Ndoanrobdoxnknai odbnnso Kahn so jdbdbbia sibndmoxux is banned isbsbbwo isjsnbdndndi six Shhbm is and she dbsbjs s ssix dish dishes sis do a Shs abks braid. Shah did
Applied micro is a joke field

Applied reg monkeys don't do nonparametric estimation. They run diffindiff, RDD, or maybe, if they're lucky, have some rainfall IV to run 2SLS.
DD is nonparametric, buddy. Again, that you guys don’t understand this is the problem.
The weaknesses of DD is that it’s model for the counterfactual may not hold. But the model is just a difference. In two periods this is a general functional form.DiffinDiff isn't nonparametric, at least not in the vanilla version popularized by Card & Krueger, and followed by millions of reg monkeys ever since. I mean, you're literally fitting a linear regression with a bunch of dummy variables. Who told you this was nonparametric?
What do you think nonparametric estimation is?

Applied reg monkeys don't do nonparametric estimation. They run diffindiff, RDD, or maybe, if they're lucky, have some rainfall IV to run 2SLS.
DD is nonparametric, buddy. Again, that you guys don’t understand this is the problem.
The weaknesses of DD is that it’s model for the counterfactual may not hold. But the model is just a difference. In two periods this is a general functional form.DiffinDiff isn't nonparametric, at least not in the vanilla version popularized by Card & Krueger, and followed by millions of reg monkeys ever since. I mean, you're literally fitting a linear regression with a bunch of dummy variables. Who told you this was nonparametric?
What do you think nonparametric estimation is?
... He's right...

DiffinDiff isn't nonparametric, at least not in the vanilla version popularized by Card & Krueger, and followed by millions of reg monkeys ever since. I mean, you're literally fitting a linear regression with a bunch of dummy variables. Who told you this was nonparametric?
What do you think nonparametric estimation is?
Nonparametrics means not parameterizing a functional form. If you model E(yx) = a + bx, then a and b are parameters describing a linear regression. In nonparametic econometrics you're not fitting a line through your entire data. You assume E(yx) = f(x), and you approximate f(x) local, say by knearest neighbors.
There are plenty of introductions to this available, so no need for me to explain this further. Check https://doi.org/10.1561/0800000009

it’s easy to see that DiD is non parametric as the model is saturated: there is one dummy variable for every possible combination of (treated, untreated) x (pre, post). you can use any nonparametric function f(D,T) you want and it will never be more flexible than DiD. think of a histogram, which is a nonparametric density estimation method. the most flexible histogram is one that has a bin for every possible value that X takes on.

it’s easy to see that DiD is non parametric as the model is saturated: there is one dummy variable for every possible combination of (treated, untreated) x (pre, post).
A regression with essentially four dummies simply splits the sample in four and calculates means. Whether you consider it flexible or not, this isn't what people call nonparametric.