Build Heterogeneous Agent Model (HAM) in R

rm(list=ls());gc()

library(data.table)

library(ggplot2)

library(doSNOW)

library(foreach)

library(tidyverse)

library(tibbletime)

library(rlang)

#-----Settings---------#

Ft=0

a=1

b=0.05

c=0.02

d=0.95

eta=0.1

lambda=0.45

v_alpha=0.0025

v_beta=0.025

v_gamma=0.0025

N0=100

K0=50

#------read table----------#

Data<-fread("price.txt",data.table = F)

#-------------------------#

colnames(Data)<-c("Code","Name","Date","Turnover","Price")

TW_names<-unique(Data$Name)

Data<-Data %>%

mutate(log_P=log(Price)) %>%

mutate(Demand_F=c*(Ft-log_P)+rnorm(n=nrow(Data),sd=v_gamma)) %>%

mutate(Demand_C=b*(log_P-lag(log_P))+rnorm(n=nrow(Data),sd=v_beta)) %>%

mutate(Attr_F=0,Attr_C=0)

#------building model-----#

Data_append<- foreach(ix = 1:length(TW_names),.combine="rbind")%do%{

test<-Data %>% filter(Name==TW_names[ix])

for(iy in 4:nrow(test)){

test$Attr_C[iy]=(exp(test$log_P[iy])-exp(test$log_P[iy-1]))*test$Demand_C[iy-2]+d*test$Attr_C[iy-1]

test$Attr_F[iy]=(exp(test$log_P[iy])-exp(test$log_P[iy-1]))*test$Demand_F[iy-2]+d*test$Attr_F[iy-1]

}

return(test)

}

Data_append<-Data_append %>% mutate(FtoC=ifelse(Attr_C>Attr_F,0.5+lambda,0.5-lambda),CtoF=ifelse(Attr_F>Attr_C,0.5+lambda,0.5-lambda))

chartist<- foreach(ix = 1:length(TW_names),.combine="rbind")%do%{

test<-Data_append %>% filter(Name==TW_names[ix])

test=test %>%

mutate(N=N0) %>%

mutate(K=K0) %>%

mutate(Deter=runif(nrow(test),min = 0,max=1))

for(iy in 2:nrow(test)){

test$K[iy]=ifelse(test$Deter[iy]<(N0-test$K[iy-1])/N0*(eta+test$FtoC[iy-1]*(test$K[iy-1])/(N0-1)),test$K[iy-1]+1,ifelse(test$Deter[iy]>(1-(test$K[iy-1]/N0*(eta+test$CtoF[iy-1]*(N0-test$K[iy-1])/(N0-1)))),test$K[iy-1]-1,test$K[iy-1]))

}

return(test)

}

C_W=chartist %>% mutate(chartist_weight=K/N0)