1 / 19

Food Labels and Weight Loss: Evidence from the National Longitudinal Survey of Youth

Food Labels and Weight Loss: Evidence from the National Longitudinal Survey of Youth. Bidisha Mandal Washington State University. AAEA ‘08, Orlando. Motivation. Who reads nutrition labels? Any link with body weight? Policy Implications NLEA enacted in 1994

barbra
Download Presentation

Food Labels and Weight Loss: Evidence from the National Longitudinal Survey of Youth

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Food Labels and Weight Loss:Evidence from the National Longitudinal Survey of Youth Bidisha Mandal Washington State University AAEA ‘08, Orlando

  2. Motivation Who reads nutrition labels? Any link with body weight? • Policy Implications • NLEA enacted in 1994 • Estimated health benefits over 20 years estimated to be $4.4 – 26.5b (FDA, 1993) • NLEA’s impact on body weight over 20 years estimated to be $63 – 166b (Variyam & Cawley, NBER WP 2006) • Nutrition labeling meets consumers need for accurate, standardized and comprehensible information (WHO)

  3. Use of nutrition information • FDA Consumer, 1995 • Survey of 1,000 individuals very soon after the 1994 enactment • ~50% of those who saw the label changed decisions • Of those, 70% cited fat content • Neuhouser et al., JADA 1999 • Label use higher among women, more than HS educated • Label use significantly associated with lower fat intake • Strong predictors of label use – belief in importance of eating low-fat diet, belief in association between diet and cancer • Kristal et al., JADA 2001 • Use of food labels strongly associated with lower fat intake, weaker association with increase in fruits and vegetables consumption

  4. Welfare effects of nutrition information • Zarkin et al., AJPH 1993 • Potential health benefits from expected change in food consumption in terms of life years gained • Life expectancy increase between 0.12-2.06 years • Teisl and Levy, JFDR 1997 • Income and substitution effects of nutrition labels • Kim et al., JARE 2000 • Endogenous switching regression techniques to control for heterogeneity in the label use decision • Teisl et al., AJAE 2001 • Nutritional labeling affects purchasing behavior • May not necessarily increase consumption of ‘healthy’ foods, but may cause substitution within ‘unhealthy’ foods

  5. Model Label preference parameter ith individual’s utility from reading label (j) at time t Proxied by time spent on buying groceries by i at t (TGit)

  6. Label Preference Aggregate effects Trying to lose weight or not Habit capital Temporal and permanent personal random effects Demographics, income Binary variable for label use

  7. Implications of the Model • Probability of reading nutrition labels • Who reads nutrition labels? • Purely out of habit? • Effect of individual’s actions regarding weight • Transition probability • Probability of transition between label use • What happens when individual starts trying to lose weight • Propensity to lose weight • Are individuals more successful if they read nutrition labels?

  8. NLSY Panel Data • Survey years - 2002, 2004, 2006; Number of observations – 6,895 • Age – 37 to 50 years • Education • Less than HS – 10.17% • HS – 43.21% • Some college – 24.29% • College and above – 22.33% • Gender • Male – 47.6% • Female – 52.4% • Race • White – 50.43% • Black – 30.73% • Hispanic – 18.42% • Other – 1.02% • Income (in $10,000) – Mean = 6.83, Within SD = 6.63

  9. Survey Questions • When you buy a food item for the first time, how often would you say you read the nutritional information about calories, fat and cholesterol listed on the label? • Don’t buy food (~ 0.01% - dropped) • Always • Often • Sometimes • Rarely • Never • Are you now trying to lose weight, gain weight, stay about the same, or are you not trying to do anything about your weight? • 47.14% trying to lose weight (within SD is 24.38%) Users – 66.58% (within SD is 23.32%) Non-users

  10. Who tries to lose weight? Percentages – average across survey years

  11. Time Cost • American Time Use Survey • Time spent on various activities – includes buying grocery • Eating and Health Module (2006) – also collects weight and height data (BMI) • Hot-deck Imputation • Impute data from ATUS-EH module and match by gender, education and BMI within age range of 37-50 years • Hot-deck imputation works better than Mean imputation • Mean (SD) of ATUS-EH only – 42.3 (29.9) minutes • Mean (SD) using mean imputation – 42.1 (4) minutes • Mean (SD) using hot-deck imputation – 40.8 (29.6) minutes • Minimum and maximum are also matched better with hot-deck

  12. Empirical Estimation • Random effects model • Data from survey years 2004 and 2006 analyzed using ‘habit’ from 2002 and 2004 waves respectively • ‘Habit’ • Calculation • Lagged label use • Two-stage probit • Alternative hypotheses: Persistence in unobserved personal effect • Serial correlation • Unobserved heterogeneity

  13. Probit Model of Reading Labels

  14. Serial Correlation • Let • Serial correlation hypothesis implies improvement in likelihood value if add in addition to (Shachar, 1994)

  15. Transition Overall, individuals are ~23%more likely to start reading food labels when they decide to try to lose weight.

  16. Transition Probability of Reading Labels 95% Confidence Intervals Dependent Variables: Starts reading food labels

  17. Predictions and Fit • Probability of reading labels • 82% accurate predictions • Predictions suffer among • African-Americans • HS or less educated • Male • Transition probability • Actual average proportion of those not reading labels in sequential wave and previously reading labels is 9% - predictions using lagged ‘habit’ is 11.5% • Actual average proportion of those reading labels in sequential wave and not reading labels previously is 14.5% - predictions using lagged ‘habit’ is 11.6%

  18. Propensity to lose weight Marginal effects (z-statistic) Dependent Variable: Weight loss in sequential wave or not * Covariates are income, exercise duration, and demographics

  19. Conclusions • Those who try to lose weight are more likely to read nutrition labels – certain demographic groups more so • Habit is still the best predictor of label use • Yet, likelihood ratio tests show weight loss preference is an important factor • Transition probabilities also support the notion that individuals who decide to lose weight are overall more likely to start reading food labels • Propensity to lose weight and label use • Those who read labels are more likely to be successful in losing weight

More Related