st: re: What to do about multiple observations for one individual in one same period in a panel

[Kit Baum <baum@bc.edu>]

<>
Adrian said

I am interested in looking at the effect of domestic macroeconomic  
conditions in a given country (e.g., GDP growth, inflation, the level  
of reserves...) as well as credit and liquidity conditions abroad  
(e.g., the US interest rate, volatility and liquidity indices...) on  
the yield spread of bonds issued by that country. The yield spread is  
a measured as the return of the bond minus the return of a "safe" or  
riskless benchmark, such as a US T-bill or bond. SO in my data, I have  
countries like Mexico, Brazil, Korea... that decide whether they want  
to issue bonds (or not) in a given period (self-selection)... but when  
they do participate in financial markets and decide to issue, they may  
actually issue multiple bonds, each with a different spread (dependent  
variable) and other individual characteristics (e.g., total amount  
issued, maturity of the bond, currency of issue, etc.) but facing the  
same macro conditions both domestically and abroad (i.e., in the same  
time period, the GDP growth, the inflation rate, the US interest  
rate, ... are all the same). Given the nature of my question, do you  
recommend I look into the nested panels literature? I have never seen  
these models before and so I have no idea what they do. If you think  
that would be a good estimation procedure, are there any references  
you could recommend?

With due respect to Jeph's suggestion of nesting issuances within  
country-years, I don't think that makes much sense from an economic  
standpoint. A sovereign borrower presumably has a target amount of  
funds to be raised in each period, and if they choose to issue  
securities they may do so (on their underwriters' recommendations)  
using several issues that appeal to different segments of the loanable  
funds market. In that sense I would consider the total amount borrowed  
and the weighted average yield spread and tenor to be more sensible  
measures in a traditional panel context. The individual issues cannot  
be considered independent, as depending on timing and information flow  
the lenders who purchase the July 20 issue are certainly taking the  
borrower's issuance of another bond on July 10 into account when they  
judge the appropriate risk premium. So I would come up with a way to  
collapse the data into country-quarter format. I do think that the  
number of issues per quarter may be worth noting, as a larger number  
of issues would seem to suggest a broader acceptance of the borrower  
in the market for sovereign lending.
Some of the time, I do play the role of a financial economist... e.g.
http://ideas.repec.org/a/bla/ecinqu/v47y2009i2p216-225.html

Kit


Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                              An Introduction to Stata Programming   
|   http://www.stata-press.com/books/isp.html
   An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html



*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/