Great paper
New papers destroys Top 5 applied micro papers falsely invoking LATE
-
Ok, paper makes valid points along the lines of the recent literature on treatment effect heterogeneity. The problem with this recent econometric literature is the following:
"We apply the decomposition to four empirical analyses and find strong evidence that the LATE interpretation of TSLS is far from accurate for the types of specifications actually used in practice."
That is, take a bunch of cases, apply a proposed solution/diagnostic test, claim that I am right and they are wrong. If instead you run a true monte-carlo to check the reliability of proposed solutions/diagnostic tests in this literature, you find that traditional methods are outperformed/correctly rejected only in very specific cases, and that most often than not a few adjustments to the standard specification that still rely on traditional estimation methods provide a reasonable solution also for these specific cases