# st: mvsumm and missing obs

 From jansoko@umich.edu To statalist@hsphsun2.harvard.edu Subject st: mvsumm and missing obs Date Tue, 10 Aug 2004 12:06:24 -0400

```Dear Stata users,

proposed solution (using -tssmooth-) doesn't work as I am interested in moving
variances, not moving averages.

How can I get create a variable that contains the variance of the five preceding
observations in a panel data set, even if one or two of the those preceding
observations are missing?  And how can I specify the minimum number of
non-missing observations within the window?

Thanks, Jan

> From   "Scott Merryman" <smerryman@kc.rr.com>
> To   <statalist@hsphsun2.harvard.edu>
> Subject   st: Re: mvsumm
> Date   Fri, 5 Mar 2004 17:29:35 -0600

> ----------------------------------------------------------------------------->
---

> -tssmooth ma- gives missing obersvations a zero weight in calculating the
> moving
> average.  This would seem to give you what you want.

> Example:

> . tsset var1
>         time variable:  var1, 1990 to 2003

> . tssmooth ma ma2= var2, w(4 1)
> The smoother applied was
>      (1/5)*[x(t-4) + x(t-3) + x(t-2) + x(t-1) + 1*x(t)]; x(t)= var2

> . l

>      +--------------------+
>      | var1   var2    ma2 |
>      |--------------------|
>   1. | 1990      .      . |
>   2. | 1991      .      . |
>   3. | 1992      3      3 |
>   4. | 1993      4    3.5 |
>   5. | 1994      5      4 |
>      |--------------------|
>   6. | 1995      5   4.25 |
>   7. | 1996      4    4.2 |
>   8. | 1997      6    4.8 |
>   9. | 1998      5      5 |
>  10. | 1999      3    4.6 |
>      |--------------------|
>  11. | 2000      4    4.4 |
>  12. | 2001      5    4.6 |
>  13. | 2002      6    4.6 |
>  14. | 2003      .    4.5 |
>      +--------------------+

> Hope this helps,
> Scott

----- Original Message -----
To: <statalist@hsphsun2.harvard.edu>
Sent: Friday, March 05, 2004 4:23 PM
Subject: st: mvsumm

> Dear Stata users,
>
> I am using Chris Baum's and Nick Fox's -mvsumm- and I'd like to know if
> there is a way to account for averages (or any other desc stats) when
> there are missing observations (e.g., the way -collapse- accounts for
> them). What I mean  is the following:
>
> 1990   .
> 1991   .
> 1992   3
> 1993   4
> 1994   5
> 1995   5
> 1996   4
> 1997   6
> 1998   5
> 1999   3
> 2000   4
> 2001   5
> 2002   6
> 2003   .
>
> If I use -mvsumm- to generate 5-year moving averages, I'd like it to
> even give me a number in year 1994 (when there are already 5
> observations to compute the average, even though 2 are missing--so the
> generated number would be the average of 3, 4, and 5) and also in 2003
> (which would be the average of 3, 4, 5, and 6). Is this possible?
>
> Thank you.