Determinants of foreign direct investment in the

African food and agriculture sector

Zaneta Kubik*

Center for Development Research (ZEF)

University of Bonn, Germany

Abstract

In this paper, we analyze the determinants of Foreign Direct Investment (FDI) in the food and agriculture
sector in Africa. Using a unique dataset on the investment projects in African countries between 2003
and 2017, we employ standard econometric analysis as well as spatial econometrics methods to
investigate the main factors affecting the location of FDI in the sector and the underlying business
models. Our results indicate that the investments are primarily driven by the potential that a domestic
emerging consumer class represents, not only in a given destination, but also in neighbouring countries,
suggesting that foreign investment in food and agriculture sector took form of regional trade platforms
throughout most of the analysed period. Among other FDI determinants, we find that supply side
factors, especially agricultural land, as well as physical infrastructure and institutional quality also play an
important role. These findings, apart from giving evidence of how FDI is responsive to new markets and
growing food demand, also suggest ways in which countries or regions can improve their attractiveness
for investment.
Keywords: foreign direct investment; food; agriculture; Africa; new economic geography
JEL codes: E22, E24, F21, F23, Q00, Q13, Q18

*Corresponding author: zkubik@uni-bonn.de

1 Introduction
The food and agriculture sector1 in Sub-Saharan Africa is undergoing a profound transformation. One
important feature of this transformation is the rapidly growing importance of the private sector,
understood here as enterprises, companies or businesses, regardless of size, ownership and structure
(FAO, 2013), in agricultural production, processing and retail. Globally, processes of liberalization and
globalization have shaped the agro-industry, especially since the mid-1980s (von Braun and Díaz-Bonilla,
2008). Thanks to trade liberalization and improvements in logistics, global food trade has doubled in this
time span and spurred investments in food production, processing and retail, both by foreign and
domestic private-sector investors (Reardon et al., 2009). These changes include a growing orientation
towards export markets, especially the food markets of industrial countries. The growth of private sector
investments also led to a consolidation of processing and retail, which induced the so-called
‘supermarket revolution’ and the spread of fast-food chains in many poor countries (Reardon et al.,
2009). A related organizational and institutional change is the rise of vertical coordination via contracts
and market linkage arrangements, as well as private grades and standards (Dolan and Humphrey, 2004;
Reardon et al., 2009; Swinnen and Maertens, 2007).
Although many of these changes have been observed in Africa, much more private sector investment will
be necessary in order to successfully deal with future challenges of providing enough jobs and food for a
rapidly growing African population. The prospects of private sector investment in the food and
agriculture sector in Africa should be considered promising especially considering the great potential
that the sector currently represents. Food demand is projected to grow by about 25 per cent in
developing countries during the next 15 years, with demand growing in Sub-Saharan Africa by 55 per
cent (Alexandratos & Bruinsma, 2012). Growing food demand as well as food system transformation,
both a consequence of rising per capita incomes and urbanization, are expected to result in a shift of
consumption from cereals to high value fresh, processed and convenience foods. The African food
market is projected to triple and reach USD 1 trillion by 2030 (WB, 2013). However, important
investments are necessary to keep pace with this growing demand. Schmidhuber, Bruinsma and
Boedeker (2009) estimate that $45 billion need to be invested annually in food and agriculture sector as
well as rural development in order to achieve ‘zero hunger’ by 2025 in Sub-Saharan Africa alone.
Foreign direct investment (FDI) has the potential to fill this gap, especially where local private sector
investment is insufficient as a result of financing constraints. Moreover, the potential contributions of
FDI to local economies reach much further than just providing capital; it is also expected to create quality
employment (Javorcik, 2012), bring new technologies that increase productivity, improve infrastructure,
and affect domestic investors through spillover effects (Zhan et al., 2018). Yet, FDI, and especially large-
scale investments, continue to raise concerns over market dominance, exclusion of smallholder farmers,
and limited linkages with the local economies in case of export-oriented projects (Zhan et al., 2018;
Karlsson, 2014). In particular, foreign investments involving land acquisition have been criticized for
negatively affecting the rights and livelihoods of local communities, leading to conflicts over resources,
or being motivated by speculative rather than productive objectives (FAO, 2011; Deininger, 2011).
Additionally, the extent to which FDI has welfare-enhancing impacts depends, to a great extent, on the
host country’s degree of openness and its business climate (Moran, 2006).

1
Note that according to FAO terminology (cf. AGROVOC Multilingual Thesaurus), the alternative terms for the food
and agriculture sector include ‘agribusiness’ and ‘agroindustrial sector’. In this paper, following the classification
adopted in our dataset, we additionally use the term ‘food and beverages cluster’. We use these terms
interchangeably.
To this day, FDI into the African food and agriculture sector has remained low compared to other
regions, at barely 10.5% of the world FDI in the sector (Fiedler and Iafrate, 2017). Indeed, there is
evidence of an adverse regional effect for Sub-Saharan Africa in terms of FDI inflows (Assiedu, 2002).
Additionally, several studies indicate that the drivers of FDI to Africa are different from the drivers of FDI
to other regions (Brunetti et al., 1997; Batra et al., 2003). While factors such as natural endowments and
market potential play the most decisive role in attracting FDI to the continent, Assiedu (2006) shows that
in some contexts, factors that are under control of policy-makers are also important. Therefore, several
important initiatives have recently been launched. These aim to create a conducive environment for
private sector investments, in particular in the form of FDI, for sustainable and inclusive growth in Africa.
Most prominent among these initiatives are the Marshall Plan with Africa and the G20 Compact with
Africa, as well as the New Alliance for Food Security and Nutrition and Grow Africa, with the last two
focusing specifically on investment in the food and agriculture sector.
Against this background, we aim to substantiate the debate about FDI in food and agriculture sector with
evidence on location of such investments in Africa. More specifically, we aim to analyze the
determinants of food and agriculture FDI inflows across African countries. This analysis is important for
several reasons. First, the literature focusing on FDI in Africa is still very scant and typically based on a
very limited sample of countries (Assiedu, 2006). Here, we cover all but five African countries. Second,
we focus on the food and agriculture sector specifically. Investment in the food and agriculture sector is
particularly relevant for the Least Developed Countries (LDCs) for two reasons: first, by improving food
availability and quality, it might enhance food security and nutrition, as well as food safety; second, it
might also raise incomes due to its growth and poverty reduction potential (WB, 2007). So far, the
literature on FDI inflows to Africa has not differentiated between sectors; and as such, previous findings
might be strongly influenced by FDI into natural resource sector, and these investments have been
shown to have the potentially negative impact on host economies (Moran, 2006). Additionally, we
employ spatial econometrics methods which enable us to provide evidence of various business models
adopted by the foreign investors.

2 Data

The principal source of data used to analyze the FDI in the African food and beverages sector stems from
fDi Markets, an intelligence unit of the Financial Times2. The dataset comprises cross-border greenfield
investments in all countries and sectors worldwide. Note that contrary to the UNCTAD or OECD datasets
on FDI, only greenfield investments are included in the fDi Markets dataset, while joint ventures are
included only if they lead to a new physical implantation, and mergers and acquisitions (M&A) are
excluded. Only new investment projects and significant expansions of existing projects are included. fDi
Markets collects data from more than 10,000 public sources worldwide, including local newspapers,
investment agencies and companies’ press releases. The databank contains all foreign direct investments
that have been publicly announced. The collection of investment projects can therefore be considered
fairly complete, as it is unlikely that an investment remains completely unnoticed.
An important caveat is that the data includes investment projects that have either been announced or
opened by a company and therefore the details such as the amount of capital invested and the number
of jobs created are mostly based on the investment plans. The data does not convey information on
whether a given project has been realized, when and to which extent, in comparison to the original

2
https://www.fdimarkets.com/; accessed 12.1.2018
3
investment plans. We conducted a reliability test of the dataset by drawing a sample equivalent to 5% of
all reported projects and verifying the data via a Google search. We were able to identify all projects, and
our test confirmed that 76.5% of the projects in this sample were implemented3. Data on the final
amounts invested and number of jobs created was largely unavailable. Even though not fully accurate,
we therefore consider this data to be relatively reliable. In particular, since the funds need to be
budgeted in the companies’ usually well-audited financial plans, we consider financial estimation of the
investment as reliable and we use it throughout our analysis. In addition, we use in the analysis the
number of investment deals as an alternative measure of FDI to investment volume in US$.
Despite these drawbacks, our data also presents an important advantage compared to the official
sources in that it strictly focuses on investments in the real economy. In contrast, standard FDI data
sources widely used in the literature, such as balance of payment data, suffers from inclusion of
investment flows that only represent bookkeeping entries in corporate accounts but no economic
activity (Lipsey & Sjöholm, 2011). Furthermore, fDi Markets data on FDI offers a more comprehensive
definition of the food and beverages cluster4 compared to the ISCO definition of food and beverage
manufacturing. Apart from food and beverage manufacturing, it also comprises crop production and
animal production (equivalent to what would be captured by agricultural sector in the standard
classification), but also other industries and services related to food and beverage sector, such as
machinery, packaging, logistics and distribution or communication. As such, the fDi Markets provides
data on FDI along the whole food and beverage value chain.
In the econometric analysis, we use additional data from different sources. In the country-level analysis,
we control for a set of variables that correspond to various strands of theory. This includes using the
logarithm of GDP per capita and population size as proxies for market potential, size of agricultural land
as a proxy for locational advantage, share of population with access to electricity and landlocked dummy
as proxies for infrastructure and locational characteristics, and finally rate of taxes and contributions and
governance indices, including delivery quality and democratic quality, as proxies for institutional
environment. Table A1 in the Appendix provides details of the data sources.
Note that the cross-sectional analysis is done for three five-year periods, 2003-2007, 2008-2012, and
2013-2017. Ideally, we would apply a panel, rather than cross-sectional, regression. However, the data
for specific variables were not available for the 15 years in all the countries. Considering that most of our
variables are of structural nature, and even in case of amenable factors such as infrastructure or
institutions, significant changes are likely to happen over years rather than on a year-to-year basis, we
preferred to opt for a cross-sectional analysis that allowed us to include relatively high, in comparison
with the literature, number of countries. We had to drop five countries5 because of the missing data. The
econometric analysis covers 48 countries for the period 2003-2007 and 49 countries, including South
Sudan, for the remaining two periods, 2008-2012 and 2013-2017.

3
Note that most of the projects for which we couldn’t confirm their implementation were announced in 2015 and
2016 and are new projects rather than expansions of existing investments, therefore, implementation may still be
underway, or media coverage may be lacking.
4
Note that the original fDi Markets classification also includes tobacco. Since we are interested in food and
beverages only, we deduct FDI inflows in the tobacco sector.
5
The excluded countries are Eritrea, Libya, Niger, Seychelles and Somalia. However, the data on FDI is available for
all countries and therefore the descriptive statistics cover all countries without exceptions.
4
3 Descriptive Statistics

A total of $48.737 billion was invested in the African food and beverages cluster between 2003 and 2017,
according to public announcements of companies’ investment plans6. Almost half of these investments,
amounting to $21.325 billion, was invested in the pesticides, fertilizer and other agro-chemical subsector
(henceforth referred to as the fertilizer sector). Since this sector substantially differs from the remaining
subsectors of the food and agriculture sector in terms of its structure, determinants and impacts, we
exclude it from the principal analysis. Figure 1 shows the change of FDI inflows in the sector over time in
terms of capital investment, number of projects, and number of companies. The figure suggests that FDI
inflows into the African food and agriculture sector broadly followed the worldwide trends in the sector
(cf. Fiedler and Iafrate, 2017). FDI inflows were low in the 2004-2007 period, when the prices of
agricultural commodities were low. This was followed by a rapid increase in 2008 and 2009 and
corresponded to the boom in agricultural commodities, indicating that foreign investors sought to
capitalize on high food prices and high expected returns.

Figure 1. FDI inflows into the African food and beverages cluster 2003-2017 (excluding fertilizer investments)

90 6.000,0

80
5.000,0
70

60 4.000,0

million USD
50
3.000,0
40

30 2.000,0

20
1.000,0
10

,0
2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017

Number of projects Number of companies Capital investment (right axis)

Source: Husmann & Kubik (2019) based on the fDi Markets data (www.fdimarkets.com; accessed January 16, 2018)

6
The dataset contains data for the food, beverages and tobacco cluster. The total amount invested in this cluster
sums up to $49.320 billion. Out of this amount, $584 million, about 1% of the total agricultural FDI inflow, was
invested in tobacco production. As tobacco does not directly contribute to food and nutrition security, we leave it
out and concentrate on the agricultural value chains of food and beverages only.
5
 Figure 2. Location of investment projects (excluding fertilizer) and investments per country

Food and agriculture FDI per country (in million

USD)

Source: Husmann & Kubik (2019) based on the fDi Markets data (www.fdimarkets.com; accessed January 16, 2018)
There was a noticeable surge in FDI inflows to the African food and beverages cluster in 2011.
Afterwards, levels of FDI inflows decreased slightly but remained high overall, especially in terms of the
number of investment projects and the number of companies involved. The singular peak in 2011 FDI
volume is, to a great extent, explained by a single investment project worth almost $2 billion. This was an
investment in palm oil production in Cameroon by Siva Group7. However, following its announcement, it
was impossible to fully trace this investment from official sources8; it is therefore not clear whether and
to which extent the $2 billion project was implemented. Nevertheless, even if we exclude this single
investment, capital investment in 2011 remains the highest over the whole period 2003-2017. We expect
that the reasons behind may be related to structural factors, mainly growth in the demand for food, both
locally and internationally, which, in turn, is a consequence of population growth, urbanization, and

7
Formally, the investing company is Biopalm Energy Ltd registered in Singapore, and operating as a subsidiary of
Siva Group.
8
The latest update dates back to 2014 and refers to a demarcation process of 3,348 ha and 21,552 ha for two pilot
plantation sites (information extracted from Biopalm Energy Ltd letter to Greenpeace). The Memorandum of
Understanding signed with the government of Cameroon in 2011 stipulated that the project would entail 200,000
ha concession.
6
rising incomes. In addition, we speculate that the rise in FDI over recent years could have been triggered
by the launch of New Alliance and Grow Africa, both of which were expected to create conducive
environment for private sector investment in agriculture, intensify partnerships between various
stakeholders and increase private companies’ commitments to invest in African agriculture. Properly
assessing the success of both initiatives is not straightforward9, as their achievements are only reported
in their respective progress reports and have not been evaluated by external independent organizations.
There are considerable regional differences in investment flows. The map in Figure 2 offers a more
detailed picture of where FDI inflows are targeted. In total, we were able to locate 84% of all projects,
representing 88% of the total capital invested10. These are visualized in the bars. Each country’s food and
agriculture FDI summarizes all investments, including those that could not be exactly located. The bars
represent the sum of investments made at a given area, i.e. a city or a village, and may thus summarize
several different investment projects. The highest amount of capital over 2003-2017 was invested in
Nigeria ($3.98 billion), followed by Egypt ($ 2.91 billion), Cameroon ($2.47 billion), South Africa ($2.46
billion), Ghana ($1.88 billion), Angola ($1.48 billion) and Ethiopia ($1.45 billion).
Because of the expected positive impacts of investment in the food and agriculture sector in terms of
food security and poverty reduction, we check whether the sectoral FDI inflows systematically differ
between LDCs and non-LDCs in Africa. We observe that out of the $27.412 billion invested by foreign
companies in the African food and agriculture sector over the 2003-2017 period, only $7.963 billion were
invested in the LDCs; even though LDCs constitute 32 out of the 54 African countries in our sample.
These patterns differ by the origin of the FDI flows: investment in LDCs constitutes 29% of the food and
beverage sector FDI from the Global North and 34% of FDI from the Global South. In contrast, 52% of
sectoral FDI from other African countries go to LDCs.
Nevertheless, there exist a wide disparity among the LDCs themselves. Three LDCs count among the ten
top food and agriculture FDI receiving countries: Angola ($1.48 billion), Ethiopia ($1.45 billion), and
Mozambique ($1.15 billion), and another five LDCs in the second tenth: Tanzania ($0.74 billion), Zambia
($0.74 billion), Liberia ($0.65 billion), Uganda ($0.45 billion) and Madagascar ($0.35 billion). At the same
time, ten LDCs received no food and agriculture FDI at all during the reference period: Benin, Comoros,
Djibouti, Eritrea, The Gambia, Guinea, Lesotho, Niger, Sao Tome and Principe, and Togo.
Whenever available, the fDi dataset provides information on the motives provided by companies for
their investments. These were disclosed for 101 out of the 680 projects. The most cited motive is the
growth potential of the domestic market (58 mentions), followed by a favorable business climate or
regulations (31 mentions) and the proximity to markets or customers (23 mentions). Natural resources
(14 mentions) and infrastructure and logistics (9 mentions) appear to be important for a few investment
projects, while lower costs (6 mentions), financial incentives or tax reasons (2 mentions), government
support (2 mentions) and the presence of suppliers or joint venture partners (2 mentions) appear
relatively unimportant. The availability of a skilled workforce did not have a single mention, indicating
that the level of human capital in Africa is not a deciding factor for FDI inflows. While the information on
markets and motives disclosed by the companies might not be fully reliable, it is generally in line with the
findings of the econometric analysis below.

9
In this paper, we do not aim at assessing the Grow Africa impacts other than on investment volumes. Note,
however, that critical points have been raised, especially regarding the impacts on smallholder farmers and food
security (e.g. De Schutter, 2015; Global Justice Now, 2015; Bergius, 2015).
10
By locating investment projects, we mean finding their (relatively) exact position within a country.

7
4 Theoretical background
In our econometric analysis of the determinants of FDI in the African food and beverages cluster, we
build upon several strands of literature, including Dunning’s eclectic paradigm (Dunning, 1979), the new
theory of trade with contributions from Markusen (1984) and Helpman (1984), institutional theory
(Mudambi and Navarra, 2002; Grosse and Trevino, 2005; Bénassy‐Quéré et al., 2007), as well as some
insights from the new economic geography (Krugman, 1991). Note that while these theories typically
focus on the structural determinants of FDI location, which is relevant to our cross-sectional analysis,
they do not account for the impact of political processes, such as the New Alliance and Grow Africa,
which, as suggested above, may have influenced the changes in FDI inflows over time.
Dunning’s eclectic or OLI (Ownership-Location-Internalization) paradigm (Dunning, 1979) accommodates
a variety of earlier economic theories of the determinants of FDI and the foreign activities of MNEs. It
posits that a firm will engage in productive activities in a foreign country in the presence of competitive
advantages, location advantages and internalization advantages. In our context, location-specific
advantages are of particular interest; they appear when locating in a foreign country provides a firm
access to the country’s natural and created endowments, as well as special tax regimes, lower
production and transportation costs, or important market size, access to protected markets, and lower
risk (Dunning and Lundan, 2008).
Within the framework of the new trade theory, FDIs are typically classified as horizontal and vertical.
Horizontal FDI (Markusen, 1984) is explained by MNEs seeking access to local markets and avoiding costs
related to transportation or protectionist policies; it can be therefore considered a substitute for exports.
In this context, the main determinants of FDI are market size and potential, as well as transportation and
commercial costs (Kinoshita and Campos, 2003). Vertical FDI (Helpman, 1984), on the other hand, results
from differences in factor prices between countries with different endowments. Further developments
of the theory include export platforms (Ekholm et al., 2007), where firms invest in production in a given
country to sell in third countries, especially when trade barriers between the host and third countries are
low; or complex vertical FDI (Baltagi et al., 2007) where a foreign subsidiary is involved in exporting to
third countries for processing before selling in a final destination. These last contributions emphasize the
importance of accounting for possible spatial interactions in the choice of FDI location.
In institutional theory, the role of institutions and the institutional environment is a primary factor
determining where MNEs locate (Bénassy‐Quéré et al., 2007). For example, in the institutional FDI fitness
theory (Wilhelms, 1998), the attractiveness of a given location for FDI inflows is, to a great extent, a
function of amenable institutional characteristics, i.e. policies, laws and their implementation, rather
than the country’s fundamental characteristics, such as population size. In particular, the importance of
government, markets, education and socio-cultural factors is emphasized.
Finally, the theory of new economic geography (Krugman, 1991) focuses on the agglomeration forces in
production, which are explained by the existence of several key elements, namely increasing returns to
scale, monopolistic competition, transportation costs, and technological externalities between
companies. These lead to an uneven distribution of productive activities across space, which is divided
into core and periphery areas. In this context, the perceived demand should have a positive effect on
investment flows towards a given destination, while production costs and the intensity of local
competition should have a negative effect.

8
5 Estimation strategy
Our estimation strategy seeks to establish the determinants of FDI location and scale in the food and
beverages cluster across African countries. In line with the different strands of literature described
above, we distinguish several sets of variables that serve as potential correlates of FDI location, namely
market potential, location advantages, infrastructure, institutional environment, and agglomeration
effects, as in eq. (1):

where is the logarithm of the total value11 of FDI in the food and beverages cluster received by
country i in a reference period; is the market potential of country i, proxied here by the logarithm
of the GDP per capita and population size; are location advantages of country i, proxied by supply-
side factors, such as natural resources, in this case agricultural land; is country i’s infrastructure,
measured by access to electricity and landlocked dummy; is the institutional framework in country
i (rate of taxes and contributions, and governance); finally, corresponds to potential
agglomeration effects, represented here by the lagged value of the dependent variable.
We expect that higher market potential (higher GDP per capita and larger population), better location
advantages (larger areas of available agricultural land), better infrastructure and better institutions
(lower corporate taxes and better regulatory quality) will attract more investments. Also, in line with the
new economic geography (Krugman, 1991), more FDI is likely to be directed towards countries with
higher levels of past investments. Since the fDi Markets dataset tracks the investment at the date of
announcement and not the date at which the capital effectively crossed the border, the dependent
variables – the volume of food and agriculture FDI and the number of projects per country – are
measures of anticipated investments. Therefore, timewise, the control variables correspond to the time
of investment decisions, and not investment realization. This characteristic of our dataset enables us to
avoid the potential problem of reverse causality.
Additionally, as a second step, we apply spatial econometric models to account for potential spatial
dependencies between the destination country and the neighboring countries. More specifically, we test
for the existence of a spatial lag using a Spatial Autoregression Model (SAR) where the spatial
dependence occurs with respect to the dependent variable as in eq. (2), and also a Spatial Durbin Model
(SDM) which accounts in addition for the impact of particular independent variables in neighboring
countries. More specifically, we include the logarithm of GDP per capita and the population size of
neighboring countries to test for the impact of neighboring markets potential on FDI location, and also
the average tariffs rate as a proxy of cross-border trade as in eq. (4):

11
In the analysis, we also use the number of FDI projects as a dependent variable.
9
Finally, we also check the robustness of results by applying Spatial Error Model (SEM) as in equation (4)
and Spatial Durbin Error Model (SDEM) as in equation (5) below:

is the spatial autoregressive term, with W a n x n inverse distance weighting matrix. is a

spatial autoregressive parameter which shows how FDI in neighboring countries affect FDI in a
destination country. If significant, its omission would bias the estimated coefficients (Anselin, 1988). is
a spatial autoregressive parameter which measures how FDI in the destination country is affected by the
shocks in FDI in neighboring countries (Coughlin & Segev, 2000). If significant, its omission would not
necessarily bias the estimated coefficients but would imply that the standard errors are wrong (Anselin,
1988). These three models enable us to distinguish between various FDI motives: horizontal FDI, vertical
FDI, and regional trade platforms (Blonigen et al., 2007).

10
6 Main results

Error! No s'ha trobat l'origen de la referència. and Table 2 present the baseline estimates of the
determinants of FDI in the food and beverage cluster in African countries. Our analysis covers 15 years of
data, split into three periods: 2003-07, 2008-12, and 2013-17. This is a cross-sectional country-level
analysis, and as such, it is exposed to several methodological flaws: first, multi-collinearity of the
independent variables, and second, the omitted variable bias. In order to address these issues, we
additionally employ fixed effects regression based on yearly panel data. However, although this is a
useful way to control for unobserved effects, it has its own limitations. Most importantly, it can
substantially reduce the variation in the explanatory variables, and this is indeed very pronounced in our
analysis, where most of the independent variables change very little over time. This leads to large
standard errors, and this effect is additionally inflated by the small number of observations.
Second, we find some evidence of structural break in the FDI data. We performed the test for multiple
structural breaks in panel data by Ditzen, Karavias, Westerlund (2021); the test confirmed the existence
of a structural breaks in 2007, in line with the descriptive data in Figure 1. The evidence is significant at
1% level in case of FDI deals, but is not significant in case of FDI inflows (see Table A2 in the Appendix). In
addition, we employed the Chow test on the cross-sectional data aggregated over the 5-year periods to
check whether the coefficients differ between these periods. The evidence is significant for FDI deals at
5% level, but is not significant for FDI inflows (not reported here). Since our data only spans 15 years of
time, the evidence of structural breaks prevents us from applying panel regression on split sample. These
considerations led us to choose the cross-sectional analysis as our preferred specification; nevertheless,
we also report the findings from panel regression for reference.
Error! No s'ha trobat l'origen de la referència. presents the results for FDI inflows, measured in millions
of US$; and Table 2 for FDI deals, measured as the number of investment projects. We expect the FDI
inflows measure tot be influenced by single big investments; while the FDI deals data to be more
representative of smaller investments. In addition, data on the number of deals is less likely to be
misreported in the context of the fDi Markets dataset, where the data on capital investment is reported
based on planned investment, as discussed in the data section. We take the logarithm of FDI inflows
which deals with skewed distribution and outliers in our FDI data. Since there are few observations with
no investments, we add one to each observation before the log transformation. FDI inflows are
aggregated over each of the 5-year periods in the cross-sectional analysis in columns 1-3 of Table 1.
Column 4 reports findings based on a pooled cross-section of yearly data, and column 5 presents findings
from the panel fixed effects regression.
In case of FDI deals in Table 2, we apply Poisson regression in columns 1-4; and Poisson regression with
fixed effects in column 5. We observe important dispersion in our count data, especially in the second
(2008-12) and third (2013-17) period. Therefore, we also apply negative binomial regression; but since
the results do not differ much quantitatively, they are not reported here. Another potential issue with
our data is the high number of zeros which might call for using the zero-inflated Poisson regression.
However, looking at our variables of interest, we do not expect that the process leading to zero FDI deals
is different than the process leading to any positive number of FDI deals. In this context, we only present
results of the basic Poisson regression here. Note that the findings in Table 2 are reported in terms of
incidence rate ratios.
The results in Table 1 and Table 2 bring support to several strands of theory and are broadly consistent
across time, albeit with changing magnitude of the impacts. In line with Dunning’s eclectic theorem
(1979) and the new trade theory on horizontal FDI (Markusen 1984), the FDI inflows can be explained by

11
the local market potential among other explanations, represented here by the population size. This is
also in line with the investment motives reported by foreign investors, as described above. However, the
magnitude of the impact is rather small, with an increase in population by one million leading to an
increase in food and beverage cluster FDI inflows by between 2.4 and 3.9% in the cross-sectional
analysis; but the impact is twice higher in case of the panel regression12. The logarithm of GDP per capita,
a proxy of purchasing power, has a negative coefficient in most cases, with the exception of the panel
regression – but is not significant for FDI inflows. On the other hand, the negative impact of the GDP per
capita is statistically significant in case of FDI deals in Table 2 where it substantially decreases the
incidence rate of investment projects; but again, with the exception of the panel data estimate. Even
though this result might be against expectations, it is nevertheless potentially important from the food
insecurity and poverty reduction perspective of recipient countries.
The most prominent finding from Table 1 and Table 2 is the weight of supply-side factors in attracting
foreign investment. In particular, countries’ natural endowments, proxied here by the logarithm of the
size of agricultural land, turn out a significant predictor of FDI inflows into the food and beverage cluster
in Africa. The elasticity of the impact is high, whereby a 1% increase in the agricultural land area leads to
an increase in FDI inflows by between 0.4% to 0.6%, depending on the period. Not surprisingly, the
magnitude of the impact is the highest in 2008-12 period when foreign investors sought to capitalize on
high food prices. The results are similar for FDI deals. While these findings are in line with expectations,
especially that land is essential for investments in agribusiness, whether directly for crop production, or
indirectly, via supply chain linkages, in other segments, they nevertheless might raise questions of long-
run sustainability if foreign investments results in excessive land expansion at the expense of forests; or
where land is appropriated by foreign investors at the expense of local populations.
Among other supply-side factors, infrastructure and geographic location play an important role as FDI
determinants. We proxy infrastructure with electricity, measured here as a share of population with
access to the grid. Availability of reliable sources of energy is crucial for the viability of businesses, and
reduction of production risk and costs; and have been found as an important predictor of FDI flows
(Inglesi-Lotz & Ajmi, 2021). While the impact of electricity access on FDI inflows is significant only in the
2003-07 period; it is significant throughout all years for FDI deals. Alternatively, we also proxied
infrastructure with other variables, including mobile phones subscriptions, share of population with
access to the Internet (both from the WB), and the total used capacity of international bandwidth from
ITU. In particular, the latter provided significant results as well13 (Table A3 in the Appendix).
Landlocked countries have a disadvantage compared to better connected countries; the coefficient of
the landlocked dummy is negative throughout both Table 1 and Table 2, and even though is not
significant for FDI inflows, it is significant and of very high magnitude for FDI deals. Landlocked countries,
because of their remoteness and lack of direct access to ports and maritime routes, are exposed to
higher transport and transit costs which then is expected to negatively affect business operations,
logistic performance and input costs of firms. Alternatively, we also control for the logistic performance
of countries with data on container port throughput from UNCTAD14 and with the logistic performance
indicator (LPI) form the WB15 (Table A3 in the Appendix). Both variables turn out to have a positive
impact on the number of FDI deals, and the magnitude of the impact is substantial: for example, an

12
Note that in order to further account for the role of population size as a determinant of FDI inflows, we also used
the FDI inflows per capita as the dependent variable. However, it did not produce consistent results, and therefore
we do not report the results here.
13
Data on the total used capacity of international bandwidth is available only from 2007 onward.
14
Data on annual container port throughput is only available from 2010 onward.
15
Data on the logistic performance indicator is only available from 2007 onward.
12
increase in LPI by one16 increases the incidence ratio of investment projects by between two to close to
four times, depending on the period.
Finally, we also looked at policy variables, including taxes, regulatory quality and democratic quality. As
expected, higher taxes discourage foreign investors; and the impact is significant for both FDI inflows and
FDI deals. The results on total tax and contribution rate in Table 1 and Table 2 seem to be mostly driven
by labor taxes and contributions, but not by taxes on profits (Table A4 in the Appendix). This is in line
with a common practice of granting various corporate tax incentives for prospective investors.
We also find evidence that quality of governance17 plays a role in attracting FDI, but the impacts differ
between measures of FDI and between periods. In case of FDI inflows, we do not find any significant
results in the cross-sectional analysis, but the impact is very pronounced in the panel regression where
an improvement in democratic quality by one18 almost doubles FDI inflows. Even higher impact of
democratic changes is observed in case of FDI deals. It is therefore important to note that it is not
necessarily countries with better governance compared to other countries, but countries which make
the biggest progress in governance that attract FDI. Alternatively, we also proxied policy and governance
with the Doing Business score from the WB, and in most specifications, we found a significant and
positive impact of improvement in the score on FDI inflows and FDI deals (Table A4 in the Appendix).

Table 1 Determinants of food and beverage cluster FDI inflows in Africa, 2003-17

2003-07 2008-12 2013-17 2003-17 2003-17

Cross-section (5-year totals) Pooled Panel

OLS OLS OLS OLS Fixed effects
FDI inflows (million US$), log
(1) (2) (3) (4) (5)

GDP pc (US$), log -0.421 -0.112 -1.034 -0.0890 0.592

(0.462) (0.581) (0.634) (0.125) (0.594)
Population (million) 0.0390*** 0.0241* 0.0311** 0.0319*** 0.0606**
(0.0125) (0.0136) (0.0117) (0.00321) (0.0260)
Agricultural land (sq. km.), log 0.425** 0.640*** 0.395* 0.220*** -2.557
(0.184) (0.205) (0.199) (0.0427) (2.027)
Electricity (% of population) 0.0376** 0.0164 0.0408 0.0121*** 0.0135
(0.0179) (0.0214) (0.0253) (0.00453) (0.0141)
Landlocked, dummy (1=yes) 0.115 -0.302 -0.619 -0.218
(0.672) (0.812) (0.859) (0.166)

16
For reference, note that the LPI ranges between 1.3 to 3.6 in our sample of countries, with a median and mean of
2.4.
17
We use two indicators, delivery quality and democratic quality. Both are based on the World Governance
Indicators (WGI), and are simple means of selected WGI indicators. Delivery quality is a mean of government
effectiveness, regulatory quality and control of corruption. Democratic quality is a mean of voice and
accountability, political stability and rule of law.
18
For reference, note that democratic quality index ranges between -2.5 to 0.95 in our sample of countries, with a
median and mean of -0.6.
13
Tax (% of profits) -0.00723* -0.00534 -0.0249** -0.00286** -0.00104
(0.00413) (0.00562) (0.0116) (0.00116) (0.00185)
Delivery quality (index) 0.427 -0.258 -0.774 -0.0403 -0.160
(1.221) (1.322) (1.455) (0.268) (0.584)
Democratic quality (index) -0.518 0.0133 0.993 0.345 0.857**
(0.982) (1.187) (1.257) (0.227) (0.392)
Country fixed effects Yes
Year fixed effects Yes Yes
Constant 1.627 -2.642 11.84 -0.803 20.07
(6.396) (7.887) (8.241) (1.654) (23.93)
Number of observations 48 49 49 682 682
R-squared 0.615 0.453 0.457 0.351 0.116

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia.

Table 2 Determinants of food and beverage cluster FDI deals in Africa, 2003-17

2003-07 2008-12 2013-17 2003-17 2003-17

Cross-section (5-year totals) Pooled Panel

Poisson Poisson Poisson Poisson Poisson, fixed
effects
FDI deals (count)
Incidence rate ratios (IRR)
(1) (2) (3) (4) (5)

GDP pc (US$), log 0.662* 0.709*** 0.361*** 0.704*** 1.461

(0.148) (0.0935) (0.0652) (0.0660) (1.040)
Population (million) 1.010*** 1.010*** 1.006*** 1.012*** 1.027
(0.00337) (0.00185) (0.00201) (0.00134) (0.0242)
Agricultural land (sq. km.), log 1.548*** 1.674*** 1.542*** 1.557*** 0.0809
(0.177) (0.101) (0.0951) (0.0632) (0.242)
Electricity (% of population) 1.026*** 1.022*** 1.034*** 1.020*** 0.999
(0.00764) (0.00450) (0.00642) (0.00316) (0.0107)
Landlocked, dummy (1=yes) 0.534* 0.721* 0.259*** 0.520***
(0.198) (0.140) (0.0702) (0.0748)
Tax (% of profits) 0.982** 0.996* 0.940*** 0.989*** 0.998
(0.00870) (0.00223) (0.0101) (0.00296) (0.00333)

14
Delivery quality 1.777 0.669 4.613*** 1.215 0.332*
(1.115) (0.253) (2.177) (0.303) (0.203)
Democratic quality 0.615 2.277*** 0.300*** 1.213 3.394***
(0.314) (0.714) (0.121) (0.248) (1.439)
Country fixed effects Yes
Year fixed effects Yes Yes
Number of observations 48 49 49 682 682
Pseudo R-squared 0.538 0.537 0.615

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia

We also estimate food and beverage cluster FDI from the Global North and the Global South19 separately
in Table A5 and Table A6 in the Appendix to check if there any consistent differences in both types of
investments. This might be relevant considering the growing importance of the FDI flows from the Global
South to the African food and beverage cluster: while in 2003-07 period, they constituted 22% of inflows,
their share rose to 38% in 2008-12 and 37% in 2013-17. Three countries from the Global South figure
among the top ten investors in the food and beverage cluster (excluding fertilizer) in Africa: India (US$
2,897.65 million over 2003-17), South Africa (US$ 1,309.8 million), and Saudi Arabia (US$ 876.91 million).
Other big investors from the Global South include Malaysia (US$ 781.8 million), the United Arab Emirates
(US$ 655.88 million), and China (US$ 415.17 million).
The results in Table A5 and Table A6 suggest that indeed, investment flows from the Global North and
the Global South differ. The Global North FDI flows largely follow the patterns presented above,
especially with respect to the supply side factors, including land availability. On the other hand, for the
Global South FDI, the main factor turns out to be population, reflecting here destination country’s
market potential. Contrary to debates on the land grabs by investors from the Global South, including
Gulf countries and China (see, for example, the discussion in Allan, Keulertz, Sojamo and Warner (eds.),
2012), we do not find evidence that agricultural land is a significant predictor of FDI inflows – but it is
significant in case of FDI deals starting from 2008 onward. Note, however, that the variance of FDI
inflows from the Global South is only half of the variance of FDI inflows from the Global North, which can
affect how precise our estimates are. Finally, the figures in Table A6 give some evidence that investors
from the Global South target countries with higher quality of public service delivery, but lower levels of
democracy.
As a further step, and in line with the new economic geography (Krugman, 1991), we controlled for
agglomeration forces in foreign investment by including the lagged values of independent variables in
Table 3. Due to the lack of available FDI data for the period prior to the year 2003, we had to exclude the
2003-07 period from the cross-sectional analysis, and the 2003 year from the panel analysis. The cross-

19
We acknowledge that the concept of the Global North and the Global South is arbitrary – but nevertheless can be
useful for the purpose of our analysis. The countries included in the Global North are: Australia, Austria, Belgium,
Canada, Croatia, Cyprus, Czech Republic, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Iceland,
Ireland, Israel, Italy, Japan, Latvia, Lithuania, Luxembourg, Malta, Monaco, the Netherlands, New Zealand, Norway,
Poland, Portugal, Romania, Singapore, Slovakia, Slovenia, South Korea, Spain, Sweden, Switzerland, Taiwan, the
United Kingdom, the United States. The remaining countries are included in the Global South.
15
sectional analysis gives clear evidence of agglomeration effects, whereby investments accumulated over
previous years are an important predictor of current investments, whether measured as FDI inflows or
FDI deals. The magnitude of the effect is high: a 1% increase in lagged FDI inflows increases new inflows
by between 0.3% to 0.6%. On the other hand, no such effects are found in the panel estimation. Even
though we might usually expect a strong effect of the lagged dependent variable in the dynamic panel
regression, note that in case of our data, an important number of countries report zero investments over
a number of years. What might matter more for FDI are investments accumulated over longer periods of
time. In most cases, inclusion of the lagged depend variable did not significantly affect the coefficients of
remaining controls (see Table A7 and Table A8 in the Appendix for full estimation results).

Table 3 Determinants of food and beverage cluster FDI in Africa, 2003-17: agglomeration effects

Panel A FDI inflows (million US$), log

2008-12 2013-17 2003-17 2003-17

Cross-section (5-year totals) Pooled sample Panel
OLS OLS OLS Fixed effects
(1) (2) (3) (4)

Lagged FDI inflows (million US$), log 0.595*** 0.496*** 0.293*** 0.0291
(0.175) (0.142) (0.0385) (0.0413)
Remaining controls Yes Yes Yes Yes
Country fixed effects Yes
Year fixed effects Yes Yes
Constant -3.042 10.72 -0.693 19.52
(7.013) (7.292) (1.635) (25.72)
Number of observations 49 49 642 642
R-squared 0.579 0.586 0.408 0.118
Panel B FDI deals (count)

2008-12 2013-17 2003-17 2003-17

Cross-section (5-year totals) Pooled sample Panel
Poisson Poisson Poisson Poisson, fixed
effects
Incidence rate ratios (IRR)
(1) (2) (3) (4)

Lagged FDI deals (count) 1.056*** 1.035*** 1.340*** 1.030

(0.0193) (0.00998) (0.0516) (0.0425)
Remaining controls Yes Yes Yes Yes
Country fixed effects Yes
Year fixed effects Yes Yes
Number of observations 49 49 642 642
Pseudo R-squared 0.551 0.639 0.408 0.118

16
Standard errors in parentheses
* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia. Remaining control variables specified as in Table 1. Full estimation
results in Table A7 and Table A8 in the Appendix.

We employ spatial econometric models to account for potential spatial dependencies between the
destination country and the neighboring countries in Table 4.20 In contrast to the model presented
above, spatial model framework assumes that foreign investment decisions are multilateral, meaning
that the decision to invest in a particular country depends on the decision to invest or not in another
country (Egelink & Elhorst, 2015). As such, spatial models of FDI allow to distinguish between different
business models used by foreign investors: pure horizontal FDI, pure vertical FDI, regional trade
platform, and complex vertical FDI with agglomeration (Hoang & Goujon, 2014). We estimate Spatial
Autoregressive Model (SAR) in columns 1, 3 and 5; and Spatial Durbin Model (SDM) in columns 2, 4 and
6. The former only accounts for spatial lag in dependent variable, the latter additionally controls for the
market potential of neighboring countries21 (GDP pc and population); in addition, the average tariff rate
is added to capture the expected cost of cross-border trade. For reference, we also present the results of
the Spatial Error Model (SEM) and Spatial Durbin Error Model (SDEM) in Table A9 in the Appendix.
The results are not conclusive and provide only weak evidence of regional trade platforms in 2003-07
period, and of complex vertical FDI in 2008-12 period. In the first case, the spatial lag in FDI is negative
(but not significant), and both the logarithm of GDP pc and population in neighboring countries have a
positive impact on FDI inflows in the host country; while the tariff rate in neighboring countries have a
negative impact, in line with expectations. This points to the importance of the market potential of
neighboring countries and suggests that FDI in the host country is used as a platform to also serve nearby
countries. In the second case, the spatial lag is positive and significant which suggests that foreign
investors locate their production processes in different countries to benefit from differences in factor
costs.
We apply the same SAR and SDM models to FDI flows from the Global North and the Global South
separately. We only report the results for FDI from the Global North in Table 5; the results for FDI from
the Global South were largely inconclusive, and spatial terms insignificant; therefore, they are not
reported here. On the other hand, the findings for FDI from the Global North are very consistent for
2003-07 and 2013-17 period, clearly indicating that foreign investment in the food and beverage cluster
in Africa took the form of regional trade platforms. The spatial lag term in columns 2 and 6 is negative
and significant; while the two proxies of market potential in neighboring countries, i.e. GDP pc and
population, are positive and also significant. The Wald test of spatial terms in columns 2 and 6 gives
strong evidence of spatial dependencies between the host country and neighboring countries in case of
FDI flows from the Global North.
Finally, spatial econometric models are known to reduce the bias in the estimated standard error in the
presence of spatial dependency (Anselin, 1998). It is therefore worthwhile that most of variables of
interest discussed before remain significant determinants of FDI flows, even after accounting for spatial
dependency between observations, which confirms the robustness of our results. Note also that our

20
Note that island countries (Cabo Verde, Comoros, Madagascar, Seychelles, and Sao Tome and Principe) are
excluded from the spatial analysis for ease of computing the inverse distance weight matrix.
21
Since our results are based on the inverse distance weight matrix, the term neighboring countries refers to all
nearby countries weighted by the distance to the host country.
17
results are robust to different definitions of the food and beverage sector, i.e. including tobacco or
including fertilizer (results not reported here).

Table 4 Determinants of food and beverage cluster FDI inflows in Africa, 2003-17: spatial models (SAR and SDM)
2003-07 2008-12 2013-17

Cross-section (5-year totals) Cross-section (5-year totals) Cross-section (5-year totals)

SAR SDM SAR SDM SAR SDM
FDI inflows (million US$), log
(1) (2) (3) (4) (5) (6)

GDP pc (US$), log -0.384 -0.265 -0.365 0.0662 -0.997 -0.827

(0.438) (0.413) (0.543) (0.658) (0.611) (0.657)
Population (million) 0.0407*** 0.0456*** 0.0234* 0.0282** 0.0281** 0.0304***
(0.0119) (0.0114) (0.0122) (0.0128) (0.0110) (0.0112)
Agricultural land (sq. km.), log 0.300 0.341 0.457 0.283 0.280 0.0870
(0.242) (0.265) (0.290) (0.366) (0.258) (0.331)
Electricity (% of population) 0.0395** 0.0422** 0.0267 0.0225 0.0379 0.0401
(0.0172) (0.0166) (0.0206) (0.0215) (0.0251) (0.0247)
Landlocked, dummy (1=yes) 0.172 -0.260 -0.549 -0.584 -1.181 -1.613**
(0.647) (0.646) (0.749) (0.806) (0.835) (0.809)
Tax (% of profits) -0.00766* -0.00571 -0.00687 -0.00593 -0.0482*** -0.0510***
(0.00400) (0.00394) (0.00527) (0.00545) (0.0165) (0.0163)
Delivery quality 0.719 0.328 0.0181 -0.444 0.397 0.458
(1.174) (1.136) (1.261) (1.408) (1.443) (1.371)
Democratic quality -0.704 -0.681 -0.260 -0.102 -0.282 -1.158
(0.943) (0.944) (1.094) (1.232) (1.298) (1.329)
Constant 2.318 -0.205 1.709 -2.610 13.94* 12.50
(6.512) (6.388) (8.170) (8.922) (8.350) (8.409)
Spatial lag in FDI inflows 0.107 -0.0786 0.338* 0.294 0.0620 -0.384
(million US$), log
(0.240) (0.289) (0.204) (0.297) (0.232) (0.309)
W.GDP pc (million US$), log 0.308* 0.0350 0.316*
(0.183) (0.220) (0.165)
W.population (million) 0.0903** 0.0256 0.0565*
(0.0424) (0.0401) (0.0337)
W.tariffs (%) -0.408** -0.0573 -0.269
(0.170) (0.227) (0.172)
Wald test of spatial terms: 0.2 8.43 2.76 3.09 0.07 5.82

18
chi square
Prob > chi square 0.66 0.77 0.096 0.542 0.789 0.213
Number of observations 43 43 44 44 44 44
Pseudo R-squared 0.61 0.66 0.45 0.45 0.48 0.58

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Cabo Verde, Comoros, Eritrea, Libya, Madagascar, Niger, Sao Tome and Principe, Seychelles, and Somalia. Results
based on the inverse distance weights matrix. SAR: Spatial Autoregressive Model. SDM: Spatial Durbin Model.

Table 5 Determinants of food and beverage cluster FDI inflows from the Global North in Africa, 2003-17: spatial
models (SAR and SDM)
FDI from the Global North
2003-07 2008-12 2013-17

Cross-section (5-year totals) Cross-section (5-year Cross-section (5-year totals)

totals)
SAR SDM SAR SDM SAR SDM
FDI inflows (million US$), log
(1) (2) (3) (4) (5) (6)

GDP pc (US$), log -0.178 -0.0743 -0.218 0.191 -1.225** -1.488***

(0.417) (0.394) (0.524) (0.633) (0.525) (0.499)
Population (million) 0.0319*** 0.0358*** 0.0248** 0.0304** 0.0275*** 0.0308***
(0.0115) (0.0110) (0.0118) (0.0123) (0.00950) (0.00867)
Agricultural land (sq. km.), log 0.460* 0.492* 0.636** 0.352 0.316 0.269
(0.241) (0.267) (0.276) (0.351) (0.222) (0.256)
Electricity (% of population) 0.0513*** 0.0520*** 0.0363* 0.0333 0.0497** 0.0589***
(0.0162) (0.0158) (0.0199) (0.0205) (0.0216) (0.0191)
Landlocked, dummy (1=yes) 0.866 0.494 0.568 0.379 -0.383 -0.844
(0.604) (0.612) (0.719) (0.769) (0.722) (0.628)
Tax (% of profits) -0.00460 -0.00287 0.0000834 0.00111 -0.0437*** -0.0426***
(0.00379) (0.00374) (0.00508) (0.00521) (0.0141) (0.0125)
Delivery quality -0.252 -0.465 -0.727 -0.724 1.018 1.340
(1.114) (1.080) (1.216) (1.350) (1.252) (1.066)
Democratic quality -0.0524 -0.261 0.671 0.534 -0.956 -1.790*
(0.896) (0.902) (1.059) (1.182) (1.124) (1.033)
Constant -2.891 -5.006 -3.570 -6.744 15.51** 18.40***
(6.287) (6.219) (7.861) (8.539) (7.144) (6.419)
Spatial lag in FDI inflows -0.551* -0.612** 0.221 0.0560 -0.0675 -0.629**
(0.301) (0.306) (0.199) (0.300) (0.239) (0.277)

19
W.log GDP pc (million US$) 0.332** 0.166 0.382***
(0.161) (0.216) (0.126)
W.population (million) 0.0495 0.0320 0.0761***
(0.0372) (0.0393) (0.0268)
W.tariffs (%) -0.399*** -0.122 -0.428***
(0.154) (0.219) (0.138)
Wald test of spatial terms: chi 3.36 11.45 1.24 2.57 0.08 14.66
square
Prob > chi square 3.591 0.022 0.266 0.633 0.778 0.005

Number of observations 43 43 44 44 44 44
Pseudo R-squared 0.59 0.64 0.48 0.48 0.56 0.7

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Cabo Verde, Comoros, Eritrea, Libya, Madagascar, Niger, Sao Tome and Principe, Seychelles, and Somalia. Results
based on the inverse distance weights matrix. SAR: Spatial Autoregressive Model. SDM: Spatial Durbin Model.

Finally, we check if Least Developed Countries (LDCs) have any disadvantage in attracting foreign
investment into the food and beverage cluster due to their serious structural challenges. We therefore
add an LDC dummy to our baseline cross-sectional specification. The figures in Table 6 suggest that LDCs
indeed received significantly less FDI inflows and FDI deals, but only in the 2008-12 period. On the other
hand, the LDC dummy is not significant in two other periods, suggesting that overall, there is no
systematic difference in how attractive LDCs are for foreign investors, all else constant. However, we
expect that the impact of specific variables might be different for LDCs compared to non-LDC countries in
Africa. We therefore conduct a separate analysis for the two groups of countries in Table A10 and Table
A11 in the Appendix. Overall, the results for both groups are largely similar, with the exception of GDP
per capita, which now mostly have a positive sign in case of LDCs, but a negative sign in case of non-LDC
countries.

Table 6 Determinants of food and beverage cluster FDI in Africa, 2003-17: LDCs

2003-07 2008-12 2013-17 2003-07 2008-12 2013-17

FDI inflows (million US$), log FDI deals (count)

Cross-section (5-year totals)
OLS OLS OLS Poisson Poisson Poisson
Incidence rate ratios (IRR)
(1) (2) (3) (4) (5) (6)

GDP pc (US$), log -0.747 -0.721 -1.383** 0.474 0.486 0.251**

(0.511) (0.617) (0.667) (0.242) (0.300) (0.167)

20
Population (million) 0.0373*** 0.0223* 0.0302** 1.038*** 1.023* 1.031**
(0.0124) (0.0130) (0.0116) (0.0129) (0.0133) (0.0119)
Agricultural land (sq. km.), log 0.401** 0.578*** 0.339* 1.493** 1.783*** 1.404*
(0.183) (0.197) (0.199) (0.272) (0.352) (0.280)
Electricity (% of population) 0.0307 0.00278 0.0250 1.031 1.003 1.025
(0.0184) (0.0212) (0.0271) (0.0189) (0.0213) (0.0278)
Landlocked, dummy (1=yes) 0.0392 -0.377 -0.721 1.040 0.686 0.486
(0.666) (0.775) (0.849) (0.692) (0.531) (0.413)
Tax (% of profits) -0.00652 -0.00388 -0.0204* 0.994 0.996 0.980*
(0.00411) (0.00539) (0.0118) (0.00409) (0.00537) (0.0116)
Delivery quality 0.250 -0.548 -0.613 1.284 0.578 0.542
(1.212) (1.267) (1.438) (1.556) (0.732) (0.779)
Democratic quality -0.276 0.454 1.076 0.759 1.575 2.933
(0.985) (1.148) (1.240) (0.748) (1.808) (3.637)
LDC -1.223 -2.297** -1.610 0.294 0.101** 0.200
(0.866) (1.024) (1.087) (0.255) (0.103) (0.217)
Constant 7.443 8.571 19.09*
(7.539) (9.028) (9.483)

Number of observations 48 49 49 48 49 49
R-squared/ Pseudo R-squared 0.634 0.516 0.486 0.634 0.516 0.486

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia. LDC: Least Developed Country.

21
7 Conclusion and implications for policy and research
In this study, we conducted an in-depth analysis of the patterns and drivers of FDI across Africa. Our
analysis shows that a total of $48.737 billion was invested in the African food and agriculture sector by
foreign private-sector investors between 2003 and 2017. While this number is important, it is still not
enough to deal with Africa’s future challenges, especially the task of providing food and jobs for rapidly
growing population. However, positive developments are observed. While the FDI started from a low
base at the beginning of the period analyzed here, it reached much higher levels in recent years. A
noticeable peak in FDI inflows is observed after the 2008/09 agricultural commodities shocks, suggesting
that international investors aim to capitalize on high food prices. Most companies aim to serve the
domestic or regional market, which shows that tapping into the growing African agricultural and
consumer market is increasingly attractive.
Our econometric analysis reveals that indeed, market potential is one of the main drivers of FDI in food
and agriculture sector in Africa. More specifically, population size consistently has a significant impact on
sectoral FDI inflows, irrespective of the model specification. However, the weight of supply-side factors
in attracting foreign investment is also high. In particular, countries’ natural endowments, proxied here
by the logarithm of the size of agricultural land, turn out a significant predictor of FDI inflows into the
food and beverage cluster in Africa. Not surprisingly, the magnitude of the impact is the highest in 2008-
12 period when foreign investors sought to capitalize on high food prices. Agglomeration effects are also
observed, with a lagged volume of FDI inflows having a very strong impact on the level of current FDI.
Finally, we show that amenable factors, i.e. factors over which policy-makers have control, such as
infrastructure or institutional quality, play an essential role in attracting investment. In line with
expectations, there is also a scope for attracting investors with various tax incentives. Overall, these
findings give support to various strands of literature that we drew upon in the theoretical framework.
The spatial econometric analysis provided additional and strong evidence of the business models used by
foreign investors from the Global North. In two of the three periods of the analysis, our findings clearly
show that FDI in the food and beverage cluster took form of regional trade platforms whereby
investment in the host country is used as a platform to also serve nearby countries. In this case, the
decision of foreign investors is multilateral, rather than bilateral, and accounts not only for the host
country factors, but also for the market potential of neighbouring countries. Further, even though
throughout the whole period of the analysis, Least Developed Countries (LDCs) within Africa received
less FDI inflows compared to non-LDC countries, the econometric analysis does not give evidence of a
systematic disadvantage of these countries in attracting foreign investment, with the exception of the
2008-2012 period.
These findings have several implications for policy-makers willing not only to attract FDI into the food
and agriculture sector of their respective countries, but also to make it beneficial to local populations.
First, while FDI location is largely determined by structural factors over which policy-makers have little
control, our analysis showed the importance of amenable factors, such as infrastructure and governance.
Improving these factors will not only encourage FDI inflows, but it will also have the potential to create
conditions under which these investments will benefit the society as a whole. It is essential to involve
multiple stakeholders including local governments, communities and non-governmental organizations in
the processes related to FDI, particularly so in case of big projects, in order to make sure that the
benefits of new investment initiatives will be fairly distributed and that the potential costs will be
reduced. Additionally, an evaluation mechanism needs to be included in all initiatives aimed at
increasing agriculture and food FDI to Africa in order to monitor progress and measure impacts over
time. Finally, more research on the welfare, employment and other socio-economic impacts on
communities located near large FDI projects is necessary.
22
23
References

African Centre for Bioderversity, Rosa Luxemburg Stiftung, 2018. The Green Innovation Centre in Zambia:
Fighting Hunger through Corporate Supply Chains? July 2018.
Ariga, J., Jayne, T.S., Nyoro, J.K., 2006. Factors driving the growth in fertilizer consumption in Kenya,
1990-2005: Sustaining the momentum in Kenya and lessons for broader replicability in Sub-
Saharan Africa. Food Security Collaborative Working Papers.
Ashburner, J.E., Kienzle, J., 2011. Investment in agricultural mechanization in Africa. Conclusions and
recommendations of a Round Table Meeting of Experts. FAO, Rome.
Baltagi, B.H., Egger, P., Pfaffermayr, M., 2007. Estimating models of complex FDI: Are there third-country
effects? Journal of Econometrics, Analysis of spatially dependent data 140, 260–281.
https://doi.org/10.1016/j.jeconom.2006.09.009
Baumgartner, P., von Braun, J., Abebaw, D., Müller, M., 2015. Impacts of Large-scale Land Investments
on Income, Prices, and Employment: Empirical Analyses in Ethiopia. World Development 72,
175–190. https://doi.org/10.1016/j.worlddev.2015.02.017
Baumüller, H., Husmann, C., von Braun, Joachim, 2013. Innovative Business Approaches for the
Reduction of Extreme Poverty and Marginality?, in: von Braun, J., Gatzweiler, F.W. (Eds.),
Marginality. Addressing the Nexus of Poverty, Exclusion and Ecology. Springer, Dordrecht
Heidelberg New York London, pp. 331–352.
Beintema, N.M., Stads, G.-J., 2008. Measuring Agricultural Research Investments. A Revised Global
Picture, ASTI Background Note. IFPRI, Washington D.C.
Bénassy‐Quéré, A., Coupet, M., Mayer, T., 2007. Institutional Determinants of Foreign Direct Investment.
The World Economy 30, 764–782. https://doi.org/10.1111/j.1467-9701.2007.01022.x
Blanas, S., Seric, A., Viegelahn, C., 2017. Jobs, FDI and institutions in sub-Saharan Africa: Evidence from
firm-level data. Department of Policy, Research and Statistics Working Paper 4, 1–52.
Blonigen, B.A., Davies, R.B., Waddell, G.R., Naughton, H.T., 2007. FDI in space: Spatial autoregressive
relationships in foreign direct investment. European Economic Review 51, 1303–1325.
CIESIN/SEDAC, 2011. Population Density Grid Future Estimates, v3: Gridded Population of the World
(GPW), v3 | SEDAC [WWW Document]. URL http://sedac.ciesin.columbia.edu/data/set/gpw-v3-
population-density-future-estimates (accessed 4.4.14).
Cotula, L., Vermeulen, S., 2011. Contexts and Procedures for Farmland Acquisitions in Africa: What
outcomes for local people&quest. Development 54, 40–48.
De Schutter, O., 2015. The New Alliance for Food Security and Nutrition in Africa., in: Directorate-General
for External Policies, European Parliament (Ed.), .
Deaton, A., 2007. Income, Aging, Health and Wellbeing Around the World: Evidence from the Gallup
World Poll, in: Research Findings in the Economics of Aging. National Bureau of Economic
Research, Cambridge, MA, pp. 235–263. https://doi.org/10.3386/w13317
Deininger, K., 2011. Challenges posed by the new wave of farmland investment. The journal of peasant
studies 38, 217–247.
Deininger, K., Byerlee, D., 2012. The rise of large farms in land abundant countries: Do they have a
future? World Development 40, 701–714.
Deininger, K.W., Byerlee, D., 2011. Rising global interest in farmland: can it yield sustainable and
equitable benefits? World Bank Publications.
Depetris Chauvin, N.M., Porto, G.G., 2010. Market Competition in Export Cash Crops and Farm Income.
Available at SSRN 2505320.
Duflo, E., 2011. Women’s empowerment and economic development. National Bureau of Economic
Research.
Dunning, J.H., 1979. Explaining Changing Patterns of International Production: In Defence of the Eclectic
Theory. Oxford Bulletin of Economics and Statistics 41, 269–295. https://doi.org/10.1111/j.1468-
0084.1979.mp41004003.x
Dunning, J.H., Lundan, S.M., 2008. Institutions and the OLI paradigm of the multinational enterprise. Asia
Pac J Manage 25, 573–593. https://doi.org/10.1007/s10490-007-9074-z
Ekholm, K., Forslid, R., Markusen, J.R., 2007. Export-Platform Foreign Direct Investment. Journal of the
European Economic Association 5, 776–795. https://doi.org/10.1162/JEEA.2007.5.4.776
FAO, 2018. FAOSTAT [WWW Document]. FAOSTAT. URL http://www.fao.org/faostat/en/#data/RA
(accessed 11.5.18).
FAO, 2017. FAOSTAT [WWW Document]. FAOSTAT. URL http://www.fao.org/faostat/en/#data/FDI
(accessed 9.8.17).
FAO, 2013. FAO Strategy for Partnership with the Private Sector. Rome, 20 March 2013. URL
http://www.fao.org/docrep/meeting/027/mg048e.pdf
FAO, 2012. The State of Food and Agriculture. Investing in agriculture for a better future. FAO, Rome.
FAO, 2011. Land tenure and international investments in agriculture. A report by the High Level Panel of
Expert on Food Security and Nutrition. July 2011.
Fiedler, Y., Iafrate, M., 2013. Trends in foreign direct investment in food, beverages and tobacco. FAO
Commodity and Trade Policy Research Working Paper No. 51.
Getahun, T.D., 2016. Industrial Clustering, Firm Performance and Employee Welfare: Evidence from the
Ethiopia Shoe and Flower Cluster, Development Economics and Policy Series. Peter Lang,
Frankfurt.
Grosse, R., Trevino, L.J., 2005. New Institutional Economics and FDI Location in Central and Eastern
Europe. MIR: Management International Review 45, 123–145.
Helpman, E., 1984. A Simple Theory of International Trade with Multinational Corporations. Journal of
Political Economy 92, 451–471. https://doi.org/10.1086/261236
ILO, 2018. World Employment and Social Outlook. Trends 2018, World Employment and Social Outlook.
International Labour Organization, Geneva.
International Monetary Fund, 2015. Regional Economic Outlook: Sub-Saharan Africa. Navigating
Headwinds, World Economic and Financial Surveys. International Monetary Fund, Washington
D.C.
Iyanda, O., 1999. The impact of multinational enterprises on employment, training and regional
development in Namibia and Zimbabwe: A preliminary assessment. International Labour Office.
Javorcik, B.S., 2014. Does FDI bring good jobs to host countries? The World Bank Research Observer 30,
74–94.
Karlsson, J., 2014. Challenges and opportunities of foreign investment in developing country agriculture
for sustainable development. FAO Commodity and Trade Policy Research Working Paper No. 48.
Kinoshita, Y., Campos, N.F., 2003. Why Does Fdi Go Where it Goes? New Evidence From the Transition
Economies. International Monetary Fund.
Krugman, P., 1991. Increasing Returns and Economic Geography. Journal of Political Economy 99, 483–
499. https://doi.org/10.1086/261763
Loayza, N.V., Raddatz, C., 2010. The composition of growth matters for poverty alleviation. Journal of
development economics 93, 137–151.
Markusen, J.R., 1984. Multinationals, multi-plant economies, and the gains from trade. Journal of
International Economics 16, 205–226. https://doi.org/10.1016/S0022-1996(84)80001-X
Naseem, A., Spielman, D.J., Omamo, S.W., 2010. Private-sector investment in R&D: a review of policy
options to promote its growth in developing-country agriculture. Agribusiness 26, 143–173.
https://doi.org/10.1002/agr.20221
25
New Alliance for Food Security and Nutrition and Grow Africa, 2015. New Alliance for Food Security and
Nutrition and Grow Africa: Joint Annual Progress Report 2014-2015.
Osgood-Zimmerman, A., Millear, A.I., Stubbs, R.W., Shields, C., Pickering, B.V., Earl, L., Graetz, N., Kinyoki,
D.K., Ray, S.E., Bhatt, S., Browne, A.J., Burstein, R., Cameron, E., Casey, D.C., Deshpande, A.,
Fullman, N., Gething, P.W., Gibson, H.S., Henry, N.J., Herrero, M., Krause, L.K., Letourneau, I.D.,
Levine, A.J., Liu, P.Y., Longbottom, J., Mayala, B.K., Mosser, J.F., Noor, A.M., Pigott, D.M., Piwoz,
E.G., Rao, P., Rawat, R., Reiner, R.C., Smith, D.L., Weiss, D.J., Wiens, K.E., Mokdad, A.H., Lim, S.S.,
Murray, C.J.L., Kassebaum, N.J., Hay, S.I., 2018. Mapping child growth failure in Africa between
2000 and 2015. Nature 555, 41–47. https://doi.org/10.1038/nature25760
Pray, C., Gisselquist, D., Nagarajan, L., 2011. Private investment in agricultural research and technology
transfer in Africa, in: Agricultural R&D: Investing in Africa’s Future. Presented at the ASTI/IFPRI-
FARA Conference, ASTI & FARA, Accra.
Pray, C.E., Fuglie, K.O., Johnson, D.K.N., 2007. Private Agricultural Research, in: Pingali, R.E. and P. (Ed.),
Handbook of Agricultural Economics, Agricultural Development: Farmers, Farm Production and
Farm Markets. Elsevier, pp. 2605–2640.
Reardon, T., Barrett, C.B., Berdegué, J.A., Swinnen, J.F.M., 2009. Agrifood Industry Transformation and
Small Farmers in Developing Countries. World Development, Agrifood Industry Transformation
and Small Farmers in Developing Countries 37, 1717–1727.
https://doi.org/10.1016/j.worlddev.2008.08.023
Sanchez, P.A., 2002. Soil Fertility and Hunger in Africa. Science 295, 2019–2020.
https://doi.org/10.1126/science.1065256
Tait, J., 2015. Project Surnrise. Final report. Oxfam: London.
The World Bank, 2018. World Development Indicators 2018. World Bank, Washington D.C.
von Braun, J., Díaz-Bonilla, E., 2008. Globalization of Agriculture and Food: Causes, Consequences, and
Policy Implications, in: von Braun, J., Díaz-Bonilla, E. (Eds.), Globalization of Food and Agriculture
and the Poor. Oxford University Press, New Delhi, pp. 1–45.
von Braun, J., Kennedy, E., 1994. Agricultural commercialization, economic development, and nutrition.
Johns Hopkins University Press.
von Braun, J., Meinzen-Dick, R.S., 2009. Land grabbing" by foreign investors in developing countries: risks
and opportunities. International Food Policy Research Institute Washington, DC.
Wangwe, S., Diyamett, B., Komba, A., 2009. Trends in research and development in Tanzania: Funding
sources, institutional arrangements, and relevance, in: Graham, M., Woo, J. (Eds.), Fuelling
Economic Growth: The Role of Public-Private Sector Research in Development. IDRC, Ottawa, pp.
159–192.
Wilhelms, S., 1998. Foreign direct investment and its determinants in emerging economies. USAID
African Economic Policy Paper 9.
Zhan, J., Mirza, H., Speller, W., 2018. International Investment and Local Food Security. In IFPRI, 2018.
2018 Global Food Security Report. IFPRI: Washington, DC.

26
 Appendix
Table A1. Variables description and data sources
Variable name Description Source

FDI inflows (million US$) FDI inflows in the food and beverage cluster. fDi Markets, Financial Times
In million US$. Planned investments.
FDI deals (count) FDI deals in the food and beverage cluster. fDi Markets, Financial Times
Number of projects. Planned investments.
GDP pc (US$) Gross Domestic Product per capita. In 2010 FAOSTAT
US$.
Population (million) Population. In millions. FAOSTAT
Agricultural land (sq. km) Agricultural land (land area that is arable, World Development Indicators,
under permanent crops, and under WB
permanent pastures). In square km.
Electricity (% of population) Percentage of population with access to World Development Indicators,
electricity. WB
Landlocked, dummy (1=yes) Dummy equal to one if a country is n.a.
landlocked.
Tax (% of profits) Taxes and contributions (including profit tax, Doing Business, WB
labor taxes and contributions, and other
taxes and contributions) as a percentage of
profits.
Labor tax and contributions (% of Labor tax and contributions as a percentage Doing Business, WB
profits) of profits.
Other taxes and contributions (% of Other taxes and contributions as a Doing Business, WB
profits) percentage of profits
Delivery quality (index) Mean of government effectiveness, World Governance Indicators, WB
regulatory quality and control of corruption
indices.
Democratic quality (index) Mean of voice and accountability, political World Governance Indicators, WB
stability and rule of law.
Internet bandwidth (Mbit/s), log Total used capacity of international International Telecommunication
bandwidth. In megabits per second (Mbit/s). Union (ITU)
Container port throughput (TEU), log Annual container port throughput. In TEU UNCTAD
(Twenty-foot Equivalent Unit).
Logistic Performance Index (index) Index based on six dimensions: efficiency of International LPI, WB
customs and border management clearance;
quality of trade and transport infrastructure;
ease of arranging competitively priced
shipments; competence and quality of
logistics services; ability to track and trace
consignments; frequency with which
shipments reach consignees within
scheduled or expected delivery times.
Doing Business score (index) Index based on ten dimensions: starting a Doing Business, WB
business; dealing with construction permits;
getting electricity; registering property;
getting credit; protecting minority investors;
paying taxes; trading across borders;
enforcing contracts; and resolving
27
 insolvency.
Tariffs (%) Unweighted average of effectively applied World Development Indicators,
rates for all products subject to tariffs WB
calculated for all traded goods.

Table A2. Test for structural breaks in the food and beverage cluster FDI

Test for multiple breaks at unknown break dates

(Ditzen, Karavias & Westerlund, 2021)

H0: no break(s) vs. H1: 1 <= s <= 3 break(s)

Panel A FDI inflows (million US$)

Bai & Perron critical values
Test statistic 1% critical value 5% critical value 10% critical value
W(τ) 6.4 12.37 8.88 7.46
Estimated break points: 2010 & 2012

Panel B FDI deals (count)

Bai & Perron critical values
Test statistic 1% critical value 5% critical value 10% critical value
W(τ) 17.34 12.37 8.88 7.46
Estimated break points: 2007

Source: Authors’ own calculations.

Table A3. Determinants of food and beverage cluster FDI in Africa: infrastructure

Panel A Infrastructure: Internet

FDI inflows (million US$), log FDI deals (count), IRR

2008-12 2013-17 2007-17 2008-12 2013-17 2007-17

Cross-section (5-year Panel Cross-section (5-year Panel

totals) totals)
OLS OLS Fixed Poisson Poisson Poisson,
effects fixed
effects
(1) (2) (3) (4) (5) (6)

Internet bandwidth (Mbit/s), log 0.126 0.647** -0.00501 1.252*** 1.390*** 0.995
(0.249) (0.297) (0.0890) (0.0658) (0.102) (0.0509)
Remaining controls Yes Yes Yes Yes Yes Yes
Country fixed effects Yes Yes

28
Year fixed effects Yes Yes
Constant Yes Yes Yes Yes Yes Yes
Number of observations 49 49 503 49 49 391
R-squared/ Pseudo R-squared 0.457 0.516 0.082 0.569 0.659
Panel B Infrastructure: Ports

FDI inflows (million US$), log FDI deals (count), IRR

2008-12 2013-17 2010-17 2008-12 2013-17 2010-17

Cross-section (5-year Panel Cross-section (5-year Panel

totals) totals)
OLS OLS Fixed Poisson Poisson Poisson,
effects fixed
effects
(1) (2) (3) (4) (5) (6)

Container port throughput (TEU), log 0.0256 0.0606 -0.0479 1.054*** 1.115*** 0.632
(0.0660) (0.0680) (0.549) (0.0195) (0.0286) (0.351)
Remaining controls Yes Yes Yes Yes Yes Yes
Country fixed effects Yes Yes
Year fixed effects Yes Yes
Constant Yes Yes Yes Yes Yes Yes
Number of observations 49 49 383 49 49 273
R-squared/ Pseudo R-squared 0.454 0.461 0.090 0.540 0.622
Panel C Infrastructure: Logistics

FDI inflows (million US$), log FDI deals (count), IRR

2008-12 2013-17 2007-17 2008-12 2013-17 2007-17

Cross-section (5-year Panel Cross-section (5-year Panel

totals) totals)
OLS OLS Fixed Poisson Poisson Poisson,
effects fixed
effects
(1) (2) (3) (4) (5) (6)

Logistic Performance Index (index) 0.656 1.897 0.365 1.936*** 3.667*** 0.633
(1.305) (1.432) (0.565) (0.489) (1.191) (0.318)
Remaining controls Yes Yes Yes Yes Yes Yes
Country fixed effects Yes Yes
Year fixed effects Yes Yes
Constant Yes Yes Yes Yes Yes Yes
Number of observations 44 45 193 44 45 131
R-squared/ Pseudo R-squared 0.518 0.496 0.123 0.545 0.616

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05

29
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia. Remaining control variables in specified as in Table 1, with the
exception of landlocked dummy in Panel B and Panel C. Note that the number of observations varies due to missing
observations or incomplete yearly data. IRR: Incidence rate ratios.

Table A4. Determinants of food and beverage cluster FDI in Africa: policy

Panel A Policy: taxes and contributions

FDI inflows (million US$), log FDI deals (count), IRR

2008-12 2013-17 2003-17 2008-12 2013-17 2003-17

Cross-section (5-year Panel Panel

totals)
OLS OLS Fixed Poisson Poisson Poisson,
effects fixed
effects
(1) (2) (3) (4) (5) (6)

Labor tax and contributions (% of profits) -0.0660 -0.0612 -0.0263 0.979* 0.907*** 0.968
(0.0408) (0.0457) (0.0387) (0.0118) (0.0192) (0.0355)
Profit tax (% of profits) -0.0119 0.00460 0.00718 1.021** 0.935*** 0.991
(0.0323) (0.0386) (0.0159) (0.0105) (0.0224) (0.0218)
Other taxes and contributions (% of -0.00734 -0.029** -0.00218 0.993*** 0.950*** 0.998
profits)
(0.00595) (0.0120) (0.00206) (0.00234) (0.0141) (0.00346)
Remaining controls Yes Yes Yes Yes Yes Yes
Country fixed effects Yes Yes
Year fixed effects Yes Yes
Constant Yes Yes Yes Yes Yes Yes
Number of observations 47 49 559 47 49 437
R-squared/ Pseudo R-squared 0.490 0.485 0.092 0.611 0.625

Panel B Policy: Doing Business

FDI inflows (million US$), log FDI deals (count), IRR

2008-12 2013-17 2003-17 2008-12 2013-17 2003-17

Cross-section (5-year Panel Panel

totals)
OLS OLS Fixed Poisson Poisson Poisson,
effects fixed
effects
(1) (2) (3) (4) (5) (6)

Doing Business Score (index) 0.0349 0.0502 0.0966** 1.032*** 1.045** 1.034
(0.0411) (0.0436) (0.0387) (0.0106) (0.0215) (0.0294)
Remaining controls Yes Yes Yes Yes Yes Yes

30
Country fixed effects Yes Yes
Year fixed effects Yes Yes
Constant Yes Yes Yes Yes Yes Yes
Number of observations 47 49 383 47 49 273
R-squared/ Pseudo R-squared 0.466 0.465 0.091 0.571 0.611

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia. Remaining control variables in specified as in Table 1, with the
exception of total tax in Panel A and delivery quality and democratic quality in Panel B. Note that the number of observations
varies due to missing observations or incomplete yearly data. IRR: Incidence rate ratios.

Table A5. Determinants of food and beverage cluster FDI from the Global North in Africa, 2003-17

FDI from the Global North

2003-07 2008-12 2013-17 2003-07 2008-12 2013-17

FDI inflows (million US$), log FDI deals (count)

Cross-section (5-year totals)
OLS OLS OLS Poisson Poisson Poisson
Incidence rate ratios (IRR)
(1) (2) (3) (4) (5) (6)
GDP pc (US$), log -0.405 0.0171 -1.399** 0.634* 0.753* 0.308***
(0.474) (0.573) (0.565) (0.174) (0.111) (0.0696)
Population (million) 0.0344** 0.0280** 0.0312*** 1.010** 1.009*** 1.005*
(0.0129) (0.0134) (0.0105) (0.00395) (0.00211) (0.00263)
Agricultural land (sq. km.), log 0.373* 0.675*** 0.339* 1.851*** 1.809*** 1.647***
(0.189) (0.202) (0.177) (0.272) (0.130) (0.131)
Electricity (% of population) 0.0526*** 0.0312 0.0584** 1.034*** 1.023*** 1.041***
(0.0184) (0.0210) (0.0226) (0.00965) (0.00503) (0.00810)
Landlocked, dummy (1=yes) 0.653 0.912 0.0681 0.459* 0.624** 0.255***
(0.689) (0.799) (0.765) (0.212) (0.142) (0.0876)
Tax (% of profits) -0.00423 0.00108 -0.0235** 0.982* 0.998 0.935***
(0.00424) (0.00553) (0.0103) (0.0109) (0.00230) (0.0130)
Delivery quality -0.254 -1.074 -0.276 0.811 0.532 4.822***
(1.252) (1.302) (1.297) (0.614) (0.233) (2.906)
Democratic quality -0.145 1.115 0.439 1.088 2.648*** 0.311**
(1.008) (1.169) (1.120) (0.666) (0.958) (0.159)
Constant 0.477 -6.726 16.09**
(6.562) (7.766) (7.342)

31
Number of observations 48 49 49 48 49 49
R-squared 0.562 0.446 0.515 0.554 0.532 0.615
Standard errors in parentheses
* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia.

Table A6. Determinants of food and beverage cluster FDI from the Global South in Africa, 2003-17

FDI from the Global South

2003-07 2008-12 2013-17 2003-07 2008-12 2013-17

FDI inflows (million US$), log FDI deals (count)

Cross-section (5-year totals)
OLS OLS OLS Poisson Poisson Poisson
Incidence rate ratios (IRR)
(1) (2) (3) (4) (5) (6)

GDP pc (US$), log 0.100 -0.324 0.179 0.627 0.535** 0.479**

(0.433) (0.612) (0.584) (0.260) (0.161) (0.146)
Population (million) 0.0246** 0.0299** 0.0388*** 1.009 1.017*** 1.007**
(0.0118) (0.0144) (0.0108) (0.00744) (0.00436) (0.00310)
Agricultural land (sq. km.), log 0.0854 0.242 0.203 1.107 1.293** 1.353***
(0.173) (0.216) (0.183) (0.206) (0.147) (0.134)
Electricity (% of population) -0.00484 -0.000444 -0.00662 1.011 1.020* 1.020*
(0.0168) (0.0225) (0.0233) (0.0139) (0.0105) (0.0107)
Landlocked, dummy (1=yes) -0.748 -1.288 -0.399 0.564 0.981 0.263***
(0.631) (0.855) (0.790) (0.386) (0.404) (0.117)
Tax (% of profits) -0.00340 -0.0107* -0.00919 0.979 0.987 0.949***
(0.00388) (0.00592) (0.0106) (0.0175) (0.00861) (0.0156)
Delivery quality 0.706 0.164 -0.704 9.089* 1.093 4.361*
(1.146) (1.392) (1.339) (10.91) (0.844) (3.303)
Democratic quality -0.528 0.0206 1.330 0.188* 1.704 0.270**
(0.922) (1.249) (1.157) (0.181) (1.107) (0.175)
Constant -1.102 4.657 -2.849
(6.004) (8.304) (7.585)
Number of observations 48 49 49 48 49 49
R-squared 0.284 0.340 0.411 0.235 0.360 0.402

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05

32
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia.

Table A7. Determinants of food and beverage cluster FDI inflows in Africa: agglomeration effects (full estimation
results)

2008-12 2013-17 2003-17 2003-17

Cross-section (5-year totals) Pooled sample Panel

OLS OLS OLS Fixed effects
FDI inflows (million US$), log
(1) (2) (3) (4)

GDP pc (US$), log 0.0989 -0.755 -0.0836 0.638

(0.521) (0.566) (0.124) (0.627)
Population (million) 0.00314 0.0223** 0.0222*** 0.0603**
(0.0136) (0.0107) (0.00336) (0.0282)
Agricultural land (sq. km.), log 0.387* 0.0549 0.161*** -2.610
(0.197) (0.201) (0.0428) (2.192)
Electricity (% of population) -0.00558 0.0200 0.00905** 0.0185
(0.0201) (0.0232) (0.00451) (0.0152)
Landlocked, dummy (1=yes) -0.302 -0.534 -0.123
(0.722) (0.759) (0.165)
Tax (% of profits) -0.000483 -0.0153 -0.00196* -0.00119
(0.00520) (0.0106) (0.00116) (0.00189)
Delivery quality -0.634 -0.365 0.0267 -0.0314
(1.180) (1.292) (0.266) (0.633)
Democratic quality 0.395 0.919 0.238 0.827**
(1.061) (1.112) (0.227) (0.418)
Lagged FDI inflows (million US$), log 0.595*** 0.496*** 0.293*** 0.0291
(0.175) (0.142) (0.0385) (0.0413)
Country fixed effects Yes
Year fixed effects Yes Yes
Constant -3.042 10.72 -0.693 19.52
(7.013) (7.292) (1.635) (25.72)
Number of observations 49 49 642 642
R-squared 0.579 0.586 0.408 0.118

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia.

33
Table A8. Determinants of food and beverage cluster FDI deals in Africa: agglomeration effects (full estimation
results)

2008-12 2013-17 2003-17 2003-17

Pooled sample Panel

Poisson Poisson Poisson Poisson, fixed
effects
FDI deals (count)
Incidence Rate Ratio (IRR)
(1) (2) (3) (4)

GDP pc (US$), log 0.745** 0.371*** 0.920 1.893

(0.0984) (0.0706) (0.114) (1.186)
Population (million) 1.006*** 1.005*** 1.022*** 1.062**
(0.00228) (0.00207) (0.00344) (0.0300)
Agricultural land (sq. km.), log 1.592*** 1.311*** 1.174*** 0.0735
(0.101) (0.0985) (0.0503) (0.161)
Electricity (% of population) 1.017*** 1.026*** 1.009** 1.019
(0.00478) (0.00681) (0.00455) (0.0155)
Landlocked, dummy (1=yes) 0.782 0.325*** 0.885
(0.154) (0.0903) (0.146)
Tax (% of profits) 0.997 0.953*** 0.998* 0.999
(0.00227) (0.0102) (0.00116) (0.00189)
Delivery quality 0.609 2.493* 1.027 0.969
(0.232) (1.268) (0.274) (0.614)
Democratic quality 2.348*** 0.424** 1.269 2.287**
(0.744) (0.179) (0.289) (0.955)
Lagged FDI inflows (million US$), log 1.056*** 1.035*** 1.340*** 1.030
(0.0193) (0.00998) (0.0516) (0.0425)
Country fixed effects Yes
Year fixed effects Yes Yes
Number of observations 49 49 642 642
Pseudo R-squared 0.551 0.639 0.408 0.118

Standard errors in parentheses

* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia.

34
Table A9. Determinants of food and beverage cluster FDI inflows in Africa, 2003-17: spatial models (SEM and
SDEM)

2003-07 2008-12 2013-17

Cross-section (5-year totals) Cross-section (5-year totals) Cross-section (5-year

totals)
SEM SDEM SEM SDEM SEM SDEM
FDI inflows (million US$), log
(1) (2) (3) (4) (5) (6)

GDP pc (US$), log -0.410 -0.246 -0.770 -0.0247 -0.938 -0.635

(0.454) (0.398) (0.645) (0.819) (0.713) (0.679)
Population (million) 0.0386*** 0.0465*** 0.0206* 0.0283** 0.0280** 0.0312***
(0.0124) (0.0113) (0.0120) (0.0134) (0.0112) (0.0110)
Agricultural land (sq. km.), 0.355 0.268 0.626** 0.334 0.312 -0.0132
log
(0.219) (0.277) (0.280) (0.381) (0.230) (0.335)
Electricity (% of population) 0.0413** 0.0414** 0.0359 0.0256 0.0344 0.0354
(0.0182) (0.0166) (0.0223) (0.0236) (0.0297) (0.0266)
Landlocked, dummy (1=yes) 0.211 -0.370 -0.436 -0.413 -1.175 -1.834**
(0.630) (0.685) (0.709) (0.781) (0.841) (0.882)
Tax (% of profits) -0.00775* -0.00562 -0.00866 -0.00632 -0.0471*** -0.0480***
(0.00399) (0.00389) (0.00560) (0.00592) (0.0177) (0.0171)
Delivery quality 0.537 0.492 -0.236 -0.628 0.455 0.598
(1.249) (1.157) (1.238) (1.434) (1.481) (1.367)
Democratic quality -0.609 -0.843 0.0764 0.130 -0.293 -1.282
(0.970) (0.982) (1.147) (1.301) (1.296) (1.286)
Constant 2.166 0.177 6.153 -1.802 13.01 10.66
(6.584) (6.034) (9.963) (10.96) (9.355) (8.349)
Spatial lag in FDI inflows 0.151 -0.230 0.525* 0.286 -0.0171 -0.499
(0.317) (0.416) (0.295) (0.419) (0.375) (0.463)
W.log GDP pc (million US$) 0.339* 0.0935 0.240*
(0.179) (0.232) (0.129)
W.population (million) 0.0925** 0.0317 0.0464
(0.0395) (0.0424) (0.0299)
W.tariffs (%) -0.431*** -0.0600 -0.210
(0.165) (0.256) (0.147)
Wald test of spatial terms: 0.23 9.15 3.17 2.45 0 6.3

35
chi square
Prob > chi square 0.63 0.057 0.075 0.654 0.963 0.178
Number of observations 43 43 44 44 44 44
Pseudo R-squared 0.60 0.66 0.40 0.43 0.48 0.55

Standard errors in
parentheses
* p<0.10 ** p<0.05 **
p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Cabo Verde, Comoros, Eritrea, Libya, Madagascar, Niger, Sao Tome and Principe, Seychelles, and Somalia. Results
based on the inverse distance weights matrix. SEM: Spatial Error Model. SDEM: Spatial Durbin Error Model.

Table A10. Determinants of food and beverage cluster FDI in Africa, 2003-17: LDCs

LDCs
2003-07 2008-12 2013-17 2003-07 2008-12 2013-17

FDI inflows (million US$), log FDI deals (count)

Cross-section (5-year totals)
OLS OLS OLS Poisson Poisson Poisson
Incidence rate ratios (IRR)
(1) (2) (3) (4) (5) (6)

GDP pc (US$), log 1.467 1.607 0.675 1.018 2.085*** 1.961*

(0.999) (1.017) (1.214) (0.906) (0.518) (0.759)
Population (million) 0.0824*** 0.0582** 0.0466* 0.892* 1.034*** 1.019***
(0.0282) (0.0255) (0.0240) (0.0549) (0.00673) (0.00758)
Agricultural land (sq. km.), log 0.0781 0.397 0.0347 14.61** 1.742*** 1.863***
(0.290) (0.305) (0.355) (15.49) (0.279) (0.351)
Electricity (% of population) -0.0331 -0.0714* -0.0342 0.949 0.973* 0.967
(0.0406) (0.0359) (0.0512) (0.0491) (0.0135) (0.0211)
Landlocked, dummy (1=yes) -0.0711 -0.129 -0.495 0.0129* 0.948 0.226***
(0.809) (0.927) (1.106) (0.0331) (0.255) (0.105)
Tax (% of profits) -0.00541 -0.000200 -0.0140 0.762** 0.996 0.920***
(0.00451) (0.00598) (0.0147) (0.0965) (0.00282) (0.0189)
Delivery quality -0.434 -1.117 0.329 49885.1** 0.373 8.593***
(1.723) (1.661) (1.907) (222217.8) (0.301) (6.820)
Democratic quality 0.821 1.237 0.587 0.00000412** 3.534** 0.330
(1.191) (1.336) (1.560) (0.0000218) (1.853) (0.235)
Constant -18.95 -22.16* -5.600

36
 (11.86) (12.72) (14.54)
Number of observations 28 29 29 28 29 29
R-squared/ Pseudo R-squared 0.540 0.576 0.370 0.708 0.585 0.612

Standard errors in
parentheses
* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia.

Table A11. Determinants of food and beverage cluster FDI in Africa, 2003-17: non-LDCs

Non-LDCs
2003-07 2008-12 2013-17 2003-07 2008-12 2013-17

FDI inflows (million US$), log FDI deals (count)

Cross-section (5-year totals)
OLS OLS OLS Poisson Poisson Poisson
Incidence rate ratios (IRR)
(1) (2) (3) (4) (5) (6)

GDP pc (US$), log -1.618** -2.241** -2.683*** 0.0905*** 0.343*** 0.181***

(0.662) (0.781) (0.785) (0.0543) (0.0806) (0.0496)
Population (million) 0.0371** 0.0142 0.0384** 1.030*** 1.010*** 1.007***
(0.0153) (0.0151) (0.0142) (0.00701) (0.00315) (0.00270)
Agricultural land (sq. km.), log 0.166 0.323 0.261 1.503** 1.535*** 1.401***
(0.312) (0.248) (0.252) (0.238) (0.110) (0.108)
Electricity (% of population) 0.0333 0.0360 0.0309 1.021** 1.036*** 1.026***
(0.0240) (0.0291) (0.0370) (0.00967) (0.00813) (0.00843)
Landlocked, dummy (1=yes) 1.058 -1.016 -0.462 2.640 0.636 0.241**
(1.447) (1.266) (1.331) (2.345) (0.300) (0.153)
Tax (% of profits) 0.00819 -0.00628 0.0459 1.039* 0.986 0.945***
(0.0436) (0.0465) (0.0565) (0.0207) (0.0124) (0.0165)
Delivery quality 1.845 -1.080 -3.134 11.31** 1.455 18.48***
(2.322) (2.032) (2.454) (13.55) (0.999) (14.01)
Democratic quality -1.497 0.882 4.358* 0.767 1.471 0.147***
(2.219) (1.929) (2.198) (0.703) (0.845) (0.0878)
Constant 22.14* 32.31** 36.29**
(11.10) (11.95) (12.17)
Number of observations 20 20 20 20 20 20
R-squared/ Pseudo R-squared 0.776 0.703 0.773 0.642 0.628 0.648

Standard errors in parentheses

37
* p<0.10 ** p<0.05 ** p<0.05
Source: Authors’ own calculations. FDI in the pesticides, fertilizers and other agro-chemicals subsector excluded. Excluded
countries: Eritrea, Libya, Niger, Seychelles, and Somalia.

38