IFRS17 Risk Adjustment Worked Example Part 1.pdf

The presentation explains the theory and practical modeling of IFRS 17 Risk Adjustment (RA) and compares it with the Solvency II Risk Margin. It begins by outlining four IFRS 17 implementation phases, Gap Analysis, Financial Impact Assessment, Systems Design and Methodology, and Implementation, highlighting that insurers’ responses range from proactive transformation to minimal compliance due to cost pressures, lack of expertise, and market fragmentation.

It then defines the RA as the compensation an insurer requires for bearing non-financial risk, emphasizing that IFRS 17 takes a lifetime-fulfilment view versus Solvency II’s one-year capital view. Regulatory commentary from SAMA (KSA) and the UAE IA shows diverse market practices, common reliance on lognormal claim distributions, and expectations for progressively more refined methodologies aligning with each company’s risk appetite.

The technical core demonstrates worked examples of Solvency II Risk Margin using analytic and simulation methods based on Mack’s chain-ladder model and Merz-Wüthrich one-year formulas. Analytical results show how lifetime reserve risk decomposes into sequential one-year risks. Bootstrap and simulated “actuary-in-the-box” re-reserving methods validate these analytics, linking the lifetime and one-year perspectives. The examples calculate capital profiles, cost-of-capital rates (6 %), and discounting (3 %), explaining how to derive margins under various confidence levels such as 99.5 % VaR.

For IFRS 17, several risk-measure techniques are illustrated: Value-at-Risk (VaR), Tail Value-at-Risk (TVaR), and Proportional Hazard Transform (PHT). IFRS 17 RA is derived as the difference between each risk measure and the mean of discounted fulfilment cash flows. The presentation discusses how TVaR and PHT handle tail risk and skewness better than VaR and provide coherent risk measures. It also explores expressing risk preference via proportional hazard transforms with parameter λ reflecting the entity’s risk tolerance. Cost-of-capital-based RA is contrasted with Solvency II, noting that IFRS 17 uses entity-specific discount and capital rates and a lifetime horizon.

Final sections integrate analytic and simulation frameworks to connect Solvency II and IFRS 17 perspectives. Key findings stress that RA estimation must be consistent across gross, reinsurance, and net levels and that reinsurance modeling cannot rely on simple net-to-gross ratios. The presentation recommends developing richer KPIs, copula-based diversification methods, and automated RA models that are computationally efficient but technically robust.

Concluding lessons emphasize the need for pragmatic vision, sufficient budgets, and leadership engagement. IFRS 17 implementation quality, not mere compliance, will distinguish insurers. True success depends on cross-functional collaboration among actuaries, finance, IT, and underwriting, supported by clear stakeholder communication and continuous improvement.

1. IFRS17 RISK ADJUSTMENT
FOR INSURANCE CONTRACTS
PART 1
ALONG WITH SOLVENCY 2 RISK MARGINS
SYED DANISH ALI
QUANTIFYING RESERVE RISK
WITH WORKED EXAMPLES

2. 1
2
3
4
IFRS17 4 Phases
Overview of Main Points covered
IFRS17 Risk Adjustment Definition & Description
Regional Regulators’ Review Points
CONTENTS
I 2

3. 1
2
3
4
Worked example modeling of Solvency 2 Risk Margin
Worked example modeling of IFRS17 Risk Adjustment
calculations
Metrics; 1) VaR2) TVaR3) PHT
2)Cost of Capital 1) Analytic 2) Simulated 3) VaR
CONTENTS
Final Notes
Data and Descriptive Analytics/Exploratory Data Analysis
3)Sources of statistical uncertainty
I 3

4. 1) IFRS17 4 PHASES
Implementing IFRS17 in a phased manner

5. IFRS17 4 Phases
5
Phase 1 –Gap
Analysis
Phase 4 -
Implementation
Phase 2 –Financial
Impact Assessment FIA
Phase 3 –Systems
Design and
Methodology
IFRS17 explained simply in 3 minutes. Part 1:
https://www.youtube.com/watch?v=9RAacCBTYc8
and Part 2 https://www.youtube.com/watch?v=LXziE9DqMxQ
IFRS4 was a patchwork that was never meant to remain as comprehensive IFRS for Insurance in
the first place. But it took 2 Decades of consultation to arrive at the comprehensive regulation of
IFRS17 that meets the aspirations of the IAIS group finally for insurance contracts.
I 5

6. IFRS17 4 Phases
DifferentvestedinterestsleadtodifferentperceptionsregardingtowhyweareimplementingIFRS17inthefirstplaceatall.Many
insurersdelayIFRS17workforaslongastheycan,whileadheringtoregulatorydeadlinesforIFRS17.Regulatorsalsodifferand
somearemorepro-activethanothers.
Evennowmanypeopleatinsurancearedismissiveorcombativeofifrs17thatitaddsverylessvalue,addstoomanycosts;and
theyaretryingtheirbesttoapplyasminimumaspossible(thinkoftheverystrongpreferenceforPAA,nobudgetforsoftware;
justwantpatchedupexcelbasednewactuarialandaccountingwork).
Ofcourse,theyhavetheirownviewwhichisthattheyfacealotofburningfiresandarefirefightingsomuchalready(increasing
lossratioovertime,theimpactofcovid19,increaseddigitization/productdevelopment,hostofotherchallenges),andthelast
thingtheyneededwasanotherregulation!
Themarketstructureisalsohighlyskewed.OtherthantheTop5insurersinagivencountry,therest20-40insurersfightover
hardlytotal20%-30%ofthemarketshare;theyaretoosmalltohavebudgetstohireanyspecializedskills,manyfunctionsare
missing,theyaretoosimpleandunsophisticated,andtheirfocususuallyissellingclonesofproductsatminimumpricestogain
marketshare.ConsultantsarefarmoreoptimisticbecauseIFRS17hasopenedsourcesofrevenueswithmagnitudespreviously
unthinkable(SomeofwhichgettrickleddowntoemployeesinformofbetterremunerationthanifIFRS17wasn'there).
Noonepartyisright,andtheviewsofallstakeholdersshouldberespectedsolet'sseehowwereachthemiddlecommonground
now.

7. 2) OVERVIEW

8. OVERVIEW

9. 3) IFRS17 RISK
ADJUSTMENT DEFINITION
& DESCRIPTION
Current Practice under IFRS4, detailed description of RA in IFRS17

10. IFRS17 Risk Adjustment Definition & Description

11. IFRS17 Risk Adjustment Definition & Description

12. IFRS17 Risk Adjustment Definition & Description

13. IFRS17 Risk Adjustment Definition & Description
https://www.aasb.gov.au/admin/file/content105/c9/AASB17_07-17.pdf
https://www.actuaries.org/IAA/Documents/Publications/IANs/IAA_IAN100_31August2021.pdf

14. IFRS17 Risk Adjustment Definition & Description
https://www.theactuary.com/2020/06/04/stochastic-claims-
reserving-models-taking-one-year-view

15. IFRS17 Risk Adjustment Definition & Description
https://www.linkedin.com/pulse/one-year-reserve-risk-robert-scarth/https://www.linkedin.com/pulse/actuary-in-the-box-
robert-scarth/

16. 4) REGIONAL
REGULATORS’
REVIEW
POINTS

17. SAMA Review of IFRS17 FIA Reports point on RA

18. SAMA Review of IFRS17 FIA Reports point on RA
Thelognormaldistributionisthemost
commonlyusedclaimsdistribution.Some
insurancecompanieshavevariedtheir
approachesbetweenLRCandLIC
calculations.Itisobviousfromtheabove
thatarangeofapproachesisbeing
adoptedinthesectortoestimatetherisk
adjustment.Theestimaterequiresinput
frombothmanagementandtheactuarial
function.SAMAexpectstheestimation
approachtoberefinedovertimeasthe
actuarialprofessiongrows in
sophisticationintheKingdom.SAMAalso
expectsmanagementtoprovideactive
inputandsteertoensurealignment
betweentheselectedconfidenceinterval
andtheCompany’sriskappetite.As
regardsthecompaniesthatareyetto
completeworkinthisregard,SAMAnotes
thatthiswasagainstitsexpectationsand
itwillfollowupwiththosecompanies

19. IA Regulator UAE Review related to Risk Adjustment

20. IA Regulator UAE Review related to Risk Adjustment

21. 5) DATA AND DESCRIPTIVE
ANALYTICS /EXPLORATORY
DATA ANALYSIS
1) Data used for both Solvency 2 Risk Margin and IFRS17 Risk Adjustment
2) Residuals’ Analysis
3) Sources of Statistical Uncertainty

22. Cumulative
Incremental
Data
Cumulative
Incremental

25. Theseareunscaledresiduals.Variouswaystoscaleresidualsbut
trendsstillvisibleas1)variablesneedtobescaledandnotonly
residuals,2)informationcompressionofmakinga2Dtableand
ignoringeverythingelseinchainladdermakesuslosealotof
information.Chainladderwasfirstmade5decadesorsoagowhen
regressionorindividuallevelreservingwasnotpossiblegenerallyso
modeldiagnosticswillbequitepoorforitnomatterwhatwedo.
Pricinghasprogressedfarbeyondreservingasitistreatedascorepart
ofbusinessandnotasregulatoryburdenofreservingwheretheonly
focusofmanagementisonminimizingthereservefigures.
25

26. These are Adjusted Scaled
residuals.
26

27. These are Zero Average Scaled
residuals.
27

28. actions.(Sources:MacroOps:UnparalleledInvestingResearch(macro-ops.com)IntroductiontoEndogeneity.Anicecreamvendorsellsicecreamon…|byashutoshnayak|TowardsDataScience23
InvestingLessonsfromGeorgeSoros|CaseyResearchhttps://datascienceplus.com/how-to-detect-heteroscedasticity-and-rectify-it/)
28

29. More sources of statistical errors
Collinearityisalinearassociationbetweentwopredictors.Multicollinearityisasituationwheretwoormorepredictorsarehighly
linearlyrelated.Ingeneral,anabsolutecorrelationcoefficientof>0.7amongtwoormorepredictorsindicatesthepresenceof
multicollinearity.
‘Predictors’isthepointoffocushere.Correlationbetweena‘predictorandresponse’isagoodindicationofbetterpredictability.But
correlation‘amongthepredictors’isaproblemtoberectifiedtobeabletocomeupwithareliablemodel.
Forexample,COVID19incidenceis5timeshigherforthosewithhealthinsurancesuminsuredSAR30thousandandabove.Doesit
meanthatweshouldthenflagthosewithhighsuminsureds?Nobecausemostofhealthcomesundergrouphealthwherelargesum
insured/highbenefitplansareforsenioremployeesonlylikeVIPPlan,PlanAetc.Attheseplans,agesarequitehighabove50usually
(thehighertheage,thehighertheCOVID19incidence),arepredominantlymeninseniorposts(menhavehigherchancesofcatching
COVID19),canhavegreaterlevelofawarenessofhealthbenefitsandmoreseriousattitudewithtakingcareoftheirhealth).Thoseare
therealreasonsandnotSuminsuredaboveSAR30thousand.
Source:https://www.statisticshowto.com/multicollinearity/https://blog.clairvoyantsoft.com/correlation-and-collinearity-how-they-can-make-or-break-a-model-
9135fbe6936a
Intuitiveunderstandingofwhatstatisticsdoes(whatiscross-validation,whatist-test,whatdoesplessthan0.5show?WhatisANOVA
practically?)plussomelevelofdomainknowledgeisfarmoreimportantinrealitythanjustcodingandchurningnumbers.Ifyoucan’t
explainitsimplyenough,yourunderstandingislacking.Oppositeattitudetothatofthrowingaroundonlyfancybigwords.Bestexample
startups.TheywillmarketasiftheyhavemadeAGI,butbehinditall,itwillbeasimplelogisticregressionbyajunioremployeewho
doesn’treallyknowwhatheisdoing).

30. 6) WORKED
EXAMPLE
MODELING of
SOLVENCY 2
RISK MARGIN

31. Summary Tables –Analytical Method 1 Results
ThefirsttableontheleftshowsanalyticresultsfortheSDofthe
reservesovertheirlifetime,andtheSDoftheclaims
developmentresult(CDR)over1year.
The two tables shown below shows analytic results for the SD of
the CDRs over a sequence of 1-year views. The results can be
shown incrementally or cumulatively.
Noticethatthesquarerootofthesumofsquaresofthe
incrementalSDsequalstheSDofthereservesovertheir
lifetime.Thisdemonstrateshowthetraditionallifetimeviewof
riskcanbepartitionedintoasequenceofone-yearviews.
Table 1
Table 3
Table 2

32. Summary Tables –Analytics Results
Mackmethodisbaseduponthechainladdermethodwhichisprobablythemostpopularmethodforthecalculationofclaim
reserves.Thistechniquemeasuresthevariabilityofthechainladderestimatesandusesittodevelopaconfidenceintervalforthe
estimatedultimateclaimamountsandtheclaimreserves.TheConfidenceintervalisimportantduetothedifferenceintheactual
ultimateamountandestimatedultimateamount.Theconfidencelevelisbaseduponanentity’spreference.Mackmethoduses
theweightedaveragelinkratios,calculatedbychainladdermethod,todeterminethestandarderroroftheestimatesofthe
ultimateclaimamounts.Themethodassumesthattheage-to-agefactorsareindependentforeachaccidentyear,buttheclaim
paymentsarecorrelatedwiththeearlierpaymentforthatyear.Forthecalculationoftheriskadjustment,wewouldhaveto
assumeanormalorlognormaldistributionfortheclaimamountsanddeterminetheparametersfortheselecteddistribution.
InTable1,thechain-ladderreservesareshown,togetherwiththestandarddeviationsoftheforecasts(RMSEPs)fromMack’s
model,givingacoefficientofvariationofthetotalreservesunderthelifetimeviewofriskof13.1%.Inaddition,theRMSEPsofthe
CDRsover1yearusingtheformulaefromMerzandWuthrichareshowninTable1.TheRMSEPsdividedbytheexpectedreserves
atthestartoftheyeararealsoshown,giving9.5%forthetotalCDR.Thisone-yearmeasureofriskislowerthanthetraditional
lifetimeview.
Table2and3showstheRMSEPs(i.e.,standarddeviations)oftheCDRsforeachfuturecalendarperiod(the“fullpicture”)using
theformulaefromMerzandWuthrich.Theresultofsquaringthevalues(togivevariances),addingupacrossallcolumnswithin
eachrow,andtakingthesquarerootisshowninthefinalcolumn.AcomparisonwithTable2showsthatthesquarerootofthe
sumofsquaresoftheCDRsgivesthesameresultastheRMSEPfromMack’smodeloverthelifetimeoftheliabilities.This
demonstrateshowthelifetimeviewofriskunderMack’smodelcanbepartitionedintoasequenceofone-yearviews.Italso
showsthattheone-yearviewofriskmustalwaysbelowerthanthelifetimeviewsincevariancescannotbenegative.Thisisan
interestingresultandlinksthelifetimeviewofriskwiththeone-yearviewofSolvencyIIusinganalyticapproaches.
32

33. Summary Tables –Bootstrap Method 2 Results
Table 4 Table 5
Simulationassumes3%discountratewhichwecanchange.1,000Simulationhasbeenrunandweshouldgenerallysimulatefrom1,000to10,000simulationruns.
BootstrapmethodforsimulationhasbeentakenasMackalthoughwecouldalsorunitonOverdispersedPoissonNon-ConstantScaleorConstantScaleaswell.
Table4showsanalogousresultstoTable2,butbyusingBootstrapsimulationinstead.Noticethatthesimulationresultsareveryclosetotheanalyticresults,
justifyingtheprocedure.Table4showstheexpectedreserves,standarddeviation(SD)(predictionerror),andcoefficientofvariationfrombootstrappingMack’s
modelusing1,000simulations.Alsoshownarethestandarddeviationsoftheone-yearaheadCDRsusingthere-reservingapproachandthestandarddeviations
expressedasaproportionoftheexpectedreservesatthestartoftheyear.ComparisonwithTable2and3showsthattheexpectedreservesareveryclosetothe
chain-ladderreserves,andthestandarddeviationsofthesimulatedreservesfrombootstrappingMack’smodelareveryclosetotheanalyticresultsgivenbyMack’s
model.Inaddition,thestandarddeviationsoftheone-yearaheadCDRsareveryclosetotheanalyticresultsgivenbytheformulaefromMerzandWuthrich.
Table 5 shows a summary of results on a discounted basis. This highlights one of the benefits of the simulation approach -a full distribution of all cash-flows is
available, which can be used to go beyond the analytic results.
33

34. Summary Tables –Simulated CDR Method 3 Results
Incremental CumulativeTable 6 Table 7
Table 8
Table 9

35. ForecastdistributionhasbeenassumedGammainthemodelingprocess.WecouldalsoselectNon-parametrictocharttheforecastdistribution.
Nonparametricmethodsarestatisticalmethodsthatrequirefewerassumptionsaboutapopulationorprobabilitydistributionandareapplicable
inawiderrangeofsituations.
Table6showstheStandardDeviation(SD)oftheCDRsoverasequenceof1-yearviews,usingthesimulationresults.Noticethatthesimulation
resultsareveryclosetotheanalyticresultsinTable3.Table6isincrementalwhereasTable7iscumulative.AcomparisonwithTable4showsthat
thesquarerootofthesumofsquaresoftheCDRsareveryclosetothestandarddeviationsfrombootstrappingMack’smodeloverthelifetimeof
theliabilities,andagaindemonstrateshowthelifetimeviewofriskunderMack’smodelcanbepartitionedintoasequenceofone-yearviews.
[email protected]%oftheCDRsoverasequenceof1-yearviews,usingthesimulationresults.Again,this
highlightsthebenefitsofasimulationapproachsinceanyriskmeasurecanbeappliedtothesimulateddistribution.Theanalyticapproachonly
providesSDs.
ThevaluesinTables3and6showremarkablesimilarity,validatingthesimulationapproachesandconnectingthelifetimeandone-year
viewsofriskforanalyticandsimulation-basedapproachesassociatedwithMack’smodel.Again,anadvantageofthesimulation-based
approachisthatafullpredictivedistributionisavailable,fromwhichanyriskmeasurecanbeobtained.Forexample,Table8and9shows
valueat-riskoftheCDRsat99.5%(whereVaRat99.5%isthenegativeofthe0.5thpercentileofthedistributionoftheCDR).
Forastatisticalmethodtobeclassifiedasanonparametricmethod,itmustsatisfyoneofthefollowingconditions:(1)themethodisusedwith
qualitativedata,or(2)themethodisusedwithquantitativedatawhennoassumptioncanbemadeaboutthepopulationprobabilitydistribution.
Incaseswherebothparametricandnonparametricmethodsareapplicable,statisticiansusuallyrecommendusingparametricmethodsbecause
theytendtoprovidebetterprecision.Nonparametricmethodsareuseful,however,insituationswheretheassumptionsrequiredbyparametric
methodsappearquestionable.
Source:https://www.britannica.com/science/statistics/Residual-analysis
Summary Tables –Simulated CDR Method 3 Results
35

36. Summary Tables –Simulated CDR Method 3 Results
BybootstrappingMack’smodel,itprovidesawayofsimulatingcumulativepaymentsforallfuturecalendarperiods,andhenceallincrementalpayments(bydifferencingthe
cumulativepayments).Foreachoriginperiod,wethereforehaveawayofsimulatingthepaymentsthatemergeoverthenextcalendarperiod.Allthatremainsistoestimate
theoutstandingliabilitiesattheendoftheyearconditionalonwhathasemerged,foreachsimulation.Thiswilldependnotonlyonthepaymentsmadeoverthenextyearin
originperiodi,butonallotheroriginperiodstoo.
Tocompletetheprocess,itisnecessarytoaugmenttheoriginalpaymentstrianglebythesimulatedpaymentsthatemergeoverthenextcalendarperiodforeachorigin
period.Thatis,theoriginalpaymentstriangleisaugmentedbyonediagonal,sincethatisallanactuaryseesoveraone-yearperiod.Conditionalonthepaymentsthat
emerge(foreachsimulation),itisthennecessarytoestimatethereservesattheendoftheperiod.Atthispoint,anautomatedreservingmethodologyisrequiredthatcan
beappliedtotheresultsforeachsimulation.Anactuaryinthecomputerisrequired,oran“actuary-in-the-box”,astheprocedureisknown1.Toremainconsistentwiththe
underlyingmethodologydescribedinthispresentation,thestandardchain-laddermethodisadoptedforthispurpose.Thatis,foreachnewsimulatedtriangle,thechain-
laddermodelisre-fittedconditionalontheclaimsthathaveemergedintheyear,givingthereservesattheendoftheyear.Thisautomaticre-fittingofthereserving
methodologyhasledtothe“actuary-in-the-box”procedurealsobeingknownas“re-reserving”.
TheActuary-in-the-Boxisageneralprocedureforestimatingone-yearreserverisk.Itassumesthatwealreadyhaveanalgorithmicmethodforsetting
reserves,andthenspecifiesaprocedureforsimulatingthenextyearofclaimsdevelopment,andre-applyingthealgorithmtogetthereservesinoneyear's
time.Themethodis:
1.ObtaintheBestEstimateoftheopeningreserve.Itisassumedthatthisisdoneaccordingtoawell-definedalgorithm,andthatitdoesnotincludeany
riskmargin.
2.Extendtheinputdataneededforthealgorithmusedinstep1bysimulatingonefurtheryearofdata.
3.Applyexactlythesamealgorithmasisstep1totheextendeddatasetgeneratedinstep2toproduceadistributionoftheclosingclaimsreserve.
Onefundamentallimitationisthatthemethodcannotadequatelycapturethejudgementusedbyareal-worldactuaryinsettingreserves,ormanyofthe
othersubtleaspectsofacomplexreservingprocess.Anotherfundamentallimitationisthattheactuary-in-the-boxmethodcannotmakeuseofinformation
notcontainedintheclaimsdatausedbytheunderlyingmodel,whichwouldlikelybeconsideredbyareal-worldactuary.
Theoutputisafulldistributionoftheultimateclaims,whichcanbeusedtocalculateanyriskstatisticdesired.Itcanalsobeiteratedtogivean
understandingofhowtheriskwillemergeupuntilthewholetriangleisfullyrun-off.
WedemonstratethatthestandarddeviationofthesimulateddistributionoftheCDRusingthere-reservingapproachmatchestheanalyticapproachofMerzandWuthrich,
connectingtheanalyticandsimulation-basedapproachesfortheone-yearviewofrisk.Andthenconnecttheone-yearviewofriskandthetraditionallifetimeview.
WedemonstratethatthestandarddeviationofthesimulateddistributionsoftheincrementalCDRsusingtherecursivere-reservingapproachmatchtheanalyticresultsfrom
theMerzandWuthrichformulae,againconnectingtheanalyticandsimulationbasedapproaches,andconnectingtheone-yearviewofriskandthetraditionallifetimeview
36

37. Histograms
Density Charts
of Discounted
Total Reserves
obtained by
Bootstrapping

38. Histograms
Density Charts
of Discounted
Total Reserves
obtained by
Bootstrapping

39. Histograms show a graphical representation of the 1,000 simulations done at each origin points. We can
see the total histogram at around normal distribution but at other origin points show left skew.
Since the bootstrap approach provides distributions of all future cash-flows (not just the reserves), it is
straightforward to obtain a distribution of the discounted reserves. The Histograms show the results of
discounting the future cash-flows at 3% (assuming payments are made mid-way through the year).
Histograms
Density Charts
of Discounted
Total Reserves
obtained by
Bootstrapping
39

40. Claims Development by Origin Period

41. Claims Development by Origin Period

42. As we can see, later origin period have lower data items
and so have greater variability in forecasts because each
successive row in the triangle has lower data points.
42

43. Risk Margins Solvency II Cost of Capital
Thecostofcapitaliscalculatedbyapplyingacostofcapitalratetothisamount.ThisratecanbedeterminedbyseveraltechniquessuchasWeighted
AverageCostofCapital(WACC)andCapitalAssetPricingModel(CAPM).ForIFRS17riskadjustment,theentity’scost-of-capitalratewouldbechosento
meetthespecificmeasurementobjectives,reflectingarateofreturnconsistentwiththeentitybeingindifferentbetweenfulfillinganinsurancecontract
liabilitywitharangeofpossibleoutcomesversusfulfillingaliabilitythatwillgeneratefixedcashflowswiththesameexpectedvalueofcashflowsasthe
insurancecontract.Theamountofcapitalusedtoestimatethecost-of-capitalwilldependonthelevelofsecuritydesired,anassessmentofthe
probabilitiesthatunfavorablecashflowoutcomeswillconsumesomeorallthecapital,andtheentity’slevelofriskaversionregardingtheuncertain,
unfavorableoutcomes.
WithintheSolvencyIIregulatoryregimeinEurope,ariskmarginisrequiredinadditiontoconsideringreservingriskwithininternalcapitalmodelsorwhen
applyingtheStandardFormula.WhereasSolvencyIIconsidersriskoveraone-yeartimehorizon,IFRS17isbasedonthefulfilmentcashflowsovertheir
lifetime.Assuch,thedefinitionsofreserveriskaredifferent,whichneedstoberecognizedandunderstood.itincludesallfourelementsneededtoestimate
capitalrequirements:1.Ariskprofile(distributionofthebasicownfunds)2.Ariskmeasure(value-at-risk)3.Arisktolerancecriterion(99.5%)4.Atime
horizon(oneyear).
SolvencyIIstipulatesthatriskmarginsmustbecalculatedusingacost-of-capitalapproach.Themechanicsoftheapproacharestraightforward.Givencapital
requirementsforeachfutureyearasthereservesrun-off,theriskmarginisthesumofthediscountedcostsofcapital,wherethecostsofcapitalarethe
capitalrequirementsmultipliedbythecost-of-capitalrate.
InTables14-17,thecolumnsare1)‘DiscFutRes’showsprojectedreserves,2)‘Capital’showsprojectedcapitalrequirements.Thesearethereserves
remainingineachfutureperiod,discountedtothestartofthatperiodat3%discountrate(assumingthatpaymentsoccurhalf-waythrougheachyear)and
evaluatedusingthecash-flowsfromthechainladdermodelapplieddeterministically.3)showsCapitalProfilewhichshowsthecapitalrequirementsateach
futureperiodexpressedasapercentageoftheopeningcapitalrequirements4)showsCostofcapitalat6%assumedmultipliedbycapitalrequirementsand
5)showsDiscountedCostofCapital,assuming3%discountrate.
43

44. Risk Margin Calculations Solvency II Cost of Capital
Table10allowscost-of-capitalriskmarginstobecalculated.Capitalamountsarecalculatedgivenaninitial
capitalrequirementanda'capitalprofile’.Thedefaultinitialcapitalrequirementistakenfromthevalue-at-
[email protected]%ofthetotalCDRover1year(showninTable8andTable9incremental/cumulative).Avarietyof
capitalprofilescanbeselected.
TheliabilitiessideoftheopeningSolvencyIIbalancesheetcontainsanestimateoftheexpectedoutstandingliabilities.Each
simulatedbalancesheetoneyearaheadalsocontainsanestimateoftheexpectedoutstandingliabilitiesatthattime,
conditionalonthepaymentsthathaveemergedintheyear.Thisintroducestheconceptoftheprofitorlossonthereserves,
whichisknownastheclaimsdevelopmentresult(CDR)orsimplytherun-offresult.
If at the end of the year, the estimated ultimate cost of claims has gone up, there is a loss on the reserves, since CDR(n+1)i<
0, which must be made up from capital. Similarly, if the estimated ultimate cost of claims at the end of the year has gone
down, there is a profit on the reserves, since CDR(n+1) i> 0. Under Solvency II, it is the change in the ultimate cost of claims
over a one-year time horizon (the profit or loss over one year) that is important, and the Solvency II definition of reserve risk
is in that context. The analogy on the assets side of the balance sheet is the change in the value of assets over one year.
Clearly, the Solvency II definition of reserve risk is different from the traditional actuarial view of risk, which considersthe
outstanding payments over their lifetime.
44

45. Risk Margin Calculations Solvency II Cost of Capital
SD Discounted Reserves SD Undiscounted Reserves
VaRReserves @98.1%
VaRReserves @99.5%
Table 14 Table 15
Table 16 Table 17

46. Risk Margin Calculations Solvency II Cost of Capital
Analytic: SD (Reverse Sum CDRs)
Simulated: SD (Reverse Sum CDRs)
VaR(Reverse Sum CDRs) @97.1%
VaR(Reverse Sum CDRs) @99.5%
Table 18 Table 19
Table 20 Table 21

47. Risk Margin Calculations Solvency II Cost of Capital
Reserves based on
different Capital
profiles in each
future year is
shown in the
graph on the left.
Reverse sum of CDR;
Reverse Sum
simply means;
___+5=11. so 6 is
the reverse sum
here.
47

48. 7) WORKED
EXAMPLE
MODELING of
IFRS17 RISK
ADJUSTMENT

49. IFRS17 Risk Adjustments Calculations
According to IFRS 17: “An entity shall adjust the estimate of the present value of the future cash flows to reflect the compensation that the
entity requires for bearing the uncertainty about the amount and timing of the cash flows that arises from non-financial risk.”
IFRS 17 is more principles based than Solvency II, and does not specify the techniques for calculating the “risk adjustment”,which is just a risk
margin by another name. Although IFRS 17 does not specify the techniques that should be used, it does state that: “If the entityuses a
technique other than the confidence level technique for determining the risk adjustment for non-financial risk, it shall disclose the technique
used and the confidence level corresponding to the results of that technique.” The “confidence level” is the percentile levelofa value-at-risk
measure, although the risk profile associated with the risk measure is not specified. We can infer from the IFRS 17 documentation that the
most appropriate risk profile is the distribution of the discounted fulfilment cash-flows over their lifetime. It is clear, therefore, that IFRS 17
takes the traditional actuarial lifetime view of reserve risk, not the one-year view of Solvency II.
The most obvious techniques to calculate a risk adjustment under IFRS 17 are therefore risk measures applied to the distributionof the
discounted fulfilment cash-flows. Several risk measures have been proposed, including:
1. VaR: Value-at-risk (“confidence level technique”)
2. TVaR: Tail value-at-risk (conditional tail estimation)
3. PHT: Proportional hazards transform
Clearly, there are other possibilities, including multiples of the standard deviation or variance. Given the choice of risk measure, the only
other input is the associated risk tolerance level (that is, percentile level for VaRor TVaR, and proportional hazards parameter for PHT). The
risk adjustment is then the risk measure evaluated at the selected risk tolerance level less the mean.
49

50. IFRS17 Risk Adjustments Calculations
Acost-of-capitalapproachisalsolikelytobepopulargivenitsuseforSolvencyII,althoughacost-of-capitalriskadjustmentunderIFRS17willbedifferent
fromacost-of-capitalriskmarginunderSolvencyII.Althoughitisopentodebate,wecontendthatthecapitalrequirementsinanIFRS17contextwillneed
toconsiderthefulfilmentcash-flowsovertheirremaininglifetimeateachperiod(nottheone-yearviewofSolvencyII),andthecost-of-capitalrateand
discountrateswillbeentityspecific.
Value-at-riskiseasytoexplaintoanon-technicalaudience,andhastheadvantageofsimplicity,butsinceitisbasedonasinglesimulation,couldbeprone
tosimulationerror(althoughtherearetechniquestomitigatethis).Ithasarangefromtheminimumsimulatedvaluetothemaximum,asthepercentile
levelchanges.Ithasbeencriticizedsinceitdoesnotadequatelyrecognizeskewnessnorextremes,norisitacoherentriskmeasuresinceitdoesnotobey
thesub-additivityproperty.
Tailvalue-at-riskisstraightforwardtocalculate.Ithasarangefromthemeantothemaximumsimulatedvalueasthepercentilelevelchanges,andisbetter
atrecognizingskewnessandextremessinceallvaluesaboveagivenpercentilelevelareincludedinthecalculation.Italsohastheadvantageofbeinga
coherentriskmeasure,andcanbeusedforallocationsofrisktosub-groups,wheredistributionshavebeencombinedbeforetheriskmeasureisapplied.
Theproportionalhazardstransform,introducedbyWanginthecontextofinsurance,alsohasarangefromthemeantothemaximumsimulatedvalueas
theassociatedparameterincreasesfrom1toinfinity.Itcouldbearguedthatitisevenbetteratrecognizingskewnessandextremessincetheweights
increaseasthesimulationvaluesincrease,unlikeTVaRwheretheweightsareconstantaboveagivenpercentilelevel.Itisalsoacoherentriskmeasure,and
againcanbeusedforallocationsofrisktosub-groups.
Onemethodofexpressingriskpreferencesisusingariskpreferencemodeltoadjusttheprobabilitydistribution.Suchamodelassignslowerpreference
adjustedprobabilityvaluestomorefavorableoutcomes,i.e.,outcomesthathavelowercashflowliabilitiesthanthemean.Forunfavorable(adverse)
outcomes,i.e.,outcomesthathavehighercashflowliabilitiesthanthemean,higherpreference-adjustedprobabilityvalueswouldbeassigned.Thisclassof
riskpreferencemodelsisreferredtoasproportionalhazardtransforms.theWangTransformprovidesafunctionaltransformationwhichassignshigher
probabilitiestothemoresevereoutcomesbyreducingthecumulativepercentileassociatedwiththelesssevereoutcomes.Thistechniqueenablesthe
probability-weightedbasedcalculationofariskadjustedvalueoftheuncertainliabilities.Ariskpreferenceparameter,lambda(λ),representsthe
compensationforbearingriskandthisparameterisappliedtotheentireprobabilitydistribution.Consequently,riskismeasuredintermsofanadjustment
totheexpectedvaluederivedfromaproportionalhazardtransformoftheprobabilitydistribution.Lambdaisthekeyparameterwhenusingthistechnique
toestimatetheriskadjustment.Thisparameterindicateshowmuchthecompensationwillincrease,whenameasureofriskincreasesbyoneunit.This
parameterisindependentofthenatureofriskandiscloselyrelatedtotheentity’soverallrisktolerance.(Source:IAAIFRS17book/monographonRisk
Adjustment).
50

51. IFRS17 Risk Adjustments Calculations –Risk Metrics
Risk Adjustments using VaR, TVaRand PHT
Equivalent Risk Tolerance Levels Required to
obtain 6% Cost of Capital
Table22showsIFRS17riskadjustmentsobtainedbyapplyingthree
differentriskmeasurestothedistributionofdiscountedoutstanding
liabilitiesovertheirlifetime.ThedistributionissummarizedinTable5andin
theHistograms.TVaR40%isusuallyneartoVaR75%.PHT1.85isalso
aroundVaR75%.VaR75%isausefulbenchmarkandisalsorequiredunder
someregulationssuchasinAustraliaandHongKong.
Table23showstheequivalentrisktolerancelevelsforthecost-of-capitalrisk
marginshowninTable10.Thisisimportantsinceunderthedisclosure
requirementsofIFRS17,the'equivalentconfidencelevel'mustbedisclosed
ifthe'confidenceleveltechnique'(i.e.value-at-risk)isnotused.RAsare
higherinTable22thaninTable23.TheBestEstimateCostofCapital4.69%
showstheanalogousconfidenceintervalsas21.20%and65.39%and1.45
forPHT.IftheseCIsareseemingtoolow,puttingCoC11%insteadof6%can
leadtosimilarlevelsofRAto75%.Analternativeviewisthatthedistribution
givenbyMack’smodelistoowide,andanarrowerdistributionwouldgivea
higherequivalentconfidencelevel.
Table 22
Table 23
51

52. IFRS17 Risk Adjustment Calculations -Cost of Capital
Expected value, standard deviation and value-at-risk of the discounted reserves at each
future period. Also, standard deviation of the undiscounted reserves. Cost-of-capital risk
adjustments are shown for each basis.
The square root of the reverse sum of the CDR MSEPs, together with the
standard deviation and VaRof the reverse sum of simulated CDRs.Cost-of-
capital risk margins are shown for each basis.
Table 24 Table
25
Tables24and25showdifferentbasesthatcouldbeusedtoobtainariskprofileinacost-of-capitalriskadjustment
underIFRS17,ifalifetimeviewofriskisusedforassessinganentity'scapitalrequirements,insteadoftheone-yearview
ofSolvencyII.Ifinsuranceentitiesuseacost-of-capitalapproachforIFRS17,theywillneedtodecidewhetheraone-year
viewisacceptableforcapitalcalculationsunderIFRS17.
InTable24,adistributionoftheremainingdiscountedreservesateachfuturetimepointisused,conditionalon
informationcurrentlyavailable.InTable25,adistributionofthereversesumofCDRsisusedateachfuturetimepoint,
whichisaprudentapproximationtothedistributionoftheremainingdiscountedreservesateachfuturetimepoint,
conditionaloninformationavailableatthattime.
52

53. IFRS17 Risk Adjustment Calculations -Cost of Capital
The risk tolerance level of VaRat 97.1% was selected such that the value in the first year is close to the opening capital
requirement in Table 14, again allowing corresponding risk adjustments to be compared.
It should also be noted that it is not clear what risk tolerance level is appropriate under a cost-of-capital risk adjustment for IFRS
17; the choice is entity specific and is not prescribed.
It should be noted that the recursive re-reserving approach is computationally expensive. Therefore, although it may be better to
use a capital profile obtained from a risk measure applied to a distribution of the reverse sum of CDRs for future capital
requirements under IFRS 17, using the distribution of the discounted outstanding future cash-flows given data up to calendar
period n may be expedient (with the risk tolerance level being used to control the level of prudence).
53

54. 8) FINALNOTES
1) Ending Notes
2) Further areas to develop in RA modeling
3) Key takeaways
4) More Key points
5) Lessons to live by
6) Recap: what we covered in this presentation

55. Ending Notes of Presentation on RA
Inthispresentation,variousconceptsassociatedwiththequantificationofreserveriskhavebeenconnected.Theanalyticformula-basedapproachesof
Mackforthelifetimeviewofreserverisk,andMerzandWuthrichfortheone-yearviewofSolvencyII,havebeencomparedtosimulation-basedresults
obtainedbybootstrappingMack’smodel,supplementedwiththere-reservingapproach.Furthermore,thelifetimeandone-yearviewswerebrought
togetherbyconsideringasequenceofone-yearviewsuntiltheliabilitiesareextinguished.Again,thiswasconsideredanalytically,usingMerzand
Wuthrich,andusingasimulation-basedapproachbyapplyingre-reservingrecursively.
IFRS17riskadjustmentsarealsorequiredonagrossandreinsurancebasis.Clearly,itisthenetpositionthatismostrelevantfortheinterpretationofan
insuranceentity’sfinancialposition,soitseemsappropriatetoestimateriskadjustmentsfromdistributionsofgrossandnetdiscountedfulfilmentcash-
flows,thentakingthedifferenceasthereinsuranceriskadjustment.Reinsurancemodellingtoobtainanaccuratedistributionofthenetdiscounted
fulfilmentcashflows(togetherwithanassessmentofcreditrisk)couldbecomplex.Inparticular,thecurrentactuarialpracticeofapplyinganapproximate
net-to-grossratiolooksincreasinglyinadequate(wherenon-proportionalreinsurancetreatiesexist),andtrianglemethodsforattritionalclaimsmayneed
tobesupplementedbyindividualclaimsmodellingforlargeclaims,withaccuratereinsurancemodelling.Furthermore,riskadjustmentsarerequiredfor
groupsofcontracts,notjustattheaggregateentitylevel(orholdingcompanylevelforamultinationalgroup),whichraisesquestionsaboutallocationof
riskanddiversification.asimulationframeworkcanbeused(usingcopulaetoapplydependencieswhenaggregating),buttheissuesarecomplex.
Ifthecost-of-capitaltechniqueisusedforIFRS17riskadjustments,itshouldberecognizedthatthiswillbedifferentfromaSolvencyIIriskmargin.
SolvencyIIconsiderstheone-yearviewofriskforcapitalrequirements,whereasthelifetimeviewofriskismoreappropriateunderIFRS17.Adistribution
oftheremainingtotalcash-flowsateachfuturetimeperiodismoreappropriateasabasisforestimatingcapitalrequirements(althoughasdiscussedin
section6andAppendix3,thetimeperspectivebecomesimportant).Furthermore,cost-of-capitalanddiscountratesareentityspecificunderIFRS17but
prescribedunderSolvencyII.Thecostof-capitaltechniqueisconsiderablymorecomplexthansimplyapplyingariskmeasuretoadistributionoffulfilment
cash-flows,andrequiresmoreparameterstoselectandjustify;itrequiresanopeningcapitalrequirement,futurecapitalrequirements,acost-of-capital
rateandayieldcurvefordiscounting.Sincetheequivalent“confidencelevel”isrequiredanywayunderIFRS17,itquestionswhythecost-of-capital
methodwouldbeusedatall.Adistributionofdiscountedfulfilmentcash-flowsisrequiredfortheequivalentconfidencelevel,soitseemsmore
straightforwardtocalculateIFRS17riskadjustmentssimplyfromariskmeasureappliedtothatdistribution.Giventhedistribution,theonlyinputtoselect
istheentityspecificrisktolerancelevel.
55

56. Further areas to develop
Insurers need to make new set of KPIs like new ratios for quantitative performance analysis.
Diversification method needs to be worked at such as Copulas, maximum allowance for diversification and so on.
ModificationinRAforLRCisneeded.IAABook/MonographonRiskAdjustmentalsocontainsmanydifferentmethodsforlifeandnon-lifelinesof
businessinChapter10CaseStudies.IAN100containsanswerstomanygeneralqueriesinIFRS17implementationwhichisawelcomesightformuch
neededclarificationandbenchmarkinginsteadofonlyrelyingonmarketconsensusthatcanmightbeaconsensusofmanyindustryplayersbutstillbe
technicalwrong.
AcomprehensiveRAmodelneedstohaveselectionmethods(selectCoC,VaR,TVaR,PHP,),claimintervals(selectmonthlytriangles,quarterly,annual)
andthenworkingforthosebasis.GrossandRItriangles(andNetornot?)needtobeworkedforwhichallclassesofbusiness.Forexample,Motorand
Medicalastheyhavehighfrequencylowseverityclaimsthataredataintensiveandinducivetocrediblemodeling.Modelsneedtobecomprehensivebut
nottakemuchtimetoreachfromdatatofinalstageotherwiseimplementationcanbecomeimpractical.RAforlongtermlifeinsuranceisadifferent
ballgamethandescribedheresoseparatemodelsneedtobedeveloptohandleRAforlongtermlifeinsurance.
RAworkingneedstobeaccommodatedforRAonreinsurancelevel.Anentityneedstocalculateriskadjustmentseparatelyongrossbasisand
reinsurancebasis.Itmustbenotedthattheriskadjustmentfornon-financialriskonthereinsurancecontractisnotthecompensationthatthereinsurer
requiresforbearingthenonfinancialriskonreinsurer’sside.Thereinsurer’sriskadjustmentisdependentuponthereinsurer’sriskappetite,and
methodologyforthisworkingwillhavenodirectbearingontheinsurer’sfinancials.Non-financialrisktransferistobereflectedintheRAfornon-
financialriskonreinsurancebasis.Theriskofnon-performancecreatedbycontractistobereflectedinestimatesoffuturecashflows.Usefultoseeif
companybenefitingfromthereinsurancearrangementornotandreinsuranceoptimizationexercisescanhelpinformthereinsuranceconsiderationsfor
RA.Source:https://www.ifrs.org/content/dam/ifrs/supporting-implementation/ifrs-17/ifrs-17-pocket-guide-on-reinsurance-contracts-held.pdf
56

57. 1
2
3
4
5
6
KEY TAKEAWAYS
Design
Scrutiny
Considerations
Kaizen
Target
Communication

58. More Key Points
All risks need to be measurable and be
quantified.
Measurable
No use if calculating Risk Adjustment needs an
unreachable budget for the company. But
reasonable budgets should also be there otherwise
patchworks can mean quality can suffer. The person
who buys expensive cries once but the person who
buys cheap cries ten times.
Cost effective
The RA needs to work across very diverse lines of
business including motor, medical, short term life,
long-term life, marine, engineering, liabilities and
so on.
Multiple lines

59. Lessons to Live by
Pragmatic Vision and
Budgets
Leadership
Quality
Deep Expertise
IFRS17 is unlike normal work like
reserving or pricing which actuaries
have repeated thousand of times.
This is being done for the first time
worldwide and no one has done A to
Z all of it before so it’s better to
over-prepare than under-prepare as
the consequences of under-
preparation are far worse than of
over-preparation. That vision needs
to be backed up by reasonable
budgets. Going for unreasonably low
budgets mean lots of pain
afterwards.
Deep expertise is needed in
order to implement solutions
that are technically sound in line
with principles of IFRS17 instead
of simple patchworks.
The binary view that
insurer is compliant with
IFRS17 or not compliant is
misleading as quality of
compliance differs
drastically across different
insurers and markets.
Unless the top management of
company takes IFRS17 seriously,
implementation will suffer
drastically. It has been noticed
across various markets that 90%
or more work is done by
consultants but there is extremely
low ownership and knowledge by
company employees of IFRS17.
Collaboration
Across different
segments of business
from Finance to
underwriting to IT and
Actuaries is crucial
Communication
tailored to specific
stakeholders is
key
Collaboration
Communication

60. 01 02 03
04 05 06
Recap -What we covered in this presentation
Phased approach. 1) Gap Analysis, 2)
Financial impact assessment 3)
system design and methodology and
4) implementation.
IFRS17 Phases
Current practice under IFRS4.
Detailed description of RA
requirements under IFRS17. one year
view Vs ultimate view.
IFRS17 Risk Adjustment
Definition & Description
SAMA regulator KSA and IA regulator
of UAE Insurance market review
points and what we can learn from
those points.
Regional Regulators’
review points
Exploratory Data Analysis of data
used in calculation of Solvency 2 Risk
Margin and IFRS17 Risk Adjustments.
Exploratory Data
Analysis
risk metrics. 1) Analytical Mack
method. 2) Bootstrap and 3) Simulated
CDR method.
Cost of Capital. 1) Analytical Mack 2)
Simulated CDR 3) VaR
Solvency 2 Risk Margins
Risk Metrics; 1) VaR2) TVaR3) PHT.
Cost of Capital 1) Analytic 2) Simulated
3) VaR.
IFRS17 Risk Adjustment
Calculations

61. THANK YOU!
A ny Q u e st i o n s ?
SYED DANISH ALI