Title: Psychiatric Disorders and Labor Market Outcomes: Evidence from the National Latino and Asian American Study (NLAAS)
1Psychiatric Disorders and Labor Market Outcomes
Evidence from the National Latino and Asian
American Study (NLAAS)
- Pinka Chatterji, PhD, Mingshan Lu, PhD, Margarita
Alegria, PhD and David Takeuchi, PhD - June 7, 2004
- Disparities in Treatment for Impact of Mental
Illness - AcademyHealth 2004 Annual Research Meeting
2Acknowledgements
- The NLAAS data used in this analysis was provided
by the Center for Multicultural Mental Health
Research at the Cambridge Health Alliance - The project was supported by NIH Research Grant
U01 MH62209 funded by the National Institute of
Mental Health (M. Alegria and D. Takeuchi, PIs)
as well as SAMHSA/CMHS and OBSSR - Pinka Chatterji additionally acknowledges
research support from grant K01 AA000328-03 from
the National Institute of Alcohol Abuse and
Alcoholism - Mingshan Lu is an AHFMR (Alberta Heritage
Foundation for Medical Research) Population
Health Investigator and thanks the foundation and
Institute of Health Economics for financial
supports
3Objective
- The objective of this paper is to investigate
the effect of recent psychiatric disorders on
three labor market outcomes (current employment
status, the number of weeks worked in the past
year, and the number of work absences in the past
month) using a nationally representative sample
of Latino and Asian Americans
4Motivation
- Mental health and labor market outcomes
literature recent psychiatric disorders appear
to detract from employment and earnings - Disparities in mental health and impact of mental
illness - - Minority individuals face distinct labor market
experiences due to factors such as
discrimination, citizenship, and language
barriers - Thus, the labor market consequences of mental
disorders may be different for Latinos, Asians
and other minority individuals the availability
of the NLAAS allows us to investigate this
question for the first time
5Background
- Previous work based on the NCS and the ECA
surveys indicates that a recent psychiatric
disorder is associated with - 21 percent reduction in earnings among men (Frank
Gertler 1991) - 11 percentage point reduction in the probability
of being employed among men and women (Ettner et
al. 1997) - Substance use and substance use disorders are
linked to reduced educational attainment, but not
necessarily to worse labor market outcomes - ECA data indicates that onset of alcohol abuse
before age 19 reduces schooling by 1.5 years,
which in turn reduces earnings (Mullahy
Sindelar 1994) - Alcoholism is associated with higher rates of
employment among white women (Mullahy Sindelar
1997) - Illicit drug use has negative or no effects on
employment, but mixed effects on wages (DeSimone
2002, Kaestner 1991, 1994a, 1994b)
6Contribution of this study
- Builds on previous research on the labor market
consequences of mental disorders - Comprehensive measures of psychiatric illness
- First study to investigate the impact of mental
illness on labor market outcomes among ethnic
minorities - Nationally representative sample of Asians and
Latinos
7Data
- NLAAS is a national psychiatric epidemiologic
study being conducted to measure psychiatric
disorders and mental health service usage in a
nationally representative sample of Asians and
Latinos - Analysis samples include NLAAS respondents
between 18 and 65 years old who are not in school
and have complete survey information - Separate analysis by Latino and Asian ethnicity
and gender - Male Latino sample (n 1,024)
- Female Latino sample (n 1,233)
- Male Asian sample (n 820)
- Female Asian sample (n 892)
- Descriptive statistics and models are weighted
using nationally representative survey weights
8NLAAS Sampling Design
- Focus on Cubans, Mexicans, Puerto Ricans other
Latinos/Latinas
- Focus on Chinese, Vietnamese, Filipinos other
Asians
9Methods
-
- Previous researchers have faced two major
methodological challenges - statistical endogeneity (unobserved
heterogeneity) - structural endogeneity (reverse causality work
affects mental health) - Standard regression methods (such as OLS) ignore
endogeneity problem - To deal with this problem, we use two approaches
- Including lifetime psychiatric disorder as a
covariate - Does not directly address structural endogeneity
- IV methods, which purge mental disorder measure
of its correlation with the disturbance term,
leading to consistent estimates - We use number of disorders with onset before age
18 as an identifying instrument (following Ettner
et al. 1997) - Should be a good predictor of current mental
disorders, but should not directly impact current
labor market outcomes if models control
adequately for indirect pathways
10Methods
- Labor market outcomes
- Employed dummy variable 1 if respondent is
currently employed, 0 if respondent is unemployed
or out of labor force - Log of weeks worked among employed individuals,
the log of the number of weeks worked in past
year - Number of absences among employed individuals,
number of full days of work respondent missed in
the past 30 days - Current Psychiatric Disorders
- Any DSM IV mental disorder in past 12 months
(major depression, dysthymia, agoraphobia, panic
disorder, social phobia, substance use disorder,
generalized anxiety disorder, PTSD, anorexia or
bulimia) - Results by type of psychiatric disorder in paper
not shown here - Other covariates ethnicity, number of family
members under 18, age, education, marital status,
state unemployment rate
11DESCRIPTIVE STATISTICS
12Variable Asian Males Asian Females Latino Males Latino Females
Employed 0.859 0.665 0.854 0.615
Out of labor force 0.090 0.243 0.071 0.318
Unemployed 0.051 0.092 0.076 0.067
Weeks worked 45.04 37.41 45.43 34.68
Number of absences in past 30 days 1.51 2.37 1.37 1.51
13Variable Asian Males Asian Females Latino Males Latino Females
Any current disorder 0.096 0.088 0.136 0.190
Current major depression 0.056 0.038 0.050 0.089
Current anxiety 0.056 0.045 0.075 0.124
Current substance abuse or dependence 0.018 0.006 0.045 0.014
Number of disorders with onset prior to age 18 0.248 0.201 0.320 0.334
14SUMMARY OF REGRESSION RESULTS
15Dependent Variable Employed (0/1) Latino Samples Coefficient (p-value) Dependent Variable Employed (0/1) Latino Samples Coefficient (p-value) Dependent Variable Employed (0/1) Latino Samples Coefficient (p-value) Dependent Variable Employed (0/1) Latino Samples Coefficient (p-value)
OLS OLS w/Lifetime IV
Males, 18-64 Males, 18-64 Males, 18-64
Any 12 month disorder -0.121 (0.021) -0.115 (0.050) -0.071 (0.401)
N 1,024 1,024 1,024
Females, 18-64 Females, 18-64 Females, 18-64
Any 12 month disorder -0.154 (0.002) -0.145 (0.027) -0.160 (0.100)
N 1,233 1,233 1,233
16Dependent Variable Employed (0/1) Asian Samples Coefficient (p-value) Dependent Variable Employed (0/1) Asian Samples Coefficient (p-value) Dependent Variable Employed (0/1) Asian Samples Coefficient (p-value) Dependent Variable Employed (0/1) Asian Samples Coefficient (p-value)
OLS OLS w/Lifetime IV
Males, 18-64 Males, 18-64 Males, 18-64
Any 12 month disorder -0.131 (0.176) -0.075 (0.522) -0.128 (0.503)
N 820 820 820
Females, 18-64 Females, 18-64 Females, 18-64
Any 12 month disorder -0.059 (0.547) 0.080 (0.610) -0.334 (0.102)
N 892 892 892
17Dependent Variable Log of weeks worked in past year Latino Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Log of weeks worked in past year Latino Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Log of weeks worked in past year Latino Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Log of weeks worked in past year Latino Samples (Limited to Employed Persons) Coefficient (p-value)
OLS OLS w/Lifetime IV
Males, 18-64 Males, 18-64 Males, 18-64
Any 12 month disorder 0.014 (0.833) 0.072 (0.437) 0.134 (0.191)
N 826 826 826
Females, 18-64 Females, 18-64 Females, 18-64
Any 12 month disorder -0.078 (0.461) -0.168 (0.148) 0.103 (0.606)
N 758 758 758
18Dependent Variable Log of weeks worked in past year Asian Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Log of weeks worked in past year Asian Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Log of weeks worked in past year Asian Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Log of weeks worked in past year Asian Samples (Limited to Employed Persons) Coefficient (p-value)
OLS OLS w/Lifetime IV
Males, 18-64 Males, 18-64 Males, 18-64
Any 12 month disorder -0.037 (0.627) 0.064 (0.560) -0.163 (0.506)
N 690 690 690
Females, 18-64 Females, 18-64 Females, 18-64
Any 12 month disorder -0.251 (0.445) -0.365 (0.283) 0.221 (0.405)
N 605 605 605
19Dependent Variable Number of work days missed in past month Latino Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Number of work days missed in past month Latino Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Number of work days missed in past month Latino Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Number of work days missed in past month Latino Samples (Limited to Employed Persons) Coefficient (p-value)
OLS OLS w/Lifetime IV
Males, 18-64 Males, 18-64 Males, 18-64
Any 12 month disorder 1.31 (0.100) 1.00 (0.242) 0.018 (0.983)
N 825 825 825
Females, 18-64 Females, 18-64 Females, 18-64
Any 12 month disorder 0.436 (0.389) 1.32 (0.015) -0.354 (0.747)
N 747 747 747
20Dependent Variable Number of work days missed in past month Asian Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Number of work days missed in past month Asian Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Number of work days missed in past month Asian Samples (Limited to Employed Persons) Coefficient (p-value) Dependent Variable Number of work days missed in past month Asian Samples (Limited to Employed Persons) Coefficient (p-value)
OLS OLS w/Lifetime IV
Males, 18-64 Males, 18-64 Males, 18-64
Any 12 month disorder 0.619 (0.587) 0.678 (0.616) -0.421 (0.644)
N 690 690 690
Females, 18-64 Females, 18-64 Females, 18-64
Any 12 month disorder -0.064 (0.938) 2.05 (0.111) 2.11 (0.332)
N 603 603 603
21Summary of Results
- Any mental disorder in past 12 months detracts
from employment for Latinos (both men and women)
but not necessarily for Asians - Effects persist in IV models and in models that
include lifetime psychiatric disorder - No effects of mental disorders on absences and
weeks worked among employed individuals - Similar results for depression, anxiety and
substance abuse/dependence (shown in the paper)
22Conclusions and Future Work
- Results for Latinos are similar to estimates
based on US nationally representative data - Results for Asians are unexpected require
further investigation - Future work will consider respondent net earnings
as a dependent variable