Judicial Behavior under Political Constraints: Evidence from the European Court of Justice
Judicial Behavior under Political Constraints: Evidence from the European Court of Justice (DOI: https://doi.org/10.1017/S0003055408080350) was written by Clifford J. Carrubba, Matthew Gabel, and Charles Hankla in 2008. It was published in American Political Science Review (vol. 102, no. 4).
Courts often rely on executive/legislative branches for enforcement of decisions. Conversely, they can be constrained by executives (e.g., noncompliance) and legislatures (e.g., overriding decisions by changing the law). Threats to constrain may also be effective.
The authors analyze this through the ECJ. A predominant theory is that the ECJ has an interest in ruling contrary to national governments so as to advance the EU complex (neo-functionalism).
They start with the set of cases from 1987-1997 and code each policy space as a distinct row. They indicate whether a national government is a litigant, and whether they submitted a brief (observation) in support/opposition. (All EU institutions and member state governments are entitled to submit observations to the ECJ before they issue a ruling.) Formally, they look for whether an observation supporting the defendant (plaintiff) increases the feasibility of overriding. They fit a probit model, regressing on...
- a weighted sum of observations supporting the defendant (model 1).
Weighted with respect to qualified majority rules.
- Results are similar with unweighted sum.
- a weighted sum of observations supporting the defendant and an indicator for if the defendant is a member state, plus an interaction effect (model 2).
Controls for whether the Commission is a litigant and the size of the judicial chamber (because ECJ may delegate to a small plenum).
Clustered standard errors for each case.
In an additional model, they additionally regress on an indicator for if the ruling part of Article 234 proceedings, plus interaction effects. This model suffers from colinearity however.
Lastly, they attempt to control for the objective validity of the ruling. Their proxy for this is the attorney general's observation.
The first model detects a positive effect, and the second less so. Suggests that noncompliance is a stronger threat than overriding.
Reading Notes
The authors say that the third model suffers from colinearity, so standard errors are inflated. My understanding is that a linear model is plain unusable in this case?