Meyers

Question 1

KS test statistic at 5%

Answer 1

D = 136/sqrt(n)

Question 2

Adjustments to Mack and ODP in Meyers

Answer 2

Variable row parameters
AY correlation
CY trend
Changing settlement rate

Question 3

Reasons for model not validating

Answer 3

Insurance loss environment has experienced changes not yet observable
Other models can validate

Question 4

Heavy tailed p-p plot

Answer 4

Steep slope in corners

Question 5

Light tailed p-p plot

Answer 5

Shallow slope in corners

Question 6

Biased upwards p-p plot

Answer 6

Curve is bowed down

Question 7

Anderson Darling results

Answer 7

All models fail

Question 8

Requirements for LCL parameters sigma and beta(10)

Answer 8

sigma 1 > sigma 2 >…> sigma 10
beta(10) = 0

This allows us to interpret column parameters as % paid to date

Question 9

Limits of Skew normal

Answer 9

delta = 1, truncated normal: mu + w*Z
delta = 0, normal distribution: mu + w*E

Question 10

Skew normal

Answer 10

Mu + (wZ)S + (wE)sqrt(1-S^2)

Question 11

Why did Meyers not select skew normal

Answer 11

maximum skewness is 0.995

Question 12

CIT model

Answer 12

u = a + b + T(w + d -1)
T = CY trend

Question 13

Why did Meyers select lognormal

Answer 13

Right skewness
Can simulate negative losses
More skew than skew normal

Question 14

CIT, LIT standard deviation restriction

Answer 14

sigma 1

Question 15

CSR model

Answer 15

u = a + b + (1 - Y)^(w-1)
Y = settlement rate change (if positive, means speedup)