Working with WIOD dataset

[karl siragusa]

Hi everyone,

my name is Karl Siragusa, I am a new user, so sorry in advance for any violation of forum rules.
I am a m.Sc student currently working on my dissertation, which involves the use of the WIOD dataset to study trade flows between countries of the European Union; it is my first time working with such database, I was wondering if any of you can share some material to help me dealing with input output tables in STATA.

Thanks in advance for any information
Karl

[Christian Otchia]

You request should be clear, with an example, if you want to receive an answer. Stata deals with matrix but not input output tables per se.

[karl siragusa]

Thank you for your answer, Christian. The dataset I am using covers around 15 years and comes in two formats: an excel file for each year, which looks like the first picture, and a .dta dataset for the whole period, which looks like the second one. I was wondering what the numbers in the second set mean and how are they derived from the excel files. Can someone point me to useful material about how to work with matrices in STATA and/or how to deal with input/output tables?

[Christian Otchia]

The data in the excel fine is for 2000 and the Stata file shows data for 1995. I think that the 3107.0305 is the intermediate demand from/to agriculture in AUS.
But still, it's not clear to me what you want to do. You should say what kind of analysis you want to do with the data, and show some command. If you want to work with matrices in Stata, you can search for : mata stata for example.
There are other software for pure input output analysis.

[karl siragusa]

Thank you for your answer Christian, I put the excel file of 2000 only for reference, to show the structure of the dataset; I will check the correspondence between the excel and dta file, as you suggested. The aim of my dissertation is to study the regional value chain that includes Germany and Eastern European countries and check the pattern of the value added by each country, showing how it increased/decreased in the last decade and trying to find an economic reason for that. I am fairly new to Stata and these tables, that's why I don't know how to work on it. Can you tell me more about this software you were talking about? Thank you for your help.

[Mike Lacy]

-findit input output table technical- from within Stata yields the following, which appears useful to your purposes:

package iot from http://fmwww.bc.edu/RePEc/bocode/i
---------------------------------------------------------------------------------------------------------------------
TITLE
'IOT': module to estimate Leontief Input-Output Table

DESCRIPTION/AUTHOR(S)

iot estimates Leontief Input-Output Table, and can estimate the
following results: (A) Technical Coefficients Matrix, Check I-O
Table Matrix, (I-A) Leontief Technical Coefficients Matrix,
Adjusted (I-A) Leontief Technical Coefficients Matrix, Input -
Output Table Multiplier, Labor Demand, and Leontief Input -
Output Table Results


I suspect, per Christian Otchia's comments, that this command provides a convenient wrapper for matrix manipulation commands in the Mata language.

Regards, Mike

[karl siragusa]

Thank you for your answer Mike, your tool seems really useful. I was wondering if it requires an I/O table with the technical coefficients already computed or if it can compute them from data about intermediate sales and total output. What I was wondering is if I can import the tables from the excel file and use the iot command or if I need some data preparation before using it with Stata.

[Mike Lacy]

Karl -- I don't have any actual experience using the -iot- program. I just found it by searching within Stata. Consequently, I don't know how it expects its input data to be formatted within Stata. I would be confident, though, that the data you have in Excel can be structured to meet the requirements of -iot-. What you should do is post back to the list with an explanation of what format -iot- requires, and what format your data currently has. With that information, people on the list should be able to help you.

Before doing that, I would encourage you strongly to read the Statalist FAQ (link at the top of this page) about how to post representations of data on Statalist. Among other things, you will find that posting images, as you previously did, is strongly discouraged on the list because they are generally unreadable. Posting Excel files is also not popular, so your best best would be to post something as plain text, a suggestion you will find in the FAQ.

Regards, Mike

[Christian Otchia]

Thank you for your answer Christian, I put the excel file of 2000 only for reference, to show the structure of the dataset; I will check the correspondence between the excel and dta file, as you suggested. The aim of my dissertation is to study the regional value chain that includes Germany and Eastern European countries and check the pattern of the value added by each country, showing how it increased/decreased in the last decade and trying to find an economic reason for that. I am fairly new to Stata and these tables, that's why I don't know how to work on it. Can you tell me more about this software you were talking about? Thank you for your help.

I think that Stata is appropriate for your analysis, as you try to look at the pattern of VA and its determinants. It's unfortunately not easy to explain you how to organize your data and conduct the analysis from stage 0. It seems that you want to conduct some econometric analysis but you may need to look first at some manual of data management in Stata.
When I talked about pure input output, I was thinking about Gams or matlab. The one Mike as mentioned is also interesting.

[Christian Otchia]

Thank you for your answer Mike, your tool seems really useful. I was wondering if it requires an I/O table with the technical coefficients already computed or if it can compute them from data about intermediate sales and total output. What I was wondering is if I can import the tables from the excel file and use the iot command or if I need some data preparation before using it with Stata.

You need to calculate the technical coefficient and organize your data before using iot.

[karl siragusa]

You need to calculate the technical coefficient and organize your data before using iot.

I was reading about technical coefficient and Leontief inverse, is it possible to do it in excel, using the matrix algebra functions? Do you know any tool/paper that I can use/read? For now I am consulting "Input - Output analysis, fundation and extensions" by MIller and Blair, trying to replicate their process in order to compute the values.

Thank you again for your help and your kind answers!

[Mike Lacy]

I suppose this might be possible in Excel, but the small effort to get your data into Stata will be rewarded by the availability of high quality and fast matrix functions and operations that have a terminology and syntax closely related to conventional practices in mathematics.

[Max Huang]

Hi Karl,

How about you research with WIOD?
Are you still using STATA to analyse the data? have you found any useful method or command?
I am also about to research the value-added issue of GVC, maybe we could discuss about it together.

[karl siragusa]

Hi Max,
thanks for your answer, I talked with my supervisor and he recommended me to work in Stata using mata, since it is easier to do everything in one program. I still have to learn Mata commands more in detail but I think I will employ this approach. Are you working on a dissertation as well? If you want, we can discuss and compare our projects, feel free to drop me a PM!

[karl siragusa]

Hi everyone,

I am writing because I made some progress using mata, which proved to be way simpler than other programs such as excel or mathematica, but I am still struggling to get the results I need. Basically, I started by working with the WIOD dataset in excel: I isolated a small set of 16 values and copy/pasted them in another sheet, I computed the sum of each row and its inverse and I imported this sample in Stata. Using mata, I inserted the row values in a 4x4 matrix (which can be called as Z) and the values of the inverse in a vertical vector . I transposed this vector in a diagonal matrix (let's call it X for convenience) and did Z*X, which gave me the technical coefficients matrix (let's call it A).
The final step should be finding the Leontief inverse, which is obtained by doing (I - A)^-1, where I is an identity matrix; I try to do this using the mata commands luinv(I - A) and invsym(I - A) but I get very strange results.

Does anyone know how to get these results? I attach to the post the small sample and the do file I used
Thank you in advance for your help!
Karl