Trade and productivity growth in
Ghana: Does Sino-Africa trade make a
difference?
Xiaolan Fu - University of Oxford
Jun Hou - University of Oxford
Pierre Mohnen - University of Maastricht
DILIC Conference, London
2 November 2015
Introduction
 China started minor trade relation with Africa in the early 90s and
intensified these relations quickly and significantly over the following years
• After 1989, China started looking for non-Western countries to boost
economic ties (Tull, 2006).
• China has become the largest trade partner for Africa in 2012, at the
same time, Africa emerges as the major import source and largest
investment destination for China.
 In 2005, China overtook the US as Ghana’s second largest trading partner
after Nigeria.
• Imports of Chinese goods moved from 3.7 per cent share of total
imports in 2000 to 18.16 per cent in 2012
• The total volume of imports from China increased more than ten–fold,
from USD 160.5 million to USD 2.2 billion in this period
Objectives
 Estimate whether the economic integration between China and Ghana has
brought about a productivity improvement for Ghanaian firms
 Compare China-Ghana with OECD-Ghana trade flows to understand
whether Sino-Africa trade made a difference
 Investigate whether industries where Ghana has a comparative advantage
benefit more from internationalization
Literature
 The inflow of foreign capital increases local technological capability and
efficiency, which eventually becomes a driver of TFP growth.
• Technology diffusion (Dunning, 1995; Javorcik, 2004; Narula and
Driffield, 2012)
• Mobility of skilled workers (Cantwell, 1989; Glass and Saggi, 1998 ;
OECD, 2008)
• Competition forces the upgrading of local production efficiency (OECD,
2002)
 Foreign presence can also have negative effects
• Foreign competitors may also crowd out local firms (Kumar and
Pradhan, 2002; Kaplinsky et al., 2007)
• Negative wage spillovers (Görg and Greenaway, 2001)
Literature
 The learning effects would be stronger when trading with countries that
share similar production capabilities
• Technological gap (Dunning, 1977; Dosi, 1990)
• Domestic learning capability constraints (Cohen and Levinthal, 1989)
 The learning effect will be greater in industries which the country has
comparative advantage than those further away from its comparative
advantage.
• Low risk and uncertainty
• Adequate resources
Estimation: computing TFP
 Olley and Pakes (1996), Levinsohn and Petrin (2004), Ackerberg, Caves and
Frazer (2006), Wooldridge (2009), Doraszelski and Jaumandreu (2014)


 3rd order polynomial for


 TFP can be retrieved as:
first-order Markov process
Estimation: the dynamics of TFP
 The impact of trade on productivity
TFPit = a + x TFPit-1 + bea EXita + benEXitn + b f IMPut + b fdi FDI ut + bhh HH ut
+ bc EXI utc + boEXI uto + b pComputc + bqComputo + f Xit + ui + eit
FDI: the proportion of foreign asset, industrial level
HH: Herfindahl index measuring the industry competition
EX, IMP: firm level exports and imports. ‘a’ denotes African countries; ‘n’ is
non-African countries
EXI, Comp: industrial level exports and imports to industry output that firm i
belongs to. the superscripts c and o denote exports to China and OECD
countries, respectively
Methodology issues
 There may be omitted variables problem due to data unavailability: hence
use of a dynamic model including lagged dependent variable
 Bringing industry level data into firm level estimation may result in
downward bias of estimated standard errors of the coefficients: hence
standard errors will be clustered at the industry level
 Possible endogeneity between industrial trade and TFP growth: OLS, FE, and
GMM
Data
 Firm-level: a panel survey comprising manufacturing firms operating in
Ghana collected by the Centre for the Study of African Economies (CSAE) at
University of Oxford.
• 201 firms, 1710 unbalanced panel during 1991-2002, 12 waves
• sectors include food processing, textiles and garments, wood products
and furniture, metal products and machinery
• At least three consecutive years
• Input and output data in 1991 price and trade statistics at firm level
 Industry-level: trade indicators from COMTRADE, which is a set of bilateral
commodity-level trade data for 1991-2002.
• 6-digits commodity level to 4-digits level, SITC Rev. 2
• The importers’ reports as the primary source: imports of Ghana record
the volume from every country in the world, but the exports will be
based on the trading partners reported imports from Ghana rather
than the Ghanaian reported exports whenever possible (Feenstra et
al., 2005)
Variables
Mean
S.D
Min
Max
Real value of manufactured output (1991 firm-specific output 17.27
prices), in logarithm
Imputed replacement value of plant and machinery (deflator 16.13
1991 Cedis, million), in logarithm
2.17
11.49
25.49
3.09
9.54
23.64
Total cost of raw materials (1991 firm-specific output prices), 3.19
in logarithm
Worker
Total number of employees, in logarithm
16.43
Expfirm_Africa
Percentage of output exported within Africa
0.02
Expfirm_nonAfrica Percentage of output exported outside of Africa
0.06
1.39
0.00
7.50
2.16
0.09
0.21
8.92
0.00
0.00
24.49
1.00
1.00
Impfirm
Industry level
FDI
Firm level
Output
Capital
Definition
Material
Herfindahl
Exp_China
Exp_EE
Exp_OECD
Imp_China
Imp_EE
Imp_OECD
Percentage of raw materials imported
0.24
0.36
0.00
1.00
Ratio of total assets owned by foreign firms in total industrial
assets, calculated with sampled firms
The sum of squared shares of firm output/industrial output,
calculated with sampled firms
Industrial level exports volume from Ghana to China, as ratio
of industrial value added
Industrial level exports volume from Ghana to the emerging
economies, as ratio of industrial value added
Industrial level exports volume from Ghana to the OECD
economies, as ratio of industrial value added
Industrial level imports volume from China to Ghana, as ratio
of industrial value added
Industrial level imports volume from the emerging economies
to Ghana, as ratio of industrial value added
Industrial level exports volume from the OECD economies to
Ghana, as ratio of industrial value added
0.41
0.29
0.00
0.91
0.30
0.18
0.09
1.00
0.23
0.43
0.00
4.32
0.40
0.63
0.00
5.56
2.64
6.61
0.00
49.17
0.25
0.06
0.00
0.44
0.12
0.29
0.00
3.26
8.83
11.16
0.00
57.18
Empirical evidence: TFP
 Obtaining firm-level TFP indices
Variables
Worker
Capital
Constant
Observations
R-squared
Number of firm
(1)
OLS
(2)
FE
(3)
LP
0.829***
(0.037)
0.289***
(0.016)
8.915***
(0.183)
0.708***
(0.048)
0.322***
(0.025)
8.792***
(0.339)
0.367***
(0.059)
0.233***
(0.082)
1,710
0.743
201
1,710
0.981
201
1,710
201
Empirical evidence: TFP robustness
 Robustness of firm-level TFP indices
Empirical evidence: China-Ghana and OECD-Ghana
VARIABLES
L.lnTFP (L: lags)
Worker
L. Exp_Africa Firm
L.Exp_nonAfrica F.
Imp_Firm
FDI
Herfindahl
Imp_China
Imp_OECD
Exp_China
Exp_OECD
Constant
R-squared
Observations
OLS
Model 1
Total Sample
FE
Model 2
GMM
Model 3
GMM
Ind. group 1 Ind. group 2
Model 4
Model 5
0.645***
(0.023)
0.016***
(0.002)
0.030
(0.024)
0.006
(0.015)
0.009
(0.007)
0.037**
(0.017)
0.025
(0.019)
-0.003
(0.011)
0.001
(0.001)
0.187***
(0.057)
-0.000
(0.001)
0.735***
(0.069)
0.342***
(0.048)
0.009
(0.008)
0.007
(0.021)
0.031
(0.024)
0.007
(0.010)
0.041**
(0.016)
0.026
(0.019)
0.005
(0.010)
0.001
(0.001)
0.166***
(0.055)
-0.000
(0.001)
1.528***
(0.117)
0.595***
(0.057)
0.019***
(0.004)
0.031*
(0.017)
-0.016
(0.028)
0.009
(0.011)
0.037***
(0.014)
0.024
(0.016)
-0.000
(0.008)
0.001
(0.001)
0.180***
(0.063)
-0.000
(0.001)
0.850***
(0.130)
0.655***
(0.053)
0.016***
(0.003)
0.013
(0.019)
-0.001
(0.018)
0.004
(0.014)
0.037**
(0.017)
0.002
(0.021)
0.043*
(0.026)
0.046*
(0.024)
0.200***
(0.069)
-0.000
(0.002)
0.748***
(0.129)
0.487***
(0.090)
0.025***
(0.007)
0.056*
(0.029)
-0.323
(0.249)
0.005
(0.015)
0.003
(0.113)
0.049
(0.051)
-0.013
(0.008)
0.002*
(0.001)
0.199**
(0.097)
-0.000
(0.001)
1.125***
(0.213)
0.584
1,464
0.123
1,464
1,464
928
536
Empirical evidence: EE-Ghana and OECD-Ghana
VARIABLES
L.lnTFP (L: lags)
Worker
L. Exp_Africa Firm
L.Exp_nonAfrica F.
Imp_Firm
FDI
Herfindahl
Imp_EE
Imp_OECD
Exp_EE
Exp_OECD
Constant
R-squared
Observations
OLS
Model 1
Total Sample
FE
Model 2
GMM
Model 3
GMM
Ind. group 1 Ind. group 2
Model 4
Model 5
0.640***
(0.023)
0.017***
(0.002)
0.029
(0.024)
0.006
(0.015)
0.008
(0.007)
0.029*
(0.017)
0.015
(0.019)
0.012
(0.009)
0.000
(0.001)
-0.006
(0.016)
0.000
(0.001)
0.759***
(0.069)
0.336***
(0.049)
0.012
(0.008)
0.005
(0.022)
0.030
(0.024)
0.006
(0.010)
0.034**
(0.015)
0.017
(0.020)
0.018*
(0.011)
0.001
(0.001)
-0.006
(0.013)
0.000
(0.001)
1.537***
(0.119)
0.576***
(0.058)
0.020***
(0.004)
0.033*
(0.017)
0.001
(0.028)
0.005
(0.011)
0.029**
(0.014)
0.013
(0.017)
0.015*
(0.009)
0.000
(0.001)
-0.008
(0.012)
0.000
(0.001)
0.909***
(0.132)
0.646***
(0.053)
0.016***
(0.003)
0.014
(0.019)
-0.007
(0.020)
0.002
(0.014)
0.031*
(0.018)
0.000
(0.021)
0.028*
(0.016)
0.042*
(0.022)
0.028
(0.057)
0.000
(0.002)
0.778***
(0.127)
0.484***
(0.092)
0.026***
(0.007)
0.053*
(0.031)
-0.369
(0.270)
0.006
(0.016)
0.024
(0.117)
0.043
(0.052)
0.008
(0.013)
0.001
(0.001)
-0.002
(0.015)
0.000
(0.001)
1.153***
(0.220)
0.581
1,464
0.118
1,464
1,464
782
682
Regional-Industrial panel analysis
• Use of the World Bank Enterprise Survey for Ghana of
2006 and 2012
• Compute mean firm level data (616 firms in 2006 and 720
in 2012) for each region and industry
• Do the same for the firm level data from the Centre for the
Study of African Economies (CSAE)
• 11 industries, 4 regions
• panel of 100 observations across 6 waves with at least 3
years apart (1992, 1995, 1999, 2002, 2006, 2012)
• Panel of 237 observations from all available years 19902002, 2006, 2012
Regional-Industrial panel analysis
Summary statistics
Year
lnTFP
Imp_China
Imp_EE
Imp_OECD
Exp_China
Exp_EE
Exp_OECD
1992
1.43
0.13
0.27
3.76
0.03
0.07
10.77
1995
1.52
0.26
0.44
4.67
0.01
0.09
10.64
1999
1.60
0.31
0.50
3.76
0.01
0.15
7.94
2002
1.56
0.80
1.19
6.23
0.05
0.43
10.08
2006
1.58
2.80
15.92
23.11
0.05
0.25
37.44
2012
1.35
1.63
7.07
64.91
0.05
0.99
24.65
Total
1.57
0.70
1.18
3.45
0.07
0.24
5.40
Regional-Industrial panel analysis
6 waves
VARIABLES
Worker
FDI
Herfindahl
Imp_China
China
Model 1
EE
Model 2
China
Model 3
EE
Model 4
0.160***
(0.033)
-0.014***
(0.002)
0.608***
(0.104)
0.095***
(0.018)
0.098***
(0.017)
-0.010***
(0.002)
0.557***
(0.090)
0.072**
(0.035)
-0.010***
(0.002)
0.353***
(0.135)
0.044**
(0.020)
0.051
(0.037)
-0.009***
(0.002)
0.237**
(0.103)
Imp_EE
Imp_OECD
Exp_China
-0.001
(0.001)
-0.404*
(0.241)
Exp_ EE
Exp_OECD
Constant
Observations
13 waves
0.010***
(0.002)
-0.000
(0.000)
-0.000
(0.001)
-0.050
(0.182)
0.010***
(0.002)
-0.000
(0.001)
-0.000
(0.001)
0.436***
(0.168)
0.003
(0.015)
0.001**
(0.000)
0.792***
(0.073)
0.001**
(0.000)
0.994***
(0.205)
0.027
(0.023)
0.001
(0.001)
1.166***
(0.200)
100
100
237
237
Conclusion
 Internationalisation fosters productivity improvement in Ghanaian
manufacturing firms.
 Compared to trade with OECD countries, forming trading activities with
China in general increased the TFP of Ghanaian firms.
 The learning effect is greater in industries which the country has
comparative advantage than those further away from its comparative
advantage
Implications
 The learning effects from trade and internationalisation seem to be stronger
when trading with countries when the technological gap is small.
 Governments of low-income countries should develop more effective
internationalization policies in directing firms to catch up with developing
and developed countries
 The trade partners and industry context are important contingency factors
for latecomers to catch up by engaging in internationalization activities
Thank you!