Table 2.

Risk Selection Based on Medical Care Use in Round 1

Variable	Coefficient	Std. Error	Coefficient	Std. Error	Coefficient	Std. Error
Non-HMO in round 1
Non-HMO in round 2 (control)	1,530**	466	1,992**	495	943	695
HMO in round 2 (switcher)	1,320**	482	1,803**	503	860	691
HMO in round 1
HMO in round 2 (control)	1,587**	472	2,058**	508	961	702
Non-HMO in round 2 (switcher)	1,957**	489	2,435**	525	1249	724
Female			632**	110	593**	104
Age (56–65 is omitted group)
Under 2 years of age			3,934**	700	4,912**	811
Between 2 and 17			−1,445**	264	−557	486
Between 18 and 40			−766**	274	−190	298
Between 40 and 55			−865**	270	−556**	281
Other potentially observable characteristics included?	NO		NO		YES
Number of observations	6,235		6,235		6,235

Notes: Dependent variable is medical care use in round 1, measured in dollars.

^**=significantly different from zero at the 5-percent level;

^*=significantly different from zero at the 10-percent level. Bold coefficients indicate that the expenditure difference in round 1 between the people switching plan types and those remaining in a particular plan type is significantly different from zero at the 10-percent level. All specifications include indicator variables for the site of the survey (usually a metropolitan statistical area). Additional potentially observable enrollee and employer characteristics include: self-reported health, marital status, children, race, education, household income, type of employer (i.e., private or government), and number of employees at the person's firm. The constant is omitted. Sample includes people who had a choice between an HMO and a non-HMO plan in round 2.