Creating a spatially lagged variable using a spatial weighting matrix

[DY Kim]

I am a Stata 16 user and am trying to create a spatially lagged variable using a spatial weighting matrix within a panel data framework. I have created the spatial weighting matrix using the following command:

. spmatrix create contiguity W.

Can you advise me how to convert this matrix into a spatially lagged variable? What would be the next codes to achieve this?

[Andrew Musau]

If you have panel data, the weighting matrix should be created using a single cross-section. This is discussed in the documentation. Not sure how you were able to create it the way you describe without getting an error.

spmatrix create contiguity W if year==XXXX

Regarding your question, see

help spgenerate

[DY Kim]

Hi Andrew,

Thank you for the help. I have followed your directions but still got the same error message. Please see below.

. spset
Sp dataset final.dta
data: panel
spatial-unit id: _ID (equal to ct)
time id: date (see xtset)
coordinates: _CX, _CY (planar)
linked shapefile: 2020_shp.dta. spmatrix create contiguity W if date == tm(2023m12), normalize(row)
. spgenerate Wy = W*y
_IDs in weighting matrix W do not match _IDs in estimation sample
There are places in W not in estimation sample and places in estimation sample not in W.
r(459);

Any advice would be greatly appreciated.

[Andrew Musau]

See #9: https://www.statalist.org/forums/for...ing-spgenerate

[DY Kim]

Hi Andrew, thank you for the help.

After generating the spatial weighting matrix for December 2023, I attempted to populate the spatial lag variable (Wy) for all months from 2017 to 2023 using the same value for each spatial unit. I followed the code you suggested in the other post. However, instead of consistent values across months for each unit, I observed unexpected variation. I would appreciate it if you could check whether my code below is correct.

spmatrix create contiguity W if date == tm(2023m12)
spgenerate Wy = W*y if date == tm(2023m12)
levelsof date, local(dates)
foreach date in `dates'{
    spgenerate XYZ`date' = W*y if date==`date'
}
egen W_y= rowmax(XYZ*)
drop XYZ*

[Andrew Musau]

Why is the variation unexpected?

spgenerate Wy = W*y

Unless the value of \(y\) is constant across months, the multiplication above would yield different values for different months.