chances of living from 33 to 99

thirtythree.R

# see install instructions here:
# https://github.com/UCBdemography/DemogBerkeley
library(DemogBerkeley)
us   <- userInput()
pass <- userInput()

mltper <- readHMDweb("USA","mltper_1x1",username = us, password = pass)

lx <- mltper$lx[mltper$Year == 2010]
lx <- lx / 1e5

# lx is at exact cutpoints, like today for me!
lx[100] / lx[34] # my chances if mortality is held constant at US 2010 levels, and if I'm totally average

# mx refers to an interval of time- 1 year
mx     <- mltper$mx[mltper$Year == 2010]

# 1% improvement
# we need mx's up through age 98 to get me to age 99:
mx1pct <- mx[34:99] * .99 ^ c(1:66)
exp(-cumsum(mx1pct)) # nifty direct path from rates to survival

# 2% improvement
mx2pct <- mx[34:99] * .98 ^ c(1:66)
exp(-cumsum(mx2pct))

# compare with Japan
mltperJ <- readHMDweb("JPN","mltper_1x1",username = us, password = pass)
# life expectancy at birth, 2012:
e02012  <- mltperJ$ex[mltperJ$Year == 2012][1]

# how long would it take US mortality to catch up with 2012 Japan at 1% improvement/year?
mx1pct2 <- outer(mx, .99 ^ c(1:50),"*")
plot(2011:2060,apply(mx1pct2,2,function(.mx){
            sum(exp(-cumsum(.mx)))
        }),type='l')
abline(h=e02012)

# 2% ?
mx2pct2 <- outer(mx, .98 ^ c(1:50),"*")
plot(2011:2060,apply(mx2pct2,2,function(.mx){
                    sum(exp(-cumsum(.mx)))
                }),type='l')
abline(h=e02012)

# end