From CS294-10 Visualization Fa08
[[Assignment 2 John De Felice]]September 24, 2008
Since I have been following the current financial turmoil. I thought I would investigate some relevant financial data. Interest rates are one of the most concise indicators of the state of the economy. I decided to investigate the relation between short-term and long-term rates.
I found data on Treasury Securities at: http://www.federalreserve.gov/datadownload/Choose.aspx?rel=H.15
I downloaded an Comma Separated Value File which I converted into an Excel workbook. I truncated parts of the file where there was missing data since this seems to confuse Tableau.
Initially I plotted the daily data for the various term securities.
There is a general tendency for short-term rates to be lower. But there are some exceptions. In order to eliminate some of fluctuation over time I next sormalized against the 20-year security. Some of the data is plotted with a baseline representing the normalized 20 year security.
This clearly shows the period when the 1-month note yields more than the longer term instruments. There is less fluctuation as the term grows longer due to the normalization.
Next I looked at possible seasonal variations.
There is some seasonal variation, including a crossing in November. Is this some sort of Christmas shopping effect?
But these monthly variations are slight and much less than one standard deviation as demonstrated by the following graph show plus and minus one standard deviation bounds on the 1-month note.
I tried to show variation relative to January 1 of each year, but this seems to be beyond the capability of Tableau.
Next I used scatter plots to show the relative correlation between different terms.
By separating into different years, distinct patterns are seen.
Finally, normalized data were plotted, to better see patterns.
Various Treasury debt security yield normalize to 20-year yield correlated with 1-month notes for August 2001 through September 2007 by year. These plots shows the varying degree of correlation for the different notes. The shorter term notes are better correlated with the 1-month notes. This is expected. What is more striking is the yearly variation. The years up to 2006 show generally tight clustering of data, but with some exceptions like the 7-year notes in 2004. In 2007 and 2008 wide scattering is apparent, with correlation seeming to disappear in some 2008 cases. 2007 shows a bifurcation in the scatter plots. These phenomena are related to the inversion of the typical relation between short-term and long term notes, with short-term yielding higher interest rates during several months in this period.