Table 5 The Effect of Advertising Expenditures on Perceived Quality across Different Advertising Channels

Variable	N3.0	N3.1	N3.2	N3.3
TELEVISION	.044	.034^†	.029^†	.020***
INTERNET	.017**	.001	.006	.003
PRINT	.015***	.022***	.011*	.009
RADIO	.020**	.013	.021**	.017*
OUTDOOR	.031***	.028*	.028**	.027***
Controls
Brand FEs	Included	Included	Included	Included
Year FEs	Included	Included	Included	Included
Quarter FEs	Included	Included	Included	Included
Brand × Year FEs	Included		Included	Included
Brand × Quarter FEs		Included	Included	Included
Brand × Year × Quarter FEs				Included
Five Control functions	Included	Included	Included	Included
Intercept	14.881^†	14.953^†	14.967^†	15.025^†
Number of observations	43,081	43,081	43,081	43,081
Number of brands	898	898	898	898

Note: The authors use five control functions to account for the endogeneity of advertising expenditures in each channel. To construct each of the control functions, the authors use an instrumental variable by utilizing average monthly advertising expenditures across all other brands in the category in a particular advertising channel. All five predictors (TELEVISION, INTERNET, PRINT, RADIO, and OUTDOOR) have been log-transformed so that the coefficients for different channels can be comparable (i.e., the coefficients capture change in perceived quality due to a 1-percent increase in advertising expenditures in that channel). FE = fixed effect; RE = random effect.
↵* p < .10;
↵** p < .05;
↵*** p < .01;
↵† p< .001.