DISENTANGLING MATERNAL DECISIONS CONCERNING BREASTFEEDING AND PAID EMPLOYMENT Bidisha Mandal, Washington State University Brian E. Roe, Ohio State University Sara B. Fein, Food & Drug Administration June 23, 2010 ASHEcon

Motivation 1  Importance of breastfeeding in developed countries is increasingly emphasized  National U.S. health objectives call for exclusive breastfeeding for at least 6 months and breastfeeding duration for at least 12 months  More U.S. mothers of infants are participating in the workforce  More mothers of infants are combining market work and breastfeeding  Postpartum return to work is associated with lower rates of initiation and shorter duration of breastfeeding in the U.S.  Conflict between economic incentives from labor market and health benefits (to both mother and child) from breastfeeding

Objective  What is the decision process regarding work leave and breastfeeding durations?  What is the decision process regarding work effort and breastfeeding intensity? 2

Data – Infant Feeding Practices Study II 3  Conducted by the FDA in collaboration with the CDC from May 2005 to June 2007  Longitudinal study of women from late pregnancy through their infant’s first year of life - 12 surveys (primarily mail)  1 prenatal questionnaire  1 telephone interview near the time of the infant’s birth  1 neonatal questionnaire at infant age 1 month  9 questionnaires sent about every 4 weeks during infant ages 2 to 7 months and then approximately every 7 weeks until infant age 12 months  4902 qualified pregnant women completed the prenatal questionnaire

Sample  Not enough information if due date was more than 3 months away or did not complete neonatal questionnaire  1885 cases deleted (3017 remaining)  Conditioning on pre-birth market work  1506 cases deleted (1511 remaining) if no prenatal work activity  Prenatal work = working for pay at any time from the 3 months before pregnancy until last trimester; does not have to be full-time worker  468 women did not initiate breastfeeding and/or worked pre-birth but did not intend to return to work  Excluded  1049 remaining  Have full information (on all variables of interest) for only N = 744 women for duration models N = 648 women for month 3 intensity model N = 586 women for month 6 intensity model 4

Selection Bias?  Breastfeeding initiation and work leave duration  Work leave duration is shorter for those who do not initiate (11.3 wks vs wks)  But, using Hausman endogeneity test, initiation is not endogenous to the determination of work leave duration  Intention to return to work and breastfeeding duration  Breastfeeding duration is shorter for those who return to work (25.3 wks vs wks)  Again, using Hausman endogeneity test, intention is not endogenous to the determination of breastfeeding duration 5 Initiation Intention Breastfeeding duration Work leave duration

Models  Breastfeeding and work leave durations  Both activities are right censored  End of survey censoring or missing data censoring  Exclusive breastfeeding and work leave durations  Both activities are right censored  End of survey censoring or missing data censoring  Breastfeeding intensity and work effort at months 3 and 6  Both activities are left censored  Observe intensity or effort only if participating in activity  4 cases – both observed, only one observed, both unobserved 6

Variables of Interest  Dependent variables  Length of time that a woman breastfeeds her infant (25.3 wks)  Length of time that a woman exclusively breastfeeds her infant (7.8 wks)  Length of time before a woman returns to employment postpartum (13.5 wks)  Number of breastfeedings per day at infant age 3 months (4.7)  Number of breastfeedings per day at infant age 6 months (3.2)  Number of hours of market work per week at infant age 3 months (10-19 hrs/wk)  Number of hours of market work per week at infant age 6 months (20-29 hrs/wk)  Independent variables  Background – age, education, race, marital status, residence  Parity – number of other children and experience with breastfeeding  Birth and breastfeeding – if mother was breastfed, comfort, pregnancy complications, problem breastfeeding, attended breastfeeding class  Prenatal job characteristics – self-employed, part-time work, available work leave, supportive at work, satisfaction with work, contribution to family income, occupation type 7

Estimation Method  Full Information MLE  Log scale 8

Data Limitations  Sample is probably not reflective of the average characteristics of the U.S. population  Higher average education  Higher income  Return to same employer  Not known 9

Work Leave and Breastfeeding Durations  Decision process is recursive rather than fully simultaneous  Longer leave from work increases duration of breastfeeding  Every percent increase in work leave increases breastfeeding duration by half percent  Duration of breastfeeding is not a significant determinant of work leave  Significant covariates  Mother’s age, education and married status are positively related to duration of breastfeeding  Women in western and northeastern states of U.S. breastfeed longer  Women with one other child who was also breastfed breastfeed longer  Women with no work leave available return to work sooner  Women who are satisfied with their work return sooner  Women who have 1 or more children and who has not breastfed before take longer leaves ; women who have more than 1 child and has breastfed before take shorter leave 10

Exclusive Breastfeeding Duration  Decision process is again recursive rather than fully simultaneous  Every percent increase in work leave increases exclusive breastfeeding duration by 0.66 percent  Duration of exclusive breastfeeding is not a significant determinant of work leave  Additional significant covariates  Self-employed women exclusively breastfeed longer  Self-employed women take longer leave from work  Women whose contributions to household income is more than 50% exclusively breastfeed for shorter duration  Change in sign  Women with one other child who was also breastfed exclusively breastfeed for shorter duration (time constraints?)  Negative correlation between unobserved factors 11

Work Effort and Breastfeeding Intensity  Recursive process (both months 3 and 6)  Work effort constraints breastfeeding intensity but the reverse is not true  Every 5-10 hour/week increase in work is associated with 1 less breast-milk feeding per day  Common significant covariates in months 3 and 6  Higher number of daily breastfeedings among higher educated women  Lower number of daily breastfeedings among women in southern states  Lower number of daily breastfeedings among women who are not comfortable with breastfeeding  No available leave is associated with less work effort  Higher contribution to family income is associated with higher work effort 12

Conclusions  Healthy People 2010  Goal: 50% women breastfeeding at month 6, 25% at 1 year  1990’s: 29% women breastfeeding at month 6, 16% at 1 year  Current: 43% women breastfeeding at month 6, 21% at 1 year  High degree of competition between breastfeeding and work  In terms of both duration and effort  Results indicate that employment decisions are made first  Longer work leave implies longer duration of breastfeeding and exclusive breastfeeding  More time spent at work means lesser number of breastfeedings  Working women probably face constraints at the workplace that are more rigid than those surrounding breastfeeding decisions  Similar results found in IFPS I ( )  Need for more workplace support 13