# REGRESSION ANALYSIS

Table 1: Mean, SD and CV Variable Wage Education Age Experience Note: CV = (SD/Mean)*100 Interpretation of Table 1: The mean value for Wage is 345.2, Education is 11.58, Age is 31.58 and for the Experience are 12.74. The variation from the mean shown by the Standard deviation (SD) for Wage is 181, for Education is 5.15, Age is 8.33 and for the Experience is 6.03. By Standard deviation (SD), wage varies much followed by Age. The coefficient of variation (CV) suggests that, the closeness of the estimates to the actual value is 52%, 44%, 26% and 47% for Wage, Education, Age and Experience respectively. By the coefficient of variation (CV), wage varies most followed by Experience. Table 2: Correlation between variables Wage Gender Education Age Experience Log (education) Wage 1 Gender 0.69 1 Education 0.92 0.69 1 Age 0.88 0.58 0.87 1 Experience 0.84 0.47 0.79 0.8 1 Log(education) 0.89 0.7 0.97 0.86 0.7 1 Log(age) 0.86 0.56 0.84 0.99 0.76 0.86 Log(experience) 0.81 0.43 0.73 0.8 0.96 0.68 Interpretation Table 2: There is positive correlation between every two variables of the equation. The dependent variable (Wage) is highly correlated with Education (0.92) (also its log form, 0.89) followed by Age (0.88) (also its log form, 0.86). In other words, Wage is highly linearly associated with Education by 0.92. Log Log (age) (experience) Mean 345.2 11.58 31.58 12.74 SD 180.9526 5.147775 8.327701 6.029892 CV 52.4196% 44.454% 26.3702% 47.3304%

Where; SD = Standard deviation and CV= coefficient of variation

1 0.78

1

1. Regression analysis between Wage (Y) and Age (X)
I. Normal linear regression analysis

Y = b1 + b2X + µ ; where, Y = Wage and X = Age Y = -256.9 + 19.07X
Se t (49.1) (-5.23) (1.5) (12.67)

N=50

The variable Age(X) is statistically significant because |t|>1.96. Interpretation Keeping other factors constant, a unit change in age brings about 19.07 units increase in wage. When all factors do not change, wage decreases by 256.9. Elasticity of Y(Wage) with respect to Age is 1.744205 which is elastic. This means that, at every 1% increase in age, wage increases by 1.74% (at every unit increase in age, wage increases by 1.74 units) Since Fcalculated(1,48) = 160.6 is greater than F0.05,1,48 = 4.043, the overall model is Statistically Significant. The variables explain the model by 77% and rest is explained by other variables. Hence the equation is reliable. II. Double log regression analysis

lnY= lnb1 +b2 lnX + µ ; where, lnY = logwage and lnX = logage lnY= -0.64 + 1.85lnX se (0.52) (0.15) t (-1.24) (12.14) N=50 R2= 0.7543 Adj R2=0.7492 Since the t-value associated with the variable log Age (lnX) |t|>1.96 then it is statistically significant Interpretation Elasticity of Y(Wage) with respect to Age is 1.85 (Elastic). This means that, at every 1% increase in log age, log wage increases by 1.85% (at every unit increase in log age, log wage increases by 1.85 units)

Since Fcalculated(1,48) = 147.37 is greater than F0.05,1,48 = 4.043, the overall model is Statistically Significant. The variables explain the model by 75% and rest is explained by other variables. . Hence the equation is reliable. III. Quadratic regression analysis

Y= b1 + b2X + b3X2 + µ ; where, Y = Wage and X = Age Y= -151.84 + 12.13X + 0.11X2 Se (167.82) (10.7) (0.16) t (-0.9) (1.13) (0.66) N=50 R2= 0.7720 Adj R2=0.7623 Interpretation Elasticity of Y with respect to X is 1.1 (Elastic). This means that at every unit increase in age, wage increases by 1.1 units Elasticity of Y with respect to X2 is 0.33 (Inelastic). This means that at every unit increase in age squared, wage increases by 0.33 units To observe if the model can actually be non-linear, Hypothesis H0: b3=0 (The equation is linear) H1: b30 (The equation is non-linear) Significance level = 0.05 Test statistics T ± test Tcalcuted = 0.66 Decision Since |t|<1.96 then the variable X2is statistically insignificant, H0 is not rejected. So, the equation can be linear. Moreover, the variable X in this quadratic equation is also statistically insignificant since the tvalue associated with its coefficient b2 is less than 1.96 as well. For this reason, the elasticity is not useful.

Since Fcalculated(2,47) = 79.56 is greater than F0.05,2,47 = 3.195, the overall model is Statistically Significant. The variables explain the model by 77% and rest is explained by other variables. . Hence the equation is reliable.

2. Regression analysis between Wage (Y) and Experience (X)
I. Normal linear regression analysis

Y = b1 + b2X + µ ; where, Y = Wage and X = Experience Y = 23.57+ 25.25X
Se (32.95) (2.34) t (0.72) (10.78)

N=50

The variable Age(X) is statistically significant because |t|>1.96 Interpretation Keeping other factors constant, a unit change in experience brings about 25.25 units increase in wage. When all factors do not change, wage increases by 23.57 Elasticity of Y(Wage) with respect to Experience is 1.931708which is elastic. This means that, at every 1% increase in experience, wage increases by 1.93% (at every unit increase in experience, wage increases by 1.93 units) Since Fcalculated(1,48) = 116.22 is greater than F0.05,1,48 = 4.043, the overall model is Statistically Significant. The variables explain the model by 71% and rest is explained by other variables. Hence the equation is reliable. II. Double log regression analysis

lnY= lnb1 +b2 lnX + µ ; where, lnY = logwage and lnX = logexp lnY= 3.37 + 0.95lnX se (0.31) (0.13) t (10.78) (7.56) N=50 R2= 0. 5432 Adj R2=0. 5337 Since the t-value associated with the variable log experience (lnX) |t|>1.96 then it is statistically significant

Interpretation Elasticity of Y(Wage) with respect to experience is 0.95 (Inelastic). This means that, at every 1% increase in log experience, log wage increases by 0.95% (at every unit increase in log experience, log wage decreases by 0.95 units) Since Fcalculated(1,48) = 57.09 is greater than F0.05,1,48 = 4.043, the overall model is Statistically Significant. The variables explain the model by 54% and rest is explained by other variables. III. Quadratic regression analysis

Y= b1 + b2X + b3X2 + µ ; where, Y = Wage and X = Experience Y= 31.3 + 24.06X + 0.04X2 Se (76.44) (10.82) (0.33) t (0.41) (2.22) (0.11) 2 N=50 R = 0.7078 Adj R2=0. 6953 Interpretation Elasticity of Y with respect to X is 0.89 (Inelastic). This means that at every unit increase in experience, wage decreases by 0.89 units To observe if the model can actually be non-linear, Hypothesis H0: b3=0 (The equation is linear) H1: b30 (The equation is non-linear) Significance level = 0.05 Test statistics T ± test Tcalcuted = 0.11 Decision Since |t|<1.96 then the variable X2is statistically insignificant, H0 is not rejected. So, the equation can be linear.

Since Fcalculated(2,47) = 56.92 is greater than F0.05,2,47 = 3.195, the overall model is Statistically Significant. The variables explain the model by 71% and rest is explained by other variables. . Hence the equation is reliable.

3. Multiple Regression Analysis
Y =b1+b2 X2 + b3X3 + b4X4 + b5X5 + µ
Where; Y = Wage X2=education X3= Age X4 = Experience X5= Gender

Y =-97.53 +15.11 X2 + 4.33X3 + 8.23X4 + 52.71X5 Se (38.9) (23.35) t (-2.51) (2.26) R2= 0.9021 (3.89) (3.88) (2.2) (1.97) (2.5) (3.29)

N=50