While reduced-form methods have gained popularity due to their simplicity and clear causal interpretation, structural methods remain an essential although often overlooked tool in the economist's toolkit.
Structural methods involve building economic models that explicitly specify the mechanisms through which variables interact. These models are then taken to the data for estimation and testing. Unlike reduced-form approaches, which focus on identifying causal effects without specifying the underlying economic mechanisms, structural methods aim to capture the full economic environment.
Why use structural methods?
- Policy Counterfactuals: With a structural model, we can simulate the effects of multiple policies that haven’t been implemented yet, or explore alternative policy designs and intensities.
- General Equilibrium Effects: With a correctly specified structural model, one can address large scale variations in an economy that affect equilibria.
- Understanding Mechanisms: Structural models allow us to understand not just that an effect exists, but how and why it occurs in an economic framework. This deeper understanding can be crucial for policy design and implementation.
- Welfare Analysis: Structural models often allow for a more comprehensive welfare analysis, considering direct and direct effects of a policy on different subsets of the population.
- Dynamics: When studying phenomena that evolve and interact over time, structural models can incorporate dynamic behavior.
Structural and reduced-form methods are not mutually exclusive. The opposite! Modern empirical work often combines the strengths of both approaches:
- Model-Guided Reduced-Form Analysis, or conversely Reduced-Form Guided Model Analysis: Economic models can suggest reduced-form relationships to estimate; conversely, reduced-form analysis can motivate and guide structural model analysis.
- Credible Identification for Structural Parameters: Reduced-form techniques such as regression discontinuity designs or instrumental variables can be used to obtain causal estimates of crucial parameters in structural models.
- Reduced-Form Validation of Structural Models: One can compare estimations from reduced-form exercises and simulations performed through structural models to check their validity.
Examples:
- Education Markets and School Choice: Consider the impact of school choice policies such as voucher programs. Reduced-form approaches can successfully estimate the immediate effects on test scores or graduation rates for students in “treated” areas. However, a structural model can capture the complex dynamics of the education market more broadly and precisely:
- Parental decision-making process - including the weights on factors like school quality, distance, peer composition (for the latter, one would need an equilibrium model)
- Schools’ strategic responses
- Equilibrium effects on housing markets
- Counterfactuals with different amounts granted for vouchers, or different allocation systems.
- General Equilibrium Effects in Trade Policy: Consider a tariff policy. A reduced-form analysis effectively compares snapshots of certain measures in different points in time. It could analyze the direct impact on domestic prices or employment in affected industries, for example. However, a structural approach can capture broader economic implications:
- Shifts in labor allocation between industries
- Effects on exchange rates and the balance of trade and other interconnected effects
- Overall welfare impact.
That’s powerful. Happy modeling!
