Tutorial 9: (Women) labour force participation
In this exercise, we are going to learn:
- How households decide on their labour supply
- How the family size affect the latter
- Estimate the effects of having children on women labour participation
Theoretical approach of labour supply
Consider an individual who derives her utility from consuming a good \(c\) and leisure \(l\). The good comes at price \(p=1\), it is the numéraire. Utility function is:
\[ U(c,l) = c^\alpha \times l^{1-\alpha} \]
The individual earns a labour wage \(w\). The number of hours worked is denoted \(h\), hence the labour income is \(wh\). She also obtains a lump-sum benefit from the governement \(y\). Hence, her total income is \(wh + y\). There is a fixed number of hours \(T\) in each day, such that \(T = l+h\).
- What are the two constraints? Can you express the budget constraint as a function of leisure \(l\), consumption, and total amount of time \(T\) only?
- Write down the maximization problem and the associated lagrangean.
- Solve the maximization program to find \(c^\star\) and \(h^\star\).
We are now going to proceed to some comparative statics, i.e. we are going to twist some parameters to understand the partial equilibrium effects.
- How does evolve \(h^\star\) with wage? with the lump-sum benefit?
Now, we are going to assume that the individual has a positive number of children \(n>0 \in \mathbb{N}\). Then, she must allocate a fixed number of hours to take care of their children. Each child requires \(K\) hours of childcare, hence the total number of hours is \(nK\). In practice, it reduces the number of hours available to leisure and work.
- How does it affect labour supply? Is there a threshold upon which the mother stops providing labour?
Empirical approach
The theoretical prediction is that mothers might drop out the labour force because they have children. We are going to estimate this effect.
- Open the dataset
lfs_women.csv
.
Variable | Description |
---|---|
idmen |
Household ID |
obs |
Household member position in the HH |
mne1 |
Number of children |
nag |
Age |
nsexe |
1 if male |
same_sexe |
1 if first two children have the same gender |
hh_twin_after |
1 if second born are twins |
ndiplo |
Scale from 1 to 9, high skilled from 5 to 9 |
ntravail |
1 if the individual is working |
ntpp |
2 if the individual is working full time |
For the sake of time, the dataset has been already cleaned. We can directly move to the estimation.
- Write a model with the labour force participation on the left hand side. This variable is explained by: being a mother, having a large family (more than 2 children), and the interaction between both.
- Create the relevant variables.
- Run this model. Intepret.
- Is there OVB? Which variables can you include to tackle it?
- Add household fixed effects. Why should we add these FE?
- Is there any endogeneity bias?
- A good instrument is to use the number of twins. Why?
- Run the IV regression with the number of twins as instrument for having a large family. Interpret.
- An alternative instrument is to have children the same gender. Why?
- Run the IV regression with an indicator that takes 1 if the children are of the same gender. Intepret.