120 likes | 312 Views
Goals of Modeling Health Care Reform. To analyze comprehensive models for increasing access to health insuranceElements to consider in designing reformCostsState and FederalOtherMovement (including crowd-out)Participation ratesRevenue estimates. Source: A. Lischko,
E N D
1. Modeling Health Care Reform An Overview of Jonathan Gruber’s Microsimulation Model Alyssa Holmgren
Presentation to the Finance Committee
November 19, 2007 In advance of contracting with a consultant to do the economic modeling for Oregon, we wanted to give you a little background on the modeling that other states have done while they were preparing to propose or implement their own health reform plans.
There aren’t many economists out there who do this type of modeling, and since Dr. Gruber has worked in both California and Massachusetts, I’ve decided to focus on his model as a template for the type of work we might want to have done. In advance of contracting with a consultant to do the economic modeling for Oregon, we wanted to give you a little background on the modeling that other states have done while they were preparing to propose or implement their own health reform plans.
There aren’t many economists out there who do this type of modeling, and since Dr. Gruber has worked in both California and Massachusetts, I’ve decided to focus on his model as a template for the type of work we might want to have done.
2. Goals of Modeling Health Care Reform To analyze comprehensive models for increasing access to health insurance
Elements to consider in designing reform
Costs
State and Federal
Other
Movement (including crowd-out)
Participation rates
Revenue estimates So, the first question to ask, is what are the goals of modeling health care reform? On the most basic level, we want to be able to analyze comprehensive models for increasing access to health insurance. We want to look at what effects the reform plan will have on costs – at both the state level and the federal level, as well as how costs will be affected for individuals, families, employers,…
We also want to look at the effect of the reform plan on movement from one plan to another such as from uninsurance to non-group insurance, or from employer coverage to public insurance (This is what’s known as crowd-out).
It will also be important to keep an eye on participation rates in the different plans, as well as revenue estimates for the state. We want this reform to be sustainable, even in long run.So, the first question to ask, is what are the goals of modeling health care reform? On the most basic level, we want to be able to analyze comprehensive models for increasing access to health insurance. We want to look at what effects the reform plan will have on costs – at both the state level and the federal level, as well as how costs will be affected for individuals, families, employers,…
We also want to look at the effect of the reform plan on movement from one plan to another such as from uninsurance to non-group insurance, or from employer coverage to public insurance (This is what’s known as crowd-out).
It will also be important to keep an eye on participation rates in the different plans, as well as revenue estimates for the state. We want this reform to be sustainable, even in long run.
3. Questions to Answer before Modeling Can Begin What policy options do you want to evaluate?
What output do you require? Who is your audience?
What data are available?
State, Federal, other
Does the work require outside expertise? But first, before we can achieve these goals, some decisions need to be made. Before we can model a reform plan, we need to have an idea what the reform plan will look like.
For example –
Are we going to assume public coverage for adults up to 200% FPL?
What about an individual mandate?
What role is the insurance exchange going to play?
These are the types of questions we need to be looking at.
We also need to determine what output we want from the model. What particular questions do we want the model to answer? Who is the audience for this output, and what are the different stakeholders going to want to know in order to back the plan?
Next we’ll need to look at what data is out there for us to use. And to determine if the work requires outside expertise, which I think we’ve determined it does. This is important for the sake of credibility.But first, before we can achieve these goals, some decisions need to be made. Before we can model a reform plan, we need to have an idea what the reform plan will look like.
For example –
Are we going to assume public coverage for adults up to 200% FPL?
What about an individual mandate?
What role is the insurance exchange going to play?
These are the types of questions we need to be looking at.
We also need to determine what output we want from the model. What particular questions do we want the model to answer? Who is the audience for this output, and what are the different stakeholders going to want to know in order to back the plan?
Next we’ll need to look at what data is out there for us to use. And to determine if the work requires outside expertise, which I think we’ve determined it does. This is important for the sake of credibility.
4. The Gruber Microsimulation Model Shows how policy changes affect the economy
Inputs
Policy parameters
Outputs
Impact on public sector costs
Distribution of insurance coverage
Effect on public sector revenues
Similar to approach used by Treasury Department, CBO, and other government entities Turning now to the specifics of Dr. Gruber’s model, he broadly shows how policy changes affect the economy. He models how individuals will respond to changes in the insurance marketplace that are the result of changes in government policy. And to do that, he inputs certain policy parameters that we give him and then outputs the impact on public sector costs, the distribution of insurance coverage, and the effect on public sector revenues.
Going back to the credibility issue, the approach Dr. Gruber uses in this model is similar to the approach used by the Treasury Dept, the Congressional Budget Office, and other government entities.Turning now to the specifics of Dr. Gruber’s model, he broadly shows how policy changes affect the economy. He models how individuals will respond to changes in the insurance marketplace that are the result of changes in government policy. And to do that, he inputs certain policy parameters that we give him and then outputs the impact on public sector costs, the distribution of insurance coverage, and the effect on public sector revenues.
Going back to the credibility issue, the approach Dr. Gruber uses in this model is similar to the approach used by the Treasury Dept, the Congressional Budget Office, and other government entities.
5. The Gruber Microsimulation Model Demonstrates the effect of government interventions in health insurance markets
Impact of tax subsidies on employer insurance
Impact of public coverage on private insurance take-up (crowd-out)
Impact of lower health insurance costs for employers on employee wages More specifically, this model demonstrates the effect of government interventions in health insurance markets in the following ways:
It shows the impact of tax subsidies on employer insurance -- How does tax subsidizing insurance for employers affect their decision to offer or not offer insurance to their employees?
It looks at the impact of public coverage on private insurance take-up. To what extent does offering public insurance to people who are currently privately insured cause them to leave their private insurance for public (crowd out)?
And what is the impact of lower health insurance costs for employers on employee wages? To what extent will lower health insurance costs for firms lead to higher wages for workers in those firms? Will the savings be passed on, or not?
More specifically, this model demonstrates the effect of government interventions in health insurance markets in the following ways:
It shows the impact of tax subsidies on employer insurance -- How does tax subsidizing insurance for employers affect their decision to offer or not offer insurance to their employees?
It looks at the impact of public coverage on private insurance take-up. To what extent does offering public insurance to people who are currently privately insured cause them to leave their private insurance for public (crowd out)?
And what is the impact of lower health insurance costs for employers on employee wages? To what extent will lower health insurance costs for firms lead to higher wages for workers in those firms? Will the savings be passed on, or not?
6. Strengths of the Model Holistic approach
Considers the effect of interventions on all firms and individuals in the state
Focuses on price
Can address multiple integrated policy approaches by converting their effects into price changes
Determines how firms and individuals will react to those price changes based on behavioral evidence from health economics
Assesses firm reactions by looking at a set of workers within the firm and aggregating impacts to the firm level So, what are the strengths of this particular model?
First of all, it looks holistically at the impact of the reform plan on all firms and all individuals in the state. It does not take a piecemeal approach.
Second, by converting the effects of multiple integrated policy approaches into price changes, the model allows us to aggregate the effects of all the different pieces of the reform plan. And then, using behavioral evidence from health economics, we can examine how firms and individuals will react to the changes in price.
For example, let’s say we’re going to expand public insurance to adults up to 200% of the federal poverty level -- this model will let us see to what extent the uninsured enroll in public insurance, to what extent individuals in the non-group market move into public insurance, the extent to which people with more expensive employer-sponsored insurance drop it and move into public coverage, and how many employers drop coverage for their employees or raise their employees’ contributions.
And finally, this is a bit of a technicality, but Gruber’s model assesses firm reactions by looking at a set of workers within the firm and aggregating the impacts on individuals’ choice to the firm level. So, what are the strengths of this particular model?
First of all, it looks holistically at the impact of the reform plan on all firms and all individuals in the state. It does not take a piecemeal approach.
Second, by converting the effects of multiple integrated policy approaches into price changes, the model allows us to aggregate the effects of all the different pieces of the reform plan. And then, using behavioral evidence from health economics, we can examine how firms and individuals will react to the changes in price.
For example, let’s say we’re going to expand public insurance to adults up to 200% of the federal poverty level -- this model will let us see to what extent the uninsured enroll in public insurance, to what extent individuals in the non-group market move into public insurance, the extent to which people with more expensive employer-sponsored insurance drop it and move into public coverage, and how many employers drop coverage for their employees or raise their employees’ contributions.
And finally, this is a bit of a technicality, but Gruber’s model assesses firm reactions by looking at a set of workers within the firm and aggregating the impacts on individuals’ choice to the firm level.
7. Potential Weaknesses of the Model Most reliable when comparing similar proposals that only differ along a small number of dimensions
Relative impacts are more reliable than absolute impacts
Estimates become more uncertain as we depart farther from existing experience in the insurance market
Some of the potential weaknesses of the model are that it’s
most reliable when comparing similar proposals that only differ along a small number of dimensions (apples to apples)
that relative impacts are more reliable than absolute impacts (impacts -- One reform proposal vs. another is more reliable than comparing either proposal to no reform)
and finally, that estimates become more uncertain as we depart farther from existing experience in the insurance market
Some of the potential weaknesses of the model are that it’s
most reliable when comparing similar proposals that only differ along a small number of dimensions (apples to apples)
that relative impacts are more reliable than absolute impacts (impacts -- One reform proposal vs. another is more reliable than comparing either proposal to no reform)
and finally, that estimates become more uncertain as we depart farther from existing experience in the insurance market
8. Data Needed for Modeling Reform State
Individual Data
Employer Data
Insurance Market
Medicaid and other public program cost data
Federal
Current Population Survey (CPS)
Medical Expenditure Panel Survey (MEPS)
Other
Behavioral responses from literature
State-specific behavioral responses Moving on to figuring out what Dr. Gruber or whichever consultant we contract with is going to need:
Individual data – age, insurance status, employment, income, family status, health status
Employer data – average price of plan, percent contribution by employers/employees, offer rate by firm size
Insurance market – price of various plans in each market, benefit coverage
CPS – adjusted for missing income data
MEPS – validated employer data
And finally, it’s going to be important to have some state-specific data on behavioral responses. Oregonians are not Californians, not from Massachusetts. And we want to how they’re different.Moving on to figuring out what Dr. Gruber or whichever consultant we contract with is going to need:
Individual data – age, insurance status, employment, income, family status, health status
Employer data – average price of plan, percent contribution by employers/employees, offer rate by firm size
Insurance market – price of various plans in each market, benefit coverage
CPS – adjusted for missing income data
MEPS – validated employer data
And finally, it’s going to be important to have some state-specific data on behavioral responses. Oregonians are not Californians, not from Massachusetts. And we want to how they’re different.
9. California-Specific Model California sample of the Feb/Mar 2001 CPS
Updated with 2005 CA Health Insurance Survey data
Premiums, cost-sharing, and offer rates by firm size updated with 2004 California-specific MEPS data, updated to 2007
Cost of non-group policies difficult to determine Finally, to close out, I want to give you a little snapshot of how the process worked in California.
Data-wise, Dr. Gruber had access to a California sample of CPS data from 2001, which he updated to 2005 using the CA Health Insurance survey.
He had data on premiums, cost-sharing, and offer rates by firm size that had been updated with 2004 CA-specific MEPS data, which was further updated to 2007.
Cost of non-group policies –
Assumption based on a typical policy holder who is a 40-44 year old male who pays $300/month
Data adjusted up or down from there based on age, gender, and health status Finally, to close out, I want to give you a little snapshot of how the process worked in California.
Data-wise, Dr. Gruber had access to a California sample of CPS data from 2001, which he updated to 2005 using the CA Health Insurance survey.
He had data on premiums, cost-sharing, and offer rates by firm size that had been updated with 2004 CA-specific MEPS data, which was further updated to 2007.
Cost of non-group policies –
Assumption based on a typical policy holder who is a 40-44 year old male who pays $300/month
Data adjusted up or down from there based on age, gender, and health status
10. CA Specific Model – Policy Decisions Public insurance expansion
To what level?
Adults and kids?
Documented and undocumented?
Central purchasing mechanism
Who is eligible?
Cost of policies available?
Cost sharing required?
Individual mandate?
What qualifies as the minimum coverage required?
Effective for both documented and undocumented populations?
Non-offering assessment
Which employers pay?
How much?
11. Population Flows Pre- and Post-Reform Population flows for Children and Documented Adults
Columns show the old source of insurance
Rows shows the new source – post-reformPopulation flows for Children and Documented Adults
Columns show the old source of insurance
Rows shows the new source – post-reform
12. Net Changes in Insurance Status Due to Reform Shows for the entire Under 65 population, where they were before Shows for the entire Under 65 population, where they were before
13. Effects of Reform on Public Spending