18: Instrumental Variables Flashcards
What is the purpose of using instrumental variables in regression analysis?
To solve endogeneity issues in regression coefficients.
Endogeneity can arise from omitted variable bias, reverse causality, and measurement error.
What are the four sources of endogeneity?
- Omitted Variable Bias
- Collider Bias
- Reverse Causality (Simultaneity)
- Attenuation Bias – measurement error
Each source can lead to incorrect regression coefficients.
True or False: Random error in the dependent variable generally affects standard errors in regression.
False.
Random error in the dependent variable does not pose a problem; it is random error in independent variables that can lead to attenuation bias.
What is an influential outlier
Strong effect on pattern and direction of the relationship
What is a non-influential outlier?
An outlier that has a weak effect on the pattern and direction of the relationship.
Non-influential outliers do not significantly alter the results of regression analysis.
How can outliers be identified?
- Scatter plots
- Data entry review
- Quantitative measures
Outliers are extreme values that do not follow the expected relationship.
What are two popular techniques to identify causation
- Instrumental Variables Approach
- Difference in Difference
Both can manaage OMV bias, attenuation biass and reverse causality - cannot necessarily solve collider bias
What do instrumental variables need to be?
Instrument must be strongly correlated with the endogenous variable
The instrument cannot be theoretically related to the dependent variable except through the endogenous variable.
This principle is crucial for the validity of the instrumental variable.
Fill in the blank: The _______ approach helps economists manage omitted variable bias and reverse causality.
[Instrumental Variables]
It is one of the popular techniques for identification in econometrics.
What is the significance of settler mortality in AJR’s instrumental variables approach?
It serves as an instrument for extractive colonial institutions that affect modern income.
High settler mortality is correlated with poor institutions, influencing GDP per capita.
What is a two-stage regression?
A method where the first stage regresses exogenous variables against the endogenous variable, and the second stage uses predicted values from the first stage.
This method helps to isolate the endogenous part of the variable being studied.
How does instrumental variables approach work
- Find a version of x which is exogenous and strongly correlated with it
- Regresss endogenous variable x on the instrument variable; gives you clean predicted part of based on instrument
- Regresss y on the predicted x, gives you clean values as exogenous x
What is are criticisms of the instrumental variables approach?
- Good instrumental variables are very rare.
- Collider Bias can arise from bad proxy variables
- In practice instrument may influence outcome bia unaccounted for channels (could be controlled through conditioning on the conounder)
The validity of the instrument is crucial, and poor measurement can lead to biased results.
What is the concern with cross-country regressions?
- They reduce units of analysis to the country, which may not reflect historical realities.
- Countries vary hugely in size but weighted equally in regression
- Don’t have a lot of observers so prone to outliers.
Historical borders and influences can vary greatly, impacting the validity of results.
What does Austin (2008) argue about long-run cross-country regressions?
They compress history and downplay historical contingency.
The significance of institutions may change over time, affecting economic outcomes.
Need to study how effects change over time
What is the relationship between institutions and economic outcomes according to North (1991)?
Institutions are humanly devised constraints that structure political, economic, and social interaction.
Examples include government types and property rights security.
