Reducing Choice Overload on Healthcare.gov

Last night, the Centers for Medicare and Medicaid Services (CMS) released the proposed Notice for Benefit and Payment Parameters -2024 (NBPP 2024).  The NBPP is an annual rule.  It is the rule book for each year of health insurance operations for the ACA individual market.  Some years it is big and meaty.  Other years it is just modifications to the previous year to assuage smooth(ish) running of the marketplaces.  It is where most big rule based policy changes are eventually implemented for the marketplaces.

The draft NBPP 2024 seeks to manage the choice space fairly aggressively:

The average number of plans available to consumers on the Marketplace has increased from 25.9 in PY2019 to 113.6 in PY2023. Such plan choice overload limits consumers’ ability to make a meaningful selection when comparing plan offerings….

CMS also proposes to limit the number of non-standardized plan options that issuers of QHPs can offer through Marketplaces on the Federal platform (including SBM-FPs) to two non-standardized plan options per product network type and metal level (excluding catastrophic plans), in any service area, for PY2024…

an alternative to… CMS could also apply a meaningful difference standard in PY2024 and subsequent PYs. Under this proposed standard, CMS proposes grouping plans by … and then evaluating whether plans within each group are “meaningfully different” based on differences in deductible amounts. With this proposed approach, two plans would need to have deductibles that differ by more than $1,000 to satisfy the new proposed meaningful difference standard.

For the past several years, choice has gotten tough on Healthcare.gov.  Patrick O’Mahen and I showed in Health Affairs Forefront that  the almost a majority of people in 2021 were exposed to at least 50 plan choices.  We know from Medicare Advantage that choice quality goes downhill after 15 choices.  We know that insurance literacy is low in the general population and there are hints of evidence that agents and brokers aren’t particularly good at keeping people out of dominated plans.

CMS wants to reduce the choice space and menu length.  The first method is to say that insurers can offer no more than 3 plans in a metal level/plan type/network combo.  Right now, Ambetter offers 22 Silver variants in Atlanta, OSCAR offers 18 Silver variants in Metro Columbus Ohio, and Quarts offers 17 Expanded Bronze variants in parts of Wisconsin.  That is absurd and abuses attention.

The other method is to make meaningful difference both meaningful and different.  The Obama Administration had meaningful difference rules that were a joke to game.  There were six elements and a plan was judged to be meaningfully different if there was one minor difference in a single element.  The Trump Administration got rid of these rules in 2019.  The proposed rule would allow for an unlimited number of plans to be offered by an insurer/plan type/metal level triad as long as the deductible is at least $1000 different from any other plan.

In 2016, I proposed a similar-ish concept where instead of deductible, I proposed using premium as the key differentiator:

 a plan offered by the same insurer is meaningfully different if the premiums vary by Z percent. That Z could be 3% or 5% or 7%. After that, CMS would not care how the difference is achieved? Is it a PPO instead of an EPO? Who cares? Is it a narrow network versus a broad network? Who cares? Is it HSA eligible or not HSA eligible? Who cares?

The insurance companies could then design products where there actually is a useful difference in the products instead of attempts to game/hack the subsidy formula for business strategy reasons.

I’m not sure which of these three variants of distinction-making I like the best.  I think there is a good argument to have more than one plan per insurer per metal level, and a very strong argument about a hard cap (realistically no more than 9 due to the $1000 deductible gap requirement) as people readily get confused.  I’m not sure if the optimal is 3 or 4 or 5 — but reasonably confident-ish it is neither 1 nor 9 nor 22.

There is a lot more to dig into the regulation including some very nifty automatic re-enrollment changes and tweaks but this is the thing that stood out to me on my first glance.