Cantor and Packer (1996) analyzed the relationship between sovereign credit ratings and yields spread. Using a cross sectional data of 49 countries, they have estimated 3 different models using OLS regression and obtained the results below.

 Dependent variable: ln(yield spread) 

                                                                                            (1)        (2)               (3) 

Intercept                                                                      2.105***     0.466        0.074 

                                                                                    (16.148)       (0.345)     (0.071) 

Average ratings (numeric)                                        -0.221***                      -0.218*** 

                                                                                     (19.715)                        (4.276)

 Per capita income (thousand $)                                                -0.144         0.226 

                                                                                                       (0.927)        (1.523) 

Indicator for economic development                                       -0.723***     -0.38 

(1 = industrialized; 0 = not industrialized)                                  (2.059)        (1.341) 

Adjusted R2                                                                 0.919        0.857           0.914 

Notes: t ratios in parenthesis; *** indicates significance at the 1% level. Models (2) and (3) include other covariates, see Cantor and Packer (1996). 

 Necessary

1. Interpret the regression coefficients of the bivariate regression model (1). What could be the difference in your interpretation if the dependent variable was yield spread instead of log yield spread? 

2. Consider models (2) and (3) and the estimates of the coefficient of Indicator for economic development. Why do you think adding the variable Average ratings changed the coefficient estimate (and level of significance) in model (3)? 

3. How can we estimate the goodness of fit of a regression? We know that the 'mean-independence' assumption will be violated if we exclude an important variable from our model. What if we include all possible variables in the model? Will that improve the goodness of fit? Explain in comparison of models (1) and (3) above.

Answer rating: 100% (QA)

1 In the bivariate regression model 1 the coefficient of the Average ratings variable is 0221 which is statistically significant at the 1 level This m... View the full answer