Study guide
5.5 Interpreting Coefficients & Multiplicative Relativities
CAS Property & Casualty Predictive Analytics PCPA
Track your progress as you read
Finished sections get a check mark.
1 Chapter 1: PCPA Exam & Project Orientation
1.1 PCPA Exam Format, Scoring & Registration 1.2 The PCPA Project: Windows, Deliverables & Grading
2 Chapter 2: Dealing with Data — Sourcing, Importing & Assessing
2.1 Requirements Gathering & Assessing Data Relevance 2.2 Importing, Joining & Reshaping Data Sets 2.3 Exploratory Data Analysis & Evaluating a Data Set 2.4 Data Quality Under ASOP 23 & Modeling Under ASOP 56
3 Chapter 3: Dealing with Data — Conditioning P&C Data for a GLM
3.1 Exposure Bases, Weights & Data Grain 3.2 Missing Data Mechanisms & Imputation Strategies 3.3 Outlier Diagnostics, Trimming & Winsorization 3.4 Variable Transformation & Scaling
4 Chapter 4: Model Diagnostics & Selection — Building a GLM
4.1 Exponential Dispersion Family & Link Functions 4.2 Poisson Regression & Exposure Offsets 4.3 Overdispersion, Quasi-Poisson & Negative Binomial 4.4 Gamma & Inverse Gaussian Models for Claim Severity
5 Chapter 5: Model Diagnostics & Selection — Model Form & Predictors
5.1 Tweedie Compound Poisson-Gamma Modeling 5.2 Binomial & Logistic GLMs for Binary Outcomes 5.3 Categorical Predictors, Base Levels & High-Cardinality Grouping 5.4 Interaction Terms & Non-Linear Effects 5.5 Interpreting Coefficients & Multiplicative Relativities
6 Chapter 6: Model Diagnostics & Selection — Evaluating and Improving the Fit
6.1 Deviance, Likelihood Ratio Tests & Residual Diagnostics 6.2 Stepwise Selection & Information Criteria Penalties 6.3 Multicollinearity, Correlated Predictors & Aliasing 6.4 Under-fitting, Over-fitting & Cross-Validation
7 Chapter 7: Model Diagnostics & Selection — Validation & Model Choice
7.1 Actual vs. Predicted, Calibration & Loss Ratio Charts 7.2 Simple Quantile Plots & Measuring Lift 7.3 Double Lift Charts & Competing Model Evaluation 7.4 Gini Index, Lorenz Curves & ROC Curves
8 Chapter 8: Model Interpretation & Presentation — Tables, Charts & Exhibits
8.1 Designing Tables & Statistical Exhibits 8.2 Chart Selection & Graphical Perception 8.3 How Charts Mislead 8.4 One-Way, Two-Way & Partial Dependence Exhibits
9 Chapter 9: Model Interpretation & Presentation — Communicating the Analysis
9.1 Communicating Methodology & Modeling Decisions 9.2 Explaining Why a Variable Was or Was Not Included 9.3 Communicating to Non-Technical Audiences 9.4 The PCPA Technical Report: Structure & Rubric
5.5 Interpreting Coefficients & Multiplicative Relativities
Key Takeaways Explain This Topic AI
In a log-link GLM, the rating relativity for level j of a categorical risk factor is obtained by exponentiating its estimated regression coefficient: Relativity_j = exp(beta_j), with the reference base category fixed at 1.000.
Actuarial best practice for selecting the reference base category requires choosing the level with the largest earned exposure, the highest statistical credibility, and a homogeneous, intuitive risk profile.
Confidence intervals for rating relativities must be computed on the linear predictor scale and then exponentiated: [exp(beta - 1.96 SE), exp(beta + 1.96 SE)], producing an asymmetric interval that reflects log-normal sampling variation.
The pure premium rating algorithm synthesizes frequency and severity models into a single multiplicative structure: Pure Premium = Base Rate * Factor_Territory * Factor_Class * Factor_Vehicle * Factor_LimitDeductible.
Re-basing rating relativities shifts all level factors by a constant divisor while adjusting the base rate inversely, leaving total indicated premium for every policyholder completely invariant.
Last updated: September 2026
From GLM Coefficients to Multiplicative Relativities
The primary objective of property and casualty predictive modeling is to construct an intuitive, statistically sound, and regulatory-compliant rating manual. Under a Generalized Linear Model with a log link, regression coefficients translate directly into multiplicative rating factors (also known as relativities ).
Consider a frequency model with K K K categorical rating variables (e.g., Territory, Driver Age Class, Vehicle Symbol). Using standard dummy coding (reference cell coding), the linear predictor for policy i i i is: ln ( μ i ) = β 0 + ∑ k = 1 K ∑ j = 1 J k − 1 β k , j X i , k , j \ln(\mu_i) = \beta_0 + \sum_{k=1}^K \sum_{j=1}^{J_k - 1} \beta_{k,j} X_{i,k,j} ln ( μ i ) = β 0 + k = 1 ∑ K j = 1 ∑ J k − 1 β k , j X i , k , j
Where:
β 0 \beta_0 β 0 is the intercept parameter.
J k J_k J k is the number of categorical levels in variable k k k .
X i , k , j = 1 X_{i,k,j} = 1 X i , k , j = 1 if policy i i i belongs to level j j j of variable k k k , and 0 0 0 otherwise.
One category for each variable is designated as the reference (base) category , for which no dummy variable is created (i.e., its parameter is fixed at β k , base = 0 \beta_{k,\text{base}} = 0 β k , base = 0 ).
Exponentiating both sides of the equation recovers the expected frequency on the original scale: μ i = exp ( β 0 ) × ∏ k = 1 K ∏ j = 1 J k − 1 exp ( β k , j X i , k , j ) \mu_i = \exp(\beta_0) \times \prod_{k=1}^K \prod_{j=1}^{J_k - 1} \exp(\beta_{k,j} X_{i,k,j}) μ i = exp ( β 0 ) × k = 1 ∏ K j = 1 ∏ J k − 1 exp ( β k , j X i , k , j )
Definitions of Core Rating Components
Base Rate (or Base Frequency) : Base Rate = exp ( β 0 ) \text{Base Rate} = \exp(\beta_0) Base Rate = exp ( β 0 )
The base rate represents the expected frequency (or loss cost) for a hypothetical baseline policyholder whose characteristics fall into the reference category for every rating factor.
Reference Category Relativity :
For the designated base level of any rating variable, X = 0 X = 0 X = 0 , which yields a relativity of exactly 1.000 1.000 1.000 : Relativity base = exp ( 0 ) = 1.000 \text{Relativity}_{\text{base}} = \exp(0) = 1.000 Relativity base = exp ( 0 ) = 1.000
Rating Factor (Relativity) for Level j j j :
For any non-base level j j j , the rating factor R k , j R_{k,j} R k , j represents the proportional multiplier relative to the base class: R k , j = exp ( β ^ k , j ) R_{k,j} = \exp(\hat{\beta}_{k,j}) R k , j = exp ( β ^ k , j )
If β ^ = 0.250 \hat{\beta} = 0.250 β ^ = 0.250 , the rating factor is exp ( 0.250 ) = 1.2840 \exp(0.250) = 1.2840 exp ( 0.250 ) = 1.2840 (a + 28.4 % +28.4\% + 28.4% surcharge).
If β ^ = − 0.150 \hat{\beta} = -0.150 β ^ = − 0.150 , the rating factor is exp ( − 0.150 ) = 0.8607 \exp(-0.150) = 0.8607 exp ( − 0.150 ) = 0.8607 (a − 13.93 % -13.93\% − 13.93% discount).
Actuarial Best Practices for Base Category Selection
In statistical software, default dummy coding frequently assigns the reference level to the first category alphabetically or numerically (e.g., Territory 01 or Driver Age 16). In actuarial pricing, allowing software defaults to dictate the base category is a serious mistake.
Actuaries adhere to three core principles when selecting the reference level for each rating variable:
1. Maximum Exposure Volume (Highest Credibility)
The reference category should be the level with the largest earned exposure (e.g., adult drivers age 35–49, standard suburban territory, mid-size passenger sedan). Because all other level coefficients represent contrasts against the base level: β ^ j = ln ( μ j μ base ) \hat{\beta}_j = \ln\left( \frac{\mu_j}{\mu_{\text{base}}} \right) β ^ j = ln ( μ base μ j )
Choosing a base category with massive exposure minimizes the variance of the intercept Var ( β ^ 0 ) \text{Var}(\hat{\beta}_0) Var ( β ^ 0 ) and minimizes the covariance between contrast coefficients. If a tiny niche category with few claims is chosen as the base (e.g., teenage drivers or exotic sports cars), sampling noise in that reference category propagates instability across every single relativity in the rating algorithm.
2. Risk Homogeneity and Stability
The base category should exhibit consistent, stable loss experience from year to year. Selecting volatile categories (such as coastal hurricane territories or high-performance motorcycles) introduces year-over-year rate turbulence into the base rate.
3. Intuitive Benchmark for Regulators and Consumers
In regulatory filings and consumer marketing, rate manuals are judged on transparency. Regulators expect the base class to represent a standard, average, or low-risk benchmark. Surcharging youthful drivers or high-density urban areas relative to a mature, suburban standard is intuitively understood; discounting all drivers relative to an ultra-high-risk teenage base class confuses regulators and consumers.
Statistical Significance and Confidence Intervals for Relativities
To determine whether a rating factor differs significantly from the baseline, actuaries evaluate the estimated coefficient's standard error.
The Wald Test
For any coefficient β ^ j \hat{\beta}_j β ^ j , we test the null hypothesis that the level does not differ from the base category: H 0 : β j = 0 ⟺ H 0 : Relativity j = 1.000 H_0: \beta_j = 0 \iff H_0: \text{Relativity}_j = 1.000 H 0 : β j = 0 ⟺ H 0 : Relativity j = 1.000 H 1 : β j ≠ 0 ⟺ H 1 : Relativity j ≠ 1.000 H_1: \beta_j \ne 0 \iff H_1: \text{Relativity}_j \ne 1.000 H 1 : β j = 0 ⟺ H 1 : Relativity j = 1.000
The Wald statistic z z z is computed as: z = β ^ j SE ( β ^ j ) ∼ N ( 0 , 1 ) z = \frac{\hat{\beta}_j}{\text{SE}(\hat{\beta}_j)} \sim \mathcal{N}(0, 1) z = SE ( β ^ j ) β ^ j ∼ N ( 0 , 1 )
Alternatively, the squared Wald statistic follows a chi-square distribution with 1 degree of freedom: W = z 2 ∼ χ 1 2 W = z^2 \sim \chi^2_1 W = z 2 ∼ χ 1 2 . If ∣ z ∣ > 1.96 |z| > 1.96 ∣ z ∣ > 1.96 (or W > 3.841 W > 3.841 W > 3.841 ), the relativity is statistically significant at the α = 0.05 \alpha = 0.05 α = 0.05 level.
Calculating Confidence Intervals for Relativities
Important
The Asymmetric Confidence Interval Rule A frequent exam error is constructing confidence intervals directly on the exponentiated relativity scale: exp ( β ^ ) ± 1.96 ⋅ SE ( β ^ ) \exp(\hat{\beta}) \pm 1.96 \cdot \text{SE}(\hat{\beta}) exp ( β ^ ) ± 1.96 ⋅ SE ( β ^ ) . This is mathematically invalid because the distribution of the exponentiated estimator R ^ = exp ( β ^ ) \hat{R} = \exp(\hat{\beta}) R ^ = exp ( β ^ ) is skewed (log-normal), not symmetric Gaussian.
The correct procedure calculates the 95 % 95\% 95% confidence limits on the linear predictor scale and exponentiates the lower and upper bounds: CI 95 % ( β j ) = [ β ^ j − 1.96 ⋅ SE ( β ^ j ) , β ^ j + 1.96 ⋅ SE ( β ^ j ) ] \text{CI}_{95\%}(\beta_j) = \left[ \hat{\beta}_j - 1.96 \cdot \text{SE}(\hat{\beta}_j), \; \hat{\beta}_j + 1.96 \cdot \text{SE}(\hat{\beta}_j) \right] CI 95% ( β j ) = [ β ^ j − 1.96 ⋅ SE ( β ^ j ) , β ^ j + 1.96 ⋅ SE ( β ^ j ) ] CI 95 % ( R j ) = [ exp ( β ^ j − 1.96 ⋅ SE ( β ^ j ) ) , exp ( β ^ j + 1.96 ⋅ SE ( β ^ j ) ) ] \text{CI}_{95\%}(R_j) = \left[ \exp\left(\hat{\beta}_j - 1.96 \cdot \text{SE}(\hat{\beta}_j)\right), \; \exp\left(\hat{\beta}_j + 1.96 \cdot \text{SE}(\hat{\beta}_j)\right) \right] CI 95% ( R j ) = [ exp ( β ^ j − 1.96 ⋅ SE ( β ^ j ) ) , exp ( β ^ j + 1.96 ⋅ SE ( β ^ j ) ) ]
Worked Numerical Example
An actuary fits a commercial property frequency GLM. For the "Wood Frame" construction class, the model yields:
- Coefficient: β ^ = 0.2850 \hat{\beta} = 0.2850 β ^ = 0.2850
- Standard Error: SE ( β ^ ) = 0.0420 \text{SE}(\hat{\beta}) = 0.0420 SE ( β ^ ) = 0.0420
Step 1: Calculate the point relativity: R = exp ( 0.2850 ) = 1.3298 R = \exp(0.2850) = 1.3298 R = exp ( 0.2850 ) = 1.3298
(Wood Frame construction carries an indicated + 32.98 % +32.98\% + 32.98% frequency surcharge relative to the Fire-Resistive base class).
Step 2: Calculate bounds on the linear scale: Lower Bound β = 0.2850 − 1.96 ( 0.0420 ) = 0.2850 − 0.08232 = 0.20268 \text{Lower Bound}_{\beta} = 0.2850 - 1.96(0.0420) = 0.2850 - 0.08232 = 0.20268 Lower Bound β = 0.2850 − 1.96 ( 0.0420 ) = 0.2850 − 0.08232 = 0.20268 Upper Bound β = 0.2850 + 1.96 ( 0.0420 ) = 0.2850 + 0.08232 = 0.36732 \text{Upper Bound}_{\beta} = 0.2850 + 1.96(0.0420) = 0.2850 + 0.08232 = 0.36732 Upper Bound β = 0.2850 + 1.96 ( 0.0420 ) = 0.2850 + 0.08232 = 0.36732
Step 3: Exponentiate to obtain relativity confidence bounds: Lower Limit R = exp ( 0.20268 ) = 1.2247 \text{Lower Limit}_R = \exp(0.20268) = 1.2247 Lower Limit R = exp ( 0.20268 ) = 1.2247 Upper Limit R = exp ( 0.36732 ) = 1.4438 \text{Upper Limit}_R = \exp(0.36732) = 1.4438 Upper Limit R = exp ( 0.36732 ) = 1.4438
The 95 % 95\% 95% confidence interval for the relativity is [ 1.225 , 1.444 ] [1.225, 1.444] [ 1.225 , 1.444 ] . Because the interval does not contain 1.000 1.000 1.000 , the surcharge is statistically significant at p < 0.05 p < 0.05 p < 0.05 . Notice that the interval is asymmetric around the point estimate ( 1.330 − 1.225 = 0.105 1.330 - 1.225 = 0.105 1.330 − 1.225 = 0.105 below, while 1.444 − 1.330 = 0.114 1.444 - 1.330 = 0.114 1.444 − 1.330 = 0.114 above).
Assembling the Multiplicative Rating Algorithm
Actuarial ratemaking typically splits pure premium modeling into two separate components: Pure Premium = E [ Loss Cost ] = E [ Frequency ] × E [ Severity ] \text{Pure Premium} = \mathbb{E}[\text{Loss Cost}] = \mathbb{E}[\text{Frequency}] \times \mathbb{E}[\text{Severity}] Pure Premium = E [ Loss Cost ] = E [ Frequency ] × E [ Severity ]
Frequency Model : Poisson or Negative Binomial GLM with log link and ln ( exposure ) \ln(\text{exposure}) ln ( exposure ) offset.
Severity Model : Gamma GLM with log link and prior weights equal to claim count.
Integrating Frequency and Severity Relativities
Because both models employ the log link, their predictions combine multiplicatively for each rating factor k k k : Pure Premium i = ( Base Freq × ∏ k = 1 K R Freq , k ) × ( Base Sev × ∏ k = 1 K R Sev , k ) \text{Pure Premium}_i = \left( \text{Base Freq} \times \prod_{k=1}^K R_{\text{Freq}, k} \right) \times \left( \text{Base Sev} \times \prod_{k=1}^K R_{\text{Sev}, k} \right) Pure Premium i = ( Base Freq × k = 1 ∏ K R Freq , k ) × ( Base Sev × k = 1 ∏ K R Sev , k ) Pure Premium i = ( Base Freq × Base Sev ) × ∏ k = 1 K ( R Freq , k × R Sev , k ) \text{Pure Premium}_i = (\text{Base Freq} \times \text{Base Sev}) \times \prod_{k=1}^K \left( R_{\text{Freq}, k} \times R_{\text{Sev}, k} \right) Pure Premium i = ( Base Freq × Base Sev ) × k = 1 ∏ K ( R Freq , k × R Sev , k ) Pure Premium i = Base Pure Premium × ∏ k = 1 K R Combined , k \text{Pure Premium}_i = \text{Base Pure Premium} \times \prod_{k=1}^K R_{\text{Combined}, k} Pure Premium i = Base Pure Premium × k = 1 ∏ K R Combined , k
Where:
Base Pure Premium = Base Freq × Base Sev \text{Base Pure Premium} = \text{Base Freq} \times \text{Base Sev} Base Pure Premium = Base Freq × Base Sev
R Combined , k = R Freq , k × R Sev , k R_{\text{Combined}, k} = R_{\text{Freq}, k} \times R_{\text{Sev}, k} R Combined , k = R Freq , k × R Sev , k
From Pure Premium to Final Gross Policy Premium
To convert modeled pure premium into final charged premium, actuaries incorporate underwriting expenses, taxes, licenses, and underwriting profit loadings: Gross Premium = Pure Premium + Fixed Expense per Exposure 1 − Variable Expense Ratio − Target Underwriting Profit Ratio \text{Gross Premium} = \frac{\text{Pure Premium} + \text{Fixed Expense per Exposure}}{1 - \text{Variable Expense Ratio} - \text{Target Underwriting Profit Ratio}} Gross Premium = 1 − Variable Expense Ratio − Target Underwriting Profit Ratio Pure Premium + Fixed Expense per Exposure
Worked Policy Rating Table
- Rating Dimension Assigned Category Frequency Relativity Severity Relativity Combined Relativity
- Base Baseline Standard Profile 0.080 0.080 0.080 claims/year $4,500 /claim $360.00 Base Pure Premium
Territory Metro Urban (Terr 12) 1.250 1.250 1.250 1.100 1.100 1.100 1.250 × 1.100 = 1.375 1.250 \times 1.100 = \mathbf{1.375} 1.250 × 1.100 = 1.375
Driver Age Youthful (Age 19) 1.800 1.800 1.800 1.050 1.050 1.050 1.800 × 1.050 = 1.890 1.800 \times 1.050 = \mathbf{1.890} 1.800 × 1.050 = 1.890
Vehicle Symbol High-Performance Coupe 1.150 1.150 1.150 1.400 1.400 1.400 1.150 × 1.400 = 1.610 1.150 \times 1.400 = \mathbf{1.610} 1.150 × 1.400 = 1.610
Deductible $1,000 Comprehensive 0.900 0.900 0.900 0.850 0.850 0.850 0.900 × 0.850 = 0.765 0.900 \times 0.850 = \mathbf{0.765} 0.900 × 0.850 = 0.765
Calculation of Policy Pure Premium: Pure Premium = $ 360.00 × 1.375 × 1.890 × 1.610 × 0.765 = $ 1 , 152.88 \text{Pure Premium} = \$360.00 \times 1.375 \times 1.890 \times 1.610 \times 0.765 = \$1,152.88 Pure Premium = $360.00 × 1.375 × 1.890 × 1.610 × 0.765 = $1 , 152.88
Re-Basing Relativities & Mathematical Premium Invariance
After model estimation, business, marketing, or regulatory requirements may dictate changing the reference base category. For instance, the marketing department may request that the most popular policy deductible ($1,000) serve as the base category (Relativity = 1.000 = 1.000 = 1.000 ), even though the GLM was originally fit with a $500 deductible reference class.
The Re-Basing Transformation Rules
Suppose variable k k k has estimated relativities R 1 , R 2 , … , R J R_1, R_2, \dots, R_J R 1 , R 2 , … , R J relative to Old Base Level 1 ( R 1 = 1.000 R_1 = 1.000 R 1 = 1.000 ). To shift the base to New Base Level m m m (so that R m ′ = 1.000 R'_m = 1.000 R m ′ = 1.000 ):
Transform every level relativity by dividing by the old relativity of the new base: R j ′ = R j R m ∀ j ∈ { 1 , … , J } R'_j = \frac{R_j}{R_m} \quad \forall \quad j \in \{1, \dots, J\} R j ′ = R m R j ∀ j ∈ { 1 , … , J }
Transform the Base Rate by multiplying by the old relativity of the new base: Base Rate ′ = Base Rate × R m \text{Base Rate}' = \text{Base Rate} \times R_m Base Rate ′ = Base Rate × R m
Mathematical Proof of Premium Invariance
For any policyholder in level j j j , the indicated premium under the re-based rating manual is: Premium j ′ = Base Rate ′ × R j ′ = ( Base Rate × R m ) × ( R j R m ) \text{Premium}'_j = \text{Base Rate}' \times R'_j = (\text{Base Rate} \times R_m) \times \left( \frac{R_j}{R_m} \right) Premium j ′ = Base Rate ′ × R j ′ = ( Base Rate × R m ) × ( R m R j ) Premium j ′ = Base Rate × R j = Premium j \text{Premium}'_j = \text{Base Rate} \times R_j = \text{Premium}_j Premium j ′ = Base Rate × R j = Premium j
The term R m R_m R m cancels out entirely. The total premium charged to every single policyholder in the book is completely invariant to the choice of base category.
Worked Re-Basing Example
Deductible Level Original Relativity ( R R R ) Re-Basing Calculation ( R / 0.800 R / 0.800 R /0.800 ) Re-Based Relativity ( R ′ R' R ′ )
- $500 (Old Base) 1.000 1.000 1.000 1.000 / 0.800 1.000 / 0.800 1.000/0.800 1.250 1.250 1.250
- $1,000 (New Base) 0.800 0.800 0.800 0.800 / 0.800 0.800 / 0.800 0.800/0.800 1.000 1.000 1.000
- $2,500 0.600 0.600 0.600 0.600 / 0.800 0.600 / 0.800 0.600/0.800 0.750 0.750 0.750
- Original Base Rate: $250.00
New Re-Based Base Rate: 250.00 × 0.800 = $ 200.00 250.00 \times 0.800 = \mathbf{\text{\textdollar}200.00} 250.00 × 0.800 = $ 200.00
Verification of policy premiums:
Policy with $500 Deductible: Old = $ 250 × 1.000 = $ 250.00 = \text{\textdollar}250 \times 1.000 = \text{\textdollar}250.00 = $ 250 × 1.000 = $ 250.00 ; New = $ 200 × 1.250 = $ 250.00 = \text{\textdollar}200 \times 1.250 = \mathbf{\text{\textdollar}250.00} = $ 200 × 1.250 = $ 250.00 .
Policy with $1,000 Deductible: Old = $ 250 × 0.800 = $ 200.00 = \text{\textdollar}250 \times 0.800 = \text{\textdollar}200.00 = $ 250 × 0.800 = $ 200.00 ; New = $ 200 × 1.000 = $ 200.00 = \text{\textdollar}200 \times 1.000 = \mathbf{\text{\textdollar}200.00} = $ 200 × 1.000 = $ 200.00 .
Policy with $2,500 Deductible: Old = $ 250 × 0.600 = $ 150.00 = \text{\textdollar}250 \times 0.600 = \text{\textdollar}150.00 = $ 250 × 0.600 = $ 150.00 ; New = $ 200 × 0.750 = $ 150.00 = \text{\textdollar}200 \times 0.750 = \mathbf{\text{\textdollar}150.00} = $ 200 × 0.750 = $ 150.00 .
Premium invariance holds with absolute mathematical precision across all policy tiers.
Actuarial Exam Traps & Pitfalls
Warning
Trap 1: Symmetric Exponentiated Confidence Intervals Never calculate confidence intervals as exp ( β ^ ) ± 1.96 ⋅ SE ( β ^ ) \exp(\hat{\beta}) \pm 1.96 \cdot \text{SE}(\hat{\beta}) exp ( β ^ ) ± 1.96 ⋅ SE ( β ^ ) . The delta method or direct exponentiation of linear bounds must be used: [ exp ( β ^ − 1.96 ⋅ SE ) , exp ( β ^ + 1.96 ⋅ SE ) ] [\exp(\hat{\beta} - 1.96 \cdot \text{SE}), \exp(\hat{\beta} + 1.96 \cdot \text{SE})] [ exp ( β ^ − 1.96 ⋅ SE ) , exp ( β ^ + 1.96 ⋅ SE )] . The resulting confidence interval is always asymmetric around the point relativity.
Warning
Trap 2: Forgetting to Adjust the Base Rate When Re-Basing Dividing rating relativities by a new base factor without adjusting the base rate alters the overall premium collected by the company. The base rate must be multiplied by the same factor used to divide the relativities to preserve portfolio revenue neutrality.
Warning
Trap 3: Credibility Weighting Relativities Directly on the Multiplicative Scale When applying classical credibility Z Z Z to complement of credibility relativities R 0 R_0 R 0 , actuaries must credibility-weight the relativities carefully. In log-link models, credibility is theoretically applied on the linear scale: β ^ cred = Z β ^ + ( 1 − Z ) β 0 \hat{\beta}_{\text{cred}} = Z \hat{\beta} + (1 - Z) \beta_0 β ^ cred = Z β ^ + ( 1 − Z ) β 0 , so that R cred = exp ( β ^ cred ) = R Z ⋅ R 0 1 − Z R_{\text{cred}} = \exp(\hat{\beta}_{\text{cred}}) = R^Z \cdot R_0^{1-Z} R cred = exp ( β ^ cred ) = R Z ⋅ R 0 1 − Z (geometric weighting), or linearly weighted as Z R + ( 1 − Z ) R 0 Z R + (1 - Z) R_0 ZR + ( 1 − Z ) R 0 with an off-balance correction.