Homework 5
1 - Introduction
This Homework is based on the paper: The China Syndrome: Local Labor Market Effects of Import Competition in the United States, AER (2013), as in your Week 12 recitation.
The authors begin by proposing a simple measure of the growth China’s local import penetration into a local market \(i\) in the US:
\[ \Delta IPW_{uit} = \sum_{j}\frac{L_{ijt}}{L_{ujt}}\frac{\Delta M_{ucjt}}{L_{it}} \]
where:
- \(\Delta M_{ucjt}\) is the growth in overall imports from China to the US in industry \(j\)
- \(L_{it}\) is total employment in local labor market \(i\) at the start of period \(t\)
- \(L_{ijt}\) is total employment in market \(i\) in industry \(j\) at the start of period \(t\)
- \(L_{ujt}\) is total employment in the US in industry \(j\) at the start of period \(t\)
The authors estimate a series of specifications of the form:
\[ \Delta Y_{it} = \gamma_{t} + \beta_{1}\Delta IPW_{uit} + \mathbf{X}_{it}'\beta_{2} + e_{it} \]
where \(Y\) are the outcomes of interest. To isolate the portion of the change in import competition that is due to the supply side, the authors construct an instrumental variable:
\[ \Delta IPW_{oit} = \sum_{j}\frac{L_{ijt-1}}{L_{ujt-1}}\frac{\Delta M_{oct}}{L_{it-1}} \]
where:
- \(\Delta M_{oct}\) is the change in China’s exports to other high income countries.
- \(L_{ijt-1}\) is lagged employment and so on.
Recall the following from recitation:
- change in manufacturing is given by the variable
d_sh_empl_mfg
- the change in import exposure is
d_tradeusch_pw
- the predicted change given global trends (the instrument) is
d_tradeotch_pw_lag
- local labor markets are defined at the commuting zone level, and so each observation in the regressions is weighted using population share of the CZ, given by
timepwt48
- standard errors are clustered by state level
- Other control variables that you may need are explained in Recitation, or listed below.
Question 1
Reproduce Figure 2 Panel B from the paper (in the same way that we replicated Panel A in recitation).
Question 2:
- Use the
fixest
package to replicate the estimates from Column (6) of Table 3. Some notes on additional variables you will need (beyond what you saw in recitation):
l_sh_routine33
is the percentage of employment in routine occupationsl_task_outsource
is the average offshoreability of occupations in the CZ at the beginning of the period.
- By reading the paper, briefly explain the logic of including the perecnt of employment in routine occupations and the average measure of offshoreability.
Question 3:
- Replicate the analysis in Table 4, Panel (C), in the first three columns. The additional variables you will need are:
lnchg_popworkage
: log change in the working age populationlnchg_popworkage_edu_c
: log change in the working age population with college educationlnchg_popworkage_edu_nc
: log change in the working age population without a college education
- List one of the three potential explanations for the fact that there does not appear to be a response in population flows to import penetration.
Question 4
- Replicate the findings presented in Columns (5) and (6) in Table 7. You will need the additional variables:
lnchg_no_empl_nmfg_edu_c
: log change in total non-manufacturing employment for college educated works.lnchg_no_empl_nmfg_edu_nc
: log change in total non-manufacturing employment for non-college educated works.d_avg_lnwkwage_nmfg_c
: average change in log weekly earnings for college educated workers in non-manufacturing jobsd_avg_lnwkwage_nmfg_nc
: average change in log weekly earnings for non-college educated workers in non-manufacturing jobs
- How do the authors interpret the fact that import penetration appears to have a negative impact on wages in the non-manufacturing sector?