What is reduced-form analysis?

Ask an Economist
Question: 

I wonder what "reduced-form analysis" is, and a non-reduced form would be. To be more specific, I'm reading an article (www.clevelandfed.org/research/review/1996/96-q1-craig.pdf) which states the following:

"Economic data usually influence policy through a reduced-form analysis...Explicit assumptions about behavior that underlie the relationship are not emphasized; rather, the researcher asserts that the “data do the talking.”"

The author later specifies what he means by "reduced-form":

"We use the term in a wider context, where the pattern in the data — not an assumed behavioral structure — forms the point of departure for estimation."

Answer: 

A “reduced-form” analysis, also often referred to as “non-structural” analysis, is the most common kind of econometric analysis performed by economists. The other kind, which you called “a non-reduced form,” is customarily referred to as “structural” analysis.

When conducting a non-structural (reduced-form) analysis, an economist uses economic theory, intuition, and other considerations (e.g., prior research results) to specify an equation describing hypothesized relationships between a dependent variable of interest and some explanatory variables. The equation is then estimated using econometric methods to see if the relationships are borne out by the data. While relatively simple to implement, non-structural analyses have a disadvantage in that they do not allow economists to rigorously quantify the effects of possible changes in public policies on individual or aggregate behavior.

In the case of a structural analysis, an economist formulates an economic model to explain some observed behavior, chooses specific functional forms for the model’s components (e.g., a particular form of a consumer’s utility function or a firm’s cost function) and explanatory variables to include, and typically estimates the model as a whole. Structural econometric analysis is usually more difficult to implement than the non-structural approach and requires more computing power and programming skill. However, it enables the simulation of the effects of alternative public policies and other hypothetical scenarios.

If you would like to compare and contrast these two kinds of analysis as used in actual studies, please see the following classical articles:

(Non-structural:) Angrist, Joshua D., and Alan B. Krueger. Does Compulsory School Attendance Affect Schooling and Earnings? Quarterly Journal of Economics, Vol. 106, No. 4 (Nov., 1991), pp. 979-1014; www.jstor.org/stable/2937954

(Structural:) Rust, John. Optimal Replacement of GMC Bus Engines: An Empirical Model of Harold Zurcher. Econometrica, Vol. 55, No. 5 (Sep., 1987), pp. 999-1033; www.jstor.org/stable/1911259