Figures & data
Figure 1. (lhs) Conditional Gaussian densities for
; (middle) conditional probability
as a function of
; (rhs) densities of claims Y for age X = 40 and genders D = 0, 1.
![Figure 1. (lhs) Conditional Gaussian densities f(x|d) for d∈D={0,1}; (middle) conditional probability P(D=0|X=x) as a function of x∈R; (rhs) densities of claims Y for age X = 40 and genders D = 0, 1.](/cms/asset/9b61f25f-df12-41ef-886b-b270444acc42/sact_a_2364741_f0001_oc.jpg)
Figure 3. Average excess premium for women D = 0 compared to men D = 1, in Example 2.14, as a function of . The dashed vertical line corresponds to the baseline scenario of
,
.
![Figure 3. Average excess premium for women D = 0 compared to men D = 1, in Example 2.14, as a function of Cor(X,D). The dashed vertical line corresponds to the baseline scenario of x0=35,x1=45, Cor(X,D)=0.447.](/cms/asset/9d10a9af-1c82-49a0-94e6-9dc800424375/sact_a_2364741_f0003_oc.jpg)
Table 1. MSEs and average prediction of the different prices in Example 2.14.
Figure 4. Example 2.14, revisited: conditional densities , for
, and two different choices for
,
; for a formal definition we refer to (Equation31
(31)
(31) )–(Equation32
(32)
(32) ).
![Figure 4. Example 2.14, revisited: conditional densities fd(x)=f(x|D=d), for d∈{0,1}, and two different choices for f+(x), x∈R; for a formal definition we refer to (Equation31(31) F+(x)=12Φ(x−x0τ)+12Φ(x−x1τ),(31) )–(Equation32(32) F+(x)=12Φ(x−(x0+x1)/2τ).(32) ).](/cms/asset/9bd90035-03c5-4090-b5ba-e18b5dc68323/sact_a_2364741_f0004_oc.jpg)
Table 2. Wasserstein distances for the two examples (Equation31
(31)
(31) )–(Equation32
(32)
(32) ) for
.
Figure 5. OT maps for examples (Equation31
(31)
(31) )–(Equation32
(32)
(32) ) of
with the original age X on the x-axis and the transformed ages
on the y-axis; the black dotted line is the 45
diagonal.
![Figure 5. OT maps Td for examples (Equation31(31) F+(x)=12Φ(x−x0τ)+12Φ(x−x1τ),(31) )–(Equation32(32) F+(x)=12Φ(x−(x0+x1)/2τ).(32) ) of F+ with the original age X on the x-axis and the transformed ages X+=Td(X) on the y-axis; the black dotted line is the 45o diagonal.](/cms/asset/0ded3f3d-4af9-40ad-aff3-0aa1e8d12eff/sact_a_2364741_f0005_oc.jpg)
Figure 6. OT input transformed model prices for examples (Equation31
(31)
(31) )–(Equation32
(32)
(32) ) of
.
![Figure 6. OT input transformed model prices μ^(X+) for examples (Equation31(31) F+(x)=12Φ(x−x0τ)+12Φ(x−x1τ),(31) )–(Equation32(32) F+(x)=12Φ(x−(x0+x1)/2τ).(32) ) of F+.](/cms/asset/2972c129-65dd-41df-9f31-b09cdd92dbf2/sact_a_2364741_f0006_oc.jpg)
Table 3. MSEs and average prediction of the different prices in Example 2.14.
Figure 7. Changed age profiles with (women) and
(men): (lhs) conditional probability
as a function of
; (middle) best-estimate, unawareness and discrimination-free insurance prices; (rhs) OT input transformed model prices
for example (Equation32
(32)
(32) ) of
.
![Figure 7. Changed age profiles with x0=45 (women) and x1=35 (men): (lhs) conditional probability P(D=0|X=x) as a function of x∈R; (middle) best-estimate, unawareness and discrimination-free insurance prices; (rhs) OT input transformed model prices μ^(X+) for example (Equation32(32) F+(x)=12Φ(x−(x0+x1)/2τ).(32) ) of F+.](/cms/asset/0febc64d-7293-4c45-99f5-e9eb2a1fee49/sact_a_2364741_f0007_oc.jpg)
Table 4. Changed role of ages of women and men, setting and
.
Figure 9. (Top) OT output post-processed prices expressed in their original features
and separated by gender
, see (Equation38
(38)
(38) ); (bottom-lhs) OT input pre-processing taken from Figure ; (bottom-rhs) unawareness price and DFIP taken from Figure .
![Figure 9. (Top) OT output post-processed prices μ+=μ+(x;d) expressed in their original features x and separated by gender d, see (Equation38(38) (x,d)↦μ+=μ+(x;d)=G+−1∘Gd(μ(x,d))∈R.(38) ); (bottom-lhs) OT input pre-processing taken from Figure 6; (bottom-rhs) unawareness price and DFIP taken from Figure 2.](/cms/asset/9355ea08-2298-4702-9989-5fb5676da3b9/sact_a_2364741_f0009_oc.jpg)
Table 5. MSEs and average prediction of the different prices in Example 2.14.
Table