Parenting, School Closures, and Student Achievement During the Pandemic
Jerry Nickelsburg, Director, UCLA Anderson Forecast
Emily Oster, Professor of Economics, Brown University
Leo Feler, Senior Economist, UCLA Anderson Forecast
- After 6.9% annualized GDP growth in Q4 2021, the economy has taken a hit with rising Omicron cases this past month. The Household Pulse Survey recorded that an estimated 8.7 million workers – approximately 5% of the labor force – were out sick themselves or caring for someone who was sick during late December/early January. See Exhibit 1. This is the highest this number has been so far during the pandemic. Real time data show that businesses tried to adapt by hiring temp workers to fill shifts. It appears that as Omicron is fading, worker absences are also coming down. Nonetheless, January has likely experienced a slowdown in employment and economic activity because of Omicron.
- With the Consumer Price Index (CPI) rising 7% in 2021 and based on information from speeches by several Federal Reserve board members, we’re expecting at least 4-5 interest rate increases over the coming year, with the Fed’s benchmark rate reaching 1.25%-1.50% by the end of 2022.
- There are several reasons why we think inflation will come down substantially in 2022. The first is that we expect the Federal Reserve to respond more aggressively and tighten monetary policy. This will reduce demand and take some of the pressure off of asset prices. Second, half of the CPI inflation we experienced in 2021 was in just two sectors: energy and vehicles. We believe that price increases in these two sectors have run their course and, while prices might not decline any time soon, they’re not likely to continue increasing as they did last year. As Matthew Klein points out in a post in The Overshoot, if the prices of just a few specific items had grown at their pre-pandemic pace, instead of jumping by 50% or more, inflation in 2021 would have been close to the 1995-2019 average. Exhibit 2 shows a chart prepared by Matthew Klein with the contributions to monthly inflation from various categories.
- Unemployment reached 3.9% in December. While the unemployment rate has almost completely recovered, the labor force participation rate has not. We still have 2.9 million fewer workers than we did before the pandemic. Even with fewer workers, GDP has more than fully recovered. That means the economy has become more productive, producing more output with fewer workers.
Exhibit 1: People not working because they had COVID or were caring for someone with COVID, Ages 18+, in millions
Source: Ben Casselman, based on Census Bureau / Household Pulse Survey Data
Notes: Colors reflect different survey phases and may not be directly comparable
Exhibit 2: Most of the volatility of monthly inflation can be explained by price spikes and declines in a small group of pandemic-sensitive categories that collectively account for only 30% of the broader economy
Source: Matthew C. Klein, Understanding Covid-flation, The Overshoot, January 19, 2022, available at: https://theovershoot.co/p/understanding-covid-flation. Based on Bureau of Labor Statistics Data.
Notes: Reopening categories are restaurants, hotels, airline fares, recreation services, personal services, club memberships, live events, school tuition and fees, childcare, personal care services, and motor vehicle insurance, rentals, and other fees.
This month, our podcast features a conversation with Professors Jerry Nickelsburg and Emily Oster regarding research on parenting decisions, COVID school closures, student achievement during the pandemic, and correlation versus causality. Below is an edited transcript of their conversation.
Jerry Nickelsburg: Please welcome Professor Emily Oster to UCLA and the Forecast Direct Podcast. Professor Oster is the Executive Director of the Covid-19 School Data Hub, the Royce Family Professor of Teaching Excellence, and Professor of Economics at Brown University.
The Millennials and Gen-Y’s who are reading this and are contemplating or actively involved in parenthood know Emily well through her trilogy: Expecting Better, Crib Sheet, and The Family Firm, which is about using data for decisions in parenting. Emily, your research is on Health Economics and Research Methodology. How did you move from this to issues on decision and parenting?
Emily Oster: I started on this part of my career when I got pregnant. Expecting Better, in particular, was really driven by my own experience of being an economist who was pregnant. I entered that phase of life thinking that I would apply the same tools that I applied in my job and use data to think about decisions. I found the landscape of decision-making in pregnancy was quite different, and sometimes less data-friendly than I had hoped it would be. My beginning there was to say, “Okay, I need to collect this data and do some of this analysis to make the right decisions for my own pregnancy,” and it moved into writing for a broader audience. I’ve always been quite interested in communicating to a broader audience the things we find in our research. That has always been something that I like to do. I think that I have a particular take on it that sometimes works. I was doing this research for myself and then thinking, “I could write about this for someone else.” That was the origin of the first book.
Jerry Nickelsburg: Economics is all about the allocation of scarce resources, and anyone who has raised children knows that raising children is all about the allocation of scarce resources. Why do you think our profession has ignored this part of economics for so long?
Emily Oster: If you go back there is a fair amount of household economics – the study of the way people allocate labor time which overlaps with parenting. Think of Gary Becker’s original “quantity / quality tradeoff.” The idea that you would have fewer children and invest more in them is fundamentally rooted in the observation that your time and your resources are finite. If you have fewer kids, you can share the resources across more of them. In terms of moving from the generic “how are we describing the way the world is operating and the demographic transition” into the question of how can we use economics to organize parenting better, I think that, although I personally feel there are many tools from economics that are useful, it’s been a little bit of a weirder transition in part because this field of study has been gendered. Economics has been a male-dominated field, there’s sometimes the feeling of “parenting is for ladies,” and it’s only more recently that studying parenting has become more acceptable.
Jerry Nickelsburg: You’ve done a lot of work in the area of sampling and methodology of analysis. I’m interested in your recent papers on bias in observed and unobserved variables as they affect how to interpret results. Can you explain more about this and what it means for interpreting data and making decisions?
Emily Oster: A huge share of my recent academic work has been around what we learn from observational data, and how do we separate causality and correlation. There is a huge feedback between the academic work and the non-academic work for me. Prior to writing Expecting Better, my work was much more on the health care space. My interest in statistical methods was influenced by the work I was doing, the papers I was reading, and writing about the public health literature in the context of pregnancy. There’s a lot of examples where it’s like “we’re seeing a lot of correlation, but it’s not obvious we’re learning about causality.” A good example is data on breastfeeding. Is breastfeeding linked to child IQ? Well, that correlation is really strong. It is definitely the case that being breast-fed is correlated with having a higher IQ score on later tests. But when you look at what else is going on in the data, you find that women who breast feed are better-educated and perform better on IQ tests, they tend to be richer, and they are more likely to be married. There are many variables that correlate with breastfeeding that we know to be independently associated with child achievement. That means it’s fairly important that you adjust for those and think carefully about holding those constant. Some of my work is just about how hard that is to do, and how in practice it’s very difficult to be confident that we are holding constant all the things that are different across people. It’s one thing to control for someone’s education or income, but it’s another to be able to say that covers all the things that are different about them. In some of my statistical work, I talk about the issue of observables, the things you can see that might be impacting your estimates, and unobservables, the things you can’t see, which may be impacting your estimates. I talk about how researchers can make assumptions about the relationships between those things that would allow them to maybe make some additional progress, with the recognition that in a lot of these settings, it’s really hard to get around the inherent problem of unobservable bias.
Jerry Nickelsburg: That’s really important because in the discussion of the application of our research to policy, thinking about causality but also thinking about these unobservables will help us frame better policy.
Emily Oster: I agree. I think we’re often trying to take our estimates and convert them to something where we’re making either policy or behavioral pronouncements. When we’re doing that, it is pretty important to be clear on causality, because the policies that we’re making are only going to work to change outcomes on the assumption that the policy and the outcomes are somehow causally linked. If in fact, they are just correlated and there’s no causality, changes to that policy aren’t really going to matter.
Jerry Nickelsburg: Let’s turn to your research of schools, which has recently ended up in the public discourse. You have studied the transmission of COVID by school children in schools that were open, and you have studied the educational impact of closures.
Emily Oster: I’ve been working on this issue of schools and COVID almost since the beginning of the pandemic. It’s worth thinking about why we ended up in the place that we did on this topic. In March 2020, virtually all schools in the entire world – with the exception of Sweden – closed for in-person learning. I think there were a lot of good reasons for that. One was that we had no idea what was going on. Another is that school closures are commonly thought to be a potentially important mitigation factor in, for example, influenza – which is a disease that shows up a lot in kids, and where we think that schools may be centers of spread. As we moved into the summer of 2020, many European countries reopened their schools realizing that COVID didn’t affect kids with serious illness in the way that it did adults. In light of that, many places in Europe opened their schools. By the fall of 2020 in the US, there was also a fair amount of variation with some schools open and some not. I started working on collecting data on COVID cases and student achievement in each of the schools. One of the results was on the question of whether schools were sources of spread. The data clearly show that as of the fall of 2020 this did not seem to be going on. The other question was whether open schools spread Covid in the community? We find in our paper in Nature Medicine, using a nationwide sample on when schools were open and not open and using data on community case rates, that in fact opening schools with appropriate mitigation did not seem to drive increases in cases. I think that was encouraging and in line with much of what we’re seeing, which is that schools don’t seems to be a significant source of Covid risk.
Jerry Nickelsburg: I’d like to hear more about your recent paper on differentials in learning.
Emily Oster: As we moved through the end of last year, our team pivoted to try to understand some of the consequences of school closures. If you look at the map of what happened across the US in the first pandemic school year, you see a huge amount of variation in school openings, and the data collected on that was not great. We spent last summer working with individual states to pull data together. For example, a lot of places in California were closed almost the whole year, whereas in Wyoming schools were mostly open for regular in-person school the entire school year. We merged that information together with data on test scores from state standardized tests at the end of the school year. We looked at the relationship between the change in test scores from the last pre-pandemic year to the test scores at the end of the 2021 and whether the school was open for in-person learning. What we observed was that there were fairly significant losses in test scores from 2019–2021, but those declines were much larger in districts that were fully remote than in districts that had in-person learning. In general, the pandemic was decreasing test scores in math, but that was even more true in districts that did not have in-person learning.
Jerry Nickelsburg: This is now the debate in New York, Detroit, Milwaukee, Chicago, and elsewhere with the new Omicron wave.
Emily Oster: One of the early priorities of the Mayor of New York City has been for schools to stay open. I think the reason for that is the duel of these two factors. First, we aren’t seeing schools as a significant source of COVID spread. Second, we are seeing very large losses, not just for learning, but also for mental health. I think it’s worth noting that the students whose lives have been most disrupted tend to be students who are low-income and in districts with larger populations of students of color. In our research on the losses here, we see not only are students in those districts less likely to have access to in-person schooling, but the losses associated with not having in-person schooling, the downsides of virtual learning, are larger for these groups. So, it’s a double hit on kids with fewer resources—they are both less likely to get to go to school in person, and it’s harder for them to learn when they’re not going to school in person. I think that’s really pushing a lot of policy makers in the direction that schools really need to stay open.
Jerry Nickelsburg: Does age matter, or is it ubiquitous across age groups?
Emily Oster: When we looked in the fall of 2020, which was pre-vaccine, there was quite a big difference between universities and virtually all K-12. There was way more spread in universities. If you look at the early maps of fall of 2020, you can see, for example, Penn State standing out. It’s a low-risk population with very high case rates. We didn’t see that in K-12 schools. It looks like there is something different about universities in that period. Currently, most universities have strong vaccine mandates for students. Vaccines have been widely available for a long time, and that changes the landscape. There are losses to college students from not being in-person, just as there are losses to kids in K-12. In this space with the availability of vaccines, there are a lot of reasons to keep things open.
Jerry Nickelsburg: I’d like to turn to another article of yours, with a fascinating title, Does Disease Cause Vaccination? You’re looking at vaccinations and the incidence of whooping cough across states. Do your results reflect on whether or not specific outbreaks of COVID in communities with low COVID vaccination rates are going to increase those vaccination rates?
Emily Oster: In that paper, we’re looking at something that was motivated by a story out of California. There was a period in which there was a medium-sized measles outbreak at Disneyland. One of the things that a bunch of people said in reaction to that was, “now I’m getting my kid vaccinated.” I heard that as “I thought measles was imaginary – I didn’t realize people got that. But now that I realize it’s a real disease, I’m going to get my kid vaccinated.” We wanted to see if that showed up in a more systematic way with whooping cough where there are almost always cases in any given year. So, we looked at the county-year level to see if people are being pushed to vaccination by the presence of cases in their county. We found this increase in vaccination rates in the data. The reason for this relationship is not something I can exactly tease out of the data, but my imagined idea for it is that it is because people realized whooping cough was a real thing, and their doctor said, “we have a baby in the ICU with whooping cough right now, don’t you want to have your kid vaccinated?” Somehow the COVID situation is much more complicated. If you thought all that was required to encourage people to get vaccinated was the realization that this was a real disease threat, you might have thought they would already have gotten vaccinated. So, there’s clearly a bunch of other stuff going on in the case of COVID. I will say as an addendum to the whooping cough analysis and thinking about California specifically – California had quite low measles vaccination rates, and at some point, after this measles outbreak, they changed the rules for school vaccine requirements. It used to be that if you said you didn’t feel like it, you could avoid vaccinating your kid for all sorts of preventable diseases. There was a bill passed that made it really difficult to not vaccinate your kid for a list of preventable diseases. That has huge impacts on vaccine rates – enormously bigger than anything we saw in our paper, or that people have seen in other interventions which try to encourage people to get vaccinated. It seems the mandate part of this is probably pretty important, even in pretty resistant areas.
Jerry Nickelsburg: Is the mandate a substitute for a lack of information?
Emily Oster: I don’t know if it would be a substitute, but it’s a different approach. I think as economists, we would like it to be the case that we can tell people “here’s the truth, here’s all the information,” and let people make the right decisions for themselves. In these cases where there is some compelling externality argument, it looks like mandating is more effective.
Jerry Nickelsburg: But you don’t expect the presence of Omicron to dramatically change vaccination rates because of the information or demonstration effect you found in the paper?
Emily Oster: I think there’s some of that. I will say that a week or two ago, I was volunteering at a pediatric vaccination clinic. There were definitely a lot of people showing up for their first dose suggesting to me that something about the fact that so many people had COVID was motivating the increase in vaccination rates. On the flip side, there’s clearly a fair amount of resistance in certain parts of the country to vaccines, and I doubt there will be high vaccination rates in those places. Those are also places that are unlikely to mandate vaccinations for kids.
Jerry Nickelsburg: Your work is incredibly fascinating, and it brings data and careful analysis about the difference between correlation and causality to decision-making, which is important today. Emily, I appreciate you taking the time to be with us for the Forecast Direct podcast. For our audience, go look at Emily’s trilogy. It’s fascinating, it’s amusing, but it’s also very informative about making good decisions in parenting.