Panel Data: How many instances shared an organization code in the past period

[Cyril Hatton]

Hi all,

I'm running into some trouble with a dataset of bank mergers.

I have data at the branch/location level, and the attributes I have are:

BranchID (unique for every location)
BankID (will be shared by locations owned by the same corporate entity)
Revenue (annual)
Year

The trouble is, the BankID appears to change every year, even if it's verifiably the same corporation there as before. Is there a way to generate a dummy variable to see if at least 1/2 of current establishments that currently share a BankID with a given branch also shared a BankID with that branch in the last period?

The dataset looks like this:

BranchID	BankID	Revenue (million)	Year
1	115	$70	1990
2	115	$65	1990
3	120	$120	1990
4	120	$55	1990
1	134	$78	1991
2	134	$70	1991
3	180	$111	1991
4	180	$62	1991
1	155	$80	1992
2	155	$66	1992
3	155	$132	1992
4	190	$71	1992

And I'd want it to look like this --

BranchID	BankID	Revenue (million)	Year	LargerPartner
1	115	$70	1990	1
2	115	$65	1990	1
3	120	$120	1990	1
4	120	$55	1990	1
1	134	$78	1991	1
2	134	$70	1991	1
3	180	$111	1991	1
4	180	$62	1991	1
1	155	$80	1992	1
2	155	$66	1992	1
3	155	$132	1992	0
4	190	$71	1992	0

Thanks so much for your time!

[Carlo Lazzaro]

Cyril:
welcome to this forum.
You may want to consider the -group- function available from -egen-:

. sort branchID (Year)

. egen wanted=group( BankID)

. list

     +---------------------------------------------+
     | branchID   BankID   Revenue   Year   wanted |
     |---------------------------------------------|
  1. |        1      115        70   1990        1 |
  2. |        2      115        65   1990        1 |
  3. |        3      120       120   1990        2 |
  4. |        4      120        55   1990        2 |
     +---------------------------------------------+

.

Otherwise you can go:

. bysort BankID (Year): gen wanted2=1 if _n==1


. replace wanted2=1 if wanted[_n]== wanted[_n-1]


.