Inner TRIM3 Masthead
TRIM3 Logo

Employer-Sponsored Health Insurance version 13.2

Version History

The EmpSponsoredHealthIns module imputes information about employer sponsored health insurance (ESI) coverage and costs, information that is needed by TRIM for federal tax simulations and poverty analyses but is not available from the March CPS input data. The module

  • Defines health insurance units and plan type for covered workers and family members
  • Determines which workers have own employer-sponsored health insurance coverage
  • Imputes whether a worker's employer offers health insurance to any workers
  • Imputes whether a worker's employer offers health insurance to the individual worker
  • Imputes whether a worker is subject to a waiting period after starting a new job
  • Matches workers with offers to premium data contained in health insurance donor files
  • On the basis of the match with premium data, determines worker and employer contribution amounts and policy characteristics
  • Produces summary tables of results

This document provides information on the following topics:

Identifying Health Insurance Units and Coverage

The EmpSponsoredHealthIns unit is the family unit that corresponds to the FamilySubfamiliesSeparate definition, modified to place unidentified persons 15+ years of age and adult children in their own separate health insurance units. [See the PersonType definitions below.]

Person Type

PersonType is set for every person to one of the following values:

  • Unidentified person; someone not in one of the other categories
  • Family head
  • Family spouse
  • Dependent child
  • Insured dependent child
  • Dependent child insured by a person outside the household
  • Adult child

Family heads are identified by HhFamilyRelation codes of 1, 3, 5, 8, 10, 14, 17, 20, 22, 25. Spouses are identified by HhFamilyRelation codes of 2, 6, 11, 18, 23, and 26.

A dependent child is a person of age DepHealthInsAgeNonStudent or younger, and whose HhFamilyRelation is 4, 7, 9, 12, 13, 15 or 27. A child, who is of age DepHealthInsAgeStudent or younger and a student (InSchoolFullOrPartTime = "In school full time" or (InSchoolFullOrPartTime != "In school part time" and (LastRemainingActivity = "Going to school part year" or LastReasonNotWorking = "Going to school last year"))) is also coded a dependent child. DepHealthInsAgeNonStudent and DepHealthInsAgeStudent should be set to 18 and 22, respectively for years prior to 2010. Beginning with the year 2010, both rules should be set to 25 (according to the Affordable Care Act (ACA)).

An insured dependent child is a dependent child who is a member of a family/health insurance unit whose head or spouse has health insurance coverage (family or dual) that includes the child.

A dependent child insured by a person outside the household is a person whose HhFamilyRelation code is none of the ones listed above (i.e., 16, 19, 21, 24, 28-37), who is of age DepHealthInsAgeNonStudent or younger, and, for simulation years < 1994, HealthChildCoveredbyOther = 2, or for later years, HealthCoveredOutsideHh = 1.

An adult child is a person whose HhFamilyRelation = 4, 7, 9, 12, 13, 15 or 17, but who does not satisfy the age and school conditions that define a child as a dependent.

AnnualContractType

  • AnnualContractType is set to one of the following values for all persons:
  • Person is a non-worker or is not covered by his/her own employer
  • Person has a family contract sponsored by his/her own employer
  • Person has a dual contract sponsored by his/her own employer
  • Person has a single contract sponsored by his/her own employer

The setting of AnnualContractType defaults to 0 unless one of the following sets of conditions is met.

Family Contract

AnnualContractType is set to family if a health insurance plan is offered through one’s current or former employer or union (HealthEmployerUnionPlan = 1) and the person covers a spouse, or spouse and children, or other combination of persons (HealthWhoElseInGroup = 1, 3, or 5). [AnnualContractType may also be set to family under conditions described in the "Dual Contract" and "Single Contract" sections that follow.]

Dual Contract

AnnualContractType is set to dual if a health insurance plan is offered through one’s current or former employer or union (HealthEmployerUnionPlan = 1) and the person covers child(ren) (HealthWhoElseInGroup = 2), is single (MaritalStatus = 1, 6, or 7), and has just one dependent child (Age < 19 and HhFamilyRelation = 4, 7, 9, 12, 13, 15, or 27). If the person meets all preceeding conditions except he/she has more than one dependent child, AnnualContractType is set to family.

Single Contract

A person’s AnnualContractType is set to single if a health insurance plan is offered through one’s current or former employer or union (HealthEmployerUnionPlan = 1) but other conditions for family and dual contracts are not met. However, if a person also covers children outside the household (HealthChildrenOutsideHhCovered = 1 for simulation years < 1994 or HealthOwnGroupCoversOutsideHh = 1 for simulation years >= 1994), then AnnualContractType is set to family, allowing for the possibility that more than one child is covered outside the household.

ESIContractType

The monthly field, ESIContractType, has the same codes as does AnnualContractType (i.e., NIU or not covered, family, dual, or single). It is set equal to AnnualContractType in months in which a person works (MonthlyWeeksWorked > 0) and is not subject to a waiting period (WaitPeriodMonthly = 0); else, it is set to "Not covered."

IsInsuredNonWorker

The monthly field, IsInsuredNonWorker, is a binary indicator of whether a person has ESI coverage during a month in which he/she does not work. It is set to true for any month MonthlyWeeksWorked = 0 and AnnualContractType > 0.

CoveredByEmpHlthIns

The monthly field, CoveredByEmpHlthIns, is a binary indicator of whether or not a person is covered by employer sponsored health insurance. It is initially set to true if HealthGroupCoverage = 1 or a person is coded as an insured dependent child.

However, if a person is an insured non-worker family head or spouse (IsInsuredNonWorker = 1) who does not have a currently (this month) working spouse with family coverage, then CoveredByEmpHlthIns is set to false.

It is also be set to false for dependent children if any of the following sets of conditions is met:

  • the child is an insured non-worker and neither the family head nor spouse has a family contract;
  • the family head has family coverage but is an insured non-worker and the family spouse has no family coverage;
  • the family spouse has family coverage but is an insured non-worker and the family head has no family coverage.

For single heads and all other persons not discussed above, CoveredByEmpHlthIns is set to false in any month the person is an insured non-worker.

EmployerCoverage

    The monthly field, EmployerCoverage, is set to one of the following values for all persons:

  • NIU, or not covered by own employer
  • Covered by own employer
  • Covered by other employer
  • Insured non-worker

EmployerCoverage defaults to 0. It is set to "Covered by own employer" for any month ESIContractType indicates family, dual, or single coverage. Else if the monthly field, CoveredByEmpHlthIns, is set to true, then EmployerCoverage is coded "Covered by other employer."

Waiting Periods

Kaiser/HRET data show that a significant number of workers are subject to a waiting period for employer sponsored health insurance (ESI) enrollment when starting a new job. During this waiting period (two months on average) the new worker is not covered by ESI.

TRIM imputes the length of the waiting period at random based on a combination of region, industry and firm size. Modeling of waiting periods is turned off by default. Set the program rule WaitingPeriod to "Waiting period is modeled" if the imputation of waiting periods is desired.

Identifying workers subject to waiting period

People who are subject to a waiting period are identified as follows:

  • They have at least one month of work (MonthlyWeeksWorked > 0) that follows a month of not working (MonthlyWeeksWorked = 0).
  • They report annual ESI coverage.
Unit members that are dependents of the person that is subject to a waiting period are affected by the waiting period in the same way as the policy holder.

Identifying dependents

Dependents of a worker, who is waiting to be eligible for ESI, are not covered by ESI during the wait period as well (ESIContractType="Not covered"), unless there is a second parent with a family contract. If there are two parents with family contracts in a unit, a child is covered by the parent that is covered by ESI the longest. If the number of months covered is the same for both parents, the child is covered by the family head. Parents are identified by the parent ids provided in the variablelist rule BioOrAdoptParentIDs.

Imputing the length of the waiting period

A "new job month" is a month of work following a month of not working. If it is determined that a person is in a "new job month", a wait period is imputed from firm size, industry type and region. In order to determine whether a "new job month" that would otherwise be covered by ESI should be subject to a waiting period, TRIM uses a lookup database of cumulative probabilities (ranging from 0 to 1). This database consists of a matrix of values for 4 regions, 5 firm sizes, 8 industries and 11 waiting period months. The fields are described below.

Waiting Period
The Waiting Period can be zero to 10 months long. There has to be at least one month coverage if the person reported annual ESI coverage and the first month of the year is never a waiting month.
Region
Region index is set equal to a person’s Region code, one of the following.

  • Northeast
  • Midwest
  • South
  • West
Firm Size
The firm size index is set according to a worker's firm size reported by LastFirmSize.

  • 1-10
  • 10-24
  • 25-99
  • 100-499
  • 500+
Industry
The industry index is set as follows according to LastWorkerClass and LastIndustryGroupRecode2002. Government workers are first identified using LastWorkerClass, and remaining workers are classified on the basis of LastIndustryGroupRecode2002.
  • Agriculture, mining, construction (LastIndustryGroupRecode2002 <= 3)
  • Manufacturing (LastIndustryGroupRecode2002 = 4-5)
  • Transportation, utilities, information (LastIndustryGroupRecode2002 = 8-10)
  • Wholesale or retail trade (LastIndustryGroupRecode2002 6-7)
  • Finance, insurance, real estate (LastIndustryGroupRecode2002 11-12)
  • Services (LastIndustryGroupRecode2002 13-20)
  • State or local government (LastWorkerClass = 4 or LastIndustryGroupRecode2002 21-22)
  • Federal government (LastWorkerClass = 2)

A random number is generated for every "new job month" and compared to the cumulative probabilities in the Lookup Database to determine the length of the waiting period. If the waiting period is 0 months, then no waiting period is imposed, and the worker (and his/her dependents if applicable) is covered by ESI for all months worked consecutively after the "new job month". If the selected waiting period is 1 or more months, then the worker (and his/her dependents) are not covered by ESI for the length of the waiting period. During a waiting period CoveredByEmpHealth is set to 0 and WaitPeriodMonthly is set to 1. If the waiting period takes out all months of ESI coverage, the waiting period is shortened by one month, and the last month of the year a person worked is set to covered. The annual result variable AnnualWaitPeriod stores the number of months a person is subject to a waiting period for ESI during the year.

ESI Offer Imputation

Even if a worker is not covered by ESI through his/her own employer, the employer may offer coverage. EmpSponsoredHealthIns separately imputes for all workers whether ESI is offered to any employees and whether ESI is offered to an individual worker.

Imputation of Whether an Employer Offers to Any Employees

The annual output field, IsESIOffered, indicates whether a worker's employer offers ESI to any employees. It is set to true for all persons who are federal or state government workers and for other persons in the work force whose AnnualContractType indicates some type of ESI coverage. For all other persons in the work force, it is set on the basis of a probabilistic logit equation.

For workers subject to the probabilistic determination, a coefficient sum is computed, a probability of participation is derived from the coefficient sum, then the probability is compared with the worker’s RandomNumHealthInsOffer. If the random number is less than or equal to the probability, then IsESIOffered is set to true, otherwise, it is set to false.

Coefficient components are computed as follows.

Wage and Salary Income

TotalWages are converted to a 1993 equivalent using an adjustment factor based on Consumer Price Indexes stored within TRIM. The 1993 equivalent wage is divided by LastWeeksWorked, multiplied by 0.01, then multiplied by the WeeksEarningsCoef program rule. [The WeeksEarningsCoef is an array rule having a single element.] If TotalWages is zero, the wage component is also zero.

The resulting wage component is the initial coefficient sum.

Age

A person with Age >= 30 is assigned an age index of 1; otherwise, his/her age index is 0. The coefficient sum is incremented by the AgeCoef array rule indexed by age index.

Dependent Status

The dependent index is set to 0 if a person is coded as a dependent child with Age < 22. Otherwise, the dependent index is set to 1. The coefficient sum is incremented by the DependentCoef array rule indexed by dependent index.

Education

The education index is set to 0 if HighestGradeCompleted = 13, or to 1 if HighestGradeCompleted = 14, or to 2 if HighestGradeCompleted >= 15. Otherwise, the education index is set to 3. The coefficient sum is incremented by the EducationCoef array rule indexed by education index.

Hispanic Origin

The Hispanic index is set to 1 if HispanicOrigin = 1; else Hispanic index is set to 0. The coefficient sum is incremented by the HispanicCoef array rule indexed by the Hispanic index.

Race

The race index is set to according to DetailedRace2002 to one of the following.

  • White
  • Black
  • American Indian, Alaskan
  • Asian, Pacific Islander

The coefficient sum is incremented by the RaceCoef array rule indexed by the race index.

Foreign Born

The foreign born index is set to according to Citizenship. If Citizenship is >= 4, it is set to 1; otherwise it is set to zero. The coefficient sum is incremented by the ForeignBornCoef array rule indexed by the foreign born index.

Part-time Worker

The part-time worker index is set to according to LastHoursPerWeek. If LastHoursPerWeek is < 35, it is set to 1; otherwise it is set to zero. The coefficient sum is incremented by the PartTimeWorkerCoef array rule indexed by the part-time worker index.

Occupation

The occupation index is set to according to LastMajorOccupation2002 to one of the following.

  • Managerial, professional (LastMajorOccupation2002 = 1-10)
  • Technical, sales, administration (LastMajorOccupation2002 = 16 or 17)
  • Service (LastMajorOccupation2002 = 11-15)
  • Farming, forestry, fishing (LastMajorOccupation2002 = 18)
  • Precision production, craft, repair (LastMajorOccupation2002 = 19-20)
  • Operator, fabricator, laborer (LastMajorOccupation2002 = 21-22)

The coefficient sum is incremented by the OccupationCoef array rule indexed by the occupation index.

Firm Size

The firm size index is set according to a worker's firm size reported by LastFirmSize. Firm size for all state and federal workers is assumed to be in excess of 100 workers (i.e., firm size index = 4). However, government workers are not subject to the imputation of whether their employers offer health insurance to some workers. Rather, such offers are assumed. (Government workers are, however, subject to the imputation of whether their employers offer coverage to the individual worker.)

  • 1-10
  • 10-24
  • 25-99
  • 100+

The coefficient sum is incremented by the FirmSizeCoef array rule indexed by firm size index.

Industry

The industry index is set as follows according to LastWorkerClass and LastIndustryGroupRecode2002. Government workers are first identified using LastWorkerClass, and remaining workers are classified on the basis of LastIndustryGroupRecode2002.

  • Agriculture, mining, construction (LastIndustryGroupRecode2002 <= 3)
  • Manufacturing (LastIndustryGroupRecode2002 = 4-5)
  • Transportation, utilities, information (LastIndustryGroupRecode2002 = 8-10)
  • Wholesale or retail trade (LastIndustryGroupRecode2002 6-7)
  • Finance, insurance, real estate (LastIndustryGroupRecode2002 11-12)
  • Services (LastIndustryGroupRecode2002 13-20)
  • State or local government (LastWorkerClass = 4 or LastIndustryGroupRecode2002 21-22)
  • Federal government (LastWorkerClass = 2)

The coefficient sum is incremented by the IndustryCoef array rule indexed by industry index.

Region

Region index is set equal to a person’s Region code, one of the following.

  • Northeast
  • Midwest
  • South
  • West

The coefficient sum is incremented by the RegionCoef array rule indexed by region index.

Gender

Gender index is 0 for males (RaceAndSex = 1-3) and 1 for females (RaceAndSex = 4-6). The coefficient sum is incremented by the MaleCoef array rule indexed by gender index.

Spouse Coverage

If a person is the family head and his/her spouse is covered by a family contract (AnnualContractType[spouse] = 1), spouse coverage index is set to 0. Else if the person is the family spouse and his/her spouse is covered by a family contract, spouse coverage index is set to 1. Else, spouse coverage index is set to 2.

The coefficient sum is incremented by the SpouseCoef array rule indexed by spouse coverage index.

Constant

Finally, the coefficient sum in incremented by a constant contained in the array program rule, ConstantCoef, which has but a single element.

Conversion of Coefficient Sum to Probability

The coefficient sum (z) is converted to a probability between 0 and 1 with the following steps:

  • x = exp(z)
  • y = x+1
  • probability = x / y
  • if probability > 1, then probability = 1

Imputation of Whether an Employer Offers to an Individual Employee

The annual output field, IsESIOfferedIndividual, indicates whether a worker's employer offers ESI to the individual worker. If a worker's HealthEmployerUnionPlan indicates the worker has employer health insurance, then IsESIOfferedIndividual is set to covered. Individual coverage for other workers whose employers are reported or imputed to offer coverage to some employees is imputed.

For workers subject to the probabilistic determination, a coefficient sum is computed, a probability of participation is derived from the coefficient sum, then the probability is compared with the worker’s RandomNumHealthInsOffer. If the random number is less than or equal to the probability, then IsESIOfferedIndividual is set to true, otherwise, it is set to false.

Coefficient components are computed as follows.

Wage and Salary Income

TotalWages are converted to a 1993 equivalent using an adjustment factor based on Consumer Price Indexes stored within TRIM. The 1993 equivalent wage is divided by LastWeeksWorked, multiplied by 0.01, then multiplied by the WeeksEarningsCoef2 program rule. [The WeeksEarningsCoef2 is an array rule having a single element.] If TotalWages is zero, the wage component is also zero.

The resulting wage component is the initial coefficient sum.

Age

A person with Age >= 30 is assigned an age index of 1; otherwise, his/her age index is 0. The coefficient sum is incremented by the AgeCoef2 array rule indexed by age index.

Dependent Status

The dependent index is set to 0 if a person is coded as a dependent child with Age < 22. Otherwise, the dependent index is set to 1. The coefficient sum is incremented by the DependentCoef2 array rule indexed by dependent index.

Education

The education index is set to 0 if HighestGradeCompleted = 13, or to 1 if HighestGradeCompleted = 14, or to 2 if HighestGradeCompleted >= 15. Otherwise, the education index is set to 3. The coefficient sum is incremented by the EducationCoef2 array rule indexed by education index.

Part-time Worker

The part-time worker index is set to according to LastHoursPerWeek. If LastHoursPerWeek is < 35, it is set to 1; otherwise it is set to zero. The coefficient sum is incremented by the PartTimeWorkerCoef2 array rule indexed by the part-time worker index.

Firm Size

The firm size index is set according to a worker's firm size reported by LastFirmSize. Firm size for all state and federal workers is assumed to be in excess of 100 workers (i.e., firm size index = 4).

  • 1-10
  • 10-24
  • 25-99
  • 100+

The coefficient sum is incremented by the FirmSizeCoef2 array rule indexed by firm size index.

Industry

The industry index is set as follows according to LastWorkerClass and LastIndustryGroupRecode2002. Government workers are first identified using LastWorkerClass, and remaining workers are classified on the basis of LastIndustryGroupRecode2002.

  • Agriculture, mining, construction (LastIndustryGroupRecode2002 <= 3)
  • Manufacturing (LastIndustryGroupRecode2002 = 4-5)
  • Transportation, utilities, information (LastIndustryGroupRecode2002 = 8-10)
  • Wholesale or retail trade (LastIndustryGroupRecode2002 6-7)
  • Finance, insurance, real estate (LastIndustryGroupRecode2002 11-12)
  • Services (LastIndustryGroupRecode2002 13-20)
  • State or local government (LastWorkerClass = 4 or LastIndustryGroupRecode2002 21-22)
  • Federal government (LastWorkerClass = 2)

The coefficient sum is incremented by the IndustryCoef2 array rule indexed by industry index.

Region

Region index is set equal to a person’s Region code, one of the following.

  • Northeast
  • Midwest
  • South
  • West

The coefficient sum is incremented by the RegionCoef2 array rule indexed by region index.

Spouse Coverage

If a person is the family head and his/her spouse is covered by a family contract (AnnualContractType[spouse] = 1), spouse coverage index is set to 0. Else if the person is the family spouse and his/her spouse is covered by a family contract, spouse coverage index is set to 1. Else, spouse coverage index is set to 2.

The coefficient sum is incremented by the SpouseCoef2 array rule indexed by spouse coverage index.

Constant

Finally, the coefficient sum in incremented by a constant contained by the array program rule, ConstantCoef2, which has but a single element.

Conversion of Coefficient Sum to Probability

The coefficient sum (z) is converted to a probability between 0 and 1 with the following steps:

  • x = exp(z)
  • y = x+1
  • probability = x / y
  • if probability > 1, then probability = 1

Match with Donor Premiums

Health insurance premium data are available in "donor" files that are matched to all workers who are offered ESI by their employers (IsESIOfferedIndividual = 1). Workers are matched to donor records in the tables selected by users via program rules PrivatePremiumDonorFile, StatePremiumDonorFile, and FederalPremiumDonorFile. Users may choose one of three state government (1995, 1999 or 2002), federal government (1995, 2001 or 2003) and private premium (1996, 2000 or 2003) donor files. Tables are tagged with a year designation to indicate the year for which data were collected. Adjustment factors contained by EmpSponsoredHealthIns program rules AdjPrivate, AdjState, and AdjFederal are used to adjust premium amounts by an inflation/deflation factor appropriate for the CPS data year that is being simulated. A factor of 1 makes no adjustment; a factor less that 1 deflates the premiums, and a factor greater than 1 inflates the premiums (e.g., a factor of 1.099 inflates all premium amounts by 9.9%). Table contents are described below.

Contents of Private Health Insurance Donor Files

The tables of private employer sponsored health insurance premiums contain representative data for private and local government workers that were obtained from KPMG for 1996 and HRET/Kaiser for 2000. The tables include the following fields.

Region

  • Northeast
  • Midwest
  • South
  • West

PlanType

  • Conventional
  • HMO
  • PPO (Preferred provider)
  • POS (Point of service)

IndustryGrp (KPMG 1996)

  • Manufacturing and High Tech
  • Services and Health
  • Local Government, Finance, Insurance and Real Estate
  • All Other

IndustryGrp (HRET/Kaiser)

  • 1 = Mining and Construction
  • 2 = Manufacturing
  • 3 = Transportation, Communication, Utilities
  • 4 = Wholesale Trade
  • 5 = Retail Trade
  • 6 = Finance, Insurance, Real Estate
  • 7 = Services and Health
  • 8 = Local Government

FirmSizeGrp(KPMG 1996)

  • 1 = 1-99
  • 2 = 100-499
  • 3 = 500+

FirmSizeGrp (HRET/Kaiser)

  • 1 = 1-24
  • 2 = 25-99
  • 3 = 100-499
  • 4 = 500-999
  • 5 = 1000+

SelfIns (Whether the employer’s plan is self insured)

  • 0 = Not self-insured
  • 1 = Fully self-insured
  • 2 = Partly self-insured

WrkPartS (Worker’s premium contribution to single plan)

WrkPartF (Worker’s premium contribution to family plan)

EmpPartS (Employer’s premium contribution to single plan)

EmpPartF (Employer’s premium contribution to family plan)

PlanWt (Plan weight)

RecordID (Sort field)

PlanID (HRET/Kaiser only)

Year (HRET/Kaiser only)

The Year field in the 2000 HRET/Kaiser data has a value of either 1999 or 2000, as the dataset combines records from those two years. 1999 premiums have been inflated to bring them into line with year 2000 premiums (single premiums by a factor of 1.0362 and family premiums by a factor of 1.0983).

In a somewhat different manner, the Year field in the 2003 HRET/Kaiser data has a value of either 2003 or 2004, as the dataset combines records from those two years. 2004 premiums have been deflated to bring them into line with year 2003 premiums. Single 2004 premiums are deflated by a factor of 0.9158, and family 2004 premiums by a factor of 0.9114.

The following private premium donor files are available for use by the ESI module. Inflation or deflation factors for both single and family plans are also indicated when more than a single year of data are included in a donor file.

  • ESIPrivateHealthPremiums1996 – KPMG data for 1996
  • ESIPrivateHealthPremiums2003—HRET/Kaiser data for 2003 and 2004. 2004 premiums deflated to 2003 by single plan factor of 0.9158 and family plan factor of 0.9114.
  • ESIPrivateHealthPremiums2005—HRET/Kaiser data for 2005 only. No adjustment factor required.
  • ESIPrivateHealthPremiums2006—HRET/Kaiser data for 2006 and 2007. 2007 premiums deflated to 2006 by single plan factor of 0.9452 and family plan factor of 0.9461.

RecordID is unique within the file, but PlanID can be replicated due both to the fact that a firm may offer more than one plan type (e.g. both Conventional and HMO) and may be represented in both 1999 and 2000. All records in all years have data for both single and family plan types.

Match with Private Premiums

Workers who either report health insurance coverage by their employers or for whom it is imputed are matched with donor premiums.

Industry and firm size group are coded for each worker to match the categories in the donor premium dataset using LastIndustryGroupRecode2002 and LastWorkerClass for industry group and LastFirmSize for firm size group. Employer share of the premium is determined from HealthEmployerUnionHelpedPay (All, Some, or None). If ESI is offered but HealthEmployerUnionHelpedPay is not coded, then it is assumed the employer pays some of the worker’s premium. Likewise, employer share of premium amounts for both family and single plans are determined for each donor record by comparing worker and employer contribution amounts.

For both private premium files, matches are made on the basis of AnnualContractType (single or family, with dual plans treated the same as family) geographic region, firm size, industry, and employer share of premium for the contract type assigned a worker. However, the match procedure differs somewhat depending on whether which donor file is chosen since the characteristics of the files differ. The 1996 firm sizes and industry groups were collapsed in an effort to eliminate empty cells in the four-way region*industry*firm size*employer share matrix, which could result in a match failure. However, 2000 and 2003 donor file categories are not collapsed, instead relying on the match procedure to deal with match failures.

With the 1996 KPMG donor file, if there is no donor record that matches all the worker characteristics (which occurs for a few workers in either the smallest or largest firm size groups), then firm size is set to the middle size of three firm size groups (99-499), and a match is again attempted. With this adjustment, a match to premiums is made for all workers.

With the 2000 and 2003 HRET/Kaiser donor files, if there is no donor record that matches all the worker characteristics, then employer share is set to "employer pays some" and a match is again attempted. If a match is still not possible, then employer share is set to its original setting ("employer pays none," "employer pays some," or "employer pays all") and firm size is adjusted upward. If a match is still not possible, then employer share is again adjusted to "employer pays some." This procedure permits a match for all records. If employer share is adjusted, then total premium is adjusted by the regional ratio of average premium for the employer share required to average premium for the employer share actually matched, and the adjusted total premium is appropriately assigned either to the worker or employer based on the actual employer share for a worker.

In all match instances, the output variables HealthInsPlanType and SelfInsurance are set equal to the PlanType and SelfIns fields of the record to which a worker is matched.

To provide for a randomness in the match with records (since there are multiple records in most four-way groupings), the match is made with the first record for which the cumulative weighted fraction within a four-way grouping is >= a workers’ RandomNumGroupHealthMatch. Cumulative weighted fraction is computed from PlanWt, which represents the number of workers covered by a plan. As an example of how it is computed, presume there are 20 donor records that match a worker’s region, industry, firm size, and plan type employer share. Each record’s fraction of PlanWt to total PlanWt for the 20 records is computed. In a loop of the matching records, these fractions are accumulated until the first instance is found where the sum of fractions exceeds or equals a worker’s random number. The match is made with that record.

Contents of the State Government Health Premium Tables

State government health premium tables contain health insurance premium data for state government workers in all states (excluding territories) in separate donor tables for the years 1995, 1999 and 2002. Records are uniquely identified by FipsStateCode and PlanType. The earlier data (1995 and 1999), contain premium data for just two plan types (Conventional and HMO). For most states, both are available, but a few states have information for just a single plan type. However, the 2002 data contain information for 1 to 4 plan types per state, also including donor data for PPO and POS plan types. State government donor data reside in the following database tables:

  • StateGovHealthPremiums1995
  • StateGovHealthPremiums1999
  • StateGovHealthPremiums2002

Table fields are shown below.

FipsStateCode

PlanType

  • 1 = Conventional
  • 2 = HMO

SelfIns (Indicator of whether the plan is self insured)

  • 0 = Not self-insured
  • 1 = Fully self-insured

RecordID (sort field)

WrkPartS (Worker’s premium contribution to single plan)

WrkPartF (Worker’s premium contribution to family plan)

EmpPartS (Employer’s premium contribution to single plan)

EmpPartF (Employer’s premium contribution to family plan)

CumProb (A probability factor used to assign a worker to either a conventional or HMO plan. It accumulates to 1 across the typically two records in a FipsStateCode * PlanType grouping. It is set to 1 for the six states offering just a conventional plan type to workers.)

Match with State Government Premiums

A state government worker is matched to a donor record that matches his state of residence if the cumulative probability of selection (CumProb) is greater than or equal to the worker’s RandomNumGroupHealthMatch. CumProb is always equal to 1 for the last (usually second) record for a given state.

Contents of the Federal Government Health Premium Tables

Federal government premium data reside in the following tables:

  • FedGovHealthPremiums1995
  • FedGovHealthPremiums2001
  • FedGovHealthPremiums2003
  • FedGovHealthPremiums2005

The tables containing health insurance premium data for federal government employees contain just 32 rows that are uniquely identified by Region, SelfIns, PlanType and PostalStatus. Fields are as follows.

Region

  • 1 = Northeast
  • 2 = Midwest
  • 3 = South
  • 4 = West

SelfIns

  • 0 = Not self-insured
  • 3 = Federal Govt plan; possibly self-insured

PlanType

  • 2 = HMO
  • 5 = Federal Govt Plan (some type of managed care)

PostalStatus

  • 1 = Is a postal worker
  • 2 = Is not a postal worker

RecordID (sort field)

WrkPartS (Worker’s premium contribution to single plan)

WrkPartF (Worker’s premium contribution to family plan)

EmpPartS (Employer’s premium contribution to single plan)

EmpPartF (Employer’s premium contribution to family plan)

CumProbF (A cumulative probability of selection used for workers assigned a family or dual AnnualContractType)

CumProbS (A cumulative probability of selection used for workers assigned a single AnnualContractType)

CumProbC (A cumulative probability of selection used for workers not assigned an AnnualContractType)

Match with Federal Government Premiums

Federal government employees are matched to donor records that agree with a workers’ geographic region and postal status, i.e., whether or not the worker is a postal employee (LastDetailedIndustry = 412). A worker’s RandomNumGroupHealthMatch is used to select one of the four potential records in each Region and PostalStatus grouping. Workers assigned either a dual or family AnnualContractType have their random number compared with CumProbF, whereas workers whose AnnualContractType is single are compared with the CumProbS. Random numbers of the remaining workers are compared to CumProbC. From a file sorted by increasing cumulative probabilities, a match is made with the record whose appropriate cumulative probability is greater than or equal to a worker’s random number. The final cumulative probability for each grouping is 1 so that a match is assured for all workers.

Summary Tables

The EmpSponsoredHealthIns module summarizes the results of the simulation described above in the following tables.

  • B1: Insurance Status of Workers by Private Sector Firm Size and public Sector Employment
  • B2: Total Annual Health Insurance Premiums Paid by Type of coverage, Private Sector Firm Size, and Public Sector Employment
  • B3: Average Monthly Workers with Employer Sponsored Health Insurance by Type of Coverage, Private Sector Firm Size, and Public Sector Employment
  • B4: Average Firm Family and Single Premiums for Private Sector by Type of Coverage and Employer Share of Premium
  • B5: Self-insured Status and Type of Plan for Private Sector by Type of Coverage and Employer Share of Premium
  • B6: Average Premiums of Plans Assigned Federal and State by Type of Coverage
  • B7: Self-insured Status and Type of Plan for Federal and State by Type of Coverage
  • B8: Total Workers and Percent in Firms that Offer Health Insurance, by Industry and Firm Size
  • B9: Workers Without Health Insurance Through Own Employer: Total and Percent with Employers Offering to Some Workers by Industry and Firm Size
  • B10: ESI covered Workers subject to Waiting Period by Firm Size, Region, Industry
  • B11: Waiting Period Lengths for new jobs with ESI coverage by Firm Size, Region, Industry
  • C1: Average Firm Family and Single Premiums for Private Sector and Local Government Workers by Type of Coverage and Firm Size
  • C2: Average Firm Family and Single Premiums for Private Sector and Local Government Workers by Type of Coverage and Industry
  • C3: Average Firm Family and Single Premiums for Private Sector and Local Government Workers by Type of Coverage and Region
  • S1: Average Family and Single Premiums for Plans Assigned State Government Workers
  • S2: Self-insured Status and Type of Plan for State Government Workers by State