### Background

According to WB estimates, Ethiopia is one of the fast-growing African economies with an average growth record of 10.6% for the past one and half decades (World Bank Group 2016). Ethiopia’s growth was induced by factors including agricultural modernization; development of new export sector; strong global demand for commodities and Government led development investments (WB 2012). Currently, the Government of Ethiopia is implementing its second phase Growth and Transformation Plan (GTP-II) for the period of 5 years (2015/16–2019/20) with an aim of transforming the country into a manufacturing hub (World Bank Group 2016). The country is the host of the second largest population in Africa with the population size of 99.39 million in 2015, according to the WB population estimates.

China with its second largest economic size is becoming the global game changer with an increasingly important role in Africa’s investment in general and Ethiopia’s FDI inflow in particular. Similarly, India, which was ranked as the 8th largest economy in real GDP terms in 2013 by WB GDP statistics, is one of the fast-changing global economies with its significant role in Africa in general and Ethiopia in particular through its FDI flow and trade relations.

Sino-Ethiopia economic cooperation began in early 1970s but the relationship got momentum after 1995 with the establishment of Joint Ethio-China Commission (Gedion 2014). Ethio-India economic relation began in 1948 but it witnessed a significant progress only after 1991, the year in which economic liberalization was introduced in Ethiopia. The relationship was further strengthened with the establishment of Bilateral Investment Promotion and Protection Agreement (BIPPA) in 2007 (Prateeksha 2015).

FDI does not comprise a major component of external finance inflow to low-income countries, but becoming increasingly important with the rise of China and India (Kinfu et al. 2010*).* FDI is important to sustain high investment rates and is essential for knowledge and technology transfer. Hence, attracting FDI is generally considered as an integral part of the development policy mix of successful emerging economies.

Ethiopia’s cheap and abundant labour, privileged access to high-income markets and growing domestic and regional markets add to its attraction as an FDI host country. But looking at the FDI levels (in % of GDP) currently observed in Ethiopia and in comparison to the successful East Asian countries, it is clear that there is an opportunity to improve the promotion of incoming FDI (WB 2012). FDI as a % of GDP in Ethiopia has been at a relatively low level of 2.0% between 2004 and 2014 somewhat contrasted by 3.9% of GDP in China for years 1991–2010 (Prateeksha 2015).

But the country’s FDI stock has been increasing since 1995 and the country has become one among the best 10 destinations in Africa (Dejene 2016). Currently, it is the 3rd largest FDI recipient in Africa. FDI flow to Ethiopia was $146.6 million in 1997 and yearly FDI inflow has varied between $146.6 million and $977 million between 1997 and 2016. This is in part because the Ethiopian Industrial Strategy is attracting Asian capital to develop its manufacturing base. Indeed FDI in light manufacturing from China, Turkey and India is the major cause of the increase in FDI inflow to Ethiopia (Selamawit Berhe 2015).

In the last few years, India and China have emerged as top foreign investors in Ethiopia and there is an increasing significance of FDI in Ethiopian economy, thus with an increasingly important role for India and China to play in Ethiopia’s development process (Prateeksha 2015).

Despite an agricultural exporter, Ethiopia has attracted significant volume of FDI from China. China’s FDI inflow to Ethiopia averaged at $88.5 million every year for the period 1997–2016 with an average growth at 202%. Similarly, India’s FDI inflow to Ethiopia averaged at $40.9 million every year for the two-decade period with an average growth of 82%. (*Own analysis from Ethiopian Investment Commission data*).

China’s and India’s manufacturing FDI in Ethiopia is still smaller in size than its potential. What policy change can help the Ethiopian Government to attract more Chinese and Indian manufacturing investment and benefit from it is a question of concern. It would be of great importance to know what factors encourage the flow of FDI from the emerging Asian giants and what policy measures could help in this regard. Hence, the objective of this research is to examine the major factors governing the FDI inflow from China and India to Ethiopia.

### Literature on FDI relation of Ethiopia with China and India

FDI is defined as an investment involving a long-time relationship and reflecting a lasting interest and control of a resident entity in an economy other than that of the foreign direct investor. It includes three components such as equity capital; re-invested earnings and intra-company loans (Anne-Lise 2014). It is conventionally defined as a form of internal inter-firm cooperation that involves a significant equity stake in or effective management control of host country enterprises (Qian, et al. 2002, pp. 4).

The main theoretical research on motivation for FDI are the production Cycle Theory by Raymond Vernon; the Internationalization Theory by Stephen Hymer and the Eclectic Paradigm by John Dunning (Anne-Lise 2014).

#### Theories on investment behaviour

The few theoretical models at work that explain investment decision behaviour in literature include the Accelerator, the Liquidity, the Expected Profits also known as the Cash Flow or Tobin’s q and the Neoclassical models of investment behaviour (Mekonnen 2010).The author in his study on private investment in Ethiopia briefly put the theories as follows:

The Accelerator theory of investment explains that investment decision of a firm is determined by changes in demand for its produces and the main implication of the model is that the investment expenditure of an investing firm is proportional to its output while its output is a function of demand (Song et al. 2001, p. 229).

The Liquidity theory of investment behaviour proposes desired capital to be proportional to the internal fund (liquidity) available for investment.

The Expected Profits (Cash Flow) theory of investment explains desired capital as a proportion of the market value of a firm. The model is regarded as a generalization of the Expected Profits model in which investment expenditure is related to the ratio of the market value of business capital assets to the replacement value of those assets. This ratio is known as Tobin’s *q*. According to the model, a value of *q* closer to 1 or greater encourages investment while a lower value of *q* discourages it.

The Neoclassical theory of investment behaviour equates desired capita stock to the value of output deflated by the price of capital services including or excluding capital gains. At the core of this model is the importance of the value of output in influencing investment decisions; thus, it is regarded as a version of the flexible accelerator model (Salahuddin and Islam 2008, p. 21–22).

The theory of multinational enterprises develops its arguments by concentrating on two questions: The issue of internalization, that is, replacement of firms’ external contracts by direct ownership due to market imperfections and the question of location, which is diversity related to the links between flows of goods and factors, that is, to locate the different activities and organizational units in a specific region (Zarotiadis n.d.).

A brief review of relevant empirical literature is done on FDI inflows specifically from China and India to Ethiopia to examine the trends and determinants of the emerging relation.

#### To summarize the overall context of the FDI literature

FDI links from the Asian Giants to Ethiopia are considered a win–win and mutually beneficial by Nazgol (2014); Gebregeorgis (2016); Asayegn (2009); Malancha (2014), etc. On the other hand the links have both opportunities and threats and there are winners and losers in the game in the host country and the outcome depends on the actors and host country’s extra collaborative engagements. These studies include Tegegne (2006), Dawit (2014), and Alemayehu and Atnafu (2011).

The studies identified determinants of FDI flow from China and India to Ethiopia as follows: Trade openness (+ve), Inflation (−ve), Labour cost (+ve), Host country economic size (+ve), Presence of Government SEZ (+ve), Political instability (−ve), Investment climate (+ve), Economic growth (+ve), real GDP growth (+ve), Policy incentive (+ve), Market size (+ve), Returns to Investment (+ve), Infrastructure (+ve), Foreign exchange instability (−ve), Government expenditure (+ve and −ve), GDP PC (−ve), Natural Resource (+ve).

Of the 10 most relevant articles identified for FDI relation of the Asian giants with Ethiopia, none of them used a methodology based on gravity model and panel data frame work for their studies.

### Methodology

The flow of FDI to Ethiopia is analyzed descriptively and comparatively in the first section based on 20-year FDI inflow data obtained from Ethiopian Investment Commission.

The second part of the analysis used gravity model-based fixed effects estimation techniques to examine the major explanatory variables for FDI inflow from China and India to Ethiopia.

#### The gravity model: theoretical formulation for trade and FDI flows

Gravity model is used in explaining the bilateral economic relations between countries for their bilateral trade and FDI flows. It is considered as a common work-horse in international trade and factor flow analysis (Eichengreen 1998).

Many gravity model applications intend to project bilateral trade (and more recently also FDI) relations between countries (Peter 2000; Estrella, Juliette n.d.).

Gravity equation is analysed in the light of a partial equilibrium model of export supply and import demand for the rationalization according to Linnemann (1966). Anderson (1979) also derives the gravity model which proposes identical Cobb–Douglas or constant elasticity of substitution (CES) preference functions for all economies and weakly separable utility functions between traded and non-traded goods and investment flows.

Further rationalization for the gravity model approach is based on the Walrasian general equilibrium model, which states that each country has its own supply and demand functions for all goods. The factor of aggregate income determines the level of demand in the importing country and the level of supply in the exporting country (Oguledo and Macphee 1994). While Anderson’s analysis was at the aggregate level, Bergstrand 1989) developed a microeconomic foundation to the gravity model. He explained that a gravity model is a reduced form equation of a general equilibrium of demand and supply systems. Bergstrand argues that since the reduced form eliminates all endogenous variables out of the explanatory part of each equation, income and prices can also be used as explanatory variables of bilateral trade or FDI. The resulting model is termed a “generalized” gravity equation (Krishna 2002). Eaton and Kortum (1997) have also derived the gravity equation from a Ricardian framework, while Deardorff (1998) derived it from a Hecksher–Ohlin perspective.

Gravity Model of International Trade was initially developed by Tinbergen in 1962 (Hui and Howard 2005). It utilises the gravitational force concept as analogy to explain the volume of trade, capital flows and migrations amongst countries. Newton gravity model states that the interaction between two heavenly bodies is proportional to the product of their masses and inversely related to the distance between them (Dinh et al. 2010).

$$ GF_{ij} = \, M_{i} M_{j} / \, D_{ij} , $$

(1)

where *M* is mass and *D* is distance for countries *i* and *j*.

In the form of natural log:

$$ {\text{Log}}GF_{ij} = { \ln }M_{i} \, + { \ln }M_{j} \, {-}{ \ln }D_{ij} ;\quad {\text{where }}i \ne j. $$

(2)

The Gravity Model of Bilateral Trade (FDI) in its basic form states that *trade(FDI) flow between country i and j is proportional to the product of GDP*_{i}*and GDP*_{j}*and inversely related to the distance between them* (Santos and Silvana 2006).

$$ F_{ij} = \, \alpha_{\text{o}} Y_{i}^{\alpha 1} Y_{j}^{\alpha 2} D_{ij}^{\alpha 3} , $$

(3)

where α_{o}, α_{1}, α_{2} and α_{3} are unknown parameters; *F*_{ij} = Flow of factors.

To account for deviations from the theory, stochastic versions of the equation are used in empirical studies. Typically, the stochastic version of the gravity equation has the form:

$$ F_{ij} \, = \, \alpha_{\text{o}} Y_{i}^{\alpha 1} Y_{j}^{\alpha 2} D_{ij}^{\alpha 3} \eta_{ij} $$

(4)

*η*_{ij} = is an error factor, assumed to be statistically independent of the regressors.

There is a long tradition in the trade/FDI literature of log-linearizing Eq. (4) and estimating the parameters of interest by least squares, using the equation

$$ {\text{Ln }}F_{ij} \, = { \ln }\alpha_{{{\text{o}} + }} \alpha_{ 1} { \ln }Y_{i} + \alpha_{ 2} { \ln }Y_{j} + \, \alpha_{ 3} { \ln }Di_{j} + { \ln }\eta_{ij} . $$

(5)

#### Estimation techniques

Estimation techniques for bilateral Trade and FDI flows *applicable to a panel data frame include* Pooled OLS, Fixed Effect Estimators, Random Effect Estimation, Hausman Test, OLS Estimation of Double log, etc., can be used as an estimation technique (Davidova 2012).

Other methods include Poisson Pseudo-Maximum Likelihood (PPML), Threshold Tobit, Least Square Dummy Variable (LSDV), Generalized Methods of Moments, Classical Minimum Distance Estimation Techniques, and Heckman Estimation Technique.

#### Fixed effect estimator

Fixed Effect approach is preferred here because it has one considerable advantage over random effects estimation. That is, little justification for treating the individual effect as uncorrelated with regressors as is assumed in the random effect model.

Random Effects model may suffer from the inconsistency due to the correlation between the individual variables and the random effect. However, the random effects treatment does allow the model to contain observed time-invariant characteristics such as demographic while fixed effect model does not (Estrella, Juliette n.d.). Fixed Effect estimation assumes that the unobserved heterogeneity component in the regressors is constant overtime.

If *Z* is unobserved, but correlates with Xit, then the least square estimator of *β* is biased and inconsistent as a consequence of an omitted variables. In such instances we use fixed effects estimators:

$$ {\text{Yit}} = {\text{Xit }}\beta + \alpha {\text{i}} + \, \varepsilon {\text{it,}} $$

(6)

where αi = *Z*i’α, embodies all the observable effects and specifies an estimable conditional mean.

This approach takes αi to be a group-specific constant term. It is fixed because the term does not vary over time.

More over this study focuses in indentifying with-in variability of the FDI inflow between a bilateral pair country (Fixed effects estimation) not between the different pair of countries which demands use of random effects estimation.

#### Panel data frame work

Green Econometrics (2002) recommends panel data models compared to cross-section for trade/factor flow analysis. Panel data allow the researcher greater flexibility in modelling difference in behaviour across individual countries.

A panel data framework reveals several advantages over cross-section analysis. It allows to capture the relationships between the relevant variables over a longer period and to identify the role of the overall business cycle phenomenon.

Through a panel approach one is able to disentangle the time-invariant country-specific effects. Above all, one should take into account that the interpretation of the estimated coefficients which is crucially different from that of cross-section analysis. In a panel framework, one checks for cross-section deviations and is thus able to interpret the parameters as elasticity of the influence of independent variables on the dependent one.

#### Empirical investigation of the FDI determinants

The inflow of FDI to Ethiopia from the two emerging Asian economies (China and India) is analysed using Panel data for the years ranging from 1997 to 2016. The data are obtained from domestic institutions such as Ethiopian Investment Commission and international institutions such as IMF, UNCTAD and WB.

The method employed for the analysis of determinants of FDI inflow to the host county is Fixed Effect Estimation. The driving factors for the emerging partners’ increased presence are examined using gravity model.

The dependent variable in our analysis is FDI inflow from China and India to Ethiopia. Explanatory Variables would include demographic, geographic and macroeconomic variables. The regression takes the functional form for some of the regressors as follows:

$$ {\text{Lnfdiflow }} = \, \beta_{ 1} + \, \beta_{ 2} {\text{lnettrad }} + \, \beta_{ 3} {\text{lnfornpopn }} + \, \beta_{ 4} {\text{lnetgrowth }} + \, \beta_{ 5} {\text{lnresource }} + {\text{ Wit,}} $$

(7)

where Lnfdinflow = log of FDI inflow; Lettrade = log of Ethiopia’s bilateral trade; Lnfornpopn = log of Foreign Population; Lnetgrowth-log of Ethiopia’s GDP growth; Lnresource = log of Resource rent in Ethiopia; Wit = the Error term; β_{1} = the intercept term; β_{2}, β_{3}, β_{4}, β_{5} = elasticity of explanatory variables.