Yesterday’s post on the potential for COVID-19 sequalae was a post that is fundamentally about uncertainty. We don’t know what is happening medium or long term because even the longest case histories in the United States are less than 90 days old. Clinicians will be operating ahead of actuaries in figuring out what the short term is as the clinicians will be seeing data earlier but less comprehensive than the actuaries who are able to see a broader but slower data set from claims. Actuaries are currently trying to price out initial rates for 2021 right now, and they are hitting a serious IBNR problem for both COVID-19 and non-COVID-19 utilization.
IBNR is Incurred claims But Not Received claims. Clinical entities and organizations won’t provide services and bill the insurer on the same day. Using Medicare data, slightly less than half claims would arrive at an insurer within a month (Table 2). By sixty days, close to 90% of all claims would have arrived at the insurer. Some clinical entities are fast billers. Larger, more administratively complex entities maybill every day. Smaller practices might only bill on the 2nd Tuesday of the month or whenever the pile gets too high. IBNR can usually be accounted for. Insurers and actuaries will take a time period worth of received claims and then multiply that value by an IBNR multiplier. That multiplier will start off as a constant with perhaps a day of the week adjustment, a seaonality adjustment and type of claim adjustment. That calculation will give a reasonable estimate of claims understanding and received. By three months out, 95% or more of all claims will have been received by the insurer and IBNR is usually a minor factor on information and financial performance.
IBNR calculations are fundamentally premised that the past is reasonably predictive of the now or near future. And usually that is good enough. IBNR will not catch a one-off multi-million dollar catastrophic claim. And it won’t deal well with a pandemic, especially early on. And we are still early on in the COVID-19 pandemic while entering a critical part of the rate setting cycle.
CDC reports that there were under 3,500 diagnosed COVID-19 cases on March 15, 2020, just a month ago. Today, there are over 600,000 diagnosed cumulative cases. Even if all March 15th claims were submitted, some insurers will have no members with COVID-19 related claims by March 15 as they either have no members diagnosed on that date, or if there was a diagnosis, the individual at that point in time was not symptomatic or not severely symptomatic.
Actuaries running COVID-19 claims based models today are seeing a tiny sliver of the actual COVID-19 claims experience. Most of the people who are currently infected have been infected in the past two doubling cycles and have not reached the point in their disease progression where hospitalization or intensive care services are going to be heavily used. Even for the people in the past two doubling cycles who are in the hospital, those claims are unlikely to have been sent out the door to the insurer.
Insurers are literally pricing on almost non-existent claims data for COVID-19 today. By the time that they need to submit their initial rates, insurers are likely to have adequate insight on the rare early March COIVD-19 hospitalizations and minimal claims knowledge of the far more frequent April hospitalizations. At the same time, we also know that March non-COVID-19 claims crashed. The IBNR models will be massively overpredicting March and April claims exposure as this March and April are nowhere close to normal. Everyone knows that the IBNR model won’t be close to right. The challenge is that estimating how badly wrong the IBNR model is mostly a shot in the dark on thirty to sixty days of run-out.