Discrepancy between UN Comtrade and WITS

tbaras...@googlemail.com

unread,

Aug 3, 2015, 5:09:18 AM8/3/15

to witsforum

Hi,

I came across a rather worrying issue - the trade data I am extracting from WITS does not match what is available on the UN Comtrade website.

For example, for the year 2013, exports and imports to World, with classification HS1996 (other produce even worse outcomes), some countries have the same values as those available in Comtrade, while others can be off by as much as 8 (China) or even 18% (Panama).

This is reflected in the total figure - in WITS the total value of exports from ALL Countries to World is $17.165tn. In Comtrade the figure amounts to $17.960tn. Imports are more closely aligned, $17.7tn in WITS and $17.917tn in Comtrade.

Still, they are off and as I understand the WITS trade database is based on Comtrade so where are these errors coming from?

Also, if one uses the ISIC3 nomenclature, the error becomes even larger (despite using HS96 as the underlying reported nomenclature).

This really undermines the usefulness of WITS for analysis and I would greatly appreciate any clarification.

Samuel Munyaneza

unread,

Aug 3, 2015, 8:53:55 AM8/3/15

to tbaras...@googlemail.com, witsforum

This is because WITS doesn't take into account none HS product codes. In the case of China for example there are imports (or exports) reported by COMTRADE under code '999999' which is not an HS code.
WITS is good when analysing data at product level but for Totals it's perhaps better to use COMTRADE or even UNCTADstat (http://unctadstat.unctad.org/wds/ReportFolders/reportFolders.aspx?sRF_ActivePath=p,15912&sRF_Expanded=,p,15912)

--
--
You received this message because you are subscribed to the Google
Groups "witsforum" group.
To post to this group, send email to wits...@googlegroups.com
To unsubscribe from this group, send email to
witsforum+...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/witsforum?hl=en?hl=en
To access WITS go to http://wits.worldbank.org

---
You received this message because you are subscribed to the Google Groups "witsforum" group.
To unsubscribe from this group and stop receiving emails from it, send an email to witsforum+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

tbaras...@googlemail.com

unread,

Aug 3, 2015, 9:04:13 AM8/3/15

to witsforum, tbaras...@googlemail.com

Hi,

That's very helpful, thank you for clarification.

I have one other question - why is there a discrepancy within WITS itself when I choose HS1996 and ISIC3 but choosing HS1996 as a reporting nomenclature? I need to use ISIC3 to achieve alignment of my product groups with some OECD data, but it WITS give me different results in these two scenarios.

How can the total be different if it supposedly pulls out the same data and only categorises it differently?

Many thanks in advance!

Samuel Munyaneza

unread,

Aug 3, 2015, 9:09:34 AM8/3/15

to tbaras...@googlemail.com, witsforum

This can happen as there is never a one to one correspondence between different nomenclatures used in WITS. However, the difference should not be huge.

Can you give an example for us to investigate

To unsubscribe from this group and stop receiving emails from it, send an email to witsforum+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

--

--
You received this message because you are subscribed to the Google
Groups "witsforum" group.
To post to this group, send email to wits...@googlegroups.com
To unsubscribe from this group, send email to
witsforum+...@googlegroups.com
For more options, visit this group at

http://groups.google.com/group/witsforum?hl=en?hl=en
To access WITS go to http://wits.worldbank.org

---
You received this message because you are subscribed to the Google Groups "witsforum" group.

To unsubscribe from this group and stop receiving emails from it, send an email to witsforum+...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

tbaras...@googlemail.com

unread,

Aug 3, 2015, 9:22:25 AM8/3/15

to witsforum, tbaras...@googlemail.com

Hi,

Thank you.

I could possibly imagine that happening if it affected some product groups I created, but the issue is the same even when looking at total trade.

For example, India's imports from China in 2011 are 50,993,083.098 when choosing ISIC3 and reported nomenclature HS1996. This same figure takes the value of 55,483,025.359 when choosing HS1996 directly.

The same problem exists for India practically across the board and affects both imports and exports although here the different is largest. I can't understand why there would be a difference in the Total figure at all since the reporting nomenclatures are the same in both instances and the figure is a total of everything anyway.

Message has been deleted

tbaras...@googlemail.com

unread,

Aug 4, 2015, 7:39:27 AM8/4/15

to witsforum, tbaras...@googlemail.com

Hi,

I found the answer to my own question. ISIC3 simply does not include the 99 product group at all and when WITS tries to reconcile HS1996 and ISIC3 it omits these values in the total. A report based straight on HS does include the figure contrary to your first answer. So, adding back group 99 from Comtrade's website to values in ISIC3 gets us the HS figures.

Now, the worrying part is that while the HS in WITS does in fact account for the 99 group, when compared to Comtrade it still misses a portion of trade. Specifically, the total export figure in HS from WITS for the United States is 207bn short of that in Comtrade. On the other hand, the import figures are completely aligned.

Looking further into the matter, I discovered that Comtrade has 77 codes that start with 9 and end in AA. I suspect that these codes are where this disparity is coming from, because while the 99 group is included in WITS, these codes are not. Perhaps these are codes that encapsulate some sensitive products, maybe defense? There is a similar discrepancy for China although here exports align perfectly, while imports in Comtrade are 8% larger than those in WITS.

Is there any place where I could find out what these codes are? All I found is that they are UN Special Codes.

Best,

tad

On Monday, 3 August 2015 14:09:34 UTC+1, samuel.m...@unctad.org wrote:

Natasha Agarwal

unread,

Dec 31, 2015, 10:11:08 AM12/31/15

to witsforum, tbaras...@googlemail.com

Hi there,

I have a related question. I was downloading import data for Timor-Leste, and interestingly I was able to download imports into Timor-Leste from all the countries for the year 2013 from the UN COMTRADE (http://comtrade.un.org/data/). This data is not even '999999'. However, when I go to WITS, I cannot find any data on Timor-Leste, let alone for the year 2013. Can you please tell me why is there such a discerpancy?

Thanks,
Natasha

WITS Admin

unread,

Dec 31, 2015, 10:48:30 AM12/31/15

to witsforum, tbaras...@googlemail.com

Dear Natasha,

The UN Comtrade data catalog shows the data is available for Timor-Leste for 2004, 2005 and 2013

The below is the screen shot from WITS Data Availability. Please let me if you have any questions

Regards

WITS Team

Auto Generated Inline Image 1

Reply all

Reply to author

Forward