Price elasticity of demand for voluntary health insurance plans in Colombia

Table 3 Results of the heterogeneity exercise on semi-elasticities

	Mean Premium	Proportion who purchased a VPHI, E[\({vphi}_{i}\)]	Semi-elasticity (AME)	Computed elasticity
Panel A: Log premium spending
HH Age:
Less than 30 y/o (1)	255,483	0,00,279	-0.00165	-0.590
			(0.00247)	(0.888)
30 to 39 y/o (2)	286,093	0,00,844	-0.0293***	-3.471
			(0.00632)	(0.749)
40 to 49 y/o (3)	452,738	0,00,861	-0.0146**	-1.699
			(0.00540)	(0.627)
50 y/o or older (4)	464,654	0,00,969	-0.0231***	-2.382
			(0.00546)	(0.564)
At least one member over 65 y/o:
No	394,215	0,00,809	-0.0197***	-2.430
			(0.00402)	(0.497)
Yes	516,750	0,0119	-0.0260***	-2.196
			(0.00623)	0.525
At least one child under 5 y/o:
No	430,069	0,00,926	-0.0258***	-2.780
			(0.00493)	(0.532)
Yes	356,767	0,00,601	-0.0171***	-2.847
			(0.00444)	(0.738)
Panel B: Log household income
HH Age:
Less than 30 y/o (1)	255,483	0,00,279	0.00623**	2.234
			(0.00299)	(1.071)
30 to 39 y/o (2)	286,093	0,00,844	0.0178***	2.105
			(0.00286)	(0.339)
40 to 49 y/o (3)	452,738	0,00,861	0.0160***	1.858
			(0.00238)	(0.277)
50 y/o or older (4)	464,654	0,00,969	0.0186***	1.920
			(0.00235)	(0.242)
At least one member over 65 y/o:
No	394,215	0,00,809	0.0158***	1.957
			(0.00168)	(0.208)
Yes	516,750	0,0119	0.0202***	1.707
			(0.00346)	(0.291)
At least one child under 5 y/o:
No	430,069	0,00,926	0.0185***	1.998
			(0.00191)	(0.207)
Yes	356,767	0,00,601	0.0125***	2.083
			(0.00254)	(0.422)
Observations			14,378

Each row presents the marginal effects associated with the parameters corresponding to the estimate shown in the columns name concerning the value of the premium and household income interacted with the age of the head of the household, if there is at least one member of the household over 65 years old [y/o] or if there are children under five y/o in the household. The estimation of the simultaneous equations model using data from the ENPH2017 includes (i) estimation of a Heckman selection model (household size is the instrument), (ii) estimation of \({\beta }_{1}\) and \({\beta }_{2}\) using the predictions of the premium considering counts the inverse of mills. The model followed is restricted to a minimum premium value of 30 USD and for the head of the household to work a non-informal job. All models include controls for the head of the household, gender dummies, age groups, and educational level. From the household composition, it is considered whether a person over 65 is present and the income level. There are also region-fixed effects present in the estimation. Standard errors in parentheses. Significance: * 90%, ** 95%, ***99%. The complete table with tests of the validity of the linear combination of parameters is in the supplemental material (see Table S4 in Additional file 1)

ISSN: 1472-6963