Economists use reduced-form and structural approaches to address questions or model various aspects of economic phenomena.
Reduced-form models focus on the effects of causes, explaining relationships between factors without delving into the exact reasons behind those connections. This approach often uses regression. For instance, imagine studying how education affects income. A reduced-form approach may involve comparing education levels and income without exploring the specific ways education impacts income, such as through skills or job market dynamics.
Specific methods—like Random Control Trials (RCT) or techniques such as Difference-in-Differences (DID), Regression Discontinuity (RD), and Instrumental Variables (IV)—also operate within this reduced-form approach. They serve as tools designed to uncover causal relationships. These methods have gained significant popularity among economic researchers, particularly after the 2021 Nobel Prize, dubbed ‘the revolution of causality. These approaches have been pivotal in advancing our understanding of cause and effect in areas like labor economics and policy evaluation. However, while isolating and estimating causal effects, they don’t fully elucidate the underlying mechanisms.
On the other hand, structural models aim to explicitly model the causes, effects, and underlying mechanisms or causal relationships between economic variables. These models focus on understanding the processes that drive observed outcomes. They aim to uncover the fundamental drivers of economic phenomena. They are powerful but can also be complex and may require strong assumptions. Examples include:
a. General Equilibrium Models: These models aim to understand the economy by simulating interactions between various markets, agents, and institutions. They explicitly model behaviors, preferences, and constraints of different economic agents to understand how changes in one part of the economy affect others.
b. Dynamic Stochastic General Equilibrium (DSGE) Models: These models incorporate time dynamics, uncertainty, and various shocks to model how individuals and firms make decisions over time in response to changes in economic conditions.
Choosing between reduced-form and structural approaches often depends on the research question, available data, and the level of understanding sought. Reduced-form models are simpler and more data-driven, while structural models aim for deeper insights but might require stronger assumptions and more complex analyses. If you have a dataset that perfectly suits a reduced-form approach to address the specific questions, adding more structure might weaken the results. However, such ideal datasets are rare occurrences.
Reduced form example: The Nobel Prize winner 2021, Angrist, authored the paper “Does Compulsory School Attendance Affect Schooling and Earnings?” In this work, Angrist and Krueger adopted a reduced-form framework to isolate the causal effect of compulsory schooling on educational outcomes and earnings. They leveraged differences in state-level laws regarding the age at which students must attend school.
For instance, they compared individuals who were just on either side of the age cutoff when the law changed. Those just above the threshold might have had to stay in school longer, while those just below it didn’t face the requirement. This setup enabled them to analyze differences in outcomes, such as education level or earnings, between these two groups, attributing any disparities to the impact of the law itself rather than other factors. The results give valuable insights into the long-term effects of educational policies on individuals’ lives.
Structural Form Example: Arcidiacono’s study, “Affirmative Action in Higher Education: How Do Admission and Financial Aid Rules Affect Future Earnings?” uses a structural model to analyze various decisions, including application submissions, school selection, and field of study choices. The model encompassed individual decisions and institutions’ decisions regarding student acceptance and financial aid. Through simulations, it revealed that race-based advantages minimally influenced earnings but significantly impacted educational outcomes for black students. Removing admission advantages reduced black enrollment in top-tier schools while removing financial aid advantages decreased overall black college attendance.
Regarding the implications of affirmative action, the paper suggests that it might not significantly affect black students’ income and could potentially lead to unintended consequences. Students with lower grades entering elite schools might struggle academically, opting for less challenging majors with lower returns. This phenomenon could result in poorer academic performance than if they had attended other institutions.
Ultimately, both empirical methods stand as pillars of significance within economic research. Their distinctiveness lies in addressing diverse analytical needs and accommodating varied data structures without a definitive declaration of superiority. Instead, each method uniquely bolsters our proficiency in effectively analyzing questions and navigating complex data structures.
References and Further Reading
Angrist, J. D., & Krueger, A. B. (1991). Does Compulsory School Attendance Affect Schooling and Earnings? The Quarterly Journal of Economics, 106(4), 979–1014. https://doi.org/10.2307/2937954
Arcidiacono, P. (2005). Affirmative Action in Higher Education: How Do Admission and Financial Aid Rules Affect Future Earnings? Econometrica, 73(5), 1477–1524. http://www.jstor.org/stable/3598881
Rust, J. (1987). Optimal Replacement of GMC Bus Engines: An Empirical Model of Harold Zurcher. Econometrica, 55(5), 999–1033. https://doi.org/10.2307/1911259
Rosenzweig, Mark, R., and Kenneth I. Wolpin. (2000). Natural “Natural Experiments” in Economics. Journal of Economic Literature, 38 (4): 827-874. https://www.jstor.org/stable/2698663