The abacus is based on a 25 percent sample of the 2013 Medical Expenditure Panel Survey (MEPS), which is a set of large-scale surveys of families and individuals, their medical providers, and employers across the United States. MEPS collects data on the specific health services that Americans use, how frequently they use them, the cost of these services, and how they are paid for, as well as data on the cost, scope, and breadth of health insurance held by and available to U.S. workers. The MEPS data allow the model to microsimulate the relationship between the various elements of insurance benefit design based on real world observations.
Specifically, each person within the data has a count of services received and a dollar value attributed to those services. The abacus takes the user’s input parameters and interacts them with the data at the person-level. Then the total value of each payment bucket (e.g. out of pocket spend) is aggregated across all observations. Output values are represented “per member per month” which takes the total spend across all individuals and divides it by the number of months of insurance coverage within the sample. Participants used within the model were required to have twelve months of private health insurance coverage (either individual, family, or a combination of the two). The model also utilizes the National Health Expenditure (NHE) data to convert the 2013 MEPS data into 2016 dollars.
The model breaks out various components of the premium (i.e., the impact of hospital, physician office, prescription drug, and other services on the premium). To do this, the model utilizes a step-wise calculation.
- First, it finds the baseline proportions of the premium components from the raw MEPS data.
- Then, it calculates the percentage change of the input variables from the default setting (which is based on Kaiser Family Foundation’s summary of 2016 silver metal level plans offered on the health insurance Exchanges[i],[ii]).
- This percentage change is then multiplied by the elasticity function, which assumes that a hypothetical decrease in cost of a variable by 10% will result in a 3% increase in utilization of that variable. Elasticity varies for different types of services and different populations; however, studies[iii] have shown that elasticity for healthcare services can range from -0.04 to -0.75. The abacus utilized a mid-point of this range which is keeping with assumptions used by CBO and others in other areas of health services research.
- Changes due to the elasticity are then multiplied by the default proportions—based on NHE data—and normalized to ensure a final sum of 100 percent.
Finally, the premium components are adjusted for the “Network Scope,” as more restrictive provider networks are likely to have a cost controlling impact than those that are less restrictive.
[i] Rae, M., L. Levitt, G. Claxton, C. Cox, M. Long, and A. Damico. 2015 (November 13). Patient Cost-Sharing in Marketplace Plans, 2016. The Henry J. Kaiser Family Foundation, available at: http://kff.org/health-costs/issue-brief/patient-cost-sharing-in-marketplace-plans-2016/ (last accessed November 8, 2016).
[ii] The Kaiser Family Foundation research only summarizes the in-network silver plans. Out of network default settings were estimated using the following method: it was assumed that using 100% out of network services would be at least 25% more expensive than 100% in-network utilization. The model used the default silver plan in-network settings and the outputs were calculated for 100% in-network utilization. These outputs were then raised by 25% to create a target output. Then the input parameters were adjusted until the target-outputs were calculated.
[iii] See Ringel J.S., S. D. Hosek, B. A. Vollaard, S. Mahnovski. 2002. The Elasticity of Demand for Health Care: A Review of the Literature and Its Application to the Military Health System. Table 3.1. Rand Corporation.