Di Zhou

This paper investigates the portrayal of economic action within popular American self-help books. By employing a computational, mixed-method analysis of best-selling titles from the New York Times over the past five decades, we explore self-help’s “promissory discourse”—that is, which actions readers are told will lead them to worldly success. Our findings reveal significant shifts in prescribed economic action, with a decline in "financialized" behavior and investment-focused advice, particularly following the Great Recession. Instead, books increasingly emphasize a “therapeutic” and self-oriented perspective, advising readers that introspection, emotions, and practices on the self are essential (and a pre-requisite) to economic success. These trends hold both across the universe of economic self-help books and within financial bestsellers. These findings expand our understanding of the transformation of finance culture, demonstrating—at least within popular financial advice books—a transition from hyper-rationalized, calculative investment behavior towards an increasingly therapeutically-inflected, self-oriented economic actor.

Social class indicates individuals’ access to economic resources and is key to understanding inequality. Current studies of social class often measure it using aggregate occupational groups. However, this measure of class may encounter the problem of using a fixed occupation-to-class mapping that neglects the important changes occurring within occupation due to technological change and organizational reform that may change an occupation’s class location over time. This study addresses this limitation by introducing a new “task-based class identification” model to evaluate an occupation's class location using text data and supervised machine learning based on Marxist class theory. Drawing on ONET’s occupational task data, I assessed class locations for detailed occupations in 2002 and 2020, linking this information to CPS surveys to map the U.S. labor force's class structure. Results reveal that while the American class structure has remained stable across four aggregate classes, significant shifts occurred within these categories. Particularly, within the non-managerial employee class, between 2002 and 2020, the share of proletarian workers declined by 8%. Decomposition analysis attributes 30% of this decline to changes in occupation size and 70% to within-occupation shifts, where proletarian workers experienced either “upskilling” or granted new supervisory tasks resulting in an upward class shifts.

2025 Erik Olin Wright Prize
2025 Robert D. Mare Graduate Student Paper Award (Honorable Mention), ASA Section on Inequality, Poverty, and Mobility
2025 Aage Sørensen Award, ISA Research Committee 28 on Social Stratification and Mobility

This study compares the neo-Weberian and neo-Marxist class frameworks to examine how theoretical choices shape our understanding of the American class structure from the 1970s to the 2020s. Using the Current Population Survey data, I apply the Erikson-Goldthorpe-Portocarero (EGP) class schema and an innovative task-based operationalization of Wright's contradictory class locations schema to map changes in the American class structure across five and a half decades. The two frameworks yield divergent portraits of the class map. The neo-Weberian schema shows a top-heavy shift driven by the expansion of the service class, while the neo-Marxist schema reveals growth concentrated in contradictory class locations – particularly managers and semi-autonomous supervisors – rather than at the top of the class hierarchy. The findings demonstrate that theoretical and measurement choices have substantial consequences for how we characterize the class structure and its trajectory.

A large body of research has documented the polarization of the American occupational structure, yet this literature relies almost entirely on skill-based classifications that say little about workplace power: who controls the labor process, who exercises authority over others, and who has voice in organizational decisions. This paper addresses that gap by examining the geographic distribution and economic fate of the American working class, defined through a relational class framework as those lacking work autonomy, supervisory authority, and organizational decision-making power, with a focus on how the transition to a knowledge economy has reshaped working-class employment and prospects across metropolitan areas from the early 2000s to 2020. Using large language model classifiers applied to occupational descriptions and merged with establishment-level employment surveys, we identify the class, occupational, and skill composition of the workforce across more than 300 metropolitan statistical areas (MSAs) from 2004 to 2020. We report three main findings. First, metropolitan areas experienced an average 4.8 percentage point decline in working-class employment share between 2004 and 2020, driven by the disappearance of middle-skill working class jobs and accompanied by a downward shift in the skill composition of remaining working-class occupations – a process of class-based deskilling. Second, cross-classifying MSAs by working class concentration and non-college employment rates reveals four distinct MSA types. Finally, to understand what leads to the flourishing of working-class employment and concentration, we modeled four dimensions of working-class prosperity: working-class concentration, non-college employment rate, median hourly wage, and a composite flourishing index. Results show that, within MSAs, the presence of highly-educated workers benefits working-class employment rates, has null effect on working class wages, while decreases the concentration of the working class. This finding point to a structural tension between the material gains of individual workers and the structural condition for working class collective action.