How to handle the missing values generated by tssmooth exponential in unbalanced panel data

[David Wang]

Dear all,

I have an unbalanced panel dataset. Panel variable is firm_id and time variable is year. The time period is between 2010 and 2015. I have been trying to compute the exponentially weighted average of past performance. Thus, I picked tssmooth exponential. However, when I applied the tssmooth exponential, all the firms without all six years generate missing values for the exponential smoothed performance. For example, if Firm 2 has only 2010, 2011, and 2012, the exponentially weighted average of past performance for Firm 2 will be missing. I get stuck on how to avoid these missing values. Are there any other ways to compute the exponentially weighted average of past performance? Really looking forward to suggestions. Thank you in advance.

Best,
David


Herere athe sample data and my codes:

 * Example generated by -dataex-. To install: ssc install dataex 
clear 
input int year int firm_id float performance
2010    1    0.053
2011    1    0.075
2012    1    0.088
2013    1    0.036
2014    1    0.049
2015    1    -0.109
2010    2    0.464
2011    2    -0.014
2012    2    -0.002
2010    3    0.120
2012    3    0.018
2010    4    0.038
2011    4    0.045
2012    4    0.051
2013    4    0.127
2014    4    0.057
2015    4    0.059
2010    5    0.134
2011    5    0.117
end

tssmooth e double p1=performance, parms(0.1)
tssmooth e double p2=performance, parms(0.2)
tssmooth e double p3= performance, parms(0.3)
tssmooth e double p4= performance, parms(0.4)
tssmooth e double p5= performance, parms(0.5)
tssmooth e double p6= performance, parms(0.6)
tssmooth e double p7= performance, parms(0.7)
tssmooth e double p8= performance, parms(0.8)
tssmooth e double p9= performance, parms(0.9)

Many thanks,
David