I have a general question regarding best practices for panel data which is non-missing for some variables and missing on Saturdays/Sundays/holidays for other variables. To give an example, here's some data I made up. Return is the stock market return for country A on a given day. Google is the number of Google searches about a particular topic for country A on a given day. There are other countries in the made-up sample that aren't shown here. I'm interested in asking Stata to create a variable "Average5" that is the average of the returns over the last 5 days on which the market was open in the given country (non-missing). I have to be careful because some countries have a holiday on a given day while others do not. Therefore, I can't simply say "give me the average over the last 7 calendar days", because the last 7 calendar days will not necessarily have 5 non-missing observations, and the number of non-missing observations may be different across countries depending on holiday differences. What is the best way to ask Stata to give me the average of the last 5 non-missing returns day-by-day, country-by-country?
Thank you, and happy Memorial Day for those in a country that has a holiday!
Stan
Day | DofW | Country | Return | |
1 | Monday | A | 0.248993 | 76 |
2 | Tuesday | A | 0.635337 | 45 |
3 | Wednesday | A | 0.511604 | 78 |
4 | Thursday | A | 0.668078 | 34 |
5 | Friday | A | 0.85938 | 94 |
6 | Saturday | A | . | 48 |
7 | Sunday | A | . | 63 |
8 | Monday | A | 0.550717 | 71 |
9 | Tuesday | A | 0.62901 | 92 |
10 | Wednesday | A | 0.930895 | 76 |
11 | Thursday | A | 0.081141 | 45 |
12 | Friday | A | 0.992014 | 78 |
13 | Saturday | A | . | 34 |
14 | Sunday | A | . | 94 |
15 | Monday | A | 0.375389 | 48 |
16 | Tuesday | A | 0.304097 | 63 |
17 | Wednesday | A | 0.428649 | 71 |
18 | Thursday | A | 0.623587 | 92 |
19 | Friday | A | 0.611869 | 55 |
20 | Saturday | A | . | 51 |
21 | Sunday | A | . | 60 |
0 Response to Panel data which is non-missing for some variables and missing on Saturdays/Sundays/holidays for other variables
Post a Comment