And the Oscar for Economics Goes To... (2/2)
And the Oscar for Economics Goes To... (2/2)
What can the Oscar nominees tell us about the economy?
The Oscars are this weekend and, as you might know, I like movies. Instead of looking at every single nominee and doing a deep-dive of the economics thereof, which would take around three months, I had a different idea: look at most of them, and do a briefer take on some economic concept related to them. Some of the films do have specific blog posts attached, which I’ll include as the post goes on.
Because there are ten movies nominated, I'll do two parts, each with five nominees. You can read part one in last week’s post, so this is part two. As a general rule, the post contains spoilers for the relevant movies.
The Holdovers - Getting schooled
The Holdovers is a delightful, funny, heartfelt movie that follows three people who have been left behind for the Christmas break at an exclusive New England school: a curmudgeonly teacher who everyone hates; a rebellious student with a complicated family life, and a cafeteria employee whose son recently died in Vietnam.
The first basic question is: is going to an elite prep school like the (fictional, I assume) Barton Academy worth it? Well, in “Diversifying Society’s Leaders? The Determinants and Causal Effects of Admission to Highly Selective Private Colleges” (2023) by Raj Chetty, David Deming and John N. Friedman, the authors look at the question of whether children from high income families are likelier to attend highly prestigious and exclusive schools (called Ivy League Plus schools: the Ivy League plus Stanford, MIT, Duke, and U. of Chicago), and whether attending those schools results in significant benefits versus another similarly rigorous and exclusive but slightly less prestigious college - to quote Bojack Horseman “Of course he was going to beat me. He went to Dartmouth! Where’s the candidate for regular schmoes like me, who went to Northwestern”1.
The authors first construct a dataset from the mid 90s to the mid 2010s that matches parental and child tax income, tax forms on college attendance, Pell grant records (aka student loans), standardized test score data, and applications and admissions to the Ivy Plus and to other highly selective universities. The data has a lot of nuance, but it largely contains a set of all the kids who applied to the Ivy League Plus over the period, alongside their test scores, family data, and data on actual attendance. Using this gigantic, highly detailed dataset, the authors analyze the probability of attending an Ivy League school based on parental income, conditional on grades - and find that kids from the top 1% have twice as many chances of attending the Ivy+ as everyone else. Secondly, they use the test data to examine whether this is because rich kids have better grades, or other reasons - and find that roughly half the difference can be explained by athletics (a difference that does not exist in the non Ivy+ selective colleges), legacy preference (who have a four-fold increase in likelihood of admission not warranted by their grades), and non-academic credentials, such as participation in extracurriculars (which are highly correlated with income). More relevantly to this point, estimating the impact of attending specific high schools on Ivy+ admission yields that attending a school with low-income “disadvantaged” types is not especially beneficial, while attending a public high school with advantaged types is - but both are swamped by the effect of going to private school, where (conditional on test scores bla bla bla) those attending non-religious private high schools have twice the probability of being admitted to elite colleges. Talk about a good investment, particularly because of the paper’s final finding: compared to diosyncratic variation in admissions for waitlisted applicants, the study finds that attending an Ivy Plus school (versus a shithole like Swarthmore or UC Berkeley) increases your chances of being in the top 1% of incomes by 60%, doubles your chances of a prestigious grad school placement, and triples your shot at a prestigious post-college career. So yeah, Barton is definitely worth it for the parents there.
Lastly, let’s look at Vietnam. Mary Lamb, the (Black) cafeteria worker who lost her son in the war, was planning on helping him go to Dartmouth, but he couldn’t afford it - but they talked about him attending after the war, with his G.I. Bill money. Was this plan likely to shake out? Well, firstly, because Vietnam service was assigned by lottery, it’s actually rather simple to verify: compare people who were in the lottery but didn’t get drafted to those who were in the lottery and did. Joshua Angrist’s 1990 paper “Lifetime Earnings and the Vietnam Era Draft Lottery: Evidence from Social Security Administrative Records” looks at this question, and finds that veterans drafted through the lottery had real lifetime incomes 15% lower than those of non-veterans who also participated in the lottery. A similar 2017 paper, titled “Unfortunate Sons: Effects of the Vietnam Draft Lottery on the Next Generation’s Labor Market” by Sarena Goodman and Adam M. Isen finds that children of Vietnam lottery draftees have lower wages and lower labor force participation, but are also more likely to volunteer for military service. Thirdly, a 2011 paper by Dalton Conley and Jennifer Heerwig titled “The War at Home: Effects of Vietnam-Era Military Service on Post-War Household Stability” finds mixed evidence for the hypothesis that Vietnam veterans had higher rates of divorce, higher rates of “moving around”, and lower odds of family cohabitation - some data sources find a positive relationship, and others a negative one. Lastly, a similar 2010 paper, but now from Argentinian data, titled “Conscription and Crime: Evidence from the Argentine Draft Lottery” by Sebastian Galiani, Martin Rossi, and Ernesto Schargrodsky (JPAL writeup here) uses a similar methodology, comparing lottery draftees to lottery non-draftees, to find that Argentinian military draftees had significantly higher crime rates than non-draftees - both for those who participated in the Falklands/Malvinas War, and for those who didn’t. This effect was larger for those in the Navy than for those in the Army or Air Force, which points to the idea that losing labor market experience is the causal channel - because Argentinian Navy conscriptions were two years long.
There is, however, one caveat. Different data sources, this time with a longer time frame (until the 2000s, rather than just the 90s) show a different conclusion. In “Long-term consequences of vietnam-era conscription: schooling, experience, and earnings” (2007) by Joshua Angrist and Stephanie Chen, the authors find that there is actually no difference in income between veterans and non-veterans after another decade or so versus Angrist’s 1990 study’s timeframe. The difference comes mostly from an increase in education, as a result of legal provisions that allow veterans subsidized access to colleges. However, there are significant health and disability consequences from serving versus non serving. So perhaps it’s good?
American Fiction - Color bind
American Fiction a bit of a mess thematically, but is really funny and has strong performances. It follows author Thelonious “Monk” Ellison, a frustrated middle class Black writer whose books are long, intellectual, and draw from Greek Mythology and French post structuralism. Tired of his lack of professional success, and with increased financial pressure, he writes a book called My Pafology to let off steam and mock the publishing industry. Compared to his cerebral doorstoppers, the book is “Blacker”, written in a tortured Black vernacular and featuring absent fathers, violent gangs, drugs, and police brutality. To Monk’s dismay My Pafology becomes an instant critical and commercial hit, and even makes waves in Hollywood.
So, in honor of a movie that asks “could it be bad to have all your stories about marginalized communities be based on arguably exploitative suffering porn” (same as The Zone of Interest, though noticeably more funny) , I’ll skip the inevitable papers about slavery or Jim Crow or discrimination or wealth inequality, and instead look at the power, for good or for evil, of names.
First, let’s look at a simple claim: do White and Black Americans have different names? And is this a recent phenomenon, as some authors claim? The paper “Distinctively Black Names in the American Past” (2013) by Trevon Logan, Lisa Cook & John Parman answers that Black naming has always been different from White naming. The authors base this on a novel database of census counts, and birth and death certificates infrom the early 20th century. The methodology was simple, and was based around two associations: conditional of being Black, which names are most popular? The second association was as follows: conditional on having a distinctively Black name, what are the person’s odds of being Black? These regressions were first run on a database of birth certificates and census questions from DC, Georgia, Michigan, and New York, and later verified with census questionnaires and death certificates from Alabama, Illinois, and North Carolina. Generally speaking, it was found that Black names were overall very distinctive from White names. A further paper by the same authors, “The Antebellum Roots of Distinctively Black Names” (2020) extends this methodology to pre-Civil War Black names, utilizing a number of data sources that were not available seven years prior. In detail, they utilize a 2000 dataset from Louisiana slaves, data on New Orleans slave sales, and data from slave shipments that came about through a quirk of US law, which banned the international, but not domestic, slave trade. These three datasets, put together and contrasted with the naming patterns of the exclusively White slaveowners included in those same datasets, independently confirm the initial findings: that Blacks have distinctive names versus Whites, and that Black names were distinctive from the get-go. All in all, this means that Black people have always had distinctive names as long as they have lived in the United States, rather than this being a more recent trend.
An interesting question, then, is if having a distinctively Black name implies a disadvantage relative to having a less racialized name - or, as Freakonomics put it, if being called Jamal Williams provides a disadvantage versus being the same person and called Derek Williams. The 2007 paper “The Causes and Consequences of Distinctively Black Names” by Stephen Levitt and Roland Fryer aims to answer this exact question. Because it’s not possible to compare a counterfactual like that directly, particularly since someone who changes their name from a distinctively Black one to a more “colorblind” name is scarcely representative of the general Black population, then something has got to give. The authors use data on all births in California, from the 60s to 2000, and pair it with data on the conditions of the family’s life: parental income and education, as well as the mother’s own life experiences. The authors utilize a similar probability association to find the most distinctively Black names. The authors find that, until the late 60s and early 70s, there were few differences in names between Whites and Blacks, with Asians used as a separate control group - however, the rise of what the authors term the “Black Power” movement resulted in an increase in popularity of Black names, and a decline in popularity of those names among Whites. Then, utilizing data on both future income and parental income, the authors try to determine whether individuals with names disproportionately given to Black people were less successful in the labor market compared to similar individuals without those names. Controlling for all determinants of family income, they found no difference: if Aliyah Smith and Molly Smith both had similarly wealthy and educated, were born in similar hospitals to similar mothers, and came from similar areas, then they would have similar job market outcomes. This does indicate that there might not be direct name-based discrimination. Or does it? A prior paper by Marianne Bertrand and Sendhil Mullainathan from 2003, titled “Are Emily and Greg More Employable than Lakisha and Jamal? A Field Experiment on Labor Market Discrimination” aims to answer the same basic question: does having a White name advantage people relative to having a Black name? To verify this, the authors sent identical fake resumes to Chicago and Boston employers, but with distinctly Black or White names. The study found was that resumes with White names had a 10.1% callback rate, while resumes with Black names had a 6.7% callback rate - that is, Emily and Greg are roughly 50% more employable than Lakisha and Jamal.
Past Lives - Going places
As of right now, Past Lives is the only Best Picture nominee I haven’t been able to see, but from reviews and trailers, the plot of the movie appears to be as follows: Nora moves from South Korea to Toronto at the age of 12, leaving behind her childhood best friend and crush of sorts, Hae Sung. As an adult living in New York with a husband, Nora reconnects with Hae Sung over the internet, and Hae Sung visits her - with, perhaps, underlying romantic tensions between the two.
Firstly, and interestingly, comes the question of whether moving countries like that is a good idea. While the conventional wisdom is that yes, it’s not really possible to compare people who emigrate to those who leave, since there are large differences regarding motivation, human capital, etc. But what if there was one migration that was random? While not international, the paper “The Gift of Moving: Intergenerational Consequences of a Mobility Shock” by Emi Nakamura, Jósef Sigurdsson and Jón Steinsson. The TL;DR of the paper’s empirical strategy is simple: in 1973, a long-dormant volcano erupted unexpectedly on the Westman Islands, off the coast of Iceland. This resulted in around a third of the population of the island of Heimaey to be forced to be relocated. This meant that, basically at random, a significant portion of the island’s population was forced to migrate elsewhere. Because Iceland keeps fairly comprehensive administrative data, the authors collected information on income and education for people who inhabited the Westman Islands before and after the volcanic eruption, as well as their descendants. Thus, they estimated the effect on income and education for Westmaners and their descendants using Instrumental Variables, where the instrument is the (assumed exogenous to income, education, etc.) destruction of homes during the 1973 eruption. The effects are straightforward: for those over 25 during the eruption, moving had slightly negative effects; however, for people under 25, it had largely positive effects on earnings and education. These effects get larger at higher incomes, too. The idea is simple: moving has high costs for individuals, but can also generate high benefits, especially for people not particularly well suited for their local labor markets.
A further question, and coming off the American Fiction section: at the beginning of the movie, Nora changes her name, from a Korean one to an English one. Was it a smart idea? The 2022 paper, “How Do You Say Your Name? Difficult-To-Pronounce Names and Labor Market Outcomes” by Qi Ge and Stephen Wu aims to answer this question. Using a sample of PhD candidates in economics. The authors start by collecting the resumes of graduates from the top 96 economics programs in 2016, 2017, and 2018. This gives fairly standardized information about each candidate. To ensure that the difficulty to pronounce a given name is held stable, the authors use different methods: difficulty rating by an algorithm, average time to pronounce, and a a subjective one from independent raters. The paper estimates the probability of the candidate being placed in academia, and in a tenure track position at that, given the pronounceability of their name, controlling for demographics, the professional characteristics of the graudate (institution, publications, experience), characteristics of their doctoral advisor, as well as fixed effects by region, field, undergrad institution, and year. The paper’s conclusions are that, controlling for all that, PhD job market candidates with harder to pronounce names have a 3% lower chance of being placed in academia, with no significant additional effect on tenure-track positions. These effects are largely consistent across definitions of “hard to pronounce”, but interestingly enough, neither the first nor last name are significant on their own.
Now, Nora went to school in Korea and, I presume, continued her education in Toronto (I haven’t seen the movie, disclaimer). This helps open an interesting line of questioning: what happens to kids when they take classes with immigrants? The paper “Diversity in Schools: Immigrants and the Educational Performance of U.S. Born Students” (2021) by David Figlio, Paola Giuliano, Riccardo Marchingiglio, Umut Özek & Paola Sapienza (summarized in Spanish here) seeks to answer this question. Because of mostly exogenous factors to the Floridian education system, in some years Florida has a large share of foreign students (of up to 91%) and some years it has a much smaller figure - around 6%. Using state-level data that matches school performance with birth records, the authors are able to put together a database of students and their relative exposure to immigrant, resulting in two estimates of the effect of immigrant presence on student performance. The first is that the same students may have different exposures to immigrant groups by year, based on the fact that the national composition of grades changes basically at random. This specification results in a small and significant negative effect on student performance - but this does not account for selection issues that depend on school “choice” or allocation. By this, the authors (and I) mean that it could be that immigrants tend to go to worse-performing schools, for instance due to having less information. A second possibility is that children from the same families can have different exposure to immigrant populations, for basically the same reason, allowing for a difference between the expected exposure to immigrant groups (based on the school or area average) and the actual exposure. This research design leads to the opposite conclusion, of a large and significant positive effect on performance, which is even larger for non-white and lower income students. A possible explanation is that immigrants tend to be “better behaved” than other students, particularly in disadvantaged schools, which might result in some spillover of their sigma grindset.
The Zone of Interest - Canoeing with Nazis
The Zone of Interest is one of the best and most disturbing movies of 2023. Set in the officers’ quarters of the Auschwitz death camp, it follows commander Rudolf Hoss, and his family in what can best be described as a series of mundane vignettes. Hoss and his wife, Hedwig, feud over a potential move to Berlin - due to Rudolf’s excellence at murdering large amounts of people. The family goes canoeing, only to be rudely interrupted by ashes from the camp’s crematoria pouring on the river. Hedwig looks after her garden, which is fertilized with human ashes. The children collect teeth as a hobby, read faintly antisemitic fairy tales, and play with the toys of murdered Jews.
Because it would be in extremely bad taste to even try to do a “the economics of concentration camps” bit, and because the movie focuses on the perpetrators and not the victims, let’s look at the Nazis: why did people join the National Socialist German Worker’s Party? The paper “War, Socialism and the Rise of Fascism: An Empirical Exploration” (2020) by Daron Acemoglu, Giuseppe De Feo, Giacomo De Luca, and Gianluca Russo develops a measure to track support from Italian Socialists, conservatives, and fascists in the run-up to World War One, as well as the activity of military veterans, landowners’ associations. The first finding of the paper is interesting: in 1919, casualties in World War One were uncorrelated to various economic or political variables, but were correlated with higher support for the Socialists, who were anti-war. Likewise, areas with higher support for the Socialists had higher fascist vote shares and activity (known as squadrismo), but not directly from Socialists - rather, from conservative and center-right elites, in a sort of “red scare”. Landowner associations and other such fora for right-wing elites were positively associated with Fascist support, but a higher number of war veterans was not correlated with fascism.
Now let’s look at the economic side of things. Contrary to popular wisdom, hyperinflation was not positively correlated with support for Nazism, but rather, with support for Communism - as shown in the paper “Spoils of War: The Political Legacy of the German hyperinflation” (2023) by Gregori Galofré-Vilà, linking city-specific price indexes and other economic indicators with vote shares by cities, and comparing the same localities across time. Another paper of the same year by the same author, titled “Scarring through the 1923 German hyperinflation” (ungated here) finds that, using a similar methodology, cities with more hyperinflation did have higher mortality rates due to deteriorating social conditions and lower standards of living. But there is an economic cause behind support for the NSDAP: austerity. The paper “Austerity and the Rise of the Nazi Party” (2017) by Gregori Galofré-Vilà, Christopher Meissner, Martin McKee and David Stuckler finds that there was consistently higher support for the Nazis in cities that enacted higher spending cuts, hiked taxes more, and had state and national transfers be reduced more deeply. Likewise, the 2021 study “Financial crises and political radicalization: How failing banks paved Hitler's path to power” by Sebastian Doerr, Stefan Gissler, Jose-Luis Peydro and Hans-Joachim Voth looks at which cities were most exposed to the failure of the Danatbank in 1931 using the stock performance of companies that had a Danatbank-linked hausbank (“house bank” - the one they did business with). Cities more exposed to bank failures had higher NDSAP vote share, especially for places with preexisting antisemitic attitudes.
One additional factor that bolstered Nazi support was social capital: membership in clubs, associations, and even sports leagues was linked to NDSAP membership and NDSAP electoral success. The paper “Bowling for Fascism: Social Capital and the Rise of the Nazi Party” (2013) by Shanker Satyanath, Nico Voigtlaender & Hans-Joachim Voth examines the membership rolls of 22,127 associations in 229 German cities (with populations over 10,000), as well as religious composition, economic variables, and occupational categories. Similarly, they look at NDSAP membership rolls and at Nazi Party support in the 1933 elections. There appear to be no systematic associations between association density or the controls, and neither appears to be the case for NDSAP rolls, meaning that selection bias can’t really be the culprit. Regressing NDSAP membership and NDSAP support by association density and the controls, they find a strong correlation between social capital and Nazism: cities with more active sports clubs, chess clubs, choirs, or veteran’s associations, among many others, were likelier to have literal card-carrying Nazis and to vote for them. Separating civic and military clubs results in no difference: both have a similar and significant association with being, or supporting, a Nazi. Looking at the timeframe of NDSAP signups, one can see that the association between the Party and clubs weakens over time, meaning that clubs and other organizations were vital in popularizing Hitler’s movement and ideas among Germans. Finally, looking at the relationship between political stability and social capital, the authors compare non Prussian2 German states with high political stability (Anhalt and Hesse, for example) with states with high political instability (such as Württemberg and Mecklenburg-Schwerin) to verify if the correlation changed. And it did: for stable states, there is no correlation between clubs and Nazis; in unstable states, it’s even stronger.
Poor Things - Tabulae rasae
Poor Things is a delightfully beautiful, very raunchy, and darkly comedic film about a woman, Bella Baxter, who commits suicide and is reanimated using an infant’s brain. She must forge her own path to freedom in a repressive Victorian context, which includes running off to Europe with a shady lawyer, doing sex work in Paris, taking a boring cruise, and then reencountering her old life in London.
The movie has, largely speaking, two themes: gender, and the idea of the blank slate, or tabula rasa in Latin. This is the notion that people come into this world without values, ideas, or skills, and gain them through nurture and experience. We must experience everything. Related to this are ideas about gender, which the movie isn’t particularly specific about, besides “rigid gender norms are bad”. While there might be something to this idea, background and environment is largely not chosen - as seen in “Childhood Environment and Gender Gaps in Adulthood” (2016) by Raj Chetty, Nathaniel Hendren, Frina Lin, Jeremy Majerovitz and Benjamin Scuderi. Using a comprehensive dataset of geolocated incomes, labor force participation, and demographic variables for people born in the 1980s, the study estimates gender gaps in major labor market variables. As I’ve pointed out before, the study finds that gender gaps in both labor market participation and wages are highest at the top of the income distribution, and lowest for low-income families. In fact, the gaps are reversed, but only for unmarried families: girls born in such families in the bottom quintile of earnings are more likely to be employed and make roughly as much, or more, as their coevals. Gaps also vary substantially across places: even poor families have “standard” gender differentials in the richest areas, but women have better outcomes than men in low income families in low opportunity areas. There are few consistent patterns, but the two key findings are that men experience a lot more variability and, at low levels of income, certain disadvantages, while women face obstacles in the top income strata. Childhood environment, however, is crucial. In fact, a similar paper by Raj Chetty, Nathaniel Hendren, and Lawrence Katz from 2015, titled “The Effects of Exposure to Better Neighborhoods on Children: New Evidence from the Moving to Opportunity Experiment” makes the exact same point. In this study, the authors follow the beneficiaries of the Moving to Opportunity experiment, where the US government offered a random subset of social housing residents a chance to move to low-poverty neighborhoods in the 90s. Tracking the participants, one can find that the children of families selected to live in better locations have better college attendance rates, higher earnings, better neighborhoods of residence as adult - but only if they are younger than 13. So perhaps the blank slate expires.
The themes of gender also manifest as explorations of sex - the movie is very explicit, particularly in the scenes involving prostitution. How does the market for sex work, then? The paper “The economics of sex work and major sporting events: Learning from the 2010 FIFA World Cup in South Africa”(2022) by Jacob Kazungu, Marlise Richter, Stanley Luchters, Matthew Chersich, and Matthew Quaifeb uses a survey of South African sex workers in 2010 to estimate the impact of market forces on the price of services offered. Because the large influx of foreign tourists raised demand, prices for sex increased, and the price premium for sexual acts without a condom also grew - but these effects were purely transitory, and were reverted after the World Cup. The movie also makes the case that the prostitutes have bad working conditions, with little ability to choose their clients and, in a deleted scene, medical risks and mistreatment. Could a different regulatory framework help? The 2014 paper “Decriminalizing Indoor Prostitution: Implications for Sexual Violence and Public Health” by Scott Cunningham and Manisha Shah examines the consequences on public health and sexual violence of a Rhode Island judge unexpectedly decriminalizing prostitution in the state. When the judge decriminalized sex work in Rhode Island, this allowed prostitution to move into the “formal economy”, and allowed a comparison between the state, which now had open advertisements for “adult services” publicly, and other states, where ads were placed in more discrete websites. Utilizing a differences in differences methodology, the authors found that decriminalization increased the number of advertisements for sex placed, but also that rates of sexually transmitted diseases among the general population, as well as a substantial decline in sexual assault.
But so, is openness to sex always a net positive? Not necessarily. A study titled “When You Can’t Tube…Impact of a Major YouTube Outage on Rapes” (2020) by Amelia Gibbons and Martin Rossi links a 2018 YouTube outage with high frequency crime data to find an increase in sex crimes in the United States. Because the outage was more or less random, the authors consider it causally identified, and instead move on to explaining why the increase in rapes happened: drug or alcohol consumption is not statistically significantly correlated to the outage, and the increase in sexual violence happened after the outage, meaning it wasn’t a “going outside” rise. Rather, the authors find, during the outage the traffic to porn website spiked, with porn consumption during the YouTube outage being significantly correlated to sexual assault. However, the results are not clear-cut: an older paper, titled “Pornography, Rape, and the Internet” (2006) by Todd Kendall finds the opposite result: using the staggered adoption of the internet across and throughout states as a source of exogenous variation, the study finds that (using differences-in-differences) having access to online adult content decreases rates of sexual assault in males 15 to 19, who had the starkest increase in availability vis-a-vis porn. However, studies from the late 90s may not be applicable anymore, since the consumption of porn itself may have changed sexual proclivities or attitudes towards consent over the long term. So the jury is out on sex work I guess.
This does not necessarily contradict the prior two papers, since Cook et al. point out that the Western United States had different historical experiences regarding Black populations, who were (largely) not enslaved, and instead voluntarily migrated.
Prussia is excluded because of its high resilience to Nazism and active anti-NDSAP maneuvers, as well as it being the most politically stable state in all of Weimar Germany.