Friday, January 13, 2012

Are We Japanese?

Most of the discussion trying to determine if the U.S. is Japan 20 years later focuses on the economy and the stock market.  However, one of the biggest and most persistent correlations between Japan and the U.S. are the Japanese Yen and the U.S. 10 Year Treasury Yield.  I think it is essential to really try to explore this relationship to help determine “How Japanese are we?”  With the US $ in decline, it is hard to imagine the continued persistence of this dynamic. (Thanks reader teramonagi for catching an error.)

From TimelyPortfolio

For additional thoughts on Japan, please see my post Japan Intentional or Accidental Pursuit of Deflation.

R code in GIST:

require(quantmod)
#get Japanese Yen daily from Fred http://research.stlouisfed.org/fred2
getSymbols("DEXJPUS",src="FRED")
#get US 10y Yield from Fred
getSymbols("DGS10", src="FRED")
Yen10y <- na.omit(merge(DEXJPUS,DGS10))
#define colors
#use derivative of indianred4 with alpha for Yen
rgbnum <- col2rgb("indianred4")
col.yen <- rgb(rgbnum[1],rgbnum[2],rgbnum[3],alpha=180,maxColorValue=255)
#use derivative of steelblue4 with alpha for US 10y
rgbnum <- col2rgb("steelblue4")
col.10y <- rgb(rgbnum[1],rgbnum[2],rgbnum[3],alpha=180,maxColorValue=255)
#2 rows and 1 column of graphs
par(mfrow=c(2,1))
par(oma=c(0,1,0,0))
opar <- par(mai = c(0, 0.8, 0.5, 0.8))
#plot the Japanese Yen
plot.zoo(Yen10y[,1], type="l",
xaxt="n", xlab=NA, ylab="Yen (US$/JPY)",
col=col.yen, col.lab=col.yen, col.axis=col.yen,
lwd=3,bty="n",fg = "gray70")
#do grid for y
for (i in seq(par("yaxp")[1],par("yaxp")[2],by=(par("yaxp")[2]-par("yaxp")[1])/par("yaxp")[3])) {
abline(h=i, col="gray70")
}
title(main="Japanese Yen and US 10y Yield %",adj=0,outer=TRUE,line=-2)
par(new=TRUE)
plot.zoo(Yen10y[,2], type="l",
xaxt="n", yaxt="n", xlab=NA, ylab=NA,
col=col.10y, col.lab=col.10y,
lwd=3,bty="n",fg = "gray70")
axis(side=4,col.axis=col.10y,fg="gray70")
usr <- par("usr")
text(usr[2] + .12 * diff(usr[1:2]), mean(usr[3:4]), "US 10y Yield %",
srt = 90, xpd = TRUE, col = "steelblue4")
par(opar)
opar <- par(mai = c(0.8,0.8,0.2,0.8))
#plot running correlation between yen and US 10y
plot.zoo(runCor(Yen10y[,1],Yen10y[,2],n=500),
xlab=NA, ylab = NA,
lwd=3,bty="n",
col.axis="gray30",col.lab="gray30",col="gray30",fg="gray30")
title(main="Rolling 2 Year Correlation", cex.main=0.9, adj=0, col.main="gray30")
for (i in seq(-1,1,by=1)) {
abline(h=i, col="gray70")
}
axis(side=4,labels=FALSE,fg="gray70")
par(opar)
view raw yen and us10y.r hosted with ❤ by GitHub

5 comments:

  1. Thank you for your good article.

    But, I think that you should use "logged" or "difference" data instead of raw data because raw data may cause "spurious regression".

    If you want to know about it more, you should check this link http://en.wikipedia.org/wiki/Unit_root

    ReplyDelete
  2. You can tell that I wrote this late on Friday afternoon. Thanks for catching my stupid error. Will fix shortly.

    ReplyDelete
  3. I always enjoy your article and your own insight to analyze financial market :)

    I'm looking forward to reading your new article!

    ReplyDelete
  4. Nice article ! about connexions between Japan and the rest of the world (say US market) check that post written a few months ago
    http://freakonometrics.blog.free.fr/index.php?post/2010/01/06/Nikkei-s-past-experience-vs.-SP500-%28in-euros%29

    ReplyDelete
  5. great article; thanks for sharing with me

    ReplyDelete