# Tim Miller (tmiller@demog.berkeley.edu)
# August 14, 2000
# 
# Econ Module
#   Simulates the annual consumption decision of households.
#   Generate consumption, earnings, income, and savings
#	by age of head, own age, and year (Out1).
#   Consumption decision is based on time series of 
#	interest rate (In1) and productivity growth rate (In2).  
#        Future needs are weighed against current needs based on expectations
#	about the future economy and the intertemporal elasticity of 
#	substitution (In3), and the rate of time preference (In4).
#       Future labor earnings are based on average earnings by age (In5) and
#       the productivity growth rate.
#       Expectiations about future mortality can be based either on current
#       survivorship rates (myopia) or future surivivorship rates (foresight)
#       (See In6).

# Attach Econ and Population Module
attach ("/home/davis/lee1/TIM.DATA/Splus.Data.Dirs/econ.module.usa",pos=1)
attach ("/home/davis/lee1/TIM.DATA/Splus.Data.Dirs/econ.module.functions",pos=2)
attach ("/home/davis/lee1/TIM.DATA/Splus.Data.Dirs/pop.module.usa",pos=3)


options(object.size=9e10) # large objects will be created

	
#Economic Parameters

# In1 and In2:	  Productivity growth rate and interest rate series.
#	1890 to 1997:	 Use U.S. historical rates
#	pre-1890:	 Use 30 year average (1890-1919)
#	post-1997:	Set prod. growth rate at 1.5% and
#			interest rate at 3.0%
#usa.int.gnp <- read.table("us.int.gnp.txt",header=T)

raw.prod.rate <- c(rep(.026,(1889-1800+1)),
	          usa.int.gnp$GNP,
	         rep(.015,(2300-1998+1)))
raw.interest.rate <- c(rep(.034,(1889-1800+1)),
	          usa.int.gnp$Interest,
	         rep(.030,(2300-1998+1)))
	
# Smooth the time-series using 31 year span
	# 15 years on either side of date.
prod.rate <- spec.smo(raw.prod.rate,span=31)
prod.rate5 <- spec.smo(raw.prod.rate,span=5) # using 5 year span
prod.rate11 <- spec.smo(raw.prod.rate,span=11) # using 11 year span

interest.rate <- spec.smo(raw.interest.rate,span=31)
interest.rate5 <- spec.smo(raw.interest.rate,span=5) # using 5 year span
interest.rate11 <- spec.smo(raw.interest.rate,span=11) # using 11 year span

# Ultimate values of productivity growth rate and interest rate
	# are used in forming expectations about these rates
ult.prod.rate <-  prod.rate[length(prod.rate)] 
ult.interest.rate <-  interest.rate[length(interest.rate)] 


#In3: Inverse of intertemporal elasticity of substitution
gamma <- 1/0.64 

#In4: Rate of time preference 
rho <- 0 

# In5: The age profile of average earnings by age in the US for 1997-99
us.earnings <- get("smooth.earnings.percap.1997.99","cps.fica.1999/Data")
producer.weights <- c(us.earnings,rep(0,20))

# Inflate earnings based on productivity growth before and after 1998
#  This means all results are based on 1998 constant dollars.
inflate.earnings <- cumprod(exp(c(0,prod.rate)))[1:run.length]
inflate.earnings <- inflate.earnings/inflate.earnings[(1998-1800+1)]

# Total earnings in the economy by age of household head
total.earnings.matrix <- matrix(0,max.age,run.length)
  for (time in 1:run.length){
    test <- household.array[,,time]*producer.weights*inflate.earnings[time]
    total.earnings.matrix[,time] <- apply(test,2,sum)
  }

# Earnings per household by age of household head
earnings.matrix <- total.earnings.matrix/head.matrix
earnings.matrix[is.na(earnings.matrix)] <- 0

# Total number of consumers distributed by age of head
consumption.weights <- c(rep(.3,5),rep(.4,5),rep(.6,5),rep(.7,4),rep(1,92))
total.consumers.matrix <- matrix(0,max.age,run.length)
    for (time in 1:run.length){
    test <- household.array[,,time]*consumption.weights
    total.consumers.matrix[,time] <- apply(test,2,sum)
  }

# Consumers per household by age of head and time
consumers.matrix <- total.consumers.matrix/head.matrix

# In6: Do households use current survivorship rates (mypoia) or
#  future rates in assessing the future?
myopia.option <- c("no")

# Wealth and consumption per household by age of head and year.
wealth.matrix <- matrix(0,max.age,(run.length-max.age))
consumption.matrix <- matrix(0,max.age,(run.length-max.age))
sim.dates <- seq(1800,,,(run.length-max.age)) #simulation dates

#Initialize wealth and consumption by repeating simulation for 50 years
for (cnt in 1:50){
  consumption.matrix[,1] <- set.consumption(1,myopia.option)
  wealth.matrix[22:111,1] <- set.wealth(2)[22:111]*
             (head.matrix[22:111,1]/head.matrix[22:111,2]) *
             exp(-interest.rate[1])}

for (time in 2:111){
    wealth.matrix[,time] <- set.wealth(time)
    consumption.matrix[,time] <- set.consumption(time,myopia.option)}

unix.time({for (time in 112:200){
    wealth.matrix[,time] <- set.wealth(time)
    consumption.matrix[,time] <- set.consumption(time,myopia.option)}})

unix.time({for (time in 201:300){
    wealth.matrix[,time] <- set.wealth(time)
    consumption.matrix[,time] <- set.consumption(time,myopia.option)}})

unix.time({for (time in 301:390){
    wealth.matrix[,time] <- set.wealth(time)
    consumption.matrix[,time] <- set.consumption(time,myopia.option)}})

# Calculate aggregate wealth (Wt), consumption (Ct),
# labor income (Ylt), interest income (Yit),
# total income (Yt), and savings rate (St).
Wt <- apply(head.matrix[,1:390]*wealth.matrix,2,sum)
Ct <- apply(head.matrix[,1:390]*consumption.matrix,2,sum)
Ylt <- apply(head.matrix[,1:390]*earnings.matrix[,1:390],2,sum)
Yit <- Wt*interest.rate[1:390]
Yt <- Ylt + Yit
St <- (Yt-Ct)/Yt

#Out1: Save results for each run.
#save("life cycle savings with mortality myopia")
save("life cycle savings without mortality myopia")