The Impact of Foreign Direct Investments on Poverty Reduction in the Western Balkans

2.1 FDI and Economic Growth

As it is considered a key ingredient of economic growth and development, it is not surprising that the literature on FDI and its impact on the economy is thriving. However, not all views totally concur on the impact of FDI on economic growth, with some authors questioning whether economic development prompts the need for FDI or vice versa. The OECD (2002) argues that there are five means through which FDI can affect economic growth in the anchor economy: transfer of technological advances and know-how; increase in competition; amelioration of human capital; consolidation of the host economy toward the world economy; and encouragement of more positive development of firms.

Apergis, Lyroudi, and Vamvakidis (2008), using a panel dataset of 27 transitional European economies, empirically showed that FDI has a better impact on economic growth, at least in transitional economies with high level of incomes and efficient privatization schemes.

In their empirical study, Borensztein, De Gregorio, and Lee (1998) find a growth impact on the host economy along with interaction with human capital. Dollar and Kraay (2004) suggest that economic development elevates the incomes of people below the poverty line in proportion with general growth. However, in the pursuit of economic growth, adequate levels of education and infrastructure are important to take full advantage of the benefits of FDI. Along the same lines, Bengoa and Sanchez-Robles (2003) argue that a country’s financial stability is another prerequisite for a positive link between FDI and growth.

Furthermore, various studies have made use of different econometric approaches such as the Granger causality test as well as the Toda–Yamamoto test to examine the direction of the causality among FDI and economic growth. One example is by Chowdhury and Mavrotas (2006), who used the Toda Yamamoto method to examine the direction of causality between FDI and GDP growth for Thailand, Chile, and Malaysia. The findings for Chile show that GDP growth caused FDI net inflows and not the opposite. However, findings for Thailand and Malaysia show strong evidence of a bidirectional causality between GDP growth and FDI inflows.

Hansen and Rand (2006) back up these evidence of bidirectional causality even in the long run. Contrary to them, Carkovic and Levine (2002) showed no evidence, in their study of 72 countries, that economic growth can be accelerated by FDI inflows.

From another perspective, Kentor (1998) determined empirically that FDI initially has a positive effect on growth; however, in the long run, this impact is negative as dependency on foreign investments causes negative externalities, such as unemployment and income inequality. Another argument for FDI is that it can be more effective in boosting economic growth than domestic investment (Borensztein et al., 1998). This finding is consistent with the neoclassical growth theory (Solow, 1956; Swan, 1956), endogenous growth theory (Romer, 1986; Lucas, 1988), and modernization theory (Bengoa & Sanchez-Robles, 2003).

Various studies, such as by Hermes and Lensink (2003), emphasize the importance of financial market development. Utilizing a dataset of 67 countries, the financial system of 37 countries was adequately developed to trigger economic growth via FDI inflows. Alfaro et al. (2010) back these findings through a theoretical model that formalizes a mechanism that highlight the role of financial markets in promoting growth through FDI inflows.

Some studies on the direct impact of FDI on growth show that FDI can also have a negative impact on economic development. Hence, Forte and Moura (2013) found that FDI can negatively impact the implementation of domestic economic policies; Mencinger (2003) highlights that funds raised via FDI inflows through mergers and acquisitions (M&As) may easily be spent on imports that can deteriorate the balance of payments and have a negative impact on the long-term economic growth of the host country; however, Meyer (2004) argues that M&As may not be effective for economic growth as they may only end in a change of ownership and not an increase in capital. Conversely, greenfield FDI, which adds new capital to the economy, may be more beneficial to the anchor economy. Mencinger (2003) analyzed eight Central and Eastern European Countries and finds a negative correlation between FDI and economic growth. Motivated by Mencinger (2003), and studying the same set of countries, Eller, Haiss, and Steiner (2006) deduce that the positive impact on the economy does not last forever, and thus that an influx of FDI is not always advantageous.

2.2 FDI and Poverty Reduction

The connection between FDI and poverty is complex. It can impact it directly through various channels like employment, output, and human capital (Tsai & Huang, 2007). However, there are also indirect channels like distribution effect or economic growth (Magombeyi & Odhiambo, 2017).

The extent to which FDI contributes to poverty reduction depends also on various characteristics such as the mode of entry of foreign investments and financing type, along with the nature of the host economy (Meyer, 2004). The mode of FDI entry is fundamental to defining whether FDI can be beneficial or not to poverty reduction. A greenfield investment or a joint venture stimulates the insertion of new capital and would likely add local spillovers (Sumner, 2005). Greenfield investments or a joint venture bring out new products that have no close replacements. This mechanism of FDI may ameliorate the prevailing unemployment and underemployment by further contributing to mitigate poverty. In this way, FDI’s impact on poverty works through its impacts on employment. This impact has been considered a major one of FDI on poverty (Chudnovsky & López, 1999, IFC, 2000, Saravanamuttoo, 1999). On the other hand, FDI by way of a merger and acquisition is more probable of giving a pessimistic result in aiming for an in increase in welfare. This is because of the lack of new capital being transferred and also that employment may not be outweighed by technology transfer, managerial skill transfers, local sourcing spillovers, or increases in productivity or exports. Therefore, FDI may contribute to more unemployment if the mode of entry is through merger and acquisition. However, if FDI comprises the merger and acquisition of companies that are close to being shut down or bankrupt enterprises, it may help prevent potentially increased unemployment and therefore poverty (Hemmer & Hoa, 2002).

Many studies that investigated the empirical link between FDI and poverty reduction (such as Fowowe & Shuaibu, 2014; Gohou & Soumaré, 2012; Sharma & Gani, 2004, etc.) confirm the positive effect of FDI on HDI and poverty reduction. Assadzadeh and Pourqoly (2013), using human development index as a proxy variable for poverty, for 21 members of Middle East and North African countries in 2000–2009, showed that FDI had a significant and positive effect on poverty reduction. The study indicated that an increase in FDI boosts employment and middle-class income. On the other hand, Agarwal et al. (2017) using data from 1981 to 2011 empirically showed that FDI has had negative effect on poverty in India. However, for the same time span, studies in the South Asian Association for Regional Cooperation (SAARC) countries showed different empirical results. Although FDI had a positive effect in poverty reduction in Nepal and Sri Lanka, it showed poverty growth in Pakistan and Bangladesh.

Abor and Harvey (2008) examined, under the Stolper–Samuelson theorem, the effect of inward FDI flows to countries that have comparative advantages in labor-intensive sectors. The results showed that FDI flows tend to give rise to employment by adding more jobs by increasing median incomes and thus reducing poverty rates. On the other hand, Agénor (2004) asserted that the presence of rigidities in the labor market may impose difficulties in the Stolper–Samuelson theorem in the short term, which result in negative impact of FDI in poverty, especially when the benefits of FDI do not reach the low-income workers (Feenstra & Hanson, 1995).

The IMF (2000) analyzes the improvements that have been made in poverty reduction while increasing real incomes in developing countries. The results are discouraging, highlighting the widening gap between the rich and the poor. Some studies have shown that many of the beneficial impacts of FDI on poverty reduction are attained from spillover effects, elevation of investment capital, and employment creation (Görg & Greenaway, 2004). A vast number of country-specific studies have enforced a general belief in the positive correlation between FDI and poverty reduction. This direct impact of FDI on poverty reduction has been found to be higher than indirect effects through GDP growth (Hung, 1999; Jalilian & Weiss, 2002). In contrast, Huang, Teng, and Tsai (2010) report a negative correlation between poverty reduction and FDI in a study of 12 countries in East Asia and Latin America between 1970 and 2005. In general, the literature that communicates a negative or insignificant effect of FDI on poverty comes under the dependency theory, which accounts for the underdevelopment of the developing countries and maintains that the nature of development triggers poverty. It is essential to emphasize that the relationship between FDI and poverty is not the same for all countries and depends on a number of factors, such as technological gap, quality of institutions, and incentives to attract FDIs, as well as the nature of FDI – greenfield or mergers and acquisitions.

2.3 FDI, Economic Growth and Poverty Reduction

An important issue that needs to be addressed is whether poor people ultimately benefit from economic growth. In other words, does economic growth alleviate poverty? The general assumption of the literature that examines the connection among FDIs and economic growth and how they influence poverty alleviation is that economic growth is an instrument to mitigate poverty but its results can differ from country to country (Hanim, 2011) and its effectiveness can differ. The possible reasons for these variable results include empirical techniques and proxy variables for poverty, sample sizes, and timeframes. Although most studies have found a positive link between economic growth and the alleviation of poverty, a few of them have reported that FDI has a negative or clearly insignificant impact on poverty reduction. Using regression analysis of the growth of average income for the poorest 20% and the poorest 40% against GDP growth per capita, Roemer and Gugerty (1997) show that, in general, poor people gain from an increase in economic growth. Put differently, a rise in GDP per capita is directly translated into income growth for the poorest 40%.

Klein Michael et al. (2001) argues that FDI is a channel through which economic growth improves welfare. Therefore, it is considered a crucial instrument for poverty alleviation (Hanim, 2011). From an economic perspective, FDI inflows mean more capital injection, which is good for economic development. Technological advancements aided by FDIs have a major impact in continuing economic prosperity and societal progress. Moreover, there is the social side of the positive consequences that foreign subsidiaries bring to host countries, which is new job creation as well as the improvement of local skills, thus contributing to welfare. However, this can be successful only if the ratio of new openings is higher than FDI-related unemployment. In other words, FDI can be efficient in job creation in sectors that depend heavily on hand-operated jobs (e.g., agriculture).

Bourguignon (2004) examines the importance of income distribution in poverty alleviation. He argues that economic growth affects poverty alleviation through effective income distribution. Furthermore, from the macroeconomic perspective, FDIs can be beneficial to host countries through positive revenue transfer.

2.4 FDI Patterns in the Balkans

The transition economies of the Western Balkans include countries that have relinquished centralized planning and have moved toward a market economy. Some other identifying elements include acknowledgment of private property, privatization, opening up to global trade and foreign investment, low GDP per capita, volatile consumption growth, and low labor unit costs. Furthermore, as stated by Akbar and Brad McBride (2004), these economies have recently seen progress in middle-income families that look for better services and better-quality products, promoting MNEs’ interest in investing in these countries. Besides, the transition to a market economy has been accompanied by the significant process of deindustrialization. All of these economies relied excessively on the industrial sector during the Soviet era (Estrin & Uvalic, 2014). Today, the Western Balkan countries have changed to become substantially service economies. Nevertheless, the aftermath of the geopolitical conflicts of BiH (1992–1995), Kosovo (1998–1999), and the disintegration of the Yugoslavian Federation generated enduring economic drawbacks associated with weak macroeconomic performance and exclusion from international trade and financial flows, all of which made investors apprehensive about investing in these countries. This set of events delayed economic reforms by prioritizing political rehabilitation. Therefore, collectively, FDI stock in the first decade (Kosovo not included) amounted to US$8.5 billion, comprising only 6.5% of total FDI inflow in the transition economies. Foreign companies also showed a pattern of being wary to invest in these countries even though they are all post-communist countries and have experienced radical changes in all aspects of their economies.

The Western Balkan countries compete to attract foreign investment, which is seen as an important source of capital given the enormous lack of domestic savings in these countries. The FDI inflows during the communist period were minimal. Albania managed to attract a few foreign investments in the first post-communist years. The countries that attracted the most investment toward the end of the decade were Serbia, Montenegro, and the RNM, which saw some FDI inflow from 1997 onward. Serbia and Montenegro had attracted US$1 billion by the end of the post-communist decade, whereas the RNM had attracted US$540 million.

Improved economic and geopolitical conditions, paired with trade liberalization among countries, privatization programs, a better business environment, and deeper economic ties with the EU, were the main reasons behind the upsurge of FDI in the 2000s (Figure 1). In Albania, BiH, and the RNM (UNCTAD, 2017), FDI flows increased by 20–27% a year. Collectively, until 2010, FDI stock amounted to US$190 billion, with the biggest recipient being Serbia. The growth of FDI in Serbia surged mainly due to the privatization of its mobile network operator Mobtel, which was acquired by the Norwegian company Telnor. Other greenfield investments, including Ball Packaging, Vip Mobile, and Microsoft’s Development Center, made a major contribution to increasing inflows. Albania saw major growth in FDI mainly due to the privatization of the oil refinery complex ARMO for about €125 million. High FDI in financial services in 2008 was the result of the purchase of shares in Banka Popullore and Union Bank by foreign investors (Foreign Direct Investment Report, Albania 2011, UNDP). The RNM also experienced an increase in FDI, mostly because of the privatization of state-owned enterprises and the acquisition of major companies and banks (UNCTAD, 2012). A major contributor was the sale of Makedonski Telecom to Magar Telekom. Since 2007, greenfield projects such as free trade zones have been the largest sources of FDI inflows into the country. Flows to the financial and insurance sectors have also been large, albeit volatile.

Figure 1

FDI inflows as a percentage of GDP (2000–2019). Sources: UNCTAD (2017), World Investment Report 2017: Investment and the Digital Economy, http://unctad.org/en/pages/PublicationWebflyer.aspx?publicationid=1782; for Kosovo: World Bank (2017), World Development Indicators 2017 (database), https://openknowledge.worldbank.org/handle/10986/26447.

For a long time, Kosovo did not have a standardized system to organize the collection of FDI data. Therefore, we have referred to the World Bank data of FDI inflows from 2004. FDI in Kosovo from 2008 to 2015 was focused on diversified economic sectors, such as real estate, construction, and the financial sector, followed by the transport and telecommunication sectors (KPMG, 2017). The amount of FDI in euros received in Kosovo in 2014 was lower than in previous years. The failed telecom privatization, a weak business environment, and domestic political turbulence in 2014 were among the factors that contributed to the substantial decline of net FDI in 2014. The state is looking to improve the factors that have impeded attracting foreign investments, such as corruption, organized crime, and the judicial system.

The higher level of FDI in 2015 indicates signs of a solid recovery. The increase in FDI in 2015 was due to a higher inflow of investments although FDI in Kosovo continues to be highly dependent on developments in the economies of the Eurozone countries, where most of the FDI comes from.

From 2008 onward, the consequences of the global financial crisis were reflected in the FDI inflows to the Western Balkans. The fragility of FDI flows was related partly to the large share of inward FDI from the EU; economic woes had particularly negative knock-on effects for FDI in the subregion, which almost halved. The industry composition of inflows to Western Balkan countries worked against it during the financial crisis; investment had not been diversified and was concentrated mainly in industries such as finance and retail (UNCTAD, 2013).

However, the database shows us that since 2013 Western Balkans have been attracting US$ 5 billion of foreign investments annually (UNCTAD, 2017), with about half going to Serbia (the largest economy in the Balkans). Table 1 displays the evolution of FDI net inflows (% of GDP) for Western Balkan countries since 2000. It follows that in 2019, buoyed by competitive production costs and access to EU markets, Montenegro, Serbia, and Albania were the largest recipients in relative terms of FDI net inflows measured as a percentage of their GDP (8.42, 8.33, and 7.86% respectively) whereas BiH, RNM, and Kosovo were the countries with the lowest ratios in the region (1.95, 3.76, and 3.84% respectively). The Western Balkan countries have since avoided the general downward trend. In fact, in 2016, FDI stocks represented over 40% of GDP in the Western Balkans, reflecting the important role that FDI plays in these countries. Table 1 shows that even in 2017, 2018, and 2019, FDI net inflows have been relatively sizeable (Table 1) providing support for economic growth, job creation, and innovation and technological progress.

Table 2 shows the top five FDI investors in the Western Balkan countries since 2000. It shows that during the last two decades, Western Europe has been the primary source of foreign investments in the Western Balkans. Investment in the Balkans also comes from the Russian Federation, Turkey, the United States, etc. emphasizing the high concentration of FDI sources, which in turn confirms the region’s high dependency. Regional cooperation plans aimed at attracting investment to the Western Balkan countries were initiated under the umbrella of the Central Eastern Free Trade Area and are expected to support FDI in the region. Albania and Serbia have also benefited from Chinese investors who have emerged as a new source of foreign investments.

Table 3 shows the key impact variable, the poverty headcount ratio (HCR), at $1.90 a day (2011 purchasing power parity [PPP]), expressed as a percentage of population for all the Western Balkan countries since 2000.

2.5 Sectoral Distribution of FDI Across the Balkans

Although FDI in the Balkan countries, determined by institutional circumstances, geographical position, and the EU membership process, has had a distinct sectoral allocation across areas, financial services and telecommunication were at the beginning of the transition of the Balkans the principal recipients of FDI distribution, notably due to privatization.

During the second half of the 1990s, FDI increased rapidly in the financial sectors of the transition economies in the Balkans, fostered by market-based reforms, financial liberalization, transformation of the one-tier bank system into a two-tiered banking system, privatization of domestic banking institutions, and preparations for EU membership. According to the OECD (2018), the manufacturing sector accounted for the greatest share of FDI stocks in North Macedonia (36% of FDI stock), Serbia (32%), Bosnia and Herzegovina (28%), and Kosovo (12%) whereas financial intermediation accounted for 17–18% of FDI stock across economies (the banking sector accounts for the majority of financial sector foreign direct investment, FSFDI) whereas communications and transport account for 2–28%. The main determinant of FDI in the Balkans is attributed to cost efficiency as EU investors locate export-oriented activities in the region to serve their home markets. With respect to FSFDI, the Balkans represent high economic growth opportunities for foreign banks as their banking systems are weak. However, as the OECD (2020) highlights, such a “composition of FDI stock in the Western Balkans underlines the region’s vulnerability against the pandemic’s impact” (p. 20).

3.1 Variables

FDI occurs when MNEs, mostly headquartered in Western industrialized countries, invest in enterprises located in another country either by buying an existing business or by starting a completely new one (Apodaca, 2010). As such, FDI in this study is conceptualized as MNEs operating in the Western Balkans.

This study aims at determining the effects of FDI on poverty in the Western Balkans. The dependent variable is the poverty level. To measure poverty, income distribution and income inequality, a number of indicators have been used in the literature, such as HCR, average monthly income/consumption expenditure of people below the poverty line, the poverty gap ratio, the Gini index, the income shares of the bottom 20% of the population, etc.^[1] Primarily, we use HCR to measure poverty, defined as the percentage of people living in households in which the consumption or income per person is below the poverty line. HCR is measured using the international poverty line of US$1.9 a day (based on 2011 PPP) provided by the World Bank development indicators (WDIs). In addition, two alternative indicators are used to measure the poverty level and income distribution, such as the poverty gap and Gini index, in order to provide some robustness checks for the estimated results of the econometric analysis. The poverty gap reflects the intensity of poverty in a nation, showing the average shortfall of the total population from the poverty line, whereas the Gini index measures the extent to which the distribution of income (or, in some cases, consumption expenditure) among individuals or households within an economy deviates from a perfectly equal distribution.

Factors that drive FDI flows into developing countries (usually from developed countries) include an abundant supply of cheap labor, growth rate, trade openness, corruption, investment freedom, etc. The data on GDP and trade openness (i.e., the sum of exports and imports of goods and services measured as a share of gross domestic product) are provided by the WDIs, data on investment freedom come from the Heritage Foundation (HF), and those relating to the perception of corruption are from Transparency International (TI), which ranks 180 countries and territories by their perceived levels of public sector corruption according to experts and businesspeople, using a scale of 0 to 100, where 0 is highly corrupt and 100 is very clean. A description of the variables and the respective information sources are provided in Table A1.

In the empirical analysis, apart from FDI inflows as our primary explanatory variable, we have employed some additional control variables, in line with the literature review provided earlier in this study.

3.2 Other Indicators

The rule of law holds to the belief that countries with well-built rules will absorb more FDI due to the strong legal and political safeguards, which eventually will guarantee high levels of financial security. Gohou and Soumaré (2012) used the rule of law as a control variable as a determinant of efficient business environment and institutional quality. Their study showed that rule of law in African studies negatively impacted FDI inflows which led to higher poverty.

Financial market development is a core determinant for FDI inflows. What we anticipate is that countries with well-established financial markets are more attractive to enterprises that would want to invest in host countries. In this case, local financing acts as a safeguard instrument. Developed financial markets contribute toward cost reductions of external finance for foreign subsidiaries as well as enable foreign investing companies to control asymmetric information by making use of the local knowledge on risk and market opportunities. In other words, host countries with underdeveloped financial markets are more likely to discourage FDI as it may impose more risks and costs to source countries. Several authors have included financial development in their model, including Gohou and Soumaré (2012).

To ameliorate our empirical analysis, we have considered labor force as a control variable. Various studies show that labor-intensive FDI is more effective in poverty reduction. Nevertheless, the level of compensation is crucial in determining the degree to which the poverty rate is lowered. In other words, if foreign companies compensate the labor force above the poverty line FDI will be successful in poverty alleviation. Several studies have included labor force in their empirical analysis, including Assadzadeh and Pourqoly (2013), Bengoa and Sanchez-Robles (2003), Ghaith, Niaz Ahmad, and Lacheheb (2017), Louzi and Abadi (2011).

3.3 Sample and Data

This article makes an empirical contribution by trying to better understand whether there is a positive link between FDI and poverty reduction in the Western Balkans. Our estimation sample consists of six Western Balkan countries over the period 2002–2021, as data before this period are scarce for some of the considered variables and often of questionable quality. In addition, during this period, the WB6 countries experienced the largest recorded inflows of FDI and made considerable progress in alleviating poverty. The data have an annual frequency given that most of the selected indicators are only published yearly. After data were extracted from the information sources such as WDIs, HF, EBRD, and TI, according to the respective selected variables as described in Section 3.2, they were compiled into a single, unified, and cohesive data set structured as a dated panel, which is unbalanced as there are still some missing values at some time observations for some of the countries. The selection of the variables to be included in the model and their combination in different specifications was guided by the theoretical and empirical literature, as well as by the technical requirements of the estimation methodology, which are explained in detail in Section 3.4.

3.4 Method

The impact of FDIs on poverty reduction is estimated using equation (1):

where POV i t represents poverty in country i at time period t and

FDI i t   represent the foreign investments; HDI i t   denotes the Human Development Index taken from the UNDP Human Development Report; IIF i t denotes the index of investment freedom in country i at time period t; IEF i t denotes the index of economic freedom in country i at time period t; TO i t denotes the trade openness in country i at time period t; FERT i t is the fertility rate defined as the number of births per woman and measured by the WDIs; REM i t indicates personal remittances received as a percentage of GDP; CREDIT i t presents the credit to private sector in percentage of GDP; LABOUR i t shows the labor force participation rate as a percentage of total population of age over 15 years old; LAW i t signifies the rule of law; PROPERTY i t refers to property rights, and CPI i t represents a measure of corruption as the corruption perceptions index in country i at time period t; however, u i and v i t represent the unobservable country-fixed effects and idiosyncratic shocks, respectively, assuming that these two components are orthogonal such that: E [ u i ] = E [ v i ] =   E [ u i v i t ] = 0 .

Two methodologies are used to estimate equation (1): fixed effects and a dynamic GMM panel data estimator (Arellano & Bond, 1991; Hansen, 1982). Fixed effects estimation corrects for problems such as omitted variable bias that may arise from pure cross section regressions and takes account of the unobservable country-specific effects, which are assumed to be fixed parameters to be estimated (Islam, 1995).

However, given the nature of our model and the selected variables, there is a possibility of a reverse causation between poverty and the explanatory variables, and such endogeneity problems could arise from several channels. For instance, feedback effects could exist between poverty and FDIs. As poverty falls, households will spend more on goods and services, which will in turn increase production and attract more investments (both domestic and foreign). Thus, poverty could also influence FDI. The fixed effects model is not able to address this endogeneity problem. Thus, in addition, we apply to our fixed-effect dynamic panel estimation the one-step difference GMM estimator devised by Arellano and Bond (1991).

The issue of the endogeneity between the lagged dependent variable ( POV i t − 1 ) and the fixed-effect u i can be easily accommodated by taking the first differences of equation (1) as follows:

where Δ v i t =   v i t −   v i t − 1 .

Although the first-difference transform of the regression equation removes the country-specific fixed effect ( u i ), the endogeneity problems still exist. For example, POV i t − 1 in Δ POV i t − 1 = POV i t − 1 − POV i t − 2 is still mathematically correlated with v i t − 1 in Δ v i t = v i t – v i t − 1 . Moreover, if any regressors other than the lagged dependent variable are endogenous, those endogenous regressors will also be correlated with the error term ( Δ v i t ) and will result in biased estimated coefficients.

In order to address this endogeneity problem, an Arellano–Bond difference GMM estimator is employed. This estimator instruments a first-differenced endogenous regressor in the transformed regression equation (2) with its lagged levels. The main idea behind this estimator is that past (lagged) levels are often predictive of current changes ( Δ x i t ). Further, second or even deeper lagged levels of an endogenous regressor ( x i t − k for k ≥ 2) are available as instruments for its first-differenced endogenous regressor ( Δ x i t ) because unlike the mean-deviation transform in standard fixed-effect estimations, second or deeper lagged levels of the endogenous regressor ( x i t − k for k ≥ 2) remain orthogonal to the error term ( Δ v i t = v i t − v i t − 1 ) (Roodman, 2009).

However, the validity of lagged levels as instruments depends on the process of the error term ( Δ v i t ) in the first-difference regression equation (2). For example, if v i t follows an AR (1) process, then x i t − 2 is no longer a valid instrument for Δ x i t and so even deeper lags ( x i t − k for k ≥ 3) need to be used as instruments. Therefore, the Arellano–Bond test is applied to control for the presence of autocorrelation in the first-differenced residuals and to determine the number of lags available for instruments. Then, the Hansen test is used to determine whether overidentifying restrictions are valid in our estimations.