Most health insurance companies want to have as many claims as possible to automatically process.  Well designed claims systems can punch out 98% or 99% of their claims without any manual intervention.  They are able to apply complex business rules at the business line, and benefit design level (for instance if a person is cut open once and two mostly unrelated surgeries are performed at the same time, that is one price, while if the person is cut open twice, the two surgeries are priced as separate events.)  Medicare contracts out most of their claims processing and ends up paying an average of 58 cents per claim.  They get the per claim rate that low because their regional claims processors are amazingly automated.

However, once a claim has to be viewed by a human, expenses start to quickly add up.  A manual review of a claim is fairly cheap, maybe a dollar or two to cover the cost of the employee’s time and associated overhead.  However if the employee has to do anything to a claim, a manually processed claim prices out at five dollars if it is simple adjustment.  If the claim is complex, the manual intervention can easily cost ten or more dollars to process.

Medicaid eligibility has always been a messy business because people move between benefit categories, they move in and out of eligibility and the rules for payment structures change.  Some of those changes are future changes, and other changes are retroactive changes which require claims readjustment.  If a person was eligible for regular Medicaid but then either gets pregnant or goes on short term disability, they get a slightly better benefit package as more services will now be paid.  Those services were performed before the eligibility redetermination was made, so previously denied or held claims have to be reassessed and repaid.  That means new explanations of benefits, new explanations of payments, and new checks need to be sent out.

The Pennsylvania waiver proposal has six new categories that would require premium payment.  Each category would fall into one of the four buckets of job search incentive payments (0%, 15%, 25%, 40%) so that means twenty four benefit design configurations.  That is a large number, but it is manageable if the population basis is stable.

However, we can also look at the labor market for some insight on the stability or instability of the less than 138% FPL portion of the market.  This is an extremely volatile labor market.  Stable, predictable schedules are now a privilege of middle and upper middle class employees.  Low wage employees are on just in time scheduling where one week an individual might have thirty seven hours on the schedule, and the following week they have twenty six hours.   And since the schedule may only come out with three days notice, there is a low probability of a stable second job where the person is always in the same hours bucket.

Furthermore, low wage work tends to see shorter periods of attachment per job than high wage work.  It is far more likely for a person make $9.00 an hour to jump jobs without a pay raise than someone making three times as much because the universe of $9.o0/hr jobs is much larger than $27.00/hr jobs.

If the state is checking employment status of Medicaid expansion recipients on a weekly or monthly basis, individuals will routinely move between premium/benefit configuration buckets multiple times in a quarter.  And that means that an individual may have satisfied all of their deductibles and co-pays on the day they received a medical service, but since their work schedule was dropped to a lower bucket, the claims have to be readjusted to bill the member for another $5.00 in co-pay or deductible.  Conversely, if a member was in a low hour bucket on the day they received a service, but gets a consistent extra shift the following few weeks, their claims will be readjusted to pay back the extra deductible and co-pays.  Providers and insurers will scream about the additional administrative complexity.

If the state checks annually, this probably is worse as I would assume the state would take a good faith hours estimate from individuals and then come back and reconcile against tax records.  This reconciliation might have fewer people bouncing between buckets.  However, the people who do bounce between buckets will have a much larger claims history to readjust.

This is important in a private option scenario as the insurers have only a 20% non-medical loss ratio.  Well run insurers have only a 10% or 11% non-medical loss ratio cost, but they get there partially by automating their claims system so almost nothing has to be readjusted.  Medicaid expansion with this job search incentive necessitates massive number of adjustments without the state paying any extra.

So between poor-shaming and needless administrative complexity and costs, I don’t think HHS will approve the Pennsylvania waiver as it is currently written.