Childcare adjusting to a new, lower-demand, equilibrium

This week, my ECONS102 class is covering supply and demand. Today, we talked about equilibrium, and tomorrow, we will talk about how the market shifts from one equilibrium to another. So, this story from the New Zealand Herald a few weeks ago is timely:

Childcare centres are offering big discounts for new customers as centres open up in fast-growing suburbs just as demand falls due to Covid-19.

Four out of 13 centres in the expanding Kumeū-Huapai area in northwest Auckland are offering 50 per cent discounts for new children who enrol in the next two, three or six months, and a fifth is advertising free fees until the end of July.

Early childhood planning consultant Logan Whitelaw says discounting is also occurring in Silverdale and Beachlands in Auckland, and at Rolleston near Christchurch.

In all those places, new centres have just opened or are about to open at a time when many parents have pulled back from daycare because they have lost their jobs, are working more from home or are still worried about the risk of children catching Covid-19.

Parents "pulling back from daycare" has led to a decrease in the demand for daycare. We can illustrate this using a diagram, shown below. The equilibrium price was initially P0, and there were Q0 days of childcare 'traded' in the market. Then demand decreases from D0 to D1. Childcare centres could try to keep their price high at the initial price of P0. The problem with that is that families will now only demand Qd days of childcare (that is the quantity demanded on the new demand curve D1, when the price is P0). The childcare centres are still expecting to supply Q0 days of childcare, so many of them are going to be empty. Rather than having an empty (or mostly empty) centre, it is better for the childcare centres to try to attract some of the families who do want childcare. They can do that by offering a discount, as the news story suggests is happening. The price of childcare starts to fall. How far will it fall? Childcare centres will still have more places available than children filling them all the way until the price has dropped to the new equilibrium price of P1. At that new, lower, price, the quantity of childcare days demanded (Q1) is exactly equal to the quantity of childcare days supplied (also Q1).

The discounts will likely be temporary though. When the world has returned to something more resembling normality, and employment has returned closer to previous levels, demand for childcare will increase, and the process will be reversed - the equilibrium price of childcare will go back up towards P0.

Quasi-rational students choosing their university major

How do university students select their major? We might hope that they would first identify all of the majors that are available, then carefully weigh up the costs and benefits (to them) of each available major, and then select the major that has the greatest net benefit to them. That is how a theoretically 'purely rational' student would choose, based on the cost-benefit principle. However, we know that real people are not purely rational. They are affected by a range of cognitive biases and make shortcuts in their decision-making. In other words, real people are quasi-rational.

So, how would a quasi-rational student select their major? This 2019 working paper by Richard Patterson (US Military Academy), Nolan Pope (University of Maryland), and Aaron Feudo (US Military Academy) provides a partial answer, at least in terms of identifying some of the biases that affect students in selecting their major.

Patterson et al. make use of data from 8777 students at the US Military Academy (USMA) at West Point, over the period from 2001 to 2015. At USMA, the first two years of the Bachelor's Degree are assigned to students automatically - they have no choice over the order of classes. Some classes are in the same semester and year for all students, while some are randomised between the first and second semester. Students must choose their major during the first semester of their sophomore year, and during that year they are randomised into whether they study economics and philosophy, or political science and geography, in each semester. So, during the semester that they first choose their major, students will have been studying one or other of those two pairs of courses.

Patterson et al. test whether the timing of courses matters for the selection of a major. For a purely rational student, the order in which they take courses should not matter, since the order doesn't affect either the costs or the benefits of each major. However, quasi-rational students are affected by availability bias. Availability bias arises when a decision-maker gives more weight to information that is readily available to them, which means information that they have received more recently, or information that is more salient or vivid. In either case (more recent information, or more salient or vivid information), that information will be more easily remembered and so it receives more weight in the decision.

In the case of selecting a major at USMA, Patterson et al. argue that the courses that students are taking in the first semester of their sophomore year will be more likely to be selected as a major, because of this availability bias. And indeed that is what they find:

...assignment to a course in the first semester of sophomore year increases the probability that a student will initially choose a corresponding major by 2.9 percentage points, or 110 percent. This result is highly precise and is significant at well beyond the 1 percent level.

The results continue to hold after controlling for student demographics, course and year fixed effects, demographic characteristics of classmates, and for the full course schedule. The results are the same regardless of the gender, race, or academic ability of the student. The results also hold regardless of how well the students performed in these classes, or how well they evaluated the courses at the end of the semester. Students at all levels of the grade distribution, and at all levels of course evaluation, were more likely to select the corresponding major of the courses they studied in the first semester of their sophomore year.

That's not the end of the quasi-rationality of these students though. After selecting their major in the first semester, there is very little that would stop a purely rational student from switching major in the second semester. The only cost is that they need to get a form signed by two academic advisors (one from the major they are switching out of, and one from the major they are switching into). However, a quasi-rational student is affected by status quo bias and loss aversion.

Loss aversion means that real people value losses much more than equivalent gains (in other words, they like to avoid losses much more than they like to capture equivalent gains). Loss aversion leads real people (including students) to avoid changing their minds, and to prefer the status quo. As Patterson et al. put it in this paper:

Once students select a major, they might experience psychological costs to switching majors. There may be several potential sources for this bias. For example, students might feel ownership of their initial choice and exhibit loss aversion...

Giving up something that students own (including our choice of a university major) is difficult, because it entails a loss (they lose the major that they were originally enrolled in). Patterson et al. find evidence for this status quo bias as well:

...the effects of first-semester assignment on graduating major are large and statistically significant... but about half the magnitude of the effects on initial major. First-semester assignment increases the probability that students select a corresponding major by between 1.4 and 1.5 percentage points (34 to 40 percent).

In other words, the effect of course order on selecting the major persists all the way through to graduation for some students. Some students do change major after the first semester of sophomore year, but not enough to eliminate the effect of course order on choice of major.

So, students are affected by both availability bias and status quo bias when they select their university major. They might also be affected by other biases as well, but this paper only provides evidence for those two biases. How should universities respond to this? If there are particular majors that universities want to steer students into (e.g. STEM), they could schedule students to do those courses at the time that students select their major. This might also be important if universities want to reduce gender or other biases in particular courses.

These results also suggest we should be cautious about having students select a university major before they begin their studies. Availability bias will lead students to be more likely to select into majors that are familiar because they are taught at high school (e.g. accounting) and less likely to select into majors that are not (e.g. marketing). They might also bias students in less affluent schools, where a full range of courses is less likely to be available, away from majors that are becoming less commonly taught (e.g. economics, sciences). Status quo bias then keeps those students in their initially-selected major, when some of them might be better off if they switched to a different major later.

Universities need to take care about how they set up course and major selection for students. The process itself, combined with the quasi-rationality of students, could easily lead to less desirable outcomes, for both students and for academic departments.

[HT: Marginal Revolution, last year]

Poor research doesn't help us understand the relationship between fast food availability and health

I've done quite a bit of research on the effects of alcohol outlets on neighbourhoods, focusing on the relationship between the number of outlets and an area and measures of harm, such as violence. The challenge in that sort of research is establishing causality. In particular, alcohol outlets tend to locate in poorer neighbourhoods, and those neighbourhoods tend to have more social problems. Also, people who want to drink more might want to live closer to places where they can easily access alcohol, and they might also be the kind of people who cause (or are victims of) social problems. Because of those reasons (and others), there is a lot of confounding of how large the relationship between alcohol outlets and harm actually is.

A similar problem is faced in the literature that attempts to link fast food outlets to health impacts at the neighbourhood level. So, even more than in the alcohol outlets literature, there isn't a consistent set of statistical results linking fast food outlets with obesity, for instance. However, it might be possible to deal with at least some of the issues by looking at the effects on children, because at least children aren't going to move to be closer to fast food outlets (probably!). However, that still doesn't solve all of the issues.

One example is this 2019 article by Matthew Pearce (NHS Gloucestershire Clinical Commissioning Group), Isabelle Bray, and Michael Horswell (both University of the West of England), published in the Journal of Public Health (open access). They look at accessibility to fast food outlets and obesity among children in South Gloucestershire, England, over the period from 2006 to 2013. Children were surveyed at the start and the end of the period. Essentially, Pearce et al. look at the relationship between obesity, or rather weight gain (proxied by whether each child gained 50 percentile points in a standardised weight distribution over the period between the two surveys) and access to fast food outlets (measured as a weighted score based on how many outlets are near where the child lives, and how close they are).

To me, this paper initially had a lot of promise. However, it then finds quite conflicting results that are difficult to reconcile. First:

The Spearman’s rank correlation coefficient between change in weight between Reception and Year 6 and accessibility score found no evidence of an association (r = −0.07, P = 0.768).

In essence, there was no correlation between fast food outlets and their measure of obesity. However, they then break their fast food accessibility measure into three groups, and find that:

The group of children (tertile 3) most exposed to fast-food outlets were more likely to gain significant weight (>50 percentile points) than those least exposed (odds ratio = 1.89, 95% confidence interval: 1.0–3.6, P = 0.04).

In other words, children in the top one-third (tertile) of the distribution of fast food accessibility experienced more weight gain than those in the bottom one-third. This is of course a very different finding than their initial result of no correlation.

The problem here stems from the choices the researchers have made in their analysis, and that they don't make those choices transparent to the reader of the article. Why do they use tertiles in the analysis, and not quartiles or quintiles, or indeed using the actual distribution of accessibility scores which is a continuous variable? As a journal reviewer and editorial board member, one of my pet hates is when researchers take a continuous variable and turn it into a categorical (or ordinal) variable instead. At the very least, I'd want some additional analyses to demonstrate that these results don't simply arise as a result of the researchers' choices about how they specify this variable in their analysis.

The lack of transparency is good reason to discount these results, and should be a good lesson to other researchers on what not to do if you want your research to be credible. This paper may have generated some buzz at the time of its release (including here in New Zealand), but it does nothing to help us actually understand the relationship between fast food availability and health.

The economic impact of universities

I've been quite critical of 'economic impact studies' in the past (for example, see here). These studies try to estimate the economic impact by aggregating up all of the spending associated with an event (for example). The key problem is that they don't usually properly take account of the counterfactual - what would have happened if the event hadn't taken place. In the case of the economic impact of a particular industry, it can be difficult to establish what the counterfactual actually is - what would have happened if that industry didn't exist? That is the case when you try to estimate the economic impact of a university, for example.

An alternative approach is to avoid trying to estimate the economic impact of a particular university, but instead to compare regions that have universities from those that don't. That can include comparing regions before and after a university is established there, as well as comparing regions with different numbers of universities.

Essentially, that is the approach taken in this 2019 article by Anna Valero (London School of Economics) and John Van Reenen (MIT), published in the journal Economics of Education Review (open access, but just in case there is an earlier ungated version here). Valero and Van Reenen use data on economic growth rates and the number of universities in 1498 regions across the world, covering the period from 1960 to 2010 in five-year time steps. They find that:

...on average, a 10% increase in the number of universities in a region is associated with around 0.4% higher GDP per person.

The results are robust to a variety of different specifications. However, that isn't all. Not only are universities associated with higher growth within a region, the higher growth also spills over into surrounding regions, such that:

...a 10% increase in universities in the rest of the country (which in most cases will represent a greater absolute increase than a 10% increase in home region universities) is associated with an increase in home region's GDP per capita of around 0.6 per cent.

They also find that the effect is greater for poorer regions, suggesting that:

...new universities have a stronger impact on laggard regions within a country.

Valero and Van Reenen then go on to investigate the important question of how universities might contribute to higher GDP per capita. They suggest four possible mechanisms:

(i) a greater supply of human capital; (ii) more innovation; (iii) support for democratic values; and (iv) demand effects.

The first two are obvious, but it turns out that changes in human capital (as measured by the share of university graduates) and changes in innovation (as measured by the cumulative number of patents) only explain a small proportion of the effect of universities on GDP per capita. In terms of the third mechanism, Valero and Van Reenen first note that:

Universities could promote strong institutions directly by providing a platform for democratic dialogue and sharing of ideas, through events, publications, or reports to policy makers.

This is over and above any effect that universities might have on human capital. Looking at the impact of universities on support for democratic institutions (using data from the World Values Survey), they find that:

...there is a highly significant association between university presence in a region and approval of a democratic system.

Interestingly, this result is not driven by university graduates, because they get the same result if they drop graduates from the World Values Survey sample.

Finally, Valero and Van Reenen look at demand effects, which is the only channel that a traditional economic impact study would consider - how much of an effect does the spending of students and staff and the university itself have on the economy? They find that:

...we can explain around 15% of the regression coefficient on universities.

In other words, the vast bulk of the economic impact of universities is not even going to be picked up in a traditional economic impact study.

Of course, this research isn't going to be the last word on this topic. Even though Valero and Van Reenen run a battery of robustness and other checks, their results are still based on correlations and so they cannot be definitive about universities causing higher GDP per capita. However, in their robustness checks they do seem to have eliminated many of the most plausible alternative explanations. They are also unable to definitively address questions about the extent to which the size or quality of universities matters. Universities do have a positive economic impact on a region (and neighbouring regions), and we are getting closer to understanding how large that impact is, and how it arises.