Lecture 21: Endogeneity and Instrument Variables

TL;DR

Instrumental variables (IV) can be used to estimate causal effects when there is endogeneity or omitted variable bias in the data.

Transcript

[SQUEAKING] [RUSTLING] [CLICKING] ESTHER DUFLO: OK, so today I want to talk about endogeneity and one new method, which is the method of instrumental variables in two stage least squares. So after our nice interlude of machine learning where we didn't think about-- we didn't have to think about causality anymore, or even about coefficients because ... Read More

Key Insights

• 🤮 Instrumental variables can be used to estimate causal effects when there are concerns about endogeneity or omitted variable bias.
• ❓ The validity of the instrumental variable depends on the relevance and exclusion restrictions.
• 👻 Instrumental variable analysis allows researchers to estimate causal effects in situations where traditional methods may be biased.

Questions & Answers

Q: What are instrumental variables?

Instrumental variables are used in statistical analysis to estimate causal effects by exploiting sources of exogenous variation in the data. They can help address endogeneity or omitted variable bias.

Q: How do instrumental variables work?

Instrumental variables work by using a variable that is correlated with the treatment variable of interest but only affects the outcome through its impact on the treatment. It allows researchers to identify and estimate causal effects when traditional methods may be biased.

Q: When would you use instrumental variables?

Instrumental variables are used when there is concern about endogeneity or omitted variable bias in the analysis. They can be particularly useful in situations where randomized experiments are not feasible or ethical.

Q: What are the key assumptions in instrumental variable analysis?

The key assumptions in instrumental variable analysis are the relevance and exclusion restrictions. The relevance assumption is that the instrumental variable is correlated with the treatment variable, and the exclusion restriction is that the instrumental variable only affects the outcome through its impact on the treatment variable.

Summary & Key Takeaways

• Instrumental variables are used to estimate causal effects when there is endogeneity or omitted variable bias.

• IV requires a valid instrument that is randomized or as good as randomized.

• The instrumental variable estimate can capture the causal effect of the treatment on those who are compelled by the instrument to get treated.