Applied Micro-econometric Techniques
2024-2025
Utrecht University
, Applied Micro-econometric Techniques
Topic 0: Introduction
What is this course about?
- Cause-and-effect relationships
Questions we ask:
- What is the effect of price on sales
- How do marketing campaigns affect sales
- How do business strategies affect returns
- How do active labour market policies affect participants
- How does trade with China affect Dutch labour market?
- How does the introduction of robots affect firm productivity
The Gold Standard:
- This term refers to methods or approaches considered the most reliable and
accurate for establishing causal relationships
- A Randomized experiment is often regarded as the “Gold Standard”
- Some reasons why (added content)
o Random assignment
To treatment and control groups
o Control of confounders
Balances observed and unobserved characteristics
o Clear counterfactuals
Control group represents what would have happened to the
treatment group in absence of treatment
- Yet, this is often infeasible in economics and business.
In this course we focus on experiments and quasi-experiments
- Natural experiments: assignment criterion occurs ‘naturally’ (without researcher
intervention)
- Quasi-experiments: criterion for assignment is selected by the researcher
Position in the program
- Regression
- ADAVE I and II looks at correlation and prediction
o Focuses on Y and Ŷ
- AMT looks at causal relationships between β and β^
o We disregard statistics like R2 in causal analysis
o We are more concerned whether our research design provides a credible
estimate of our population parameter
2
, Applied Micro-econometric Techniques
Topic 1: Regression
1. Correlation versus causality
Correlation does not imply causality
- Left panel shows a correlation between US spending on science and suicides.
Even though there is a correlation, it doesn’t necessarily imply a causal effect of
increased spending on suicides
- Right panel shows a less close correlation of Japanese cars sold in the US and
suicides by the crashing of motor vehicles.
- We call this spurious correlation
A lack of correlation does not imply lack of a causal effect
- Example: mandatory face masks in public transport in NL from June 2020
o No apparent change in COVID-19 cases, even an increase in the autumn of
2020
- Concluding question: Do face masks have an effect on less COVID cases?
o No: we do not know what would have happened had there been no rule to
wear masks
o There is no clear counterfactual
3
, Applied Micro-econometric Techniques
Vaccinations:
- No clear correlation between vaccine rates and infection numbers (fluctuates
positively and negatively)
- Can we conclude vaccinations have no effect?
o No: We do not know what would have happened if there had been no
vaccinations
o Further studies show vaccinations are effective. It’s just that other things
happen simultaneously.
Threats to the identification of causal effects
Reverse Causality
- Example: Middle Ages
- Europeans believed lice to improve health
- Reasoning: They observed that sick people do not have lice, whereas healthy
people do
o No lice sick
- However, causality is reversed
- Lice are sensitive to body temperature and leave sick hosts
o No lice fever
Selection bias and Omitted variables
- Example: health status of people who have (not) been hospitalized) in the past
12 months
- Do hospitals make people sick? E.g. due to germs etc? Not necessarily
- Alternative explanations
o Selection bias
o Omitted variable bias
- Example: a study comparing hospital visits and health status might miss that
sicker people are more likely to visit hospitals (selection bias)
Summary of Causal relationships
4
2024-2025
Utrecht University
, Applied Micro-econometric Techniques
Topic 0: Introduction
What is this course about?
- Cause-and-effect relationships
Questions we ask:
- What is the effect of price on sales
- How do marketing campaigns affect sales
- How do business strategies affect returns
- How do active labour market policies affect participants
- How does trade with China affect Dutch labour market?
- How does the introduction of robots affect firm productivity
The Gold Standard:
- This term refers to methods or approaches considered the most reliable and
accurate for establishing causal relationships
- A Randomized experiment is often regarded as the “Gold Standard”
- Some reasons why (added content)
o Random assignment
To treatment and control groups
o Control of confounders
Balances observed and unobserved characteristics
o Clear counterfactuals
Control group represents what would have happened to the
treatment group in absence of treatment
- Yet, this is often infeasible in economics and business.
In this course we focus on experiments and quasi-experiments
- Natural experiments: assignment criterion occurs ‘naturally’ (without researcher
intervention)
- Quasi-experiments: criterion for assignment is selected by the researcher
Position in the program
- Regression
- ADAVE I and II looks at correlation and prediction
o Focuses on Y and Ŷ
- AMT looks at causal relationships between β and β^
o We disregard statistics like R2 in causal analysis
o We are more concerned whether our research design provides a credible
estimate of our population parameter
2
, Applied Micro-econometric Techniques
Topic 1: Regression
1. Correlation versus causality
Correlation does not imply causality
- Left panel shows a correlation between US spending on science and suicides.
Even though there is a correlation, it doesn’t necessarily imply a causal effect of
increased spending on suicides
- Right panel shows a less close correlation of Japanese cars sold in the US and
suicides by the crashing of motor vehicles.
- We call this spurious correlation
A lack of correlation does not imply lack of a causal effect
- Example: mandatory face masks in public transport in NL from June 2020
o No apparent change in COVID-19 cases, even an increase in the autumn of
2020
- Concluding question: Do face masks have an effect on less COVID cases?
o No: we do not know what would have happened had there been no rule to
wear masks
o There is no clear counterfactual
3
, Applied Micro-econometric Techniques
Vaccinations:
- No clear correlation between vaccine rates and infection numbers (fluctuates
positively and negatively)
- Can we conclude vaccinations have no effect?
o No: We do not know what would have happened if there had been no
vaccinations
o Further studies show vaccinations are effective. It’s just that other things
happen simultaneously.
Threats to the identification of causal effects
Reverse Causality
- Example: Middle Ages
- Europeans believed lice to improve health
- Reasoning: They observed that sick people do not have lice, whereas healthy
people do
o No lice sick
- However, causality is reversed
- Lice are sensitive to body temperature and leave sick hosts
o No lice fever
Selection bias and Omitted variables
- Example: health status of people who have (not) been hospitalized) in the past
12 months
- Do hospitals make people sick? E.g. due to germs etc? Not necessarily
- Alternative explanations
o Selection bias
o Omitted variable bias
- Example: a study comparing hospital visits and health status might miss that
sicker people are more likely to visit hospitals (selection bias)
Summary of Causal relationships
4
