R/rafi.r
In rexmacey/TaxAwareAA: Tax Aware Asset Allocation
Defines functions
rafi.cma
rafi.cma.v1
rafi.load.v1
rafi.read.csv.v1
acname_lookup
pretax_return
rafi.data.check
rafi.cma.v2
rafi.load.v2
rafi.read.xls.v2
Documented in
acname_lookup
pretax_return
rafi.cma
rafi.cma.v1
rafi.cma.v2
rafi.data.check
rafi.load.v1
rafi.load.v2
rafi.read.csv.v1
rafi.read.xls.v2
#' Load RAFI data into a cma
#'
#' This function reads a set of returns and correlations
#' downloaded from the RAFI website. Combined with assumptions about
#' the investor, a cma object is returned representing capital
#' market assumptions. RAFI files are in CSV format
#' through April 2017. From May 2017 they use a XLSX format.
#'
#' @param version Version (either v1 or v2). Use v1 if RAFI files are in CSV format
#' through April 2017. From May 2017 they use a XLSX format.
#' @param rafi.data.loc Folder in which the RAFI input files are located
#' @param acnametable Name of csv or xlsx file containing asset class details
#' @param xls.file.name Name of XLSX RAFI files
#' @param file.ret CSV file name with return assumptions
#' @param file.corr CSV file name with correlation assumptions
#' @param inflation.rate Default calls the function
#' \code{\link{Get10YrBEInflationRate}}
#' @keywords asset allocation efficient frontier
#' @export
#' @return A cma object which is a list containing the as_of_date of the
#' assumptions, the nclasses (number of classes), ac.data which is a table in
#' which each row represents and asset class and columns contain return, risk
#' and tax information. This table also includes box and custom constraints
#' There is also corr the correlation matrix, and/or cov for the covariance
#' matrix and nconstraints for the number of constraints.
#'
rafi.cma <- function(version="v2",rafi.data.loc,acnametable="acname_table.xlsx",
xls.file.name="Asset-Allocation-Interactive-Data.xlsx",
file.ret="core_asset_class_expected_returns.csv",
file.corr="core_asset_class_correlations_forecasted.csv",
inflation.rate=NULL){
inflation.rate <- ifelse(is.null(inflation.rate), Get10YrBEInflationRate()$rate, inflation.rate)
return(switch(toupper(version),
V1 = rafi.cma.v1(rafi.data.loc,acnametable,file.ret,file.corr,inflation.rate),
V2 = rafi.cma.v2(rafi.data.loc,acnametable,xls.file.name),
stop("Invalid version in call to rafi.cma function. Should be v1 or v2")))
}
#' Load RAFI data into a cma
#'
#' This function reads a set of returns and correlations from CSV files
#' downloaded from the RAFI website. Since RAFI may change its file formats,
#' this function may not work as expected in the future. This data takes the raw
#' RAFI estimates wihch are real geometric.
#'
#' @param rafi.data.loc Folder in which the CSV files are located
#' @param acnametable Name of csv file containing asset class details
#' @param file.ret CSV file name with return assumptions
#' @param file.corr CSV file name with correlation assumptions
#' @param inflation rate Default calls the function
#' \code{\link{Get10YrBEInflationRate}}
#' @keywords asset allocation efficient frontier
#' @return A cma object which is a list containing the as_of_date of the
#' assumptions, the nclasses (number of classes), ac.data which is a table in
#' which each row represents and asset class and columns contain return, risk
#' and tax information. This table also includes box and custom constraints
#' There is also corr the correlation matrix, and/or cov for the covariance
#' matrix and nconstraints for the number of constraints.
rafi.cma.v1<-function(rafi.data.loc,acnametable="acname_table.csv",file.ret="core_asset_class_expected_returns.csv",
file.corr="core_asset_class_correlations_forecasted.csv",
inflation.rate=Get10YrBEInflationRate()$rate){
ac_names<-read.csv(file=paste0(rafi.data.loc,acnametable),stringsAsFactors = FALSE)
first.constraint.col<-match("Max",colnames(ac_names))+1 # Constraints begin after Max Col
row.names(ac_names)<-ac_names$rt_class_names
rafi.data<-rafi.load.v1(rafi.data.loc,acnametable,file.ret,file.corr)
ac_names<-ac_names[rafi.data$ret$Asset.class,] #re order to match ret which should match corr
geom.ret<-rafi.data$ret$Expected.Return../100 +inflation.rate - ac_names$Expense
arith.ret<-geom.ret + (rafi.data$ret$Volatility../100)^2/2
yield<-rafi.data$ret$Yield../100
idx<-yield<=0
yield[idx]<-yield[idx]+inflation.rate
growth<-rafi.data$ret$Growth../100
growth[!idx]<-rafi.data$ret$Growth..[!idx]/100+inflation.rate
val.change<-geom.ret-yield-growth
ac.data<-data.frame(ret=arith.ret,geom.ret=geom.ret,yld=yield,growth=growth,
valChg=val.change,risk=rafi.data$ret$Volatility../100)
ac.data<-cbind(ac.data,ac_names[rafi.data$ret$Asset.class,c("IntOrd","IntTE","DivQual","DivOrd","Turnover","LTCG","STCG","ForeignTaxWithheld","Expense","Min","Max")])
nconstraints<-ncol(ac_names)-first.constraint.col+1
if (nconstraints>0) {
ac.data<-cbind(ac.data,ac_names[rafi.data$ret$Asset.class,first.constraint.col:ncol(ac_names)])
}
out<-list()
out$as_of_date<-rafi.data$as_of_date
out$classes<-row.names(ac_names)
out$nclasses<-rafi.data$nclasses
out$ac.data<-ac.data
out$ret <- ac.data$ret
out$corr<-as.matrix(rafi.data$corr)
out$cov<- as.matrix((ac.data$risk %*% t(ac.data$risk)) * out$corr)
if(!matrixcalc::is.positive.semi.definite(out$cov)) {
out$cov <- Matrix::nearPD(out$cov, corr=FALSE, keepDiag=TRUE, maxit=1000)$mat
}
out$cov<-as.matrix(out$cov)
#out$boxMin<-cma$ac.data[,"Min"]
#names(out$boxMin)<-cma$classes
#out$boxMax<-cma$ac.data[,"Max"]
#names(out$boxMax)<-cma$classes
out$nconstraints<-nconstraints
out$inflation<-inflation.rate
class(out)<-"cma"
return(out)
}
#' Load RAFI data
#'
#' This function reads a set of returns and correlations from CSV files
#' downloaded from the RAFI website. Since RAFI may change its file formats,
#' this function may not work as expected in the future.
#'
#' @param rafi.data.loc folder in which the CSV files are located
#' @param acnametable Name of csv file containing asset class details
#' @param file.ret CSV file name with return assumptions
#' @param file.corr CSV file name with correlation assumptions
#' @keywords asset allocation efficient frontier
#' @return A list with four items: ret is the return data, corr is the correlation matrix and
#' as_of_date is the date of the assumptionsfrom the csv file, nclasses is the number of classes
#' correlation data
#'
rafi.load.v1<-function(rafi.data.loc,acnametable="acname_table.csv",file.ret="core_asset_class_expected_returns.csv",
file.corr="core_asset_class_correlations_forecasted.csv"){
# The order and text of the asset class names of the returns and correlations may not match. We want the classes to be
# in the same order and have corresponding names.
# A csv file is used to lookup new names for the asset classes. It also contains other information
ac_names<-read.csv(file=paste0(rafi.data.loc,acnametable),stringsAsFactors = FALSE)
#first.constraint.col<-match("Max",colnames(ac_names))+1 # Constraints begin after Max Col
rafi<-rafi.read.csv.v1(rafi.data.loc,file.ret,file.corr)
rafi$ret[rafi$ret=="-"]<-0 # converts hyphens to zeros
idx<-is.na(rafi$ret[,"Expected.Return.."])
rafi$ret<-rafi$ret[!idx,] #remove extraneous rows
rafi$ret$Asset.class<-unlist(sapply(rafi$ret$Asset.class,acname_lookup,type="ret",ac_names=ac_names))
row.names(rafi$ret)<-rafi$ret$Asset.class
rafi$as_of_date<-rafi$ret$As.of.Date[1]
rafi$nclasses<-nrow(rafi$ret)
rafi$corr<-rafi$corr[1:rafi$nclasses,4:ncol(rafi$corr)]
rownames(rafi$corr)<-sapply(colnames(rafi$corr),acname_lookup,type="corr",ac_names=ac_names)
colnames(rafi$corr)<-rownames(rafi$corr)
rafi$ret<-rafi$ret[rownames(rafi$corr),] # put in same order as correlations
for (i in 3:ncol(rafi$ret)){
rafi$ret[,i]<-as.numeric(rafi$ret[,i])
}
return(rafi)
}
#' Read RAFI csv files
#'
#' This function reads a set of returns and correlations from CSV files
#' downloaded from the RAFI website. No processing is done. This function
#' allows one to see the data in an unprocessed form.
#'
#' @param rafi.data.loc folder in which the CSV files are located
#' @param file.ret CSV file name with return assumptions
#' @param file.corr CSV file name with correlation assumptions
#' @keywords asset allocation efficient frontier
#' @export
#' @return value A list with two items: ret is the return data, corr is the
#' correlation data
#'
rafi.read.csv.v1<-function(rafi.data.loc,file.ret="core_asset_class_expected_returns.csv",
file.corr="core_asset_class_correlations_forecasted.csv"){
out<-list()
out$ret<-read.csv(paste0(rafi.data.loc,file.ret),stringsAsFactors = FALSE)
out$corr<-read.csv(paste0(rafi.data.loc,file.corr),stringsAsFactors = FALSE)
return(out)
}
#' Look up asset class name in a table
#'
#' RAFI does not provide consistent names for their asset classes. That is the names in the
#' return CSV file do not match the names in the correlation CSV. Plus we may not prefer their names.
#' This function is called to look up a RAFI name and return a preferred name.
#'
#' @param ac is the RAFI name of the asset class
#' @param type is ret if the name comes from the return CSV else we assume it comes from the corr CSV
#' @param ac_names is the table. This is loaded in \code{\link{rafi.load}}
#' @return A character string with the name of the asset class
#'
acname_lookup<-function(ac,type,ac_names){
#type is ret or corr for rafi_corr_class_name or rafi_ret_class_name
if (type=="ret"){
i<-which(ac_names[,1]==ac)
} else {
i<-which(ac_names[,2]==ac)
}
return(ac_names[i,3])
}
#' Calculates the pre-tax return
#'
#' Simulates the investment value without taxes
#'
#' @param yld is initial yield in decimal (e.g. 0.03 for 3 percent)
#' @param growth is the annual growth rate of the income in decimal.
#' @param valchange is the annual percentage change in the valuation in decimal
#' @param horizon number of years to simulate
#' @return Pretax annual return in decimal
#'
pretax_return<-function(yld, growth, valchange, horizon=10){
price<-100
div0<-yld*price
div<-div0
shares<-1
for (i in 1:horizon){
div<-div*(1+growth)
if (div==0 | div0==0){
price<-(valchange+1)^i*100
} else {
price<-(valchange+1)^i*100/div0*div
}
income<-div*shares
shares<-shares+income/price
}
return((shares*price/100)^(1/horizon)-1)
}
#' Check RAFI data
#'
#' This function checks the 2 CSV files from RAFI.
#' It checks that the names of the assets classes are defined in the acnametable
#'
#' @param rafi.data.loc folder in which the CSV files are located
#' @param acnametable Name of csv file containing asset class details
#' @keywords asset allocation efficient frontier
#' @export
#' @return A list with errors or and empty list is no errors
#'
rafi.data.check<-function(rafi.data.loc,acnametable){
ac_names<-read.csv(file=paste0(rafi.data.loc,acnametable),stringsAsFactors = FALSE)
rafi<-rafi.read.csv.v1(rafi.data.loc)
n_acnames<-nrow(ac_names)
n_acret<-nrow(rafi$ret)
n_accorr<-nrow(rafi$corr)
err_list<-list()
n_err<-0
if(n_acnames!=n_acret){
n_err<-n_err+1
err_list[[n_err]]<-"# Assets in return file not equal to # in acname_table"
}
if(n_acnames!=n_accorr){
n_err<-n_err+1
err_list[[n_err]]<-"# Assets in correlation file not equal to # in acname_table"
}
for(i in 1:n_acret){
if(length(which(ac_names[,1]==rafi$ret$Asset.class[i]))==0){
n_err<-n_err+1
err_list[[n_err]]<-paste("Unrecognized asset class in RAFI return file",rafi$ret$Asset.class[i])
}
}
for(i in 1:n_accorr){
if(length(which(ac_names[,1]==rafi$corr$Index[i]))==0){
n_err<-n_err+1
err_list[[n_err]]<-paste("Unrecognized asset class in RAFI corr file",rafi$corr$Index[i])
}
}
return(err_list)
}
#' Load RAFI data into a cma
#' Version 2 uses RAFI file format which was adopted around May 2017
#' This new format includes a nominal return from which the inflation rate
#' may be inferred.
#'
#' This function reads a set of returns and correlations from XLSX files
#' downloaded from the RAFI website. Since RAFI may change its file formats,
#' this function may not work as expected in the future.
#'
#' @param rafi.data.loc Folder in which the CSV files are located
#' @param acnametable Name of XLSX file containing asset class details
#' @param xls.file.name name of XLSX file with RAFI assumptions
#' @keywords asset allocation efficient frontier
#' @export
#' @return A cma object which is a list containing the as_of_date of the
#' assumptions, the nclasses (number of classes), ac.data which is a table in
#' which each row represents and asset class and columns contain return, risk
#' and tax information. This table also includes box and custom constraints
#' There is also corr the correlation matrix, and/or cov for the covariance
#' matrix and nconstraints for the number of constraints.
rafi.cma.v2<-function(rafi.data.loc,acnametable="acname_table.xlsx",
xls.file.name="Asset-Allocation-Interactive-Data.xlsx"){
rafi.data<-rafi.load.v2(rafi.data.loc,acnametable,xls.file.name)
inflation.rate<-rafi.data$ret[1,"Expected Return (Nominal)"]-rafi.data$ret[1,"Expected Return (Real)"]
ac_names<-data.frame(readxl::read_xlsx(paste0(rafi.data.loc,acnametable),sheet="acname_table"))
row.names(ac_names)<-ac_names$rt_class_names
first.constraint.col<-match("Max",colnames(ac_names))+1 # Constraints begin after Max Col
ac_names<-ac_names[row.names(rafi.data$ret),] #re order to match ret which should match corr
geom.ret<-rafi.data$ret$`Expected Return (Nominal)` - ac_names$Expense
arith.ret<-geom.ret + rafi.data$ret$Volatility^2/2
yield<-rafi.data$ret$`Average Net Yield`
idx<-yield<=0
yield[idx]<-yield[idx]+inflation.rate
growth<-rafi.data$ret$`Capital Growth`
growth[!idx]<-growth[!idx]+inflation.rate
val.change<-geom.ret-yield-growth
ac.data<-data.frame(ret=arith.ret,geom.ret=geom.ret,yld=yield,growth=growth,
valChg=val.change,risk=rafi.data$ret$Volatility)
ac.data<-cbind(ac.data,ac_names[row.names(rafi.data$ret),c("AssetClass","LongName","IntOrd","IntTE","DivQual","DivOrd","Turnover","LTCG","STCG","ForeignTaxWithheld","Expense","Min","Max")])
nconstraints<-ncol(ac_names)-first.constraint.col+1
if (nconstraints>0) {
ac.data<-cbind(ac.data,ac_names[row.names(rafi.data$ret),first.constraint.col:ncol(ac_names)])
}
idx <- ac_names$Max !=0
out<-list()
out$as_of_date<-rafi.data$as_of_date
out$classes<-row.names(ac_names[idx,])
out$nclasses<-sum(idx)
out$ac.data<-ac.data[idx,]
out$ret <- ac.data$ret[idx]
out$corr<-as.matrix(rafi.data$corr[idx,idx])
out$cov<- as.matrix(rafi.data$cov[idx,idx])
if(!matrixcalc::is.positive.semi.definite(out$cov)) {
out$cov <- Matrix::nearPD(out$cov, corr=FALSE, keepDiag=TRUE, maxit=1000)$mat
}
out$cov<-as.matrix(out$cov)
out$nconstraints<-nconstraints
out$inflation<-inflation.rate
class(out)<-"cma"
return(out)
}
#' Load RAFI data
#'
#' This function reads a set of returns and correlations from XLSX files
#' downloaded from the RAFI website. Since RAFI may change its file formats,
#' this function may not work as expected in the future.
#'
#' @param rafi.data.loc folder in which the CSV files are located
#' @param acnametable Name of xlsx file containing asset class details
#' @param xls.file.name Name of Excel file downloaded from RAFI
#' @keywords asset allocation efficient frontier
#' @return A list with four items: ret is the return data, corr is the correlation matrix and
#' as_of_date is the date of the assumptionsfrom the csv file, nclasses is the number of classes
#' correlation data
#'
rafi.load.v2<-function(rafi.data.loc,acnametable="acname_table.xlsx",xls.file.name="Asset-Allocation-Interactive-Data.xlsx"){
# The order and text of the asset class names of the returns and correlations may not match. We want the classes to be
# in the same order and have corresponding names.
# A csv file is used to lookup new names for the asset classes. It also contains other information
ac_names<-readxl::read_xlsx(paste0(rafi.data.loc,acnametable),sheet="acname_table")
#first.constraint.col<-match("Max",colnames(ac_names))+1 # Constraints begin after Max Col
rafi<-rafi.read.xls.v2(rafi.data.loc,xls.file.name)
row.names(rafi$ret)<-unlist(sapply(rafi$ret$`Asset Class`,acname_lookup,type="ret",ac_names=ac_names))
rafi$nclasses<-nrow(rafi$ret)
rownames(rafi$corr)<-sapply(colnames(rafi$corr),acname_lookup,type="corr",ac_names=ac_names)
colnames(rafi$corr)<-rownames(rafi$corr)
rownames(rafi$cov)<-rownames(rafi$corr)
colnames(rafi$cov)<-rownames(rafi$corr)
rafi$ret<-rafi$ret[rownames(rafi$corr),] # put in same order as correlations
return(rafi)
}
#' Read RAFI XLS file
#'
#' This function reads a set of returns and correlations from XLS file
#' downloaded from the RAFI website. No processing is done. This function
#' allows one to see the data in an unprocessed form.
#'
#' @param rafi.data.loc folder in which the files are located
#' @param xls.file.name name of XLSX file
#' @keywords asset allocation efficient frontier
#' @export
#' @return value A list with two items: ret is the return data, corr is the
#' correlation data
#'
rafi.read.xls.v2<-function(rafi.data.loc,xls.file.name="Asset-Allocation-Interactive-Data.xlsx"){
if (file.exists(paste0(rafi.data.loc,"xlranges.yaml"))){
xl<-yaml.load_file(paste0(rafi.data.loc,"xlranges.yaml"))
rng.return<-xl$rng.return
rng.corr<-xl$rng.corr
rng.cov<- xl$rng.cov
rng.date<-xl$rng.date
} else {
rng.return<-"B4:L32"
rng.corr<-"C4:AD32"
rng.cov<- "C35:AD63"
rng.date<-"C50"
}
out<-list()
out$as_of_date<-as.character(readxl::read_xlsx(paste0(rafi.data.loc,xls.file.name),
sheet="Expected.Returns",range=rng.date,col_names = FALSE))
out$as_of_date<-unlist(out$as_of_date)
out$ret<-as.data.frame(readxl::read_xlsx(paste0(rafi.data.loc,xls.file.name),sheet="Expected.Returns",range=rng.return))
rafi.corr<-readxl::read_xlsx(paste0(rafi.data.loc,xls.file.name),sheet="Expected.Risk.Matrix",range=rng.corr)
row.names(rafi.corr)<-names(rafi.corr)
rafi.cov<-readxl::read_xlsx(paste0(rafi.data.loc,xls.file.name),sheet="Expected.Risk.Matrix",range=rng.cov)
row.names(rafi.cov)<-names(rafi.cov)
out$corr<-rafi.corr
out$cov<-rafi.cov
return(out)
}
rexmacey/TaxAwareAA documentation built on Dec. 3, 2019, 7:54 a.m.
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Defines functions rafi.cma rafi.cma.v1 rafi.load.v1 rafi.read.csv.v1 acname_lookup pretax_return rafi.data.check rafi.cma.v2 rafi.load.v2 rafi.read.xls.v2
Documented in acname_lookup pretax_return rafi.cma rafi.cma.v1 rafi.cma.v2 rafi.data.check rafi.load.v1 rafi.load.v2 rafi.read.csv.v1 rafi.read.xls.v2
#' Load RAFI data into a cma
#'
#' This function reads a set of returns and correlations
#' downloaded from the RAFI website. Combined with assumptions about
#' the investor, a cma object is returned representing capital
#' market assumptions. RAFI files are in CSV format
#' through April 2017. From May 2017 they use a XLSX format.
#'
#' @param version Version (either v1 or v2). Use v1 if RAFI files are in CSV format
#' through April 2017. From May 2017 they use a XLSX format.
#' @param rafi.data.loc Folder in which the RAFI input files are located
#' @param acnametable Name of csv or xlsx file containing asset class details
#' @param xls.file.name Name of XLSX RAFI files
#' @param file.ret CSV file name with return assumptions
#' @param file.corr CSV file name with correlation assumptions
#' @param inflation.rate Default calls the function
#' \code{\link{Get10YrBEInflationRate}}
#' @keywords asset allocation efficient frontier
#' @export
#' @return A cma object which is a list containing the as_of_date of the
#' assumptions, the nclasses (number of classes), ac.data which is a table in
#' which each row represents and asset class and columns contain return, risk
#' and tax information. This table also includes box and custom constraints
#' There is also corr the correlation matrix, and/or cov for the covariance
#' matrix and nconstraints for the number of constraints.
#'
rafi.cma <- function(version="v2",rafi.data.loc,acnametable="acname_table.xlsx",
xls.file.name="Asset-Allocation-Interactive-Data.xlsx",
file.ret="core_asset_class_expected_returns.csv",
file.corr="core_asset_class_correlations_forecasted.csv",
inflation.rate=NULL){
inflation.rate <- ifelse(is.null(inflation.rate), Get10YrBEInflationRate()$rate, inflation.rate)
return(switch(toupper(version),
V1 = rafi.cma.v1(rafi.data.loc,acnametable,file.ret,file.corr,inflation.rate),
V2 = rafi.cma.v2(rafi.data.loc,acnametable,xls.file.name),
stop("Invalid version in call to rafi.cma function. Should be v1 or v2")))
}
#' Load RAFI data into a cma
#'
#' This function reads a set of returns and correlations from CSV files
#' downloaded from the RAFI website. Since RAFI may change its file formats,
#' this function may not work as expected in the future. This data takes the raw
#' RAFI estimates wihch are real geometric.
#'
#' @param rafi.data.loc Folder in which the CSV files are located
#' @param acnametable Name of csv file containing asset class details
#' @param file.ret CSV file name with return assumptions
#' @param file.corr CSV file name with correlation assumptions
#' @param inflation rate Default calls the function
#' \code{\link{Get10YrBEInflationRate}}
#' @keywords asset allocation efficient frontier
#' @return A cma object which is a list containing the as_of_date of the
#' assumptions, the nclasses (number of classes), ac.data which is a table in
#' which each row represents and asset class and columns contain return, risk
#' and tax information. This table also includes box and custom constraints
#' There is also corr the correlation matrix, and/or cov for the covariance
#' matrix and nconstraints for the number of constraints.
rafi.cma.v1<-function(rafi.data.loc,acnametable="acname_table.csv",file.ret="core_asset_class_expected_returns.csv",
file.corr="core_asset_class_correlations_forecasted.csv",
inflation.rate=Get10YrBEInflationRate()$rate){
ac_names<-read.csv(file=paste0(rafi.data.loc,acnametable),stringsAsFactors = FALSE)
first.constraint.col<-match("Max",colnames(ac_names))+1 # Constraints begin after Max Col
row.names(ac_names)<-ac_names$rt_class_names
rafi.data<-rafi.load.v1(rafi.data.loc,acnametable,file.ret,file.corr)
ac_names<-ac_names[rafi.data$ret$Asset.class,] #re order to match ret which should match corr
geom.ret<-rafi.data$ret$Expected.Return../100 +inflation.rate - ac_names$Expense
arith.ret<-geom.ret + (rafi.data$ret$Volatility../100)^2/2
yield<-rafi.data$ret$Yield../100
idx<-yield<=0
yield[idx]<-yield[idx]+inflation.rate
growth<-rafi.data$ret$Growth../100
growth[!idx]<-rafi.data$ret$Growth..[!idx]/100+inflation.rate
val.change<-geom.ret-yield-growth
ac.data<-data.frame(ret=arith.ret,geom.ret=geom.ret,yld=yield,growth=growth,
valChg=val.change,risk=rafi.data$ret$Volatility../100)
ac.data<-cbind(ac.data,ac_names[rafi.data$ret$Asset.class,c("IntOrd","IntTE","DivQual","DivOrd","Turnover","LTCG","STCG","ForeignTaxWithheld","Expense","Min","Max")])
nconstraints<-ncol(ac_names)-first.constraint.col+1
if (nconstraints>0) {
ac.data<-cbind(ac.data,ac_names[rafi.data$ret$Asset.class,first.constraint.col:ncol(ac_names)])
}
out<-list()
out$as_of_date<-rafi.data$as_of_date
out$classes<-row.names(ac_names)
out$nclasses<-rafi.data$nclasses
out$ac.data<-ac.data
out$ret <- ac.data$ret
out$corr<-as.matrix(rafi.data$corr)
out$cov<- as.matrix((ac.data$risk %*% t(ac.data$risk)) * out$corr)
if(!matrixcalc::is.positive.semi.definite(out$cov)) {
out$cov <- Matrix::nearPD(out$cov, corr=FALSE, keepDiag=TRUE, maxit=1000)$mat
}
out$cov<-as.matrix(out$cov)
#out$boxMin<-cma$ac.data[,"Min"]
#names(out$boxMin)<-cma$classes
#out$boxMax<-cma$ac.data[,"Max"]
#names(out$boxMax)<-cma$classes
out$nconstraints<-nconstraints
out$inflation<-inflation.rate
class(out)<-"cma"
return(out)
}
#' Load RAFI data
#'
#' This function reads a set of returns and correlations from CSV files
#' downloaded from the RAFI website. Since RAFI may change its file formats,
#' this function may not work as expected in the future.
#'
#' @param rafi.data.loc folder in which the CSV files are located
#' @param acnametable Name of csv file containing asset class details
#' @param file.ret CSV file name with return assumptions
#' @param file.corr CSV file name with correlation assumptions
#' @keywords asset allocation efficient frontier
#' @return A list with four items: ret is the return data, corr is the correlation matrix and
#' as_of_date is the date of the assumptionsfrom the csv file, nclasses is the number of classes
#' correlation data
#'
rafi.load.v1<-function(rafi.data.loc,acnametable="acname_table.csv",file.ret="core_asset_class_expected_returns.csv",
file.corr="core_asset_class_correlations_forecasted.csv"){
# The order and text of the asset class names of the returns and correlations may not match. We want the classes to be
# in the same order and have corresponding names.
# A csv file is used to lookup new names for the asset classes. It also contains other information
ac_names<-read.csv(file=paste0(rafi.data.loc,acnametable),stringsAsFactors = FALSE)
#first.constraint.col<-match("Max",colnames(ac_names))+1 # Constraints begin after Max Col
rafi<-rafi.read.csv.v1(rafi.data.loc,file.ret,file.corr)
rafi$ret[rafi$ret=="-"]<-0 # converts hyphens to zeros
idx<-is.na(rafi$ret[,"Expected.Return.."])
rafi$ret<-rafi$ret[!idx,] #remove extraneous rows
rafi$ret$Asset.class<-unlist(sapply(rafi$ret$Asset.class,acname_lookup,type="ret",ac_names=ac_names))
row.names(rafi$ret)<-rafi$ret$Asset.class
rafi$as_of_date<-rafi$ret$As.of.Date[1]
rafi$nclasses<-nrow(rafi$ret)
rafi$corr<-rafi$corr[1:rafi$nclasses,4:ncol(rafi$corr)]
rownames(rafi$corr)<-sapply(colnames(rafi$corr),acname_lookup,type="corr",ac_names=ac_names)
colnames(rafi$corr)<-rownames(rafi$corr)
rafi$ret<-rafi$ret[rownames(rafi$corr),] # put in same order as correlations
for (i in 3:ncol(rafi$ret)){
rafi$ret[,i]<-as.numeric(rafi$ret[,i])
}
return(rafi)
}
#' Read RAFI csv files
#'
#' This function reads a set of returns and correlations from CSV files
#' downloaded from the RAFI website. No processing is done. This function
#' allows one to see the data in an unprocessed form.
#'
#' @param rafi.data.loc folder in which the CSV files are located
#' @param file.ret CSV file name with return assumptions
#' @param file.corr CSV file name with correlation assumptions
#' @keywords asset allocation efficient frontier
#' @export
#' @return value A list with two items: ret is the return data, corr is the
#' correlation data
#'
rafi.read.csv.v1<-function(rafi.data.loc,file.ret="core_asset_class_expected_returns.csv",
file.corr="core_asset_class_correlations_forecasted.csv"){
out<-list()
out$ret<-read.csv(paste0(rafi.data.loc,file.ret),stringsAsFactors = FALSE)
out$corr<-read.csv(paste0(rafi.data.loc,file.corr),stringsAsFactors = FALSE)
return(out)
}
#' Look up asset class name in a table
#'
#' RAFI does not provide consistent names for their asset classes. That is the names in the
#' return CSV file do not match the names in the correlation CSV. Plus we may not prefer their names.
#' This function is called to look up a RAFI name and return a preferred name.
#'
#' @param ac is the RAFI name of the asset class
#' @param type is ret if the name comes from the return CSV else we assume it comes from the corr CSV
#' @param ac_names is the table. This is loaded in \code{\link{rafi.load}}
#' @return A character string with the name of the asset class
#'
acname_lookup<-function(ac,type,ac_names){
#type is ret or corr for rafi_corr_class_name or rafi_ret_class_name
if (type=="ret"){
i<-which(ac_names[,1]==ac)
} else {
i<-which(ac_names[,2]==ac)
}
return(ac_names[i,3])
}
#' Calculates the pre-tax return
#'
#' Simulates the investment value without taxes
#'
#' @param yld is initial yield in decimal (e.g. 0.03 for 3 percent)
#' @param growth is the annual growth rate of the income in decimal.
#' @param valchange is the annual percentage change in the valuation in decimal
#' @param horizon number of years to simulate
#' @return Pretax annual return in decimal
#'
pretax_return<-function(yld, growth, valchange, horizon=10){
price<-100
div0<-yld*price
div<-div0
shares<-1
for (i in 1:horizon){
div<-div*(1+growth)
if (div==0 | div0==0){
price<-(valchange+1)^i*100
} else {
price<-(valchange+1)^i*100/div0*div
}
income<-div*shares
shares<-shares+income/price
}
return((shares*price/100)^(1/horizon)-1)
}
#' Check RAFI data
#'
#' This function checks the 2 CSV files from RAFI.
#' It checks that the names of the assets classes are defined in the acnametable
#'
#' @param rafi.data.loc folder in which the CSV files are located
#' @param acnametable Name of csv file containing asset class details
#' @keywords asset allocation efficient frontier
#' @export
#' @return A list with errors or and empty list is no errors
#'
rafi.data.check<-function(rafi.data.loc,acnametable){
ac_names<-read.csv(file=paste0(rafi.data.loc,acnametable),stringsAsFactors = FALSE)
rafi<-rafi.read.csv.v1(rafi.data.loc)
n_acnames<-nrow(ac_names)
n_acret<-nrow(rafi$ret)
n_accorr<-nrow(rafi$corr)
err_list<-list()
n_err<-0
if(n_acnames!=n_acret){
n_err<-n_err+1
err_list[[n_err]]<-"# Assets in return file not equal to # in acname_table"
}
if(n_acnames!=n_accorr){
n_err<-n_err+1
err_list[[n_err]]<-"# Assets in correlation file not equal to # in acname_table"
}
for(i in 1:n_acret){
if(length(which(ac_names[,1]==rafi$ret$Asset.class[i]))==0){
n_err<-n_err+1
err_list[[n_err]]<-paste("Unrecognized asset class in RAFI return file",rafi$ret$Asset.class[i])
}
}
for(i in 1:n_accorr){
if(length(which(ac_names[,1]==rafi$corr$Index[i]))==0){
n_err<-n_err+1
err_list[[n_err]]<-paste("Unrecognized asset class in RAFI corr file",rafi$corr$Index[i])
}
}
return(err_list)
}
#' Load RAFI data into a cma
#' Version 2 uses RAFI file format which was adopted around May 2017
#' This new format includes a nominal return from which the inflation rate
#' may be inferred.
#'
#' This function reads a set of returns and correlations from XLSX files
#' downloaded from the RAFI website. Since RAFI may change its file formats,
#' this function may not work as expected in the future.
#'
#' @param rafi.data.loc Folder in which the CSV files are located
#' @param acnametable Name of XLSX file containing asset class details
#' @param xls.file.name name of XLSX file with RAFI assumptions
#' @keywords asset allocation efficient frontier
#' @export
#' @return A cma object which is a list containing the as_of_date of the
#' assumptions, the nclasses (number of classes), ac.data which is a table in
#' which each row represents and asset class and columns contain return, risk
#' and tax information. This table also includes box and custom constraints
#' There is also corr the correlation matrix, and/or cov for the covariance
#' matrix and nconstraints for the number of constraints.
rafi.cma.v2<-function(rafi.data.loc,acnametable="acname_table.xlsx",
xls.file.name="Asset-Allocation-Interactive-Data.xlsx"){
rafi.data<-rafi.load.v2(rafi.data.loc,acnametable,xls.file.name)
inflation.rate<-rafi.data$ret[1,"Expected Return (Nominal)"]-rafi.data$ret[1,"Expected Return (Real)"]
ac_names<-data.frame(readxl::read_xlsx(paste0(rafi.data.loc,acnametable),sheet="acname_table"))
row.names(ac_names)<-ac_names$rt_class_names
first.constraint.col<-match("Max",colnames(ac_names))+1 # Constraints begin after Max Col
ac_names<-ac_names[row.names(rafi.data$ret),] #re order to match ret which should match corr
geom.ret<-rafi.data$ret$`Expected Return (Nominal)` - ac_names$Expense
arith.ret<-geom.ret + rafi.data$ret$Volatility^2/2
yield<-rafi.data$ret$`Average Net Yield`
idx<-yield<=0
yield[idx]<-yield[idx]+inflation.rate
growth<-rafi.data$ret$`Capital Growth`
growth[!idx]<-growth[!idx]+inflation.rate
val.change<-geom.ret-yield-growth
ac.data<-data.frame(ret=arith.ret,geom.ret=geom.ret,yld=yield,growth=growth,
valChg=val.change,risk=rafi.data$ret$Volatility)
ac.data<-cbind(ac.data,ac_names[row.names(rafi.data$ret),c("AssetClass","LongName","IntOrd","IntTE","DivQual","DivOrd","Turnover","LTCG","STCG","ForeignTaxWithheld","Expense","Min","Max")])
nconstraints<-ncol(ac_names)-first.constraint.col+1
if (nconstraints>0) {
ac.data<-cbind(ac.data,ac_names[row.names(rafi.data$ret),first.constraint.col:ncol(ac_names)])
}
idx <- ac_names$Max !=0
out<-list()
out$as_of_date<-rafi.data$as_of_date
out$classes<-row.names(ac_names[idx,])
out$nclasses<-sum(idx)
out$ac.data<-ac.data[idx,]
out$ret <- ac.data$ret[idx]
out$corr<-as.matrix(rafi.data$corr[idx,idx])
out$cov<- as.matrix(rafi.data$cov[idx,idx])
if(!matrixcalc::is.positive.semi.definite(out$cov)) {
out$cov <- Matrix::nearPD(out$cov, corr=FALSE, keepDiag=TRUE, maxit=1000)$mat
}
out$cov<-as.matrix(out$cov)
out$nconstraints<-nconstraints
out$inflation<-inflation.rate
class(out)<-"cma"
return(out)
}
#' Load RAFI data
#'
#' This function reads a set of returns and correlations from XLSX files
#' downloaded from the RAFI website. Since RAFI may change its file formats,
#' this function may not work as expected in the future.
#'
#' @param rafi.data.loc folder in which the CSV files are located
#' @param acnametable Name of xlsx file containing asset class details
#' @param xls.file.name Name of Excel file downloaded from RAFI
#' @keywords asset allocation efficient frontier
#' @return A list with four items: ret is the return data, corr is the correlation matrix and
#' as_of_date is the date of the assumptionsfrom the csv file, nclasses is the number of classes
#' correlation data
#'
rafi.load.v2<-function(rafi.data.loc,acnametable="acname_table.xlsx",xls.file.name="Asset-Allocation-Interactive-Data.xlsx"){
# The order and text of the asset class names of the returns and correlations may not match. We want the classes to be
# in the same order and have corresponding names.
# A csv file is used to lookup new names for the asset classes. It also contains other information
ac_names<-readxl::read_xlsx(paste0(rafi.data.loc,acnametable),sheet="acname_table")
#first.constraint.col<-match("Max",colnames(ac_names))+1 # Constraints begin after Max Col
rafi<-rafi.read.xls.v2(rafi.data.loc,xls.file.name)
row.names(rafi$ret)<-unlist(sapply(rafi$ret$`Asset Class`,acname_lookup,type="ret",ac_names=ac_names))
rafi$nclasses<-nrow(rafi$ret)
rownames(rafi$corr)<-sapply(colnames(rafi$corr),acname_lookup,type="corr",ac_names=ac_names)
colnames(rafi$corr)<-rownames(rafi$corr)
rownames(rafi$cov)<-rownames(rafi$corr)
colnames(rafi$cov)<-rownames(rafi$corr)
rafi$ret<-rafi$ret[rownames(rafi$corr),] # put in same order as correlations
return(rafi)
}
#' Read RAFI XLS file
#'
#' This function reads a set of returns and correlations from XLS file
#' downloaded from the RAFI website. No processing is done. This function
#' allows one to see the data in an unprocessed form.
#'
#' @param rafi.data.loc folder in which the files are located
#' @param xls.file.name name of XLSX file
#' @keywords asset allocation efficient frontier
#' @export
#' @return value A list with two items: ret is the return data, corr is the
#' correlation data
#'
rafi.read.xls.v2<-function(rafi.data.loc,xls.file.name="Asset-Allocation-Interactive-Data.xlsx"){
if (file.exists(paste0(rafi.data.loc,"xlranges.yaml"))){
xl<-yaml.load_file(paste0(rafi.data.loc,"xlranges.yaml"))
rng.return<-xl$rng.return
rng.corr<-xl$rng.corr
rng.cov<- xl$rng.cov
rng.date<-xl$rng.date
} else {
rng.return<-"B4:L32"
rng.corr<-"C4:AD32"
rng.cov<- "C35:AD63"
rng.date<-"C50"
}
out<-list()
out$as_of_date<-as.character(readxl::read_xlsx(paste0(rafi.data.loc,xls.file.name),
sheet="Expected.Returns",range=rng.date,col_names = FALSE))
out$as_of_date<-unlist(out$as_of_date)
out$ret<-as.data.frame(readxl::read_xlsx(paste0(rafi.data.loc,xls.file.name),sheet="Expected.Returns",range=rng.return))
rafi.corr<-readxl::read_xlsx(paste0(rafi.data.loc,xls.file.name),sheet="Expected.Risk.Matrix",range=rng.corr)
row.names(rafi.corr)<-names(rafi.corr)
rafi.cov<-readxl::read_xlsx(paste0(rafi.data.loc,xls.file.name),sheet="Expected.Risk.Matrix",range=rng.cov)
row.names(rafi.cov)<-names(rafi.cov)
out$corr<-rafi.corr
out$cov<-rafi.cov
return(out)
}
R Package Documentation
Browse R Packages
We want your feedback!
Note that we can't provide technical support on individual packages. You should contact the package authors for that.
Embedding an R snippet on your website
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.