Question 1
Each observation in the dataset represents a worker. Variables in the dataset are:
- expenditure: monthly expenditure on entertainment
- time: commuting time to workplace, in hours
- pass: explained later
Consider the regression: expenditure = α + β × time + ε.
a) Suggest a confounding factor that makes β not a causal effect of commuting time on expenditure on entertainment.
b) Suppose that the City of Sydney gave a "Sydney motorway pass" to randomly chosen workers in Sydney, allowing pass holders to use motorways free of charge. Explain why it is a valid instrument for commuting time in the regression in (a).