Skip to main content

Table 3 Model estimates of lockdown effects on employment probabilities.

From: Lockdown stringency and employment formality: evidence from the COVID-19 pandemic in South Africa

 

(1)

(2)

(3)

(4)

(5)

(6)

(7)

 

POLS

POLS

POLS

FE

DiD (POLS)

DiD (POLS)

DiD (FE)

Treatment

 − 0.046***

 − 0.046***

 − 0.036***

 

 − 0.030***

 − 0.020***

 
 

(0.008)

(0.008)

(0.007)

 

(0.008)

(0.008)

 

Post

 

 − 0.133***

 − 0.123***

 − 0.117***

 − 0.122***

 − 0.111***

 − 0.105***

  

(0.004)

(0.004)

(0.005)

(0.005)

(0.005)

(0.006)

Treatment × post

    

 − 0.028***

 − 0.028***

 − 0.029***

     

(0.008)

(0.008)

(0.008)

Time-varying controls

N

N

Y

Y

N

Y

Y

Time-invariant controls

N

N

Y

N

N

Y

N

Individual FE

N

N

N

Y

N

N

Y

Constant

0.720***

0.799***

 − 0.259***

1.107*

0.792***

 − 0.267***

1.079*

 

(0.005)

(0.005)

(0.053)

(0.603)

(0.005)

(0.053)

(0.602)

Observations

26,286

26,286

26,069

26,069

26,286

26,069

26,069

R2

0.002

0.023

0.193

0.800

0.023

0.193

0.800

  1. This table presents estimates of specification (1) with a binary employment variable serving as the dependent variable. Sample restricted to the working-age population (15–64 years) as of 2020Q1. POLS = pooled ordinary least squares, FE = fixed effects, DiD = Difference-in-Differences. Standard errors presented in parentheses and are clustered at the panel level. All estimates weighted using sampling weights. Time-varying controls include age, highest education level, and employment type. Time-invariant controls include sex, racial population group, province of residence, a binary urban residence indicator, an ‘industry or occupation job-mover’ indicator, and a ‘formality job mover’ indicator as described in Sect. 3.3
  2. ***p < 0.01, **p < 0.05, *p < 0.10