Egg prices will rise in New Zealand, even without a major avian flu outbreak

Last year, I posted about avian flu in the US and the impact on egg prices, noting that prices will rise. Thankfully there hasn't been a major outbreak of avian flu in New Zealand as yet, although it seems likely there will be soon. Domestic birds, such as chickens, are at risk, and as I noted in that earlier post, that affects the supply of eggs. And New Zealand egg suppliers are acting now, as the New Zealand Herald reported earlier this week:

It comes as New Zealand’s largest egg supplier Mainland Poultry, accounting for nearly 40% of the country’s eggs, is putting hundreds of thousands of free-range chickens into lockdown after the deadly bird flu virus was detected in the country last week.

Putting free-range chickens into lockdown will raise the costs of production for free-range eggs. The effect on the market for free-range eggs is shown in the diagram below. Before the chickens were locked down, the free-range egg market was in equilibrium, where demand D0 meets supply S0, with a price of P0 and a quantity of free-range eggs traded of Q0. The lockdown increases the costs of producing free-range eggs, which decreases supply to S1. This increases the equilibrium price of free-range eggs to P1, and reduces the quantity of free-range eggs traded to Q1.

Free-range eggs and colony eggs are substitutes. Once free-range eggs become relatively more expensive, some consumers will switch to colony eggs. The effect on the colony eggs market is shown in the diagram below. Before the change in the price of free-range eggs, the market for colony eggs was in equilibrium, where demand DA meets supply SA. The equilibrium price was PA, and the quantity of colony eggs traded was QA. Since some consumers switch to the relatively cheaper colony eggs, that increases the demand for colony eggs from DA to DB, increasing the equilibrium price of colony eggs from PA to PB, and increasing the quantity of colony eggs traded from QA to QB.

Overall, eggs are going to cost more, regardless of whether they are free-range eggs or colony eggs. And even without a major outbreak of avian flu. If avian flu does take hold in New Zealand, the price of eggs of both varieties will go up even further.

This week in research #136

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

• Baker et al. (with ungated earlier version here) find that following legalisation, sports betting does not displace other gambling or consumption but significantly reduces savings, as risky bets crowd out positive expected value investments, and that these effects are concentrated among frequent bettors and low-savings households
• Alfani (open access) summarises the state of research on social mobility and income inequality in preindustrial societies, and identifies several promising avenues for new research
• Bostwick and Nguyen (with ungated earlier version here) find that when a student enrols in a triweekly (rather than biweekly) university class, they earn lower grades and are less likely to take a subsequent course in that same field
• Kaffine and Rao (with ungated earlier version here) find that the observed distribution of satellite orbits is well explained by profit-maximising satellite operator behaviour, and that collision risk plays a key deterrent role
• Yang et al. (open access) find that having daughters slightly increases the preference for gender equality and homosexual rights, using data across 39 countries from the European Social Survey
• Di Tella, Gàlvez, and Schargrodsky (with ungated earlier version here) find, using a laboratory experiment, that two strategies often proposed to reduce political polarisation on social media, removing access to the platform and exposing users to counter-attitudinal content, do not reduce polarisation and may instead produce mild increases
• Perroni et al. (open access) find, using a sample of 40,000 job vacancies in Vietnam, that physically attractive women are offered higher salaries, whereas physically attractive men are not
• Sentana and Gómez-Bengoechea comparing the behaviour of a groups of professional soccer players and inexperienced college students when they play the same two-person zero-sum game in both a soccer stadium (the field) and a video game (the lab), and find that while both groups behave consistently with the implications of mixed strategy equilibrium in the field, their behaviour rejects the theory in the lab

Are men's and women's soccer complements or substitutes?

Both my ECONS101 and ECONS102 classes touched on the subject of complementary and substitute goods this week (in different model contexts). Two goods are complements if consumers tend to consume them together. In that case, a decrease in the price of one good would increase the quantity that the consumer buys of both goods. Two goods are substitutes if consumers tend to consume one or the other. In that case, a decrease in the price of one good would increase the quantity that the consumer buys of the now-cheaper good, but decrease the quantity that the consumer buys of the other good (which is now relatively more expensive).

Often, it is easy to tell if goods are complements or substitutes. However, sometimes it is not straightforward. Consider the example of men's and women's soccer matches. Are they complements, or substitutes? If, when faced with the choice of whether to attend a men's or a women's soccer match, or both, fans tend to choose one or the other (and not both), then the matches are substitutes. On the other hand, if fans tend to go to both, then the matches are complements. Another way of thinking about this is that, when the price of one of the matches goes up, what happens to attendance at the other. So, if the ticket price for a men's soccer match increases and attendance at women's matches goes up, then they are substitutes, whereas if attendance at women's matches goes down, then they are complements.

Ultimately, whether men's and women's soccer are substitutes or complements is an empirical question. Fortunately, this 2025 article by Galila Nasser and Christian Deutscher (both Bielefeld University), published in the Journal of Sports Economics (open access), provides us with an answer. Or rather, they provide us with an answer in one particular context, which is German soccer.

Specifically, Nasser and Deutscher use data from the 2009/10 to 2018/19 seasons of the Frauen-Bundesliga, and look at the impact on match attendance when a Frauen-Bundesliga match is played on the same day as a men's Bundesliga match. They also consider whether the effect is larger when the overlapping men’s and women’s matches involve teams belonging to the same club. Their dataset contains 1,256 Frauen-Bundesliga matches, including 851 played on the same day as a men's Bundesliga match and 118 played on the same day as a match involving the men's team of the same club.

Controlling for the day of the week, week of the season, the weather, whether a UEFA Champions League match was also being played that day, and a variety of variables capturing the popularity of the match, Nasser and Deutscher find that there is:

...an approximately 15 percentage points decrease in attendance when women’s games coincide with men’s games on the same day.

A minor quibble with the paper is that when they say a 15 percentage points decrease, they really mean a 15 percent decrease. And the effect for matches played by the same club on the same day is somewhat larger, with attendance lower by about 16 percent. So, these results are consistent with men's and women's top-league soccer matches in Germany being substitutes (fans tend to go to men's or women's games, and not both). However, we can't conclude this for certain as the results are based on observational data so they are correlations, not causal. Nevertheless, Nasser and Deutscher conclude that:

For matches on the weekend, it is essential for clubs that have both men’s and women’s soccer teams in the first Bundesliga to avoid scheduling their matches on the same day.

Given that the seasons overlap substantially, and clubs in both leagues understandably want weekend matches, another option might be to make joint attendance at both men's and women's matches more attractive. Clubs with both men’s and women’s teams could offer a combined ticket covering matches played on different days, or even arrange occasional double-headers. As I note in my ECONS101 class, this sort of bundling can be an effective pricing strategy when there is heterogeneous demand across multiple products. Provided the variation in fans' willingness to pay for the ticket to the combined event is lower than the variation in fans' willingness to pay for the tickets separately, then bundling has the potential to increase total revenue overall. And that higher total revenue can then be shared between the men's and women's teams. Whether that would work here is another empirical question. Perhaps Bundesliga clubs could indulge us by running the experiment?

Farmers can't avoid high synthetic nitrogen fertiliser prices by switching to organic fertiliser

The New Zealand Herald reported yesterday:

New Zealand farmers face hefty increases in the price of fertiliser this spring as a result of the escalating US-Iran conflict and the war in Ukraine.

The Middle East plays a big role in the global fertiliser market because of its supply of natural gas and mineral resources.

Russia is also a major supplier of fertiliser.

Renewed hostilities in the Persian Gulf – and the virtual closure of the Strait of Hormuz – have driven oil prices up to about US$90 ($154) a barrel for Brent crude, the international benchmark.

Synthetic nitrogen fertiliser is generally manufactured from ammonia created using the Haber-Bosch process. This requires hydrogen, which is often derived from natural gas (mainly methane). Since the Middle East is a major supplier of natural gas, a lot of nitrogen fertiliser is manufactured in the Middle East. The current conflict in the Middle East is constraining the transport of nitrogen fertiliser from the Persian Gulf, reducing the supply of nitrogen fertiliser.

The effect of this on the market for nitrogen fertiliser is shown in the diagram below. The market was initially in equilibrium, where demand D0 meets supply S0, with a price of P0 and a quantity of nitrogen fertiliser traded of Q0. The Middle East conflict reduces shipping of nitrogen fertiliser, which decreases supply to S1. This increases the equilibrium price of nitrogen fertiliser to P1, and reduces the quantity of nitrogen fertiliser traded to Q1.

Can farmers avoid the higher price of nitrogen fertiliser by switching to an alternative product, such as organic fertiliser (compost, or manure)? Not really. Consider what happens in the market for organic fertiliser, shown in the diagram below. Before the change in the price of nitrogen fertiliser, the market for organic fertiliser was in equilibrium, where demand DA meets supply SA. The equilibrium price was PA, and the quantity of organic fertiliser traded was QA. Since nitrogen fertiliser and organic fertiliser are substitutes, and nitrogen fertiliser is now relatively more expensive (as shown above), farmers switch to the relatively cheaper organic fertiliser. That increases the demand for organic fertiliser from DA to DB, increasing the equilibrium price of organic fertiliser from PA to PB, and increasing the quantity of organic fertiliser traded from QA to QB.

So, the effect overall is that the price of both nitrogen fertiliser and organic fertiliser increase. Farmers cannot easily avoid high fertiliser prices. We can expect that to flow through into higher prices for farm produce, as well as lower profits for farmers.

How career stereotypes shape students' choice of major

What jobs do accounting majors get? How about psychology majors? Or economics majors? If you answered, respectively, 'accountant', 'psychologist', and 'economist', you're probably far from alone. When most people think about particular fields of study, they have stereotypical jobs in mind, and they're far more likely to believe that majors get the stereotypical job than any other job. Even in the case of economics, where very few graduates will go into a job with the title 'economist' (many will go into a job with some sort of 'analyst' title, like a business analyst, market analyst, or financial analyst, etc.).

A new article by John Conlon (Ohio State University) and Dev Patel (Brown University), published in the Quarterly Journal of Economics (open access) demonstrates the extent of this stereotyping. They also show using a simple survey experiment that students' stereotypical views can be changed, affecting their choice of major.

The first part of the paper compares students’ beliefs about the careers associated with different majors with actual major-career combinations in the 2017-2019 American Community Survey. Conlon and Patel then use the CIRP Freshman Survey, covering more than nine million first-year students between 1976 and 2015, to compare students’ expected careers with the occupations subsequently observed among graduates from the same cohorts. In that, they find:

...large, systematic, and persistent differences between the careers that freshmen expect to attain and the actual occupations they go on to have... We see that twice as many students expect to become artists, counselors, and lawyers (about 5% each) than actually do (2%–3% each). Four times as many students expect to become writers and doctors (2.7% and 11.1%) than do (0.7% and 2.8%).

Interestingly, the occupations that students most overestimate themselves as having are those that are rare and representative of particular majors (like writers or artists), while those that are most underestimated are those that are common alternative occupations that many majors may later hold (like teaching or business). Conlon and Patel also show using implicit association tests that people:

...strongly and systematically associate majors with their representative careers: implicit associations are 0.30–0.36 standard deviations higher for representative major–career pairs than for nonrepresentative pairs...

This supports Conlon and Patel’s interpretation that stereotyping contributes to students’ exaggerated beliefs about the connection between majors and the corresponding representative careers. Conlon and Patel then show using a theoretical model that stereotyping can increase misallocation of labour. What that means is that, given the occupation in which a graduate eventually works, that person might have been better off studying a different field. They also present suggestive evidence that links greater stereotyping with job dissatisfaction and regrets about the chosen field of study.

The welfare loss, job dissatisfaction, and regret, then motivated a survey experiment where Conlon and Patel attempt to correct for the stereotyping. The survey experiment proceeded as follows, using students from Ohio State University:

The survey began by asking students the percent chance that they would graduate with the two majors they selected as being most likely to pursue (their “top-ranked” and “second-ranked” majors). It then asked their self and population beliefs about the likelihood of each career group conditional on these two majors. Students were randomly sorted into a control group and a treatment group. Those in the control arm answered questions about their classes so far that semester and how they had (or had not) contributed to their major and career plans...

In the treatment arm, information modules provided students with the actual distribution of careers conditional on each of their top two majors according to data from the [American Community Survey]. For each major, it told them several headline numbers about the frequency of the careers they had listed as their most likely jobs if they graduated with that major...

Conlon and Patel then looked at whether assigning a student to the treatment group, where they received accurate information about the chance their top-ranked majors would lead to particular careers, affected students' beliefs about their own chances of working in each major's 'representative career'. They found that the information partially (but not fully) corrects students' misbeliefs about the chance of attaining the representative career, and that:

...reducing students’ self beliefs about their chance of having their top-ranked major’s representative job by 10 percentage points decreases intentions toward that major by 0.11 standard deviations (about 3.5 percentage points, p < .05).

So, when accurate information lowers students' beliefs about their chances of obtaining the representative career associated with their top-ranked major, their intentions towards that major, and their enrolment in that major, decline. Interestingly though, correcting beliefs about a student’s second-ranked major can make that alternative more attractive. The response therefore depends on both the information that students receive and how much they value the different careers.

That would seem to be good news for some university majors, where the 'representative occupations' are relatively common (e.g. accounting or marketing) and bad news for other majors where the 'representative occupations' are less common (e.g. journalism or film studies).

Where does economics fit in as a major? Sadly, Conlon and Patel don't answer that question, as they combine economics with accounting, finance, marketing, and other business fields. In that broad group, 73 percent of prospective majors expected to enter a business career, compared with 47 percent of graduates who actually did so. So economics may not escape stereotyping. It may simply be stereotyped as a route into 'business', rather than more narrowly as a route to a job called 'economist'. However, we don't know for sure from these results.

Would better information help prospective economics students to make better choices? On the one hand, it might deter students who see economics as a guaranteed path to one particular career, but it might attract others by showing that economics is not tied to a single occupation. Perhaps the problem for economics is not a lack of possible career destinations, but that its many potential destinations are less vivid than the single job title of 'economist'.

[HT: Marginal Revolution, back in 2022 when it was Conlon's job market paper]

Will generative AI mean the end of rational ignorance?

In this Substack post back in March, Andy Hall made the case for generative AI to create 'political superintelligence':

The more I work with and study AI, the more I believe it can give every human being on the planet access to a sort of political superintelligence, if we shape it right. And that intelligence, in turn, can make governments smarter and more effective, representatives more faithful, and institutions more responsive than anything we’ve built in over 2,000 years of experimenting with democracy.

Hall's post is worth reading in its entirety, but I want to explore a related point - will generative AI mean the end of rational ignorance for voters? Rational ignorance is the idea that it may be better for voters to not know what decisions policymakers are making on their behalf. That's because it's costly (in terms of time and effort) for voters to keep track of how the decisions that policymakers (and politicians) make on their behalf will affect them (economists call those monitoring costs). The benefit that a voter would receive by becoming informed of what policymakers (and politicians) are doing is relatively small, because their ability to change an election (and therefore policy) is very small. When the monitoring costs are greater than the benefits of being better informed, then voters would be better off not paying the monitoring costs. That is, voters would be better off not paying attention to what the policymakers (and politicians) are doing - the voters would be better off remaining rationally ignorant. This theory of rational ignorance was introduced in the 1950s by the late economist Anthony Downs.

Where does generative AI fit into this? Generative AI could meaningfully lower the monitoring costs for voters, as it gives the opportunity for voters to ask for quick summaries of policy proposals that may affect them. This will be even more effective as generative AI understands more about users' preferences. Moreover, agentic AI offers voters even greater opportunity to investigate what policymakers (and politicians) are doing, at relatively low cost.

When the monitoring costs decrease, then the rationale for voters to remain rationally ignorant weakens. We might expect voters to become more engaged with what the government is doing on their behalf, and to be more active in engaging with government to make their preferences known. Or, at least, maybe voters will delegate these activities to their favourite agentic AI model.

There are, of course, some reasons for caution. Generative AI might reduce the cost of obtaining political information without reducing the cost of checking whether that information is accurate or unbiased. Moreover, an overly sycophantic generative AI that knows the voter's preferences might reinforce the voter's existing views rather than challenging them. So, perhaps generative AI simply moves the monitoring costs from monitoring the government to monitoring the generative AI?

Hall makes the point that political superintelligence has the potential to increase the quality of governance. If generative AI enables voters to become better informed at low cost, it could strengthen political accountability. Policymakers (and politicians) who know that voters can easily scrutinise their decisions may be less willing to act against voters’ interests, or may face greater consequences when they do.

We may not have political superintelligence yet, and large numbers of voters may still be rationally ignorant. However, it may not be long before we start to see some substantive changes in the political process, driven in part by the emergence of generative AI.

[HT: Marginal Revolution for the Andy Hall post]

This week in research #135

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

• Katz and Jung (open access) estimate the macroeconomic impact of generative AI using cross-country data over the period 2022 to 2025, finding that generative AI contributed to increasing the productivity of most workers, regardless of their education, contract type, full or partial work time, and vulnerability level
• Dills and Raghav (with ungated earlier version here) find that state legalization of recreational marijuana in the US substantially reduced arrests and disciplinary incidents for drug law violations on college and university campuses
• Wang and Wong find an overall negative association between financial literacy and use of Buy-Now-Pay-Later services in the US
• Funahashi and Cardazzi (with ungated earlier version here) find that sports stadiums increase the value of nearby properties in Japan
• Sayre (with ungated earlier version here) finds no evidence that Airbnb entry leads to a substantial increase in eviction filings overall, but that there is a modest increase in eviction filings in neighbourhoods with a high poverty rate and/or a high concentration of renter households
• Hampole, Truffa, and Wong (with ungated earlier version here) find that female MBAs are 24 percent less likely than male MBAs to enter senior management within 15 years of graduation, and that having a larger proportion of female MBA section peers increases the likelihood of entering senior management for women but not for men

Could prosecuting STI transmission increase infections?

In my ECONS102 class this week, we covered unintended consequences - where an incentive is created that works against what was originally intended. One of my favourite examples is the familiar (but possibly apocryphal) story about cobras in Delhi, as I noted in this 2015 post:

The government was concerned about the number of snakes running wild (er... slithering wild) in the streets of Delhi. So, they struck on a plan to rid the city of snakes. By paying a bounty for every cobra killed, the ordinary people would kill the cobras and the rampant snakes would be less of a problem. And so it proved. Except, some enterprising locals realised that it was pretty dangerous to catch and kill wild cobras, and a lot safer and more profitable to simply breed their own cobras and kill their more docile ones to claim the bounty. Naturally, the government eventually became aware of this practice, and stopped paying the bounty. The local cobra breeders, now without a reason to keep their cobras, released them. Which made the problem of wild cobras even worse.

Just because the consequences of a policy are unintended, that doesn't necessarily mean that they are unforeseen. Sometimes, we can anticipate what will go wrong with a particular policy. And it's not just policies that can go wrong. Any change in costs or benefits that alters people’s incentives can produce unintended consequences. As an example, consider this recent article in The Conversation by Bridget Haire and David Carter (both University of New South Wales):

In an Australian first, a Canberra man has been convicted for giving genital herpes to a sexual partner...

This recent case represents a significant expansion of criminal law into sexual health. It sets an unhelpful legal precedent, and undermines successful public health messages.

Decades of research have concluded that prosecuting disease transmission doesn’t reduce infection and may make things worse...

But criminalising transmission can create perverse incentives not to seek medical care and treatment. If a person genuinely doesn’t know their status, it can be more difficult to prove “reckless” transmission.

The intuitive case for punishment is especially strong in this case: the man knew his status, denied having an STI when directly asked, and repeatedly had unprotected sex with his partner. However, the punishment itself will change incentives for other people.

Ideally, we want people to know their STI status. For curable STIs, diagnosis enables treatment. For example, for infections such as herpes, it allows people to use medication and other precautions that reduce the risk of further transmission.

At one level, it makes sense to punish people who knowingly infect others with an STI. That creates a strong disincentive to transmit STIs to other people. However, criminalising STI transmission also reduces the incentive to get tested, because a person not knowing that they are infected might be able to use their lack of knowledge of their infection status as a defence in a criminal case. So, we might expect that fewer people would get tested for STIs. So, on the one hand there are disincentives to transmit STIs, but on the other hand there are disincentives to find out whether you are infected with an STI, which leads to move STI transmission. If the latter effect is larger, then overall there could be higher prevalence of STIs and greater incidence of new infections.

And so, rather than reducing STI infections, criminalising those who transmit STIs may have the unintended consequence of increasing STI infections overall.

Why rising honey prices may increase kiwifruit orchard costs

This week, my ECONS102 class covered rational behaviour, one aspect of which is the cost-benefit principle: that when evaluating mutually exclusive alternatives, a rational decision-maker will choose the alternative that offers the greatest net benefit (the greatest difference between benefits and costs). So, it was interesting to see a good example of this in The New Zealand Herald just last month:

There’s growing competition for beehives as honey prices sweeten again and kiwifruit orchards continue to grow...

[Beekeeper Liam Gavin] said renewed confidence in honey production is seeing some pivot away from pollination.

“I sort of describe it as the tug of war between honey and pollination.

“Both are needing more beehives. So which one, where are they going to go? And that’ll all be down to, like, region-specific [stuff], and what people like to do in terms of how they beekeep.”...

With honey prices coming back up, [New Zealand Kiwifruit Growers Incorporated chief executive Colin] Bond expected more beekeepers would prioritise honey over pollination, which would create a challenge for kiwifruit growers.

Beekeepers can position their hives primarily to generate income from honey production, or primarily to generate income by providing pollination services. Thus, for a particular hive at a particular time, honey production and paid pollination are mutually exclusive alternatives.

A rational beekeeper, applying the cost-benefit principle, would compare the expected net benefit from using their hives for pollination with the expected net benefit from using them for honey production. That comparison would include pollination fees, expected honey revenue, transport and feeding costs, risks to hive health, and other relevant costs and benefits. As honey prices increase, the opportunity cost of committing hives to pollination increases. Ceteris paribus (holding all else constant), as honey prices increase fewer hives will be offered for pollination.

So, if kiwifruit growers (and other farm and orchard businesses that depend on pollination) want to secure enough hives for pollination, they will probably need to offer higher pollination fees. That would raise their pollination costs and, consequently, their overall orchard operating costs.

Generative AI and grade inflation

If generative AI can produce assessed work that earns higher marks than a student would earn by themselves, widespread use of generative AI should increase measured grades even in the absence of equivalent learning gains. In other words, we might expect grade inflation to occur as a result of students using generative AI, and that grade inflation to not reflect improved learning. To what extent is this generative AI-induced grade inflation occurring? That is the question addressed by this recent working paper by Igor Chirikov (UC Berkeley). 

Specifically, Chirikov looks at the change in the grade distribution across 319 courses (and over 500,000 enrolments) at "a large, selective public research university in Texas", covering the period from 2018 to 2025. He follows the labour economics literature by measuring 'task exposure' to AI, using the share of required tasks in each course's Fall 2022 syllabus that involved writing or coding - areas in which generative AI is particularly capable and could be used as a substitute for students' own efforts. Using a difference-in-differences (DID) approach, Chirikov compares the difference in the share of A grades and GPA between years before and including 2022 (when ChatGPT was released) and more recent years up to 2025, between courses more or less exposed to generative AI. He finds that:

...grades rose substantially in high-exposure courses after 2022: the share of A grades increased by 13 percentage points (about 30% relative to the 2022 baseline) and GPA by 0.12 points, accompanied by compression of the grade distribution.

Looking at the shares of other grades, Chirikov finds that:

The share of A- grades fell by 4 percentage points, the share of B+ grades by 3 percentage points, and the shares of B and below show smaller and mostly insignificant changes.

So, courses that were more exposed to generative AI have seen a greater increase in the share of A grades and a greater decrease in the share of lower grades than courses less exposed to generative AI. Chirikov then turns to exploring the mechanism that underlies the change, by extending the DID approach to also compare courses that place greater or lesser assessment weight on homework tasks (in what is called a 'triple differences' approach). In that analysis, he finds that:

...above-median homework courses show an additional 16 percentage point increase in the share of A grades relative to below-median courses with the same level of AI exposure. The effect on GPA follows a similar pattern, with an additional 0.13 point increase in high-homework courses, though this estimate is less precisely estimated...

In above-median homework courses, the share of other grades declines significantly with AI exposure relative to below-median homework courses.

So, the observed pattern of grade inflation, with grades near the top of the distribution shifting towards As to a greater extent in courses that have greater homework weight, provides strong evidence consistent with generative AI contributing to the observed grade inflation, principally by substituting for student effort on unsupervised assessment tasks.

We should be cautious about over-interpreting the results from this study. It is based on results from a single university, and it measures course-level task exposure to generative AI, rather than students' actual use of generative AI. Nevertheless, the results are consistent with what we would expect, and are likely to hold in other contexts.

Given that, how should universities adapt to solve this issue? The obvious response is to move to more invigilated, in-person assessment. However, Chirikov cautions that:

Not all skills can be meaningfully evaluated under exam conditions: the ability to produce a well-researched essay, develop a software project, or conduct an empirical analysis requires sustained engagement that timed in-person assessments cannot capture. Restricting assessment to formats that are AI-proof risks measuring a narrower set of capabilities than the ones courses are designed to develop, potentially undermining the learning goals that graded work is meant to serve. A more promising direction is to redesign assessments so that AI use is either structurally constrained by the task or purposefully incorporated into it, for example, by requiring students to document their process, justify their choices, or demonstrate understanding through follow-up interaction.

In my view, the optimal response for universities is a combination of invigilated in-person assessment and authentic assessments in which AI use is permitted (or even encouraged) and evaluated. The appropriate balance depends on the specific learning outcomes for each course. However, as I have argued before, one approach is to scaffold students through their studies, with lower-level courses that rely more on developing core knowledge, relying less on generative AI and therefore using more secure assessment, while higher-level courses increasingly incorporate generative AI use explicitly into the assessment. This also requires scaffolding students through learning how best to apply generative AI at each level.

It almost goes without saying that we cannot simply continue to assess students as we always have done. Generative AI has broken key elements of the assessment toolkit that we previously used. This is not a new observation. However, evidence that generative AI may be accelerating grade inflation makes it even more imperative that assessment practices are updated to better reflect the availability of generative AI.

Read more:

• Teaching evaluations and grade inflation
• Grade inflation is harming students' learning
• A few papers on grade inflation at universities
• Grade inflation and college completion in the U.S.
• More on teaching evaluations and grade inflation
• Grade inflation at New Zealand universities, and what can be done about it
• Who is morally responsible for grade inflation?

Book review: Econometrics for Dummies

I just finished reading Roberto Pedace's 2013 book Econometrics for Dummies. As you might expect from a book written as part of Wiley's 'Dummies' series, the book is written as a basic introduction to its topic, starting from the basics of probability distributions, and ending with a brief primer on how to conduct an econometrics research project, as well as common mistakes to avoid in applied econometrics.

The book mostly hits the mark as a good introduction. However, Pedace clearly has a high opinion of dummies, because he assumes a great deal of statistical understanding. The book also has a lot of mathematical formulae to negotiate. To be fair to Pedace, it would be difficult to teach econometrics without the formulae without turning it into a 'recipe book' of steps to follow that would not help readers to understand. Nevertheless, I feel like the book could have been pitched perhaps a little lower, as more of a stepping stone between basic statistics and a full introduction to econometrics.

Nevertheless, the book is well written and easy to follow. Pedace does use some unusual terminology though. I struggled with his reference to categorical variables as "qualitative variables". In my mind, qualitative is something quite different. The book is also a little repetitive at times, in part because Pedace has written it in a way where the reader need not necessarily linearly follow through each chapter, but instead can jump directly to the bits of most relevance without missing out on important details that are hidden earlier in the text.

The book may be an introduction for dummies, but Pedace certainly stretches the dummies. I really appreciated that it included a discussion of difference-in-differences (and regular readers of this blog will recognise that this is a research approach that is applied quite commonly in research papers). I also thought that Pedace gave the clearest description of the difference between fixed effects and random effects models for panel data, as well as the Hausman test. Although applications of econometrics to panel data are a feature of every econometrics textbook, in my mind most do not clearly explain these models. At least, not as well as Pedace does.

Finally, the book offers an online 'cheat sheet', although I was disappointed that this seemed to simply be a static webpage, and not really a sheet that can be downloaded or printed.

Overall, this book is a good accompaniment to a full econometrics textbook, or as a memory aid for those who did econometrics some years ago and want the simple details quickly. Or, for those who want an intermediate step between introductory statistics and a full econometrics book such as Mastering 'Metrics (which I reviewed here) or Mostly Harmless Econometrics (which I reviewed here).

Broadband coverage and rural fertility

I recently expressed some scepticism at a paper showing that the release of the iPhone explained a third or more of the decline in US fertility. So, I was interested to read this new working paper by Gokhan Kumpas (California State University, Los Angeles), which looks at the effect of broadband coverage on fertility. 

Specifically, Kumpas studies broadband expansion to rural areas through the USDA's Broadband Initiatives Program (BIP) and the National Telecommunications and Information Administration’s Broadband Technology Opportunities Program (BTOP), which accepted applications in 2009-10 for broadband expansion. Kumpas identifies counties where applications for the programmes were rejected, using a subset of those counties as a control group to compare with counties where applications were successful. Moreover, Kumpas limits the control group counties to those that most closely matched the pre-treatment trajectory of the treated counties in terms of population growth, in order to deal with any problems of mean reversion.

Kumpas compares fertility among teenage women (aged 15 to 29 years) between 2010-13 and 2018-19, leaving out the intervening years where broadband was being expanded. In his main specification, he finds that broadband rollout:

...reduced the rural teen birth rate by approximately 1.6 per 1,000 in the pre-COVID post period (CHR release years 2018–2019), or about 3 percent of the pre-period baseline of 48.6 per 1,000.

The teenage birth rate for the treated counties fell from 48.6 to 34.0 per 1,000, so based on these results the effect is equivalent to about 11 percent of the 14.6-point decline in the teenage birth rate. Notice that this is a much more modest estimated contribution to fertility decline than that in the iPhone and fertility paper. Moreover, Kumpas has a good theoretical basis for believing that broadband would affect fertility, based on the opportunity cost of fertility work of Kearney and Levine (see here, for example), which:

...predicts that any local shock that meaningfully expands the perceived economic or informational opportunity set young women face should depress teen fertility. Broadband Internet expands the perceived opportunity set through several channels.

First, broadband expands informational access to contraceptive methods, to family-planning service locations and scheduling, and to the comparative costs and consequences of different reproductive choices...

Second, broadband expands access to schooling and credentialing options beyond what is locally available, including online community-college coursework, remote tutoring and test preparation, financial-aid information, and credential-program advertising. The downstream consequence is to raise expected returns to schooling and to deferring family formation.

Third, broadband expands the labor-market opportunity set by making non-local jobs visible and (with telework) accessible.

Kumpas finds results consistent with the first two of these channels, with the effects concentrated in counties that had at least one Title X family planning clinic (with no statistically significant effect in counties without a clinic), and individual-level evidence that high school completion and college attendance increased in treated counties. However, there was no evidence for changes in the adult labour market. These results are suggestive that the contraception-access and education channels explain the results, although they are not definitive.

Now, the analysis relies critically on comparing counties covered by successful applications with those covered by unsuccessful applications. This rejected-applicant comparison is appealing, but it relies on the assumption that the treated and control counties would have followed similar trends in the absence of funding. Kumpas provides considerable evidence in support of that assumption, including by selecting control counties with similar pre-treatment population trends, but it cannot be tested conclusively. Funding decisions were based partly on project benefits, viability, and sustainability, and the applications would have included information about subscriber projections and local demographics. It is therefore possible that factors related to anticipated demographic change influenced both the likelihood of receiving funding and subsequent fertility trends. This is similar to the concern I raised about the iPhone and fertility paper.

Nevertheless, if we take the results at face value it appears that broadband access contributed to reduced teenage births in rural US counties. Let's not get carried away though - broadband explains only around 11 percent of the decline in the teenage birth rate. That is a meaningful proportion, but most of the decline in teenage births happened for other reasons. In other words, fertility would have declined substantially even without the contribution of broadband access.

Read more:

• The pandemic 'baby bust' in other countries
• Cohort effects and the downturn in US fertility since the Great Recession
• The new economics of fertility
• The economics of the falling total fertility rate in New Zealand
• The economics of fertility in high-income countries
• Can fertility return to replacement levels?
• You can make future population decline disappear just by changing the way you categorise people and fertility
• The modest impact of Australia's baby bonus on fertility timing
• My take on that iPhone-fertility paper

This week in research #134

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

• Wu and Lee use a theoretical model to show that higher military spending stimulates aggregate demand, thereby promoting employment and income growth, particularly among low-income groups, and consequently reducing income inequality
• Wang (open access) finds using data from British Columbia that having larger proportions of female peers has a large positive effect on students' choice of a STEM major
• Clark and Nielsen (open access) conduct a meta-analysis on the returns to education, including 79 studies that use changes in the minimum school leaving age to identify effects, finding that the average return to an additional year of education is 8.2 percent
• Giuranno and Manni (open access) review the literature on the impact of TikTok on elections, concluding that TikTok functions as a fast-moving marketplace for political ideas in which algorithmic incentives may shape conditions relevant to electoral integrity
• Collischon and Zimmermann (open access) unsurprisingly find zero effects of western Zodiac signs on wages, education, and managerial status among workers in Germany
• Garcia et al. find that both social trust and civic pride contribute to the willingness to pay to prevent the relegation of a professional football (soccer) team, with social trust more relevant for those who attend games

Also, I am delighted to report that my Master's student, Josh McNamara, won the Jan Whitwell Prize for the best student research paper at the New Zealand Association of Economists conference last week. Josh has blossomed into a standout research student over the last couple of years, and will no doubt do great things in a PhD programme in the near future. Congratulations Josh!

Finally, I can also report a marker of my own mild fame. While in Scotland last week, I visited my ancestral family lands at Achnacarry in the western highlands, and the Clan Cameron Museum. My wife asked if they would add me to their list of famous Camerons, and after looking me up online (including my blog!), they agreed! So, I join the likes of former UK prime minister David Cameron, Pearl Jam drummer Matt Cameron, and others on the list.

Spark's new overseas roaming charges and price discrimination

I've just gotten back home from three weeks in Europe. One irksome but necessary aspect of travelling is mobile phone roaming. While I was away, Spark introduced new roaming charges, and their new options both increase the price per day of roaming for most overseas trips, and price discriminate so that those staying overseas for longer pay a higher price for roaming. As the New Zealand Herald reported:

Spark customers travelling overseas for the school holidays face new charges to stay connected, with the telco scrapping its cheapest $25 fortnightly roaming pack.

The company is overhauling its roaming plans, with the new charges taking effect this Friday, including an automatic $10-a-day fee if customers don’t turn roaming off...

Previously, pay monthly customers could use 2GB of data on one of the provider’s 14-day roaming packs, priced at $25 and $30...

Three new packs will replace the old plans, alongside a new daily roaming option.

A $30 14-day pack will still be available to prepaid customers.

The other options will provide travellers with 20GB to use over 30 days, a change the company believes will make roaming simpler and more predictable.

“This helps our customers to stay connected for longer, with fewer top-ups, less uncertainty, and greater confidence about what they’ll pay.”

While the $50 and $65 packs have a higher upfront cost, customers would receive five times more data to use than under the previous plans, the spokesperson said.

If you look at the price per gigabyte of data, the new roaming packs are clearly much better value. Customers are paying twice the price, but getting ten times the data. So, high data users are likely to be better off under these plans. I want to focus instead on travellers who are not using large amounts of data (and for simplicity, I'm going to focus on the data-only packs, not the more expensive packs that include roaming calls and texts). For those travellers, when you look at the cost per day of roaming, the new packs are far more expensive. This is illustrated in the diagram below, which shows the costs for up to 30 days of roaming. The bold green line shows the existing pricing for a 14-day data-only roaming pack ($25 for each 14-day period). The light blue dashed line shows the cost using the new $10 daily roaming rate. The orange dashed line shows the cost for the new 30-day data-only roaming pack ($50 for each 30-day period).

For a Spark customer roaming for one or two days only, the new daily roaming pack is the cheapest option. So, if you're travelling to Australia for a day or two of shopping or to attend a concert or a sporting event, the new option is a better deal than what was previously on offer. With the new options, daily roaming is lower cost than buying a 30-day pack for up to four days of roaming, and the same cost as the 30-day pack for five days of roaming. Beyond that, you would be better off buying the 30-day roaming pack, even if you are only roaming for seven days.

The comparison between the old 14-day roaming pack and the 30-day roaming pack makes it clear that anyone roaming between five days and 14 days will now be paying twice as much as before. From 15 to 28 days, the cost of roaming with the new packs is the same as for the old packs. For someone like me, who typically goes overseas for a conference and might be away for 10-14 days at a time, this is clearly going to increase the cost of roaming.

It may be that Spark has determined that the new pricing options better reflect actual customer usage. That is what a Spark spokesperson argues in the New Zealand Herald article. However, it is also clearly an example of price discrimination in action. Travellers going overseas for a few days likely have more elastic demand for roaming than travellers going overseas for a longer time. That's because of the availability of close substitutes. If you go overseas for a few days, you could make use of free hotel and airport WiFi, or be prepared to just switch off mobile data for the time you are away, rather than paying for roaming. So, travellers who go overseas for a few days are likely to be relatively price sensitive. Travellers going overseas for a longer time are less likely to be able to switch off mobile data for that length of time, making them less price sensitive. The optimal pricing therefore is to set a higher price for travellers going overseas for a longer time than for those going overseas for a few days.

Price discrimination is very common in practice. In this case, Spark is using price discrimination and that will likely increase their profits. And that means that many travellers who are not high data users, myself included, will be paying more for roaming in the future.

This week in research #133

This week I attended the 18th International Conference on Education and New Learning Technologies in Palma de Mallorca. Often, conferences in holiday locations are little more than an academic junket, but this conference certainly bucks that trend, which is why I keep coming back to it every few years. As you might expect given the title of the conference, this year it was heavily dominated by research on generative AI in education (at all levels, from K-12 to postgraduate). My own presentation was on the impact of introducing Harriet, our ECONS101 tutor, on student performance and experiences in ECONS101. I'll be giving a longer version of that presentation at a seminar at the University of Exeter in the UK later today. Anyway, here are some of the highlights I found from the conference:

• Chris Godfrey outlined the factors associated with student persistence at Manchester Business School, showing that at-risk students can be identified early by the extent of their engagement in their studies, as well as quantitative difficulties
• Ioannis Famelis presented on designing a custom GPT and Perplexity Space that can guide students on questions related to their graduate studies, specifically on curriculum regulations and identifying the most appropriate thesis supervisor
• Deniz Iren presented in detail on the development of custom AI tutors using a bespoke platform that looks very promising, called WiseTutor.ai
• Jonathon Cohen discussed how scenario-based learning and assessment could be used to maintain authenticity of assessment and real-world contexts in the face of generative AI
• Soeren Dressler looked at student perceptions of oral examinations as an assessment tool, and found that students had a strong preference for written examinations over oral examinations, potentially due to fear of the oral examination format, but students also believed that oral examinations were better at ensuring their learning
• Marian Hurley presented on student perceptions of continuous assessment, showing that students recognised the challenges to continuous assessment arising from generative AI, and showed a strong preference for a move to more invigilated assessment

Aside from the conference, here's what caught my eye in research over the past week:

• The Nobel lectures by Joel Mokyr, Philippe Aghion, and Peter Howitt
• Bluhm, Lessmann, and Schaudt (with ungated earlier version here) find that cities that gain the status of being a capital city grow faster in the medium term, and this growth spills over to nearby cities, and this arises through migration of educated individuals to capital cities and increased public and private investment

Airports and regional development

Most large regional cities have their own airports. Is that because growing regions are more likely to open an airport, or because having an airport leads to faster population growth for small regions? Probably, it is a combination of both, but empirically they are difficult to disentangle. However, this 2025 article by Jørn Rattsø (Norwegian University of Science and Technology) and Nicholas Sheard (Deakin University), published in the Journal of Economic Geography (open access) attempts to answer the question of how much regional airports contribute to growth.

Rattsø and Sheard focus on the example of Norway, where the number of regional airports grew rapidly from the 1950s, with fifty new airports opening between 1950 and 2019. They apply an event study difference-in-differences approach with synthetic controls. That means that they compare regions where an airport opened with synthetic controls made up of a weighted average of other regions, between the time before and the time after the opening of the airport. The outcome variable they concentrate on is the regional population, but they also look at employment (in total and by broad industry category).

Rattsø and Sheard find that:

...regions where airports were opened subsequently experienced growth in both population and employment, relative to otherwise similar regions that had been on similar growth paths before the airports were opened...

The size of the effect is relatively modest, with population growth about 0.4 percent higher after 1-5 years, 0.9 percent highers after 6-10 years, 0.5 percent higher after 11-15 years, and no difference after 16-20 years of the airport opening. Rattsø and Sheard also report a number of heterogeneity analyses, which are interesting too:

The population growth effects of new airports are largest and most significant for airports established in the first decade studied (the 1950s) and for new airports opened where there were no other airports within 100km... the growth effects are relatively large and more often statistically significant for airports that are physically larger (measured by length of runway) and that have a connection to at least one of the four largest cities in the country.

The first of those heterogeneity results points to a potential problem with the analysis. Airports are not opened randomly. Governments are more likely to open airports in regions where those airports are likely to have the largest effects first. And so, the effects being largest for the airports that were opened in the 1950s may be because those regions were going to grow rapidly regardless of whether an airport was located there or not. The synthetic control method attempts to deal with this by comparing regions with an airport with a weighted average of other regions without an airport, where the weighted average control 'looks like' the region that received an airport. However, this approach can only ever provide an imperfect control, because the reality is that the regions that are part of the control did not receive an airport, and if airports are allocated first to regions that are likely to grow faster, then the comparison with the synthetic control may simply pick up that fact.

The other heterogeneity results are consistent with what we would expect if regional airports do lead to faster population growth. If airports increase growth, then larger airports should increase growth by more. And connectivity matters, particularly to larger regions (although it is worth noting that when you have an airport, the flights go in both directions, and so it is by no means a given that increasing connectivity leads to net in-migration). The results for employment are also consistent with expectations, with increases in employment in the 'transport and communications' sector, as well as services.

Rattsø and Sheard rightly conclude that:

...the effects were concentrated in the early era of expansion when the air network was much less developed and similar benefits are not likely to be available today. In addition, the effects of having a small airport are limited: having an airport with little air traffic and few connections is not helpful for regional development. For peripheral regions, it may be better to improve road and other infrastructure to reduce travel times to larger airports with better connections, rather than building their own airports.

None of those conclusions should be surprising. However, the results from this study should caution against small regions in modern times arguing strongly for the opening of a new airport. Taking the results from this study at face value, where the air network is already extensive, adding an additional small airport will have little effect on population growth. There may be other good reasons to open a small regional airport, but expecting an increase in population growth should not be among them.

This week in research #132

Here's what caught my eye in research over the past week (a very quiet one, as I've been travelling in the UK, which also explains the lack of blog posts this week):

• Skryabin tests the proposition that 'stolen food tastes better', finding that 'high-risk covert taking' increases pleasantness of food by 39.3 percent compared with legitimate consumption
• Cortinhas et al. (open access) find that lecture absenteeism at UK universities is strongly associated with access to recorded lectures, inconvenient scheduling, less engaging sessions, and high student workload (no surprises there), and that tutorial attendance is higher when tutorials feature exclusive content, interactive problem-solving, and opportunities to ask questions

This week in research #131

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

• Chan (open access) finds that, between 1870 and 1910, ports which increased their proportion of steam in shipping volumes increased trade by diversifying their trade flows in terms of the range of trading partner countries and products traded
• Brodeur, Kattan, and Musumeci (with ungated earlier version here) study the relationship between statistical significance and placement outcomes for 200 empirical economics job market papers from 2018-2021, finding that marginally significant results are associated with higher academic placement likelihoods, providing a strong incentive for young researchers to 'p-hack' for statistical significance
• Gershoni and Stryjan (with ungated earlier version here) find significant declines in both exam attendance and demonstrated knowledge following the switch to online instruction during the COVID-19 pandemic in Israel

My take on that iPhone-fertility paper

If you've been reading the news over the last week, you may have seen talk about new research linking fertility decline in the US to the release of the iPhone. For example, the New Zealand Herald reported that:

Middlebury College economist Caitlin Myers and her student Ezekiel Hooper tested a hypothesis that smartphones - which emerged with the arrival of the first iPhone in 2007 - might have something to do with it.

Until 2011, iPhones were available from a single US cellular network, AT&T, so they compared US counties that had near-universal AT&T coverage with those that had little or none during those years.

And they found that access to the iPhone correlated with reductions in births by 4.5% to 8% at ages between 15 and 19, and by 3.2% to 6.6% at ages between 20 and 24.

There were also statistically significant but smaller declines among older women.

Other news sources picked up that the research attributed 33 to 52 percent of the decline in fertility to the iPhone's release (see here and here, for example). That result made me sceptical, and my concerns really echo those of Tyler Cowen here:

In 2008, 1.9% is the share of the mobile-subscribing population with an iPhone wireless subscription.  As a percent of all adults that is 1.6%.

In 2009, it is 4.3%.  3.6% of all adults.

In 2010, 6.8%.  5.5% of all adults...

So when the authors talk about diffusion explaining 33–52% of the decline in the general fertility rate among American women 15–44, I still do not get how that is supposed to operate.

If less than six percent of all adults have an iPhone by 2010, how could iPhones reduce fertility by between one-third and half? This requires very large spillovers from a small group of early adopters, and I am not convinced the paper has made those spillovers quantitatively plausible (we'll get to the authors' views on that later).

The research is reported in this NBER Working Paper by Caitlin Myers and Ezekiel Hooper (both Middlebury College). They use data on national wireless broadband coverage at the census block level to categorise US counties into those where less than 10 percent of the population have coverage by AT&T ('control' counties) and those where more than 90 percent of the population have coverage by AT&T ('treated' counties). Their sample includes 1399 'control' counties, and 914 'treated' counties (with 794 counties excluded from the sample). The reason that Myers and Hooper chose AT&T is because AT&T had an exclusive arrangement with Apple for almost the first four years after it was first launched in June 2007. The first Android phones didn't become available until October 2008, and didn't become widespread in the 'control' counties until a year later. So, there was a period where AT&T coverage is a reasonable proxy for the prevalence of iPhones.

Myers and Hooper then compare control counties with treated counties in terms of annual age-specific fertility rates (in five-year age groups). However, they recognise a key problem, which is that the treated and control counties differ in meaningful ways, the most obvious of which is that the treated counties are more urban than the control counties. This is a problem for their analysis because fertility rates have been declining more rapidly in urban areas than in rural areas, and therefore this would lead to overstatement of the measured effect of iPhone coverage on fertility. Specifically, the CDC reports that from 2007 to 2017, the total fertility rate fell by 12 percent in rural counties (many of which will be in the control sample), but by 18 percent in large metro counties (which are almost certainly in the treated sample).

Myers and Hooper try to deal with this problem by re-weighting their data in two ways. The first is by using an "entropy balanced Poisson event study", which effectively re-weights the control counties by giving more weight to those that are most similar to the treated counties in terms of their cross-sectional characteristics at the time of the iPhone launch. The second is by using a "synthetic difference-in-differences estimator", which creates a set of synthetic control counties by re-weighting the control counties so that the time series of fertility most closely matches each of the treated counties.

Using those methods, Myers and Hooper find the results that the news media has picked up. Specifically:

Both estimators imply large, statistically significant declines in births to young women. The post-gestation ATT ranges from −4.5 to −8.0% at ages 15–19 and −3.2 to −6.6% at ages 20–24 (the entropy-balanced Poisson at the lower-magnitude end, SDID at the higher), with smaller effects at older ages. Scaled to the U.S. county universe, these estimates imply the iPhone accounts for between 33 and 52% of the 2007–2011 decline in the general fertility rate. The pattern is similar across race, parity, marital status, and education, with the exception of Black women, for whom we estimate no effect.

The key results are summarised in Figure 3 from the paper (for the entropy balanced Poisson event study):

And in Figure 4 from the paper (for the synthetic difference-in-differences (SDID) estimator):

In both cases, the point estimates from the time before 2008 show no statistically significant difference between treated and control counties, while there is a negative (and increasing) difference between treated and control counties from 2008 onwards. However, notice that in Figure 3 (the first figure above), it seems clear visually that the downward trend starts before 2008, even if it is statistically insignificant. In Figure 4, there is no pre-trend, but remember that in the SDID analysis, the controls are reweighted to replicate the pre-treatment time series of fertility for the treated counties, so there should be no difference in the pre-treatment values by construction.

Myers and Hooper run various robustness checks that address some of the more obvious criticisms of their approach, including sensitivity to the choice of treatment and control cutoffs, using a continuous treatment variable, estimating the model in levels rather than logs, various placebo treatments, and truncating the sample to exclude any contamination from the release of Android phones. Among the placebo tests, they run analyses using Verizon's and Sprint’s pre-2011 coverage, and find no effects. So, their findings are not general to the difference between counties that attract mobile operators and those that don't. They also address the plausibility of the results, noting that:

The iPhone is not a treatment that operates at the individual level. Whether one’s own phone matters likely depends on whether one’s peers have phones; a phone in a friend group full of non-owners is a different intervention than a phone in a group where everyone has one. Spillovers run between phone-owning peers and their non-owning friends, and operate at the level of the group, not just the match: if smartphones reduce friend-group meetups and parties, then matches that would have formed under no-iPhone simply never do—the unformed match is itself the outcome.

That may be so, but the implied size of the spillovers is far larger than is plausible. If, as Cowen suggests, less than 15 percent of the population have iPhones, unless iPhone ownership and the spillovers from iPhone ownership were heavily concentrated among women of childbearing age, the overall effect simply can't be that large.

So, what has gone wrong. The overall approach that Myers and Hooper apply seems valid on the face of it, and re-weighting of controls to better match the treated sample is a common method of causal inference. The problem here is that the weighting is extreme. Myers and Hooper note that, in relation to the entropy balanced Poisson event study approach:

Balance comes at a cost: equalizing the marginal means requires putting high weight on a small number of treated-like controls. The Kish (1965) effective sample size of the balanced control pool is 77 out of 1,399 raw controls...

So, basically the analysis is heavily skewed towards a comparison between the treated counties and a small number of control counties, which are the control counties that are most like the treated counties (which also makes them the most unlike the other control counties). Those control counties are doing a lot of the work in this analysis.

There are also other possible differences between urban and rural counties that are approximately contemporaneous with the release of the iPhone. First among these is the 'Great Recession' and the housing slump around that time. Myers and Hooper do control for county-level changes in house prices, so that reduces concerns about contamination from that source. They also control for unemployment and poverty rates, which might pick up differential changes in labour markets. However, there was a change in contraceptive availability that directly affects young women's fertility, which is expanded access to the 'morning after pill' for 17-year-olds, although that occurred in 2009. Finally, after the 'Great Recession' there was a slowdown in Hispanic immigration, which might have affected urban and rural counties differently. Given that Hispanic immigrants tend to have relatively higher fertility than the US-born, so if the decline in Hispanic immigration was greater in control counties (and especially for the small number of heavily weighted control counties), then that might explain the effect. Myers and Hooper control for county Hispanic population share. However, it would be better to control for Hispanic population share among the age group that is being analysed, or to control for changes in Hispanic immigration.

This paper has certainly gotten people talking. Smartphones might be part of the story of why fertility has declined, but I don't think that we should uncritically take away from this study that the iPhone caused half of the decrease in US fertility between 2007 and 2011. More likely, it had a modest effect (if at all), and is confounded by a number of other changes that differentially impacted rural and urban US counties at around the same time.

[HT: Marginal Revolution]

Read more:

• The pandemic 'baby bust' in other countries
• Cohort effects and the downturn in US fertility since the Great Recession
• The new economics of fertility
• The economics of the falling total fertility rate in New Zealand
• The economics of fertility in high-income countries
• Can fertility return to replacement levels?
• You can make future population decline disappear just by changing the way you categorise people and fertility
• The modest impact of Australia's baby bonus on fertility timing

Book review: How to Think Like an Economist (Roger Arnold)

If you ask many economics teachers, they will tell you that they really want to teach students how to think like an economist. However, in amongst the supply and demand curves, the elasticities, and the multiplier effects, the core goal of teaching students to actually think like an economist gets lost, overwhelmed by a lot of do this stuff like an economist. So, it's interesting when a book actually tries to get behind the models and teach the underlying thinking.

That's what the 2005 book How to Think Like an Economist, by Roger Arnold, tries to do. Arnold explains that:

To teach students how economists think, we must tell them stories. While we tell the stories, we must point out just what is "running through the economist's head." In this book, I have tried to focus on what goes through the economist's head as he or she looks at the world.

And mostly, Arnold is successful, although it isn't always the case that every economist would think in the same way. For example, Arnold makes a big deal about ratios. And while ratios are important, I for one am never thinking about the ratio of marginal benefit to marginal cost, when I can simply think about which one is larger. The ratio is redundant.

There is a lot to like about this book, and Arnold surfaces some of the more surprising (to non-economists) ways that economists would think about problems. For example, who but an economist would even ask the question, "What is the optimum amount of hitting yourself in the head with a hammer?". And yet, Arnold treats us to a consideration of exactly that question in the second chapter.

Having said that, I felt like the book was quite uneven. Although Arnold warns readers at the beginning that the book is intended as a companion to a more thorough textbook economics treatment, and gives examples of how the chapters can be mixed and matches with various styles of economics courses, a reader reading the book chapter by chapter is constantly confronted with terminology that is left unexplained until later chapters. This was most jarring in the case of the 'equilibrium price', which came with no explanation of what equilibrium is, nor why the equilibrium price is important at all. Similarly, Arnold uses the term ceteris paribus first, without explaining what it means. And if you want to understand how the economist thinks, understanding the meaning of ceteris paribus (which, for the record, means holding all else constant) is kind of important.

Arnold also betrays a lack of understanding of some real-world context. Blackjack is provided as an example of a zero-sum game played between the players. However, blackjack in the real world is not at all like that. Blackjack players are playing against the house, not against each other. One blackjack players win does not in itself entail a loss to the other players.

So, although understanding how economists think is important, and I applaud the effort and the approach that this book takes, I feel like it fell a bit short of the mark. This book is long out of print, but that might not be such a bad thing.

This week in research #130

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

• Fumarco and Groero (open access) describe a Stata package that reduces a dataset down to just those variables that are used in a particular .do file (useful for creating replication packages while minimising data bloat)
• Cox (open access) describes three Stata commands that creates a new dataset of the quantiles, percentiles, or confidence intervals for a particular variable or result (if you've ever needed to do this, you will know how frustrating it is)
• Yarashov, Baryshnikova, and Kakhkharov find that military expansion exerts a significant negative impact on fertility across 15 post-Soviet countries between 1992 and 2022
• Chatterjee, Dimova, and Ojha (open access) find, using a correspondence study in urban India, that equally qualified single mothers are much less likely to receive interview callbacks than unmarried women without children, married women, and married mothers
• Charness et al. (with ungated earlier version here) provide a convincing argument of the virtues of lab experiments in economics
• In a companion piece, Gneezy examines the principles of experimental economics
• Wang finds that China's policy to limited young peoples’ access to online video games did not produce detectable effects on academic performance, study time, or health
• Pritchett and Viarengo (open access) demonstrate that ad hoc poverty lines, including the World Bank's poverty lines, are far too low to be plausible candidates for an inclusive global poverty line