Reshaping a Shapefile?

[Branon Xerxes]

I have a shapefile that in addition to the coordinates etc, has monthly data on nighttime lights for districts in India. I am trying to import the shapefile into Stata 15 SE , and convert it to a panel while preserving the spatial properties. I want to use census_no (district identifiers) for later work.
The main code is pasted below

spshape2dta "`where'\lights\rad_dist", saving(rad_dist) replace
use rad_dist
label data "rad_dist excludes pixels>2800 and Delhi has been merged to one dist/state"
rename  census_no censusno
drop census_*
rename censusno census_no
spset census_no, modify replace
spset, modify coordsys(latlong, kilometers) 
save, replace
*Beginning reshape
reshape long avg, i(census_no) j(yyyymm)
rename avg rad
tostring yyyymm, generate(monthstr)
*Breaking up Years and Months
gen yearstr=substr(monthstr,1,4)
gen mthstr=substr(monthstr, 5,2)
destring yearstr, generate(year)
destring mthstr, generate(month2)
gen double yrmonth=ym(year, month2)
label var yrmonth "Stata Formatted(SIF) YYYYMM"
xtset census_no yrmonth
spset 
label var rad "Night Lights"
*note rad: These are the cleaned Night lights- i.e. pixel values greater than 2800 changed to null and values lower than zero replaced with zero.
drop yearstr mthstr monthstr 
rename month2 month
save radpanel, replace

After spset (sixth-last line), I get an error, "variable _ID does not uniquely identify the observations"

. xtset census_no yrmonth
       panel variable:  census_no (strongly balanced)
        time variable:  yrmonth, 627 to 695
                delta:  1 unit

. spset 
  Sp dataset
                data:  cross sectional
     spatial-unit id:  _ID (equal to census_no)
         coordinates:  _CY, _CX (latitude-and-longitude, kilometers)
    linked shapefile:  rad_dist_shp.dta
variable _ID does not uniquely identify the observations
    Do these data need to be xtset?
r(459);

end of do-file

I also tried spset census_no but that does not help either:

. xtset census_no yrmonth
       panel variable:  census_no (strongly balanced)
        time variable:  yrmonth, 627 to 695
                delta:  1 unit

. spset census_no
  Sp dataset
                data:  panel
     spatial-unit id:  _ID (equal to census_no)
             time id:  yrmonth (see xtset)
         coordinates:  none
    linked shapefile:  none

You can see now that it has lost all the coordinate and linked shapefile info.

Thanks.