Accountant CEOs and firm innovation

When you find out that someone is an accountant, it evokes certain stereotypes. [*] Adjectives like risk averse, conventional, conservative, even boring, may come to mind. Many accountants go on to become leaders of firms. Given that the stereotypical traits of accountants are inconsistent with creativity and innovation, does that make firms that have accountant CEOs less innovative?

That is the question addressed by this recent article by Jian Cao (Florida Atlantic University) and co-authors, published in the journal Research Policy (sorry, I don't see an ungated version online). They collated data from financial statements from CompuStat, and their data on the backgrounds of CEOs and other board members comes from BoardEx. Their final sample includes:

...41,020 firm-year observations from 2001 through 2018, and accountant CEOs are present in 9.8 % of the firm years.

Their main variable of interest is whether the CEO has an accounting background, which they define as:

...equal to 1 if the CEO is a CPA or has worked in accounting-related roles (including “accountant,” “accounting,” “controller,” or “audit” in the job title) and 0 otherwise...

The main outcome measures (as proxies for innovation) are:

...the number of patent applications filed by each firm in each year (Patent), the number of citations subsequently received by the filed patents in each year (Citation), and the number of citations per patent (CPP).

Those output measures are the shares of patents within each "four-digit CPC technical scheme (used by the USPTO to classify patents) and year". Apparently, that deals with some bias issues in the patent data (although it isn't clear to me whether they then simply add up the shares across multiple different technical schemes, or average them, or something else). The analysis controls for a range of other CEO attributes (like tenure, age, and whether they have an MBA) and firm attributes (like board size, and various financial indicators and industry variables). Overall, Cao et al. find that:

Consistent with our hypothesis (H1), across all specifications, the coefficients of Acct. CEO are negative and significant (p-value <0.01), both statistically and economically. Model (1) implies that firms with accountant CEOs generate 46 % fewer patents... Models (2) and (3) indicate that firms with accountant CEOs are associated with fewer patent citations and fewer citations per patent. The coefficients imply that firms with accountant CEOs are associated with patent citations and citations per patent that are lower by 40 % and 27 %, respectively.

So far, so consistent with stereotypes. Next, they look at the effects on patents that are exploratory (involving new knowledge) or exploitative (involving the exploitation of existing knowledge). In that analysis, they find that:

...the coefficient on Acct. CEO is not statistically significant for the Exploitative ratio variable... and the relationship between Acct. CEO and the Exploratory ratio is negative and significant... Firms with accountant CEOs exhibit an Exploratory ratio that is 21 % lower than that of their non-accountant counterparts...

Cao et al. then look at 'innovation efficiency', in terms of the cost per patent, and find that:

...the coefficients of Acct. CEO are negative and significant, both statistically and economically. Model (1) implies that firms with accountant CEOs produce patents at an average cost per patent that is lower by $0.14 million... than the cost incurred by firms with non-accountant CEOs...

Models (2) and (3) indicate that firms with accountant CEOs are associated with lower R&D capital inputs for patents and a lower average cost per citation. The coefficients imply that accountant CEOs reduce R&D capital per patent by $0.11 million or the cost per citation by $0.09 million...

Finally, Cao et al. differentiate their results between high-growth industries and low-growth industries. The hypothesis is that industry growth moderates the effect of having an accountant CEO, and indeed that is what they find:

Results... indicate that accountant CEO–led firms are associated with lower innovation output only in low-growth industries. The coefficient on Acct. CEO across all three IO proxies... is more negative and statistically significant for low-growth industries. Accountant CEOs are significantly associated with a 61 % lower patent count, a 53 % lower citation count, and a 49 % lower citation count per patent than non-accountant CEOs whose firms operate in low-growth industries. By contrast, in high-growth industries, no coefficient on the IO proxies is statistically significant...

Results... show that, in low-growth industries, firms with accountant CEOs are associated with a smaller fraction of both explorative and exploitative patents. Still, the effect on the Exploratory ratio... is three times greater than that on the Exploitative ratio... In high-growth industries, neither coefficient is statistically significant...

...firms led by accountant CEOs exhibit greater innovation efficiency, but only in high-growth industries. The coefficient on Acct. CEO for all IE proxies... is much greater in magnitude and is statistically significant for high-growth industries. Accountant CEOs are significantly associated with a reduction of $0.19 million in R&D spending per patent, $0.17 million in R&D capital per patent, and $0.16 million in R&D spending per citation. None of the three coefficients is statistically significant for the IE proxies in low-growth industries.

Taken altogether, these results are consistent with the stereotype of the accountant CEO. Cao et al. conclude that:

...CEOs’ backgrounds in accounting shape their firms’ innovation styles and performance. Firms with accountant CEOs tend to generate lower innovation output, particularly in exploratory innovation, which aligns with the idea that accountants primarily represent the conventional personality type. Meanwhile, firms led by accountant CEOs achieve greater transformational efficiency in turning R&D investments into innovation output... Our results imply that accountant CEOs facilitate innovation efficiency in high-growth industries while constraining potentially undesirable innovation in low-growth industries.

At this point though, it is worth going back to the beginning. Where does the stereotype of the conservative accountant come from? In their background section, Cao et al. point to this 1994 book chapter by Holland et al. and say:

Holland et al. (1994) code personality types for a multitude of occupations, and the accountant stereotype aligns well with the conventional personality type in Holland’s hexagon. Conventional character descriptions include “conformity, defensiveness, inflexibility, inhibition, obedience, prudishness, and a lack of imagination” (Holland et al., 1994, p. 6).

Ok, but where is the evidence that accountants are really like that at all? Cao et al. cite four studies, all of which were published before the Holland et al. book chapter. In other words, all of the evidence that Cao et al. provide for accountants being this 'conventional character type' is all over thirty years old! Maybe that's not such a problem though, because the average accountant CEO in their sample is 53 years old, so was probably a junior accountant in the 1990s. However, the average length of 'accounting career' was only 1.26 years for the CEOs in this sample. That's hardly long enough to have much of an effect, unless Cao et al. are claiming that 'conventional character types' select into an accounting career. But what character types select into a career, but only spend 1.26 years in that career before doing something else (and becoming a CEO later)?

So, maybe the news isn't so bad for accountants after all. I'm not sure I believe this study shows quite what it purports to show. A CEO must surely spend more than 1.26 years as an accountant, before you classify them as an 'accountant CEO'. Perhaps it is instead picking up that CEOs who move quickly between positions (the average CEO tenure in the sample was 4.01 years, and statistically significantly shorter than for non-accountant CEOs) are more likely to want to invest in exploitative patents, and less willing to invest in patents overall?

*****

[*] And when you find out that someone is an economist, it also evokes certain stereotypes. I was once an accountant. Now I'm an economist. Does that evoke the intersection of those two stereotypes?

Minimum wages and student summer employment

The research literature is starting to coalesce around minimum wages have small disemployment effects, consistent with the standard demand-and-supply model (see the meta-analysis in this post, or the links at the end of this post, for more details). To add to that evidence, this recent article by Adam Wright, Darius Martin, and John Krieg (all Western Washington University), published in the journal Contemporary Economic Policy (), looks at the impact on student summer employment. Specifically, Wright et al. use data from Washington state, and focus on students enrolled in Western Washington University in Bellingham, which has about 15,000 undergraduate students.

Wright et al. match student data from 2013 to 2019 to employment records from the Washington State Employment Security Division (ESD) from 2009 to 2019. The ESD data includes both earnings and hours worked, but only for employers that pay into the state's unemployment insurance programme. This excludes workers for the federal government and the self-employed, who are unlikely to be students. However, it also excludes students "students working under financial assistance programs provided by the school", so Wright et al. supplement the ESD data with data on student employees at WWU. Because the employment data starts before the student data, Wright et al. can control for past work experience. And because the data are quarterly, they are able to look separately at employment effects over summer, and during term time.

There are over 31,000 students in the analysis sample, and nearly 260,000 student-by-quarter observations. Wright et al. apply a fairly straightforward panel regression model, controlling for work experience and for the unemployment rate of Whatcom County (where the WWU campus is located). They look at three outcome variables: (1) hours worked; (2) wage income; and (3) a binary variable indicating whether a student worked any hours at all. The overall effects of the minimum wage are not statistically significant. However, when looking at the results by season, Wright et al. find that:

In summer, when students are most likely to work, higher minimum wages significantly predict reduced work hours and the probability of employment, whereas the relationship with income is negative but imprecisely measured. In particular, the coefficient estimates imply that a 100% increase in the minimum wage is associated with 90.08 fewer hours worked in the summer and a reduction in the probability of work by 34.2% points. This suggests that the 16% minimum wage increase experienced in 2016–2017 was linked to a 14.4 h decrease in summer hours worked (an 8.5% decrease relative to the average) and a 5.5% point decrease in summer employment (an 8% decrease)... Minimum wage policy does not appear to predict disemployment in non-summer quarters, leading wage income to rise with increasing minimum wages in winter and spring...

So, minimum wages reduce student summer employment. It is unclear why Wright et al. decided to illustrate the size of the effect with a 100% change in the minimum wage (although that is what the coefficient in the table represents), since that size of change is never observed in the data. It is better to say that a 10% increase in the minimum wage would reduce hours worked in summer by about nine hours (compared with a mean of 170 hours of work in the summer quarter). It's a small effect, but not zero. And since student summer employment tends to be concentrated in low-wage industries like retail or hospitality, that makes sense. Turning to the effect of previous work experience, Wright et al. find that:

...higher wage income in winter and spring quarters when minimum wages increase only holds for students with prior work experience. However, there is statistically significant drop in employment in summer for both students with and without work experience. We estimate that a 100% minimum wage increase for workers with no pre‐matriculation work experience is associated with 115 fewer hours of work and a 38.8% point decrease in the probability of employment during the summer quarter. These estimates imply that the 16% minimum wage increase in 2017 corresponded with inexperienced students working 18.1 fewer hours and being 12.8% points less likely to be employed in the summer. These results are attenuated for those who entered WWU with work experience: a 100% minimum wage increase corresponds to 21.4% point reduction in summer employment for this group whereas the relationship between minimum wages and hours worked is negative but statistically insignificant.

The key finding is that students with no prior work experience and more negatively affected by the minimum wage increase than students with prior work experience. This is to be expected, since employers facing a higher minimum wage would be likely to concentrate employment in more experienced (and more productive) workers, even within summer student workers.

Next, comparing the effects between students who are local to Whatcom County and those who are not, Wright et al. find that:

...the negative relationship between minimum wages and summer employment only occurs for non‐locals, with estimates comparable to those experienced by students with no prior pre‐matriculation work history... Taken together, the full sample and quarterly results indicate that those with higher search costs may be more negatively impacted by minimum wage changes.

The mention of search costs here is important. In a search model of the labour market, workers face a search cost, made up of the time and effort spent looking for a job. Locals face lower search costs, because they likely have networks of local acquaintances and friends who can more easily help them find work, compared with non-locals. These results show that, in the context of higher minimum wages, those differences in search costs really matter.

Wright et al. are careful to point out that their results are correlations rather than causal. Specifically, their analysis lacks a control group. Nevertheless, it provides some descriptive evidence that is consistent with the emerging consensus of small but significant disemployment effects of the minimum wage. However, it would be interesting to see whether these results stood up to a more careful analysis using methods designed to elicit causal impacts. Nevertheless, Wright et al. conclude that:

Our results suggest that minimum wages particularly hurt inexperienced workers in summer, the quarter in which students tend to work most.

As someone who teaches students who rely on summer employment to build up reserves that they can draw on during term time, these results, even if they are not definitively causal, are a worry.

Read more:

• A bunch of research findings on minimum wages in New Zealand
• The disemployment effects of Canadian minimum wages
• Jeffrey Clemens on the disemployment effects of the minimum wage
• The impact of an online minimum wage
• Seattle's minimum wage, revisited
• The latest evidence supports negative employment effects of the minimum wage
• Latest research suggests the minimum wage DOES reduce employment
• More empirical support for the disemployment effects of the minimum wage
• The minimum wage and job vacancies
• The minimum wage and hiring standards
• The employment effects of the minimum wage, from American Community Survey data
• The minimum wage and homelessness
• New York restaurants find a new way to respond to the minimum wage
• The minimum wage and workplace safety
• How employers respond to minimum wage increases
• Minimum wage increases and employer noncompliance
• Higher minimum wages and poverty revisited
• Minimum wages and health
• Effects of the minimum wage on the nonprofit sector
• 15 years of US research on the minimum wage elasticity of employment
• Local minimum wages and low-quality housing rents in Japan

Internal rates of return for college majors in the US

This week, my ECONS102 class covered the economics of education. Part of that topic involves understanding the private education decision - the decision an individual makes in weighing up the costs and benefits of further education, acknowledging that most of the benefits of education happen in the future, and therefore must be discounted in order to be compared to the costs, which are mostly incurred in the present or near future. As part of that topic, I get my students to evaluate the costs and benefits of their own education decision. For most students, the benefits far outweigh the costs, using a discount rate of 10 percent. However, if the discount rate were higher (and so the future was discounted more heavily), the present value of the benefits would be lower, and more of the students would end up with costs exceeding benefits. The discount rate that perfectly balances benefits and costs is called the internal rate of return (IRR). A higher internal rate of return means that the benefits outweigh the costs by more (since it requires a higher discount rate to equalise the benefits and costs).

Anyway, that is a long introduction to this 2024 article by Liang Zhang (New York University), Xiangmin Liu (Rutgers University) and Yitong Hu (New York University), published in the American Educational Research Journal (ungated version here). They estimate the distribution of internal rates of return for college majors in the US, using earnings data from the American Community Survey from 2009 to 2021. Importantly, in terms of benefits they look at the incremental (extra) earnings over and above what high school graduates earn. The costs side of the equation uses data from the National Postsecondary Student Aid Study (NPSAS).

Zhang et al. use the benefits and costs data to estimate IRRs. Overall, in their preferred specification:

...the IRR for a college education is estimated to be 9.88% for women and 9.06% for men.

However, those overall results hide a lot of heterogeneity. Looking across different majors:

...there are large variations in IRRs across college majors. Specifically, engineering and computer science majors command the highest IRRs among all majors, exceeding 13%. Additionally, a few other majors such as business, health, and math and science have higher IRRs than the overall college population, ranging from 10% to 13%. The next tier, including biology, social science, and other majors, have IRRs of 8% to 9%. Finally, education and humanities and arts majors have the lowest IRRs, especially for men in those fields.

Interestingly, on the gender difference in IRR, Zhang et al. note that:

...on average, women tend to have higher IRRs than men for college education in general. However, this does not necessarily mean that female graduates earn more than their male counterparts who are men. In fact, the opposite is true: Women in our final sample earned approximately 28% less than men among college graduates and about 33% less than men among high school graduates. The lower earnings for women among high school graduates also means that women generally have lower opportunity costs while attending college.

I still read that as saying that women gain more from a college education than men do, even if the gender wage gap means that they don't earn as much as men. The difference in IRRs between women and men would be even greater if there was no gender wage gap.

Further exploring the heterogeneity, Zhang et al. use quantile regression to evaluate the distribution of internal rates of return. The resulting distribution can then be summarised graphically, as in their Figure 2:

The axes are difficult to read on that figure, but the values are somewhat less interesting than the distributions. The x-axis shows (I believe) the decile of the earnings distribution, so this figure shows how IRRs differ between those graduates who earn at the bottom, middle, and top of the earnings distribution (overall at the top, and for each major at the bottom). The interesting ones are those that somewhat counterintuitively have lower IRRs for those at the top of the earnings distribution than for those in the middle or at the bottom (like education, engineering, and health). What is likely happening there is that, in those fields, the difference in earnings between the top earners who went to college and the top earners who have only a high school education, is less than the difference for those lower down the earnings distribution. That makes sense in engineering (top engineers without a degree might still earn a lot) and health (nurses don't necessarily require a college degree), but is less easy to explain for education. Zhang et al. describe the results, but unfortunately don't offer a good explanation for them.

Overall, and for each major, it does seem that a college education is a good investment. A nine percent return is difficult to match. There is, as you would expect, some variation between different majors, and even within majors. That variation also needs to be taken into account by anyone carefully considering the private education decision.

This week in research #94

Here's what caught my eye in research over the past week (in what must have been a slow week!):

• Gebrewolde, Rockey, and Ullah (open access) introduce a new measure of economic gender inequality (EGI) based on the ratio of women’s share of national labour income to men’s, then show that population-weighted average EGI increased between 1994 and 2014, with much of the higher EGI in poorer, more populous, countries explained by the lower rates of female employment

The business economics of The Summer I Turned Pretty

I tell my students that, once they start to understand some economics, they start to notice it everywhere. It's not just a throwaway line. It really is true. As an example, one of my ECONS101 students excitedly shared with me a short example on the business economics of Prime Video's show The Summer I Turned Pretty. That show is not really my cup of tea (I prefer something like The Witcher). However, the pricing strategy that Amazon employed with The Summer I Turned Pretty is quite interesting to tease out. Specifically, when season 3 of The Summer I Turned Pretty was released on Prime Video, Amazon simultaneously released seasons 1 and 2 for free on YouTube. What was Amazon trying to do?

I believe that this was an example of Amazon using customer lock-in to increase the number of subscribers to Prime Video. Customer lock-in occurs when consumers find it difficult to change once they have started purchasing a particular good or service. High switching costs (the cost of switching from one good or service to another, or from one provider to another) are likely to generate customer lock-in, because a high cost of switching can prevent customers from changing to substitute products.

Where are the switching costs here? With a television show, viewers get invested in their favourite characters and in following particular storylines. If a viewer was to watch something else instead, they face a switching cost of missing out on knowing what their favourite characters are doing, or how the storylines that they were following play out. So, once a viewer starts watching a particular television series that they like, they are reluctant to stop. This is the switching cost in action - the viewer is locked into watching that series.

By releasing the first two seasons of The Summer I Turned Pretty for free on YouTube, Amazon is hoping that will attract new viewers, who will become locked into watching it, and then pay for a subscription to Prime Video in order to continue watching season 3. More Prime Video subscribers equals more revenue (and profits) for Amazon. And because very few consumers would be attracted to Prime Video for the first two seasons of this show, making them available for free didn't really have a high opportunity cost for Amazon (and the challenge of cancelling subscription services creates a further degree of lock-in). 

This strategy is essentially a form of multi-period pricing - setting the price low initially (free for the first two seasons), before raising the price once consumers are locked in (since they have to have a paid subscription to watch season 3). This works because locked-in customers have less elastic demand for a product (they are less price sensitive). So, charging a higher price to locked-in customers than to those who are not (yet) locked in is a profit-maximising strategy.

There is a further aspect of this strategy that I find equally interesting. The student I was speaking with noted that some of her friends had waited until the last episodes of The Summer I Turned Pretty were released, before subscribing to Prime Video for one month and binge-watching the whole season and then cancelling their subscription. In contrast, my student was more impatient and watched each episode as it was released. However, that meant paying for three months of Prime Video subscription.

This sounds a lot like price discrimination - charging different prices to different consumers for the same good or service (and where the difference in price doesn't reflect a difference in costs). In this case, super-fans of the show will be impatient and wanting to watch each episode as it is released. They have short time horizons (they want to watch now), so their demand is less elastic. And with less elastic demand, the profit-maximising price is higher. In contrast, casual fans of the show will be more patient, and happy to wait and binge-watch the whole season in a day. They have longer time horizons, so their demand is more elastic. And with more elastic demand, the profit-maximising price is lower.

By releasing one episode a week, Amazon is able to effectively price discriminate for both groups. The impatient fans (with inelastic demand) pay for three months of Prime Video (a higher price), while the patient fans (with more elastic demand) pay for one month (a lower price). Even better, Amazon doesn't even need to be able to tell these fans apart, because the fans make the decision themselves about what price to pay.

Economics is all around us. You just need to keep your eyes open, and you will see it.

[HT: Georgie from my ECONS101 class]

The effect of studying economics on political attitudes

Every year, I get my ECONS102 class to complete the Political Compass test as an extra credit task. Having done this for 16 years, I have noticed some general trends. First, the mean (and median) result for each class is always to the economic left and social libertarian. I put this down to the centering of the test being relevant for a general US population, which would likely be to the right of a general New Zealand population. Second, there is a strong positive correlation between the economic left-right axis and the social libertarian-authoritarian axis. Those on the economic left are more likely to be social libertarian (as is the mean and median for my class).

While this exercise is interesting and provides some descriptive evidence on the political attitudes of my economics class, it doesn't really answer a broader question: do political attitudes vary systematically between different fields of study? I can certainly identify some of my past students who were outliers in the test, and link them with their majors, but those anecdotes don't really hold much value.

Fortunately, this 2019 article by Mira Fischer (University of Cologne) and co-authors, published in the European Journal of Political Economy (ungated earlier version here), offers more definitive descriptive evidence. They make use of a large dataset collected over a long period of time:

We use data from a student survey administered by the Research Group on Higher Education which is supported by Germany's Federal Ministry of Education and Research. Beginning in the winter semester of 1982/83, data on about 8000 university students have been collected every two or three years...

The questions inquiring about the students' political attitudes were first included in the second wave and have been asked ever since. The dataset contains 100,420 observations collected in twelve waves and comprises 1032 variables, most of which were included in several waves...

The questionnaire distinguishes eight fields of study: humanities, social sciences, law, economics, medicine, engineering, natural sciences, and other subjects... About 15% of the students studied humanities, 9% social sciences, 7% law, 15% economics, 9% medicine, 22% engineering, 18% natural sciences, and 4% other subjects. The term “economics” is an abbreviation of “economic sciences” which includes both economics and business students who, in Germany, study the same courses in their basic studies. 

Fischer et al. first look at the descriptive evidence of the relationship between field of study and political attitudes. Here they find that:

The coefficients of the LAW, ECONOMICS and MEDICINE variables have a positive sign and are statistically significant at the 1% level... This means that the incoming law, economics, and medical students are more prone to espouse liberal-democratic policy positions than the engineering students (reference category). The coefficients of the HUMANITIES and SOCIAL SCIENCES variables have a negative sign and are... statistically significant at the 1% level (Social sciences) and at the 10% level (Humanities). Not surprisingly, incoming humanities and social science students are less prone to espouse liberal-democratic policy positions than engineering students.

Interestingly, there are no differences between male and female students in terms of the relationship between field of study and political attitudes. Then, extending their analysis to look at the particular attitude towards free markets, Fischer et al. find that:

First, incoming economics students are not only more likely to support liberal-democratic policies than the reference engineering students; they also are more likely to favor free markets. Second, incoming law and medicine students support the liberal-democratic position mainly because of the socio-political dimension of classical liberalism. With respect to their evaluation of economic liberalism, these students are not different from engineering students. Third, students who begin to study humanities, social sciences, natural science, or one of the “other” fields are significantly more adamant in rejecting economic liberalism than incoming engineering students.

Turning to specific political attitudes, Fischer et al. find that:

...students who just begin to study law, economics, and medicine are significantly more in favor of Christian-conservative policies than our benchmark group, the engineering students... As compared to the engineering students, the incoming humanities, social science, natural science, and “other” students are all much less enthusiastic about Christian-conservative policies... The green political attitudes are an almost perfect mirror image of the liberal-democratic and Christian-conservative positions. But only the social science and economics students have clear left-right preferences. The other students do not measure the two left ideologies with the same yardstick.

Finally, Fischer et al. look at whether the differences are the result of selection (students with particular attitudes are more likely to choose particular fields of study) or socialisation (the field of study affects political attitudes). To do this, they look at how political attitudes change during students' studies. Here, things get really interesting:

In contrast to the political persistence of the medical and law students, and the rather slight polarization of the engineering, humanities, natural science, and social science students, the economics students' political attitudes do appear to change more dramatically during the course of their studies: the further economics students progress in their studies, the more they appear to favor liberal-democratic and Christian-conservative policies. Economics students also become more favorably disposed towards green policies which are advocated in Germany by the left-wing party Alliance '90/The Greens. By contrast, economics students appear to turn away from social-democratic policy positions as they become more proficient in their chosen field of study.

That seems like a somewhat odd mix, with economic students increasingly favouring market-liberal and Christian-conservative positions and more favourable views on Green policies, and away from social-democratic positions. In the German context, that means viewing the FDP or CDU and Greens parties more favourably, and the SPD less favourably. In terms of the Political Compass, that suggests a shift towards the economic right, and down towards social libertarianism. All the more surprising given the strong positive correlation between the two axes on the Political Compass I noted earlier. What others might find surprising is that:

... the economics curriculum does not make economics students adopt more right-wing policy views...

Given the changes in political attitudes that economics students experience, Fischer et al. note the importance of socialisation, as well as selection. Fischer et al. conclude that:

Economics students are thus quite special in terms of their political attitudes because of self-selection and socialization effects. Our preferred interpretation of the socialization result is not that economics students are brainwashed or indoctrinated by their instructors. The available empirical evidence rather suggests that economics students acquire analytical tools that make them see the world in a different light. We thus agree with George Stigler who firmly believed that the cause of the economists' political “conservatism” derives from their training: “It simply becomes impossible for the trained economist to believe certain absurd arguments which are, however, often used in the political discourse with resounding success”...

I would buy into that interpretation.

Read more:

• Does education in economics make politicians corrupt?
• The impact of business and economics education on moral competence
• This study tells us nothing about whether studying economics make students less ethical
• Does studying economics make you selfish?
• Do economists act like the self-interested decision-makers from our models, and if so, why?

Book review: Microeconomics Made Simple

How simple can microeconomics be? Can a person learn the basics by reading a book of only 100 pages? If we look at Microeconomics Made Simple, by Austin Frakt and Mike Piper, the answer is clearly yes.

Frakt and Piper set themselves an ambitious goal. However, they are careful not to over-pitch the book, at least to students:

For any students using this book in an academic setting: If your professor expects you to read a several-hundred-page textbook, please do not think that you can read this book instead and learn all of the same information. This book may serve as an introduction - a way to get a grip on the basics so that the textbook is easier to understand - but it's not meant to be a replacement for a comprehensive text.

Although the book was published back in 2014 (which is when I bought it, and it has sat unread on my bookshelf ever since), it has aged well, helped by the fact that key economic concepts don't change much over time. Across 11 chapters, Frakt and Piper cover the basics of utility and opportunity cost, production possibilities and the gains from trade, demand and supply and government intervention in markets, costs of production, and market structures (perfect competition, monopoly, oligopoly, and monopolistic competition). And they do a great job. The book is clearly written and, although it doesn't go into the depth that a textbook would, or provide the same range of examples as a textbook, it likely achieves its goal of providing a basic level of understanding of microeconomics.

I see a different use of the text than Frakt and Piper though. Rather than encouraging students to use this as an introduction, which they would build on in by later reading a more comprehensive textbook, I think this book would serve as an excellent refresher for those who studied microeconomics some years ago and want to be reminded of the key points. If that describes you, then I highly recommend this book.

This week in research #93

Here's what caught my eye in research over the past week:

• Abbate and Jiménez (with ungated earlier version here) find that the 2008 increase in the minimum wage in Argentina triggered a 4.8 percentage point (19%) decrease in job separations (the case against the disemployment effects of the minimum wage is shaky, but it's not dead)
• Birch and Preston (open access) find that there has been a significant decline in the wage premia (returns to education) within all education groups in Australia between 2001-2011 and 2012-2023, except for PhD and Graduate Certificates
• Baylis et al. (open access) look at decadal migration data at the county level in the US from the 1950s to the 2000s, and find that warm temperatures induce net out-migration, while cooler temperatures do not
• Chiovelli, Michalopoulos, and Papaioannou (open access) find that demining in Mozambique increases economic activity at the local level, and that this arises from increases in market access rather than increases in productivity

More on taxing the super-rich, and exit taxes

In a weird coincidence following yesterday's post about the challenges of taxing the super-rich, the Financial Times published an article on the same topic overnight, making some of the same points:

Income taxes and social security contributions, along with sales taxes, tend to be the main revenue-raisers in developed countries. But they do not address the capital wealth of the super-rich, which is often concentrated in real estate, investments or equity in businesses.

Yet imposing higher capital taxes on a relatively small number of very wealthy individuals often prompts changes in their behaviour that limit or even reduce the amounts raised. Raising taxes on the moderately wealthy, a much larger and less mobile cohort, usually has consequences at the ballot box.

The history of wealth taxes provides a prime example. In the mid-1980s, about half of OECD countries imposed an annual net wealth tax on their richest inhabitants. Today, in Europe only Spain, Norway and Switzerland retain taxes on individuals’ overall net wealth — and they raise relatively small amounts.

“Given the rich are extremely mobile and less and less attached to the country that made their wealth, they can shift and they do,” says Pascal Saint-Amans, a former head of tax at the OECD. “I suspect if you were to ask most billionaires, ‘Where is your loyalty, with your country or with your money?’, most would say, ‘My loyalty is with my money.’”

Interestingly though, the article also presents a potential solution:

One option to address the issue of the rich simply moving their assets elsewhere is the exit tax. Australia, Canada, France, Germany and Japan are among the 14 OECD countries that tax unrealised capital gains for those who change their tax residence, while the US taxes individuals who relinquish their citizenship.

“Tax flight happens less than most people think, but it does happen,” says Arun Advani, director of CenTax, a UK based think-tank, and professor at the University of Warwick. But, he adds: “It’s a policy choice to let them emigrate tax free.”

An OECD working paper on capital taxation this year agreed that exit taxes could curb revenue leakage and discourage tax-induced migration, though it added that these objectives needed to be balanced against other policy aims “such as attracting and retaining talent and entrepreneurs”.

That OECD working paper is available here, and is well worth a read. Exit taxes do sound like a potential solution to the negative incentive effects of wealth taxes or higher taxes on capital income. However, the working paper does not offer a strong endorsement of the effect of exit taxes on curbing out-migration of wealthy taxpayers. Instead it notes that:

...no empirical research on the impact of exit taxes on inward migration and entrepreneurship is yet available.

That definitely suggests a relevant research gap for someone to fill. Since we already have evidence (from the paper I discussed yesterday) of the large incentives for out-migration created by wealth taxes, it would be good to be able to quantify how much that incentive could be offset using an appropriately designed exit tax.

Read more:

• Newsflash! The super-rich are mobile, and higher taxes incentivise them to move away

Newsflash! The super-rich are mobile, and higher taxes incentivise them to move away

The super-rich are super-mobile. So, if a country decides to increase taxes on the super-rich (for example, with a wealth tax), some (but not all) of the super-rich will simply move elsewhere. This should not be a surprise to anyone. And yet, simplistic proposals to tax the super-rich are a favourite policy for some political and advocacy groups.

So, in case there was any doubt about how the super-rich respond to tax incentives, this recent article by Enea Baselgia and Isabel Z. Martínez, published in the Economic Journal (open access), provides some useful empirical evidence. They look at the case of Switzerland which, in addition to being an attractive place for the super-rich to locate themselves, also has taxes at the local (canton) level. Baselgia and Martínez make use of a special tax regime for wealthy foreigners, which was removed by five cantons between 2010 and 2014. As they describe:

One important reason why Switzerland is so attractive for wealthy individuals from around the globe is a special tax privilege the small country offers to wealthy foreigners: the so-called expenditure-based taxation. Those eligible for this special tax regime pay taxes on their (and their spouse’s and dependants’) global living expenses, rather than on their true income and wealth. Living expenses are defined broadly and include all expenditures for food, clothing and housing, taxes and social security contributions (around 25,000 CHF per adult and year), alimony payments, remunerations paid to household employees (in cash and in kind), expenses for education and leisure (sports, travel, cultural events, hobbies), for health and wellness cures, and costs of keeping pets (riding horses, etc.), as well as maintenance and operating costs of cars, motorboats, yachts, aeroplanes, etc.

Then:

Expenditure-based taxation has become the subject of heavy criticism over the past decade, both from outside and within the country. In light of these discussions, several cantons proposed to abolish this practice, usually holding a popular vote... Zürich (2010), Schaffhausen (2012),Appenzell Ausserrhoden (2012), Basel Stadt (2014) and Basel-Landschaft (2014) adopted corresponding proposals and removed the option of expenditure-based taxation. Seven other cantons held a popular vote between 2011 and 2014 that did not find a majority...

Baselgia and Martínez use data from a rich list compiled by BILANZ (the equivalent of the Forbes rich list internationally, or the NBR rich list in New Zealand). They apply a difference-in-differences approach, looking at the difference in the number foreign-born super-rich between the time before and the time after the special tax privilege was withdrawn, between cantons that withdraw the privilege and those that did not. They also apply a different approach, based on a location choice model. Both models result in similar estimates, such that:

...removing this preferential tax treatment reduces the stock of super-rich foreigners by approximately 43% five years after the abolition.

There were no corresponding results for Swiss-born super-rich individuals, who were not affected by the tax change (because they weren't eligible for the special tax privilege). That is an important aspect of the results, because if the Swiss-born also responded to the tax change, it should make us wonder if something else changed at the same time.

Baselgia and Martínez then conduct some back-of-the-envelope calculations, finding that:

...the elasticity of the stock of super-rich taxpayers in a canton with respect to the total net-of-tax rate on wealth lies in the range of 28.4–32.2.With respect to a revenue-equivalent tax on capital income (rather than on wealth), our estimates would imply an elasticity of the stock of super-rich taxpayers of 1.4–1.5.

What that first elasticity means is that a one percent increase in wealth tax rate decreases the number of super-rich taxpayers in a canton by between 28.4 and 32.2 percent. So, as we might expect, tax increases are a big disincentive to the super-rich. And the way that the results are expressed probably understates the magnitude of the effect. Notice that the elasticity is expressed as a one percent increase in wealth tax, not a one percentage point increase. That difference is important. Increasing a wealth tax from two percent to three percent is a one percentage point increase in the tax rate, but it is a 50 percent increase in the tax rate. The elasticity tells us the impact of the latter, which is massive. Now, of course the effect is not likely to be linear, but nevertheless that elasticity is enormous.

The implied elasticity on capital income is also large - a one percent (not one percentage point) increase in tax on capital income decreases the number of super-rich taxpayers in a canton by between 1.4 and 1.5 percent. The effect is not as large, but taxes on capital income are much less consequential for super-rich taxpayers than are taxes on wealth.

Incentives matter, and proponents of wealth taxes and other taxes on the super-rich must not be allowed to ignore the incentive effects. Is it better to tax a larger number of super-wealthy foreigners a little bit, or to tax a much smaller number of super-rich foreigners (or no super-rich foreigners at all) by much more? This research doesn't tackle that question, but that is what needs to be considered.

Principal-agent and adverse selection problems among Senegalese taxi drivers

A principal-agent problem (or agency problem) may arise in interaction that involve a person or group (the agent) being given the power to decide how to use resources that ‘belong’ to someone else (the principal). The problem that may arise is when the principal wants the agent to use the resources in a particular way that is contrary to the interests of the agent (who has their own goals and motivations), but the agent uses the resources in some other way that is detrimental to the principal. This principal-agent problem is an example of moral hazard. The market failure here is that the principal becomes unwilling to engage an agent, if they can't be sure that the agent will act in the principal's best interests.

An interesting example is described in this post from the Development Impact blog late last year, which was based on the job market paper of Deivy Houeix (MIT). They describe the problem as follows:

The Senegalese taxi industry exemplifies common principal-agent challenges faced by small firms in lower-income countries. The typical arrangement involves a car owner (employer) and a single driver (employee) linked by a relational contract. The driver keeps any revenue exceeding a rental fee paid weekly and sometimes receives an upfront payment from the owner. Due to limited liability, drivers can default on the rent by claiming low earnings for the week. Importantly, the owner has no way to observe whether this is due to bad luck, lack of effort from the driver, or whether the driver misreports revenue. This creates scope for moral hazard in both driver's effort and reported output and allows drivers to capture informational rents, contributing to inefficiencies common in informal arrangements. Default may lead to the (costly) termination of the relationship to mitigate moral hazard.

Notice that this has all the features described above. The car owners are the principals. The drivers are the agents. The resources belonging to the principal, that the agent has the power to decide how to use, are the taxis. The car owner wants the taxi drivers to pay the full rent for the car. However, the drivers have an incentive to claim low earnings and default on paying the rent. The market could fail, because the car owners choose to terminate the agreement with the drivers.

Houeix's post was about how to solve the principal-agent problem, and describes the results of two field experiments that were conducted to test whether adopting digital payment technology could reduce the principal-agent problem. Since closer monitoring of the agent by the principal is one solution to principal-agent problems, using digital payment technology to facilitate this monitoring seems like it could be beneficial.

I recommend reading the entire post (and following up with the job market paper itself, for further detail), but in short the results show limited success. On the one hand, the digital payment technology:

...significantly cuts drivers' cash-related costs (e.g., small-change shortages) by half and serves as effective monitoring tools even with partial digitalization of transactions (about 13% of revenue).

The digital payment technology made it easier for drivers to accept payment, and easier for car owners to monitor what drivers were doing. However, drivers had the option not to adopt the digital payment technology, by simply not providing the car owner's details:

I find that observability is an important barrier to technology adoption, especially for the worst-performing and poorest workers. Initially, 50% of drivers did not want to adopt the technology, citing various privacy concerns for not sharing owners' information.

This should be no surprise. The lowest-productivity workers are likely to be the ones hurt the most by closer monitoring, so of course they wouldn't want to adopt the digital payment technology.

However, there may be an upside. Since car owners don't know who the high-productivity drivers are (productivity is private information), there is also an adverse selection problem here. That problem is that, because the car owners can't tell high-productivity and low-productivity drivers apart, they must assume that all drivers are low-productivity drivers, and will treat them accordingly. This is a pooling equilibrium (because all drivers are pooled together and treated the same). The high-productivity drivers don't want to be treated as if they are low-productivity, so they will be more likely to leave the market. If this continues, eventually only low-productivity drivers are left. The market fails.

This adverse selection problem could be solved by the car owners asking drivers to adopt the digital payment technology (this is a type of screening, since it involves the uninformed party (the car owners) attempting to credibly reveal the private information). The answer to the question about whether they will agree to install the digital payment technology reveals the productivity of the drivers. The drivers who don't agree are likely to be the low-productivity drivers, and so the car owners should choose not to rent their car to those drivers. Instead, the drivers who accept the digital payment technology will be the high-productivity drivers, and the car owners should be happy with them. That creates a separating equilibrium (since the low-productivity and high-productivity drivers can now be separated), and solves the adverse selection problem for the car owners.

Houeix didn't note the potential solution to the adverse selection problem in the post, nor in their job market paper. That was a missed opportunity, and may be the real benefit of the digital payment technology that they investigated.

Book review: Gender and the Dismal Science

The gender gap in economics is large and persistent (see the links at the end of this post for details). They have also been around since the founding of the discipline. In her 2022 book Gender and the Dismal Science, Ann Mari May documents the gender gap in economics between the founding of the American Economic Association (AEA) in 1885, and 1948. As May describes it, the book uses:

...novel data sets to offer new information on the proportion of women members in the AEA, their backgrounds, and their limited role in the association's work in its first sixty-three years of existence. At the same time, they provide information on the "old boy network" in publishing - in monographs and in scholarly journals such as the AER and the QJE.

May has exhaustively investigated the women in economics during those early decades of the AEA, and compares them with a sample of randomly selected male economists from the same era. This offers insights into the number of women (compared with men) earning doctorates in economics, joining the AEA, receiving faculty appointments, becoming officeholders in the AEA, as well as co-authorship (within and between genders).

Readers who are familiar with the gender gap in economics will not be surprised at all by the results. However, some aspects of the gap do make for uncomfortable reading. For example, May notes the perniciousness of marriage bar policies and antinepotism rules, which had the effect of limiting the hiring of married women and were used to fire women from existing positions if their marriage was discovered. Unsurprisingly, this meant that successful female economists typically were unmarried. In contrast, nearly all male economists, including successful male economists, were married.

Another result that should be more widely known is that the first woman to serve as president of the AEA was Alice Rivlin, as late as 1986. And similarly, the first woman to serve as editor of the association's premier journal, the American Economic Review (AER), was even later, with Pinelopi Goldberg taking on the role in 2011.

The quantitative analysis in the book is quite descriptive and relatively shallow. Given the small sample size of women in the early years, it would be difficult for it to be otherwise. And to be honest, we don't even need the quantitative analysis to recognise that economics had a real problem in those early years, and that was the genesis of the problems that the discipline faces today. The more qualitative elements of the book, being the stories of the women who were trying to make their mark on economics in the early years of the AEA, stand out as important testimony of the problems. May has done a great job of surfacing and summarising this information, and for anyone interested in gender in economics, this book is an important read.

The moral hazard of bailing out scam victims

Earlier this year, bowing to pressure from media and advocacy groups, banks acted to help out victims of financial scams. As the New Zealand Herald reported in April:

Banks will be required to reimburse fraud victims up to $500,000 and introduce new rules to crack down on scammers in a suite of measures unveiled today.

The changes include new technology to identify risky or unusual transactions based on a customer’s banking history and the ability to freeze payments and suspect accounts.

The moves are in response to Government demands to improve customer protections or be regulated in the face of Kiwi victims losing hundreds of millions of dollars to scammers each year...

The New Zealand Banking Association says it is rolling out a package of new protections in line with international best practice, which will be in place by November.

However, by itself the reimbursement of fraud victims creates a problem that might actually result in there being more victims of fraud overall. That is because of what economists call moral hazard. Moral hazard arises when one of the parties to an agreement has an incentive, after the agreement is made, to act differently than they would have acted without the agreement. Importantly, the agreement doesn't have to be a formal contract. It can be an implicit understanding or expectation.

In this case, there is an implicit agreement between banks and their clients, that makes it clear that the bank will reimburse the client when the client is impacted by a financial scam. Without that implicit agreement, bank clients have a strong financial incentive to avoid being scammed. If they are scammed, they lose a lot of money. However, now that banks will bail them out, bank clients have less incentive to act carefully and avoid financial scams. Scammers may be more successful as a result, leading to more scam victims.

This is not to say that bank clients will be flagrantly imprudent with their money, only that at the margin, clients will act a little less cautiously. The moral hazard problem here is that the risky actions of the bank clients end up costing the banks money, in the case when the bank client is scammed and the bank needs to reimburse them. If banks couldn't do anything about the risk, they would be less inclined to take on depositors. If the risk turned out to be extreme, the market for bank deposits could fail entirely.

Fortunately, banks can respond to this increased moral hazard in various ways. The first way is through increased monitoring of their clients. Notice in the quote from the article above that banks will employ "new technology to identify risky or unusual transactions based on a customer’s banking history". By monitoring clients' transactions, banks can hopefully head off any scam activity. Banks could also use incentives to reward their clients for not being scammed. Perhaps they could pay slightly higher interest rates to those that pass occasional bank-delivered 'scam checks' (where the bank employs someone to test whether the client will fall for a scam). None of the banks are proposing this yet, but it is a solution that is open to them.

Overall, this change is likely to be positive for bank clients. However, with these new scam protections in place, people will be less careful. Banks will need to remain very vigilant, or the number of scam victims will increase.

This week in research #92

Here's what caught my eye in research over the past week:

• Kaicker finds that the pandemic induced lockdowns resulted in a sharp increase in the share of alcohol in total expenditure across rural and urban India, and for all income levels
• Yildirim and Bilman investigate whether hosting the return leg in two-legged soccer ties improve a team’s advancement chances, and find that the video assistant referee (VAR) offsets the second-leg home advantage solely in the presence of the away goals rule (AGR), while AGR creates that advantage solely in VAR’s absence
• Wienhold and Roberts find that the upgrading by producers who grow coffee for differentiated specialty markets is mainly (but variably) appropriated by downstream actors (so farmers don't really benefit much from growing specialty coffee)
• Anderson and Zaber (with ungated earlier version here) find that when students receive additional financial aid, on average they reduce labour earnings dollar for dollar (so financial aid would reduce student working, but not improve incomes)
• Liu, Geng, and Chen (with ungated earlier version here) find that two-year colleges in the US experience an approximately 10 percent decline in both enrolment and the number of degrees and certifications awarded within a decade following a hurricane, whereas four-year institutions exhibit no significant effects
• Gicheva et al. (open access) find that, among North Carolina tertiary students, female college students have been passing fewer classes and experiencing slower credit accumulation since 2021, compared with male students
• Brunello et al. (open access) find evidence of strong inter-generational persistence in choice of major in Italian universities, especially in medicine and health professions, followed by economics and law, and STEM
• Barnes, Dischman, and Mendez find that postseason officiating disproportionately favours the Mahomes-era Kansas City Chiefs, coinciding with the team’s emergence as a key driver of TV viewership/ratings and, thereby, revenue (was there more financial pressure from Philadelphia in the last Super Bowl?)

Sellers of natural diamonds are in big trouble

As I noted in yesterday's post, sellers of natural diamonds are in trouble. Lab-grown diamonds are undercutting their market. The problem here is that consumers can't easily tell lab-grown diamonds and natural diamonds apart. The two types of diamonds are perfect substitutes. And when faced with the option of buying one of two products that are perfect substitutes, consumers will generally choose the product that is lower-priced. In this case, that's the lab-grown diamonds.

To see why, consider the diagram of the consumer choice model below. The two goods are X (natural diamonds) and Y (lab-grown diamonds). The consumer's budget constraint is shown by the black line. The budget constraint is relatively steep, which means that the price of lab-grown diamonds is relatively lower than the price of natural diamonds. The consumer's indifference curves are shown by the two red lines, I0 and I1 (with I1 representing a higher level of utility, or satisfaction, for the consumer). The indifference curves are straight lines when the two goods are perfect substitutes (which is the case here). The consumer's best affordable choice (the consumer's optimum) is the bundle of goods E0, (it's on the highest indifference curve that they can reach, I1), where the consumer spends all of their income on lab-grown diamonds, and spends nothing on natural diamonds. [*] This makes sense, given that lab-grown diamonds and natural diamonds are exactly the same good in the mind of the consumer (they are perfect substitutes), and lab-grown diamonds are relatively less expensive than natural diamonds.

So, how can natural diamond sellers respond to this problem? One way is to lower their prices to match the lab-grown diamond price. That would cause the consumer's budget constraint to pivot outwards and become flatter (just like in this example), and then the highest indifference curve would exactly match the budget constraint. However, lowering prices could easily escalate into a price war, and is unlikely to end well for anyone.

A better option for the sellers of natural diamonds arises when they recognise that the real problem here is not the price, it is that the two goods are identical in the mind of the consumer. If the sellers of natural diamonds can somehow convince the consumer that the two goods are different rather than identical, then they may be able to keep some sales, even if the price of natural diamonds is higher than the price of lab-grown diamonds.

This situation is shown in the diagram below. When the goods are differentiated, the consumer's indifference curves are curves (not straight lines - straight line indifference curves only happen when the goods are perfect substitutes). The highest indifference curve that the consumer can get to is I1'. They will buy the bundle of goods E1, which contains Y1 lab-grown diamonds, and X1 natural diamonds. Even though natural diamonds are relatively more expensive, the consumer chooses to buy some of them.

So, how can the sellers of natural diamonds differentiate their diamonds from the lab-grown diamonds? That is the tricky thing, because there is an asymmetric information problem here. The sellers know whether diamonds are lab-grown or natural, but buyers don't know. The origin of a diamond is private information. Because buyers can't tell the two diamonds apart, they assume that all diamonds are lab-grown [**]. This creates a pooling equilibrium (because all diamonds are pooled together and treated the same). Buyers would only be willing to pay low prices for diamonds, because they assume that the diamonds are lab-grown. Natural diamond sellers don't want to sell their diamonds for the lower lab-grown diamond price, so they drop out of the market. Only lab-grown diamonds would be left in the market. The market for natural diamonds would fail. Economists call this an adverse selection problem. And that is what seems to be happening, since the Financial Times article I referred to yesterday notes that:

By the end of 2024, De Beers had amassed an inventory of unsold diamonds worth $2bn, the largest stockpile since the 2008 financial crisis.

The diamonds are unsold in part because they cannot be distinguished from the lower-priced lab-grown diamonds. How can the natural diamond sellers solve this adverse selection problem? When the informed party (the party that knows the private information) credibly reveals that information to the uninformed party, we call that signalling. To be effective, a signal needs to meet two conditions. First, it must be costly. And second, it must be costly in such a way that the sellers of lab-grown diamonds wouldn't want to attempt the signal.

Unfortunately, it is difficult to identify a signal that meets those two conditions for the sellers of natural diamonds. If they try some recording the chemical signature of their diamonds, those structures can probably be easily copied by makers of lab-grown diamonds. Similarly, microscopically etching a serial number onto each natural diamond is something that makers of lab-grown diamonds can do as well. That rules out branding diamonds. Advertising is not likely to be very effective either, because while advertising natural diamonds and making consumers want them more seems like a good strategy, it won't turn into extra sales if consumers can't tell the lab-grown diamonds and natural diamonds apart. Sellers can often use warranties to signal quality. However, when a good has a warranty, it's quality is eventually revealed to the consumer (because, if the good is low quality, they end up having to claim on the warranty). That isn't the case for diamonds.

So, the natural diamond sellers are not just in trouble. They are in big trouble, unless they can find some way of signalling that their diamonds are natural diamonds (and at the same time hoping that buyers continue to be willing to pay a premium for natural diamonds).

*****

[*] Yes, this model is assuming that the consumer spends all of their budget on only two goods, natural diamonds and lab-grown diamonds. If it makes you feel better, think of it as the consumer spending all of their diamond budget on those two goods.

[**] I'm treating the lab-grown diamonds as if they are lower quality than natural diamonds. That is what the sellers of natural diamonds would argue, anyway, and given that there is a slight price premium for natural diamonds, the buyers seem to think that way too.

Read more:

• Sellers of natural diamonds are in trouble
• Market power and disruption in the diamond industry

Sellers of natural diamonds are in trouble

The Financial Times reported earlier this year (paywalled):

Over 70 per cent of the world’s lab-grown diamonds for jewellery — many destined for the ring fingers of newly engaged couples — originate in a Chinese factory, with Henan at the centre of the synthetic trade...

For the natural diamond industry, Feng’s factories and others like them have been devastating. The explosion of lab-grown diamonds on the international jewellery market has coincided with a slump in demand, sending the price of smaller natural diamonds to their lowest levels in a decade.

Marty Hurwitz, head of the Grown Diamond Trade Organisation, says lab-grown diamonds have “been a massive disruption. People in the industry at first didn’t believe it and, second, couldn’t accept it.

“This has been the first competitive product that mined diamonds have ever faced.”

To see the impact of the growth of the lab-grown diamond market on natural diamonds, let's first consider the market for lab-grown diamonds, shown in the diagram below. The 'explosion of lab-grown diamonds' is demonstrated by the increase in the supply, from SA to SB. The equilibrium price of lab-grown diamonds has decreased from PA to PB, and the equilibrium quantity of lab-grown diamonds traded has increased from QA to QB.

Next, consider the effect on the market for natural diamonds, shown below. Lab-grown diamonds and natural diamonds are close substitutes (so close that no regular consumer can tell them apart!), and lab-grown diamonds are now cheaper. So, as the quote from the Financial Times article notes, the demand for natural diamonds has decreased, from D0 to D1. The equilibrium price of natural diamonds has decreased from P0 to P1, and the equilibrium quantity of lab-grown diamonds traded has decreased from Q0 to Q1.

This is a real problem for the producers of natural diamonds. The product that they are selling is decreasing in price, and they are selling a smaller quantity. That must make those sellers worse off. And, to make matters worse, they should have seen this coming. I wrote this post back in 2019, wondering why De Beers wasn't investing in its own production facilities for lab-grown diamonds. It turns out that they were, as the Financial Times article notes:

In 2018, De Beers established its own lab-grown diamond company, Lightbox, which started churning out cheap synthetic stones. Part of the thinking was to create a bifurcated market that would ensure the luxury appeal of expensive natural stones was maintained while undercutting synthetic rivals. 

Instead, it sparked a price war that also dragged down the price of natural diamonds, which were simultaneously hit by a slump in demand due to lower marriage rates during the pandemic. By the end of 2024, De Beers had amassed an inventory of unsold diamonds worth $2bn, the largest stockpile since the 2008 financial crisis.

Again, that outcome could and should have been anticipated. De Beers (and other natural diamond producers) need to find some way of differentiating natural diamonds from lab-grown diamonds (I will return to this point in a future post). Otherwise, the natural diamond sellers are in real trouble.

Read more:

• Market power and disruption in the diamond industry

The AI arms race in hiring may cause the market for high-quality job applicants to fail

Two years ago, which is a lifetime or more ago in generative AI years, I wrote a post about the impact of large language models on job interviews. I concluded that post with:

So, perhaps there are offsetting benefits of AI on the job interview process, if the AI is being used by the employer. If AI leads to job interviews that are more effectively able to screen for high-quality job candidates, by reducing bias in the interview process, then that can be a good thing. The AI might also be better able to ask the searching questions that distinguish high-quality and low-quality job candidates.

Of course, that assumes that the AI is interviewing a human job candidate. What happens when a job candidate, being interviewed by an AI job interviewer, is being given real-time prompts on how to answer by an AI chatbot? Or, in a Zoom job interview, the job candidate simply replaces themselves with an avatar or a 'deep fake' video of themselves generated in real time, and using an AI-scripted voiceover. If it hasn't happened already, it is going to be happening soon. Will that be the death of the job interview as a screening tool? Time will tell.

Time has told, indeed. The Financial Times reported back in May (paywalled):

You can almost hear the howls of frustration from HR departments. Jobseekers have discovered artificial intelligence and they’re not afraid to use it. Employers have become snowed under by people using the new tools to churn out impersonal applications. Some applicants are using AI to bluff their way through online assessments, too...

So it should be no surprise that jobseekers have turned to new generative AI tools such as ChatGPT to speed up or “game” a process that already felt dehumanised. Videos have even appeared on TikTok in which people demonstrate how to use ChatGPT to provide answers to questions in asynchronous video interviews, which the applicant then simply reads out.

The problem here is that the online tools that employers have been using in recent years as a tool for screening job applicants, have suddenly become less effective due to generative AI. The issue is one of adverse selection, which my ECONS102 class covered last week.

Job applicants know their quality as a worker (how hard-working, conscientious, smart, etc. they are), but employers do not. Quality is private information. Faced with a lack of information about the quality of each job applicant, employers would assume that every job applicant is low quality. This is referred to as a pooling equilibrium (because all job applicants are pooled together and treated the same). Employers would offer low wages to each new hire, because they assume that the new hire is low quality. High quality job applicants don't want to be treated as if they are low quality and receive low wages, so they would reject any low wage offer, and stop applying for jobs. Only low-quality job applicants would be left in the market. The market would fail for high-quality job applicants.

Of course, employers have found ways to deal with this situation. When the uninformed party (the party that doesn't know the private information, which in this case is the employers) try to reveal the private information (about the quality of job applicants), economists call that screening. Screening in a job application setting will be effective if it accurately reveals the quality of the job applicant. In-person job interviews used to be used as a screening tool. More recently, as noted in the Financial Times article, employers have moved to more technological solutions, such as 'asynchronous video interviews'.

With the rise of generative AI, those new screening tools have suddenly become ineffective, because they can no longer distinguish the high-quality job applicants from low-quality job applicants who are using generative AI. Generative AI has broken the current approach to hiring. As the Financial Times article notes:

Some online assessments are, for now, less vulnerable to AI use, such as those that involve playing short games. But I wouldn’t be surprised to see the return of mass in-person test centres for technical skills assessments.

I wouldn't be surprised to see the return of in-person job interviews as well. Of course, smart glasses combined with real-time generative AI assistance may render even in-person methods of assessment vulnerable, if not now then certainly in the near future. The unfortunate outcome of all of this is that the market for high-quality job applicants will soon be failing, and no one (not employers, not high-quality job applicants, and not even low-quality job applicants) will benefit from that.

Read more:

• The impact of large language models on job interviews, from both sides

This week in research #91

Here's what caught my eye in research over the past week:

• Ajzenman, Ferman, and Sant’Anna (with ungated earlier version here) investigate discrimination in #EconTwitter, finding that follow-back rates by users in #EconTwitter were 12 percent higher when followed by White students compared to Black students, 21 percent higher for students from top-ranked universities compared to those from lower-ranked institutions, and 25 percent higher for female compared to male students
• Hwang, Jakob, and Squires (with ungated earlier version here) find, using US genealogical data on couples married over a century ago, that children whose parents were first cousins experience a more than a two-year reduction in age-five life expectancy
• Daher et al. (with ungated earlier version here) find that Saudi women who receive driver training are 41 percent more likely to be employed and yet 19 percent less likely to be able to make purchases without family permission
• Loewenstein and Wojtowicz (with ungated earlier version here) summarise the research on the economics of attention
• Strehl-Pessina, Bergolo, and Leites (with ungated earlier version here) find a marked decline in support for redistribution among the top 1% of the income distribution in Uruguay, and that support for redistribution is not solely explained by current income or demographics
• Fu, Zhang, and Zhong propose two novel scoring rules for multiple-choice questions based on the test-takers’ propensity to hedge across possible answers
• Cai, Cheng, and Jiao find that students in classrooms with greater economic inequality tend to have lower test scores and cognitive outcomes, and that the effects are especially significant among students from lower socioeconomic backgrounds, male students, and those preparing for high-stakes examinations
• Livermore and Major (with ungated earlier version here) examine the low levels of participation and diversity in economics at Australia's high schools, and find that while students typically have positive perceptions of economics as a field, the perceptions of economics as a subject of study tend to be negative (could be relevant to New Zealand as well)
• Sentana conducts a (quasi) field experiment in the soccer field with junior players, and finds that results are consistent with mixed strategy, even though the least experienced goalkeepers tend to replicate each other’s actions
• Masso et al. (with ungated earlier version here) find that male academics obtain higher returns than women academics from the same outside option during salary negotiations, using data from the University of Tartu in Estonia
• Adler et al. (open access) find that students who are the first in their family to attend university sort less into study fields based on their earnings expectations, which leads to sizable gaps in expected earnings

Is free employment mediation really free if you have to wait for it?

The New Zealand Herald reported last month:

An employment lawyer is advising businesses to bypass the government’s free employment mediation service.

There was currently a seven-week waiting list to access the Ministry of Innovation, Business and Employment’s mediation service, which was supposed to be a way to avoid drawn-out disputes between employers and their employees.

“Don’t wait,” Rotorua employment lawyer Michelle Urquhart said, adding the cost of accessing private services was well worth it given the risks associated with leaving a dispute to fester.

MBIE advised availability was limited due to high demand and apologised for the inconvenience, though wait time was an improvement from the peak 11-week wait in February.

When the price of a good or service is zero (free), there is often a shortage (excess demand) for the good or service - there are more people wanting to access the good or service than there is available. This is illustrated in the diagram below. If the market for public mediation services operated at equilibrium, the market price would be P0, and the quantity of mediation services would be Q0. That quantity (Q0) is both the quantity of mediation demanded, and the quantity of mediation supplied (the number of mediation appointments available for businesses). We can say that the market clears, because quantity demanded is equal to quantity supplied (the market is in equilibrium).

However, the price is below equilibrium, at zero (free) [*]. At that zero price, the quantity of mediation demanded is QD, while the quantity of mediation supplied is QS. Since QD is greater than QS, there is excess demand (a shortage). That is what we are seeing, with long waits for mediation services.

That excess demand needs to be managed. Ordinarily, we would expect the price to rise when there is excess demand, but since the government has set the price at zero, that cannot happen. So, the alternative is that the excess demand is managed with a waiting list. When a business wants to access the free employment mediation service, it is added to the waiting list, and then needs to wait until the service is available.

Ironically, the operation of the waiting list means that the 'free' mediation service is no longer 'free'. It just has no monetary cost. There is a cost associated with waiting for the mediation, because in the meantime whatever employment dispute necessitated mediation is not being resolved (and festering, as the employment lawyer in the quote above notes). The costs of that unresolved situation might be much less than the cost of paying for private mediation services. The free public services are not really free at all. It should be little wonder that some businesses are opting for private mediation services instead.