REGULATORY ORDER NUMBER R 20-09

APPROVAL WITH MODIFICATION OF THE

2021^[1] WORKERS’ COMPENSATION LOSS COST FILING

AND ASSIGNED RISK RATE FILING

BACKGROUND:

On August 24, 2020, the Division of Insurance (division) received the 2021 Alaska Workers’ Compensation Filing for Voluntary Loss Costs and Assigned Risk Rates from the National Council on Compensation Insurance, Inc. (NCCI)^[2]. This filing proposed an overall 17.5% decrease in voluntary loss costs and an overall 11.2% decrease in assigned risk rates from the current approved levels.

On August 14, 2020 the director issued Notice of Public Hearing H20-01 notifying interested parties that, in accordance with Alaska Statute (AS) 21.39.043, a hearing would be held on September 15, 2020.  Due to public health and safety concerns, the hearing was held telephonically and by video conference.  The purpose of the hearing was to allow interested parties to provide testimony or evidence as to whether the filing’s prospective loss costs met the requirements of AS 21.39. Interested parties were also invited to include a recommendation for approval, disapproval, or modification of the filing. 

No requests for information or interrogatories were submitted to NCCI by member or subscriber companies prior to the hearing, and no written testimony was received by the division prior to the hearing. At the hearing, Alaska National Insurance Company (ANIC) provided oral testimony and related supporting exhibits. The hearing record was held open for 10 days after the hearing; no  written comments were received during that time.

The division requested and received additional supporting information from NCCI as allowed under AS 21.39.043(e) as detailed below.

DISCUSSION OF FILING METHODOLOGY

1. Consistent with what was approved in the 2020 filing, NCCI again:

1. used a 50/50 weighting of indications derived using limited and unlimited losses (“limited indication” and “unlimited indication”, respectively);
2. for each indication, used an average of paid and paid-plus-case experience;

i. paid loss development factors (LDF) are based on a 3-year average;

ii. paid-plus-case LDFs are based on a 5-year average;

1. used paid-plus-case experience to derive the tail (19^th report-to-ultimate) LDFs; and
2. used excess ratios (in the limited indication) that limit losses to $50 million, as losses over $50 million are funded via a separate catastrophe loss cost.
3. indemnity benefit changes due to changes in the State Average Weekly Wage (SAWW) are captured through the indemnity trend selection rather than through explicit benefit changes.

2. Changes in NCCI’s methodology versus the 2020 filing include:

1. overall indication is now calculated using four  policy years (PY) of experience instead of three (i.e. PY2015, PY2016, PY2017, and PY2018 are used in this filing);
2. adjusting and other expense (AOE) provision now excludes policies that use Third-Party Administrator (TPA) agreements.

3. The limited indication is based on NCCI’s Large Loss Procedure (LLP)^[3]. To accommodate the LLP, annual updates to the large loss limit threshold and excess ratio^[4] were made, based on Alaska premium and loss data.

4. NCCI proposed to increase the loss adjustment expense (LAE) provision from 20.0% to 20.4%, as a percent of losses. This change is due to an increase, from 8.5% to 8.9%, in the AOE portion of the LAE, while the defense and cost containment expense (DCCE) portion remains at 11.5%. The impact of this change on loss costs is +0.3%.

5. NCCI proposed using annual trends of -5.5% and -4.0% to adjust historical indemnity and medical loss ratios, respectively. Consistent with last year’s filing, the loss ratios used to determine trends are based on experience data limited by the LLP and use an average of paid and paid-plus-case based severities. Losses are adjusted to a common wage level so these trends are in excess of wage inflation. These selections represent a 1.0% decrease compared to the loss ratio trends approved in the 2020 filing.  The combined impact of this change on loss costs is -4.1%.

6. NCCI included adjustments to incorporate the expected impact of benefit changes as follows.  The combined impact of these changes on loss costs is -1.0%:

1. medical benefit costs are expected to decrease 5.3% due to the change in Alaska medical fee schedule conversion factors established by the Department of Labor and Workforce Development, effective 1/1/2020.  Half of this impact was included in the 2020 filing; the other half is incorporated in this filing via including the whole impact in the benefit on-level factors.
2. medical benefit costs are expected to increase 0.8% due to automatic adoption, by reference, of annual updates to the Medicare fee schedules other than conversion factor changes, effective 1/1/2020.  This impact is incorporated via adjusting projected medical cost ratios.  The method and assumptions used to estimate the impacts of this automatic adoption have not changed versus previous filings.

7. The assigned risk portion of the filing includes the following items. The combined impact of these changes on assigned risk rates is +7.6%:

1. servicing carrier allowance; the filing uses a value based on the servicing carrier allowances applicable to the 2018 – 2020 contract period, since the 2021 – 2023 contract period servicing carrier contracts were not finalized at the time the filing was submitted. bid process (+0.2%^[5]);
2. assigned risk plan commission rate (no change);
3. assigned risk plan administration expenses (-0.4%);
4. excess of loss reinsurance expense (+4.5%);
5. profit and contingency provision (+2.9%);
6. offset for expected reinsurance recoveries (+0.1%); and
7. uncollectible premium provision (+0.3%).

8. Various values within the “Footnotes” and “Advisory Miscellaneous Values” loss cost and rate pages and Experience Rating and Retrospective Rating plan manual pages were revised:

1. based on previously approved formulas to reflect changes identified elsewhere within the filing, such as changes in the SAWW, relative changes in state and federal benefits, and proposed loss cost level changes; or
2. to reflect changes approved in other filings.

9. NCCI did not make any explicit adjustments for potential impacts related to COVID-19.

REQUESTED MODIFICATIONS, COMMENTS, AND INTERROGATORIES

At the hearing, NCCI provided a presentation summarizing the components and impacts proposed in the filing.  ANIC then provided oral testimony and presented related supporting exhibits[6] at the hearing and made requests for modifications.  The testimony focused on implications related to declines in frequency, including questioning whether NCCI's trend selections are optimal considering the potential volatility in year-to-year loss ratios as claim frequency declines and larger claims increasingly make up a larger component of overall loss costs.  ANIC pointed to their internal experience, the increasing cost of the reinsurance purchased by the assigned risk pool, and the August 2020 multi-bureau Countrywide Mega Claims report[7] as evidence of the increasing frequency of larger claims.  ANIC suggested that when predicting future cost levels, NCCI should explicitly consider (and document within the loss cost filing) differences in frequency trends by size of loss and generally consider the greater uncertainty associated with having fewer claims in the experience.

No other interested parties provided testimony or posed questions to NCCI during the hearing. 

The hearing record remained open for 10 days after the hearing date, in accordance with AS 21.39.043(d)(7). No additional written testimony or proposed modifications to the filing were received by the division within that timeframe.

Questions and concerns raised during the hearing were incorporated by the division into an interrogatory letter sent to NCCI on September 30, 2020. The letter requested additional supporting information related to both the hearing testimony and other issues identified by the division, including:

• whether the limited indication is appropriate considering the loss ratios calculated using the LLP have consistently been lower than the unlimited loss ratios over the past number of PYs;
• whether the DCCE selection is adequate considering an apparent upward trend in DCCE, as a percent of losses;
• whether the assigned risk plan administrative expense provision is appropriate considering an apparent upward trend, as a percent of premium.

NCCI responded with the requested information and rebuttal, as required by AS 21.39.043(e) and (f) on October 14, 2020.

NCCI RESPONSE AND REBUTTAL

NCCI was unable to provide trend indications by size of loss for Alaska specifically within the fifteen day timeframe of AS 21.39.043(f).  They did, however, discuss a recent analysis performed by their Actuarial Committee using countrywide data (for NCCI states) that found that, despite the apparent higher average annual change in frequency for claims above $1MM at 2nd report (i.e. the latest report shown in the Large Loss Research section of the filing reference in ANIC's hearing exhibits), after claims mature to 5th report, the average annual change in frequency for claims above $1MM was "in line with" that of claims below $1MM.  NCCI also pointed out that segregating losses by size for trend analysis purposes would introduce additional volatility and potential error due to lower credibility in each component and having to make multiple selections.  Their method implicitly considers and reflects changes in frequency (and severity) by size of claim.

Regarding the issue of decreasing claim frequency making predictions more volatile and uncertain, including whether the accident year (AY) 2019 loss ratios require more moderate trend selections, NCCI explained:

• they do not explicitly incorporate credibility into their trend analysis, but stated their belief that the change in trend selections compared to the prior filing's approved trends is appropriate and conservative;
• they rely on longer-term fits to smooth out volatility;
• their loss ratio trend selections are less negative than almost every trend fit (based on using up to 15 years of historical PY data);
• their loss ratio trend selections are in line with fits calculated excluding the latest PY's loss ratio; and
• recent years' changes in frequency appear to be less volatile, even though there are fewer claim counts.

Regarding the observation that limited indications have been consistently less than unlimited indications in recent years (e.g. PYs 2012 - 2018), NCCI explained that the differences are generally small and the differences in four of the seven years are related to large claim activity.  Since the limited indications temper the impact of individual large claims in those years, they conclude that the LLP is doing its job.  NCCI also pointed out that their method helps promote year-to-year stability in loss cost indications.

Regarding the adequacy of the DCCE ratio selection, NCCI explained that the DCCE ratios in four of the last five PYs fall within a narrow range, it is unclear whether the latest PY is an outlier or a reliable indication of the future, that the selected provision is above all but the latest PY's ratio, and that their selection balances stability and responsiveness.

Regarding the adequacy of the assigned risk plan administration expense provision, NCCI explained that the longer-term, 10-year weighted average smooths year-to-year changes as the administration expenses are relatively fixed compared to the premium volume of the Plan, which is of "cyclical nature".  NCCI also disclosed that they adjust the calculated 10-year weighted average based on projections for the prospective period.

FINDINGS

After fully reviewing and considering the supporting documentation and testimony, the director finds:

1. The change to a four PY experience period is acceptable.  Adding another year to the experience period helps smooth the impact that year-to-year variability in loss ratios has on the overall loss cost level change.  As frequency continues to decline, this additional year makes increasingly more sense.

2. The use of a 50/50 weighting of limited and unlimited indications is acceptable. 

Implicit in the LLP is a threshold for limiting losses and an excess ratio to build back in a provision for the expected value of losses removed by the limiting threshold.  The threshold is set by finding the size of an individual claim that would cause a certain impact on the overall loss cost indication.  The threshold used in this year's filing has once again decreased, consistent with decreasing aggregate loss cost volume.  The excess ratio increased, reflecting that a larger percentage of losses are removed due to the lower threshold for limiting losses, despite that excess ratios are generally decreasing for a fixed limit[8].

Despite that the threshold and excess ratio are generally moving directionally as expected, the limited loss ratios[9] for six of the seven most recent PYs (PY 2012 through PY 2017) are lower than the unlimited loss ratios, and the loss ratios in PY 2018 are equal.  NCCI notes that the differences are related to large claim activity and the lower limited loss ratios in those years thus reflect that the LLP is doing its job.  However, smoothing variation in large losses is different than tempering the impact of large losses without building back an appropriate load.  The consistency of the low estimates over many recent years suggests the excess load is inadequate.

Still, considering the desire for stability in the loss cost indications, the desire to maintain a consistent approach from year-to-year, and the relatively small impact of implementing reasonable alternatives (for example, using only unlimited indications), maintaining the 50/50 weighting of limited and unlimited indications—which is consistent with the treatment used in the four most recent annual filings—is acceptable for this 2021 filing.  This weighting incorporates the desirable characteristics of each method in a balanced way. 

3. NCCI’s  loss ratio trend selections should be less negative. Considering the potential for increased volatility in loss ratios and the increased uncertainty in predicting loss ratios due to continually declining frequencies, the additional uncertainty in the coming year(s) due to impacts related to COVID-19 (for which no separate explicit adjustment has been made), and the trend selection's increased leverage on the indication introduced this year via adding an additional year to the experience period, a smaller change in trend selections versus those approved in the 2020 filing will reduce dislocation due to changing assumptions while still reflecting that the universe of trend fits presented in the filing are more negative this year versus last year.  A slightly tempered change in trend selections also ensures they remain conservative relative to the presented universe of trend fits if adjusted to either exclude the latest PY 2018 loss ratio or include the latest AY 2019 loss ratio.

Also, as discussed in Finding #2 above, it is not entirely clear that the LLP's parameters are accurately reflecting large loss expectations in recent years.  Accordingly, it is also informative to look at trend fits based on unlimited data, provided the focus is on the longest-term fits and there are no obvious outliers.  These longest-term trend fits are less negative than the same fits based on limited data, providing additional reason for selecting slightly less negative trends.  Correspondingly, the observation that the universe of trend fits presented in this filing are more negative than those presented in last year's filing is partially driven by the significant decrease in the LLP's large loss threshold.  But changes in trend fits stemming from changes to the LLP's parameters are not true changes in loss ratios, therefore a tempered change in trend selections in comparison to those selected last year is appropriate.

4. The exclusion of data corresponding to policies associated with TPA agreements within the AOE provision calculation is acceptable.  This change ensures the AOE analysis is not distorted by differences in carrier reporting.

5. The selected DCCE provision appears inadequate.  The LAE provisions are calculated as ratios to losses.  Accordingly, a constant percentage implicitly implies that LAE costs will track with loss costs.  However, the DCCE portion of LAE costs, in particular, does not appear to have exhibited the same decreasing trend as losses, as evidenced by increasing DCCE ratios.  Selecting ratios based on relatively short historical timeframes is not sufficient to adjust for prospective expectations; trends in LAE should be reflected similar to how trends in losses are reflected.  Or rather, when historical trends in LAE dollars do not align with historical trends in loss dollars, loss trends should not be imputed on LAE.

6. NCCI's decision to not include an explicit adjustment for the impact of the COVID-19 pandemic is acceptable.  As discussed within the Disclosure and Filing Narrative sections of the Filing, NCCI has considered direct and indirect pandemic-related factors, but there remains such uncertainty and the unavailability of necessary quantitative data prevents determining credible estimates of impacts for the proposed effective period.  Additional discussion can be found in the document "2020-2021 Rate Filing Season, What You Need To Know"[10].  NCCI has been, and continues to, monitor, research, and analyze COVID-19 impacts on the workers' compensation system.  Details, including current articles and analyses, can be found on their dedicated COVID-19 "Resource Center" webpage.[11]

The above findings do not represent a comprehensive list of the information reviewed and considered by the director, nor does the list imply relative importance or materiality.  Also, these findings are specific to the subject filing and are not meant to apply generally to past or future filings or to provide guidance for future filings unless specifically noted.

ORDER

For the reasons set forth above and in accordance with AS 21.39.043, the director orders:

A. The following modifications should be made to the 2021 loss costs and assigned risk rates:

1. The indemnity loss ratio trend should be revised from -5.5% to -5.0%.
2. The medical loss ratio trend should be revised from -4.0% to -3.5%.
3. The combined LAE provision should be revised from 20.4% to 21.2% of losses (12.3% DCCE and 8.9% AOE).
4. The expense provisions of the assigned risk rate will be updated to include the actual 2021 servicing carrier allowance.

Based on the information provided within the filing, these modifications should result in approximately a -14.9% change in voluntary loss costs, and approximately a -9.6% change in assigned risk rates.

B. NCCI should continue to provide alternate indications in the 2022 filing. The alternative indications should include combinations of the following:

1. Experience Period: 3-year, 4-year, and 5-year;
2. Losses: unlimited and limited;
3. Loss development averages: 2-year, 3-year, and 5-year paid; 3-year, 5-year, and 5-year xhilo paid-plus-case; and
4. Trend assumptions: the assumptions approved in this 2021 filing, at least one alternative that is higher than the trend assumption selected for the 2022 filing and at least one alternative that is lower.

C. In the 2022 filing, NCCI should include:

1. Supporting information related to whether, and to what extent, the frequency and severity of the largest claims are changing.  This can be done by providing pages similar to those included in the 2021 filing in Section E of the Supplemental Data, showing the frequency and severity of claims above $1 million, both in Alaska and countrywide.  NCCI should also provide an update on any research that is being performed related to this issue.
2. Documentation of frequency trend analyses by claim size.

D. All carriers issuing workers’ compensation insurance in Alaska shall use and apply, in strict accord, the loss costs, rating plans, rules, and classifications approved for NCCI, except to the extent a carrier has a deviation approved.

This order is effective October 27, 2020.

Lori Wing-Heier

Director