interpolate monthly data from quarterly

[kevin smithy]

Hi,
I have time series data. Quarterly real GDP. I want to interpolate monthly data from the quarterly data. My data is already sorted in quarterly. I understand I can use ipolate to generate this. But I'm not sure how to do this. How do I assign each quarter to the corresponding month? How do I then interpolate linearly?
Please help me.
Thanks

* Example generated by -dataex-. For more info, type help dataex
clear
input double RGDP float quarters
1.967413e+13 207
1.804428e+13 208
1.792831e+13 209
1.890856e+13 210
2.100401e+13 211
1.869884e+13 212
1.898774e+13 213
1.967019e+13 214
2.110525e+13 215
1.893856e+13 216
 1.94016e+13 217
 2.06766e+13 218
2.333754e+13 219
2.879804e+13 220
2.723715e+13 221
2.969374e+13 222
2.994094e+13 223
3.074136e+13 224
2.842663e+13 225
3.056173e+13 226
3.178569e+13 227
 3.29908e+13 228
2.966174e+13 229
 3.23917e+13 230
3.432189e+13 231
3.421635e+13 232
3.121491e+13 233
3.381562e+13 234
3.310769e+13 235
3.416841e+13 236
3.066814e+13 237
3.421977e+13 238
3.300644e+13 239
 3.38609e+13 240
2.612069e+13 241
3.471957e+13 242
3.277599e+13 243
3.519921e+13 244
3.038855e+13 245
 3.26951e+13 246
3.343517e+13 247
3.463101e+13 248
end
format %tq quarters

[Nick Cox]

You could use ipolate. It seems simpler just to map each quarterly value to the three months to which it applies. Otherwise you could map quarterly values to months 2 5 8 11 and then interpolate, but the extra detail is spurious.

GDP reported to 7 significant figures is upstream of you, but astonishing any way. They got the measurements right on all the shadowy parts of the economy?

[kevin smithy]

You could use ipolate. It seems simpler just to map each quarterly value to the three months to which it applies. Otherwise you could map quarterly values to months 2 5 8 11 and then interpolate, but the extra detail is spurious.

GDP reported to 7 significant figures is upstream of you, but astonishing any way. They got the measurements right on all the shadowy parts of the economy?

Hi,
thank you for the reply.
Reported in local currency. Therefore I think figures should be correct. Can you please tell me which commands I should use to do this

Otherwise you could map quarterly values to months 2 5 8 11 and then interpolate, but the extra detail is spurious.

Thank you very much for your reply

[Nick Cox]

Here is a sketch, which follows from standard date-time functions:

   

gen mdate = mofd(dofq(quarters)) + 1  

format %tm mdate  

tsset mdate

tsfill  

gen work = log10(RGDP)

ipolate work mdate, gen(work2)  

list

I add the refinement that interpolating in log GDP may make a little more sense than using the raw data. Other way round, there is a question of whether your data are seasonally adjusted and if not whether interpolation should pay attention to seasonality.

[kevin smithy]

thank you very much