Modeling Health Care Reform: What States Need to Know

1
Modeling Health Care Reform What States Need to
Know

• Amy Lischko
• Assistant Commissioner, DHCFP Massachusetts

2
Todays presentation

• Goals/purpose of MA analysis
• Questions
• Data
• Partners
• Sample output
• Challenges

3
Goals of Analysis

• To analyze comprehensive models for increasing
  access to health insurance in Massachusetts by
  estimating
• Costs
• State and Federal
• Other
• Movements (including crowd-out)
• Participation rates
• Revenue estimates

4
Questions

• What options do you want to evaluate?
• Comprehensive vs. incremental
• What data are available?
• State, Federal, other
• What output do you require? Who is your
  audience?
• Does the work require outside expertise?
• How much can you do yourself?
• credibility

5
Data

• State
• Individual Data (age, insurance status,
  employment, income, family status, health status)
• Employer Data (average price of plan,
  contribution, offer by size)
• Insurance Market (price of various plans in each
  market, benefit coverage)
• Medicaid and other public program cost data

6
Data

• Federal
• CPS (adjusted for missing income data)
• MEPS (validated employer data)
• Other
• Behavioral responses from literature
• Massachusetts-specific behavioral responses

7
Partners

• Other agency staff
• Actuaries (Mercer, Lewin)
• price alternative health insurance plans
• Economist (J. Gruber)
• modeling

8
Actuary output
9
Econometric Model Output
10
Challenges

• Data
• Missing values become very important
• Never used data for policy-making purposes per
  se, gaps in understanding was more apparent
• You know your state better than national experts,
  check carefully and ask lots of questions!
• Assumptions
• Modeling is an art not a science, make sure
  youre involved and know what assumptions are
  being made.
• National responses from the literature may not
  always be pertinent to your state, again,
  question assumptions!

11
A Few Final Words

• We had a lot of our own data so it was easier to
  manage this process than if youre using federal
  data sets that you dont know very well
• Once you have a model, lots of interest to use it
  
• Own your model. We dont have staff trained to
  use it, costs more money to manipulate it.