Table 4.
Results of regression for advertising time or space
| Number of observations = 214 | ||||
| Adjusted r2 = 0.6116 | F statistic = 56.90 | P of F statistic < 0.0001 | ||
| Variable | Coefficient | SE (Robust)* | T (Robust)* | P (Robust)* |
| ln GDP per capita | -0.592 | 0.0967 (0.1310) | -6012 (-4.52) | 0.000 (0.000) |
| ln of population in service area | 0. 425 | 0.0461 (0.0482) | 9.20 (8.81) | 0.000 (0.000) |
| Dummy if costs are for TV | 2.215 | 0.2559 (0.2705) | 8.66 (8.19) | 0.000 (0.000) |
| Dummy if costs are for newspaper | 2.055 | 0.2286 (0.2093) | 8.99 (9.82) | 0.000 (0.000) |
| Dummy for Eastern Europe | -8.668 | 4.1304 (4.1141) | -2.10 (-2.11) | 0.037 (0.039) |
| Dummy for Eastern Europe * ln GDP per capita | 1.0670 | 0.4688 (0.4673) | 2.28 (2.28) | 0.024 (0.026) |
| Constant | -5.1640 | 1.1805 (1.4728) | -4.37 (-3.51) | 0.000 (0.001) |
*The results as reported after a robust and country clustering are reported in the parenthesis.
Dependent variable: log ratio of price of media and advertising to GDP per capita
Breusch-Pagan test of heteroskedasticity: 1.91 (p = 0.1667 (Chi2)).
VIF test for multicollinearity is not appropriate for this model due to the inclusion of interaction variables.