Productivity Spillovers from Foreign Direct Investment to Ethiopian Manufacturing Sector

Abstract

The study investigates productivity spillover from foreign direct investment (FDI) to domestic firms in the Ethiopian manufacturing industries. The system generalised method of moments (SYS-GMM) estimator using panel data of manufacturing firms for the years 2011–2016 with 12,006 observations grouped under 102 industries was employed. The results show a coexistence of both negative and positive productivity spillover effects from FDI to domestically owned firms at a moderate level of absorptive capacity. Specifically, foreign presence in the industries contributed to a positive and significant horizontal and backward productivity spillover effect on domestic firms on an average level. The horizontal productivity spillover is transmitted to local firms through demonstration and competition effects at a moderate level. Likewise, vertical productivity spillover occurred through the channel of sales of intermediate goods and services to foreign firms. We have observed that the technology gap is a critical factor among those factors that determine the productivity spillover occurrence. Finally, policy measures aimed at minimising the technology gap between foreign and domestic firms to maximise the productivity spillover effect are suggested. Since the result shows that the spillover effects in all firms are not equal, prioritisation as per their promise is recommendable in an FDI-attracting framework.

JEL: C33, D29

Keywords

Foreign direct investment foreign firm domestic firm horizontal productivity spillover vertical productivity spillover labour competition

I. Introduction

Foreign direct investment (FDI) in developing countries has been perceived not solely as an inflow of capital but conjointly as a vehicle for both fast-trendy technology and the necessary managerial skills that these countries need for long-term development. Görg and Greenway (2004) identified three horizontal spillovers (intra-industry) channels through which FDI-induced externalities to domestic firms: (a) Demonstration effect, (b) competition effect and (c) labour mobility effect. In contrast, the occurrence of inter-industry spillovers (vertical spillover) through customer–supplier relationships gives rise to two distinct types of connections between domestic and foreign firms, namely backward and forward linkages (Javorcik, 2004).

To realise this, developing countries have provided several investment incentive packages, such as lowering income taxes, import duty exemptions, tax holidays and subsidies for infrastructure facilities to attract FDI (Glass & Saggi, 2002; Waldkirch & Ofosu, 2010). According to the United Nations conference report on the trade and development of the 2013 year, the share of FDI flows destined for developing countries reached 54 per cent of the world FDI flow compared to 39 per cent for developed countries.

Similarly, the FDI capital flow to Ethiopia shows an increasing trend from birr 971 million in 2000 to birr 63.6 billion in 2016. In the last 4-year period, the flow increased from $1.3 billion in 2013 to $3.2 billion in 2016 in Ethiopia (UNCTAD, 2016). The potential reason for the increment might be the amendment of the investment proclamation in 2014. It brought incentives and subsidies, including business income tax exemptions and import duty exemptions for FDI. In contrast, between 1981 and 2016, Ethiopia’s share of manufacturing value added in gross domestic product (GDP) fluctuated between 3.22 per cent, the lowest recorded in 1992, and 7.8 per cent, the highest recorded in 1997 (UNCTAD, 2016).

The government’s efforts notwithstanding, empirical studies present conflicting findings on the presence of a productivity spillover effect from FDI to domestic firms. For instance, Tomohara and Yokota (2006), Vahter (2004), Ayyagari and Kosová (2010) and Damijan et al. (2003) have documented positive and statistically significant horizontal productivity spillover effects. Likewise, Tomohara and Yokota (2006), Atieno (2015), Boly et al. (2015), Javorcik (2004) and Lenearts and Merlevede (2018) have identified a positive and significant backward spillover effect.

In contrast, empirical literature highlights the adverse impact of FDI on domestic firms in terms of productivity spillover. For instance, Sanfilippo and Seric (2016), Atieno (2015) and Kong (2003) have documented negative horizontal spillover effects, while Di Ubaldo et al. (2018) and Dogan et al. (2017) have pointed out negative backward spillover effects. In a similar vein, Turi (2015) discovered the detrimental consequences of forward productivity spillovers from FDI to domestic enterprises by analysing unbalanced panel data from the large and medium-scale manufacturing sectors in Ethiopia spanning from 2004 to 2010.

Apart from these two extremes, Bruhn and Calegario (2014) documented the coexistence of negative and positive productivity spillover effects from FDI on domestic firms. The potential explanation lies in the fact that spillover effects are conditional on various factors such as absorptive capacity, labour quality, the presence of supportive structures and institutions, market orientation, trade openness, technological gap, financial markets, sectorial competition and firm size (Crespo & Fontoura, 2007; Gachino, 2010; Girma & Gorg, 2005; Görg & Greenaway, 2004; Javorcik, 2004).

The existence of these contradicting empirical studies makes the productivity spillover effects of FDI on domestic firms inconclusive. Furthermore, this study differs from the previous studies undertaken in developing countries’ contexts, especially in Ethiopia, because, for instance, the study by Turi (2015) pooled both domestic and foreign firms for the analysis, which may result in parameter overestimation. Likewise, studies of Negash et al. (2020) and Seyoum et al. (2015) did not investigate the vertical spillover effect and the spillover effect channels, respectively.

Therefore, the objective of this study is to investigate the productivity spillover effect from FDI to domestic manufacturing firms of Ethiopia. Additionally, the study also investigates the impact of labour market competition and the channels through which vertical spillover occurs as a result of foreign firms operating in the domestic manufacturing sector. Lastly, the study analyses the determinants of the horizontal spillover effect.

This study makes four contributions to the current body of literature and policy-making. First, by investigating the impact of foreign firms on domestic manufacturing productivity, this study adds empirical evidence to the literature on productivity spillovers. The findings will contribute to a better understanding of whether and to what extent foreign presence leads to positive outcomes for domestic firms, thereby filling a gap in the current knowledge base. Second, by analysing the impact of labour market competition and exploring the channels of vertical spillover (backward and forward), the study sheds light on the intricate dynamics of the labour market within the manufacturing sector. Understanding how labour market competition influences productivity spillovers can inform policy interventions aimed at fostering a more competitive and productive workforce. Third, by examining the determinants of horizontal spillover effects, such as the demonstration effect, competition effect and labour mobility effect, the study provides valuable insights into the mechanisms through which knowledge and technology transfer occur among domestic firms. This understanding can guide policymakers in designing targeted interventions to enhance horizontal spillovers and promote knowledge diffusion within the domestic manufacturing sector.

Lastly, the study offers an opportunity for policymakers to critically evaluate the effectiveness of government policies aimed at attracting FDI and facilitating the transfer of productivity-enhancing technologies. By assessing whether such policies have resulted in tangible productivity spillovers for domestic firms, policymakers can refine existing strategies or develop new initiatives to maximise the benefits of FDI for the domestic economy.

The study is structured as follows. Section II provides the theoretical and empirical literature reviews. Section III outlines the materials and methodology employed. Section IV describes the results of the dynamic panel data model and dictates the results. Finally, Section V presents the conclusions.

II. Literature Review

Theoretical Literature

A foreign investment can be a direct investment or a portfolio investment. Direct investment is the acquisition or construction of an actual capital asset by an enterprise from a source country in the host country. Thus, FDI is an investment involving a long-term relationship and controlled by a resident entity of one country located in a country other than the investment country (Duce & España, 2003). Based on the definition of IMF (1993) and OECD (1996), foreign firms are defined as a firm in which there is at least 10 per cent foreign equity, and otherwise, they are considered domestic.

According to Wei and Liu (2001), the presence of multinational firms has the potential to accelerate and lower the cost of technology transfer. Competition by multinationals may encourage local firms to innovate and operate more efficiently. As mentioned earlier, competition, demonstration, learning-by-imitation, contagion effects and the training of workers by foreign firms may help to facilitate the speed of the transfer of technology to domestic firms. Görg and Grееnаwау (2004) also theorised possible mechanisms through which spillover may occur, such as imitation, worker mobility, competition and linkages. However, a theory developed by Gachino (2010) states that spillover effect occurrences have a strong relationship with the potential of hosting enterprises to absorb know-how, skills and technologies.

In contrast, Gorodnichenko et al. (2014) argued that foreign companies might have negative effects on domestic firms’ output and potency if they take over their market or take over their best competent workers. If the experienced workers leave for foreign companies, potency within the domestic firms declines, which eventually affects the productivity of the domestic companies.

Vertical Productivity Spillovers

When inter-industry spillovers are primarily the result of a customer–supplier relationship, there exist two types of linkage between the domestic and foreign firms, that is, backward and forward linkage. The backward linkage occurs when the local supplier firms have to meet the demands of the foreign firm in the form of higher quality, price and delivery standards (Javorcik, 2004). When these local firms supply certain raw materials, the high quality, reliability and speed of delivery that multinational corporation (MNC) affiliates demand forces them to enhance productivity. In some cases, foreign firms deliver technical and managerial training in the production of required inputs, called the backward spillover effect.

Horizontal Spillovers

According to Görg and Greenway (2004), domestic firms can benefit from horizontal spillovers through three channels. First, via demonstration effects, the local enterprises become familiar with superior technologies, marketing and managerial practices used by their foreign affiliates. Thus, spillovers can take place in the form of imitating the foreign subsidiaries’ technology (Sasidharan, 2006). Second, labour turnover occurs when employees from foreign affiliates leave multinationals to join local firms. The last channel of transmission is the competition effect, which occurs when the presence of a foreign firm exerts pressure on local firms to adopt more efficient methods (Sasidharan, 2006).

Empirical Evidence on FDI Productivity Spillover

Assessing the effect of productivity spillovers is becoming an essential device in designing country-wide regulations regarding inward FDI. Thus, a considerable number of empirical studies have attempted to examine the impact of productivity spillover effects from foreign-owned firms (FDI) on domestic firms and have found contradictory results. The first strand of empirical studies shows a positive impact of FDI on the productivity of domestic firms. For instance, Tomohara and Yokota (2006) reported a positive productivity spillover effect of FDI to domestic firms through horizontal and backward channels on average using plant-level data in Thailand. Similarly, Vahter (2004), Ayyagari and Kosová (2010) and Damijan et al. (2003) found that FDI had a positive and significant horizontal productivity spillover effect on local firms. On the other hand, Atieno (2015), Boly et al. (2015), Javorcik (2004) and Lenearts and Merlevede (2017) found positive and significant backward spillover effects.

Moreover, Turi (2015) studies spillover effects resulting from FDI with a focus on the manufacturing firms in Ethiopia based on the Central Statistics Agency’s (CSA) survey for the years 2004 up to 2010 and finds econometric evidence for positive backward spillovers and negative forward spillovers to the total productivity of the manufacturing firms in the country. In this study, the author used both domestic and foreign firms for the analysis. This may cause an overestimation of the parameters. Negash et al. (2020) utilise firm-level data from Ethiopia and find that Chinese firms were more productive than local firms and that their presence may bring positive potential spillover effects for domestic firms. This study did not address the inter-industry (vertical) spillover. The other shortcoming of this study is that we cannot be sure of the horizontal spillover reported because the spillover may occur from other firms owned by non-Chinese that were not controlled in the model. Seyoum et al. (2015) also conducted a study using firm survey data of 1,033 manufacturing firms operating in Ethiopia in 2011 and found that domestic firms with higher absorptive capacity experienced positive spillovers. The channels through which the spillover effect happened were not addressed in this study. Thus, a clear policy recommendation was not forwarded.

In contrast, the second strand of empirical work found a negative productivity spillover effect from FDI to local firms. For example, Sanfilippo and Seric (2016) scrutinised the spillover effect of FDI on domestic firm productivity in sub-Saharan Africa and found negative and significant horizontal spillover effects stemming from subsidiaries’ taking away of the domestic firm market shares by FDI. Atieno (2015) also claimed a negative horizontal productivity spillover effect from FDI on Kenyan manufacturing firms. There are also a considerable number of studies that find negative empirical evidence on backward spillover effects (Di Ubaldo et al., 2018; Dogan et al., 2017). Turi (2015) reported the existence of negative forward productivity spillovers from FDI to domestic firms using unbalanced panel data of large and medium-scale manufacturing industries in Ethiopia for a period between 2004 and 2010.

Employing moderated multiple regression (MMR) and a generalised linear model (GLM) on panel data sets, Bruhn and Calegario (2014) documented the coexistence of negative and positive productivity spillover effects from FDI to domestic firms. The negative impact concerns the firms that are performing below the average level of the technological gap between domestic and foreign firms, whereas the positive signals concern the firms that have a moderate and above-average technological gap. Apart from these, there are a few studies that miss the mark in figuring out a significant effect (Damijan et al., 2003).

On the other hand, Crespo and Fontoura (2007), Schoors and Van der Tol (2002), Javorcik (2004), Gachino (2010), Görg and Greenaway (2004), Girma and Gorg (2005), Kokko et al. (1996), Blomstrom (1986), Mugendi (2014) and Bruhn and Calegario (2014) conducted a study to identify the determinants of spillover effects in different contexts. They documented the absorptive capacity, quality of labour, presence of supportive structures and institutions, presence of interactions, firms’ market orientation (import and export), sectoral competition and size of the firm as determinants of productivity spillover effects. The positive spillover effects also depend on the host country’s openness to trade, the existence of relatively developed local financial markets, the technological gap between foreign and local firms and the ability of sectors to support learning (De Mello, 1999; Görg & Greenaway, 2004).

Conceptual Framework

For a more comprehensive understanding of the phenomenon of productivity spillover from FDI to local firms, it is advisable to conceptualise the relationship between the characteristics of local firms in terms of absorptive capacity and the status of foreign firms in the host countries. The theoretical framework of this study is based on the sensitivity of domestic firms to absorb spillover effects, commonly referred to as conditional spillover. Previous research conducted by Schoors and Van der Tol (2002), Javorcik (2004), Gachino (2010) and Görg and Greenaway (2004) has identified several key factors that influence the spillover-absorptive capacity of a firm or an industry, including sectoral competition, firm size, foreign participation and export orientation.

The study by Kokko et al. (1996) and Blomstrom (1986) indicated that when there is a minimal technological difference between domestic and foreign companies, the spillover of productivity from FDI has a favourable influence on domestic firms. Additionally, Kokko (1994), Blomström and Kokko (1998), Gachino (2010) and Görg and Greenaway (2004) have put forth arguments highlighting the significance of the absorptive capacity of local firms in determining the occurrence of spillover effects.

Conversely, the absorptive capacity of local firms can be affected by various factors, which can be classified into two main categories: Internal and external factors. Internal factors encompass the quality of labour, which can be assessed through variables like the disparity in skill intensity and ownership ratio, as well as the discrepancy in capital intensity and embodied technology between domestic and foreign firms. Additionally, other internal factors comprise the market orientation of firms and their organisational capabilities, such as the utilisation of skilled labour, accumulated experience, size (in terms of financial resources and workforce) and involvement in export activities. These factors have been recognised in existing literature as significant determinants of absorptive capacity (Blomström & Kokko, 1998; Kokko, 1994).

Moreover, the horizontal and vertical productivity spillover occurs when the domestic and foreign firms engaged in the exchange of inputs or finished materials, either through procurement or sales.

III. Data and Method

Data Description

This article relied on an annual survey of large and medium-scale manufacturing firms and industrial data, conducted by the Ethiopian Central Statistical Agency from 2011 to 2016. CSA has employed stratified sampling methods in each region and sub-sector based on International Industrial Classification version 3.1. The nature of the data set was unbalanced panel data that covered 12,006 domestic and 2,550 foreign firms, with a total number of 14,556 observations at the firm level. By aggregating the firms under 102 industries, a data set with 443 observations at the industry level was also used to conduct vertical spillover effect analysis by assuming that spillovers to firms categorised under the same industry are the same.

Econometric Model Specification

Productivity spillovers from FDI can lead to improvements in efficiency, technology adoption and managerial practices in local manufacturing firms (Blomström & Kokko, 1998). Through partnerships, collaborations and knowledge diffusion, local firms can leverage FDI-induced technological advancements to develop new products, processes and services (Kokko et al., 1996). Enhanced productivity levels can increase the competitiveness of domestic manufacturers both domestically and in international markets, driving export growth and attracting further investment (Gachino, 2010).

Endogenous growth theory posits that economic growth is driven by factors such as knowledge accumulation, increasing returns to scale and the role of government in promoting innovation and growth (Acemoglu, 2009; Romer, 1986, 1990). In the context of studying productivity spillovers from foreign firms to domestic manufacturing firms, endogenous growth theory offers several advantages. First, it recognises that productivity spillovers can arise from interactions between firms and are not solely dependent on exogenous factors (Romer, 1994). For instance, FDI can lead to productivity gains through various channels, including technological diffusion, human capital accumulation and increased competition (Aitken & Harrison, 1999; Bloom et al., 2013; Haskel & Slaughter, 2002). The presence of foreign firms can enhance the skills and knowledge of domestic workers, leading to productivity gains (Bloom et al., 2013). Moreover, foreign firms may introduce new technologies and managerial practices that can diffuse to domestic firms, further enhancing productivity (Keller, 2004).

Second, endogenous growth theory emphasises the role of innovation and knowledge accumulation in driving sustained economic growth (Romer, 1986, 1990). Productivity spillovers under this framework are expected to be more persistent and long-term due to the ongoing accumulation of knowledge within the economy (Acemoglu, 2009). Finally, endogenous growth theory highlights the importance of policies and institutions in fostering innovation and growth (Acemoglu, 2009). Policies that encourage FDI, promote education and research and development (R&D) and enhance competition can amplify productivity spillovers and contribute to overall economic development. Therefore, given its focus on endogenous factors driving innovation, knowledge accumulation and the role of policies, endogenous growth theory is chosen as the framework to study productivity spillovers from foreign firms to domestic manufacturing firms.

This study follows the econometric specification adopted by several scholars, including Вuсklеу et al. (2010), Наlе and Lоng (2011), Nhamo (2011), Xu and Sheng (2011), Pham (2016) and Negara and Adam (2012). The firm-level data analysis in this study is conducted using a dynamic panel data model as a foundation.

{VAD}_{i j t} = γ V A D_{i j t - 1} + Q_{1} K_{i j t} + Q_{2} L_{i j t} + Q_{3} {MR}_{i j t} + Q_{4} {FP}_{i j t} + Q_{5} {EXT}_{i j t} + Q_{6} {SZ}_{i j t} + Q_{7} {TG}_{i j t} + Q_{8} {CI}_{i j t} + Q_{9} {SI}_{i j t} + Q_{10} {Horz}_{i j t} \times {TG}_{i j t} + Q_{11} {Horz}_{i j t} \times {CI}_{i j t} + Q_{12} {Horz}_{i j t} \times {SI}_{i j t} + Q_{13} {Horz}_{i j t} + Q_{14} {HEF}_{i j t} + α_{i} + ε_{i j t}

(1)

Where, i denotes the firm (i = 1, 2, …, 12,006), t denotes the period (t = 2011, …, 2016) and j (j = 1, 2…102) denotes the industry in which the firms exist; also, assuming that N is large and T is small (short panel). VAD _ijt is the observation on the dependent variable for firm i in industry j and period t and denotes the natural logarithm of labour productivity of firms; VAD _ijt _–1 is the natural logarithm of one lag period home firm productivity, K _ijt is the fixed asset intensity, L _ijt is the labour input, MR _ijt is the input materials intensity, FP_ijt is the foreign firm presence, EXT _ijt is export of firm, SZ _ijt is the size of the firm, TG _ijt is a technological gap between foreign firm and domestic firm, CI _ijt is the capital intensity gap, SI _ij is the skill intensity gap, Horz _ij is the horizontal spillover, HEF _ij is the Herfindahl index and Horz _ij × TG _ij , Horz _ij × CI _ij and Horz _ij × SI _ij are the interaction of horizontal spillover with technological gap, capital intensity gap and skill intensity gap, respectively.

The dynamic panel data model specified in Eq. (1) is characterised by the presence of lagged dependent variable among regressors. However, by construction, the inclusion of lagged dependent variable as explanatory variable correlates with the disturbance term. The resultant is autocorrelation and individual effects characterising heterogeneity among the individual intercepts, which were allowed to vary among different cross-sections. To mitigate the potential bias and inaccuracy associated with the use of ordinary least square (OLS), system generalised method of moment (SYS-GMM) estimator developed by Arellano and Bover (1995) has employed to estimate Eq. (1).

By following similar procedures to the above model specification, the following dynamic econometrics model is specified to evaluate the FDI productivity spillover effect on the domestic industry level as follows:

{VAD}_{j t} = γ {VAD}_{j t - 1} + Q_{1} K_{j t} + Q_{2} L_{j t} + Q_{3} M R_{j t} + Q_{4} F P_{j t} + Q_{5} E X T_{j t} + Q_{6} S Z_{j t} + Q_{7} T G_{j t} + Q_{8} C I_{j t} + Q_{9} {SI}_{j t} + Q_{10} F P_{j t} \times T G_{j t} + Q_{11} {FP}_{j t} \times {CI}_{j t} + Q_{12} {FP}_{j t} \times {SI}_{j t} + Q_{13} H o r z_{j t} + Q_{14} {Forward}_{j t} + Q_{15} {Backward}_{j t} + Q_{16} {Labcom}_{j t} + Q_{17} {HEF}_{j t} + α_{j} + ε_{j t}

(2)

Where, j denotes the industry (j = 1, 2, …, 443) and t denotes period (t = 2011, …, 2016). VAD _jt denotes the mean of total value added per labour at the industry level in each year. MVA _jt _–1 denotes the lag value of the total value added of industry production per labour in each year. Forward _jt stands for forward spillover, Backward _jt stands for backward spillover, and Labcom _jt stands for labour competition effect. The other remaining explanatory variables denote the mean of the observed variable in Eq. (1).

Similarly, to meet objective three, the spillover identified in objective one (horizontal spillover) was regressed against the horizontal spillover channels, technology gap and skill using the SYS-GMM estimator based on Green (2006) and Baltagi et al. (2015). The results are presented as follows:

{Horz}_{j t} = α_{j} + γ {Horz}_{j t} + Q^{'} {MX}_{j t} + U_{j t}

(3)

Where Horz_jt denotes the mean of horizontal spillover of the domestic firm from foreign firms in the same industries. Horz_jt−1 denotes the mean of the lag value of horizontal spillover of domestic industries from foreign firms in the same industry. X_jt denotes a vector of the mean of explanatory variables such as skill intensity gap, technology gap, demonstration effect, and labour turnover and competition effects. The measurement of variables and the expected signs for the coefficients of the explanatory variables included in the system GMM model are presented in Table 1.

Table 1.

Explanation of Variables Measurements.

Symbol	Variable name	Measurement	Expected signs
VAD	Natural logarithm of home firm productivity	Measured by proxy, the logarithm of the ratio of total value added to its employment figure
K	Fixed capital	The logarithm of the ratio of the net asset of a firm to their labour (Cuyvers et al., 2008; Hale & Long, 2011; Nhamo, 2011)	+
L	Labor input	The logarithm of the total number of labourers in firms	+
MR	Input materials	The logarithm of the ratio of the value of material input purchases to their employment (Cuyvers et al., 2008)	+
HEF	Herfindahl index	Measured by proxy variable, which is the square of the ratio of the value added of the firm to the total value added of the sub-sector	+
EXT	Export of firm	The logarithm of the export value of the firm	+
SZ	Size of the firm	The firm’s value added as a share of the average value added in the sub-sector to which the firm belongs	+
FP	Foreign presence	The ratio of foreign firms’ employment in a sub-sector to the total employment of that sub-sector	+
CI	Capital intensity	It is defined as the percentage difference in the capital–labour ratio between a domestic firm and that of an average foreign firm in the same sub-sector (Van Thanh & Hoang, 2010)	+
SI	Skill intensity	The difference between the average wage of a worker in a domestic firm and foreign-owned firms	+
Demos	Demonstration effect	Captured by the percentage of firms who have changed production techniques (process) due to competitive pressure (to keep up) from FDI within the same industry	+
Labcom	Labour market competition	Proxied by the percentage of respondents who answered ‘Yes’ to the question: ‘Do you face competition from foreign enterprises in the labour market (difficulties in hiring skilled workers)?’	+
Compn	Competition effect	Proxied by the percentage of firms in the same industry that directly adopt or imitate production techniques by observing or copying from FDI competitors	+
LabMob	Labour mobility	Proxied by the percentage of firms in the same industry who hired employees with experience in FDI firms in the same industry	+
Skill	Skill	Captured by the average wage of workers	+
TG	Technological gap	Proxied by the difference between the domestic firm’s labour productivity and the average labour productivity in foreign firms in the same industry (productivity measured in natural logarithms)	–
Horz	Horizontal spillover	${Horz}_{i j t} = \frac{{FP}_{i j t} \times {VAD}_{i j t}}{\sum {VAD}_{i j t}}$ Where FP _ijt × VAD _ijt is foreign firms’ value-added and $\sum {VAD}_{i j t}$ is total value-added	+
${Forward}_{i j t}$	Forward spillover	Measured by the percentage of respondents who answered ‘Yes’ to the question: ‘Have your customer (buyer) relations with FDI brought customer-related technology transfer to your firms this year?’ This percentage is calculated based on the total number of respondents from firms in the same sub-sector	+
${Backward}_{i j t}$	Backward spillover	Proxied by the percentage of respondents who answered ‘Yes’ to the question: ‘Have your supplying relations with FDI brought supplier-related technology transfers to your firms this year?’	+

Note: FDI, foreign direct investment.

Econometric Issues

Eqs. (1) and (2) in the aforementioned models include variables that may be subject to the endogeneity problem. Several studies, such as Olley and Pakes (1996) and Levinsohn and Petrin (2003), have shown that firms can select production inputs, such as capital and labour, based on their productivity. However, the latent nature of this experience and ability makes it difficult for econometricians to accurately determine the required magnitude of capital and labour inputs in an exogenous sense. Consequently, the OLS method is inappropriate for estimating the impacts of these inputs as it treats capital and labour as exogenous variables. Failure to account for endogeneity can result in biased estimates of the coefficients.

The relationship between foreign presence in domestic firms and domestic firm productivity is likely to run in both directions. This is because FDI not only affects domestic firm productivity, but more productive firms also attract FDI. The issue of endogeneity between foreign presence and productivity spillovers in a panel data framework has been extensively addressed in the existing empirical literature. For instance, Busse and Hefeker (2007), Campos and Kinoshita (2008), Demekas et al. (2007) and Carstensen and Toubal (2004) have attempted to address it.

Furthermore, the relationship between domestic firm productivity and FDI has been subject to examination through the lens of Granger causality. For example, the studies conducted by De Mello (1999), Hansen and Rand (2006), Piyaareekul (2008) and Chowdhury and Mavrotas (2006) have demonstrated a two-way causality between domestic firm productivity and FDI.

To address these issues, studies have utilised the system GMM estimator as a solution. For instance, Carkovic and Levine (2002) used the system GMM estimator to examine the role of FDI on economic growth. Tondl and Fornero (2008) utilised the system GMM approach to tackle the endogeneity problem in analysing the effect of productivity spillover from FDI to domestic firms in Latin America. In conclusion, it is crucial to correct for endogeneity when estimating the effects of productivity spillover effect from FDI to local firms using system GMM system estimator and other appropriate methodologies.

Moreover, instrumental variables procedures have been suggested for panels with fixed effects, starting with Anderson and Hsiao (1981), who recommended estimating differences and using the lagged observation from one period prior as an instrument. Later, Arellano and Bond (1991) popularised the GMM difference estimator, which became widely applied. They also estimated the model in differences to eliminate the fixed effects but employed all available lagged observations in levels as instruments, claiming a substantial increase in estimation efficiency.

In this study, a suitable instrumental variables estimator needs to tackle two main challenges. First, the lagged dependent variable on the right-hand side of Eq. (1) is anticipated to have a high level of persistence with a significant autoregressive coefficient. Second, other variables in our model may also display similar levels of persistence. This situation can lead to weak instrument problems when using lagged levels as instruments for equations in differences, such as with the GMM difference estimator. Moreover, the Arellano and Bond (1991) difference GMM method is prone to substantial downward finite-sample bias due to short time periods (Blundell & Bond, 1998). To overcome this issue, the GMM system estimator introduced by Blundell and Bond (1998) provides a viable solution by expanding the set of instruments to include those for the level equation, effectively addressing the problem of finite-sample bias. For instance, Carkovic and Levine (2002) and Tondl and Fornero (2008) have employed the system GMM estimator with moment restriction assumptions instead of assuming normality. This methodology is feasible because of the standard assumption regarding additional moment conditions, which permits the utilisation of endogenous lagged variables as valid instruments for multiple periods. It is important to highlight that this assumption remains valid in the absence of serial correlation, as demonstrated by Arellano and Bover (1995) and further supported by Blundell and Bond (1998).

IV. Result and Discussion

Descriptive Result Analysis

Tables 2 and 3 provide descriptive statistics of the variables and enable us to compare domestic firms with foreign firms. As shown in the two tables, the mean gross value of production of domestic and foreign firms over the period of 2011–2016 was about 120 million and 150 million Ethiopian birr, respectively. The mean productivity (total value of production per worker) of foreign firms is 790,624.8 ETB, whereas the domestic average is 623,405.7 ETB over the specified period. This means that foreign firms are 13 per cent more productive than firms owned by citizens of the host country. Similarly, the average capital and raw material consumption intensity on average are 2,443,425 ETB and 1,925,665 ETB in foreign firms and 664,942 ETB and 4,183,345 ETB in domestic firms, respectively. This figure shows that the employment of capital and raw material inputs in foreign firms was larger than the counterpart.

Table 2.

Descriptive Statistics of Domestic Firms’ Performance.

Variable	Mean	Std. Dev.	Min	Max
TVP	1.20E+08	8.20E+08	18,325	1.22E+10
VAD	623,405.7	3,215,798	839.279	6.17E+07
K	664,942.3	8,753,521	0.007246	2.92E+08
RM	418,334.6	2,526,024	26.98795	4.26E+07
L	212.1343	1,821.707	10	52,013
AVEWE	43,191.58	276,168.5	123.8399	7,411,137
Export	1.39E+07	3.18E+08	0	9.44E+09
FP	0.131537	0.11988	0.017142	0.517291
HEF	0.011394	0.06414	6.19E-07	0.915361
SZ	0.892574	5.913551	0.000072	144.4533
Horz	0.132083	0.170475	0.001439	0.519199
SI	–6,972,471	8.73E+07	–1.76E+09	1.25E+07
CI	–469,372	8,144,890	–2.92E+08	1,020,408
TG	879,829.1	4,214,955	–5.39E+07	1.63E+07

Source: Own Computation from Central Statistical Agency Survey data (2011–2016).

As presented in the summary Tables 2 and 3, the average number of people engaged in the firms was 274.5234 in domestic firms and 212.1343 in foreign firms. Likewise, FDI productivity spillover to domestic manufacturing firms through horizontal linkage is 0.132083 on average. The value of the variable ‘foreign presence’ in an industry ranges from a minimum value of 1.7 per cent to a maximum value of 51 per cent with a mean value of 13 per cent. Since foreign presence at the industry level is proxied by the share of foreign firm employment in the total employment of an industry, we can understand from this figure that foreign firms have a 13 per cent employment share in the manufacturing sector of Ethiopia on average.

Table 3.

Descriptive Statistics of Foreign-owned Firms’ Performance.

Variable	Mean	Std. Dev.	Min	Max
TVP	1.50E+08	6.18E+08	322,606	6.54E+09
VAD	793,726.3	2,285,062	2,979.329	1.74E+07
K	2,455,056	1.44E+07	0.111111	1.72E+08
RM	1,934,321	1.67E+07	294.7852	2.24E+08
TE	275.7887	850.0101	10	12,013
AVEWE	21,939.58	84,903.57	312.9267	834,071.7
Export	5.90E+07	5.96E+08	0	6.37E+09
HEF	0.0192816	0.0560266	1.44E-05	0.473184
SZ	1.020681	4.700731	0.000518	55.89814

Source: Own Computation from Central Statistical Agency Survey data (2011–2016).

The significant differences in skill, capital and technological intensity between foreign and domestically owned firms range from –1.76E+09 to 1.25E+07, –2.52E+08 to 10,020,408 and –5.39E+07 to 1.63E+07, respectively. The negative sign in the skill intensity gap shows that the average wage of foreign firms in that specific industry is greater than the average wage of domestic firms. The negative sign in the capital intensity gap shows that the average capital per labour in a specific industry is greater than the domestic firm capital per labour. Similarly, the negative sign in the technological gap shows that the average productivity of foreign-owned firms in the same industry is greater than that of domestic firms. The opposite is true for the positive values in the preceding case.

As far as the market concentration is concerned in this study, its value ranges from a minimum of 6.19E-07 to a maximum of 0.9153, with a mean value of 0.0114, as portrayed in Table 2. Since the average value of the HEF is low, it may indicate that on average, firms do not have greater market power. Firm size also ranged from 0.000072 to 144.4533, with a mean value of 0.893, suggesting the existence of huge variation among firms’ total value of production in the industries.

SYS-GMM Estimation Pre-test

Table 4 presents the correlation matrix of the variables used in the econometrics model. Capital intensity, labour intensity, material input intensity, foreign presence, Herfindahl index, firm size and horizontal spillover effect are all positively correlated with firm productivity. In contrast, firm productivity is negatively correlated with the technological gap, capital intensity gap and skill intensity gap between domestic and foreign-owned firms. Except for the strong correlation between the Herfindahl index and the size of firms, the correlation score between other variables does not show a strong correlation. The strong correlation between the Herfindahl index and the size of the firm may prove difficult to use at the same time in a single regression because of multicollinearity problems. Based on economic theory, this study uses the Herfindahl index only for analysis.

Table 4.

Correlation Matrix for Bivariate.

	VAD	KL	RM	HEF	SZ	FP	Horz	SI	TG	CI	FP × CI	FP × TG
VAD	1.00
K	0.35	1.00
L	0.61	0.34	1.00
RM	0.54	0.35	0.55	1.00
HEF	0.48	0.03	0.12	0.06	1.00
SZ	0.44	0.02	0.07	0.04	0.92
FP	0.02	–0.26	–0.02	–0.01	0.00	0.01	1.00
Horz	0.18	0.29	0.20	0.17	0.02	–0.02	–0.02	1.00
SI	–0.05	–0.04	–0.08	–0.03	0.00	0.00	0.00	–0.06	1.00
TG	–0.28	0.03	–0.02	0.00	–0.47	–0.40	–0.01	0.10	–0.01	1.00
CI	0.13	0.18	0.16	0.09	0.08	0.00	0.00	0.35	–0.05	–0.04	1.00
FP × CI	–0.02	–0.03	–0.03	–0.03	0.01	0.00	–0.54	–0.01	–0.01	–0.01	–0.01	1.00
FP × TG	–0.01	0.05	–0.03	0.03	0.03	0.10	–0.01	0.12	–0.02	–0.14	0.03	1.00

Source: Author’s estimation based on data set discussed in the methodology.

Diagnostic Test

Since the author considered only domestic firms in the analysis, the Heckman two-step selection model was employed to correct the selection bias resulting from the exclusion of foreign firms. Table 5 presents the result of the Heckman two-step selection model estimation. The result shows that the Mill ratio coefficient is not statistically significant at any conventional level of significance. Thus, we do not reject the null hypothesis. This implies that there is no selection bias in the specified model.

Table 5.

Heckman Two-step Sample Estimation.

					Censored obs = 1,532 Uncensored obs = 12,804 Wald chi² (5) = 332.80 Prob > chi² = .0000
	Coef.	Std. Err.	Z	P > \|z\|	[95% Conf. Interval]
VAL	0.186846	0.0742083	2.52	.012	0.0414003	0.3322916
K	0.184516	0.0678397	2.72	.007	0.0515527	0.3174793
L	0.2897508	0.1154152	2.51	.012	0.0635413	0.5159604
RM	0.4634927	0.0768132	6.03	.000	0.3129417	0.6140438
Antrho	0.357754	0.5227002	0.68	.494	–0.6667196	1.382228
L	0.2234048	0.0352504	6.34	.000	0.1543152	0.2924944
K	0.0552008	0.0225705	2.45	.014	0.0109635	0.0994381
RM	–0.0324187	0.0279976	–1.16	.247	–0.087293	0.0224557
ISIC	–0.0000525	0.000072	–0.73	.466	–0.0001936	0.0000885
Year	–0.0011946	0.0001855	–6.44	.000	–0.0015582	–0.000831
FP	3.70e-08	1.10e-08	3.38	.001	1.55e-08	5.85e-08
HEF	0.0080278	0.0050524	1.59	.112	–0.0018747	0.017930
Mills	Lambda	0.4679708	0.631398	.74	0.459	–0.7695464
	Rho	0.00
	Sigma	0.00

Source: Author’s estimation based on data series discussed in the methodology.

The Econometric Model Estimation Result and Discussion

Model 1, Model 2 and Model 3 in Table 6 columns represent empirical model estimation results for the models specified in Eq. (1) at the firm level, Eq. (2) at the industry level and Eq. (3) at the industry level for horizontal spillover occurrence determinants analysis, respectively.

Table 6.

Result of Dynamic Panel-data Estimation, One-step System Generalised Method of Moment (GMM).

Model 1	Coefficients	Model 2	Coefficients	Model 3	Coefficients
L.VAD	0.309***	L.VAD	0.318***	L.Horz	0.6471162***
	(0.0256)		(0.0669)		(0.080)
K	0.0933***	K	0.294***	Demos	0.8361643**
	(0.0225)		(0.0829)		(0.373)
L	0.290***	L	0.774***	Compn	0.664389**
	(0.0553)		(0.206)		(0.579)
RM	0.240***	RM	0.240***	LabMob	–3.82E-09
	(0.0354)		(0.0656)		(5.790E-10)
SI	–1.51E-10	SI	–1.97E-08	Skill	0.2838385
	(9.67e-11)		(1.19e-08)		(0.086)
TG	–1.84e-07***	TG	–0.651**	TG	–8.98E-09***
	(4.14e-08)		(0.287)		(–1.28E-09)
FP	4.86e-09**	FP	8.39e-10**
	(2.35e-09)		(3.24e-10)
CI	–0.00459*	CI	–0.475**
	(0.00250)		(0.206)
Horz	0.0107***	Horz	1.276**
	(0.00230)		(0.527)
HEF	1.178	Labcom	0.0598
	(1.684)		(0.320)
		EXT	0.0094537
			(0.420)
Horz × TG	4.17e-07***	FP × TG	3.01e-09***
	(1.08e-07)		(1.01e-09)
Horz × CI	8.11E-09	FP × CI	0.00608
	(1.54e-08)		(0.00486)
Constant	5.480***	Forward	–1.12E-08
	(0.373)		(2.49e-08)
		Backward	0.452**
			(0.206)
Arellano–Bond test for AR (2) (p value)	.111		.128		.209
Hansen J-test	0.293		0.741		0.801
Hansen tests excluding the differenced Instruments on the level equation  (p values)	.872		.890		.856
Hansen tests excluding the system GMM instruments (p value)	.61		.923		.789

Notes: Robust standard errors in parentheses. ***p < .01, **p < .05, *p < .1 (denote significance at the 1%, 5% and 10% level, respectively).

Source: Own computation from CSA survey data (2011–2016).

For both firm and industry-level estimations of the model, the p values of the Arellano–Bond test for AR (2) are greater than 5 per cent. Thus, the null hypothesis of no autocorrelation in the AR (2) process is not rejected for the three estimations. This test therefore supports the validity of the model specifications. As shown in Table 6, the p values for the Hansen J-test are greater than the conventional level of significance. Hence, do not reject the null hypothesis that the sets of instruments used are appropriate. Therefore, the lags used in the model estimation have valid instrumentation.

The coefficients of the control variables such as capital, labour and material input per labour show positive and significant results at a 1 per cent level of significance. In contrast, the coefficients of the other control variables, such as skill intensity and technology gap, show negative and significant levels at conventional levels of significance. Since their logarithmic values have been taken, the coefficients are the elasticity. However, we have not probed further into them because they are not in our interest.

As shown in Table 6, column 2, the foreign presence has the expected positive sign and is significant at the 5 per cent level of significance. The coefficient shows that a 1 per cent increase in foreign presence will lead to a 4.86E-09 percentage increase in domestic firm productivity, ceteris paribus. This might be a sign of the occurrence of the direct effect of FDI on domestic firm productivity in the manufacturing sector.

To capture the effect of foreign presence on domestic firm productivity because of the absorptive capacity of the industry in which the firm was operating, we made an interaction between the industry mean of foreign presence and the means of technology gap, capital intensity gap and skill intensity gap.

The estimated coefficient of the interaction between foreign presence and technology shows a positive and statistically significant at a 1 per cent level. This implies that a 1 per cent increase in foreign presence in industry results in an 8.39E-10 per cent increase in industry productivity, in moderate levels of technology gaps (i.e., both variables are centred on the mean), keeping other factors constant. Even though the magnitude of the spillover effect in the industry because of the presence of the foreign firm seems small, it has similarities with the empirical findings of Blomström and Sjöholm (1999) for Malaysia and Liu et al. (2001) for China. Indeed, given China’s and Malaysia’s high levels of economic development, and the total FDI in both countries is much higher than in Ethiopia, this may not come as a surprise. However, regarding developing countries, this finding is consistent with the study of Boly et al. (2015) conducted in selected sub-Saharan countries. They reported that foreign presence has a positive effect on domestic firm productivity for those firms that have better absorptive capacity. The finding is also plausible with the theory that firms’ absorptive capacity is a determinant for FDI productivity spillover occurrence. This explains why the productivity spillover effect from FDI depends on the industry’s absorptive capacity. This implies that foreign presence in the manufacturing industry of Ethiopia appeared to be a determinant of domestic firm productivity, putting absorptive capacity as a condition.

As observed in Table 6, column 2, the horizontal spillover effect has a positive and significant coefficient at a 1 per cent level. The coefficient for the interaction between the horizontal spillover and technological gap is positive and significant at a 1 per cent level of significance. This implies that the effect of horizontal productivity spillover on domestic firms’ productivity depends on the level of the technology gap. This means that a one-percentage-point increase in horizontal productivity spillover causes a 0.0107-percentage-point increase in domestic firm productivity at an average level of technology gap (i.e., has a score of zero on the centred technology gap variable), while all other variables remain constant. This is quite similar to the findings of Kokko (1994), Kokko et al. (1996) and Flores et al. (2000). Meyer (2004) and Mugendi (2014) also point out that when the technology gap between domestic and foreign firms reduces, the spillover effect from FDI to domestic firm’s increases. Turi’s (2015) empirical study on Ethiopian manufacturing firms contradicts this finding. The disparity in estimation methods and the incorporation of foreign firms in Turi’s study could account for this discrepancy.

Table 6 Model 2 represents the empirical model estimation results for models specified in Eq. (2) using the industry-level data set. Except for the slight changes, the estimated coefficients observed in column 4 show similar results using the firm-level data set presented in Table 6 Model 1. Thus, we delve into the vertical spillover (forward and backward productivity spillover) and labour competition effects only here. The result shows that the backward spillover has a positive and significant effect at a 5 per cent level of significance. Forward spillover and labour competition, on the other hand, are statistically insignificant at any conventional significance level. This shows that industries in which firms have backward linkage with foreign firms are more productive than industries with no linkage. To be more precise, the movement of backward linkage among the industries from zero to one percentage point results in a 0.00425 percentage point change in domestic firms under that industry, holding other factors constant. These findings are consistent with the findings of Turi (2015) for Ethiopia and Tomohara and Yokota (2006) for Thailand using firm-level unbalanced panel data.

Finally, on average, FDI improves domestic firms’ productivity through the horizontal and backward channels but does not affect the increase in productivity of domestic firms through forward linkage. The horizontal productivity spillover effect comes into existence through demonstration and competition effects, whereas the backward productivity spillover effect has been transmitted through local supplier firms’ related technology transfer.

SYS-GMM Estimation Result for Determinants of Horizontal Spillover

Table 6 Model 3 provides the SYS-GMM estimation result for the horizontal spillover Eq. (3) at the industry level. As shown in column 6, the coefficients of demonstration and competition are positive and significant at the conventional level of significance. According to the estimates, a 1 per cent increase in demonstration effect translates into a 0.84 percentage point increase in horizontal productivity spillover on average. Similarly, a percentage point increase in competition effect results in a 0.67 percentage point increase in horizontal productivity spillover, all else being equal.

The estimate for the coefficient of technology gap indicates a negative and significant level at 1 per cent. Conversely, the coefficients of skill and labour mobility show insignificant results. The negative and significant coefficient of the technology gap implies that an increase in one unit of the technology gap between the domestic and FDI firms results in an 8.98E-09-unit decrease in horizontal spillover, ceteris paribus. This implied that domestic firms with low technology gaps were bigger recipients of horizontal spillover and vice-versa. This is consistent with the findings of Mugendi (2014) in Kenya.

V. Conclusion

The focus on attracting FDI in developing nations has garnered significant interest from policymakers in recent times. Despite the existence of several studies on the spillover effects of FDI productivity in different countries, there remains a lack of empirical research in Africa (Boly et al., 2015). Given this context and the gaps in previous studies, this research aims to assess the impact of productivity spillover from foreign-owned firms to domestically-owned firms in the manufacturing sector of Ethiopia.

Accordingly, the empirical result shows the coexistence of both negative and positive effects arising from foreign presence on Ethiopian manufacturing firms. This implies that foreign presence leads to positive productivity spillover effects in low technological gap firms and negative effects in high technological gap firms concerning foreign firms. It confirms the theory contributed by Buckley et al. (2010, p. 192) for such a complex spillover. They argued that the complexity of spillover effects challenges the fallacious expectation of an equal spillover effect in all firms. The results are consistent with the Boly et al. (2015) and Bruhn and Calegario (2014) findings that FDI has a conditional effect on the host firm’s characteristics. From this, we can deduce that moderate technology gaps between foreign and domestically owned firms are an important determinant of productivity spillover occurrence. It is also similar to the result of Girma and Gorg’s (2005) argument that absorptive capacity is a crucial factor in diffusing advanced technology to the local industry.

Similarly, this study also shows evidence of an interaction between the horizontal spillover from FDI and the technology gap. Local firms with technology gaps at an average level and below average level have benefited from FDI productivity spillover. This study also found evidence of the channels through which the horizontal spillover effect occurs in domestic firms. It occurs through demonstration and competition effects. The estimates also reveal that foreign presence positively and significantly influences the productivity of local firms through backward linkage. The occurrence of these spillovers, however, is determined by the level of technological disparity between the two counter firms.

Based on the findings, five recommendations are made. First, policymakers should recognise that the impact of foreign presence on domestic firms varies depending on their technological gap. For low technological gap firms, policies should focus on attracting and facilitating foreign investment to capitalise on positive spillover effects. Conversely, for high technological gap firms, strategies should be developed to mitigate negative spillover effects, possibly through targeted support programmes or incentives.

Second, given the importance of absorptive capacity in diffusing advanced technology to local firms, efforts should be made to enhance the technological capabilities of domestic firms. This could involve investments in education and training, technology transfer programmes and fostering collaboration between foreign and domestic firms to facilitate knowledge exchange.

Third, the study highlights the positive influence of competition and demonstration effects on horizontal spillover from FDI. Policymakers should consider measures to promote healthy competition within the domestic manufacturing sector and facilitate opportunities for knowledge sharing and learning among firms. This could include initiatives to encourage networking, collaboration and information exchange platforms.

Fourth, the study indicates that foreign presence positively influences the productivity of local firms through backward linkages. Policymakers should explore ways to facilitate and strengthen backward linkages between foreign and domestic firms, such as promoting supplier development programmes, fostering local sourcing initiatives and facilitating technology transfer agreements.

Lastly, continuous monitoring and evaluation mechanisms should be established to assess the effectiveness of policies aimed at promoting FDI and enhancing productivity spillover effects in the manufacturing sector. This will allow policymakers to adjust strategies and interventions based on empirical evidence and feedback from stakeholders.

It would be interesting to explore potential extensions for future research. One possible avenue is to consider moving towards a scaled data type, if available, as this could provide more reliable ways of determining the spillover effect caused by vertical linkage between firms in the same industry and across other sectors.

In this study, the variable skill intensity was found to be insignificant in the analysis of all three models. The study suggests that this may be due to using the wages of workers as a proxy. For future studies, it is recommended to use the educational level of workers as a proxy for skill intensity. This is because the wage of workers is not necessarily based on their productivity, especially in developing countries.

Acknowledgements

The author would like to show my warm thanks to the Central Statistical Agency for the contribution in survey data delivery.

Availability of Data and Materials

The data sets used and/or analysed during the current study will be available when requested.

Consent for Publication

This manuscript does not include details, images or videos relating to individual participants.

Footnotes

Declaration of Conflicting Interests

The author declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Ethics Approval and Consent to Participate

This study does not involve human subjects, human material or human data.

Funding

The author received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Temesgen Woldamanuel Wajebo

References

Acemoglu

(2009). Introduction to modern economic growth . Princeton University Press.

Aitken

B. J.

, & Harrison

A. E.

(1999). Do domestic firms benefit from direct foreign investment? Evidence from Venezuela. American Economic Review , 89(3), 605–618.

Anderson

T. W.

, & Hsiao

(1981). Estimation of dynamic models with error components. Journal of the American Statistical Association , 76(375), 598–606.

Arellano

, & Bond

(1991), Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations. Review of Economic Studies , 58, 277–297.

Arellano

, & Bover

(1995). Another look at the instrumental variable estimation of error-components models. Journal of Econometrics , 68(1), 29–51.

Atieno

O. B.

(2015). Productivity spillovers from foreign direct investment in Kenya (Doctoral Dissertation). University of Nairobi.

Ayyagari

, & Kosová

(2010). Does FDI facilitate domestic entry? Evidence from the Czech Republic. Review of International Economics , 18(1),14–29.

Baltagi

B. H.

, Bun

M. J. G.

, & Sarafidis

(2015). Dynamic panel data models. In The Oxford handbook of panel data . Oxford University Press. https://doi.org/10.1093/oxfordhb/9780199940042.013.0003

Blomstrom

(1986). Foreign investment and productive efficiency: The case of Mexico. Journal of Industrial Economics , 35(1), 97–110.

10.

Blomström

, & Kokko

(1998). Foreign investment as a vehicle for international technology transfer. In Creation and transfer of knowledge: Institutions and incentives (pp. 279–311). Springer.

11.

Blomström

, & Sjöholm

(1999). Technology transfer and spillovers: Does local participation with multinationals matter? European Economic Review , 43(4–6), 915–923.

12.

Bloom

, Draca

, & Van Reenen

(2013). Trade induced technical change? The impact of Chinese imports on innovation, IT, and productivity. Review of Economic Studies , 80(2), 423–462.

13.

Blundell

, & Bond

(1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics , 87(1), 115–143.

14.

Boly

, Coniglio

N. D.

, Prota

, & Seric

(2015). Which domestic firms benefit from FDI? Evidence from selected African countries. Development Policy Review , 33(5), 615–636.

15.

Bruhn

N. C. P.

, & Calegario

C. L. L.

(2014). Productivity spillovers from foreign direct investment in the Brazilian processing industry. BAR-Brazilian Administration Review , 11(1), 22–46.

16.

Buckley

P. J.

, Wang

, & Clegg

(2010). The impact of foreign ownership, local ownership and industry characteristics on spillover benefits from foreign direct investment in China. International Business Review , 16(2), 142–158.

17.

Busse

, & Hefeker

(2007). Political risk, institutions, and foreign direct investment. European Journal of Political Economy , 23, 397–415.

18.

Campos

, & Kinoshita

(2008). Foreign direct investment, structural reforms and institutional quality: Panel evidence from Eastern Europe and Latin America (IMF Staff Papers 2008/26). IMF.

19.

Carkovic

, & Levine

(2002). Does foreign direct investment accelerate economic growth? Development Economics , 195–220.

20.

Carstensen

, & Toubal

(2004). Foreign direct investment in Central and Eastern European countries: A dynamic panel analysis. Journal of Comparative Economics , 32(1), 3–22.

21.

Chowdhury

, & Mavrotas

(2006). FDI and growth: What causes what? The World Economy , 29(1), 9–19.

22.

Crespo

, & Fontoura

M. P.

(2007). Determinant factors of FDI spillovers—what do we really know? World Development , 35(3), 410–425.

23.

Cuyvers

, Soeng

, Plasmans

, & van den Bulcke

(2008). Productivity spillovers from foreign direct investment in the Cambodian manufacturing sector: Evidence from establishment-level data (Working Paper 2008004). University of Antwerp, Faculty of Applied Economics.

24.

Damijan

J. P.

, Knell

, Majcen

, & Rojec

(2003). The role of FDI, R&D accumulation and trade in transferring technology to transition countries: Evidence from firm panel data for eight transition countries. Economic Systems , 27(2), 189–204.

25.

De Mello

(1999). Foreign direct investment-led growth: Evidence from time series and panel data. Oxford Economic Papers , 51(1), 133–151.

26.

Demekas

, Balázs

, & Ribakova

E. (2007)

Foreign direct investment in European transition economies—the role of policies. Journal of Comparative Economics , 35, 369–386.

27.

Di Ubaldo

, Lawless

, & Siedschlag

(2018). Productivity spillovers from multinational activity to indigenous firms in Ireland (Working Paper No. 587. The Economic and Social Research Institute (ESRI).

28.

Dogan

, Wong

K. N.

, & Yap

M. M.

(2017). Vertical and horizontal spillovers from foreign direct investment: Evidence from Malaysian manufacturing. Asian Economic Papers , 16(3), 158–183.

29.

Duce

, & España

B. D.

(2003). Definitions of foreign direct investment (FDI): A methodological note . Banco de Espana.

30.

Flôres

R. G.

Jr , Fontoura

M. P.

, & Santos

R. G.

(2000). Foreign direct investment spillovers: What can we learn from Portuguese data?

31.

Gachino

(2010). Foreign direct investment and firm productivity: Evidence from Kenyan manufacturing .

32.

Girma

, & Gorg

(2005). Foreign direct investment, spillovers and absorptive capacity: Evidence from quantile regressions (Paper Discussion). Kiel Institute for World Economics , 1(1248), 1–31.

33.

Glass

A. J.

, & Saggi

(2002). Multinational firms and technology transfer. The Scandinavian Journal of Economics , 104(4), 495–513.

34.

Gorodnichenko

, Svejnar

, & Terrell

(2014). When does FDI have positive spillovers? Evidence from 17 transition market economies. Journal of Comparative Economics , 42(4), 954–969.

35.

Görg

, & Greenaway

(2004). Much ado about nothing? Do domestic firms really benefit from foreign direct investment? The World Bank Research Observer , 19(2), 171–197.

36.

Green

W. H.

(2006). Econometric analysis . Prentice-Hall International.

37.

Hale

, & Long

(2011). Are there productivity spillovers from foreign direct investment in China? Pacific Economic Review , 16(2), 135–153.

38.

Hansen

, & Rand

(2006). On the causal links between FDI and growth in developing countries. The World Economy , 29, 21–41.

39.

Haskel

, & Slaughter

M. J.

(2002). Does inward foreign direct investment boost the productivity of domestic firms? (Working Paper 8724). National Bureau of Economic Research.

40.

IMF. (1993). Balance of payments manual (5th ed.). http://www.imf.org/external/np/sta/bop/BOPman.pdf

41.

Javorcik

B. S.

(2004). Does foreign direct investment increase the productivity of domestic firms? In search of spillovers through backward linkages. American Economic Review , 94(3), 605–627.

42.

Keller

(2004). International technology diffusion. Journal of Economic Literature , 42(3), 752–782.

43.

Kokko

(1994). Technology, market characteristics, and spillovers. Journal of Development Economics , 43(2), 279–293.

44.

Kokko

, Tansini

, & Zejan

M. C.

(1996). Local technological capability and productivity spillovers from FDI in the Uruguayan manufacturing sector. The Journal of Development Studies , 32(4), 602–611.

45.

Lenaerts

, & Merlevede

(2018). Indirect productivity effects from foreign direct investment and multinational firm heterogeneity. Review of World Economics , 154(2), 377–400.

46.

Levinsohn

, & Petrin

(2003). Estimating production functions using inputs to control for unobservables. Review of Economic Studies , 70(70), 317–341. https://doi.org/10.1111/1467-937X.00246

47.

Liu

, Parker

, Vaidya

, & Wei

(2001). The impact of foreign direct investment on labor productivity in the Chinese electronics industry. International Business Review , 10, 421–439.

48.

Meyer

K. E.

(2004). Perspectives on multinational enterprises in emerging economies. Journal of International Business Studies , 35(4), 259–276.

49.

Mugendi

C. N.

(2014). Foreign direct investment spillovers and productivity of domestic firms in Kenya (Doctoral Dissertation). Kenyatta University.

50.

Negara

S. D.

, & Adam

(2012). Foreign direct investment and firms’ productivity level. ASEAN Economic Bulletin , 29(2), 116–127.

51.

Negash

E. S.

, Zhu

, Lu

, & Wang

(2020). Does Chinese inward foreign direct investment improve the productivity of domestic firms? Horizontal linkages and absorptive capacities: Firm-level evidence from Ethiopia. Sustainability , 12(7), 3023.

52.

Nhamo

(2011). MNC-borne FDI, absorptive capacity and economic growth: An empirical investigation . University of the Witwatersrand. http://wiredspace.wits.ac.za/handle/10539/10670

53.

OECD. (1996). Benchmark definition of foreign direct investment (3rd ed.).

54.

Olley

G. S.

, & Pakes

(1996). The dynamics of productivity in the telecommunications equipment industry. Econometrica , 64(6), 1263. https://doi.org/10.2307/2171831

55.

Pham

(2016). Foreign direct investment, productivity and crowding-out: Dynamic panel evidence on Vietnamese firms. Proceedings of the 5th Economic & Finance Conference, Miami . https://doi.org/10.20472/EFC.2016.005.022

56.

Piyaareekul

(2008). Productivity spillover effects of FDI in ASEAN countries. In International conference on applied economics (pp. 751–757). ICOAE.

57.

Romer

P. M.

(1986). Increasing returns and long-run growth. Journal of Political Economy , 94(5), 1002–1037.

58.

Romer

P. M.

(1990). Endogenous technological change. Journal of Political Economy , 98(5), S71–S102.

59.

Romer

P. M.

(1994). The origins of endogenous growth. Journal of Economic Perspectives , 8(1), 3–22.

60.

Sanfilippo

, & Seric

(2016). Spillovers from agglomerations and inward FDI: A multilevel analysis on sub-Saharan African firms. Review of World Economics , 152(1), 147–176.

61.

Sasidharan

(2006). Foreign direct investment and technology spillovers: Evidence from the Indian manufacturing sector . Maastricht Economic and Social Research Institute on Innovation and Technology (MERIT).

62.

Schoors

, & Van Der Tol

(2002). Foreign direct investment spillovers within and between sectors: Evidence from Hungarian data . Ghent University, Faculty of Economics and Business Administration.

63.

Seyoum

, Wu

, & Yang

(2015). Technology spillovers from Chinese outward direct investment: The case of Ethiopia. China Economic Review , 33, 35–49.

64.

Tomohara

, & Yokota

(2006). Does foreign direct investment benefit domestic companies via increased productivity: Horizontal, backward, and forward linkages . The International Centre for the Study of Queens College and CUNY Graduate Center.

65.

Tondl

, & Fornero

J. A.

(2008). Sectoral productivity and spillover effects of FDI in Latin America (Working Paper No. 53). FIW—Research Centre International Economics.

66.

Turi

A. N.

(2015). Productivity spillovers from foreign direct investment: The case of Ethiopia (Working Paper No. 29/2015). Institute of Economic Studies.

67.

UNCTAD. (2016). World investment report 2016: Investor nationality: Policy challenges . UN.

68.

Vahter

(2004). The effect of foreign direct investment on labour productivity: Evidence from Estonia and Slovenia (Working Paper 32). Faculty of Economics and Business Administration, University of Tartu.

69.

Van Thanh

, & Hoang

P. T.

(2010). Productivity spillovers from foreign direct investment: The case of Vietnam. In Causes and consequences of globalization in East Asia: What do the micro data analyses show (pp. 228–246). ERIA.

70.

Waldkirch

, & Ofosu

(2010). Foreign presence, spillovers, and productivity: Evidence from Ghana. World Development , 38(8), 1114–1126.

71.

Wei

, & Liu

(2001). Foreign direct investment in China: Determinants and impact . Edward Elgar Publishing.

72.

, & Sheng

(2011). Productivity spillovers from foreign direct investment: Evidence from firm-level data in China. World Development , 40(1), 62–74.