Variable

Tobit

Probit

Truncated regression

Heckman


Coef.

t

Coef.

z

Coef.

z

Coef.

z


Age

0.219

1.21

− 0.015**

− 2.13

0.085

2.46

0.091**

2.63

Age squared

− 0.003

− 1.54
  
− 0.001

− 2.38

− 0.001**

− 2.38

Female

0.358

0.39

− 0.105

− 0.60

0.055

0.41

0.039

0.25

Married
  
− 0.403

− 1.82*
    
Education attainment

0.140

0.47

− 0.019

− 0.28

0.147

2.63

0.157**

2.42

Local market

− 0.765

− 0.46

− 0.308

− 0.84

0.677

2.09

0.885**

2.31

Distant market

1.873*

1.78

0.337

1.41

0.517

2.58

0.356

1.44

Knowledge
  
0.557

2.38**
    
Own savings account

− 0.858

− 1.19

− 0.279

− 1.72*

0.036

0.27

0.169

0.99

Make decisions on income
  
− 0.365

− 2.15**
    
ln (Fishing assets)

− 0.067

− 1.08
  
0.022

1.84

0.024**

2.03

ln (Fish processing assets)

− 0.072

− 0.96
  
0.011

0.79

0.008

0.59

Madzedze

− 0.483

− 0.46

0.046

0.17

− 0.802

− 4.2

− 0.805***

− 3.56

Malembo

− 2.359**

− 2.24

− 0.438

− 1.71*

− 0.497

− 2.57

− 0.279

− 1.09

Msaka

− 2.154**

− 2.00

− 0.425

− 1.68*

− 0.814

− 4.08

− 0.598**

− 2.30

Chikombe

− 1.322

− 0.86

− 0.360

− 0.99

− 0.425

− 1.51

− 0.261

− 0.77

Lambda
      
− 1.128*

− 1.87

Constant

4.095

1.05

1.854

3.25***

8.054

10.44

8.240***

10.09

Number of observations

382
 
382
 
284
 
382
 
LR/Wald Chisquare

25.34**
 
31.61***
 
62.17***
 
51.27***
 
Log likelihood

− 1008.58
 
− 201.712
 
− 383.964
   
Pseudo R^{2}

0.0124
 
0.0752
     
 The same probit model has been used as a selection equation in the Heckman selection model and participation equation in the double hurdle model
 ***Denotes statistically significant at 1%
 **Denotes statistically significant at 5%
 *Denotes statistically significant at 10%