winners. The Top 1% includes 60 million people, mostly populated by the top 12% Americans and between 3 and 6% of other major OECD countries. The biggest losers are among the bottom 5 per cent, mainly African countries, and those between the 75th and 90th percentiles, including mainly former communist countries and Latin America. It's important to remark that this does not mean that the poorest have remained stuck at the bottom. Since this analysis does not follow people over time, it is possible that the poorest in 1988 were not the poorest in 2008. That has to do with intergenerational mobility which is out of the scope of this research. Nevertheless, this shows that the very bottom has not moved much, while the middle and very top have moved up.
Figure 2: Change in real income between 1988 and 2008, by percentile of global income distribution.
In the 19th century, global inequality appears to have had a lot to do with people's class within their countries. The country itself didn't matter much. Nowadays, what matters the most is in which country you were born. The current consequence of this is clear in Figure 3, where the population of each country is divided in groups of 5 per cent (ventiles) and their income is placed within the world's percentile distribution. For example, the top panel of Figure 3 shows that the poorest 5% in the US are around the 60th percentile in the worldwide distribution (among the world's top 40%). It's important to remark that this may be considered the poverty threshold in the US. For comparison with other countries, the dashed line shows their position. The top 5% of Indians barely reach this level. So the top 5% of Indians are almost at the same level as the bottom 5% of Americans. (Obviously there are very rich Indians, but the top 5% of Indians includes 60 million people.) Focusing on the second panel, the bottom 5% in Italy is similar to the one from the US. However, Germany's poorest are shown to be a lot better off. Both Argentina and Brazil mimic the worlds distribution of income. The poorest Argentinians/ Brazilians are among the worldwide poorest, their middle classes are between the 70th and 80th percentiles and their high classes are among the worldwide top.
Figures 3: Global inequality in US and Italy vs other countries.
"Location, location, location" is usually said to remind people that location is the most important factor for the price of a house or the revenue of a shop. But location, the place people are born in, also seems key for income levels across the world. What are the advantages provided by a better place? They probably include education, health care, financial systems and inherited wealth. Regardless of your political views, countries usually try to limit inequalities within their frontiers (through progressive taxation, subsidized education, etc) while most of the differences are actually reflected across those borders. Maybe open borders is the solution to much of the inequality we see around the world...
Based on an article by Branko Milanovic. Subjective political comments are my own.
Last week I posted on how much inequality has increased in the last 4 decades. The main conclusion then was that demand for skills, recently polarized by computerization, had driven much of the increase in wage inequality. Figure 1 shows that skills are actually rewarded all over the world, so this story is potentially not unique to the US. Cognitive skills are substantially rewarded across all 22 countries, with an average return of 18% for a "unit" (one standard deviation) of skills. The US is at the top of this with a premium of around 28%, implying a difference of around 50-60% between some one standard deviation above and someone one standard deviation below the average. If wages were determined mainly by luck, beauty or family connections, we would expect to see little connection between cognitive ability and wages. The high returns to skills all over the world confirm that this is not the case. And so inequality of income, is most likely highly related to inequality of skills on top of the demand shifts I explored last week.
Figure 1: Wage returns to skill, 2011-2013.
Education attainment has increased substantially in the US during the previous century, but it appears that has not been enough to catch up with the high increase in demand for some of those skills. For example, from 1963 to 1982 the fraction of hours provided by by college educated workers grew by around 1% per year. But since 1983 this increase has reduced to less 0.5%, which could partially explain the steep increase in the college premium (and particularly the post-college premium). Why has this happened? One hypothesis is that the Vietnam War might have over-boosted college attendance as it helped people skip the draft. However, this sharp (and unnecessarily high for the market) increase in supply of highly educated workers reduced the wage of college educated workers. As the returns to college became smaller, the following young generations may have been misled to believe that college was not worth it. Even nowadays, there are claims of a "college bubble" where young people are wasting too much money to get a degree. Even though it is possible that some the knowledge acquired is not useful, the market rewards it. Figure 1 showed that skills are rewarded, and Figure 2 shows that these rewards outweigh the extra costs of college. The net value of college is at its peak. Nowadays, the message of the need to get more education appears to have gotten through as the supply of college graduates has started to increase faster.
Figure 2: Lifetime value of college relative to high school degree, net of tuition cost.
But if inequality itself - through the difference in wages - is what partially leads young generations to educate themselves, what is wrong with it? A market economy may need inequality to provide incentives but this may have other costs. Intergenerational mobility is a way to evaluate the degree to which individual economic fortunes depend on their parents - for example the likelihood that children born to a low-income family become high-income adults and vice-versa. High economic inequality need not have low intergenerational mobility if the society is dynamic, with lots of movement up and down the economic ladder. But one concern is that high inequality at one point in time may serve to reduce mobility over time, and hence create a dynastic society if currently wealthy households are able to "buy" success for their children. This way inequality could become self-perpetuating, even if it originally starts from simple high market returns to skill. Figure 3 shows that this is a real concern, as societies with more income inequality have lower intergenerational mobility. And countries with higher college premiums also tend to display lower mobility.
Figure 3: Income inequality and intergenerational mobility.
Original figure is from Corak (2013)
When the return to education is high, children of better-educated parents have two advantages. They are more likely to receive a higher education and they are more likely to be more rewarded for that when they become adults. And recent work by Chetty and others (also in a previous post), suggested that intergenerational mobility does not seem to have improving over time in the US. I think this is the biggest risk of inequality. Inequality per se is not the problem. The problem is poverty and lack of opportunities, but inequality could be a big driver of the persistence of these issues among a group of people who are "doomed" from the very beginning of their lives.
Based on an article by Autor.
A lot of attention (myself included) has been recently put on the Top 1% income and wealth. However, there is also substantial inequality in the other 99% that is worth exploring. To get an idea, if we took all the Top 1% income growth between 1979 and 2012 and distributed it among the other 99%, each of us (I assume you also belong to the other 99%...) would earn around $7000. However, the increase in the earnings gap between a college-educated and a high-school educated household is four times that in the same period. Hence, here we will focus on wage inequality among the other 99%, but particularly between the bottom 10% and top 90%, so as to exclude the very extreme cases (which deserve a different attention). But first, Figure 1 shows how wages have changed between 1963 and 2005 by wage percentile. Here we see that generally there was a much bigger increase in wages among the top half than the bottom one.
Figure 1: Change in real wages by percentile, 1963-2005.
A common measure for overall inequality is the ratio of those at the 90 and 10 percentiles. A typical issue is that the population might be changing its structure, with more people getting educated or more work experience. As this happens, the typical person in either of these percentiles might be changing, hence changing our standard interpretation of increasing inequality. Another take on inequality is to look at between-groups inequality, where the typical comparison is those with a college degree and those with a High School degree. This tries to avoid the issues of other characteristics of the population changing as in the overall inequality measure. However, another alternative look at inequality is to look within-groups, hence evaluating how much variation there is among small groups (for example: college educated, 25-30 years old, male). This three measures of inequality are displayed in Figure 2, where we see that even though the three of them have increased over the long haul, they have done so at different paces and through different paths. Particularly the college premium follows a strange path, increasing in the 1960s, decreasing in the 1970s and increasing very fast since the 1980s. This suggests that a simple, unique explanation for the recent increase in inequality is not likely to work.
Figure 2: Three measures of Income Inequality.
But has inequality changed more among the top or the bottom? An easy way to look at this is to compare the 90 and 50 percentiles (upper-tail inequality) and, separately, the 50 and 10 percentiles (lower-tail inequality). Note this still excludes the very bottom and very top. Figure 3 shows that even though lower-tail inequality grew in the 1980s, it has not grown since then. On the other hand, upper-tail inequality has increased continuously.
Figure 3: Upper- and Lower-Tail Inequality.
So what is behind this monumental increase in inequality? Identifying the cause of this change is very hard or probably impossible, but what we can do better is identify the proximate causes, meaning what seems to be closely associated with this change, even if we do not understand what led to the first thing. And this is where the skills of economists Autor, Katz and Kearney comes into play. They argue that we cannot simply think of people as belonging to one of two groups - skilled and unskilled - where the top one is associated with higher education. Figure 4 shows that from 1979 to 2005 the wages of those with a post-college education grew by a lot more than those with college degree. Moreover, the difference between those with exactly college and high-school degrees slowed down significantly since the 1980s. And finally, the difference between those with high school degrees and those without one has flattened or even decreased since the mid-1990s. All this suggests that, since the 1990s we are in a situation where the income among the very high- and very low-skilled workers has increased relative to those in the middle. Income has polarized.
Figure 4: Changes in wages by Education.
What explains this? The main hypothesis is that computerization has changed the demand for job tasks and affected the demand for skills in such a way that explains this polarization of income. Computers are good at doing routine tasks which are codifiable, like bookkeeping, clerical work or repetitive production tasks. (If you have interacted with a call-center lately, you will probably know how computers have improved in voice recognition and seem to have taken over those tasks that require gathering the same information all the time). On the other hand, abstract tasks like those performed by "high-skills" managers or educated professionals are hard to automatize since they require cognitive and interpersonal skills and adaptability. Similarly, manual tasks used in many "low-skilled" jobs like security guards, cleaners and servers are hard to computerize and hence have not been affected much by the advance of computers. Figure 5 confirms this intuition that low-skill jobs (taking the average education of those performing such jobs) usually have manual tasks. On the other end, high-skill jobs are mainly filled with abstract tasks. However, routine tasks are concentrated between the 20th and 60th percentiles.
Figure 5: Task intensity by Occupational Skill.
The conclusion is that the change in wage inequality may be substantially explained by changes in the demand of skills, which has been lately polarized by the introduction of computers. As the demand for these jobs increased, so did their wages. But why haven't workers matched the increase in demand by educating themselves more? Well, most likely this change was very hard to predict and so not enough people found higher-education to be "worth it." However, recent trends in education attainment suggest that young people are catching up to this increased demand.
Based on an article by Autor, Katz and Kearney.
Most professionals, from doctors to policy makers, are subject to biases that can prevent them from helping the people they are actually seeking to assist. Previous beliefs or personal culture can prevent them from objectively analyzing the information they have in front of them. This can be particularly serious in professions where data analysis plays a major role, like development professionals. Fortunately, the World Bank - an institution full of such people - decided that instead of pretending such issues were not present, they would release a report where they collected data examining how their staff behaved, testing if the biases were actually there.
Decisions can be very complex. Suppose you have a problem at hand, which is particularly messy. How do you compare alternatives in such situations? And what if alternatives are almost infinite? Even in simpler cases, going from two to three alternatives has been suggested to "irrationally" change professionals actions: doctors were outlined a situation where a patient had two alternatives: 1) Ibuprofen + Referral (to specialist); or 2) Just referral. In such a case 53% of doctors chose option 2. Another set of doctors was given a third alternative: 3) Piroxicam + Referral. Paradoxically, now many more doctors (72%) chose the previously available option of just Referral. Complexity modifies their behavior, leading them more to the simpler solution of just referral. Since the only difference between the two cases was that a new (possibly "irrelevant") alternative was added, we should expect less doctors choosing any of the previously available ones. But this did not happen. And this are highly educated and experienced professionals!
Framing (basically how something is described) is also unbelievably important: people seem to have much negative views of a policy that says 1/3 of people will die versus one that says 2/3 of them will survive. Obviously, the two are exactly the same. Moreover, when framed in terms of gains, people prefer certainty ("1/3 of people survive"). However, when framed in terms of losses ("2/3 of people die"), people are more willing to take a gamble and try to save more people (1/3 chance no one dies and 2/3 chance everyone dies"). Does this happen at the World Bank as well? Certainly.
Confirmation Bias means that people gather or value information selectively in order to support their previous beliefs. We all have cultural and ideological priors, which leaves us susceptible to analyze or interpret information with motivated reasoning, so as to arrive to the conclusions we like. A very nice experiment was held where professionals were given the same data to analyze. To some people they told it referred to the effectiveness of a skin cream (something for which we don't have priors or political beliefs) and others were told it was about the impact of the minimum wage laws on poverty. Figure 1 shows that World Bank professionals were more likely to identify the answer supported by the data (i.e. the "correct" one), when talking about skin cream. (I am also shocked by how low the percentage of right answers is when talking about skin cream, but I guess that's another issue...). When dealing with the minimum wage they were more likely to get the right answer when this matched their cultural views. And differences in seniority or cognitive ability did not improve the interpretation of the data. This was also done with people outside the World Bank and they actually found that skill helps only when the right answer matches your ideology...
Figure 1: Subjective interpretation of data.
We could take all this information in a negative way: some of the best people in charge of world development are not doing a good job. However, I believe that's not correct. This bias is present in all of us. And the positive thing is that the people at the World Bank now seem to be willing to recognize it and hopefully move forward to try to make progress. Identifying the problem is the first step, or as John Dewey said a problem well put is half solved.
After a long long time devoted to education, economists do need to look for a job. But they (generally) do not do it in the standard way: calling, sending CVs and so on. There is something called the Job Market that takes place every year early in January. Obsessed with efficiency, the Economics Job Market has a particular advantage: applications and initial interviews are centralized. Most of them take place at the American Economic Association annual meeting. And after some very stressful days, interested employers call back and schedule fly-outs for February-March. After meeting them, going for drinks and dinner, and also presenting your research, job offers are determined. But the question I have is what determines the outcome of this stressful process? As a person that will hopefully eventually go through this ordeal, I wondered if there was any data about it.
Even though all economists have experienced this, I wasn't able to find much research about the job market unfortunately. But I found one paper where they asked what aspects of education are associated with good outcomes in the job market. They collected data on graduates from Top departments (Harvard, MIT, Princeton, Stanford and Chicago) and checked what was associated with the best job outcomes. Obviously, the sample of graduates coming from those departments is not representative of all economics graduates. Since they were accepted in such departments, they are most likely representative of the very top of the distribution of applicants to PhDs. Studied by academics, another caveat is that job placements in the business sector were generally assigned a much lower ranking than university ones (a good business sector job was similar to a university in the 200-250 rank). But well, the questions are:
0) What is the typical PhD graduate? 1990-1999.
He (only 25% female) is a foreigner (63% non-US) who might come from a foreign undergrad school (49%). Most likely he does not come from a top undergrad school though (22% coming from top-15) nor does he have a masters degree (24% with masters). 3 out 4 admitted students do graduate the PhD. And around 26% of (this very selective group of) graduates end up in a Top-20 school. The sample is a bit old and selective unfortunately and some things might have changed. Unfortunately, one has definitely not. It is still mainly male students.
1) Do admissions requirements matter for grades?
Before entering, a standardized exam called GRE is required. This has three parts: math, verbal and analytical. GRE math and analytical grades - even within this group of people with really high ones - are highly positively associated with good core grades in the PhD program. I always thought this was more of a filter requirement: once above it, all students would be pretty similar. But it seems not.
Coming from a Top-15 US university is not associated with better grades. A masters degree helps slightly. And coming from a foreign school is correlated with better grades. But this may be due to a much more selective procedure for students coming from abroad. Or from them being more devoted since they are willing to leave their home countries.
2) Do grades matter for graduation?
First, grades are highly correlated: if you do well in one, you also do well in others. I find this very interesting since we are looking at people who will later on focus on a very very tiny part of the world of economist, so we could have expected that people doing great in one Micro would not do well in Macro, or viceversa. Let me clarify that grades are only a small part of the PhD. Most of it is actually doing research, which is what most graduates will do afterwards in their careers. But core micro and macro - sorry econometrics! - grades certainly seem to matter for graduation. Even (sort of) when restricting to those who passed the courses requirement, goods grades were associated with graduation.
3) And finally, what matters for job placement?
A) Observable before starting the PhD.
Once again, coming from a foreign university is positively associated with landing a Top 20 job. Coming from a Top school in the US is also good. GRE not so much anymore. (Being a man or a woman does not seem important either, so maybe we have a hope.)
B) Observable after starting the PhD.
Micro and Macro core grades are good predictors of job placement - sorry econometrics again. Admissions rank does not seem relevant, which might question the capacity of departments to rank students. Conditional on grades, coming from a foreign school does not seem to matter as much. But coming from a Top US school still does. I wonder if a language or culture bias could be behind this...
The questions that remains are why are some characteristics much stronger predictors of grades than of job placements? If what really matters for the outcome of PhD students or the evaluation of the department is the placement, why does the admission procedure seem quite ineffective in predicting it? And, finally, what's wrong with econometrics?
Based on article from AEA.
Follow me on twitter at @diedaruich
News and posts for an active mind.