Modeling and forecasting age-specific mortality: Lee-Carter method vs. Functional time series
1. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Modeling and forecasting age-specific mortality:
Lee-Carter method vs. Functional time series
Han Lin Shang
Econometrics & Business Statistics
http://monashforecasting.com/index.php?title=User:Han
2. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Outline
1 Lee-Carter model
2 Nonparametric smoothing
3 Functional principal component analysis
4 Functional time series forecasting
3. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model
1 Lee and Carter (1992) proposed one-factor principal
component method to model and forecast demographic data,
such as age-specific mortality rates.
4. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model
1 Lee and Carter (1992) proposed one-factor principal
component method to model and forecast demographic data,
such as age-specific mortality rates.
2 The Lee-Carter model can be written as
ln mx,t = ax + bx × kt + ex,t , (1)
where
5. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model
1 Lee and Carter (1992) proposed one-factor principal
component method to model and forecast demographic data,
such as age-specific mortality rates.
2 The Lee-Carter model can be written as
ln mx,t = ax + bx × kt + ex,t , (1)
where
ln mx,t is the observed log mortality rate at age x in year t,
6. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model
1 Lee and Carter (1992) proposed one-factor principal
component method to model and forecast demographic data,
such as age-specific mortality rates.
2 The Lee-Carter model can be written as
ln mx,t = ax + bx × kt + ex,t , (1)
where
ln mx,t is the observed log mortality rate at age x in year t,
ax is the sample mean vector,
7. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model
1 Lee and Carter (1992) proposed one-factor principal
component method to model and forecast demographic data,
such as age-specific mortality rates.
2 The Lee-Carter model can be written as
ln mx,t = ax + bx × kt + ex,t , (1)
where
ln mx,t is the observed log mortality rate at age x in year t,
ax is the sample mean vector,
bx is the first set of sample principal component,
8. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model
1 Lee and Carter (1992) proposed one-factor principal
component method to model and forecast demographic data,
such as age-specific mortality rates.
2 The Lee-Carter model can be written as
ln mx,t = ax + bx × kt + ex,t , (1)
where
ln mx,t is the observed log mortality rate at age x in year t,
ax is the sample mean vector,
bx is the first set of sample principal component,
kt is the first set of sample principal component scores,
9. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model
1 Lee and Carter (1992) proposed one-factor principal
component method to model and forecast demographic data,
such as age-specific mortality rates.
2 The Lee-Carter model can be written as
ln mx,t = ax + bx × kt + ex,t , (1)
where
ln mx,t is the observed log mortality rate at age x in year t,
ax is the sample mean vector,
bx is the first set of sample principal component,
kt is the first set of sample principal component scores,
ex,t is the residual term.
10. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model forecasts
1 There are a number of ways to adjust kt , which led to
extensions and modification of original Lee-Carter method.
11. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model forecasts
1 There are a number of ways to adjust kt , which led to
extensions and modification of original Lee-Carter method.
2 Lee and Carter (1992) advocated to use a random walk with
drift model to forecast principal component scores, expressed
as
kt = kt−1 + d + et , (2)
where
12. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model forecasts
1 There are a number of ways to adjust kt , which led to
extensions and modification of original Lee-Carter method.
2 Lee and Carter (1992) advocated to use a random walk with
drift model to forecast principal component scores, expressed
as
kt = kt−1 + d + et , (2)
where
d is known as the drift parameter, measures the average
annual change in the series,
13. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model forecasts
1 There are a number of ways to adjust kt , which led to
extensions and modification of original Lee-Carter method.
2 Lee and Carter (1992) advocated to use a random walk with
drift model to forecast principal component scores, expressed
as
kt = kt−1 + d + et , (2)
where
d is known as the drift parameter, measures the average
annual change in the series,
et is an uncorrelated error.
14. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Lee-Carter model forecasts
1 There are a number of ways to adjust kt , which led to
extensions and modification of original Lee-Carter method.
2 Lee and Carter (1992) advocated to use a random walk with
drift model to forecast principal component scores, expressed
as
kt = kt−1 + d + et , (2)
where
d is known as the drift parameter, measures the average
annual change in the series,
et is an uncorrelated error.
3 From the forecast of principal component scores, the forecast
age-specific log mortality rates are obtained using the
estimated age effects ax and estimated first set of principal
component bx .
15. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Construction of functional data
1 Functional data are a collection of functions, represented in
the form of curves, images or shapes.
France: male log mortality rate (1899−2005)
0
−2
Log mortality rate
−4
−6
−8
−10
0 20 40 60 80 100
Age
16. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Construction of functional data
1 Functional data are a collection of functions, represented in
the form of curves, images or shapes.
2 Let’s consider annual French male log mortality rates from
1816 to 2006 for ages between 0 and 100.
France: male log mortality rate (1899−2005)
0
−2
Log mortality rate
−4
−6
−8
−10
0 20 40 60 80 100
Age
17. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Construction of functional data
1 Functional data are a collection of functions, represented in
the form of curves, images or shapes.
2 Let’s consider annual French male log mortality rates from
1816 to 2006 for ages between 0 and 100.
3 By interpolating 101 data points in one year, functional curves
can be constructed below.
France: male log mortality rate (1899−2005)
0
−2
Log mortality rate
−4
−6
−8
−10
0 20 40 60 80 100
Age
18. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Smoothed functional data
1 Age-specific mortality rates are first smoothed using penalized
regression spline with monotonic constraint.
19. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Smoothed functional data
1 Age-specific mortality rates are first smoothed using penalized
regression spline with monotonic constraint.
2 Assuming there is an underlying continuous and smooth
function {ft (x); x ∈ [x1 , xp ]} that is observed with errors at
discrete ages in year t, we can express it as
mt (xi ) = ft (xi ) + σt (xi )εt,i , t = 1, 2, . . . , n, (3)
where
20. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Smoothed functional data
1 Age-specific mortality rates are first smoothed using penalized
regression spline with monotonic constraint.
2 Assuming there is an underlying continuous and smooth
function {ft (x); x ∈ [x1 , xp ]} that is observed with errors at
discrete ages in year t, we can express it as
mt (xi ) = ft (xi ) + σt (xi )εt,i , t = 1, 2, . . . , n, (3)
where
mt (xi ) is the log mortality rates,
21. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Smoothed functional data
1 Age-specific mortality rates are first smoothed using penalized
regression spline with monotonic constraint.
2 Assuming there is an underlying continuous and smooth
function {ft (x); x ∈ [x1 , xp ]} that is observed with errors at
discrete ages in year t, we can express it as
mt (xi ) = ft (xi ) + σt (xi )εt,i , t = 1, 2, . . . , n, (3)
where
mt (xi ) is the log mortality rates,
ft (xi ) is the smoothed log mortality rates,
22. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Smoothed functional data
1 Age-specific mortality rates are first smoothed using penalized
regression spline with monotonic constraint.
2 Assuming there is an underlying continuous and smooth
function {ft (x); x ∈ [x1 , xp ]} that is observed with errors at
discrete ages in year t, we can express it as
mt (xi ) = ft (xi ) + σt (xi )εt,i , t = 1, 2, . . . , n, (3)
where
mt (xi ) is the log mortality rates,
ft (xi ) is the smoothed log mortality rates,
σt (xi ) allows the possible presence of heteroscedastic error,
23. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Smoothed functional data
1 Age-specific mortality rates are first smoothed using penalized
regression spline with monotonic constraint.
2 Assuming there is an underlying continuous and smooth
function {ft (x); x ∈ [x1 , xp ]} that is observed with errors at
discrete ages in year t, we can express it as
mt (xi ) = ft (xi ) + σt (xi )εt,i , t = 1, 2, . . . , n, (3)
where
mt (xi ) is the log mortality rates,
ft (xi ) is the smoothed log mortality rates,
σt (xi ) allows the possible presence of heteroscedastic error,
εt,i is iid standard normal random variable.
24. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Smoothed functional data
1 Smoothness (also known filtering) allows us to analyse
derivative information of curves.
France: male log mortality rate (1899−2005)
0
−2
Log mortality rate
−4
−6
−8
−10
0 20 40 60 80 100
Age
25. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Smoothed functional data
1 Smoothness (also known filtering) allows us to analyse
derivative information of curves.
2 We transform n × p data matrix to n vector of functions.
France: male log mortality rate (1899−2005)
0
−2
Log mortality rate
−4
−6
−8
−10
0 20 40 60 80 100
Age
26. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Functional principal component analysis (FPCA)
1 FPCA can be viewed from both covariance kernel function
and linear operator perspectives.
27. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Functional principal component analysis (FPCA)
1 FPCA can be viewed from both covariance kernel function
and linear operator perspectives.
2 It is a dimension-reduction technique, with nice properties:
28. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Functional principal component analysis (FPCA)
1 FPCA can be viewed from both covariance kernel function
and linear operator perspectives.
2 It is a dimension-reduction technique, with nice properties:
FPCA minimizes the mean integrated squared error,
K 2
E f c (x) − βk φk (x) dx, K < ∞, (4)
I k=1
where f c (x) = f (x) − µ(x) represents the decentralized
functional curves, and x ∈ [x1 , xp ].
29. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Functional principal component analysis (FPCA)
1 FPCA can be viewed from both covariance kernel function
and linear operator perspectives.
2 It is a dimension-reduction technique, with nice properties:
FPCA minimizes the mean integrated squared error,
K 2
E f c (x) − βk φk (x) dx, K < ∞, (4)
I k=1
where f c (x) = f (x) − µ(x) represents the decentralized
functional curves, and x ∈ [x1 , xp ].
FPCA provides a way of extracting a large amount of variance,
∞ ∞ ∞
Var[f c (x)] = Var(βk )φ2 (x) =
k λk φ2 (x) =
k λk , (5)
k=1 k=1 k=1
where λ1 ≥ λ2 , . . . , ≥ 0 is a decreasing sequence of
eigenvalues and φk (x) is orthonormal.
30. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Functional principal component analysis (FPCA)
1 FPCA can be viewed from both covariance kernel function
and linear operator perspectives.
2 It is a dimension-reduction technique, with nice properties:
FPCA minimizes the mean integrated squared error,
K 2
E f c (x) − βk φk (x) dx, K < ∞, (4)
I k=1
where f c (x) = f (x) − µ(x) represents the decentralized
functional curves, and x ∈ [x1 , xp ].
FPCA provides a way of extracting a large amount of variance,
∞ ∞ ∞
Var[f c (x)] = Var(βk )φ2 (x) =
k λk φ2 (x) =
k λk , (5)
k=1 k=1 k=1
where λ1 ≥ λ2 , . . . , ≥ 0 is a decreasing sequence of
eigenvalues and φk (x) is orthonormal.
The principal component scores are uncorrelated, that is
cov(βi , βj ) = E(βi βj ) = 0, for i = j.
31. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Karhunen-Lo`ve (KL) expansion
e
By KL expansion, a stochastic process f (x), x ∈ [x1 , xp ] can be
expressed as
∞
f (x) = µ(x) + βk φk (x), (6)
k=1
K
= µ(x) + βk φk (x) + e(x), (7)
k=1
where
1 µ(x) is the population mean,
32. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Karhunen-Lo`ve (KL) expansion
e
By KL expansion, a stochastic process f (x), x ∈ [x1 , xp ] can be
expressed as
∞
f (x) = µ(x) + βk φk (x), (6)
k=1
K
= µ(x) + βk φk (x) + e(x), (7)
k=1
where
1 µ(x) is the population mean,
2 βk is the k th principal component scores,
33. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Karhunen-Lo`ve (KL) expansion
e
By KL expansion, a stochastic process f (x), x ∈ [x1 , xp ] can be
expressed as
∞
f (x) = µ(x) + βk φk (x), (6)
k=1
K
= µ(x) + βk φk (x) + e(x), (7)
k=1
where
1 µ(x) is the population mean,
2 βk is the k th principal component scores,
3 φk (x) is the k th functional principal components,
34. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Karhunen-Lo`ve (KL) expansion
e
By KL expansion, a stochastic process f (x), x ∈ [x1 , xp ] can be
expressed as
∞
f (x) = µ(x) + βk φk (x), (6)
k=1
K
= µ(x) + βk φk (x) + e(x), (7)
k=1
where
1 µ(x) is the population mean,
2 βk is the k th principal component scores,
3 φk (x) is the k th functional principal components,
4 e(x) is the error function, and
35. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Karhunen-Lo`ve (KL) expansion
e
By KL expansion, a stochastic process f (x), x ∈ [x1 , xp ] can be
expressed as
∞
f (x) = µ(x) + βk φk (x), (6)
k=1
K
= µ(x) + βk φk (x) + e(x), (7)
k=1
where
1 µ(x) is the population mean,
2 βk is the k th principal component scores,
3 φk (x) is the k th functional principal components,
4 e(x) is the error function, and
5 K is the number of retained principal components.
36. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Empirical FPCA
1 Because the stochastic process f is unknown in practice, the
population mean and eigenfunctions can only be approximated
through realizations of {f1 (x), f2 (x), . . . , fn (x)}.
37. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Empirical FPCA
1 Because the stochastic process f is unknown in practice, the
population mean and eigenfunctions can only be approximated
through realizations of {f1 (x), f2 (x), . . . , fn (x)}.
2 A function ft (x) can be approximated by
K
¯
ft (x) = f (x) + βt,k φk (x) + e(x), (8)
k=1
where
38. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Empirical FPCA
1 Because the stochastic process f is unknown in practice, the
population mean and eigenfunctions can only be approximated
through realizations of {f1 (x), f2 (x), . . . , fn (x)}.
2 A function ft (x) can be approximated by
K
¯
ft (x) = f (x) + βt,k φk (x) + e(x), (8)
k=1
where
¯ 1 n
f (x) = n t=1 ft (x) is the sample mean function,
39. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Empirical FPCA
1 Because the stochastic process f is unknown in practice, the
population mean and eigenfunctions can only be approximated
through realizations of {f1 (x), f2 (x), . . . , fn (x)}.
2 A function ft (x) can be approximated by
K
¯
ft (x) = f (x) + βt,k φk (x) + e(x), (8)
k=1
where
¯ 1 n
f (x) = n t=1 ft (x) is the sample mean function,
βk is the k th empirical principal component scores,
40. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Empirical FPCA
1 Because the stochastic process f is unknown in practice, the
population mean and eigenfunctions can only be approximated
through realizations of {f1 (x), f2 (x), . . . , fn (x)}.
2 A function ft (x) can be approximated by
K
¯
ft (x) = f (x) + βt,k φk (x) + e(x), (8)
k=1
where
¯ 1 n
f (x) = n t=1 ft (x) is the sample mean function,
βk is the k th empirical principal component scores,
φk (x) is the k th empirical functional principal components,
41. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Empirical FPCA
1 Because the stochastic process f is unknown in practice, the
population mean and eigenfunctions can only be approximated
through realizations of {f1 (x), f2 (x), . . . , fn (x)}.
2 A function ft (x) can be approximated by
K
¯
ft (x) = f (x) + βt,k φk (x) + e(x), (8)
k=1
where
¯ 1 n
f (x) = n t=1 ft (x) is the sample mean function,
βk is the k th empirical principal component scores,
φk (x) is the k th empirical functional principal components,
e(x) is the residual function.
42. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Decomposition
0.2
0.2
−1
0.2
0.20
0.1
0.1
−2
Basis function 1
Basis function 2
Basis function 3
Basis function 4
Mean function
0.15
0.0
0.1
−3
0.0
−0.1
0.10
−4
0.0
−0.1
−0.2
−5
0.05
−0.3
−0.1
−6
−0.2
0.00
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100
Age Age Age Age Age
8
10
0.5
6
1
5
Coefficient 1
Coefficient 2
Coefficient 3
Coefficient 4
0
4
0.0
0
−5
2
−0.5
−1
−10
0
−15
−1.0
−2
−2
1850 1900 1950 2000 1850 1900 1950 2000 1850 1900 1950 2000 1850 1900 1950 2000
Year Year Year Year
1 The principal components reveal underlying characteristics
across age direction.
43. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Decomposition
0.2
0.2
−1
0.2
0.20
0.1
0.1
−2
Basis function 1
Basis function 2
Basis function 3
Basis function 4
Mean function
0.15
0.0
0.1
−3
0.0
−0.1
0.10
−4
0.0
−0.1
−0.2
−5
0.05
−0.3
−0.1
−6
−0.2
0.00
0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100 0 20 40 60 80 100
Age Age Age Age Age
8
10
0.5
6
1
5
Coefficient 1
Coefficient 2
Coefficient 3
Coefficient 4
0
4
0.0
0
−5
2
−0.5
−1
−10
0
−15
−1.0
−2
−2
1850 1900 1950 2000 1850 1900 1950 2000 1850 1900 1950 2000 1850 1900 1950 2000
Year Year Year Year
1 The principal components reveal underlying characteristics
across age direction.
2 The principal component scores reveal possible outlying years
across time direction.
44. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Point forecast
Because orthogonality of the estimated functional principal
components and uncorrelated principal component scores, point
forecasts are obtained by
K
¯
fn+h|n (x) = E[fn+h (x)|I, Φ] = f (x) + βn+h|n,k φk (x), (9)
k=1
where
1 fn+h|n (x) is the h-step-ahead point forecast,
45. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Point forecast
Because orthogonality of the estimated functional principal
components and uncorrelated principal component scores, point
forecasts are obtained by
K
¯
fn+h|n (x) = E[fn+h (x)|I, Φ] = f (x) + βn+h|n,k φk (x), (9)
k=1
where
1 fn+h|n (x) is the h-step-ahead point forecast,
2 I represents the past data,
46. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Point forecast
Because orthogonality of the estimated functional principal
components and uncorrelated principal component scores, point
forecasts are obtained by
K
¯
fn+h|n (x) = E[fn+h (x)|I, Φ] = f (x) + βn+h|n,k φk (x), (9)
k=1
where
1 fn+h|n (x) is the h-step-ahead point forecast,
2 I represents the past data,
3 Φ = (φ1 (x), . . . , φK (x)) is a set of fixed estimated functional
principal components,
47. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Point forecast
Because orthogonality of the estimated functional principal
components and uncorrelated principal component scores, point
forecasts are obtained by
K
¯
fn+h|n (x) = E[fn+h (x)|I, Φ] = f (x) + βn+h|n,k φk (x), (9)
k=1
where
1 fn+h|n (x) is the h-step-ahead point forecast,
2 I represents the past data,
3 Φ = (φ1 (x), . . . , φK (x)) is a set of fixed estimated functional
principal components,
4 βn+h|n,k is the forecast of principal component scores by a
univariate time series method, such as exponential smoothing.
48. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Point forecast
Point forecasts (2007−2026)
0
−2
Log mortality rate
−4
−6
−8
Past data
−10
Forecasts
0 20 40 60 80 100
Age
Figure: 20-step-ahead point forecasts. Past data are shown in gray,
whereas the recent data are shown in color.
49. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Conclusion
1 We revisit the Lee-Carter model and functional time series
model for modeling age-specific mortality rates,
50. Lee-Carter model Nonparametric smoothing Functional principal component analysis Functional time series forecasting
Conclusion
1 We revisit the Lee-Carter model and functional time series
model for modeling age-specific mortality rates,
2 We show how to compute point forecasts for both models.