FDIp
Step 1: Computation of similarity-dissimilarity index
This is achieved by computing similarity index in which a low value means close similarity (or low dissimilarity) with Nigeria while high value implies low similarity (or high dissimilarity) with Nigeria. There are three aspects to this similarity-dissimilarity calculation, viz: CS (CS); NS (NS); and Facility similarity (FS). Each of these is explained below:
a. Cultural characteristics similarity
This is computed as the simple average of the standardised score of each of the five CS variables. The CS variables are represented byC ; ik
where
i = 1-1511 for each of the West African Countries except Nigeria; K = 1 for contiguity;
2 for official language; 3 for non-official language; 4 for colonial master; 5 for the distance.
These are relevant cultural variables in international trade literature (Ghemawat, 2001;
Gallego and Llano, 2013; Christen, 2012).
All the variables are obtained from CEPII GeoDist Database12 (Mayer and Zignago, 2011) and are defined as follows:
Contiguity is a dummy variable that equals one (1) when Country i and Nigeria are contiguous and equal zero (0) otherwise.
11 i=1-15: [i=1 for Benin; i=2 for Burkina Faso; =3 for Cape Verde; 4 for Cote d'Ivoire; =5 for Gambia, The; =6 for Ghana; =7 for Guinea; =8 for Guinea-Bissau; =9 for Liberia; =10 for Mali; =11 for Mauritania; =12 for Niger; =13 for Senegal; =14 for Sierra Leone; =15 for Tog]
12 See (http://www.cepii.fr/anglaisgraph/bdd/distances.htm)
UNIVERSITY OF IBADAN
Official language is a dummy variable that equals one (1) when Country i and Nigeria share a common official language and equal zero otherwise.
Non-Official language is a dummy variable that equals one (1) when there is/are common language(s) spoken by at least 9 per cent of the population in Country i and Nigeria and equal zero otherwise.
Colonial master is a dummy variable that equals one (1) when Country i and Nigeria have had a common colonial master after 1945 and equal zero otherwise.
Distance is a measure of bilateral distances between the biggest cities in Country i and that of Nigeria. Those inter-city distances being weighted by the share of the city in the overall country‟s population13.
The average and the standard deviation of each variable represented by C.k and
C.k
respectively, are used to obtain the standardised values (Cik
) as follows:
15 k C , -C C
C.k
.k ik
ik
(48) To prevent the summation of the standardised values from approaching zero, the highest absolute value of a negative standardised score ( min Cik
) is doubled (2min Cik
) and then
UNIVERSITY OF IBADAN
added to each standardised score (2min Cik Cik
). By this transformation, the minimum value for each variable is the absolute value of the highest negative value before adjustment.
Note that, the smaller the distance measures, the more similar the destination. The interpretation of the other 4 CS variables requires caution as small values imply high dissimilarity. To account for this, the standardised score of these four variables are inverted such that small value of CS can be interpreted as higher CS with Nigeria.
The CS value is then obtained as the average of the 5 variables as follows:
) C C min 2 ( C C min 2
1 5
CS 1
4
1
i5 i5 ik
ik
i
k
(49)
b. Natural characteristics similarity
This is obtained by comparing the standardised score of each of the three groups of natural variables for Nigeria with that of the other 15 West African countries. The three groups are:
Climatic similarity (CLi), Elevation similarity (ELi), and Biomes similarity (BIi).
i. Climatic similarity
The climate similarity (CL ) is measured as the average of nine climate characteristic i variable presented byCLil
;
where i 1-16for each West African Country, including Nigeria; l1-9 for per centage of land area in square kilometre that falls in each of the following nine climatic group. l 1 for tropical with no dry season, over 60mm rain in driest month and animal range temperature of less than 50C; l 2 for tropical monsoon type with short dry season and wet ground all year; l3 for tropical monsoon type with short dry season and wet ground all year annual range temperature less than 5°C; l 4 for tropical distinct dry season with one month of less than 60mm rainfall; l 5for tropical distinct dry season with one month of less than 60mm rainfall and annual range temperature of below 50C; l 6 for temperate winter dry season with at least ten times as much rainfall in
UNIVERSITY OF IBADAN
wettest month as driest month; l 7 for temperate winter dry season with at least ten times and average annual temperature of below 220C in warmest month; l 8 for dry steppe vegetation type of subtropical desert with average temperature of below 180C; and l 9 for desert vegetation type of subtropical desert with any temperature of above 180C.
The mean value (CL.l) and the standard deviation (
l
CL.l
) are used to obtain the standardised values as follows:
l
CL.
. i i
L C -CL CL
l
l l
l
(50)
As for the CS computation, the standard values are adjusted by the highest minimum value to prevent zero sum value. The climate similarity score is thus obtained as:
) L C L C min 2 9 ( CL 1
9
1
i i
i
l
l l
(51)
ii. Elevation similarity
The elevation similarity (EL ) is measured as the average of ten elevation variables i represented by EL ; i
where, i1-16 for each West African country, including Nigeria; J 1-10 for per centage of land area in square kilometre that falls in each of the following elevation range. J 1 for each elevation of below 5 metres; J 2 for between 5 and 10 metres; J 3 for elevation of between 10 and 25 metres; J 4 for elevation between 25 and 50 metres; J 5 for
UNIVERSITY OF IBADAN
10
1
i i
i (2min EL EL )
10 EL 1
J
J J
(52) iii. Biomes similarity
This is calculated as the average of 7 biomes variables represented byBIiL:
Where: i1-16 for each West African country, including Nigeria; L1-7 for per centage land area in each country that falls under the following biome classes: L1 for tropical and subtropical moist broadleaf forests; L2 for tropical and subtropical dry broadleaf forests;
3
L for tropical and subtropical grasslands, savannahs and shrublands; L4 for flooded grasslands and savannahs; L5 for montane grasslands and shrublands; L6 for deserts and xeric shrublands; and L7 for mangroves.
The mean value (BI.L) and standard deviation (
BI.L
) are used to calculate the standardised values as follows:
) I B I B min 2 7 ( B 1
7
1
i i
i
L
L
I L
(53) The average of the above three NVi is calculated as follows:
)/3 BI EL CL (
NVi i i i (54)
The average natural variables (NVi) is converted to NS (NS ) index by finding the absolute i deviation of values between Nigeria and each other country as follows:
16 i , NSi NViNV16
(55)
UNIVERSITY OF IBADAN
c. Facility similarity
This is calculated as the average of the standardised score of 5 facility variables which are defined as FiN;
where i116 for each West African country, including Nigeria and N15 for each measure of tourism facility (UNCTAD, 2007) as follows: N 1 for passenger cars per thousand people; N 2 for motor vehicles per thousand people; N 3 for road paved as a per centage of total road; N 4 for personal computer per hundred people; N 5for mobile and fixed lines telephone subscribers; N 6 for households with television; and
7
N for population covered by mobile cellular network per centage.
The average F.N and standard deviation
F.N
of the variables are used to compute the standardised FiN
as follows:
FN
N iN iN
F F F
.
.
(56) After adjusting for negative values as earlier explained, the average facility score for each country is obtained as:
2min
7 1 7
1
i
iN iN
i F F
S
F
(57)
This average facility score is converted to similarity index as follows: