#Global Warming

#There is an upward trend since about 1970.  Is this man-made or natural cycle??

 globtemp = ts(scan("I:/AAEC 6311/lectures/R_demos/globtemp.txt"), start=1856)  

 gtemp = window(globtemp, start=1900)     # use data from 1990                    

 plot(gtemp, type="o", ylab="Global Temperature Deviations")





#Download S&P data

sp500 <- read.table("I:/AAEC 6311/lectures/R_demos/sp500.csv", header=TRUE, 

  sep=",", na.strings="NA", dec=".", strip.white=TRUE)



#Use settlement logged difference of settlement prices

returns <- ts(diff(log(sp500[,5])), start=c(1950,2), frequency=12)

adf.test(returns) #Augmented Dickey Fuller Test



plot(returns)           #Plot of returns

acf(returns,400)        #ACT of returns

pacf(returns,400)       #PACF of returns



ar1 <- arima(returns,order=c(1,0,0))  #Estimate an AR(1) model

tsdiag(ar1)  #plots diagnostic tests for estimated model



ma1 <- arima(returns,order=c(0,0,1)) #Estimate an MA(1) model

arma11 <- arima(returns,order=c(1,0,1)) #Estimate an ARMA(1,1) model



#Compare models

aic <- rbind(ar1$aic, ma1$aic, arma11$aic)

model <- c("AR1", "MA1", "ARMA1,1")

output.r <- matrix(data=round(aic,3), ncol=1,nrow=3, dimnames=list(model))

output.r



garch11 <- garch(returns,order=c(1,1))  #Estimate with GARCH(1,1)

AIC <- 2*garch11$n.likeli - 2*(sum(garch11$order)+1)  #Compute AIC for GARCH

SIC <- 2*garch11$n.likeli - log(length(returns))*(sum(garch11$order)+1)  #Compuate SIC for GARCH



plot(garch11$fitted.values[,1])  #Plot conditional variance

title("Conditional Variance for GARCH(1,1)")  #Insert Title into plot







___________________________________________________________________________________________________________________________________

##Evaluating Corn and Live Cattle Prices

data <- read.table("I:/AAEC 6311/lectures/R_demos/corn_cattle.csv", header=TRUE, sep=",", na.strings="NA", dec=".", strip.white=TRUE)

corn <- ts(data[,2]); lc <- ts(data[,3]);



#Plot Corn and Cattle prices stacked

par(mfrow=c(2,1))  #Defines the plotting environment to stack 2 plots

plot(corn); title("Plot of corn prices 1980 - 2005")

plot(lc); title("Plot of live cattle prices 1980 - 2005")



adf.test(corn); 

adf.test(lc)



y <- ts.union(corn, lc)  #Combines corn and live cattle prices into one mv variable

dly <- diff(log(y))

adf.test(dly[,1]) #ADF test on corn



library(vars)

VARselect(dly,lag.max=10, type="none") #Provides normalized IC measures for VAR(1)-VAR(8)



var_8 <- VAR(dly, p=8, type="none") #Estimate VAR(8)

var_8

summary(var_8) #View Results

causality(var_8, cause="corn") #Test H0: Corn Prices Granger Causes Cattle Prices

causality(var_8, cause="lc")   #Test H0: Cattle Prices GC Corn Prices



var_8_pred <- predict(var_8, n.ahead=30, ci=.95, dumvar=NULL) #Forecast 30 periods ahead

pred <- var_8_pred$fcst$corn[,1]



var_8_irf <- irf(var_8, n.ahead=100) #Impulse Response Functions

plot(var_8_irf)