Re: st: Currency_Time-series

["Michael S. Hanson" <mshanson@mac.com>]

On Apr 3, 2007, at 12:51 PM, b.qureishi wrote:

> I will be conducting an empirical study to asses the affects of “The 
> release of US Macro data on daily currency prices”- and whether or not 
> such fundamental releases start certain trends. I have about 30 years 
> of data for cable, eur/usd and usd/jpy.-dependant variables. My 
> independent variables consist of GDP, PPI, CPI, INTEREST RATES and 
> UNEMPLOYMENT DATA.
> The frequency of my independent variables in monthly, excluding GDP 
> which is quarterly, over 30 years. The frequency of my dependant 
> variables is daily over 30 years.
>
> In light of the above, my data is no-doubt time-series. My question 
> is, how to I get my independent variables to reconcile with my 
> dependant variables? I.e how can I get for example quarterly gdp 
> figures to translate to a daily fx figure?- my lecturer who is not a 
> currency whiz suggests that I should convert my daily rates into 
> quarterly rates. I could do this, but believe that it will not 
> directly contribute to what I want to do.
>
> Your thoughts, suggestions and direction would be most appreciated.
	Your description of the proposed study is sufficiently ambiguous that 
I don't believe meaningful suggestions can be offered before you 
clarify your objectives.  I see two possibilities:

(1) You want to estimate a macroeconomic model of exchange rate 
determination.

(2) You want to investigate the efficiency of the foreign exchange 
market to macroeconomic "news."

	In the first case, the natural frequency of the analysis would be 
quarterly (if you include NIPA data) or monthly (otherwise).  The 
release of the data is not what's relevant in this case;  rather, the 
exchange rate is viewed as a function of the macroeconomic activity 
that occurred over the current quarter (month), and possibly preceding 
ones.  In that case, I would recommend the last observation of the 
quarter (month) for the exchange rate, although other researchers have 
used the average.  There are a variety of econometric approaches one 
can use for this investigation, but an important issue will be 
simultaneity / endogeneity of the macro variables to the value of the 
exchange rate.  Neely & Sarno (2002) might be a place to start 
exploring this question.

	In the second case, you likely want to know how the FX market responds 
to the release of macroeconomic "news" -- that is, the day (or even 
minute) that the information about the macroeconomic variable is 
released to the public.  In this case, you will need data on the exact 
day (and time) of the release of the macroeconomic information, which 
for NIPA data can be up to three quarters following the quarter in 
which the measured activity actually occurred, not including subsequent 
revisions.  The natural frequency for this study would be daily (or 
higher).  See chapter 4 of Campbell, Lo, and MacKinlay (1997) for 
examples of such an event study approach.

	Hope this helps.

Neely, Christopher J. & Sarno, Lucio, "How Well Do Monetary 
Fundamentals Forecast Exchange Rates?" Federal Reserve Bank of St. 
Louis Economic Review, September-October, 2002, pp. 51-74.

Campbell, John Y., Lo, Andrew W. & MacKinlay, A. Craig, "The 
Econometrics of Financial Markets," Princeton University Press, 1997.

                                        -- Mike

P.S.  The correct spelling is "dependent."


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/