Hello,
I want to know how to OLS regress panel data with fixed-effect, when more than 2 dimensions are defining the panel.
Dataset
I have a dataset mocked below that contains panel data across 3 dimensions: year (temporal), as well as specie and observer (which are independent categorical variables). I.e. on each year, each observer has reported the count and (average) size of each specie.
Problem
So far, I have been running “OLS regression with fixed-effects” using i.<categorical> as independent variables for the fixed-effect (leaving aside the clustering of standard errors):
However, I gather this is not the correct way of calculating fixed-effect with panel data.
I have thus started porting my regressions to xtreg. I have been able to replicate regressions 1–3 as follows:
However, I struggle with scenarii 4 and 5. Indeed, the following command fail:
As for case 5, -help xtset- mention only 1 panelvar (singular) in addition to timevar, so I fully expect -xtset specie_id observer_id year- to fail too.
Question
How to run an OLS regression with the following fixed-effects, when the data are panelled on 3 dimensions (1 temporal + 2 categorical):
System
I am using Stata 17.0 MP.
I want to know how to OLS regress panel data with fixed-effect, when more than 2 dimensions are defining the panel.
Dataset
I have a dataset mocked below that contains panel data across 3 dimensions: year (temporal), as well as specie and observer (which are independent categorical variables). I.e. on each year, each observer has reported the count and (average) size of each specie.
Code:
* Example generated by -dataex-. For more info, type help dataex clear input int year str11 specie str7 observer byte count float size 2010 "Anchovie" "Alice" 98 69.97352 2010 "Anchovie" "Bob" 98 32.766758 2010 "Anchovie" "Charlie" 12 11.658455 2010 "Beaver" "Alice" 57 19.060453 2010 "Beaver" "Bob" 54 .04540863 2010 "Beaver" "Charlie" 97 94.63239 2010 "Caterpillar" "Alice" 6 3.7774425 2010 "Caterpillar" "Bob" 0 0 2010 "Caterpillar" "Charlie" 42 32.16354 2010 "Deer" "Alice" 83 53.52138 2010 "Deer" "Bob" 4 1.3835385 2010 "Deer" "Charlie" 86 77.69691 2011 "Anchovie" "Alice" 10 1.3469838 2011 "Anchovie" "Bob" 10 6.67327 2011 "Anchovie" "Charlie" 15 13.41947 2011 "Beaver" "Alice" 82 38.37632 2011 "Beaver" "Bob" 1 .4180519 2011 "Beaver" "Charlie" 35 19.670977 2011 "Caterpillar" "Alice" 18 1.374207 2011 "Caterpillar" "Bob" 58 41.45398 2011 "Caterpillar" "Charlie" 16 10.51918 2011 "Deer" "Alice" 48 30.710747 2011 "Deer" "Bob" 44 15.254408 2011 "Deer" "Charlie" 100 13.260367 2012 "Anchovie" "Alice" 0 0 2012 "Anchovie" "Bob" 23 6.768587 2012 "Anchovie" "Charlie" 96 25.88619 2012 "Beaver" "Alice" 88 59.88318 2012 "Beaver" "Bob" 18 3.859546 2012 "Beaver" "Charlie" 36 27.21044 2012 "Caterpillar" "Alice" 29 20.154434 2012 "Caterpillar" "Bob" 41 22.805317 2012 "Caterpillar" "Charlie" 43 22.87975 2012 "Deer" "Alice" 71 47.48964 2012 "Deer" "Bob" 0 0 2012 "Deer" "Charlie" 63 38.72793 2013 "Anchovie" "Alice" 65 47.36024 2013 "Anchovie" "Bob" 21 12.862803 2013 "Anchovie" "Charlie" 98 81.47609 2013 "Beaver" "Alice" 90 81.25338 2013 "Beaver" "Bob" 30 29.611444 2013 "Beaver" "Charlie" 28 18.24471 2013 "Caterpillar" "Alice" 53 34.682957 2013 "Caterpillar" "Bob" 4 .4044559 2013 "Caterpillar" "Charlie" 28 23.134344 2013 "Deer" "Alice" 33 12.742416 2013 "Deer" "Bob" 68 67.77893 2013 "Deer" "Charlie" 54 2.1096578 2014 "Anchovie" "Alice" 43 17.796349 2014 "Anchovie" "Bob" 23 9.286314 2014 "Anchovie" "Charlie" 35 21.72264 2014 "Beaver" "Alice" 49 15.02113 2014 "Beaver" "Bob" 0 0 2014 "Beaver" "Charlie" 77 5.291418 2014 "Caterpillar" "Alice" 46 28.128736 2014 "Caterpillar" "Bob" 22 16.836308 2014 "Caterpillar" "Charlie" 96 81.99249 2014 "Deer" "Alice" 80 35.820347 2014 "Deer" "Bob" 36 18.96105 2014 "Deer" "Charlie" 36 4.982587 2015 "Anchovie" "Alice" 14 5.368907 2015 "Anchovie" "Bob" 22 13.315876 2015 "Anchovie" "Charlie" 80 18.309069 2015 "Beaver" "Alice" 92 25.678885 2015 "Beaver" "Bob" 48 20.29379 2015 "Beaver" "Charlie" 55 49.40905 2015 "Caterpillar" "Alice" 41 27.920774 2015 "Caterpillar" "Bob" 65 26.987574 2015 "Caterpillar" "Charlie" 10 1.2208078 2015 "Deer" "Alice" 37 4.282799 2015 "Deer" "Bob" 45 33.59572 2015 "Deer" "Charlie" 62 9.016172 2016 "Anchovie" "Alice" 1 .9934739 2016 "Anchovie" "Bob" 94 36.00561 2016 "Anchovie" "Charlie" 38 7.68219 2016 "Beaver" "Alice" 60 48.01498 2016 "Beaver" "Bob" 84 44.79958 2016 "Beaver" "Charlie" 11 5.028427 2016 "Caterpillar" "Alice" 72 51.12494 2016 "Caterpillar" "Bob" 36 12.705547 2016 "Caterpillar" "Charlie" 17 14.800262 2016 "Deer" "Alice" 28 15.861224 2016 "Deer" "Bob" 49 24.369453 2016 "Deer" "Charlie" 49 38.53837 2017 "Anchovie" "Alice" 39 23.66544 2017 "Anchovie" "Bob" 64 38.87928 2017 "Anchovie" "Charlie" 64 36.107174 2017 "Beaver" "Alice" 76 73.23235 2017 "Beaver" "Bob" 74 50.15491 2017 "Beaver" "Charlie" 100 10.102198 2017 "Caterpillar" "Alice" 46 5.233415 2017 "Caterpillar" "Bob" 28 27.837124 2017 "Caterpillar" "Charlie" 17 2.9475856 2017 "Deer" "Alice" 22 7.284397 2017 "Deer" "Bob" 55 4.100322 2017 "Deer" "Charlie" 64 21.89705 2018 "Anchovie" "Alice" 89 55.62903 2018 "Anchovie" "Bob" 55 40.11375 2018 "Anchovie" "Charlie" 25 20.121956 2018 "Beaver" "Alice" 70 13.271093 2018 "Beaver" "Bob" 28 17.943098 2018 "Beaver" "Charlie" 89 79.80712 2018 "Caterpillar" "Alice" 94 48.44288 2018 "Caterpillar" "Bob" 60 21.88108 2018 "Caterpillar" "Charlie" 16 9.971855 2018 "Deer" "Alice" 88 2.622402 2018 "Deer" "Bob" 2 .6319003 2018 "Deer" "Charlie" 41 2.404092 2019 "Anchovie" "Alice" 74 37.69833 2019 "Anchovie" "Bob" 5 1.8495693 2019 "Anchovie" "Charlie" 49 35.447426 2019 "Beaver" "Alice" 58 17.566992 2019 "Beaver" "Bob" 61 39.27077 2019 "Beaver" "Charlie" 73 40.09311 2019 "Caterpillar" "Alice" 23 13.877348 2019 "Caterpillar" "Bob" 15 12.655438 2019 "Caterpillar" "Charlie" 55 35.99815 2019 "Deer" "Alice" 36 31.71533 2019 "Deer" "Bob" 6 3.630006 2019 "Deer" "Charlie" 83 21.364994 2020 "Anchovie" "Alice" 43 32.283325 2020 "Anchovie" "Bob" 0 0 2020 "Anchovie" "Charlie" 59 12.265055 2020 "Beaver" "Alice" 68 45.07475 2020 "Beaver" "Bob" 37 1.69084 2020 "Beaver" "Charlie" 18 4.1585655 2020 "Caterpillar" "Alice" 77 74.63856 2020 "Caterpillar" "Bob" 83 26.62211 2020 "Caterpillar" "Charlie" 99 6.484401 2020 "Deer" "Alice" 39 16.99574 2020 "Deer" "Bob" 15 7.349376 2020 "Deer" "Charlie" 94 92.99251 end
So far, I have been running “OLS regression with fixed-effects” using i.<categorical> as independent variables for the fixed-effect (leaving aside the clustering of standard errors):
Code:
* Encode categorical variables encode specie , generate(specie_id) encode observer , generate(observer_id) * 1/ Regression with no fixed-effect regress count size * 2/ Regression with Year fixed-effect regress count size c.year * 3/ Regression with Specie fixed-effect regress count size i.specie_id * 4/ Regression with Year & Specie fixed-effect regress count size c.year i.specie_id * 5/ Regression with Year & Specie & Observer fixed-effect regress count size c.year i.specie_id i.observer_id
I have thus started porting my regressions to xtreg. I have been able to replicate regressions 1–3 as follows:
Code:
* 1/ Regression with no fixed-effect regress count size * 2/ Regression with Year fixed-effect xtset year xtreg count size, fe * 3/ Regression with Specie fixed-effect xtset specie_id xtreg count size, fe
Code:
* 4/ Regression with Year & Specie fixed-effect xtset specie_id year // repeated time values within panel // r(451);
Question
How to run an OLS regression with the following fixed-effects, when the data are panelled on 3 dimensions (1 temporal + 2 categorical):
- temporal & categorical_1,
- temporal & categorical_1 & categorical_2?
System
I am using Stata 17.0 MP.
Comment