Research Papers

Different conceptions of merit are disambiguated to consider whether higher education in the United States is currently or could plausibly become meritocratic. We argue that contemporary societies are not meritocratic in the commonly popularized sense, and that because society is not meritocratic, college admissions could not be either––even if reformed to reward only grades and test scores.

Institutions of higher education have been criticized across the political spectrum for failing to be appropriately meritocratic, with recent demands from the federal government that elite universities reform admissions to be more merit-based. Yet it is not entirely clear what is meant by merit. Answering that question requires a more robust understanding of merit and meritocracy than public debate currently offers.

Calls for meritocratic admissions demand rewarding developed abilities, not native ones, which is impossible while the family exists in its current form. Even a nonmeritocratic society can produce harms described by critics because those harms arise from the false belief that society is meritocratic. A belief that reforming admission around “merit” risks deepening rather than correcting this problem.

We examine emotions children encounter in educational materials. Using LMMs, we classify eight emotions across text and images in three decades of California and Texas public school textbooks and a century of award-winning children’s books. We examine heterogeneity by context and character identity, validate results against non-generative methods and human annotators, and merge book emotions with purchasing data.

Emotional intelligence is a key component of human capital, yet little is know about childhood inputs (e.g., textbooks and children’s media) that shape emotional norms. It is not obvious ex ante whether educational materials expose children to broad emotional range or reproduce narrower, socially rewarded displays, and evidence on emotions across text and images together is limited.

Text exposes children to a broad emotional range, but images overwhelmingly show happiness and calm regardless of the text on the same page. Purchasers are less likely to choose books with “negative” emotions on their covers, suggesting market forces narrow the emotional landscape. Results point to the value of more intentional emotional diversity in children’s materials.

We construct a framework for measuring distributional between two groups that compares their full distributions directly, letting researchers build normative judgements into the measure. Defining inequality as the smallest cumulative effect of transfers to a state of equality, we derive tractable measures via optimal transport and apply them to the gender wage gap using 2005-2024 ACS data.

Researchers typically summarize group distributions with a mean or quantile and compare those, but this ad-hoc approach can obscure important dissimilarities between distributions and lacks principled grounding. No existing framework directly compares full distributions while allowing researchers to build their own normative judgments about inequality into the resulting measure.

Difference between group means can mask important differences between distributions and is not a valid measure of distributional inequality in general. For Black men and women, the CDFs cross throughout 2005-2024, and the measure exceeds the difference in means. Varying the sensitivity parameter emphasizes different parts of the distributions, extending to racial and ethnic disparities in health and education.

This paper examines how workers reallocated as new technologies destroyed old jobs and created new ones during Britain’s Second Industrial Revolution. Using 170 million full-count British census observations (1851-1911), I construct new task-level occupational data and study English bootmaking as it mechanized, tracking entry, exit, and mobility across occupations.

Despite widespread technological change during Britain’s Second Industrial Revolution, we know little about how workers actually reallocated as old jobs disappeared and new ones emerged. Existing accounts rarely trace individual-level movememnt between occupations at this scale, leaving open whether displaced workers retrained into new roles, exited the labor force, or were replaced by new entrants altogether.

Understanding how workers reallocate amid technological change is newly urgent given generative AI. I propose two forms of obsolescence: swift, comprehensive shocks displace incumbents, while slow-diffusing technologies let incumbents stay, shifting adjustment onto the flow of new workers. Early evidence suggests generative AI operates on this second margin where entry–not incumbency–is what absorbs disruption.

We develop a life-course model distinguishing timing, accumulation, and sequencing of exposure to concentrated poverty from adolescence through midlife. Using Add Health data, we introduce a novel class of marginal structural models parameterized via principal components analysis using residual balancing weights, to estimate age-specific effects on self-rated health, BMI, CRP, and uncontrolled hypertension.

Prior research rarely test long-arm, contemporaneous, and cumulative-disadvantage perspectives because it typically measures neighborhood conditions at one time point or uses restrictive duration-weighted averages. Fully flexible stage-specific models suffer from collinearity since exposures are highly correlated over time. This mismatch between theory and method leaves unclear when neighborhood poverty matters most.

Efforts to deconcentrate poverty and promote moves to opportunity have produced uneven results. Interventions targeted at a single stage, or that fail to alter long-term exposure from childhood through midlife, are unlikely to fully mitigate health effects. The largest health improvements are likely to come from policies preventing prolonged exposure to high-poverty neighborhoods from childhood onward.

We study how much an ancestor’s social status in 19th-century China shaped the fortunes of descendants. Using a population dataset from Liaoning province (CMGPD-LN), we track families across six generations. We apply Markov chain models and measure mobility in two ways, addressing how long the memory of family origins shapes descendants’ outcomes, and how often sons moved into a different class than their fathers.

Late Qing China is a unique historical period, encompassing eras of relative stability as well as conflict and political crisis, making it a rare test case for whether crises reshape ordinary families’ chances of improving or losing social standing. The CMGPD-LN registry’s six generations let us observe five transition matrices, allowing us to model mobility as a nonhomogeneous Markov chain that evolves over time.

Most of what we know about intergenerational mobility comes from stable, modern economies. Studying how mobility evolves under upheaval, over a long historical horizon, broadens our understanding of mobility itself. Our findings echo existing theories linking declining upward mobility to political instability, leaving open a number of questions about the underlying mechanisms for future research.

We use a randomized controlled trial in Title X clinics to evaluate how eliminating out-of-pocket contraceptive costs affects reproductive outcomes among uninsured U.S. women. By randomly assigning vouchers for free contraception, we estimate the effects on contraceptive use, pregnancies, births, and abortions.

Research links unintended pregnancies to diminished outcomes for women and children. While RCTs have been done on this topic for low- and middle-income countries, evidence for high income countries relies on uncontrolled or quasi-experimental research designs. Data constraints limit the ability to estimate the magnitude of treatment effects following policy changes regarding contraceptive or pregnancy choices.

By providing causal evidence on how financial access to contraception affects reproductive outcomes among uninsured women, this study deepens our understanding of the role cost plays in contraceptive use, pregnancies, and pregnancy outcomes. The findings contribute to broader discussions about reproductive health access and the effects of reducing financial barriers to care.

This article defines meritocracy as three linked principles (hiring by merit, fair equality of opportunity, and fair reward), then tests it against three major critiques: that talent and family make advantage undeserved, that “merit” is an elite construct, and that meritocracy humiliates low-skilled workers. It uses conceptual and philsophical arguments, drawing on empiricial studies where relevant.

Academics have attacked meritocracy in a wave of publications, with critiques often blurring together. It’s unclear whether meritocracy is judged as unacheivable or undesirable, and which flaw is to blame. Critics also rarely reject it outright, while the general public remains quite popular.

Meritocracy is popular with politicians and the general public but fiercly attacked by academics. I argue that critics risk overlooking that meritocracy, rightly understood, can itself serve as a pwoerful critique of the unjust inequalities they highlight.

I examine two justifications for market meritocracy: that it justifies unequal outcomes through equality of opportunity, and that it builds an elite whose skills serve the common good. It tests each on its own terms, using evidence on unequal educational investment, elite admissions, wages, and skill-biased technical change.

The familiar objection to meritocracy is that meritocrats don’t deserve their talents, training, or networks. Less familiar is meritocracy’s own role in blocking fair equality of opportunity, and the failure of the argument that it creates an elite whose quest for private advantage serves the public good, which is not commonly appreciated.

Meritocracy is self-defeating: its successes undermine the values that give it its appeal. Reforms that purify economic and social life of aristocratic elements will not seriously reduce the inequalities that lie at the root of populist discontent. Repairing meritocratic inequality requires a rejection of meritocracy itself, and institutions must remake themselves in a fundamentally more democratic image.

This article distinguishes between “thin” and “thick” meritocracy. Thin meritocracy focuses exclusively on fairness at the point of selection—neutral rules and race-blind procedures. Thick meritocracy retains procedural neutrality while also addressing the formative institutional and relational conditions under which merit is cultivated.

The meritocratic idea as commonly invoked risks becoming a convenient totem for a political ideology that legitimates inequality by seeing social outcomes as due to individual qualities alone Typical debates on meritocracy focus on fairness in selection, but ignore fairness in formation, overlooking the conditions under which such qualities are developed and evaluated.

In a racially stratified society, exclusive reliance on race-blind selection risks legitamizing inherited disadvantage. A defensible meritocracy must be race conscious in diagnosis, race neutral in allocation, and relational in method. Meritocracy remains indespensable, but cannot stand alone.