Question with regard to nested loop

[remonrace]

Hi guys,

I've got a question on my specific set of data.

It is ordered like this:

Description	Manager number	Application year	Holding firm identifyer	Cited firm 1	Cited firm 2	Cited firm 3		Cited firm 1356
Variable name	mgrno	appyear	cusip	cusipcited 1	cusipcited 2	cusipcited 3	…	cusipcited 1356
	1	1980	A	G	.	H
	1	1980	B	B	D	I		Z
	1	1980	C	A	.	.		.
	1	1980	D	D	G	H		I
	1	1981	E	E	B	.		.
	1	1981	A	B
	1	1981	B	F	X	Y		Z
	1	1981	C	G
	1	1982	D
	1	1982	E
	1	1982	A
	1	1982	B
	…	…	…
	1	2006	B

So for 1 manager number (mgrno), I have a list in which year (appyear) he held which firms (cusip).

Example (in bold): in 1981, manager 1 held firm E. In that year, firm E cited patents held by firm E and B. I want to know, if firm E and B were also owned by manager 1, in the years before 1980 (which is true for both in this example).

Unfortunately, I have absolutely no idea where to start. Even on paper I can't solve it without using mutliple nested loops. Does anyone have an idea on how to take on such an issue?

[William Lisowski]

You might find it helpful to start by using the reshape command to transform your data from wide to long format. Copy this sample into your do-file editor and run it to see the entire output; I've just shown the final result.

// set up sample data changing "." to string missing value ""
clear
input mgrno appyear str1 (cusip cusipcited_1 cusipcited_2 cusipcited_3) 
1 1980 A G . H 
1 1980 B B D I 
1 1980 C A . . 
1 1980 D D G H 
1 1981 E E B . 
1 1981 A B . .
1 1981 B F X Y 
1 1981 C G . . 
1 1982 D . . .   
1 1982 E . . .   
1 1982 A . . .   
1 1982 B . . .   
end
foreach cited of varlist cusipcited* {
    replace `cited' = "" if `cited'=="."
    }
list, noobs sepby(appyear)
// demonstrate reshape
reshape long cusipcited_, i(mgrno appyear cusip) j(cite_num)
list, noobs sepby(appyear)
// clean up and delete extraneous stuff
rename cusipcited_ cited
drop if cited=="" & cite_num > 1
drop cite_num
list, noobs sepby(appyear)

  +---------------------------------+
  | mgrno   appyear   cusip   cited |
  |---------------------------------|
  |     1      1980       A       G |
  |     1      1980       A       H |
  |     1      1980       B       B |
  |     1      1980       B       D |
  |     1      1980       B       I |
  |     1      1980       C       A |
  |     1      1980       D       D |
  |     1      1980       D       G |
  |     1      1980       D       H |
  |---------------------------------|
  |     1      1981       A       B |
  |     1      1981       B       F |
  |     1      1981       B       X |
  |     1      1981       B       Y |
  |     1      1981       C       G |
  |     1      1981       E       E |
  |     1      1981       E       B |
  |---------------------------------|
  |     1      1982       A         |
  |     1      1982       B         |
  |     1      1982       D         |
  |     1      1982       E         |
  +---------------------------------+

[remonrace]

Thank you William. In the meantime, I was working on the file and found out the "." observations don't interfere with any codes I run on the file.

I have come quite far with writing a code in the wide format, but it is becoming extremely complicated. I had to use the wide format to match patent data (appyear - cusip - cusipcited) to intistutional holdings (manager number - appyear), but it might indeed be easier to reshape into long for this format. There is a difficulty with the dimensionality of the data, as some firms in some years cite up to 460 other firms but I can at least give it a try.

[Robert Picard]

This is not really a mata question and a variant of the same question was posted on the general forum so I posted a reply there.