Re: st: panel data/no observations

[Andrew Reed <drew.reed1@gmail.com>]

Nick,

I have checked for strings that should be numeric. This issue is nonexistent. Xtset is also specified correctly. I believe my major problem is missings. What might be some methods to overcome this problem?

I do have gaps that imply missings but I never encountered any problems when running fixed effect panel regressions for event effects on event exchange rates. It was only when I created specifications for credit rating levels for non-event and event (which contained a plethora of missings) that I began experiencing this problem.

I assumed that in creating a dummy variable which would account for event windows that I would be able to measure how 'non event exchange rates' were affected by effectively dropping out the observations of exchange rate fluctuations for 'event' entities. Is this an acceptable way of approaching investigations of contagion/spillover effects?

Andrew

Den 16/08/2013 kl. 17.21 skrev Nick Cox:

> Odd things that can happen:
> 
> Missings
> 
> Strings that should be numeric
> 
> Gaps implying missings
> 
> Incorrect -tsset- or -xtset-, e.g. you have declared daily dates, but
> you have days once per week or or days once per month
> 
> Nick
> njcoxstata@gmail.com
> 
> 
> On 16 August 2013 16:13, William Buchanan <william@williambuchanan.net> wrote:
>> Given the reported error, it'd be helpful to see exactly what you typed into Stata as well as the number of observations in your data set.  It is possible that while you were recoding variables you did not populate all of the observations that should have valid values which would negatively affect your estimation sample.
>> 
>> HTH,
>> Billy
>> 
>> 
>> 
>> On Aug 16, 2013, at 10:01 AM, Andrew Reed <drew.reed1@gmail.com> wrote:
>> 
>>> I am trying to run a panel regression in Stata in order to determine spillover effects of an 'event' entity on 'non event' entities. The basic idea is to see if credit rating news announcements on one entity affect the exchange rates of other entities. I have therefore created an event dummy variable where I use '1' to define the event, and events can range from [-1,+1] to [-1,+7].
>>> 
>>> I have furthermore created a modification of my delta.exchange_rate variable where I omit changes in the exchange rate of an event entity in order to measure the effect of the credit rating news announcements only on 'non-event' changes in exchange rates. The problem is that now when I run the panel data regression I am served with an r(2000) error report saying that I lack observations. Does anyone know how to overcome this problem? My econometric equation takes the following form, where i denotes an 'event' entity and j represents all other entities that are defined as 'non-event'.
>>> 
>>> delta.exchange_rate.j,t = alpha + beta.1(delta.change in credit rating)i,t + beta.2(event credit rating level)i,t + beta.3(non-event credit rating level)j,t + lamda(controls, etc) + epsilon
>>> 
>>> If anyone has any suggestions or clarifications I would greatly appreciate it. Also, if clarifications for my own problem are needed, please feel free to let me know.
>>> 
>>> Sincerely,
>>> 
>>> Andrew W. Reed
>>> *
>>> *   For searches and help try:
>>> *   http://www.stata.com/help.cgi?search
>>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>>> *   http://www.ats.ucla.edu/stat/stata/
>> 
>> 
>> *
>> *   For searches and help try:
>> *   http://www.stata.com/help.cgi?search
>> *   http://www.stata.com/support/faqs/resources/statalist-faq/
>> *   http://www.ats.ucla.edu/stat/stata/
> 
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/


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