A Central Bank Digital Currency?

Most money is already digital: for example, my paycheck flows automatically into my bank account, my mortgage payments automatically flow out, and I authorize my bank to pay credit card charges online. So what would be different about a central bank digital currency?

Just to be clear, the idea here is not that the Federal Reserve, the European Central Bank, the Bank of Japan, and others would start up competitors for the cryptocurrencies like Bitcoin, Ethereum, or Dogecoin. Instead, a central bank digital currency would be a form of the nation’s existing currency. Like currency in the form of coins or paper money, you would be able to use the digital current as a medium of exchange when buying and selling. However, unlike the examples of digital money I mentioned a moment ago, a central bank digital currency would not need to move through your bank.

Thinking about a central bank digital currency in this way raises obvious practical questions. What form would it take? How would it be accessed when buying or selling? What security would protect it? Would digital currency be anonymous, like physical currency, or could it be tracked? It also raises a bigger conceptual question: What benefits might result from central bank digital currency?

The subject is hot right now. For example, the Bank of International Settlements, in its most recent annual report, devotes a chapter to “CBDCs: an opportunity for the monetary system.” Randal K. Quarles of the Fed Board of Governors just gave a talk on the subject called “Parachute Pants and Central Bank Money” (June 28, 2021). A subcommittee of the Senate Committee on Banking, Housing, and Urban Affairs held hearings on “Building A Stronger Financial System: Opportunities of a Central Bank Digital Currency” earlier this month (June 9, 2021), including testimony from  Neha Narula, Director of the Digital Currency Initiative at MIT, Lev Menand of Columbia Law School, and Darrell Duffie of the Stanford Graduate School of Business.

The main potential benefit claimed for a central bank digital currency is an improvement in “payment systems,” which is the jargon-heavy term for the idea that many of the current ways of making digital payments through banking involve fees of some kind. Current digical payments are mostly “bank-railed,” to use a term from Darrell Duffie–that is, they flow through connections set by banks. Sometimes the fees are explicit, from banks or credit card companies or payment services. Other times the fees are implicit, like when a financial firm takes a day or two or three before a payment clears–time when someone is potentially earning interest on that money. In many places, these fees for making payments can be 1% of GDP or more.

Here’s a figure from the BIS report on the size of payment-related fees around the world, where EMEA is an abbreviation referring to a selection of countries in Europe, the Middle East, and Africa.

Here’s a similar figure from Duffie on costs of payment systems. Duffie writes: “It takes too long for U.S. merchants to receive their payments, often more than a day. Based on McKinsey data, moreover, Americans pay about 2.3% of GDP for payment services, far more than Europeans, particularly because of extremely high fees for credit cards, as illustrated in Figure 1. This is not because Americans are getting better quality service. Further, the primary payment instrument of Americans, their bank deposits, is compensated with extremely low interest rates.”

There are various concerns about these fees. If relatively few companies are providing payment services, the fees may be higher than they need to be because of lack of competition. In addition, a substantial number of people are “unbanked,” and without a bank account, the payment fees charged by nonbank payment service companies (say, those sending funds across borders) can be quite high. There are also questions about whether the current digital payment systems do a good job of protecting people’s personal information and privacy.

How would a central bank digital currency address the problem. For example, would people and businesses have individual accounts at the Federal Reserve? This would avoid banks, but it would also be a major change in the function of the Federal Reserve, which serves the function of coordinating payments across banks, but not being a bank itself. The BIS describes what a retail-level central bank digital currency might look like this way:

Retail CBDCs come in two variants. One option makes for a cash-like design, allowing for so-called token-based access and anonymity in payments. This option would give individual users access to the CBDC based on a password-like digital signature using private-public key cryptography, without requiring personal identification. The other approach is built on verifying users’ identity (“account-based access”) and would be rooted in a digital identity scheme. This second approach is more compatible with the monitoring of illicit activity in a payment system, and would not rule out preserving privacy: personal transaction data could be shielded from commercial parties and even from public authorities by appropriately designing the payment authentication process. …

From the public interest perspective, the crucial issue for the payment system is how the introduction of retail CBDCs will affect data governance, the competitive landscape of the PSPs [payment services providers] and the industrial organisation of the broader payments industry.

At least to me, other advantages sometimes cited for a central bank digital currency often miss the point. For example, one will sometimes hear claims that the Fed needs a digital currency to compete with the cryptocurrencies like Bitcoin and Ethereum. But it’s not at all clear to me that these cryptocurrencies are anywhere near unseating the US dollar as a mechanism for payments, and it’s quite clear to me that competing with Bitcoin is not the Fed’s job. Or one will hear that because other central banks are trying digital currencies, the Fed needs to do so, also. My own sense is that it’s great for some other central banks to try it out, and for the Fed to wait and see what happens. There is a hope that zero-cost bank accounts at the Federal Reserve might help the unbanked to get bank accounts, but it’s not clear that this is an effective way to reach the unbanked (who are often disconnected from the financial sector and even the formal economy in many ways), and there are a number of policy tools to encourage banks to offer cheap or even zero-cost no-frills bank accounts that don’t involve creating a central bank digital currency.

Quarles at the Fed sums up the current case against a CBDC in this way (footnotes omitted):

In brief, the potential benefits of a Federal Reserve CBDC are unclear. Conversely, a Federal Reserve CBDC could pose significant and concrete risks. First, a Federal Reserve CBDC could create considerable challenges for the structure of our banking system, which currently relies on deposits to support the credit needs of households and businesses. An arrangement where the Federal Reserve replaces commercial banks as the dominant provider of money to the general public could constrict the availability of credit, fundamentally alter the economy, and expose the public to a host of unanticipated, and undesirable, consequences. Among other potential problems, a dominant CBDC could undermine the consumer and other economic benefits that accrue when commercial banks compete to attract customers.

A Federal Reserve CBDC could also present an appealing target for cyberattacks and other security threats. Bad actors might try to steal CBDC, compromise the CBDC network, or target non-public information about holders of CBDC. The architecture of a Federal Reserve CBDC would need to be extremely resistant to such threats—and would need to remain resistant as bad actors employ ever-more sophisticated methods and tactics. Designing appropriate defenses for CBDC could be particularly difficult because, compared to the Federal Reserve’s existing payment systems, there could be far more entry points to a CBDC network—depending on design choices, anyone in the world could potentially access the network. Critically, we also would need to ensure that a CBDC does not facilitate illicit activity. … [I]t may be challenging to design a CBDC that respects individuals’ privacy while appropriately minimizing the risk of money laundering. 

The Federal Reserve is preparing what it calls a “comprehensive discussion paper” on central bank digital currency, so the subject is sure to remain a live one. My own sense is that, at present, the proposal can be seen as a shot across the bow from US policymakers of US banks and the US financial system: basically, “help us figure out a way to make the payments system cheaper and more accessible to all, or we might do something drastic.” Duffie argues: “U.S. banks are capable of providing an effective low-cost payment system but have not done so. Regulations, network effects that limit entry, and profit incentives have not promoted an open, innovative, and competitive market.”

Interview with Lant Pritchett: Topics in Development

The Centre for Development and Enterprise has published “Lant Pritchett in Conversation with Ann Bernstein” (June 2021), a transcript of a one-hour conversation. Here are a few of Pritchett’s comments that caught my eye, but see the interview for thought-provoking comments on education, corruption, and other topics.

On prioritizing national development over poverty reduction:

Development is a process that happens at the level of countries. The four transformations a country should make are: (1) to a productive economy, (2) to a capable state, so that it is able to do what it sets out to do, (3) to a government responsive to the needs and wishes of citizens, and (4) to a society where equal treatment of all before the law and of each other is a bedrock principle. I think those four characterise the transformation that takes a country from chaos and poverty to the levels of prosperity and well-being that we see in developed countries.

I strongly favour a focus on prosperity over a focus on poverty. You often hear the phrase ‘this or that isn’t a panacea’. My argument is: national development is a panacea. If your country manages to undergo the four transformations of national development, then all problems get solved because that is a machinery for nominating and solving problems.

Yet the current focus in development is on what I call ‘kinky development’, which involves tinkering on the margins to help the poorest of the poor. That is the wrong focus. If you achieve national development, you will solve poverty and provide prosperity for the general population, whereas focusing on poverty alone often is at odds with getting you to desirable levels of prosperity. We should ask ourselves with everything we do: “Is this contributing to one of the four transformations we need to do, and if so, how?” …

No country has high levels of human wellbeing without having achieved national development; and every country that has high national development achieves very high levels of human well-being. So, the only path to high human well-being is through national development.

Why following “best practice” is a poor and even counterproductive development strategy:

The AK47 is the world’s most popular weapon. The M16, which is the standard weapon in the US army, is far and away a more accurate weapon than the AK47, which beyond a few hundred yards, cannot hit a thing. The AK47 emerged from the Soviet Union, where they designed their weapons for the soldiers they had, low capability with little training. They also designed the AK47 to be unbelievably robust; no matter what you do to it, when you pull the trigger, it fires. You can basically hand anybody an AK47 and it will be a reasonably effective weapon. The United States took the opposite approach of designing the best possible weapon and training soldiers to match the weapon. It is an excellent weapon, but if you do not keep it clean and in good functioning order, it will misfire.

The problem is when you give the M16 with its perfect design to a poor soldier it won’t work. This mirrors a lot of what has happened in development – the desire to adopt best practice, leads to a gap between practice design and the capability for implementation. Rather than organically building designs that work in a low-implementation environment, policymakers have tried to borrow designs and fit them into countries, and it just does not work. When well-designed programmes are poorly implemented the reason is obvious, but the problem repeats itself, because no one ever admits that what they need is an AK47. You need to design the programme for the soldiers you have.

The Shrinking US Role in World Car Markets

Within the US, I often find a widespread sense that what happens in US (and perhaps also European) car markets will be the key factor in shaping world car markets. But that’s highly unlikely to be true. The US accounted for 62% of all global car registrations in 1960. Even by 2000, the US included 22% of all global car registrations. But by 2018, the US share of global car registrations was 11% and still falling. The US and European car markets are essentially stagnant in terms of total quantity of cars, but rapid growth in cars is happening in the rest of the world, including in China.

The estimates of the US share of global car registration is from the latest version of the Transportation Energy Data Book, put together by Stacy C. Davis and Robert G. Boundy for Oak Ridge National Laboratory (vol. 39, updated April 2021). Here’s some other data from that volume.

This table shows car production by country for 2000 and 2018. The US now runs a distant fourth, behind China, Japan, and and Germany, just ahead of India, Brazil, and Spain.

This figure shows total ownership of cars, comparing 1960, 2000, and 2018. Here, the US still runs second, behind China, but of course this is in part because of the much larger US population compared to most of the other countries listed here. It’s also true that US cars are lasting much longer, so the gap between annual production and total ownership has risen. For example, in the US back in 1970, half the cars on the road were four years old or younger, and only 2.9% of cars on the road were 15 years old or more. In 2018, only about one-quarter of the cars on the road were four years old or less, and 19% of the cars on the road were 15 years old or more.

Of course, the fundamental driver of the rising share of global car production and usage outside the US is a result of faster-growing economies in those places. How much farther do other countries have to go before they begin to approach US levels of car ownership?

This figure helps put US car ownership in perspective with other countries. The dark line in the figure shows US vehicles per 1000 people over time from 1900 to the present. In the upper right, you can see how this measure flattens out for the US in recent decades. Then the car ownership of other nations per 1000 people is plotted on this same line. Thus, you can see that Canada in 2018, or western Europe in 2018, had about the same number of vehicles per 1000 population as the US had in the early 1970s. Further down the line, you can see that Mexico in 2018 had about the same number of vehicles per 1000 population as the US had in 1950.

What about China and India and others? To see where they are on this line, the second figure is a blow-up of the bottom left corner of the above figure. Here, the rise in US car consumption goes only from 1900 to 1930. You can see that Brazil’s vehicle ownership in the 2018 was at about the same level (per 1000 population) as the United States in the late 1920s, while China’s vehicle ownership in 2018 per 1,000 people was at about the US level of the early 1920s.

In short, the future of the global car industry in terms of sales and technology and how automotive technology affects the world’s environment is going to be written largely outside US borders. A just-released study of demand for cars in China found an income elasticity of demand for cars in China of 2.5: that is, every 10% rise in incomes in China (roughly what has been happening every year or so) has been leading to a 25% rise in quantity of cars demanded.

It’s a Zoom World After All

The pandemic has pushed all of us into Zoom World: that is, a world where many more of our interpersonal connections, both work and personal, happen online. This de-emphasis on direct interpersonal connections and greater emphasis on long-distance online connections has some inevitable tradeoffs.

For example, there is a tradeoff that geographically close relationships have been to some extent devalued relative to long-distance ones. There is a tradeoff that the more formal and pre-planned interactions have increased in importance, while informal and unplanned interactions and conversations have been reduced. There is a tradeoff that those with a strong network of pre-existing long-distance relationships find it easier to continue in those relationships, while those without such a pre-existing network will find it harder to break into such relationships. There is “superstar” effect that many of us now have greater access to high-profile people than we did before, but the tradeoff is that this access is via an online channel that often lacks interactivity, and in the meantime we may be neglecting the mid-profile and lower-profile people around us with whom we could have a higher level of interactivity.

Here, I’ll point out some thoughts on how these tradeoffs and Zoom World in general are playing out in several contexts: academic seminars, family balance, and urban workplaces. The comments are all drawn from a recent e-book book by Luis Garicano: Capitalism After COVID: Conversations with 21 Economists (June 2021, CEPR Press) However, I should emphasize that Garicano’s book is not focused on this narrow topic, and indeed is very broad in scope. I’ll provide a full list of interviewees at the bottom of this post.

Here’s Jesús Fernández-Villaverde commenting in his interview with Garicano on the tradeoffs of on-line academic seminars, in an interview titled ” Economists and the pandemic:”

You and I waste an enormous amount of time in airports, and now we’re doing everything by Zoom. It’s very hard to figure out how much is being lost in terms of output. That’s true. Let me give you the positive and the negative. In my view, the positive is that this is opening up, for instance, seminars to people that are ‘out of the circuit’. The top people in the profession are hard to attract to your seminar series if you are a relatively small university – often because they don’t have the money. Over the last few weeks, however, I have given my paper on Covid in a lot of South American universities. The average South American university may not have the budget to bring someone from the US and put them in a hotel. And I will not have the time. So suddenly it’s easier for them to do a conference. On the other hand, Zoom is not a great substitute for person-to-person. So, if the host is the London School of Economics, the department of economics there is getting less now from its seminars than it did on a personal basis. There are winners and there are losers.

And the second thing that is important is the difference between maintaining existing relations and creating new relations. We can spend a year on Zoom because we already know each other. We have some dynamics, we can interact. But what is going to happen with the new assistant professors and the new graduate students trying to get into the circuit through zoom? It is going to be much harder. Zoom gives a premium to eloquence that maybe in a person-to-person relationship is not so strong.

You worked at Chicago with Sherwin Rosen. Rosen wrote the famous paper on the economics of superstars (Rosen, 1981) and his example was, of course, music. Why are you going to go and listen to a good opera singer when, thanks to a DVD, you can listen to an amazingly extraordinary opera singer? This may happen with economics. Why are you going to go to a seminar by a good professor of economics when you can listen to the seminar by Ivan Werning?

Here’s Fernández-Villaverde on the tradeoffs of Zoom World from in the balance between family and work life:

One thing that I find disappointing for societies is that either you work eight hours a day or you work zero. Some societies, especially in the north of Europe, have made progress in terms of solutions, but it’s something we should push very hard. For instance, let’s think about fertility. One of the reasons why fertility is so low these days is because having a kid and trying to keep a full-time job is very difficult. That usually has a gender component because women usually assume most of the cost of it, and that generates many tensions and unhappiness in society that I fully understand. So, imagine we are in a society where we have more flexible forms of work. Thanks to Zoom, I can go from being expected to work eight hours a day, to being expected to work six hours and half hours, to say a random number. Then, it’s much easier to reconcile work with family. In the old world where you had a factory, that’s difficult to accomplish.

Governments here should have a leading role. A lot of what we do is about coordination. I want to be in the office at 9:00 because someone else is going to be at there at that time. You just need to have a focal point, and governments can help to coordinate us in good focal points. I can imagine many people who are aged 65 or 66 and are reluctant to work eight hours a day, but at the same time they are not very happy working zero hours a day. If we could get to a society where you can flexibly work four or five hours a day, maybe we could extend the working life of many people, contributing to GDP and helping us a lot to transition. It will also have positive health benefits. We have actually quite a bit of evidence that when people suddenly retire, they start an exponential decay, and especially if they are men they tend to drink and gain weight. So, I think that these solutions will help society a lot. Thanks to telecommuting and Zoom, we may be able to do that much better than in the past.

Finally, here are some comments from Esteban Rossi-Hansberg about the question of how Zoom world and online interactions will ultimately affect the density of urban areas.

This is highly speculative, of course, but I think the evidence from urban economics and from all the studies of cities is that systems of cities tend to be very resilient to shocks. Cities tend to come back after shocks, and the distribution of cities doesn’t tend to change that much. For example, if you look at the evolution over time of the size distribution of cities in the United States, or in many countries in Europe (certainly in France), it looks very similar today to what it did 50 years ago. The overall size of the cities has changed, but the distribution itself – how much bigger the biggest city is than the second and so on – is very similar.

So cities have proven very resilient to shocks. The question is: is this time different? I do think there’s a potential for it to be different, in particular because of telecommuting technologies. These are not new, and we’ve seen them improve over time. Telecommuting was already growing before the pandemic, but it was at very low levels (around 5%). But it’s rising, as you would expect from the technology getting better over time, or a lot better over time. That’s the standard evolution and it is perhaps not going to change cities in a dramatic way, at least in the nearby future. With Covid of course, a lot of us were sent or chose to go home. In a world in which there’s multiple equilibria, can this change the equilibrium?

Let me be a little bit more explicit. How much we get out of going to the office depends on how many of our colleagues go to the office. If we all go to the office and we all interact in the office, we get more out of the office. We learn from our co-workers and there’s externalities, knowledge spillovers. That’s why there’s offices. Of course, these spillovers also operate across firms. And that’s why it costs so much money to rent an office in the middle of Palo Alto or in the middle of Manhattan. It pays to be there. This creates a coordination problem. Going to the office works if everyone else goes to the office too. But if no one goes to the office, it’s not so good because you have to pay all of the commuting costs and there will be no one there. Now, suppose the vaccine comes and the whole pandemic goes away. Are we going to go back to the office? Well, if we’re coordinating in the equilibrium of not going to the office, we’re going to stay there. Why? Because no one wants to be the first mover and go to the office.

Can we solve that coordination problem and make everyone go to the office? It’s not going to be that easy. There are individual advantages of staying at home. Most people are saving an hour on average a day not having to go to work and back, and that hour is valuable. Unless firms and governments are proactive in moving us back to an equilibrium where we all go to the office, we’re going to stay home. Now is this new equilibrium bad? Maybe it is a great advantage of the modern world. Maybe it is great to telecommute. But all the evidence that we have is that the interactions we would be missing are valuable. Hence, if the economy doesn’t get all those interactions and to the extent that telecommuting doesn’t exploit and encourage those interactions, the economy is going to suffer because productivity is going to suffer in the long run. Furthermore, cities are going de-agglomerate as a result of this. So, some of these externalities are external, not just to the worker, but even to the firms – externalities that firms aren’t capturing. The firm is saving itself some money by not paying the rent. But the city and the productivity of the whole economy could suffer. …

We were in a good equilibrium in which we were paying a lot of commuting cost. But at that cost we were getting a big benefit, which was all these productivity enhancements from all of us being there. It’s a quantitative question to what extent the benefits from not commuting and all the potentially large environmental benefits of having fewer people driving and so on are going to compensate for the productivity losses that we’re going to see from people not going to the office. Now, if we’re in the equilibrium where we don’t go to the office, that may also encourage biased technological change to improve those interactions online. It is a possibility. But it is something that we still don’t know. We don’t know how good these technologies can be.

Finally, here’s the full list of Garicano’s interviewees, and the broad subject areas of the interviews, taken from the Table of Contents of the book:

Debt sustainability
Markus Brunnermeier: Let’s compare the central bank to a race car 11
John Cochrane: Throwing money down ratholes 17
Jesús Fernández-Villaverde: Economists and the pandemic 23
Agnès Bénassy-Quéré: How to design a recovery plan 29

Tackling inequality
Oriana Bandiera: Overcoming poverty barriers 39
Stefanie Stantcheva: Taxes and social economics 45
Esteban Rossi-Hansberg: Will working from home kill cities? 53
Atif Mian: The savings glut of the rich 59

A more balanced globalisation
Dani Rodrik: Globalisation after the Washington Consensus 69
Pol Antràs: Is globalisation slowing down? 75
Michael Pettis: Trade wars are class wars 83

Containing the new leviathan
Daron Acemog˘lu: The Great Divergence 93
Wendy Carlin: The Third Pole 99
Lucrezia Reichlin: Democratising economic policy 107
Carol Propper: Targets and terror 111
Raffaella Sadun: Management for the recovery 117

Promoting innovation and curbing the power of digital giants
Philippe Aghion: Is ‘cutthroat’ capitalism more innovative? 127
John Van Reenen: The Lost Einsteins 133
Fiona Scott Morton: What should we do about big tech? 141

Combatting global warming
Nicholas Stern: Zero-emissions growth 149
Michael Greenstone: The real enemy here is carbon 157


The Reality of Attractiveness Bias

Being more attractive is associated with having a higher income. What one chooses to make of this fact can be a source of controversy. But the fact itself is well-established. Ellis P. Monk Jr., Michael H. Esposito, and Hedwig Lee discuss the topic in “Beholding Inequality: Race, Gender, and Returns to Physical Attractiveness in the United States” (American Journal of Sociology, July 2021). For example, they describe some of the earlier research in this way (citations omitted):

While a one-standard-deviation increase in ability is associated with 3%–6% higher wages, attractive or very attractive individuals earn 5%–10% more than average-looking individuals. Another study even finds that returns to perceived attractiveness unfold over the life course and are robust to a wide array of potentially relevant controls, such as educational attainment, parental background, personality traits, IQ, and so on.

Of course, some obvious questions in this line of research include who is deciding what is “attractive,” and whether “attractive” is primarily a marker for other categories like race/ethnicity or gender. For example, if one compares within the category of, say, black women, is there still a positive association between attractiveness and income? How is the benefit of attractiveness distributed across groups, and it is larger for some groups than others? The authors describe their results in this way (references to figures omitted):

[W]e must emphasize, however, that perceived physical attractiveness is a major factor of inequality and stratification regardless of one’s race or gender. In fact, our analyses suggest that the magnitude of the earnings gap among White men along the perceived attractiveness continuum rivals that of the canonical Black-White wage gap and the attractiveness earnings gap among White women actually exceeds, in real dollars, the Black-White wage gap.

This, however, is not at all to say that race and gender do not matter. Quite the contrary. We find that while the returns to perceived physical attractiveness are similar for most race-by-gender combinations, the slope of the returns to perceived physical attractiveness is steepest among Black women and Black men. The returns to attractiveness among Black women, for instance, are so immense, that the earnings of the most attractive Black women appear to converge or even overlap those of white women of similar levels of attractiveness. Notably, the returns to perceived physical attractiveness, in terms of real dollars, are similar between White males and Black females even though Black females make significantly less than White males on average. Similarly, the returns to attractiveness among Black men are quite substantial as well, though not enough to see a convergence or cross-over with white men. Among Black women and Black men, the wage penalties associated with perceived physical attractiveness are also so substantial that, taken together, the earnings disparity between the least and most physically attractive exceeds in magnitude both the Black-White wage gap and the gender gap.

Here’s a figure to illustrate some of the results. The horizontal axis is a measure of attractiveness. The vertical axis measures income. The lines on the graph are three shades of gray: the darkest line is the white population, the in-between gray is the black population, and the lightest shade of gray is the Hispanic population. For all groups the lines slope up: that is, those who are more attractive have higher income. But the slope of the “attractiveness” line is especially steep for black women and men.

An obvious question in this kind of research is what data was used, and how “attractiveness” was defined. The data is from the National Longitudinal Study of Adolescent to Adult Health, which is “an ongoing, nationally representative survey of a group of individuals who were in grades 7–12 in 1994.” The original sample included more than 20,000 individuals from 132 schools across the country, and the survey was an in-depth home interview. There were four waves of surveys, in 1994-95, 1996, 2001-2, and 2007-8. They focus on the 9300 people who were in all four waves.

The survey asked the interviewers to evaluate the physical attractiveness of the respondent, but did not give them any guidelines for doing so. However, one can compare attractiveness rankings given by male and female interviewers, and across black and white interviewers. The short answer to these questions is that while it is hard to define “attractive,” people tend to know it when they see it. The authors write:

Overall, then, there seems to be overwhelming consensus on ratings of attractiveness across interviewers regardless of race/ethnicity and gender. Furthermore, our analyses reveal virtually no evidence of same-race bias in their attractiveness ratings. Thus, taken together, even though these measures are subjective and there is some variation, interviewers exhibit consensus, which is in line with the findings of research on judgments of attractiveness …

One can hypothesize about the possible reasons for a link between attractiveness and income. For example, perhaps attractiveness is more likely to get certain people hired or promoted, even if those people have otherwise identical skills. Alternatively, perhaps attractive people are more likely to have better skills when entering the workplace, perhaps because when they were younger or going through school their attractiveness meant that they got more attention and support. Or perhaps being “attractive” is in part the result of a set of soft social skills and behaviors–say, ability to notice and follow social cues about what attractiveness means, together with disciplined patterns of diet, exercise, and other personal habits–which are correlated with skill-sets that are desired in the workplace. Or perhaps the bias arises in part because potential customers in certain industries are more likely to give their eyeballs or their spending power, or both, to more attractive people.

The authors do not discuss the topic of how society might respond to attractiveness bias, and I will follow their example and sidestep the subject here. But at a personal level, it is worth being aware that when you are responding to other people–whether as a co-worker, a customer, a manager, a loan officer, or just someone you walk past on the street–it is likely for most of us that attractiveness bias is shaping our responses.

Industry Concentration: Is It Rising Overall?

One of the simplest ways to measure the degree of competition in an industry is the “four-firm concentration ratio.” Take the market share of the four largest firms in the industry. Add them up. You’ve done it! For example, an industry where the biggest four firms each have 10% of the market would have a four-firm concentration of 40 percent. As a very rough rule of thumb, a four-firm ratio above 80 percent is commonly considered to be “high,” while a four-firm ratio of 50 to 80 percent is medium, and a ratio below 50 percent would be considered “low.”

Of course, one can immediately raise a number of concerns with this simple measure, which is why economists often use either a slightly more complex formula called the Herfindahl-Hirschman Index, or instead just skip past these overall measures of industry concentration and instead work with more detailed models of individual markets. In addition, when talking about industry concentration, a key question is always how one defines the relevant market. For example, does the measure of industry concentration look only at US firms, or also at imports? Is the measure of industry concentration looking at market share across the entire country, or at market shares within certain regions: for example, if there are only three or four big supermarkets chains near where I live, it’s may not matter much for practical competition that there are also completely different big supermarket chains in other regions.

The economic field of industrial organization spends a substantial chunk of its mental energy trying to think through alternative measures of concentration in a variety of contexts. Here, I want to make a simpler point. If there is a large and widespread increase in industry concentration in the US economy, it should presumably show up even in a relatively crude measure like the four-firm concentration index.

Every five years the US Census Bureau does an “economic census,” which is a census of firms in the economy. The most recent economic census was done in 2017, and the results are being released over time as they are tabulated. In December 2020, detailed data on four-firm concentration ratios by industry was released. Robert D. Atkinson and Filipe Lage de Sousa summarize the results in “No, Monopoly Has Not Grown” (Information Technology & Innovation Foundation, often known as ITIF, June 7, 2021)

To understand the specific measure being used here, you need to know that the Economic Census classifies firms according to what is called the NAICS, which stands for North American Industry Classification System. The NAICS starts off by classifying industries into big categories, which are then subdivided into smaller group, and then subdivided again, and again, and again. The biggest categories are called “two-digit”–for example, manufacturing categories all start with a 31, 32, or 33. The three-digit category 335 is “Electrical Equipment, Appliance, and Component Manufacturing.” The four-digit category 3352 is “Household Appliance Manufacturing.” And continuing down the chain, 33510 is “Small Appliance Manufacturing” while 33520 is “Major Household Appliance Manufacturing.”

Obviously, when looking at the extent of competition in an industry, it makes sense to look at the greatest possible level of detail. Not all two-digit manufacturing firms will compete against each other. But “Small Appliance Manufacturing” firms are likely to have similar product line–or the ability to shift easily into similar product lines when it seems profitable to do so. The NAICS divides the US economy into 850 six-digit categories. Thus, a logical question when asking about industry concentration is to look at the extent of concentration and shifts in concentration across these 850 categories.

Here’s the overall breakdown from the ITIF report on level of concentration in 2017 as measured by the four-firm ratio:

One can then compare this level of concentration with the level found in the 2002 Economic Census. Here’s are some results of such a comparison from the ITIF report:

  • Despite widespread claims of widespread monopolization, just 4 percent of U.S. industries are highly concentrated, and the share of industries with low levels of concentration grew by around 25 percent from 2002 to 2017.
  • Overall, Census data show U.S. industries have not become more concentrated: The average C4 ratio (the share of sales that the top four firms capture in an industry) increased just 1 percentage point from 2002 to 2017.
  • The more concentrated industries were in 2002, the more likely they were to become less concentrated by 2017.
  • Prices rose less in industries with higher levels of concentration than they did in the overall economy from 2002 to 2017.
  • There was no relationship between industry concentration and profitability in that period.

I am of course aware that the message of this report runs counter to a wave of claims that there has been a large wave of increased concentration of industry in the United States. But as I have pointed out in the past, these claims about a a meaningful overall increase in industry concentration have been open to question. When you look more closely, it often turns out that the increase in concentration is that the share of the biggest four firms as measured at the six-digit NAICS level rose from, say, 20% to 25%, which is not likely to bring about a meaningful decrease in the extent of competition.

Pointing out specific concerns about lack of competition in certain industries or the growth of certain firms–say, antitrust issues raised by some of the big tech companies–does not prove that there has been an overall decrease in competition for the economy as a whole. The overall lesson here, perhaps, is to focus attention on specific companies and industries that raise concerns about lack of competition, but to be quite cautious about overall claims that competition has increased more broadly in all sectors.

Some Economics of Black America

The McKinsey Global Institute has published “The economic state of Black America: What is and what could be” (June 2021). Near the beginning, the report points out:

Dismantling the barriers that have kept Black Americans from fully participating in the US economy could unleash a tremendous wave of growth, dynamism, and productivity. This research identifies critical gaps Black Americans face in their roles as workers, business owners, savers and investors, consumers, and residents served by public programs.

Much of the focus of the report is on pointing out gaps in various economic statistics. In terms of income. While such comparisons are not new, they do not lose their power to shock. For example:

Today the median annual wage for Black workers is approximately 30 percent, or $10,000, lower than that of white workers … We estimate a $220 billion annual disparity between Black wages today and what they would be in a scenario of full parity, with Black representation matching the Black share of the population across occupations and the elimination of racial pay gaps within occupational categories. Achieving this scenario would boost total Black wages by 30 percent … The racial wage disparity is the product of both representational imbalances and pay gaps within occupational categories—and it is a surprisingly concentrated phenomenon.

Here’s a figure showing a set of occupations, where the circles show the number of occupation. The horizontal axis shows typical wages in the occupation, while the horizontal axis shows the share of workers in each occupation who are black. Some of the larger circles are highlighted and labelled. The pattern that emerges is that blacks are overrepresented in lower-paying occupations.

These gaps in income levels and differences in occupations are part of a broader and interrelated picture, including difficulties that affect black-owned businesses, and lower accumulation of wealth for black households. Here are a few comments on black-owned businesses:

The US Census Bureau’s Annual Business Survey identified some 124,000 Black-owned businesses with more than one worker—which means that they constitute only 2 percent of the nation’s total, far below the 13 percent Black share of the US population. … White entrepreneurs start their businesses with $107,000 of capital on average, but the corresponding figure for Black founders is $35,000. Starting from behind in this way can create a heavier debt burden; in a McKinsey survey, almost 30 percent of Black-owned businesses reported directing more than half their revenues to debt service in 2019.23 Without the funding to launch and sustain their operations, many Black-owned businesses do not survive the startup stage.

These challenges may be one reason that Black entrepreneurship is stronger in less capital-intense industries. Roughly one-third of Black-owned employer businesses are in healthcare and social assistance. Many are home healthcare providers, daycare providers, and physician’s offices. The “low cost of failure” emerged as a theme in interviews we conducted with business owners in the home healthcare field. Access to patients and other providers can come from existing connections, and no capital-intensive equipment is needed. Scaling up such enterprises can be a difficult task, but starting can be relatively simple.

As expected, the differences in income, occupation, and business ownership translate into differences in wealth:

The median Black household has only about one-eighth of the wealth held by the median white household. The actual dollar amounts are striking: while the median white household has amassed $188,000, the median Black family has about $24,000. … Black households start with less family wealth and are constrained in their ability to save We estimate a $330 billion disparity between Black and white families in the annual flow of new wealth, some 60 percent of which comes from inheritances. Every year there is a massive intergenerational transfer of family wealth, creating an effect that is both profound and self-perpetuating. Black families are less likely to receive inheritances, and when they do, the amounts are smaller. The gap in inheritances between Black and white recipients is some $200 billion annually. …

The Black home ownership rate at the end of 2020 stood at 44 percent, which is 30 percentage points below the 74 percent home ownership rate of white Americans.50 While 18.6 percent of white households own stocks, the rate for Black households is 6.7 percent. Consequently, Black households are not positioned for gains when homes appreciate in value or the stock market has an upswing.

Finally, these economic differences translate into even more fundamental measures of well-being, like health and life expectancy:

An important measure of health outcomes is life expectancy. In 2018, life expectancy at birth was 76.2 years for white men but 71.3 years for Black men; it was 81.1 years for white women but 78.0 years for Black women.62 The average gap across both genders was about 3.5 years. If we apply those lost years across the entire Black population, the painful result is that 2.1 million more Black Americans could be alive today with parity in life expectancy. The COVID-19 pandemic has widened the disparity to a five-year gap. If we again apply those lost years across the Black population, the result is now 3.4 million Black Americans who would otherwise be alive today.

It’s fairly to list a bunch of policies that might reduce these gaps, but harder to quantify which policies would have a lasting effect at reasonable cost. Here, I won’t try to unpack those issues. I’ll just note that sometimes the start of a response can be recognizing that there is a problem.

Interview: Amartya Sen on a Bicycle

Christina Pazzanese interviews 87 year-old Amartya Sen (Nobel ’98) for the Harvard Gazette (June 3, 2021), with an emphasis on the long arc of his life and career (‘I’ve never done work that I was not interested in. That is a very good reason to go on.’ June 3, 2021). The interview is full of interesting nuggets, like the time he co-taught a Harvard course on social choice theory with Kenneth Arrow and John Rawls. One point that caught my eye was Sen’s passion since his boyhood for bicycling: 

I was a bicyclist of quite an extreme kind. I went everywhere on bicycles. Quite a lot of the research I did required me to take long bicycle trips. One of the research trips I did in 1970 was about the development of famines in India. I studied the Bengal famine of 1943, in which about 3 million people died. It was clear to me it wasn’t caused by the food supply having fallen compared with earlier. It hadn’t. What we had was [a] war-related economic boom that increased the wages of some people, but not others. And those who did not have higher wages still had to face the higher price of food — in particular, rice, which is the staple food in the region. That’s how the starvation occurred. In order to do this research, I had to see what wages people were being paid for various rural economic activities. I also had to find out what the prices were of basic food in the main markets. All this required me to go to many different places and look at their records so I went all these distances on my bike.

And when I got interested in gender inequality, I studied the weights of boys and girls over their childhood. Very often, it would happen that the girls and boys were born the same weight, but by the time they were five, the boys had — in weight for age —overtaken the girls. It’s not so much that the girls were not fed well — there might have been some of that. But mainly, the hospital care and medical treatment available were rather less for girls than for boys. In order to find this out, I had to look at each family and also weigh the children to see how they were doing in terms of weight for age. These were in villages, which were often not near my town; I had to bicycle there. …

When the Nobel committee after you get your prize asks you to give two mementos or two objects connected with your work, I chose two. One was a bicycle, which was an obvious choice. And the other was a Sanskrit book of mathematics from the fifth century by Aryabhata. Both I had a lot of use for.

Also, although I do not expect to be saying anything similar about my ongoing intellectual in 2047, when I have every intention of turning 87, one cannot help but appreciate Sen’s ongoing zest for what he does. 

I’m planning to do a book on gender. There should be one in about a year or two. There are so many different problems people get confused that I thought I might put together the problems that make up gender disadvantage. It will draw on prior research, but there will be a number of new things in it. …

People have given up hope that I might retire. But I like working, I must say. I’ve been very lucky. I’ve never done, when I think about it, work that I was not interested in. That is a very good reason to go on.

I’m 87. Something I enjoy most is teaching. It may not be a natural age for teaching, I guess, but I absolutely love it. And since my students also seem not unhappy with my teaching, I think it’s a very good idea to continue doing it.

For another interview with Sen, this one from summer 2020, see “Interview with Amartya Sen: 
Economics with a Moral Compass?”
 (August 5, 2020). 

From Pandemic to Digitalization to Productivity?

We know that the pandemic caused people and firms to make much more widespread use of digital technologies: working from home, ordering on-line, tele-medicine, education from K-12  to college delivered on-line, and so on. Indeed, it seems likely that this surge of digital activity is also providing an incentive for substantial investments in physical capital, intangible capital (like software), and complementary human skills to make use of these investment. Might these shifts in patterns and investments provide a boost to improved productivity growth in the next few years? 
The Group of Twenty has published a report (prepared by staff at the IMF) on these subject: “Boosting Productivity in the Aftermath of COVID-19” (June 2021). The report suggests the possibility that while many people will be better off because of the shift to digital technologies, these gain in well-being may not be well-reflected in conventional economic statistics like GDP. 
It’s worth noting that nothing in the report seeks to put a happy face on the economic side of the pandemic experience. Unemployment has soared. Worker skills have been unused, and in some cases will have depreciated. Firms and communities have suffered, many of them grievously. As the report notes: “For instance, the so called `jobless recoveries’ from previous US recessions were driven by contractions in routine occupations, which account for about 50 percent of total employment, that are never recovered. More recently, the COVID-19 shock has also hit sectors that are more vulnerable to automation much harder and lowered the share of low-skilled and low-wage workers in the workforce. As we look ahead, the productivity and earnings of low-skilled workers that have lost their jobs in sectors vulnerable to automation are therefore at risk …”

But it is also true that use of digital technologies has increased, in ways that seem likely to persist, at leas in part, as the pandemic recession faced. Indeed, this shift to heavier use of digital technologies is one reason why stock prices of leading tech companies have done so well in the last year or so. Here are a couple of interesting illustrations from the report. The first shows the pattern of new US patent applications related to remote work and e-commerce, and how it has risen. The second shows the results of a survey of business executives, emphasizing that for most of them, the pandemic recession led to heightened efforts to digitize and automate their operations. 

The report discusses the extent to which this shift may increase productivity: for example, the reallocation of resources away from less-productive to more-productive firms should boost productivity. The report expresses cautious and hedged optimism about the chances for productivity: for example, “In sum, the impact of reallocation so far looks beneficial for productivity, but much remains to be learned and it is associated with several concerns.”

The report also raises the difficult question of productivity measurement. Workers who have greater flexibility to work from home may benefit, for example, from less time spent commuting. But shorter commutes don’t provide a direct boost to JEP. If I have groceries delivered more often, but my purchase of groceries is pretty much the same, the benefits to me may not be well-captured by conventional economic statistics. If I see my doctor on-line, or children see a K-12 teacher online, or college students attend classes remotely, there will be a mixture of effects on the quality of what is provided and the costs of providing it that will not translate in a simple way into productivity statistics. These kinds of issues have been lurking in the productivity statistics for years, but the economic after-effects of the pandemic may strengthen them.

Mismeasurement of the digital economy has been an often-cited contributor to the prolonged slowdown in measured productivity growth prior to the COVID-19 pandemic. As the productivity slowdown occurred alongside a fast pace of innovation in the hard-to-measure digital economy, a commonly mentioned contributor to the measured slowdown is the inability to capture well in price statistics and deflators the increases in convenience, varieties, free online products, and lower quality-adjusted prices that arises from the digital economy. … Looking forward, if the pandemic accelerates growth in the digital economy, its contribution to mismeasurement may become more salient. For example, greater prevalence of remote work and online interactions across borders may reduce travel costs, which, if not properly captured, may lead to an underestimation of productivity growth. A shift to digital and peer-to-peer platforms could also bring added convenience, making it feasible to access an increasing number of varieties and lower prices, which, if not properly accounted for, would also result in mismeasurement.

Finally, it’s worth noting that the pandemic will affect future productivity growth in a number of ways, not just via the effects on digitalization. For example, many students around the world have experienced a severe disruption of their education. The report notes: 

School closures affected 1.6 billion learners globally at the peak of the pandemic and continue to disrupt learning for millions. These disruptions had disproportionately adverse impacts on schooling in economies with preexisting gaps in infrastructure (such as access to electricity and internet), which constrained their ability to implement remote learning. Girls and learners in low-income households faced disproportionately greater risk of learning losses as they lost a boost from peer-effects that occur in school and may have been less likely to have parental support for remote learning. Women may also have needed to take on additional caregiving and teaching responsibilities while at home, putting them at a disadvantage in the labor market. These interruptions to learning and work will likely set back human capital accumulation—with such effects spread unevenly across generations, genders, and income levels, and with adverse implications for longer-run productivity.

Posted by Timothy Taylor at 12:59 PM Email ThisBlogThis!Share to TwitterShare to FacebookShare to Pinterest

FRIDAY, JUNE 11, 2021

The Social Nature of Government Actions

Economics famously begins with an idea of individuals pursuing their own interests, and then discusses both the positive and negative dynamics that can emerge. But there has been a long-time pattern in human affairs, going back to the days of the hunter-gatherers, that certain outputs have been produced socially–by families, communities, and in modern times also by government. Emmanuel Saez explores this issue in his American Economic Association Distinguished Lecture at the virtual AEA meetings last January on the subject, “Public Economics and Inequality: Uncovering Our Social Nature” (AEA Papers and Proceedings 2021, 111: 1-26, subscription required, but freely available at Saez’s website here). Saez writes: 

[O]ur social nature, absent from the standard economic model, is crucial for understanding our large modern social states and why concerns about inequality are so pervasive. Taking care of the young, sick, and elderly has always been done through families and communities and likely explains best why education, health care, and retirement benefits are carried out through the social state in today’s advanced economies. Behavioral economics shows that we are not very good at solving these issues individually, but descriptive public economics shows that we are pretty good at solving them socially. …  Even though an individual solution through markets is theoretically possible, it does not work well in practice without significant institutional or government help. Human societies are good at providing education, health care, and retirement and income support even though individuals are not.

Although Saez offers a brisk overview of earlier human societies, his main focus is on what he calls “”the rise of the social state in the twentieth century:” He writes: 

Perhaps the most striking fact in modern economies illustrating both our social nature and concerns for inequality is the size of government and the large direct impact it has on the distribution of economic resources. In advanced modern economies, we pool a large fraction of the economic output we produce through government. In the richest countries today, taxes generally raise between 30 and 50 percent of national income and are used to fund not only public goods needed for the functioning of the economy but also a wide array of transfers back to individuals, both in cash and in kind. Even though modern economies generally allocate the fruits of production to workers and owners through a capitalistic market system with well-defined property rights, as societies, a significant fraction of market incomes, typically between one-third and one-half, is shared (that is, effectively “socialized”) through government.

Here’s are a couple of figures showing the rise in government spending in advanced economies in the 20th century:

(In the figure “Regalian public goods” is a category that Saez defines as the basic roles of a very limited government, including defense, law and order, administration, and infrastructure).
As Saez notes, the US economy is near the lower end of this range–but it’s still a substantial share. I would add that a significant part of the difference is that the US has kept a large portion of its health care spending in a heavily regulated private sector. Saez also notes that there is relatively little cross-border redistribution, and when it happens, it’s often in the form of disaster relief. People seem to define their circle of sharing within their country, or to some extent within a lower-level jurisdiction like a state or city, 

Again, the big four social categories on which Saez focuses are education, retirement benefits, health care, and income support. To get a sense of the tone of his argument, here are a few of his comments on these categories: 

Historically, mass education is always government driven through a combination of government funding (at all levels including higher education) and compulsory schooling (for primary and then secondary education). … 

Before public retirement programs existed, a large fraction of the elderly was working (80 percent of men aged 65 or older were gainfully employed in the United States in the late nineteenth century …). The elderly who could no longer work enough to support themselves had to rely on family support. Public retirement systems were a way to provide social insurance through the state instead of relying on
self-insurance or family insurance. … 

[U]niversal health insurance creates significant redistribution by income and also, of course, by health and health-risk status. One important question is why health-care quality is the same for all in such universal health-care systems (at least as a principle, not always realized in practice). Why isn’t health insurance offered in grades, with cheap insurance covering only the most cost-effective treatments. Probably because humans are willing to spend a lot of resources to save a specific life, that is, an actual person with a condition that can be treated. This is likely a consequence of our social nature shaped by evolution: taking care of the sick or injured was helpful for group survival. This makes withholding treatment to the poorly insured socially unbearable. …

People make mistakes in health- care utilization and treatment choices. Copayments and deductibles lead consumers to reduce demand for high-value care. This may explain why universal health-care systems have low copays and deductibles and why health-care decisions for patients are made primarily by health-care professionals. Like for education, the difficulty for users to understand and navigate health-care choices implies that the market does not necessarily deliver efficiency. In sum, the problem of health care is also primarily resolved at the social level rather than the individual level. … 

Everywhere, there is strong social reprobation against “free loaders” who could work and support themselves but decide to live off government support This is why income support is concentrated among groups unable or unexpected to work, such as the unemployed, the disabled, and the elderly.

As Saez discusses, the fact that advanced societies have decided that government provision will play such a large role in these four areas is rooted in other social judgements: for example, judgements about the fairness and importance of widespread education for children, judgements about whether the elderly should need to work (and how to define who is “elderly”), judgements about whether the sick and injured will have access to care, and judgements about which groups of people deserve income support and under what conditions. Of course, this kind of social consensus can shift. We saw a shift in the 1980s and 1990s about whether single mothers with small children were expected to work, or not. As another example, back in the 1980s, 50-55% of Americans in the 16-19 age bracket were in the labor force; now, it’s about 35%. A substantial part of that shift is in our sense of what people in that age group should be doing with their time. Saez writes: 

However, the social state also intentionally reduces labor supply by design through various regulations: child labor prohibitions and compulsory education limit work by the young, retirement benefits sharply reduce work in old age, and overtime hours-of-work regulations and mandated paid vacation (for example, five weeks in France) reduce work across the board. This implies that labor supply should be seen partly as a social choice, with society having disutility of labor for the very young, the old, and very long hours with no vacation break.

There’s much more in the lecture itself. But the main theme deserves attention. Saez writes: “Therefore, social organization does seem to come naturally to us. We can easily take a group perspective and act accordingly.” Understanding the group perspective and the social organizations that form as a result seems like an important tool for understanding what we expect from government–and what some of the barriers are to redesigning government programs to operate more effectively 

How to Improve College Completion Rates: The Time Commitment Problem

The US higher education system does an OK job of enrolling US high school students. About 70% of US high school graduates enroll in a two-year or four-year college. But the higher education system does a poor job actually producing graduates who have completed college. About half of students who enroll at a four-year college graduate within six years; the completion rate is lower for two-year colleges. If the goal of getting more high school student to attend college is to be a meaningful one, it needs to be accompanied by efforts to raise the college completion rate. 

Philip Oreopoulos discusses these issues in  a review article “What Limits College Success? A Review and Further Analysis of Holzer and Baum’s Making College Work” (Journal of Economic Literature 2021, 59:2, 546–573, subscription required). As Oreopolous details, Holzer and Baum provide an overview of steps to encourage college enrollment and completion. In particular, some of the steps to encourage college enrollment can be fairly low-cost, like requiring high school students as part of their coursework to fill out at least one college application and to take the SAT or ACT, and having states dol a better job of communicating about available financial aid to low-income households. 

Here, I want to focus on policies more directly aimed at improving college completion. For example, one approach discussed in the Holzer and Baum book is a comprehensive set of support services for first-year students. Oreopoluos describes perhaps a prominent example of such a program this way (citations omitted): 

Exhibit A for demonstrating how to improve college access and success is the Accelerated Study in Associate Program (ASAP). MCW [Making College Work] and many other researchers point to it as the central example worth considering. ASAP provides incoming freshman an envelope of comprehensive support services, including tutoring, counseling, career advising, free public transportation passes, and funding for textbooks. Taking advantage of the potential benefits of more structure, students are required to meet regularly with their advisor and tutors, attend a student success seminar, and enroll full-time to participate. The program was experimentally tested on low-income students with remedial needs at CUNY in colleges where the three-year graduation rate was only 20 percent. ASAP doubled graduation rates at CUNY, and similar impacts on persistence were replicated in Ohio ….Among the evidence we have, comprehensive support programs such as ASAP offer the most promise for improving college completion, at least among community college freshman from disadvantaged backgrounds. The impact of ASAP is the largest I know of, compared to other college program evaluations … The program represents an impressive “proof of concept” for how much we could help if we offered a gamut of student support and made participation mandatory. As impressive as the results are—doubling completion rates from 20 to 40 percent— they also highlight serious policy limitations. Even with a full range of proactive mandatory support services and financial incentives to stay engaged, 60 percent of ASAP participants still did not complete their degrees. The best program we know, which … many administrators feel is unaffordable, still fails to help more than half of its target population.

One problem underlying low college completion rates is that the incoming students lack necessary skills to do college-level work. Such students may be admitted to college but then required to take remedial classes before they can begin the classes that lead to their desired degree. Oreopoulous describes the tradeoffs this way: 

Many community colleges provide open access, meaning that they admit any applicant with a high school degree into at least a general studies program. This level of access increases opportunity for all graduating high school seniors to pursue higher education at a relatively low cost. The downside is that many entrants are not well prepared to handle the academic standards of their program. The same colleges therefore often require entrants to take remedial mathematics and English courses before being allowed to take courses that would contribute toward a degree or certificate in their desired program. “About 68 percent of students entering public two-year and 40 percent of those entering public four-year colleges in 2003–2004 took at least one remedial class by 2009” (p. 21). Freshmen find themselves feeling stuck working on subjects they covered earlier and concerned about the longer road they face to completion.
College dropout rates for those taking remediation courses are shockingly high—Jaggars and Stacey (2014) report a 72 percent dropout rate among community college students who take a remedial education course. Adams et al. (2012) use data from 33 participating states and find a 65 percent overall dropout rate by sixth year for students taking remedial courses. Those who require remediation are obviously less prepared and less likely to graduate compared to those who don’t require it, but a consensus of policy researchers agree that reform is needed to avoid discouraging these marginal students facing long delays to complete their degrees.There may be ways to make such remedial classes feel like less of a hurdle to students: for example, by figuring out ways that students can at least start their desired course of study at the same time as the remedial course, and thus do them side-by-side, rather than being required to start their college experience completely focused on remedial courses. Of course, the better answer would be for high schools to produce fewer graduates who need remedial courses.

Oreopolous also focuses on  theme that I have often found myself emphasizing to prospective or newly-arrived college students: making the necessary time commitment. As he writes: “Many college administrators and faculty recommend two or three hours of study for each hour a student spends in class, implying 25 to 35 hours of effort outside of class for someone enrolled full-time (there is a reason
they call it “full-time” enrollment).” However, a typical college student actually studies about 15 hour/week (or so they say), which  means that a sizeable minority study less than that. 
Oreopoulous discusses the results of some polling he carried out among first-year students at the University of Toronto about their expectations of outside-of-class study time. He  writes: 

Low-performing students admit to time management problems and procrastination, but even when asked to plan their hours in advance, they often set low goals. .,. If students entered a plan with fewer than 15 hours of routine study [as their personal plan on the survey form] , we asked “[W]e’d like to better understand how and why you decided on this number. Is it because you did not expect to gain much from studying more, or because you did not think you would haveenough time, or some other factor? Please share your thoughts in a paragraph or two” … [A]mong those who eventually ended up with a fall grade average less than 60 percent. … [a] majority said they felt their target was fair and reasonable. Some justified their answer based on their successful high school experience;
others said they wanted to leave room for sports, extracurricular activities, and friends. Very few of these students anticipated doing so poorly and none said they felt constrained from work. In fact, about half said they were intending to complete graduate studies in the future, 58 percent expected to receive above average fall grades, and the average expected economics grade was 76 percent. It seems as though these students had the wrong reference point for sufficient study time. By the end of the semester … these kinds of students update their academic expectations downward, but rather than respond by planning to study more, they tend to accept their academic fate and plan to study about the same the following semester.

Of course, some college students have highly limited time to study because of job or family responsibilities. But those examples are not the core of the time commitment problem. Moreover, Oreopolous and his co-authors have found no noticeable effects on grades from trying to encourage more study time with an online program of information, reminders, and coaching. Trying to raise college graduation rates, or levels of academic achievement, for full-time students who are only putting in 15 hours or less of study time per week will inevitably be an uphill battle.