HW05

Part 1:Reading and questions

What is the causal link the paper is trying to reveal?

The effect of minimum wage increase on employment. Specifically, how employers in a low wage labor market react to increase in the minimum wage.

What would be the ideal experiment to test this causal link?

An unanticipated policy change in minimum wage regulation.

What is the identification strategy?

The authors use a minimum wage regulation in New Jersey that took place in 1992 to identify the treatment group (NJ) and control group (PA).

What are the assumptions / threats to this identification strategy?

There could be some confounding interaction from the competition brought about by new firms that may have sprung up after the policy change but have not been taken into account. Moreover, taking PA as control group may pose difficulty for the parallel trends assumption.

Part 2: Replication analysis

library(tidyverse)
library(dplyr)
library(stargazer)

Loading data

CardKrueger <- read.csv("~/AgEconR_Filipski_Spring2021/Assignment5/CardKrueger1994_fastfood.csv", header=TRUE)

Verify that the data is correct

summaryvars <- CardKrueger %>% group_by(state) %>% 
  summarize(a.BurgerKing = sum(bk), b.KFC = sum(kfc), c.Roys = sum(roys), d.Wendys =
              sum(wendys))

table <- as.matrix(summaryvars) 
table <- prop.table(t(table[,-1]), margin=2)*100 # proportion across columns after transposing
colnames(table) <- c("PA", "NJ") #Renaming columns
print(table)

##                    PA       NJ
## a.BurgerKing 44.30380 41.08761
## b.KFC        15.18987 20.54381
## c.Roys       21.51899 24.77341
## d.Wendys     18.98734 13.59517

summaryvars2 <- CardKrueger %>% group_by(state) %>% 
  summarize(FTEempWave1= mean(emptot, na.rm = TRUE), 
FTEempWave2 = mean(emptot2,na.rm = TRUE)) # mean of employment
table2 <- as.matrix(summaryvars2) 
table2 <- t(table2[,-1]) # Transpose
rownames(table2) <- c("FTE employment (Wave 1)", "FTE employment (Wave 2)")
colnames(table2) <- c("PA", "NJ")

print(table2)

##                               PA       NJ
## FTE employment (Wave 1) 23.33117 20.43941
## FTE employment (Wave 2) 21.16558 21.02743

Using OLS to obtain their Diff-in-diff estimator

regols <- lm(demp ~ state, data= CardKrueger)

# Formatting the table

stargazer(regols,  align=TRUE, type="text", 
          dep.var.labels=c("Difference, NJ-PA"))

## 
## ===============================================
##                         Dependent variable:    
##                     ---------------------------
##                          Difference, NJ-PA     
## -----------------------------------------------
## state                         2.750**          
##                               (1.154)          
##                                                
## Constant                     -2.283**          
##                               (1.036)          
##                                                
## -----------------------------------------------
## Observations                    384            
## R2                             0.015           
## Adjusted R2                    0.012           
## Residual Std. Error      8.968 (df = 382)      
## F Statistic            5.675** (df = 1; 382)   
## ===============================================
## Note:               *p<0.1; **p<0.05; ***p<0.01

The ols coefficient of state is 2.75 with 1.154 as standard error. Both the OLS coefficient as well as s.e are lower than the DiD estimates obtained in the paper.

Equation of a standard “difference in difference” regression

\[ FTE employment_{i,t} = \alpha + \beta state_i + \tau wave_t + \gamma (state_i * wave_t) + \epsilon_{i,t} \]