1. Introduction

This document provides a step-by-step view of the data preparation and calculation stages for a specific analytical task, as well as visualizations of some preliminary findings. It is part of a research project concerned with assessing the involvement of developing economies in Global Value Chains of non-renewable resource (NRR) sectors during the 2005-2014 period.

This project is part of my dissertation on value capturing through NRR-based industries, which partially fulfills the requirements for a doctoral degree in Public Policy at The New School University, NYC.

1.1 The analytical task

The main task is to calculate sector-level indexes of participation and position in NRR-based Global Production Networks (GPN). On the one hand, the GPN participation index estimates the share of gross sector-related exports and output involved in a vertically fragmented production process. Our index focuses on inter-industry trade (among different sectors) to account for the inputs that make possible the exploitation of own NRR; that is, it signals the integration of domestic inter-industry supply chains in GPN:

\[\sf{GPNPRT_{c,j}}=\frac{\sum_{i=1,i≠j}^{I} {IM_{i,j,c}}}{\sf{Y_{j}+\sum_{i=1,i≠j}^{N} {IX_{i}}}}+\frac{\sum_{i=1,i≠j}^{I} {IX_{c,i,j}}}{\sf{Y_{j}+\sum_{i=1,i≠j}^{N} {IX_{i}}}}\]

where \(\sf{GPNPRT_{c,j}}\) denotes our GPN participation index for country c and industry j, composed from buying (left) and selling (right) metrics. For the buying perspective of c, \(\sf{IM_{i,c,j}}\) denotes the foreign inputs from industry i of origin used by NRR-based industries j in buyer c of destination. For the selling side of c, \(\sf{IX_{c,i,j}}\) refers to inputs i from seller c exported for use by foreign sectors j. These buying and selling metrics are expressed as percentage of the sum total of sector gross output (\(\sf{Y_{j}}\)) and inputs from i in c exported for use by foreign sectors j (\(\sf{IX_{c,i,j}}\)) (our common denominator).

On the other hand, the GPN position index measures the nature of the integration of c in the GPN, defined as the log ratio of the selling to the buying metric for c through j:

\[\sf{GPNPSN_{c,j}}=ln\left(1+\frac{\sum_{i=1,i≠j}^{I} {IX_{i,j,c}}}{\sf{Y_{j}+\sum_{i=1,i≠j}^{N} {IX_{j}}}}\right)+ln\left(1+\frac{\sum_{i=1,i≠j}^{I} {IM_{c,i,j}}}{\sf{Y_{j}+\sum_{i=1,i≠j}^{N} {IX_{j}}}}\right)\]

the results of \(\sf{GPNPSN_{c,j}}\) are interpreted between two opposing poles: predominantly seller or buyer. That is, if the country-sector c,j is predominantly a seller of inputs along the global supply chain, the numerator tends to be large; on the contrary, if the country-sector is mainly a buyer in the GPN, then the numerator tends to be small. For instance, if the US specializes in providing inputs for extractive firms in Perú, the index tends to take on a higher value for the US and a lower value for Perú.

For our case, we examine the inputs used by the entire value chain of NRR production (both upstream and downstream segments) for 2005.

1.2 Data

B. Data sources:

  1. United National Global Material Flows data-set, UNGMF UN, 2019.
  2. Inter-country input-output tables data-cubes, ICIOv18 OECD-WTO, 2018:
    1. Matrices of intermediate transactions icio_z
    2. Data-sets on Trade in Value-added indicators icio_tiva

C. Relevant coding systems:

  1. Countries and regions: United Nations M49 UNSD, 2019.
  2. Industry/sectors: UN ISIC Rev. 4.

2. Data preparation

2.1 The study universe

Our unit of analysis is the domestic industry engaging indirectly (through inputs) in NRR production; for simplification purposes, we refer to it from now on as “countries”. Two criteria orient the delimitation of our study universe:

  1. data availability
  2. production of NRR.

2.1.1 Delimiting the study universe

Please click on the titles below to reveal the contents of each subsection.

2.1.1.1 Data availability: Cases in ICIOv18

NRR producers based on ICIO data: The universe of cases is determined by ICIOv18 (our main data source for data on intermediate transactions), which covers 64 economies (codes_cntrs_icio_all) + 1 Rest of the World (ROW) global-region for the 2005-2015 period:

# We count the number of unique country codes in ICIOv18:
n_distinct(unique(codes_cntrs_icio_all$country_code_iso3))
## [1] 64

2.1.1.2 Inclusion criteria: Data availability on DE of NRR

Data availability on DE of NRR for ICIOv18 cases:

Since our concern is with NRR production, the sole inclusion criterion of the study is defined as those countries engaging in Domestic Extraction (DE) of NRR for our starting year 2005; where NRR include Fossil Fuels and Metal Ores (tonnes). As mentioned before, our main data source is UNGMF UN, 2019.

The file codes_cntrs_intgrtd_0515 shows data availability on DE of NRR (UNGMF) for the universe of cases covered in ICIOv18. In sum, it integrates ICIOv18 cases, UNSD codes, and data from UNGMF relevant to our inclusion criteria; on the latter, it includes data on DE of NRR (tonnes) by country, and a dummy variable for NRR-production when DE of NRR > 0.

The main attributes of codes_cntrs_intgrtd_0515:

## Rows: 693
## Columns: 7
## $ country_code_iso3 <chr> "ARG", "ARG", "ARG", "ARG", "ARG", "ARG", "ARG", "A…
## $ country_name_mvf  <chr> "Argentina", "Argentina", "Argentina", "Argentina",…
## $ development_level <chr> "Developing", "Developing", "Developing", "Developi…
## $ subreg_name       <chr> "Latin America and the Caribbean", "Latin America a…
## $ year              <fct> 2005, 2006, 2007, 2008, 2009, 2010, 2011, 2012, 201…
## $ amount_nrr        <dbl> 111964302, 117591462, 114697002, 112168192, 1158527…
## $ nrr_prod_dummy    <fct> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, …
Overview of data availability from UNGMF, by country:
country_code_iso3 country_name_mvf development_level subreg_name year amount_nrr nrr_prod_dummy
ARG Argentina Developing Latin America and the Caribbean 2005 111964302 1
ARG Argentina Developing Latin America and the Caribbean 2006 117591462 1
ARG Argentina Developing Latin America and the Caribbean 2007 114697002 1
ARG Argentina Developing Latin America and the Caribbean 2008 112168192 1
ARG Argentina Developing Latin America and the Caribbean 2009 115852769 1

A closer look at codes_cntrs_intgrtd_0515 shows that there is only data on 63 out of the total 64 economies covered in ICIOv18; Taiwan, China (TWN), is not included.

# Number of countries from ICIOv18 on which there is data on DE of NRR (UNGMF):
n_distinct(unique(codes_cntrs_intgrtd_0515$country_name_mvf))
## [1] 63
#From the total cases covered in ICIOv18:
n_distinct(unique(codes_cntrs_icio_all$country_code_iso3))
## [1] 64

2.1.2 Our study universe

Based on the above, we create the file codes_stdy_unvrs_allnrr where we only include those cases meeting our inclusion criteria (i.e., DE of NRR > 0 in 2005 through our nrr_prod_dummy). In practical terms, codes_stdy_unvrs_allnrr serves as filter when making calculations for buying and selling sides; in this way, ICIO stays a raw data-cube that is filtered from an external source (our object) before calculations are made.

# we create codes_stdy_filter only those countries displaying DE of NRR > 0 in 2005 (our inclusion criteria) through our dummy variable:
codes_stdy_unvrs_allnrr <- codes_cntrs_intgrtd_0515 %>%
  filter(year == 2005 & nrr_prod_dummy == 1)

We show below an overview and detailed list of our study universe. In sum, we focus on 55 economies that were NRR producers by our starting year of 2005.

Please click on the titles below to reveal the contents of each subsection.

2.1.2.1 Overview of cases

a. Size: NRR producers in ICIOv18

To get a sense of the size of our study universe, we compare those cases that meet our inclusion criteria (NRR producers in 2005) to all cases covered in ICIOv18.

The size of our study universe (NRR-producers in 2005) is:

# we count the number of unique country codes in our file `codes_stdy_unvrs_allnrr`
n_distinct(unique(codes_stdy_unvrs_allnrr$country_code_iso3))
## [1] 55

While the number of cases that are covered in ICIOv18 (codes_cntrs_icio_all) but are excluded from our study universe (non NRR-producers in 2005) is:

# we count the number of cases [countries] from ICIOv18 that do not meet our inclusion criteria (i.e., our study universe)
n_distinct(unique(codes_cntrs_icio_all$country_code_iso3[!codes_cntrs_icio_all$country_code_iso3 %in% codes_stdy_unvrs_allnrr$country_code_iso3]))
## [1] 9

Based on the above, we focus on 55 economies that were NRR producers by our starting year of 2005. The remaining 9 economies and the ROW region are included as partner countries (i.e., countries of origin and destination).

b. Composition: Development level

The composition of our study universe by development level:

# We select only developed economies from our study universe
icio_stdy_unvrs_dvlpd <- filter(codes_stdy_unvrs_allnrr, development_level=="Developed")
# and we count the number of unique cases that are "Developed"
n_distinct(unique(icio_stdy_unvrs_dvlpd$country_code_iso3))
## [1] 33
# We select only developing economies from our study universe:
icio_stdy_unvrs_dvlpng <- filter(codes_stdy_unvrs_allnrr, development_level=="Developing")
# and we count the number of unique cases that are "Developing"
n_distinct(unique(icio_stdy_unvrs_dvlpng$country_code_iso3))
## [1] 22

2.1.2.1 Detailed list of cases

NRR producers in ICIOv18: List of included countries

A detailed list of countries in our study universe
development_level subreg_name country_code_iso3 country_name_mvf year amount_nrr
Developed Australia and New Zealand AUS Australia 2005 1026275900
Developed Australia and New Zealand NZL New Zealand 2005 14020343
Developed Eastern Asia JPN Japan 2005 5197602
Developed Eastern Europe BGR Bulgaria 2005 40338301
Developed Eastern Europe CZE Czech Republic 2005 65968060
Developed Eastern Europe HUN Hungary 2005 13718933
Developed Eastern Europe POL Poland 2005 242399520
Developed Eastern Europe ROU Romania 2005 51245996
Developed Eastern Europe RUS Russian Federation 2005 1383600542
Developed Eastern Europe SVK Slovakia 2005 3195885
Developed Northern America CAN Canada 2005 509819932
Developed Northern America USA United States of America 2005 2280372819
Developed Northern Europe DNK Denmark 2005 27129360
Developed Northern Europe EST Estonia 2005 15557400
Developed Northern Europe FIN Finland 2005 12679086
Developed Northern Europe GBR United Kingdom 2005 171220229
Developed Northern Europe IRL Ireland 2005 9035738
Developed Northern Europe LTU Lithuania 2005 622000
Developed Northern Europe LVA Latvia 2005 841865
Developed Northern Europe NOR Norway 2005 199448421
Developed Northern Europe SWE Sweden 2005 57067947
Developed Southern Europe ESP Spain 2005 21310260
Developed Southern Europe GRC Greece 2005 73557616
Developed Southern Europe HRV Croatia 2005 2565842
Developed Southern Europe ITA Italy 2005 14490922
Developed Southern Europe PRT Portugal 2005 287418
Developed Southern Europe SVN Slovenia 2005 4542880
Developed Western Asia ISR Israel 2005 1557645
Developed Western Europe AUT Austria 2005 4480691
Developed Western Europe BEL Belgium 2005 1200
Developed Western Europe DEU Germany 2005 222039759
Developed Western Europe FRA France 2005 2363050
Developed Western Europe NLD Netherlands 2005 49388936
Developing Central Asia KAZ Kazakhstan 2005 285920346
Developing Eastern Asia CHN China 2005 3108319155
Developing Eastern Asia KOR South Korea 2005 4718356
Developing Latin America and the Caribbean ARG Argentina 2005 111964302
Developing Latin America and the Caribbean BRA Brazil 2005 449802842
Developing Latin America and the Caribbean CHL Chile 2005 546513349
Developing Latin America and the Caribbean COL Colombia 2005 116659198
Developing Latin America and the Caribbean CRI Costa Rica 2005 170269
Developing Latin America and the Caribbean MEX Mexico 2005 375051265
Developing Latin America and the Caribbean PER Peru 2005 311726595
Developing Northern Africa MAR Morocco 2005 2371110
Developing Northern Africa TUN Tunisia 2005 5670457
Developing South-eastern Asia BRN Brunei Darussalam 2005 19209294
Developing South-eastern Asia IDN Indonesia 2005 922988475
Developing South-eastern Asia MYS Malaysia 2005 85658635
Developing South-eastern Asia PHL Philippines 2005 15712964
Developing South-eastern Asia THA Thailand 2005 50454805
Developing South-eastern Asia VNM Viet Nam 2005 62595007
Developing Southern Asia IND India 2005 684112336
Developing Sub-Saharan Africa ZAF South Africa 2005 407221218
Developing Western Asia SAU Saudi Arabia 2005 553444073
Developing Western Asia TUR Turkey 2005 71497096

2.2 Intermediate transactions data

For the calculation of our numerators and denominators, we use icio_z from ICIOv18:

# this are the dimensions of our datafile:
icio_z_dim
##   icio_z_dim
## 1         11
## 2       2484
## 3       2484
# where first row is number of individual matrices of intermediate transactions
# for each year in the '05-'15 period;
# the second and third rows correspond to the size of each individual matrix.

2.2.1 Transformations to our data

To facilitate calculations, we make a series of transformations to icio_z:

  1. We subset icio_z to our year of interest.
  2. We use codes_stdy_unvrs_allnrr to filter and focus only on our study universe (see 2.1.2).
  3. We simplify the original format of the data-file into an indexed format.
  4. We re-code special country codes.

The following subsections expound on how these transformations are conducted. Please click on the titles below to reveal the contents of each subsection.

2.2.1.1 Sub-setting a matrix to year of interest

We create icio_z_2005 from icio_z to focus on year 20051. To further facilitate data wrangling, we also transform the rows of icio_z_2005, which refer to the country/industry of origin for intermediate transactions, into a column with cntry_sct_name as header.

# Size of matrix
dim(icio_z_2005)
## [1] 2484 2484
First elements of our sub-matrix icio_z_2005:
cntry_sct_name AUS_01T03 AUS_05T06 AUS_07T08 AUS_09 AUS_10T12 AUS_13T15
AUS_01T03 3425.26872 102.45226 24.00948 0.0000000 11642.146396 99.9047580
AUS_05T06 107.55441 2687.92862 124.16880 0.5155636 131.895222 4.8860440
AUS_07T08 15.33443 381.54107 6454.67654 0.6550312 8.373702 0.2551598
AUS_09 0.00000 3879.36360 1850.74698 2724.6356920 0.000000 0.0000000
AUS_10T12 2015.42505 24.10272 27.23894 5.3428780 6341.098155 52.5556643
AUS_13T15 28.53818 24.56546 25.76046 3.6107721 47.444875 471.7913376

In sum, icio_z_2005 is a 2484x2484 matrix of intermediate transactions for year 2005, where the left row correspond to the country-industry of origin (sellers), and columns to the country-industry of destination (buyers).

2.2.1.2 Filtering by study universe

To facilitate calculations of buying and selling sides for our study universe (i.e., NRR-producers in 2005), we further subset icio_z_2005. We create two data-files where NRR-producers act as buyers (columns) or sellers (rows), respectively. Specifically, we filter columns (for buyers) and rows (for sellers) according to our criteria of inclusion2; such yields icio_z_2005_nrr_buyers and icio_z_2005_nrr_sellers.

First, we filter the rows to include those countries that are part of our study universe, where codes_stdy_unvrs_allnrr includes the country codes for our study universe:

# we select only those columns from the matrix whose countries match our study universe
icio_z_2005_tbl_allnrr_buyers <- select(icio_z_2005_tbl, contains("cntry_sct_name"),
  contains(paste(codes_stdy_unvrs_allnrr$country_code_iso3)), contains(paste(codes_cntrs_icio_mx_cn$country_isoalpha3_code)))
First elements of our sub-matrix icio_z_2005 for NRR producers as buyers:
cntry_sct_name ARG_01T03 ARG_05T06 ARG_07T08 ARG_09 ARG_10T12 ARG_13T15
AUS_01T03 1.2384974 0.0003615 0.0009479 0.0002585 4.1980736 0.0760311
AUS_05T06 1.8442650 0.1867868 0.3516953 0.7059199 1.1247149 0.1999119
AUS_07T08 0.0003883 0.0001985 0.0008452 0.0000763 0.0001857 0.0000437
AUS_09 0.0000006 0.0000045 0.0000004 0.0000004 0.0000001 0.0000000
AUS_10T12 0.0161091 0.0012495 0.0010919 0.0015021 0.0213997 0.0070686
AUS_13T15 0.0072439 0.0009814 0.0007429 0.0008683 0.0084536 0.1746471

Second, we also filter rows to include our study universe:

# we select only those rows from the matrix whose countries match our study universe
icio_z_2005_tbl_allnrr_sellers <- icio_z_2005_tbl %>%
  filter(grepl(paste(codes_stdy_unvrs_allnrr_mx_cn$country_isoalpha3_code, collapse = "|"), cntry_sct_name))
First elements of our sub-matrix icio_z_2005 for NRR producers as sellers:
cntry_sct_name AUS_01T03 AUS_05T06 AUS_07T08 AUS_09 AUS_10T12 AUS_13T15
AUS_01T03 3425.26872 102.45226 24.00948 0.0000000 11642.146396 99.9047580
AUS_05T06 107.55441 2687.92862 124.16880 0.5155636 131.895222 4.8860440
AUS_07T08 15.33443 381.54107 6454.67654 0.6550312 8.373702 0.2551598
AUS_09 0.00000 3879.36360 1850.74698 2724.6356920 0.000000 0.0000000
AUS_10T12 2015.42505 24.10272 27.23894 5.3428780 6341.098155 52.5556643
AUS_13T15 28.53818 24.56546 25.76046 3.6107721 47.444875 471.7913376

2.2.1.3 Creating distinct simplified versions

We also conduct transformations to the original format of our data-files (i.e., matrices) into an indexed format to facilitate calculations 3. The result, is icio_z_2005_nrr_buyers_indexed and icio_z_2005_nrr_sellers_indexed. These last two become our main data-files to calculate our buying and selling metrics.

# Main attributes of our newly-crated datafile for buyers in indexed format:
glimpse(icio_z_2005_tbl_allnrr_buyers_pvt)
## Rows: 5,276,016
## Columns: 5
## $ cntry_c  <chr> "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS…
## $ sctr_c_i <chr> "01T03", "05T06", "07T08", "09", "10T12", "13T15", "16", "17…
## $ cntry_p  <chr> "ARG", "ARG", "ARG", "ARG", "ARG", "ARG", "ARG", "ARG", "ARG…
## $ sctr_p_j <chr> "01T03", "01T03", "01T03", "01T03", "01T03", "01T03", "01T03…
## $ value    <dbl> 1.238497e+00, 1.844265e+00, 3.882531e-04, 6.126050e-07, 1.61…
First elements of data on buyers in indexed format:
cntry_c sctr_c_i cntry_p sctr_p_j value
AUS 01T03 ARG 01T03 1.2384974
AUS 05T06 ARG 01T03 1.8442650
AUS 07T08 ARG 01T03 0.0003883
AUS 09 ARG 01T03 0.0000006
AUS 10T12 ARG 01T03 0.0161091
AUS 13T15 ARG 01T03 0.0072439

where column cntry_c and sctr_c_i are the selling country and industry of origin, respectively; while cntry_p and sctr_p_j are the buying country and industry of destination. Finally, value refers to the intermediate transaction between country-industry seller and buyer.

2.2.1.4 Re-coding special codes from ICIOv18

ICIOv18 splits MEX and CHN into MX1 & MX24 and CN1 & CN25, respectively. As a consequence we have + 4 countries listed as buyers and partner countries:

# Total number of NRR-producing countries as buyers (should be 55):
n_distinct(as_factor(icio_z_2005_tbl_allnrr_buyers_pvt$cntry_p))
## [1] 59
# Total number of partner countries of NRR-producers (should be 65):
n_distinct(as_factor(icio_z_2005_tbl_allnrr_buyers_pvt$cntry_c))
## [1] 69

To solve this issues, we re-code MX1/MX2 to MEX, and CN1/CN2 to CHN in our buyer indexed file icio_z_2005_nrr_buyers_indexed:

# we rename CN1,CN2 as CHN for buyers_files
icio_z_2005_tbl_allnrr_buyers_pvt_CHN <- icio_z_2005_tbl_allnrr_buyers_pvt %>%
  mutate_at(vars("cntry_c", "cntry_p"),funs(str_replace_all(., c("CN1|CN2"), "CHN")))

# we rename MX1,MX2 as MEX for buyers_files, taking the above file as input
icio_z_2005_tbl_allnrr_buyers_pvt_CHNMEX <- icio_z_2005_tbl_allnrr_buyers_pvt_CHN %>%
  mutate_at(vars("cntry_c", "cntry_p"),funs(str_replace_all(., c("MX1|MX2"), "MEX")))

After re-coding both of our buying and selling indexed files, we confirm that our data-files match the size of our universe of study (55) and their partner countries (64 + 1 [ROW]):

Data-files for NRR-producers as buyers:

# 65 partner countries of buyers [c]:
n_distinct(as_factor(icio_z_2005_tbl_allnrr_buyers_pvt_CHNMEX$cntry_c))
## [1] 65
# NRR-rpdoucers in 2005 as buyers [study universe]:
n_distinct(as_factor(icio_z_2005_tbl_allnrr_buyers_pvt_CHNMEX$cntry_p))
## [1] 55

Data-files for NRR-producers as sellers:

# 65 partner countries of sellers [p]:
n_distinct(as_factor(icio_z_2005_tbl_allnrr_sellers_pvt_CHNMEX$cntry_p))
## [1] 65
# NRR-rpdoucers in 2005 as sellers [study universe]:
n_distinct(as_factor(icio_z_2005_tbl_allnrr_sellers_pvt_CHNMEX$cntry_c))
## [1] 55

2.2.2 Overview of final files

The following two data-files are the result of the transformations in subsection 2.2.1; these are to calculate buying and selling metrics of NRR producers during 2005:

icio_z_2005_allnrr_buyers main attributes:

## Rows: 5,276,016
## Columns: 5
## $ cntry_c  <chr> "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS…
## $ sctr_c_i <chr> "01T03", "05T06", "07T08", "09", "10T12", "13T15", "16", "17…
## $ cntry_p  <chr> "ARG", "ARG", "ARG", "ARG", "ARG", "ARG", "ARG", "ARG", "ARG…
## $ sctr_p_j <chr> "01T03", "01T03", "01T03", "01T03", "01T03", "01T03", "01T03…
## $ value    <dbl> 1.238497e+00, 1.844265e+00, 3.882531e-04, 6.126050e-07, 1.61…
First elements of our data-files on buyers:
cntry_c sctr_c_i cntry_p sctr_p_j value
AUS 01T03 ARG 01T03 1.2384974
AUS 05T06 ARG 01T03 1.8442650
AUS 07T08 ARG 01T03 0.0003883
AUS 09 ARG 01T03 0.0000006
AUS 10T12 ARG 01T03 0.0161091
AUS 13T15 ARG 01T03 0.0072439

icio_z_2005_allnrr_sellers main attributes:

## Rows: 5,276,016
## Columns: 5
## $ cntry_c  <chr> "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS…
## $ sctr_c_i <chr> "01T03", "05T06", "07T08", "09", "10T12", "13T15", "16", "17…
## $ cntry_p  <chr> "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS", "AUS…
## $ sctr_p_j <chr> "01T03", "01T03", "01T03", "01T03", "01T03", "01T03", "01T03…
## $ value    <dbl> 3425.26872, 107.55441, 15.33443, 0.00000, 2015.42505, 28.538…
First elements of our data-files on sellers:
cntry_c sctr_c_i cntry_p sctr_p_j value
AUS 01T03 AUS 01T03 3425.26872
AUS 05T06 AUS 01T03 107.55441
AUS 07T08 AUS 01T03 15.33443
AUS 09 AUS 01T03 0.00000
AUS 10T12 AUS 01T03 2015.42505
AUS 13T15 AUS 01T03 28.53818

In sum, our datafles in indexed formats can be understood as follows. In both cases, cntry_c and sctr_c_i are the country-industry of origin, while cntry_p and sctr_p_j are the country-industry of destination. In case of data on buyers, all cntry_p are NRR producers; in the case of sellers, only those countries producing NRR are included in cntry_c.

2.3 Defining NRR-based GPN: Which sectors?

We associate NRR (Fossil Fuels + Metal Ores) to sectors in order to define what constitutes an NRR-based GPN; in turn, we split the NRR VC in upstream (NRR extraction) and downstream segments (NRR beneficiation):

Materials to sectors: Correspondence table
flow_category_UNGMF upstream_sectors upstream_codes_OECD downstream_sectors downstream_codes_OECD
Fossil fuels Mining and extraction of energy producing products; Mining support service activities D05T06 ; D09 Coke and refined petroleum products D19
Metal ores Mining and quarrying of non-energy producing products D07T08 ; D09 Basic metals D24

Since we are interested only in inter-industry trade, this correspondence becomes our way to filter sectors in future calculations and focus on the entire VC as destination sectors; this involves excluding any intra-industry trade between sector of the NRR VC.

3. Calculations

3.1 Buying and selling metrics for the NRR VC

We calculate below the buying and selling metrics of our GPN indexes.

3.1.1 Buying metrics: imported inputs

The buying side involves calculating the total value of intermediate imports (IM) destined for use by the entire NRR-production VC of NRR-producers. This implies the following steps:

  • exclude NRR-producers as countries of origin (i.e., buyers of imports);
    • focus on NRR VC’s corresponding sectors as destination while excluding them as sectors of origin (sector buyers of imports);
    • aggregate the value of all inter-industry intermediate transactions.

We use our indexed version of icio_z_2005_allnrr_buyers as input:

# where nrrprd_vc is the sector codes associated with the NRR-production VC
indxd_icio_z_2005_allnrr_buyers_nrrprd_vc_im_ind <- indxd_icio_z_2005_allnrr_buyers %>%
  filter(cntry_p != cntry_c,
  sctr_p_j %in% codes_nrrprd_vc,
  !sctr_c_i %in% codes_nrrprd_vc)

#Our destination sectors are those associated above with the NRR-production VC:
  unique(indxd_icio_z_2005_allnrr_buyers_nrrprd_vc_im_ind$sctr_p_j)
## [1] "05T06" "07T08" "09"    "19"    "24"
# The total number of NRR-producers acting as buyers (55):
  n_distinct(unique(indxd_icio_z_2005_allnrr_buyers_nrrprd_vc_im_ind$cntry_p))
## [1] 55
# once filters are in place, we aggregate the value of intermediate imports (im) for buyers (cntry_p):
indxd_icio_z_2005_allnrr_buyers_nrrprd_vc_im_ind <- indxd_icio_z_2005_allnrr_buyers_nrrprd_vc_im_ind %>%
  group_by(cntry_p) %>% summarise(im_ind = sum(value))
Buyer metrics: First elements
cntry_p im_ind
ARG 1318.8245
AUS 3223.0204
AUT 2078.7317
BEL 4990.3792
BGR 396.9364
BRA 4298.4926

where cntry_p is the buyer country of inter-industry imported inputs im_ind destined for NRR-based sectors.

3.1.2 Selling metrics: exported inputs

The selling side involves calculating the total value of intermediate exports (IX) destined for the NRR-production VC of foreign NRR-producers. Since our data-files icio_z_2005_allnrr_sellers only includes NRR-producers as countries of origin (cntry_c) we follow the same steps for buying metrics.

We replicate the above for icio_z_2005_allnrr_sellers:

# The total number of NRR-producers acting as sellers of ix equals our universe of study (55):
n_distinct(unique(indxd_icio_z_2005_allnrr_sellers_nrrprd_vc_ix_ind$cntry_c))
## [1] 55
Selling metrics: First elements
cntry_c ix_ind
ARG 740.0962
AUS 6089.9520
AUT 2300.6573
BEL 3741.1325
BGR 241.3122
BRA 2634.4856

where cntry_c is the selling country exporting inter-industry inputs ix_ind destined for foreign NRR-based sectors.

3.2 The GPN index

3.2.1 Calculating sides of indexes

GPN indexes include buying and selling metrics for exporting country c and its sector j expressed as shares of gross sector output and sector-related exports (i.e., \(\sf{Y_{j}+\sum_{i=1,i≠j}^{N} {IX_{j}}}\)). In our case, country c is a NRR producer and sector j includes all NRR-based sectors.

To exemplify our calculations, we focus on buying metrics.

For the denominator, the data sources used were the following:

  • NRR-based sectors’ output: from icio_tiva data-set
    • NRR-related intermediate exports: previously calculated in 3.1.2

For the numerators, we use as data our selling metrics from

Both of these metrics are included our the data-file allnrr_gpn_sllng_nrrprd_vc_2005, where countries and sectors are coded in terms of country c and sector j (\(\sf{GPN_{c,j}}\))6.

# buying side nrrprd_vc, where im_ind is our buying metric, and gpn_dnmntr our denominator;
# we perform this operation for NRR-producers acting as buyers (cntry_c) who produce and export through NRR-based sectors (sctr_sgmnt_c_j)
allnrr_gpn_byng_nrrprd_vc_2005_tmp <- allnrr_gpn_byng_nrrprd_vc_2005_tmp  %>%
  group_by_at(vars("cntry_c","sctr_sgmnt_c_j")) %>%
  mutate(gpn_byng_im_ind = im_ind / gpn_dnmntr)
Selling side of the GPN: First elements of the datafile for for 2005:
cntry_c sctr_sgmnt_c_j year gpn_dnmntr im_ind im_total gpn_byng_im_ind
ARG nrrprd_vc 2005 41485.396 1318.8245 2404.312 0.0317901
AUS nrrprd_vc 2005 156702.852 3223.0204 9192.028 0.0205677
AUT nrrprd_vc 2005 24323.657 2078.7317 7380.586 0.0854613
BEL nrrprd_vc 2005 55573.532 4990.3792 19737.938 0.0897978
BGR nrrprd_vc 2005 6884.312 396.9364 3092.977 0.0576581
BRA nrrprd_vc 2005 149619.786 4298.4926 14852.954 0.0287294

where cntry_c is the country NRR-producing country acting as buyer in the GPN, sctr_sgmnt_c_j is the producing and exporting NRR-based sector, gpn_dnmntr is the sum of sector output (\(\sf{Y_{j}}\)) and sector-related intermediate exports (\(\sf{IX_{c,i,j}}\)) (our denominator), im_ind is the imported intermediates destined for NRR-based sectors (\(\sf{IM_{i,c,j}}\)), and gpn_byng_im_ind is our buying metric in terms of the GPN (im_ind expressed as share of gpn_dnmntr).

3.2.2 Calculating GPN indexes

Following our main analytical task mentioned above, we calculate our indexes on participation and position in GPN of NRR-based sectors.

We use a data-file gpn_nrrprd_vc_2005, which compiles our buying and selling metrics calculated previously; with these metrics, calculating our indexes is fairly simple:

gpn_nrrprd_vc_2005_tmp <- gpn_nrrprd_vc_2005_tmp %>%
  group_by_at(vars("cntry_c","sctr_sgmnt_c_j")) %>%
  mutate(gpn_prtc_ind = gpn_byng_im_ind + gpn_sllng_ix_ind) %>%
  mutate(gpn_pstn_ind = (log((1 + gpn_sllng_ix_ind))) - (log((1 + gpn_byng_im_ind))))
# where gpn_prtc_ind is the result of calculating our index on GPN participation;
# and gpn_pstn_ind on the GPN position.
GPN indexes: First elements of the datafile for 2005
cntry_c sctr_sgmnt_c_j year gpn_byng_im_ind gpn_sllng_ix_ind gpn_prtc_ind gpn_pstn_ind
ARG nrrprd_vc 2005 0.0317901 0.0178399 0.0496300 -0.0136126
AUS nrrprd_vc 2005 0.0205677 0.0388631 0.0594308 0.0177678
AUT nrrprd_vc 2005 0.0854613 0.0945852 0.1800465 0.0083704
BEL nrrprd_vc 2005 0.0897978 0.0673186 0.1571164 -0.0208426
BGR nrrprd_vc 2005 0.0576581 0.0350525 0.0927106 -0.0216050
BRA nrrprd_vc 2005 0.0287294 0.0176079 0.0463373 -0.0108698

where cntry_c is the NRR-producing involved in the GPN, sctr_sgmnt_c_j is the producing and exporting NRR-based sector, gpn_byng_im_ind is our buying metric, and gpn_sllng_ix_ind our selling metric. Finally, gpn_prtc_ind is GPN participation index for cntry_c, and gpn_pstn_ind its GPN position index.

4. Visualization of results

In this section, we provide some visualizations of some preliminary findings for year 2005.

4.1 Buying and selling sides of NRR producers

A relationship on our buying and selling metrics based on development levels of NRR producers in the GPN:

## `geom_smooth()` using formula 'y ~ x'

4.2 GPN participation index

Results from the GPN participation index, ranked from higher the lower:

4.2 GPN position index

Results from the GPN position index, ranked from higher the lower:

  1. The research project focuses on the ’05-’14 period; however, we only focus on 2005 as an example in this document.↩

  2. This filtering stage takes into account that in ICIOv18, countries MEX and CHN are further broken down as MX1 & MX2, and CN1 & CN2, respectively.↩

  3. Matrices can be thought as resembling the format of a Cartesian product between two sets (columns of buyers of inputs and rows of sellers of inputs), where their intersection yields the value of transactions (i.e., cells). To facilitate data manipulation in R, we transform our data-files into an indexed format, where variables are columns, and rows are observations. In this way, we have variables for “country of origin” and “of destination”, which correspond to buyers and sellers, respectively; a final column corresponds to the value of transactions. Note that this same data format is adopted by the published version of OECD’s ICIO-TiVa data-sets (matrices of intermediate transactions in VA terms).↩

  4. Global Manufacturing activities.↩

  5. Export processing activities.↩

  6. Notice that within matrix icio_z, country c was coded in terms of country of origin of traded intermediates; for \(\sf{GPN_{c,j}}\) country c refers to the producing and exporting country and its associated sector j.↩