X-Sender: arose@haas.berkeley.edu
X-Mailer: QUALCOMM Windows Eudora Version 5.1
Date: Mon, 28 Jan 2002 08:49:31 -0800
To: tenreyro@fas.harvard.edu
From: Andrew Rose <arose@haas.berkeley.edu>
Subject: More on Missing Currency Union Observations
Cc: arose@haas.berkeley.edu
X-Security: The Haas email server has filtered this message.
X-Security: For details see http://groups.haas.berkeley.edu/hcs/howdoi/email_filtering.htm

I've now checked a little further into the issue of missing data (I'll post my data set on the web within the hour).  The IMF's Direction of Trade seems like it may have fewer missing observations than the WTDB.

 

In my (DoT) data set, I have observations for 18 of the 20 years for bilateral trade between CAR and Chad (you assert that the WTDB only has 9), and for 17 of the 20 for trade between CAR and Cote d'Ivoire (you only have 12).  I list the observations in my program pst2 (which I both attach and append at the bottom of this message).

 

Now I don't have the most up to date version of the WTDB CDs.  I have the 1997 ones distributed to NBER members which only go to 1992.  For that, there are observations for only 4 years between CAR (country code 141400) and Chad (141480), and 8 years between CAR and Cote d'Ivoire (163840).  The years also line up closely (though not exactly) with the ones you mention in your paper on p 14.  The following are extracts from the file WT_BILAT.ASC:

 

141400 141480        177 78

141400 141480        226 79

141400 141480        253 80

141400 141480       1669 89

 

141480 141400       1635 78

141480 141400        563 79

141480 141400        485 80

141480 141400         90 89

 

141400 163840        715 78

141400 163840       1504 79

141400 163840        523 80

141400 163840        687 81

141400 163840        379 82

141400 163840        385 83

141400 163840        473 85

141400 163840        996 89

 

163840 141400         85 78

163840 141400        271 79

163840 141400        959 80

163840 141400        275 81

163840 141400        307 82

163840 141400        136 83

163840 141400         16 85

 

So it seems like the WTDB simply has fewer observations that the DoT data set.  Using either data sampled every 5 years, or temporally averaged data, the DoT gives no indication that missing observations are the source of any bias for the effect of CU on trade.

 

What do you think?  Over to you --

 

-------------------------------------------------------------------------------

       log:  C:\res\cuts\progs\pst2.log

  log type:  text

 opened on:  28 Jan 2002, 08:26:05

 

. *

. * This STATA program tabulates data for the CFA countries that

. * Tenreyro lists on her page 14.

. *

. set more 1

 

. set mem 100m

(102400k)

 

. *

. * Read in the data set

. *

. use \res\cuts\data\data7ast

 

. *

. * List the data.  CAR is 626; Chad is 628 and Cote d'ivoire is 662.

. *

. * CAR and Chad post '77

. *

. list trade year if cty1==626&cty2==628&year>1977

 

           trade      year

1274831.  13135.78      1978

1274832.   5194.03      1979

1274833.   4709.69      1980

1274834.  15586.78      1981

1274835.  15437.03      1982

1274836.         0      1983

1274837.         0      1984

1274838.  1260.603      1985

1274839.  1327.313      1986

1274840.  1536.399      1987

1274841.  1845.313      1988

1274842.  1936.965      1989

1274843.  2112.938      1990

1274844.  2549.745      1991

1274845.  2861.125      1992

1274846.  2040.026      1993

1274847.  4803.166      1994

1274848.  6077.008      1995

1274849.   6947.66      1996

1274850.  7467.592      1997

 

. *

. * CAR and Cote d'Ivoire post '77

. *

. list trade year if cty1==626&cty2==662&year>1977

 

           trade      year

1275481.  4536.978      1978

1275482.  10388.06      1979

1275483.  7823.185      1980

1275484.  5707.478      1981

1275485.  3040.401      1982

1275486.  2811.715      1983

1275487.         0      1984

1275488.  4135.688      1985

1275489.  16186.92      1986

1275490.         0      1987

1275491.         0      1988

1275492.  3899.983      1989

1275493.  2558.077      1990

1275494.  1420.624      1991

1275495.  27314.09      1992

1275496.  21383.65      1993

1275497.  29571.58      1994

1275498.  51312.94      1995

1275499.  128071.6      1996

1275500.  137655.7      1997

 

. *

. * Clean up and close down

. *

. drop _all

 

. set more 0

 

. set mem 1m

(1024k)

 

. log close

       log:  C:\res\cuts\progs\pst2.log

  log type:  text

 closed on:  28 Jan 2002, 08:27:51

-------------------------------------------------------------------------------



********

Andrew K. Rose, Rocca Professor Tel: (510) 642-6609

Haas School of Business         Fax: (510) 642-4700

University of California                        E-mail: arose@haas.berkeley.edu

Berkeley, CA 94720-1900         URL: http://faculty.haas.berkeley.edu/arose

 

C:\Program Files\Qualcomm\Eudora\attach\pst2.logpst2.log