Session 26 PD, The ACA and Risk Adjustment: Round Two. Moderator/Presenter: Gregory G. Fann, FSA, FCA, MAAA

Transcription

1 Session 26 PD, The ACA and Risk Adjustment: Round Two Moderator/Presenter: Gregory G. Fann, FSA, FCA, MAAA Presenter: Hans Leida, FSA, MAAA Erica Rode, PhD, ASA, MAAA SOA Antitrust Disclaimer SOA Presentation Disclaimer

2 The ACA and Risk Adjustment: Round Two Session 26 - SOA Health Meeting 6/12/2017 Presented by: Greg Fann Greg.fann@axenhp.com

3 Risk Adjustment: A Key ACA Element 1. Only permanent R of 3 Rs Necessary with market rules Not a transitional element 2. Different actuarial perspectives 3. Health Section Council Strategic Initiatives One of first two strategic initiatives was ACA and the Exchanges Outcome was web exclusive for The Actuary: Differing perspective but similar themes: Shortcomings in methodology, suggested improvements Unfinished ACA business new strategic initiative 6/12/2017 Slide 2

4 HHS-HCC Risk Adjustment Ideals 1. Budget neutral zero-sum game Battle for larger piece of pie, not a larger pie Generally would suggest less alignment within industry 2. Adjust for constrained rating flexibility ASOP 12 (Risk Classification) recognition 3. Bridge cost difference between: A. actuarial (required) rate B. allowed rate 4. Remove enrollee selection incentive 6/12/2017 Slide 3

5 Competing Carrier Desires 1. Budget neutrality and Inequity Equal ($ not plans) share of winners and losers 2. Winners: large established plans 3. Losers: new, growing plans 4. Reactions/Responses States: NY - protect SG insurers from risk adjustment State Associations/Alliances (i.e. MA) Coalitions (i.e. Choices) Regulatory Comments 2018 NBPP mms/oct2016/comment%20letter%20appendix%20i_part1.pdf Litigation - 6/12/2017 Slide 4

6 HHS-HCC Methodology A Two Step Process 1. Individual Risk Assessment Important to get right (equity) Insurers have ceded risk factor calculations to government 2. Transfer Formula Budget neutral Should reflect actuarial cost differences Equity also important 6/12/2017 Slide 5

7 HHS-HCC Methodology Risk Assessment 1. Scoring should be precise and impartial Data/Methodology 2. Data Used Large group experience without large concentration of short duration members or coding incentives Scheduled to be replaced w/ ACA market in Methodology: HCC Coefficients Overstates costs of individuals (20%) with HCCs Post Risk Adjustment Medical Loss Ratios: No HCCs 126% / HCCs 72% no adjustments at this time, continued exploration 6/12/2017 Slide 6

8 HHS-HCC Methodology Risk Assessement 4. HCC coefficients also neglect to consider: Partial Year Enrollment (incorporate in 2017) Pharmacy claims (incorporate in 2018) Care Management Effectiveness 5. Methodology penalizes small, growing plans that have not adequately diagnosed members Note: Diagnoses don t transfer as they do in Medicare 6/12/2017 Slide 7

9 HHS-HCC Methodology Transfer Formula 1. Required to be unbiased even if risk assessment is perfect 2. Complaints on Statewide Average Premium model Budget neutral decision by HHS S - Statewide nature of formula Regional coding can vary A Average inflates transfers for efficient carriers, limited networks P Premium transfer incorporates transfer of nonrisk administrative expenses (14% in 2018) 3. Discourage plans who attract low cost individuals 6/12/2017 Slide 8

10 HHS-HCC Methodology Transfer Formula 4. Result: formula with items extraneous to actuarial risk Individual market enrollment choices Competitors rates and coding patterns Enrollment shifts in market Numerical Example in March 2017 Health Watch Unpredictability Unstable market New surprise financial item (vs. IBNR) Surprise for regulators as well as health plans -> Complicated, unreasonable pricing exercise 6/12/2017 Slide 9

11 Summary and Looking Ahead 1. Risk adjustment (RA) models require equity, predictability, and removal of selection incentives; we don t care who we enroll (some exceptions) 2. RA models should be based on actuarial risk. 3. RA models should not dampen incentives for costeffective operations and innovation. 4. Today, there are incentives for high premiums and enrolling high risk individuals. 5. RA challenges will be magnified with legislative changes to the ACA. 6/12/2017 Slide 10

12 Summary and Looking Ahead 6. RA relies on ACA provisions: Essential health benefits, uniform age curve, metal levels. All of these could be disrupted with AHCA state waivers. 7. New mechanics of federal assistance should theoretically impact risk adjustment. Could double-dip Reinsurance under ACA State Stability Funds Invisible High Risk Pools 8. President Trump's experts will have an even harder job than did President Obama's. As more variation is permitted in insurance policies, and as the disparity in risk levels among insurance buyers increases, risk adjustment becomes more necessary and difficult. - Seth Chandler, Mercatus Center and University of Houston Law Center 6/12/2017 Slide 11

13 Q&A and Wrap-Up

14 The ACA and Risk Adjustment: Round Two Hans Leida, PhD, FSA, MAAA Erica Rode, PhD, ASA, MAAA June 12, 2017

15 Limitations This presentation is intended for informational purposes only. It reflects the opinions of the presenters, and does not represent any formal views held by Milliman, Inc. Milliman makes no representations or warranties regarding the contents of this presentation. Milliman does not intend to benefit or create a legal duty to any recipient of this presentation. 2

16 Agenda Lightning review of ACA risk adjustment and program results to date Sampling of challenges and successes Coming changes to the model Potential changes to the program under the new administration Strategies for success and management 3

17 ACA changes to individual and small group markets Creation of Exchanges Single risk pool Guaranteed issue Elimination of underwriting Limitation of rating variables for carriers Only plan design/network, age, geography, family size, and tobacco Prevented charging higher premium for health status 4

18 ACA risk adjustment program Goals: Stabilize Premiums Competition based on benefits offered, networks, etc. Transfer payments between carriers in each state = Premium w/ Risk Selection Premium w/o Risk Selection = Actual population risk Risk charged for in premiums Zero-sum program in each market 5

19 Risk score calculation HHS-HCC Model Scores based on age, gender, and filtered Medical diagnosis data Claims filtered based on procedure codes and bill type codes Diagnoses grouped into Hierarchical Condition Categories (HCCs) 127 HCCs developed by modifying CMS-HCCs for commercial population Coefficients developed to predict net plan liability Coefficients reflect Medical and Rx claims Separate model calculated for each metal and age group (adult, child, infant) Concurrent model 6

20 So, how s it going? 7

21 Results to date Overview % Change Individual Market Total Market Premium $35,359 M $59,261 M 67.6% Total Transfer Payments $3,506 M $5,615 M 60.2% Transfer Payments/Premium 9.9% 9.5% -4.4% Small Group Market Total Market Premium $19,338 M $38,270 M 97.9% Total Transfer Payments $1,131 M $2,159 M 90.9% Transfer Payments/Premium 5.9% 5.6% -3.6% Source: Busch, F., et. al., (August 2016) On second thought: Initial insights from year 2 of the ACA risk adjustment program, from 8

22 Consistency of carrier year over year results Individual Market 2014 Transfer Payment Distribution 2015 Transfer Payment Distribution Receipt Payment Receipt Payment Graphs from: Busch, F., et. al., (August 2016) On second thought: Initial insights from year 2 of the ACA risk adjustment program, from 9

23 At a broad level, the incentives to focus on risk selection have been attenuated by the risk adjustment program. -Paul D. Jacobs et al, Risk Adjustment, Reinsurance Improved Financial Outcomes for Individual Market Insurers with the Highest Claims Amount of paid claims is strongly correlated with risk scores. -Summary Report on Transitional Reinsurance Payments and Permanent Risk Adjustment Transfers for the 2015 Benefit Year 10

24 Source: Summary Report on Transitional Reinsurance Payments and Permanent Risk Adjustment Transfers for the 2015 Benefit Year 11

25 Risk transfers are inaccurate predictors of claim costs. -Alan Vandagriff, et al, Sizing Up ACA Risk Adjustment Volatility The current HHS-HCC risk adjustment model established by CMS is known to understate risk scores for relatively healthy individuals and to overstate them for those with significant health conditions. - Richard Foster, Method to Address Estimation Bias in the HHS-HCC Risk Adjustment Model 12

26 Risk Adjustment Transfer Relative to Paid Claims Incurred Claims State Average 250% 200% 150% 100% 50% $(150) $(100) $(50) $0 $50 $100 $150 $ Transfer PMPM CHOICES Coalition white paper: Impact of Risk Adjustment on Carriers 2015 Financial Performance in the Individual Market 13

27 Sampling of challenges and successes 14

28 Changes to the model Past and Proposed Initial Model 2015 Minor logic changes 2016 Minor logic changes Recalibrate 2017 Enrollment duration factors Recalibrate Prescription drug data High-cost enrollee adjustment Recalibrate 2019 Recalibrate on ACA data 15

29 Changes to the risk adjustment model Data and Methods Sample of approximately 200K members enrolled in the individual market in 2016 RxC mapping developed based on information in the 2018 Notice with clinician input Members scored under 2016, 2017, and 2018 models, using enrolled metal level and 2016 data Impact of CSRs removed 16

30 Changes to the model Ranges of scores Silver Minimum Median Maximum Bronze Minimum Median Maximum

31 Components of model changes Raw scores 18

32 Components of model changes Normalized scores 19

33 Components of normalized risk score changes 2017 to

34 Components of (normalized) risk score change 2017 to

35 Components of normalized risk score changes 2017 to

36 Proposed changes 2019 recalibration on ACA data Data and Methods Goal: compare risk adjustment and claim cost curves by age for individual market ACA Individual Market Data Approximately 1.5 M lives in the 2015 individual market from Milliman s internal research database 2016 silver HHS-HCC model scores calculated, excluding CSR impact Compared to paid claims for same members Large Group Data Sample of large group data from Milliman s Health Cost Guidelines - Commercial Represent nationwide average billed charge levels for individuals/groups with little to no underwriting Allowed claims estimated using Milliman benchmark nationwide discounts applied to billed charges 23

37 Age curves 24

38 Proposed changes under new administration 25

39 Proposed changes under new administration Loosening issue and rating rules Strictest issue requirements Need for risk adjustment to lessen incentives to avoid high-cost Greatest need for risk adjustment Loosest rating rules Least need for risk adjustment Need for risk adjustment to lessen incentives to avoid high-cost Strictest rating rules Loosest issue requirements American Academy of Actuaries Issue Brief: How Changes to Health Insurance Market Rules Would Affect Risk Adjustment 26

40 Proposed changes under new administration High-risk pooling Different ways to structure Traditional provide coverage in separately run pool Invisible enrollees remain in the individual market, but a portion of their claims are reimbursed. Typically condition-based. Reinsurance a portion of costs over some threshold are reimbursed, regardless of condition Risk adjustment needs to coordinate with pool to avoid overcompensation E.g., use different conditions for invisible pool and risk adjuster Method finalized in 2018 Notice: Exclude 60% of claims > $1 M from PLRS and fund from nationwide pool Separate pool for small group and individual American Academy of Actuaries Issue Brief: How Changes to Health Insurance Market Rules Would Affect Risk Adjustment 27

41 Proposed changes under new administration Increased flexibility in cost-sharing levels/ehb requirements Current model assumes uniform EHB coverage and narrow AV ranges If EHBs are still covered, can still use an AV calculator Would require recalibration of risk adjustment model Could continue to use ranges of AVs or switch to one model with AV as input Increases selection risk if <60% AV allowed American Academy of Actuaries Issue Brief: How Changes to Health Insurance Market Rules Would Affect Risk Adjustment 28

42 Proposed changes under new administration Allowing sales across state lines Challenges revenue-neutral aspect Differences in age curves or other requirements Enrollees must be part of licensing states risk pool Unclear how to handle geographic factors for out-of-state regions American Academy of Actuaries Issue Brief: How Changes to Health Insurance Market Rules Would Affect Risk Adjustment 29

43 Proposed changes under new administration Increased state variation Could introduce different high risk pools May need to develop state-specific risk adjustment models Some elements of transfer formula could vary easily, but fundamental changes to methodology would make it more difficult for CMS to administer More administratively complicated, less cost-effective for multi-state insurers American Academy of Actuaries Issue Brief: How Changes to Health Insurance Market Rules Would Affect Risk Adjustment 30

44 Proposed changes under new administration Elimination of cost sharing reduction subsidies CSR subsidies future is very uncertain Risk adjustment currently assumes carriers will be reimbursed for CSRs Risk adjustment currently attempts to spread the induced utilization associated with CSRs across the single risk pool Not yet clear how carriers would be required to implement rate changes for CSR shortfalls Rate up silver plans only, spread across the pool, or??? Risk adjustment program will need to be altered to dovetail with rating rules Policy question: should risk adjustment be used to spread unfunded CSRs across the single risk pool? Pedro Alcocer, et al, A bridge too far? The most likely fates of ACA CSR payments and impacts on the individual market 31

45 Comprehensive Risk Adjustment Management Program (RAMP) Facilitate, document and code it right Prospective Data Collection Retro Data Correction Make sure it s right Risk Score Verification Data Submission Data and score match Timely Accurate Complete 32

46 Thank you Hans Leida Erica Rode