The structural and experimentalist approaches are similar in that both seek to identify quantitative impacts of policy changes. In this way, they differ from the descriptive approach that can make no such claim to identification. Papers the descriptive approach usually begin with some discussion of economic theory and empirical implications that can be derived from the theory, the latter typically in the form of correlations or monotonicities we expect to observe when one variable is plotted against another. The signs and statistical significance of empirical relationships are often the main focus of the empirical analysis (as opposed to the magnitude), as the goal is usually to discriminate across alternative theories that differ in the signs of comparative statics relationships. Regression parameter estimates obtained in this style of work are typically referred to as reduced-form estimates, as opposed to underlying structural parameter estimates.
这是原文。谢谢楼上各位!