Good Economics for Hard Times

by Abhijit V. Banerjee

  Second, in most developing countries, the average person would also certainly like a stable job with a good income and benefits, but it is not what they think they are entitled to. A very large proportion of the world’s poor and near-poor, who essentially all live in the developing world, are self-employed. They don’t like being self-employed, but they are used to it. They know they might have to switch from one occupation to a very different one within the space of a month or even a day, depending on what opportunities are available. They sell snacks in the morning and work as seamstresses in the afternoon. Or work as farmers during the monsoon and as brick makers during the dry season.

  Partly for that reason, they do not build their lives around their work; they are careful to maintain connections with their neighbors, their relatives, their caste and religious groups, their formal and informal associations. In Abhijit’s native West Bengal, the club (or in Bangla pronunciation, the klaab) is a key institution; most villages and urban neighborhoods have at least one. The members are men between the ages of sixteen and thirty-five; they meet nearly every day to play cricket or soccer or cards or the uniquely South Asian tabletop sport carrom. They often describe themselves as social workers, and when, say, there is a death in some family, they show up and help. But they also practice a mild form of extortion in the name of “social work” or religious observances, and this, along with contributions from local politicians who use them as foot soldiers, pays for the club and its occasional celebrations. Mostly, however, it serves as a way to keep the local bloods from getting into more trouble, in a setting where most of them are either not working or working at a job they don’t enjoy. It provides a modicum of meaning.

  BEYOND FLEXICURITY

  If UBI won’t solve the disruption caused by our current economic model, what will? Economists and many policy makers like the Danish model of “flexicurity.” It allows for full labor market flexibility, meaning people can be laid off quite easily whenever they aren’t needed anymore, but the laid off are subsidized so they do not suffer much of an economic loss, and there is a concerted effort by the government to get the worker back into employment (perhaps after meaningful retraining). Compared to a system where workers are essentially on their own (like in the United States), flexicurity is meant to ensure job loss is not a tragedy, but a normal phase of life. Compared to a system that makes it difficult to fire workers on permanent contracts (like in France), flexicurity makes it possible for employers to adjust to changes in circumstances, and avoids the conflict between the “insiders,” those lucky enough to have strongly protected jobs, and the “outsiders,” who have no jobs at all.

  This is consistent with the economist’s basic reflex: we should let the market do its job and insure people who end up holding the short end of the stick. In the long run, preventing reallocation of labor from shrinking sectors to growing ones is both impractical and costly. For many people in the economy, particularly the younger worker, any help to seriously retrain is valuable. We saw earlier that the TAA program worked.

  Nonetheless, we don’t think that flexicurity is the entire answer. This is because of what we already discussed; job loss is clearly much more than income loss. It is all too often about being yanked from a settled life plan and a particular vision of the good life. In particular, older people and those who have worked in a particular location or for a particular firm for many years probably find it more difficult to switch to another career. Retraining them is costly, given they have relatively few years of work life left. They have a lot to lose and little to gain from moving to another career (and even more to another place). The only relatively easy transition would be to move to another role in the same area and in a similar position.

  This is why at the end of chapter 3 we proposed the somewhat radical idea that some workers should be subsidized to stay in place. When a whole sector is disrupted by trade or by technology, the wages of the older workers could be partly or fully subsidized. Such a policy should only be triggered when a particular industry in an area is in decline, and reserved for older employees (above fifty or fifty-five) with at least ten (or eight or twelve) years of experience in a comparable position.

  Economists are instinctively critical of opening up such a large space for governmental discretion. How will the government know what the declining industries are?

  We don’t doubt there will be some errors and some abuse. However, this has been the excuse for not intervening all these years when trade has been robbing people of their livings while claiming to make everyone better off. If we want to claim trade is good for everyone, we need to design mechanisms to make it actually so, and those will involve identifying losers and compensating them. In fact, trade economists (including those in government) have the numbers to know where imports are growing fast and where outsourcing is growing apace; the round of tariffs imposed by the United States in 2018 were computed from this data. A trade war risks hurting a lot of other people in the economy, whereas a much more targeted subsidy would protect the most vulnerable groups without creating new forms of disruption. A similar policy for identifying sectors and locations where automation is happening fastest, and intervening, could also be put together.

  Prominent urban economists, like Moretti, are suspicious of place-based policies because they worry the policy will just end up redistributing activity from one region to the other, and possibly away from the most productive regions to less productive regions. But if people over a certain age cannot or will not move, then it is not clear what choice we have. Today, large pockets of left-behind people are dotted across the US landscape, with hundreds of towns blighted by anger and substance abuse, where everyone who can afford it has either left or is contemplating leaving. It will be very difficult to help people in these places. The goal of social policy, therefore, should be to help the distressed places that exist, but perhaps more importantly avoid ending up with many more.

  In a sense, this is what Europe has done with its Common Agricultural Policy. Economists hate it, because a dwindling number of European farmers have gotten a great deal as a result of subsidies at the expense of everyone else. But they forget that by preventing many of the farms from shutting down, it has kept the countryside in many European countries much more verdant and vibrant. In the past, since farmers were paid more to produce more, they had a tendency to intensify agriculture, giving rise to large ugly fields. But since 2005–2006, the amount of assistance given to farmers has not been linked to production. It is based instead on environmental protection and animal welfare. The result is that small artisanal farms are able to survive, and from them we get high-quality produce and pretty landscapes. This is something most Europeans probably think is worth preserving and certainly contributes to the quality of their lives and the sense of what it is to be a European. Would French GDP be higher if agricultural production were more concentrated and farmhouses were replaced by warehouses? Possibly. Would welfare be higher? Probably not.

  The analogy between protecting manufacturing employment in the United States and protecting nature in France may seem strange. But pretty countrysides attract tourists and keep young people around to take care of their aging parents. Similarly, the company town can ensure there is a high school, some sports teams, a main street with a few shops, and a sense of belonging somewhere. This is also the environment, something we all enjoy, and society should be ready to pay for it, just as it is willing to pay for trees.

  SMART KEYNESIANISM: SUBSIDIZING THE COMMON GOOD

  In 2018, a very different approach based on subsidizing work is gaining ground in the US Democratic party. In 2019, presidential candidates Cory Booker, Kamala Harris, Bernie Sanders, and Elizabeth Warren have all proposed some kind of federal guarantee, whereby any American who wanted to work would be entitled to a good job ($15 an hour with retirement and health benefits on par with other federal employees, childcare assistance, and twelve weeks of paid family leave) in community service, home care, park maintena
nce, etc. The Green New Deal proposed by Democratic members of Congress includes a federal job guarantee. The idea is of course not new; the Indian National Rural Employment Guarantee Act works along the same lines, as did the original New Deal.

  Such a program is not easy to run well if the experience in India is any guide. Creating and organizing enough jobs would probably be even harder in the United States, given that very few people want to dig ditches or build roads, which is what they are asked to do in India. Also, the jobs would need to be useful. If they were transparently some form of make-work, they would not boost the employees’ self-esteem. Between pretending to work and going on disability, they might still choose the latter. Finally, given the required scale of the program, it would have to be implemented by private companies bidding on government contracts, known for delivering low quality at a high price.

  A more realistic strategy may be for the government to increase the demand for labor-intensive public services by increasing the budget for those services without necessarily providing them directly. An important consideration, especially in the developing world, is to not create jobs where people are underworked and overpaid. As we already saw, the presence of such jobs freezes up the labor market, because everyone queues up to get them. This has the result that overall employment may actually go down. The jobs need to be useful and compensation needs to be fair. There are many possibilities. Elder care, education, and childcare are all sectors where the productivity gains from automation are, at least for the moment, limited. Indeed, it seems likely that robots will never be completely able to replace the human touch in caring for the very young or the very old, though they may well complement it effectively.

  Another reason why humans will be very hard to replace in schools and preschools is that if robots take over all the jobs requiring narrow technical skills (from screwing in bolts to accounting), people will be increasingly valued for flexibility and natural empathy. Indeed, research shows social skills have become more valued in the labor market in the last decade compared to cognitive skills.62 There is very little research on how social skills can be taught, but it seems common sense that human beings will retain some comparative advantage over software in teaching social skills. Indeed, an experiment conducted in Peru shows boarding-school students who were randomly assigned beds near highly sociable students gained social skills themselves. In contrast, being assigned a neighboring student with good test scores did not help them get better grades.63

  The comparative advantage of humans in care and teaching means the relative productivity of these sectors will increasingly lag behind as machines gain hold elsewhere, and they may also attract less private investment than sectors where more rapid productivity gains can be achieved. At the same time, the care of the elderly is definitely a worthwhile social goal that is currently underserved, and there are enormous potential gains for society from investing in better education and early-childhood care. This will cost money; these two sectors alone could probably absorb as much money as a government would be willing to spend. But if this is money spent paying people well for stable, well-respected jobs, it will achieve two important goals: producing something useful for society and providing a large number of meaningful occupations.

  HEAD STARTS

  The intergenerational mobility of children is tightly linked to the neighborhoods in which they grow up. A child born in the bottom half of the income distribution in the United States will on average reach the forty-sixth percentile of income if he grew up in Salt Lake City, Utah, but only the thirty-sixth percentile if he happened to be from Charlotte, North Carolina. These spatial differences emerge well before an individual starts working: children in the low-mobility zones are less likely to attend college and are more likely to have children of their own early.64

  In 1994, the US department of Housing and Urban Development launched a program called Moving to Opportunity (MTO) that offered public housing residents the opportunity to participate in a lottery giving them the chance to move from high-poverty housing projects to lower-poverty neighborhoods. About half of the families who won the vouchers took advantage of it, and those who did ended up in much less poor neighborhoods.

  A team of researchers was able to follow winners and losers of the voucher lottery to see if anything changed as a result. The early results for children were somewhat disappointing: while girls were in a better mental state and did better in school, the same was not true of boys.65 However, in the longer term, some twenty-odd years after the initial lottery, large differences in their life outcomes were evident. Young adults whose parents won the vouchers earned $1,624 more per year than those whose parents did not. They were more likely to have gone to college, they lived in better neighborhoods, and the girls were less likely to be single mothers. Some of these effects will therefore likely transmit to the next generation as well.66

  What explains why some neighborhoods are “better” than others for mobility? Researchers are far from settling this, but there are clearly features of the environment that seem to be correlated with higher mobility, including most importantly the quality of schools. The map of social mobility, it turns out, is closely related to the map of performance in standardized education tests.67

  Thanks to decades of research on education, we know a fair amount about what can be done to improve learning outcomes. In 2017, a study summarized 196 randomized studies conducted in developed economies on interventions (both in schools and with parents) to improve school achievement.68 While there was a wide variation in how effective these interventions were, a good preschool education and intensive tutoring in schools for disadvantaged children seemed to work best. Some children have a higher chance of falling behind grade level and then getting totally lost; preparing them ahead of time in preschool, and then being ready to identify and address any major gaps in their learning before they become too large, stops it from happening. This is entirely consistent with what we have found in our own work in developing countries.69

  There is also evidence that short-term gains in school outcomes translate into long-term differences in opportunities. For example, an RCT in Tennessee that cut class sizes from 20–25 to 12–17 led to an increase in test scores in the short run and a higher chance of going to college later on. Students assigned to small classes had better lives later on, as measured by home ownership, their savings, their marital status, and the neighborhood they lived in.70 High-dosage tutoring and small class sizes require staff, which would provide employment as well as helping kids throughout their school career.

  The constraint in the United States comes from the local funding of education. This has the consequence that the places in the most desperate need of good public education have the least money to pay for it. A substantial financial effort could make a big difference. More generally, one consequence of the low levels of government funding in the US is that pre-kindergarten education is not federally subsidized, and as a result only 28 percent of children attend some sort of subsidized pre-K program in the United States,71 in contrast with France, say, where pre-K is subsidized, attendance has been near universal for years,72 and it was recently made mandatory.

  The original evidence in support of pre-K programs came from some early randomized controlled trials that found large effects of high-quality preschool interventions both in the short and long term, leading Nobel Prize–winner James Heckman to advertise them as the best solution to reduce inequality.73 However, some of these experiments were tiny, making it possible to ensure the programs were run exactly as they should be.

  Two larger RCTs evaluating more realistic “at scale” pre-K programs (the national Head Start program and the Tennessee pre-K experiment) have been more disappointing; both of them found effects in the short run, but the effects on test scores faded or were even reversed after a few years.74 This has led many to conclude pre-K programs are overrated.

  But in fact a key finding of the national Head Start study is that the effectiveness of
Head Start seems to vary tremendously with the quality of the program. In particular, programs that run for the full day are more effective than half-day programs, and those that include home visits and other forms of engagement with parents are also more effective. There is also separate evidence from RCTs in both the United States and other countries of the effectiveness of home visits, during which preschool teachers or social workers work with parents to show them how to play with their children.75

  The general takeaway right now is that there needs to be more research so we know exactly what works in early childhood. But what we do know suggests resources matter; when Head Start was scaled up, many centers tried to reduce costs by cutting services, making them ineffective. Maintaining quality is crucial and has the added advantage of offering a massive expansion of what would surely be attractive jobs for many people, especially if they were adequately paid. These jobs would be both rewarding and impossible to robotize (one cannot really imagine a robot visiting parents at home).

  Equally importantly, it seems possible to train someone to be an effective pre-K teacher cheaply and fairly quickly, as long as there are the necessary materials to support them. In India, we worked with Elizabeth Spelke, a psychology professor at Harvard, to create a preschool math curriculum involving games that build on the intuitive knowledge of mathematics to prepare those who have not yet learned to read or write or even count for primary school. This was evaluated in an RCT in several hundred preschools in the slums of Delhi.76 Liz was initially horrified by the conditions in Delhi—the tiny porches overcrowded with students of various ages and the teachers’ low level of training, many of whom had barely completed high school. It was a far cry from the conditions in her lab at Harvard. But it turns out those teachers, with one week’s worth of training and good materials, were able to sustain the slum children’s attention, who played math games for several weeks, progressing through the games fast and with gusto, learning a good deal of math in the process.