Employment Sorting by Size: The Role of Health Insurance (Text Version

On September 16, 2009, Barbara Schone made this presentation at the 2009 Annual Conference. Select to access the PowerPoint® presentation (998 KB) (Plugin Software Help).

Slide 1

Employment Sorting by Size: The Role of Health Insurance

Lan Liang and Barbara Schone
 

Slide 2

Goals of Our Analysis

• To add to the literature on the labor market effects of obesity
• To investigate whether obesity has an impact on worker sorting
• Two types of sorting of key interest:
  • Sorting by firm size
  • Sorting by insurance availability at a job
• Use similar approach to Kapur et al. (2008)

Slide 3

The Policy Relevance of Obesity

• According to the CDC ~ 34% of all adults are obese (2005 - 2006)
• Obesity is correlated with a number of serious health conditions
• Expected medical expenses are higher for obese individuals

Slide 4

Measuring Obesity

• Based on Body Mass Index (BMI)
  • BMI = Weight (lb)/Height2 (in) x 703
• BMI is correlated with body fat (but not perfectly)

Slide 5

What Do We Know about Labor Market Outcomes & Obesity?

• Some evidence that obesity adversely affects wages
• Mixed evidence that employment is affected
• Obesity may adversely affect worker productivity
• Some evidence of discrimination against obese workers
• Many of the effects vary across men and women

Slide 6

How Might Weight Affect Sorting across Firms for Insurance?

• With higher expected medical costs, obese workers might have greater demand for insurance
• Higher expected medical costs might lead firms to avoid obese workers since obesity is observable (if there are no full wage offsets)
• Normal weight people might find insurance less attractive and have reduced demand
• Net Effect: Insurance coverage could either increase or decrease for obese workers

Slide 7

How Might Weight Affect Sorting across Firms by Firm Size?

• If absenteeism is higher for obese workers, smaller firms might have a harder time adjusting to absenteeism and may be more inclined to avoid obese workers
• Obese workers may be more attracted to firms with generous benefits (e.g., sick leave) and may be more inclined to work in large firms as a result
• Net Effect: Obese workers are expected to be more likely to be employed in large firms

Slide 8

The Interaction of Firm Size and Insurance

• Large firms will also be more attractive because they are more likely to offer insurance
• Due to greater risk-pooling opportunities, an obese worker will have a smaller effect on the pool in a large firm than a small firm
• Net Effects: Conditional on offering insurance, obese workers may be more likely to be employed in large firms

Slide 9

Specific Research Questions

• How does weight affect the likelihood of being employed in a small firm?
• How does weight affect the likelihood of being employed in a job that offers insurance?
• Does weight affect the interaction effects of being in a small firm and being offered insurance?
• Do the patterns differ between men and women?

Slide 10

Data

• MEPS Full Year Data from 2002-2004
• Single Employed Persons Aged 21-64
• N = 14,150
• Results are weighted and adjust for the complex survey design

Slide 11

Measuring BMI

• Based on self-reported height and weight information
• 4 Weight Categories
  • Underweight (BMI < 18.5)
  • Normal (18.5 = BMI <25)
  • Overweight (25 = BMI < 30)
  • Obese (BMI = 30)

Slide 12

Key Dependent Variables

• Whether a worker is offered insurance from his main job
• Whether a worker is employed in a small firm
  • Use firm size = 25 for main results
  • Data report establishment, not firm size
  • Use establishment size and whether a firm has multi-establishments to derive a conservative measure of small firm

Slide 13

Descriptive Information

Slide 14

Interactions between Firm Size and Offers

Slide 15

Key Patterns

• Relative to Normal Weight Persons:
• Obese women are less likely to work in a small firm
• Obese workers are more likely to be offered insurance
• Obese workers are more likely to work in a large firm that offers insurance
• Obese men are more likely to work in a small firm that offers insurance (p < .10)

Slide 16

Multivariate Analysis

• Logit Models
  • Being employed in a small firm
  • Being employed in a firm that offers insurance
• Multinomial Logit with 4 Outcomes:
  • Small firm & offered
  • Small firm & not offered
  • Large firm & offered
  • Large firm & not offered
• Overall and by Gender
• Controls include sex, age, race/ethnicity, education, children, health status, region, MSA, unemployment, year dummies

Slide 17

Logit Results Odds Ratios

Slide 18

Multinomial Logit Results Odds Ratios

Slide 19

Key Findings from Logits

• Obese workers are (relative to normal weight):
  • Less likely to work in a small firm
  • More likely to work in a firm that offers insurance
  • Small firm result is not statistically significant for men
• Overweight workers are (relative to normal weight):
  • Less likely to work in a small firm
  • More likely to work in a firm that offers insurance
  • Not statistically significant for men

Slide 20

Multinomial Logit Findings

• Relative to normal weight, obese:
  • Workers are less likely to be in a small firm without insurance than a small firm with insurance
  • Workers are less likely to be in a large firm without insurance than a large firm with insurance
  • Workers are less likely to be in a small firm without insurance than a large firm with insurance offered

Slide 21

Implications

• Firms that offer insurance are not avoiding obese workers
  • Demand effects outweigh supply effects or
  • Maybe there is a full wage offset
• No statistically significant evidence that small firms are avoiding obese workers (with or without insurance being offered)
• Driven by women more than men
• Could we observing behavior driven by normal weight workers rather than obese or overweight workers?

Slide 22

Relevance

• Efficiency of labor markets
• Efficiency of health insurance

Slide 23

Future Steps

• Add married workers as an additional control group
• Compare obesity effects to unobservable conditions
• Consider hires of workers and worker separations