David Card: ‘They thought that we were being traitors to the cause of economics as a whole.’
‘Economics as a whole is really a combination of two kinds of people: those who are very practically oriented and those who are more like mathematical philosophers,’ says David Card, professor of economics at the University of California, Berkeley. As a labour economist, he has always thought of himself as the first kind – the kind that mathematical philosophers are never happy with. ‘They come up with a broad general theory, and we tell them it doesn’t fit the evidence.’
David Card was the 2021 Nobel Prize winner in economics, jointly with Joshua Angrist and Guido Imbens. Card landed himself one-half of the prize, with Angrist and Imbens each receiving a quarter, for empirical research of the labour market. Card’s commitment to the empirical approach was highlighted by, among others, Harvard economist Richard Freeman: ‘If one unifying principle runs through David Card’s work, it is a belief in the power of empirical economic science – in the ability to use statistics creatively to make inferences about how the economy operates.’
‘I grew up on a farm’
Card was born in 1956 in Canada to a family of animal farmers. ‘I grew up on a farm near a university town,’ Card recalled later, ‘and felt that the university had a strong impact on the overall intellectual environment in that small town’. The memory later motivated Card to study how the proximity of a college drives children’s educational success, especially if their parents are not well educated.
Card completed a bachelor’s degree in physics – and it was by chance that he ventured into economics: ‘But my girlfriend through college was taking an economics class, and she was having some trouble with the textbook’s chapter on elasticities of demand. I started reading it and thought it was quite informative. I grew up on a farm and there’s a puzzle in the agricultural businesses: Why is a good year for farmers really a bad year? [farmers’ revenues are higher in a year of crop failure]? It’s basically because the elasticity of demand is less than one. And so reading about that was quite enlightening and I went through half the textbook over the next few days. Given that I was also probably not going to be the greatest physicist of all time, I decided to switch to economics’. As a man who grew up on a farm, Card still takes an interest in applied rather than purely theoretical studies.
On graduating from Queen’s University (Kingston, Ontario), Card defended his PhD thesis at Princeton. His supervisor was Orley Ashenfelter, a famous economist in the field of labour economics and econometrics, who also worked as a ‘practitioner’ at the US Department of Labor. Card considers Ashenfelter his main teacher.
Before the Nobel Prize
Card was awarded several major prizes in economics, including the John Bates Clark Medal and the IZA Prize. A 2011 survey of ( .pdf) 299 US university economics professors ranked Card fifth most popular economist under age 60 (at the time), after Paul Krugman, Greg Mankiw, Daron Acemoglu and Steven Levitt. Card holds the 12th position in the RePEc ranking of highly-cited researchers, as of January 2022. Card was a faculty member at Chicago University, Columbia University, and Princeton University, and has been at Berkeley since the late 1990s. He was editor of the Journal of Labour Economics, Econometrica and The American Economic Review, among others.
Famous and controversial
Card first came to prominence in the early 1990s when he found, jointly with the Princeton economist Alan Krueger, that a rise in minimum wages results in a rise, rather than a drop, in employment. This is the finding that Card and Krueger came up with in a case study of fast-food jobs in New Jersey and eastern Pennsylvania when in 1992 New Jersey raised its minimum wage, but Pennsylvania did not. As a result, the number of fast-food jobs in New Jersey went up relative to Pennsylvania. It was a hugely influential finding in contradiction with commonly accepted demand and supply models. A rise in minimum wages had been believed to cause an increase in unemployment since employers would not be willing to pay wages at higher rates to the same number of employees.
Following a rise in 1996 of the federal minimum wage, minimum wages went up in Pennsylvania but did not rise in New Jersey (its minimum wage was higher than the new federal level). No drop in employment was recorded in Pennsylvania. The rise in the federal minimum wage had no adverse employment impact in other states. In California, an increase in 1988 of the minimum wage from $3.35 to $4.25 (the wages of 11% of the state’s labour force were lower than the minimum) not only brought about a rise in the wages of teenagers and those employed in low-wage industries, but also pushed up employment. In the mid-1990s, Card looked into a non-standard situation in the labour market during the 1980s. While wages dropped and employment went up in the US, both wages and employment in Europe remained at the same level. That largely came as a result of labour market regulation, showed Card.
His research came in for heavy criticism. The data in the New Jersey case were not that good, having been collected by telephone, notes David Henderson, a research fellow at Hoover Institution and editor of the Concise Encyclopaedia of Economics, in a WSJ article. Following in the footsteps of Card but analysing administrative data, David Neumark of the University of California, Irvine, and William Wascher of the Federal Reserve Card and Krueger looked into the case of New Jersey restaurants, and discovered that the rise in the minimum wage resulted in what most economists thought it would: a drop in employment. In 2000, Card and Krueger conducted a follow-up study in response, which relied on Bureau of Labour Statistics data among other evidence, and adjusted their key finding that still ran counter to conventional wisdom: a higher minimum wage in New Jersey had no tangible employment impact.
Card and Krueger’s research sparked a rift among economists. In 1992, a survey of the American Economic Association’s members found that 79 per cent agreed that a minimum wage law increased unemployment among younger and lower-skilled workers. Those views were largely based on traditional economic views. By 2000, however, that view had been shared by just 46 per cent of the association’s members. Card and Krueger’s research fundamentally altered economists’ view of employment effects of the minimum wage.
Card argued that the demand and supply model that describes how employers operate in the labour market is untenable since it fails to account for market constraints and the fact that market players have imperfect information: individuals have to spend time looking for job opportunities and employers have to spend time finding employees.
According to Card, if a firm can readily go to the market and buy a worker, there is no such thing as a vacancy, or at least not a continuous vacancy. In the early 1990s, when Card and Krueger were studying minimum wages, many employers had had vacancies for months on end, and many fast-food restaurants had policies that said ‘Bring in a friend, get him to work for us for a week or two and we’ll pay you a $100 bonus.’ ‘These policies raised the question to us: why not just increase the wage?’ Card says. The simple model failed to explain this. In an alternative paradigm, a range of wage offers co-exist in the market at any one time, similar to product markets where two firms that sell very similar products may not charge exactly the same price.
Card and Krueger’s research also came in for criticism from Joshua Angrist, who shared the Nobel Prize with Card. Card admitted that his work cost him a lot of friends. ‘People that I had known for many years, for instance, some of the ones I met at my first job at the University of Chicago, became very angry or disappointed. They thought that in publishing our work we were being traitors to the cause of economics as a whole.’ There have been ongoing debates about this research. And yet, each new work in the field relies in some way on Card’s methodology.
I think my research is mischaracterised both by people who propose raising the minimum wage and by people who are opposed to it.
David Card in a 2006 interview to The Region of the Federal Reserve Bank of Minneapolis
Many economists who challenged Card and Krueger’s views were concerned that such works give voice to populist politicians. The authors themselves were disappointed over the politicisation of their research. In his column for The New York Times in 2015, Krueger lamented that the findings in the New Jersey study he presented jointly with Card were interpreted as the case for a wage rise, although they never made that argument; moreover, there is a difference between the increase in the minimum wage of a little less than 20%, as was in the New Jersey case, and the twofold increase under discussion, from $7.25 to $15, said Krueger. ‘A $15-an-hour national minimum wage would put us in uncharted waters, and risk undesirable and unintended consequences’, he warned. This is what Card himself had explained ten years before: ‘There are frictions in the market and some imperfect information. It doesn’t mean that if we raised the minimum wage to $20 an hour we wouldn’t have massive problems.’
Card admitted that he generally avoided speaking out on political issues, since active participation in public discussions could question the rigour and impartiality of research findings. However, this does not prevent him picking topics of high public importance for his studies.
‘In my opinion, no-one’s view in economics is overturned by any single paper, or even any collection of papers. And so the best you can do is what I would call ‘professional work’. <…> I don’t want to publish a controversial result and have it possibly the case that somebody could spend less than, say, three years on the project and find an error,’ said Card in answer to the question about how controversial results affect how he writes the paper. You must realise that a bunch of very smart people, who have very strong incentives are going to pour over your results and try and figure out what you did wrong – this is how he explained the need to make sure there are no errors. His point is, the researcher should be as specific as possible: ‘All you can report is, “in this particular circumstance, what we found is...” Don’t say that your results are anything other than what they are.’
You must realise that a bunch of very smart people, who have very strong incentives are going to pour over your results and try and figure out what you did wrong.
David Card, 2010 interview to Simon Bowmaker of New York University
Card recalled that top academic journals would reject most of his papers at least one or two times. ‘Usually, it’s very annoying. Younger people tend to look at the referee report and ... think to themselves that there’s something in there that’s saying why their paper didn’t make it. But oftentimes that’s not really why. The reason was because it’s not quite interesting enough, or not quite decisive enough, or not quite fitting with the scheme of how things are going in the field, or it doesn’t seem plausible.’ The ‘revise and resubmit’ is the most important stage of the project, he noted: ‘It’s your job to address every single thing that’s raised by the referee and nail it. … Somebody takes the comments seriously, follows the advice, tries to nail down the loose ends and even fesses up a problem by saying: ‘There is one thing we can’t figure out. But we’re going to rewrite the paper to acknowledge it.’ Those who were less well-trained would send back only a few pages of comments, and the referee would then be alienated, which means the end of the paper. I learnt a lesson.’
Faces of inequality
One hundred years ago, the US was on track for universal secondary schooling. But even at the time, mobility rates for black families were substantially lower than for whites, Card shows. Black students were disadvantaged by the low quality of segregated schools (the difference in teachers’ salary went up after many southern states had set minimum wages). Graduates from low-quality schools could not improve social mobility and in the future find higher-skilled and better-paid jobs than their parents.
Segregation has a strong effect on inequality, Card shows in another article. The higher ethnic segregation in a particular American city (which shows in inter-city distributions and the composition of schools), the greater the difference between educational results for white and black students. A move from a highly segregated to a more integrated city – where whites and blacks at least to a certain extent live and study together – leads to a one-fourth drop in the difference of educational outcomes.
In the US, the inequality of earnings has been on the rise and among men and, conversely, has declined among women over the last fifty years, writes Card in his recent article. Comparisons are made between inequality levels, not earnings of men and women. The trend has essentially emerged as more women have chosen to work. 80% of the decline in female earnings inequality is explained by increasingly higher labour force participation of women, regardless of the number of young children and spousal earnings.
Immigration accounts for a small share of 5 per cent in the increase in US wage inequality, Card calculates; immigration makes no effect on the local wage structure, and does not cause native outflows, with cities attracting migrants recording a strong rise in the supply of low-education labour. Card’s most cited paper was dedicated to the immigration of Cubans to Florida, where some 125,000 Cuban refugees arrived in the first six months of 1980. The population of Florida grew 7%, with huge inflows of unskilled labour, yet unskilled local workers’ wages between 1979 and 1985 were virtually unchanged. <
And so the best you can do is what I would call ’professional work’. This means no errors and no obvious omissions.
David Card, 2010 interview to Simon Bowmaker of New York University
As in the case of the minimum wage, Card’s interest in migration was not driven by the question of the benefits or harm it brings, but by the question of how labour markets adjust to migration. For instance, immigrants account for almost half of all Los Angeles inhabitants, of which the proportion of low-educated people is high. Meanwhile, with few migrants in Pittsburgh and Cleveland, low-skilled workers are in short supply there. Card demonstrates that the US labour market is very adaptive and quickly ‘absorbs’ migrants. Therefore, the inflow of low-educated migrants leads to a very small drop in the wages of local workers with similar qualifications.
Unlike wealth inequality, inequalities in education could be beneficial for students. Card found that participation in a ‘gifted/high achiever’ class led to significant achievement gains for non-gifted participants, in addition to high-ability students. This effect was sustained and long-term.
In education studies, Card’s key contribution was analysis of how various models of school education influence the education premium. Card proved that gifted and talented programmes were best allocated not on the basis of cognitive abilities of the sudents but on the basis of their past achievements. This can boost the performance of students, especially for black and Hispanic students from low-income families. Jointly with Krueger, Card wrote that the education premium (the difference in earnings between the more and less highly educated) depended on, beyond the years of education and stages, school quality (for example, the pupil-teacher ratio). The schooling choices of children from disadvantaged family backgrounds are of particular importance.
Card’s concern is that parents’ education makes a strong impact on their children’s education. ‘One of the interesting things that I’ve witnessed in my 25 years as a professional economist is the changing character of the students that I see in graduate school. Every year I do a poll of students and ask how many of their parents have PhDs. That ratio is now close to 60%, even at Berkeley. We are getting a very select group of people going into PhD programmes who have very different backgrounds from the average person in the country. I think that is a little troubling. It means our students don’t really understand what a big leap it is for some families to send a child to college. They don’t think about college as a difficult choice, because everybody in their family has high education, and all their friends went to college and will probably go further. We live in a stratified society.’
Social policy
Card wrote many papers with analysis of various social policy aspects that reduce inequality. Central to this are programmes such as Medicare (health insurance independent of income). In 2007, before Barack Obama decided to expand the programme, this type of insurance had covered all Americans above 65 years of age and led to, Card calculated, a 20% reduction in the mortality rate of such patients admitted to hospital who needed urgent medical intervention. The reason is higher insurance rates for medical care, making hospitals more willing to perform all the necessary procedures for insured patients and to transfer patients to other departments for further treatment. Insurance of the over 65s had similar positive effects in other countries, Card found.
Jointly with Raj Chetty of Harvard, Card looked into job search behaviour in Austria in the context of generous unemployment insurance (€2,300). Entitlement to benefits for 2 months reduces the job-finding rate by 8–12%; entitlement to benefit for 5–7 months, by 5–9%. These payments did not have any effect on the quality of subsequent jobs.
Emotions, neighbourhood effects and urban policy
Role models are critical during adolescence, when teenagers tend to emulate the behaviour of their friends. If their friends tend to engage in a risky activity (sex, smoking, marijuana use, truancy), adolescents are on average 30% more likely to imitate their behaviour, Card calculated jointly with Laura Giuliano, his regular co-author. For adults, access to information on peers’ wages has serious implications for individual pay satisfaction, and in an asymmetric way. Employees report dissatisfaction and job search intentions if they find out that their wages are lower than those of their colleagues. However, those earning above the median report no higher satisfaction: they take it for granted.
In their highly acclaimed paper, Card and Gordon Dahl of the University of California, San Diego, showed how losses by the home football team during the National Football League season led to an 8 per cent increase in at-home intimate partner violence. It was further found that unexpected losses in highly salient or frustrating games rates of family violence were much higher and increased by12-16%. The work was another contribution to behavioural economics, providing evidence that family violence is better characterised as a breakdown of control rather than as rationally instrumental violence.
Card collaborated with Alexandre Mas (Princeton) and Jesse Rothstein (Berkeley) to investigate urban policy. They made the case that neighbourhoods with mixed racial composition can be as stable as all-white neighbourhoods, provided that their minority share remains within a certain range (5%–20%). If a minority share is above that level, the neighbourhood loses its appeal to the racial/ethnic majority. The entailing conflict is likely to result either in white flight, with such neighbourhoods rapidly transitioning to all-minority composition, or minority flight, making the neighbourhoods all-white. If a minority share is within a tipping point, the neighbourhood remains equally attractive to the majority and minority.
Card wrote that settlement patterns are also affected by labour market forces. For instance, poor labour market conditions expanded the share of youth living with their parents in Canada in the 1970s–1980s relative to the US, where labour market conditions also declined. This ‘move back home’ saved Canadian youth extra spending, to the effect that their relative position in family income distribution did not deteriorate as fast as in the US.
Good and bad selection
Card takes a keen interest in various types of inequality – gender, racial or regional. Racial inequality is created not only at the market level, but also by some firms, Card argues. Firms’ wage-setting policies may be characterised by racial pay differences. Moreover, higher-paying firms tend to have a preference for white workers. To investigate racial inequality factors, Card took the case of Brazil where the levels of this inequality are comparable to the US (27–33%). Non-whites are less likely than whites to work at higher-wage establishments, even in the absence of discriminatory employment practices, and are likely to find themselves in lower-pay positions. Overall differences in wages paid by a given employer for similar jobs (‘firm wage effects’) contribute on the order of 20% of labour market inequality.
Card detected similar components of inequality in Germany: the increasing variability of West German wages resulted from rising heterogeneity in skills and in the wage premiums at different establishments. Higher-skilled workers seek to find jobs at higher-paying establishments.
The combination of sorting and bargaining effects explain the racial and gender wage gaps: women are less likely to work at high-paying firms and are offered 10% lower wage bargains with their employers than men.
Adding to inequality, workforce selection improves firm-level productivity, Card concludes in a paper written jointly with Nicholas Bloom (Stanford), John Van Reenen (LSE) and others. According to Card’s co-authors’ prior studies, much of the cross-firm variation in measured productivity is due to differences in use of management practices. These practices – including monitoring, goal setting and the use of incentives – involve the talent of top managers. Therefore, more productive firms recruit top managers with higher management scores. It is this selection of managers and their pay premiums, not the difference in average wages, that explains 30% of the measured impact of management practices on productivity.
Labour unions matter
In a recently published paper, Card investigated the impact of sectoral bargaining agreements between labour unions and employers on real wages in a specific sector. The influence of labour unions on wages is a priority research area for Card, who published many papers on the subject in the 1980s and early 1990s. The US and European labour union systems are different. In the USA, labour unions make arrangements with employers on wages for certain jobs. An increase in wage rates, if labour unions manage to secure it, is converted into increased wages of workers, unless employers are forced to cut jobs because of the wage rise. In Spain, Italy, France, Belgium, the Netherlands, and Portugal, sectoral contracts set wage floors for some occupation groups rather than a wage rate, with employers paying wages above the floor. The European system is more flexible in terms of compensation, while the US system affords more opportunities to dismiss workers.
In a conventional view, such agreements are bad in undermining employer opportunities to adjust to downturns. If a national financial crisis breaks out, as was the case with Portugal in 2011, creditors represented by the IMF even recommend that the country go back on sectoral agreements. Card explains why in the case of Portugal – where about 90% of jobs are covered by sectoral agreements – the IMF is wrong. He looked into the ratio Portuguese workers’ wages to sectoral wage floors for 2008–2016 and showed that sectoral agreements do not prevent a strong variation of wages both within and between occupation groups. A typical worker receives a 20% premium over the floor, and this wage cushion increases with the company’s productivity. As floors rise, cushions are compressed. In times of crisis, both wage floors and real cushions went down. This suggests that wage floors in Portugal do not erode wage flexibility during economic upturns or downturns.
In the US, Canada and the UK, ‘unionisation’ works to reduce wage inequality among men but has little impact on wage inequality for women since many women are employed in education and other sectors where wages are regulated without involving labour unions. Recent decades saw ‘de-unionisation’of the US labour market, which contributed to growing inequality. Card estimates the share of this driver at 10–20% in the 1970s–1990s. Meanwhile, the contribution of still active unions to reducing wage inequality in the US is no more than 10%. Importantly, unions in the public sector (education, healthcare and social services) are much more influential than in business.
Similar firm-level agreements – the subject of Card’s previous research – account for approximately a 5–10% wage premium, with larger premiums for more highly paid workers.
Card counters the common view that inequality has increased in recent decades on the back of a widening gap between high and low-skilled workers. This hypothesis was not very helpful in understanding the gender gap, the racial wage gap and their changes, or between-sector wage gaps in the end of the 20th century.
