Printer Friendly

Can aid stimulate productivity in Sub-Saharan Africa? A dynamic panel approach.

1. Introduction

The debate on the effectiveness of development cooperation remains an important issue in the development economics literature. For a number of countries, particularly in Sub-Saharan Africa and South Asia, aid is an important source of capital and foreign currency. Some studies suggest that aid contribute to economic growth (Irandoust & Ericsson, 2005; Banerjee, 2007; Ssozi & Asongu, 2016). However, other researchers argue that development assistance is ineffective in economic growth stimulation, mainly due to its inherent nature and mismanagements (e.g. Papanek, 1972; Boone, 1996; Radelet et al., 2004; Easterly, 2009; Loxley & Sackey, 2008; Rodrik, 2009; Rajan & Subramanian, 2011). Undoubtedly, aid transfers are susceptible to the rent seeking problem in both donor and recipient countries. They serve a plethora of goals in domestic and foreign policy (Easterly, 2009). At the same time, the development cooperation system has undergone major policy changes since the 2000, thanks to the establishment of Millennium Development Goals and subsequent agreements.

Due to the dichotomy of the attitudes to development aid, in the article we reconsider the impact of aid and investigate whether the declarations to increase aid effectiveness were fulfilled. The goal of the study is to assess recent (1995-2014) impact of aid on fundamental economic indicators in Sub-Saharan Africa. Particularly, we look for the answers to the following research questions:

(a) Does technical ODA flow contribute to the increase of the productivity?

(b) What is the impact of total ODA on productivity growth (previous studies show that the flows are ineffective - see e.g. Alvi & Senbeta 2012)?

(c) Is the impact of selected forms of ODA on productivity more important than the impact of total ODA (which forms of ODA are most effective)?

Our results suggest that some aid flows, particularly technical cooperation, have a significant and positive effect on the productivity in the region. This implies that this form of international cooperation is indeed conducive for technology development. Technology development, often approximated with productivity, is expected to enable structural change by moving labour force from low- to high-productivity sectors (McMillan et al., 2014; Solow, 2016). Because of that, increasing productivity in Sub-Saharan Africa is of a paramount importance for its economic development.

This paper adds to the literature in two major ways. First, we decompose aid and verify the impact of different aid instruments on productivity in Sub-Saharan African states. Quite a lot of papers have addressed the issue of country characteristics that are crucial to increase the level of productivity and FDI impact on TFP (Chengchun & Sailesh, 2018). Less attention has been put on the choice of channels of aid flows (Kumar & Pacheco, 2012; Ssozi & Asongu, 2016). Hence, in this study we examined which types of aid are more likely to raise TFP in Sub-Saharan Africa.

Secondly, based on our results, we formulate policy implications for the development cooperation strategy and technology development in the analysed region.

As suggested by Solow (2001), in order to explore the determinants of TFP, we concentrate on a group of fairly similar countries (*). We estimate a panel model for 24 countries for the region, check different specifications and include a set of control variables, what makes our results fairly robust.

The reminder of the paper is as follows. In the following section (2), we review development aid literature and explain why productivity was selected as the dependent variable in the research. Next, we present the data (section 3), and the empirical model (section 4) applied in the study. In section 5 we discuss the results of our research and provide some policy implications. We include our main research conclusions in the last section (6).

2. Literature review

Development literature devoted a lot of attention to the aid and growth nexus, as aid was supposed to solve saving gap and poverty issues in low-income countries. However, currently the development economics concentrates not only on direct sources of growth (and development), but also on its indirect causes such as a productivity (Klenow & Rodriguez, 1997; Hausman, 2009). This is especially useful when a limited time series data is available. In this research we approximate economic progress with technology development, which can be expressed by the productivity concept, rooted in the Solow-Swan exogenous growth model.

There are two major groups of factors which foster productivity: technological and non-technological factors. The first group encompasses technology embodied in new production processes, new tools, new and higher quality goods and services, new methods and solutions (Klenow & Rodriguez 1997; Comin, Hobijn, & Rovito, 2006; Comin, 2008; Comin & Mestieri, 2014). The non-technological factors include i.a. the structure of the economy, human capital and institutions (Chanda & Dalgaard, 2008; Ascari & Di Cosmo, 2004; Byrne et al., 2005; Comin, 2008; Jajri, 2007; Frija et al., 2015). In this research, productivity is represented by TFP - the residuals from growth regression equation. It measures the growth of income per worker, which is not due to factor accumulation (physical capital, labour) (Lipsey & Carlaw, 2004). TFP allows to follow the change in efficiency of the usage of factor inputs and change in technology mix (Byrne, 2005; Comin, 2008; Jajri, 2007). As such, it has the potential to contribute to economic development in laggard economies. Consequently, there arises a question, whether development cooperation can stimulate the efficiency of the utilization of existing resources.

In theory, aid combines measures of capital injection and capacity building, which makes it an instrument suitable to address both groups of productivity factors. It is committed to sectors of social infrastructure, economic infrastructure and production in forms of grants and loans. Moreover, technical cooperation, includes free-standing technical cooperation (FTC, the provision of resources aimed at the transfer of technical and managerial skills or of technology for the purpose of building up general national capacity without reference to the implementation of any specific investment projects) and investment-related technical cooperation (IRTC, the provision of technical services required for the implementation of specific investment projects). Generally, it is dedicated to improve technology use in a recipient country (IMF, 2003).

Dual nature of aid (capital and non-capital instruments) is rooted in the main concepts of development economics, and as a policy instrument it was designed to fit into various kinds of development thinking. In the context of exogenous theories of growth, aid was supposed to compensate the shortage of capital in order to stimulate growth. It was supposed to contribute to input driven growth based on capital accumulation. As the practical realisation of the Rosenstien Roden "big push" idea, aid was meant to help low income countries escape a low-income equilibrium state in which they were stuck (the poverty trap) (Sachs et al., 2004; Graham et al., 2006). According to the second influential theory, aid could fill in the financing gap based on the two gap model delivered by Chenery & Strout in 1966 (Chenery, 1969). Aid was supposed to eliminate the distance between low domestic saving rate and the investment required to keep up with population growth and the depreciation of capital.

Justification of donating aid is also found in the endogenous growth models. Namely, it involves investments in human capital and the overall recipient capacity building, indispensable to launch economic growth (Romer, 1990; Grossman & Helpman, 1991; Aghion, 1992). Increase of technical capacity in the recipient countries was supposed to stimulate productivity and growth by increasing endogenous capacity to adopt and develop technology. While grants and program aid engage distribution of financial flows, project aid and technical cooperation involves direct actions in recipient countries.

According to Nachega & Fontaine (2006), aid can contribute to TFP growth especially if it involves investments in infrastructure (roads, power points, irrigation). In such cases aid may limit the strains on the domestic tax base, and thus prevent costly distortions. Therefore, aid has the potential to addresses the technology and non-technology aspects of TFP. However, Byrne et al. (2005) warn that aid in forms of projects and financial investments is not effective in achieving long-term convergence, when it is not successfully targeting the structural differences. Notably, it is the lack of compliance of aid flows with recipient's development strategies, which may constitute a major potential limitation for aid impact on productivity in Sub-Saharan Africa.

The functioning of aid received some heavy criticism (e.g. Bonne, 1996; Easterly, 2009; Cohen & Easterly, 2009) based on both theoretical and practical arguments. While some researchers consider aid ineffective in growth stimulation due to its primordial inadequacy for the purpose, including fungibility (Bonne, 1996; Rodrik, 2008; Easterly, 2009; Rajan & Subramanian, 2011), others associate it with the donors' approach (e.g. Easterly, 2009; Andrzejczak & Kliber, 2015). Alvi & Senbeta (2012) claim that aid negatively affects TFP; it reduces the efficacy of financial institutions and causes efficiency losses which undermine its overall effect on growth. Durlauf et al. (2005) did not include aid in their list of 145 separate variables which they found to be significant in growth regressions (with a typical sample of around 100 observations). The instrumentalization of aid and lack of consistency in terms of what it should be doing and how, seem to explain much of its indolence in solving poverty issues. Aid projects were criticized for not integrating with the recipient's infrastructures, or the societies ("white elephants" projects, e.g. Brunel, 1993), or simply not working at all (Banjerlee, 2007; Cohen & Easterly, 2009). Aid failures were often caused by the lack of funds for long-term sustainability. There is evidence of the aid initiatives which provided fleeting improvements, which eroded after the project finished (Ovadia, 2015). The aid's image crisis reached its peak in the context of the "lost decade" of the 90s, when especially African countries faced rising problems of poverty, debt, and conflicts.

It is worth noting, that not only aid itself, but also aid research receives a lot of criticism. Easterly (2009) divides approaches in aid literature to transformational and marginal, while Rodrik (2009) distinguishes between micro-evaluation enthusiasts and growth-diagnosticians. Transformational approach is criticized because it is based on low quality data and provides little guidance about how to adjust any intervention to work better. According to Easterly: "there are too many things changing at once to know what caused the failure" (2009, p. 377). As opposed to the large sample cross-country studies, development economics turned towards randomized evaluation (RE) of particular interventions in the academic literature (which lies in the scope of Easterly's marginal approach) (Duflo, Kremer, & Robinson, 2008; 2009). The RE research is based on micro data (as opposed to aggregate econometrics), brings the risk of bias in "random" project selection (Ravaillon 2009) and the possibility of groundless extrapolation of its conclusions to general aid policies. Ravaillon (2009) claims that the "randomistas" methodology may help solving some issues in development economics, but the random field experiment method may not be sufficient to determine that a given development intervention is or is not working in general.

Finally, there is the science and policy nexus, which impacts the issue of development aid. Donors, particularly from DAC, follow the pendulum of views on desirable development paths presented in economic literature. The lines of Berg report, Washington consensus, Burnside & Dollar "sound policies paper" (2000), or lately Lin's New Structural Economics, were setting the standards for development policies (not necessarily thinking). As noticed by Easterly (2009), aid ideas have often been cyclical, with the same ideas going out of fashion only to come back again many years later. Volatility of ideology/paradigm behind the aid system (e.g. market paradigm vs. state paradigm) indicates lack of long-term strategies which would allow any development reforms to be actually implemented, ameliorated and tested. Aid and development recipes try to keep up with the state and market failure debate, unfortunately without serious reflection. This was exemplified by the structural adjustment episode in Africa (Devarajan & Kanbur 2013).

To sum up, based on the literature query, we find evidence, that aid flows have the potential to address both factor accumulation and productivity growth (Abegaz, 2005). For the countries with insufficient inflow of FDIs and trade, or limited capacity to absorb technology coming via commercial channels, aid may be an alternative technology source. Aid flows combine the non-profit and semi-profit capacity building instruments of technology promotion in the less economically advanced countries. A range of different aid flows may impact both tangible and intangible assets in recipient countries, with a potential to improve the quality of inputs in production process. Hence, it is plausible that aid may contribute to the overall economic system' capacity to develop by increasing its productivity. Since the new development economics should be concentrated on seeking "what works for development" (Rodrik, 2009, p. 2), an attempt to evaluate aid flows impact on productivity seems a justifiable task, especially in the context of Sub-Saharan Africa. Based on the TFP literature (e.g. Ascari, 2004; Byrne et al., 2005, Comin, 2008, Ssozi & Assongu, 2016) we intend to apply a cross-country GMM panel model, which seems to better suit the research problem stated in our study and examine the hypothesis that flows of aid are susceptible to impact productivity.

3. Data

In the research, we used data from Penn World Tables (PWT), OECD International Development Statistics, and World Development Indicators (WDI). The measure of TFP was taken from the PWT (see Feenstra et al., 2013 for the details of the measure computation). The data covered 24 countries from Sub-Saharan region over the years 1995-2014. The selection of 24 out of over 40 Sub-Saharan countries was motivated by the availability of TFP data in PWT. The choice of the time frame is dictated by the idea to:

(a) verify the impact of aid on productivity since the general launch of democratic systems in a number of Sub-Saharan states in mid 90s and

(b) test the hypothesis, that aid evaluation programs inspired by severe criticism in the 90s increased its efficiency (Carbone, Memoli, & Quartapelle, 2016).

We expect, that improvements in donor policy and in recipients' governance could have solved the fungibility problem at least partially (see Easterly, 2009), and hence could have increased aid flows efficiency. With the spread of democracy in Africa the improvements in structure of available resources allocation is more likely (McMillan et al., 2014).

As explanatory variables we used the data on total ODA received by African countries over the period 1995-2014, technical ODA, and loans from DAC countries - according to OECD statistics (see Appendix for detailed data description). The decomposition of ODA was based on OECD methodology in reporting aid flows; such approach was earlier adopted in similar research (e.g. Kang & Won, 2017). All the explanatory variables were expressed in US 2015 dollars. Next, using the PWT data on population we expressed the variables in per-capita values. In Table 1 we present the descriptive statistics of the data. We notice that the most volatile are the inflows of loans (with either positive or negative sign, when the country repaid its debt).

Total productivity factors were expressed in categories of purchasing power parity, where US=1. When we analyse the values in Table 1, the variable does not appear to vary substantially across the sample. However, we may detect the actual variability of the data in Figure 1. The maximum value was obtained in the case of Gabon in 1995, and the country appears to be the productivity leader in the region. Very high values of productivity are noted also in the case of Kenya, Nigeria, Tanzania, Swaziland, while the lowest - in Burundi, Niger and Rwanda.

In Table 2 we present the list of the countries which are covered by the study. We associated a number with the name of each country in the sample. These numbers are used in some of the following charts, where the usage of the country's names would make the figure unclear.

In Figure 2 we present the amount of technical aid received by Sub-Saharan African countries over the 1995-2014 period. The highest volume of such help per capita was given to Gabon (number 8), Namibia (number 13), Senegal (number 17) and Botswana (number 2). Nigeria and Sudan were the countries which received the lowest amount of technical aid per capita. We checked, if the per capita amounts of help are simply associated with the population variables in these countries. Indeed, Gabon, Namibia, and Botswana are relatively low populated, but on the contrary, Senegal has a relatively high population and density of population. Nigeria and Sudan are highly populated economies, which may decrease their per capita values, compared to the sample average. On the other hand, it is noteworthy, that Gabon, Namibia and Botswana are among best performing Sub-Saharan countries in terms of Human Development Index and Global Competitiveness Index (2017). Particularly, Botswana is perceived the "growth miracle" African economy, with well-functioning democracy.

In Figure 3 we present the box plot of total net ODA per capita received by each country. We observe that the highest values of the overall foreign help were sent to Gabon (number 8) and Namibia (number 13). Large economies, such as South Africa (regional leader in terms of economic development and industrialization) and Nigeria (largest economy) received much smaller amounts of aid. However, also low income countries like Burkina Faso and Niger were among moderate per capita aid recipients. We may assume that the population density does not seem to constitute a bias. Interestingly, neither does the income of recipients. A closer look at aid data provides more detailed picture. For example, Botswana was generally obtaining comparably moderate foreign aid over the studied period - however the dots in the picture inform us that it received large irregular "injections" of official development aid, such as debt forgiveness in 2008, or infrastructure grant in 2012. Comparing the two plots we can see that for instance Niger received small amount of help in total but compared to other economies - most of the money received was devoted for technological development.

Eventually in Figure 4 we present the net value of loans received by the economies. Compared to the amount of foreign help the values are not high. Lots of countries had to re-pay its foreign debt in the period over study, such as Gabon, Cameroon, Cote d'Ivoire or Mauritania. We may assume that it will be difficult to interpret the impact of loans net value. When we compare the total productivity factors (Figure 1) and the amount of help devoted to technological development, received by SSA economies, we can notice that the two productivity leaders: Gabon and Botswana are also the receivers of very high amount of technological aid. Thus, the research questions formulated in the introduction of the article seems to be empirically justified.

4. The model

In order to utilize all available information, changes of variables in time and across countries we composed a panel of the data, including T = 20 time periods (1995-2014) and N = 24 cross-section units (24 countries listed in Table 2). As the data exhibited strong autocorrelation, we used dynamic panel approach. Thus, the model has the following form:

[y.sub.i,t] = [alpha][y.sub.i,t-1] + [[beta][X.sub.i,t] + [[eta].sub.i] + [u.sub.i,t] (1)

Where, [y.sub.i,t] denotes total factor productivity of the i- th African country at time t. In our case t = 1, ...,20, while i = 1, ...,24. [beta] is a 1 x k vector of the coefficients, [alpha] is the vector of autoregressive coefficients, [X.sub.it] is a k x 1 vector of explanatory variables observed for country i at time t, [[eta].sub.i] are so called individual effects (time-invariant), while [u.sub.i,t] - the disturbance specific for country i at time t. Such models are estimated using either Arellano & Bond (1991) or Blundell & Bond (1998) estimators, especially in the case of macro-panels when T is much smaller than N.

In our case the difference between N and T was not high, therefore we decided to test also other specifications to check robustness of our results. For instance, Judson & Owen (1999) showed that when T is not very small compared to N, the fixed-effect (FE) estimator is not very biased (see also Nickell, 1981). They performed Monte Carlo experiments and showed that the bias of FE increases with [alpha] and decreases with T (see also Baltagi & Kao, 2001). Below we describe three estimators used in our study.

4.1. Arellano and Bond (1991) difference estimator

The model can be estimated using the so called "difference estimator". The main idea of the approach is to first difference the data. Then, one obtains:

[DELTA][y.sub.i,t] = [alpha][DELTA][y.sub.i,t-1] + [beta][DELTA][X.sub.i.t] + [DELTA][u.sub.i,t] = [gamma][W.sub.i,t] + [DELTA][u.sub.i,t] (2)

The error term of the equation (2) is by definition autocorrelated and also correlated with the lagged dependent variable. All values of [y.sub.i,t-k] with k > 1 can be used as instruments for [DELTA][y.sub.i,t-1]. One-step estimator of equation (2) amounts to computing (Cottrel & Lucchetti, 2014):

[mathematical expression not reproducible] (3)

where:

[mathematical expression not reproducible] (4)

Once the 1-step estimator is computed, the 2-step estimated are obtained through replacing the matrix H with the sample covariance matrix of the estimated residuals. The 2-step estimator is consistent and asymptotically efficient.

4.2. GMM-system estimator of Blundell and Bond (1998)

Blundell & Bond (1998) proposed so-called "system" estimator that complements the differenced data with data in levels, so the lagged differences are used as instruments. The system GMM estimator shows dramatic efficiency gain over the "difference"-one, especially when [alpha] [right arrow] 1 (Baltagi & Kao, 2001). The key equation of the system estimator is as follows (Cottrel and Lucchetti, 2014):

[??] = [([N.summation over (i=1)] [[??]'.sub.i][[??].sub.i])[A.sub.N][([N.summation over (i=1)] [[??]'.sub.i][[??].sub.i])].sup.-1][([N.summation over (i=1)] [[??]'.sub.i][[??].sub.i])[A.sub.N]([N.summation over (i=1)] [[??]'.sub.i][[DELTA][[??].sub.i])] (5)

[mathematical expression not reproducible]. (6)

The choice of matrix H* is not trivial. The details are presented for instance in Roodman (2009) or Hsiao (2014).

4.3. Fixed-effect estimator

Let us return to the specification (1). The model can be estimated in two ways. In the first method [[eta].sub.i]s are treated as fixed parameters and a dummy variable is introduced into the model for each group i. This is called Least Squares Dummy Variables (LSDV) method. In the second approach the data are de-meaned and a model without constant [[eta].sub.i] is estimated. The dependent variable is defined as:

[[??].sub.it] = [y.sub.it] - [[??].sub.i], (7)

and the group-mean:

[[??].sub.i] = 1/[T.sub.i][[T.sub.i].summation over (t=1)][y.sub.it]. (8)

Where, [T.sub.i] is the number of observations for unit i. The two approaches are numerically equivalent (see: Hsiao, 2013). However, we can compute two kinds of [R.sup.2]. LSDV [R.sup.2] is the measure of fit of the model with binary variable, while "within" [R.sup.2] - the measure of goodness of fit of the model for [[??].sub.it], i.e. assuming that we are interested in the influence of other factors apart from individual effects (Cottrel & Lucchetti, 2014, see also: Greene, 2011).

5. Results

As an opening step the data was tested for non-stationarity through Levin-Lin-Chu (1992) test, and the null hypothesis of unit root was rejected in the case of the test without intercept, with intercept, as well as with intercept and linear trend (and lag=1). Due to data autocorrelation, we estimated the dynamic panel model. This method has also empirical explanation - we expect that the today's total factor productivity will depend on its value from previous year. Such model is also commonly applied in the case of macro-data (Anoruo & Elike, 2015; Foster-McGregor et al., 2016). A dynamic panel model was estimated of the following form:

CTF[P.sub.sub.i,t] = [[eta].sub.i,t] + [[beta].sub.1]CTF[P.sub.i,t-1] + [[beta].sub.2]LOANsP[C.sub.i,t] + [[beta].sub.3]techODAp[c.sub.i,t] + [[beta].sub.4]netODAp[c.sub.i,t] + [u.sub.i,t], (9)

here CTFP_(i,t) CTF[P.sub.i,t]--denotes the value of total productivity factor in county i at time t, techODAp[C.sub.i,t] is the amount of technical ODA in 2015 US dollars received per 1 mln citizens, netODAp[c.sub.i,t] - net ODA received per 1 mln citizens and LOANsP[C.sub.i,t] - value of loans received per 1 mln citizens.

In the first approach, two-step system-GMM approach with asymptotic errors was applied. To check the robustness of the results, also the following specifications were tested:

a. Two-step approach with asymptotic errors,

b. Two-step approach with normal errors,

c. Two-step GMM-system approach with normal errors,

d. One-step estimation - which did not pass the Sargan overidentification test and is not reported here,

e. Fixed-effect approach with lagged dependent variable.

The results of the performed estimations are presented in Table 3. Depending on the specification different sets of significant explanatory variables were obtained. However - regardless of the specification technical ODA per capita appeared to be a significant explanatory variable (in the case of Specification 4, when the model was estimated using GMM-system approach and asymptotic errors are not assumed - with p-value of 0.062 and in the case of fixed-effect specification - of 0.056). ODA loans received by the beneficiary country proved to be significant variable only in the case of the first specification and based upon the sign of the coefficient we could say that they contributed to the increase of productivity. Eventually, total ODA received was a significant variable only in the Specification 2, and the negative coefficient could suggest that their usage is not effective when increase of productivity is considered.

To summarize - although the total net ODA received and net loans received do not seem to significantly contribute to the productivity growth (which corroborates the results obtained by other researchers i.a. Byrne et al., 2005; Alvi & Senbeta, 2012), technical ODA received by African economies proved to stimulate productivity growth, which extends the results of Ssozi & Asongu (2016). As these flows are devoted to technical development, we can suppose that technical cooperation contributes to productivity growth through technology absorption and development including increasing skills and capacity.

All the specifications passed the panel diagnostic tests. The tests for Specifications 1-4 are presented in Table 4. In each case order 2 autocorrelation was not present in the errors; all models passed Sargan overidentification test, as well as Wald test. In the case of the Specification 5 the LSDV [R.sup.2] amounted to 0.96, while the within-one to 0.63. The null hypothesis of Hausman (1978) test (that the groups have a common intercept) was rejected with p-value lower than 0.001. Eventually, the specification 6 was estimated using general FGLS estimator of Wooldridge (2002). This estimator is robust against any type of intragroup heteroscedasticity and serial correlation (Croissant & Millo, 2008). Additionally, to account for cross-sectional dependence, we computed robust errors (using Arellano estimator) in the case of specification 5. The test confirmed significance of ctfp(-1) with p-value <0.0001 and techODApc with p-value of 0.002.

Further robustness check in this research was done in two steps. First, we controlled for the business cycle, next we included several control variables into estimated equations.

In order to take into account possible influence of the business cycle on the results we re-estimated the model for averages of variables over 5-years periods (see e.g. Islam, 1995). According to the results of Hausman (2009) and Breusch-Pagan (1980) tests, we estimated the random-effect model. The results are presented in Table 5.

The Pesaran (2015) test for cross-sectional dependence returned p-value of 0.1 (we do not reject the null hypothesis of no cross-sectional dependence), while the Breusch-Godfrey (Breusch, 1978; Godfrey, 1987) test for serial correlation: p-value <0.001 (we reject the null hypothesis of no serial correlation). Thus, we re-estimated the model using robust covariance matrix of parameters for a panel model according to the Beck & Katz (1995). The resulting errors and p-values are presented in two last columns of Table 7.

In order to extend the robustness check of our results, we tested yet another specification of the model, including a list of control variables associated with technological change and growth. List of such variables used in empirical studies is very long. For instance, Durlauf, Johnson, & Temple (2006) present a number of variables used in growth equations in Solow and not-Solow models. The variables are usually associated with such categories, as institutional quality, political stability, human capital, financial development, initial level of technological development, trade openness, etc. To check the robustness of our results we included control variables from the categories presented in Table 6 in the model.

We assumed that high values of indices of political stability, as well as corruption control, should be associated with high values of total factor productivity. The good political climate supports entrepreneurship and foreign investment, lowering the so called "political risk". The impact of institutions on the expected levels of productivity is confirmed in the literature (i.a. Chanda & Dalgaard, 2008).

Initial level of infrastructure development and technology readiness is approximated by the variable "access to electricity" (percentage of population with access to electricity). The variable is much diversified across the group - see Figure 5. For instance, in the case of Burundi it takes value lower than 10% (number 4), while in the case of Gabon and South Africa - over 80%. We are aware of the fact that the variable does not encompass the technological level development - therefore we define the variable "lack of electricity access". At the same time we assume that it may approximate limited access to basic infrastructure for business. We assume that high values of this variable will be associated with low values of total factor productivity (Frija et al., 2015).

Next, we wanted to include human capital as a control variable. Usually secondary or even tertiary school enrolment is taken into account (Mankiw, Romer, & Weil, 1992; Easterly & Levine, 1997; Barro & Sala-i-Martin, 2004), but for Sub-Saharan countries relatively good quality data was available only for primary education. We consulted the literature and based on the evidence presented by Klenow & Rodriguez (1997) and Krueger & Lindahl (2001) decided to accept primary enrolment as a human capital variable. This was also justified by the fact that the study covers a region were primary enrolment ratio is very diversified across countries, which increases its discrimination value (see Figure 6). It takes the lowest values in the case of Niger (where also the literacy rate is one of the lowest in the Sub-Saharan Africa), while the highest - in the case of Gabon. Again, as we are aware of the fact that the variable does not fully encompass the level of human capital, we define the opposite variable - no enrolment to primary school lackOfPrimaryEdu = (100 - PrimarySchoolEnrollment). The variable catches the underdevelopment of human capital, and we assume that it will be associated with low values of total factor productivity (see: de la Fuente & Donenech, 2006).

When it comes to financial market development, we took the variable of access of private enterprises to credit. It measures the ability of financial sector to provide financial services and the density of such operations in the market perspective (to GDP ratio). Such quantification of the extent of financial market development is acceptable from the point of view of the study. The financial development and productivity correlation are presented in Figure 7.

We observe that the most financially developed economy is South Africa. Banking systems and the use of credit in all other economies in the region is lesser - even in the case of Gabon that was the leader in the previous categories. Apart from the South Africa, the highest value of the variable is observed in the case of Namibia and Zimbabwe. We assumed that high value of financial development should be associated with higher value of total factor productivity, as the access to credit stimulates individual entrepreneurship.

Eventually, we take into account trade openness (Chanda & Dalgaard, 2008), measured through the ratio of trade to GDP (Figure 8). The highest values of the variable were observed for Swaziland, while the lowest for Burundi, Niger and Rwanda. No data were available for Lesotho. We assume that the higher value of trade openness should be associated with higher value of total factor productivity. Following Chanda & Dalgaard (2008) and Erken, Donselaar, & Thurik (2018) we assume that the openness of an economy implies a higher level of competition from abroad. This functions as an incentive for firms to innovate, given they dispose of a certain amount of R&D capital (see also Edwards, 1998). Furthermore, more competition stimulates firms to reduce their X-inefficiencies (Leibenstein, 1966).

Results of the estimation for each specification are presented in Table 7. In the case of each specification technical aid per capita appeared to be significant (in the case of models where asymptotic errors were not assumed - at least at 10% significance level). In each specification, the signs of parameters were economically significant, i.e. lack of electricity and lack of primary education were associated with lower productivity, while trade openness, development of financial market, political stability - with higher productivity values. This suggests that in a state where government increases access to education and infrastructure and fosters the development of market institutions, technical aid may be absorbed and contribute to total factor productivity. In the case of each specification high values of technological aid received were associated with higher productivity.

Eventually, to check the validity of the estimates, we re-estimated the model, taking into account 5-year averages of the TFP and the control variables and using the robust covariance matrix. The results are presented in Table 8. We observe that technical aid still remains the significant explanatory variable.

In order to explain the lack of the impact of the overall net ODA provided by DAC countries to Sub-Saharan Africa in the studied period we refer to the concepts of humanist interpretation and public choice theory. Both of these theoretical frameworks assume rationality of agents in realization of their goals (Buchanan & Tullock, 1962; Kmita, 1991). Generally, aid policies tend to be motivated by donor interests rather than recipient needs (Dreher et al., 2016). Hence, limited effects of aid are in fact consistent with the development cooperation policy. Only by changing the motivation of the system, aid effectiveness may be increased. However, positive impact of technical aid on productivity detected in the period 1995-2014 suggests, that direct actions may be less affected by the general malfunctions of the development cooperation system. Well-designed instruments may avoid syndromes described i.a. by Moyo (2009) or Fosu (2013). Technical cooperation which involves person to person relations, appeared to be more productive, than the financial aid transfers administrated by both donors and beneficiary bureaucratic structures. It seems that the individuals are motivated by utility-maximizing considerations in the technical projects realisation. Successful projects create an opportunity for mutual gains for the agents (project employees and local community) in the field project operations (Buchanan & Tullock, 1962).

It seems that the grass root level of "trading" aid assets may contribute more to the overall economy, than the state level allocation of financial assets where bureaucratic interests and top-down planning overwhelm markets and fail. This is consistent with Malunga (2014), who claims that it is the clarity of the issues of culture, identity, and spirituality that enables people to articulate their agenda and the type of development they want. This issue is especially pivotal while engaging in relations with external entities, often stronger parties, like aid donors. Technical cooperation, as a type of intervention involves human interactions, which can evolve in time. People working on projects get more involved and find it rationale to engage in aid activities. They gain the advantage of a local perspective, over the "European office" perspective. Effectiveness of technical cooperation may be further explained in the light of the results of Walley & Cushing (2013) research'. They indicate that development aid directed at activities which enhance knowledge accumulation and skills development of the work force are more likely to improve productivity and long-run growth in Sub-Saharan African countries. Therefore, technical aid is an instrument likely to empower local communities, which are involved in project realisation.

At the end we want to stress that there are several limitations of this study, especially the data availability for the countries from Sub-Saharan Africa. It would be very interesting to analyse the impact of sectoral aid on productivity. However, the data on decomposed aid is available only from 2005. Also, due to data unavailability not all Sub-Saharan African economies were included in the study (see Table 2). Next, net ODA is itself aggregated and technical aid flows are included in this aggregate. Similarly, the data on loans received is also of aggregated nature - the loans may be received for various purposes - not only to stimulate productivity.

6. Conclusions

The goal of the study was to assess the empirical influence of OECD DAC aid flows on productivity growth in Sub-Saharan Africa over the period 1995-2014. 24 countries from the region were included in the study. Based on the dynamic panel model estimation we found that - regardless of the model specification - technical aid flows are significantly linked to increased productivity in the region. Thus, we suppose that technical cooperation spurs technological progress in Sub-Saharan Africa. This may suggest that the ongoing criticism of aid may have resulted in some types of aid activities.

Based on the development cooperation experiences from the 20th century in Sub-Saharan Africa, not many scholars (nor politicians) expect development aid to actually work. Therefore we decided to check if the declarations of DAC countries to improve aid effectiveness in the New Millennium were implemented in recent aid practice. Despite the initial optimism, the GMM panel analysis of the impact of aid flows on productivity in 24 Sub-Saharan countries suggests, that indeed the aid system generally does not work. However, based on our results, we claim that technical cooperation is the exception. A positive and significant effect of technical cooperation on productivity in the studied period suggests, that actions which are oriented directly towards technical progress in recipient countries create better effects for productivity than other aid instruments. Hence, we suppose that the mechanisms of aid distribution and the degree of agents' involvement in the process are crucial for aid effectiveness. This is valid in the context of the rent seeking phenomena in the recipient and the donor countries and the problem of subordination of aid agenda to foreign policy and personal gains. We argue that it is vital to reflect over aid practice rather than compromise it a priori.

Some policy implications may be formulated based on the results of this study. First, decreasing the poverty gap by short-term aid initiatives is unlikely. The effectiveness of technology transfer via development aid channels requires establishment of long-term mechanisms for knowledge, know-how and technology absorption. Second, growing threat of terrorism, social exclusion, and illegal economic migrations prove, that refusing to admit aid subjection to donors' economic and geostrategic policies has backfired. If not for good reasons, then at least from common-sense, aid system should be concentrated on recipients' needs and aimed especially at development of productivity-oriented technology. Fostering the capacity of aid recipients to develop on their own technology may initiate the indispensable structural changes of their economies. Based on the evidence presented in this paper, the adjustment of foreign impact to local demand and capacity appears to be indispensable to translate technology into productivity increase in the recipient country.

Some states are more independent in shaping their policies, while others depend on foreign entities (transition countries in Central Europe being a clear example of that) (Berglof, 2015). In the case of the low- and middle-income African countries, their policies were under a strong influence of international financial institutions and the DAC community in the modern history. Development assistance system was exploited instrumentally for that purpose (Chauvet & Collier, 2009). However lately, the hegemony of Western thought has been destabilized under growing pressure of China presence in Africa. The idea of economic liberalization without democracy grows to be as an alternative for Sub-Saharan states and their leaders. The dominance of the Western type of thinking in international economics in the following years is threatened and there appears temptation to use the aid as a tool of interests (not as a development driver). Development cooperation once again may play an important role in the international geopolitics, not in poverty fight.

Acknowledgment

The research is supported by Polish Ministry of Science and Higher Education through the Project SONATA: 2013//09/D/HS4/01849 (Mechanisms of technology absorption in Sub-Saharan Africa). The authors would like to thank the participants of the International Conference on Emerging Markets Economies--Euroconference 2016 (29.06-01.07.2016, Porto, Portugal) for the discussion and comments on the early version of this paper.

References

Abegaz, B. (2005). Multilateral development aid for Africa. Economic Systems, 29(4), 433-454.

Acemoglu, D., Johnson, S., Robinson, J. (2001). The colonial origins of comparative development: An empirical investigation. American Economy Review, 91(5), 1369-1401.

Aghion, P. H. (1992). A model of growth through creative destruction. Econometrica, 60(1), 323-51.

Alvi, E., Senbeta, A. (2012). Foreign aid: good for investment, bad for productivity. Oxford Development Studies, 40(2), 139-161.

Andrzejczak, K., Kliber, A. (2015), The model of French development assistance: who gets the help? Dynamic Econometric Models, 15, 89-109.

Anoruo, E., Elike, U. (2015). Human capital-economic growth nexus in Africa: heterogeneous panel causality approach. International Journal of Economics and Financial Issues, 5(4), 1017-1023.

Anyanwu, J. C. (2014). Factors affecting economic growth in Africa: are there any lessons from China? African Development Review, 26(3), 468-493.

Arellano, M., Bond, S. (1991). Some tests of specifiction for panel data: Monte Carlo evidence and an application to employment equations. Revue of Economic Studies, 58, 277-297.

Ascari, G., Di Cosmo, V. (2005). Determinants of total factor productivity in the Italian regions, Scienze Regionali/Italian Journal of Regional Science, 2, 27-49.

Baltagi, B. H., Kao, C. (2001). Nonstationary panels, cointegration in panels and dynamic panels: A survey. In: Baltagi, B.H., Fomby, T.B., Hill, C. (Eds.) Nonstationary panels, panel cointegration and dynamic panels. Bingley: Emerald Insight.

Barro, R., Sala-i-Martin, X. (2004). Economic growth (2nd ed.), Cambridge: MIT Press.

Beck, N., Katz, J. (1995). What to do (and not to do) with time-series cross-section data in comparative politics. American Political Science Review, 89(3), 634-647.

Berglof, E. (2015). New structural economics meets European transition. Journal of Economic Policy Reform, 18(2), 114-130.

Birdsall, N., Rodrik, D., Subramanian, A. (2005). How to help poor countries. Foreign Affairs, 84(4), 136-152.

Blundel, R., Bond, S. (1998). Initial conditions and moment restrictions in dynamic panel data models. Journal of Econometrics, 87(1), 115-143.

Boone, P. (1996). Politics and the effectiveness of foreign aid. European Economic Review, 40, 289-329.

Breusch, T. S. (1978). Testing for autocorrelation in dynamic linear models. Australian Economic Papers, 17(31), 334-355.

Breusch, T. S., Pagan, A. R. (1980). The Lagrange multiplier test and its applications to model specification in econometrics. Revue of Economic Studies, 47(1), 239-253.

Brunel, S. (1993). Le gaspillage de l'aide publique. Paris: Editions du Seuil.

Buchanan, J. M., Tullock, G. (1962). The calculus of consent: logical foundations of constitutional democracy, Ann Arbor: University of Michigan Press.

Burnside C, Dollar, D. (2000). Aid, policies and growth. American Economic Review, 90(4), 847-869.

Byrne J., Fazio G., Piacentino D. (2005). Convergence in TFP among Italian regions: Panel unit roots with heterogeneity and cross sectional dependence, ERSA conference paper No. 591, European Regional Science Association.

Carbone, G. M., Memoli, V., Quartapelle, L. (2016). Are lions democrats? The impact of democratization on economic growth in Africa, 1980-2010. Democratization, 23(1), 27-48.

Chanda, A., Dalgaard, C. J. (2008). Dual economies and international total factor productivity differences. Economica, 75(300), 629-661.

Chauvet, L., Collier, P. (2009). Elections and economic policy in developing countries. Economic Policy, 24(59), 509-550.

Chenery, H. B. (1969). The two gap approach to aid and development: reply to Bruton, American Economic Review, 59(3), 446-449.

Chengchun, L., Sailesh, T. (2018). The impact of foreign direct investment on productivity: New evidence for developing countries. Economic Modelling, Available online 1 December 2018, https://doi.org/10.1016/j.econmod.2018.11.028

Cohen, J., Easterly, W. (2009). Introduction, In: Cohen, J., Easterly, W. (Eds.), Thinking big versus thinking small. What works in development?. Brookings Institution Press.

Comin, D. A., Mestieri, M. (2014). Technology diffusion: Measurement, Handbook of Economic Growth 2, 565-622.

Comin, D. A. (2008). Total factor productivity. In: Durlauf S. N., Blume L. E. (Eds.) The New Palgrave dictionary of economics, Second Edition, New York: Palgrave Macmillan.

Comin, D. A., Hobijn, B., Rovito, E. (2006). Five facts you need to know about technology diffusion (NBER Working Paper 11928). National Bureau of Economic Research.

Cottrel, A., Lucchetti, R.J. (2014). Gretl User's Guide. GNU regression. Econometrics and time-series library. Retrieved August 01, 2017, from http://gretl.sourceforge.net

Croissant, Y., Millo, G. (2008). Panel data econometrics in R: The plm package. Journal of Statistical Software, 27(2), 1-43.

de la Fuente, A., Donenech, R. (2006). Human capital in growth regressions: how much difference does data quality make? Journal of the European Economic Association, 4(1), 1-36.

Delechat C., Ramirez G., Wagh S., Wakemann-Linn J. (2010). How global financial markets affect Sub-Saharan Africa (IMF Staff Papers, 57-1), Washington D.C.: International Monetary Fund.

Devarajan, S., Kanbur, R. (2013). The evolution of development strategy as balancing market and government failure (Working Paper 09), Charles H. Dyson School of Applied Economics and Management, Cornell University.

Dreher, A., Eichenauer, V. Z., Gehring, K. (2016). Geopolitics, aid, and growth: The impact of UN Security Council membership on the effectiveness of aid (World Bank Policy Research Working paper, WPS 7771). Washington, D.C.: World Bank Group.

Duflo, E., Kremer, M., Robinson, J. (2008). How high are rates of return to fertilizer? Evidence from field experiments in Kenya. American Economic Review, 98(2), 482-88.

Duflo, E., Kremer, M., Robinson, J. (2009). Nudging farmers to use fertilizer: Evidence from Kenya. American Economic Review, 101, 2350-2390.

Durlauf, S. N., Johnson, P. A., Temple, J. R. W. (2005). Growth econometrics. In P. Aghion, S. N. Durlauf (Eds.), Handbook of Economic Growth, Vol. 1A, Amsterdam, San Diego: Elsevier.

Easterly, W., Levine, R. (1997). Africa's growth tragedy: Policies and ethnic divisions. Quarterly Journal of Economics, 112(4), 1203-50.

Easterly, W. (2009). Can the West save Africa? Journal of Economic Literature, 47(2), 373-447.

Edwards, S. (1998). Openness, productivity and growth: what do we really know? Economic Journal, 108(447), 383-398.

Erken, H., Donselaar, P., Thurik, R. (2018). Total factor productivity and the role of entrepreneurship. Journal of Technology Transfer, 43(6), 1493-1521

Farahmand, S., Sameti, M., Sasan, S. S. (2014). Spatial variations of [beta]-convergence coefficient in Asia (The GWR Approach), Iranian Economic Review, 18(1), 81-101.

Feenstra, R. C., Inklaar, R., Timmer, M. P. (2013). The next generation of Penn World Table. Online Appendix. Retrieved July 26, 2017, from https://assets.aeaweb.org

Fosu, A. K. (2013). Growth of African economies: productivity, policy syndromes and the importance of institutions. Journal of African Economics, 22(4), 523-551.

Foster-McGregor, N., Isaksson, A., Kaulich, F. (2016). Importing, productivity and absorptive capacity in Sub-Saharan African manufacturing and services firms. Open Economies Review, 27(1), 87-117.

Frija, A., Dhehibi, B., Aw-Hassan, A., Akroush, A., Ibrahim, A. (2015). Approaches to total factor productivity measurements in the agriculture economy. CGIAR Research on Dryland Systems, Retrieved April 5, 2019, from: http://repo.meil.cgiar.org

Godfrey, L. G. (1978). Testing against general autoregressive and moving average error models when the regressors include lagged dependent variables. Econometrica, 46(6), 1293-1301.

Graham, B. S., Temple, J. R. W. (2006). Rich nations, poor nations: how much can multiple equilibria explain? Journal of Economic Growth, 11(1), 5-41.

Greene, W. H. (2011). Econometric analysis. 7th edition, New York: Pearson.

Grossman, G. M., Helpman, E. (1991). Innovation and Growth in the Global Economy. Cambridge, MA: MIT Press.

Hausman, R. (2009). The other invisible hand: high Bandwidth development policy. In: Easterly, W., Cohen, J. (Eds.) What Works in Development? Thinking Big and Thinking Small. Washington, D.C.: Brookings Institution Press.

Hausman, J. A. (1978). Specification tests in econometrics. Econometrica, 46(6), 1251-1271.

Hsiao, Ch. (2014). Analysis of panel data. 3rd edn., Cambridge: Cambridge University Press DOI: http://dx.doi.org/10.1017/CBO9781139839327.

IMF (2003). External debt statistics: Guide for compilers and users - Appendix III, Glossary, International Monetary Fund, Washington DC.

Irandoust, M., Ericcson, J. (2005). Foreign aid, domestic savings, and growth in LDCs: An application of likelihood-based panel cointegration. Economic Modelling, 22(4), 616-627.

Islam, N. (1995). Growth empirics: A panel data approach. Quarterly Journal of Economics, 110(4), 1127-1170.

Jajri, I. (2007). Determinants of total factor productivity growth in Malysia, Journal of Economic Cooperation, 28 (3), 41-8.

Judson, R.A., Owen, A.L. (1999). Estimating dynamic panel data models: A guide for macroeconomists. Economic Letters, 65(1), 9-15.

Kang, G. S., Won, Y. (2017). Does Korea's Official Development Assistance (ODA) promote its FDI? Journal Of Economic Research, 22(1), 23-46.

Kaufmann, D, Kraay, A., Mastruzzi, M. (2010). The worldwide governance indicators: Methodology and analytical issues (World Bank Policy Research Working Paper No. 5430). Washington D.C.: World Bank.

Klenow, P., Rodriguez, A. (1997). The neoclassical revival in growth economics: Has it gone too far? In: Bernanke B. S., Rotemberg J. (Eds.), NBER Macroeconomics Annual 1997, MIT Press.

Kmita, J. (1991). Essays on the theory of scientific cognition, Warszawa, Dordrecht, Boston, London: PWN - Polish Scientific Publishers, Kluwer Academic Publishers.

Krueger, A. B., Lindahl, M. (2001). Education for growth: Why and for whom? Journal of Economic Literature, 39(4), 1101-1136.

Kumar, S., Pacheco, G. (2012). What determines the long-run growth rate in Kenya?. Journal of Policy Modeling, 34(5), 705-718.

Leibenstein, H. (1966). Allocative efficiency vs. "X-efficiency", American Economic Review, 56(3), 392-415.

Levin, A., Lin, C.F., Chu, C.-S.J. (2002). Unit root test in panel data: asymptotic and finite sample properties. Journal of Econometrics, 108(1), 1-24.

Lin, J. Y., Monga, C. (2010). The growth report and new structural economics (World Bank Policy Research Working Papers Series, 5336). Washington, D.C.: World Bank.

Lipsey, R. G., Carlaw, K.I. (2004) Total factor productivity and the measurement of technological change. Canadian Journal of Economics, 37(4), 1118-1150.

Lipton, M. (2012). Income from work: the food-population-resource crisis in the 'Short Africa', Leontief Prize lecture, Tufts University, Medford, MA.

Loxley, J., Sackey, H. (2008). Aid effectiveness in Africa, African Development Review 20(2), 163-199.

Malunga, C. (2009). Understanding organizational leadership through Ubuntu. London: Adonis and Abbey Publishers.

Mankiw, D., Romer, D. W. (1992). A contribution to the empirics of economic growth. Quarterly Journal of Economics, 107(2), 407-437

McMillan, M., Rodrik, D., Verduzco-Gallo, I. (2014). Globalization, structural change, and productivity growth, with an Update on Africa. World Development, 63, 11-32.

Moyo, D. (2009). Dead aid: why aid is not working and how there is a better way for Africa, Vancouver: Douglas & Mcintyre.

Nachega, J. C., Fontaine, T. (2006). Economic growth and total factor productivity in Niger (IMF Working Paper, WP/06/208). Washington D.C.: International Monetary Fund.

Nickell, S. (1981). Biases in dynamic models with fixed effects. Econometrica, 49(6), 1417-1426.

Ovadia, J. S. (2015). The petro-developmental state in Africa: Making oil work in Angola, Nigeria and the Gulf of Guinea. London: Hurst.

Papanek, G. (1972). The effect of aid and other resources on savings and growth in less developed countries. Economic Journal, 192, 934-50.

Pesaran, M. H. (2015). Testing weak cross-sectional dependence in large panels. Econometric Reviews Econometric Review, 34(6-10), 1089-1117.

Radelet, S., Clemens, M., Bhavnani, R. (2004). Aid and growth: the current debate and some new evidence. Centre for Global Development, February 2004.

Rajan, R. G., Subramanian, A. (2011). Aid, Dutch disease, and manufacturing growth. Journal of Development Economics, 94(1), 106-118.

Ravallion, M. (2009). Should the Randomistas rule? Economists' Voice, 6(2). Retrieved July 28, 2017, from http://EconPapers.repec.org

Rodrik, D. (2009). The new development economics: We shall experiment, but how shall we learn?. In: Easterly W., Cohen J. (Eds.) What works in development? Thinking big and thinking small. Washington, D.C.: Brookings Institution Press.

Rodrik, D. (2014). An African growth miracle? Retrieved July 15, 2017, from http://drodrik.scholar.harvard.edu

Romer, P. M. (1990). Endogenous technological change. Journal of Political Economy, 98(5), part II, S71-102.

Roodman, D. (2009). How to do xtabond2: An introduction to difference and system GMM in Stata. The Stata Journal, 9(1), 86-136.

Sachs, J. D., McArthur J. W., Schmidt-Traub G., Kruk M., Bahadur Ch., Faye M., McCord G. (2004). Ending Africa's poverty trap. Brookings Papers on Economic Activity, 2004(1), 117-216.

Solow, R. (2001). What have we learned from a decade of empirical research on growth? Applying Growth Theory across Countries. World Bank Economic Review, 15, 283-288.

Solow, R. (2016). Resources and economic growth. American Economy, 61(1), 52-60.

Ssozi, J., Asongu, S. A. (2016). The comparative economics of catch-up in output per worker, Total factor productivity and technological gain in Sub-Saharan Africa. African Development Review, 28, 215-228.

Swierczynska, K. (2019). Uwarunkowania i zroznicowanie rozwoju gospodarczego krajow Afryki Subsharyjskiej [Conditions and diversification of economic development of Sub-Saharan African countries], Warszawa: CeDeWu (original - in Polish).

Walley, B., Cushing, M. (2013). Development ad and foreign R&D spillovers in Sub-Saharan Africa. International Journal of Applied Economcs, 10(2), 10-31.

Wooldridge, J. M. (2002). Econometric analysis of cross-section and panel data. Cambridge, London: MIT Press.

Appendix
TABLE 9. DESCRIPTION OF THE DATA USED IN THE STUDY

VARIABLE                                 DESCRIPTION

ctfp                    TFP level at current PPPs (USA=1)

population              Population in millions

techODApc               Technical ODA received in constant US dollars
                        (2015)--per capita

netODApc                Total ODA received in constant US dollars (2015)
                        --per capita

netLoansPc              Net loans received in constant US dollars (2015)
                        --per capita

                  Control variables

PoliticalStability      Political Stability and Absence of Violence/
                        Terrorism measures perceptions of the likelihood
                        of political instability and/or politically-
                        motivated violence, including terrorism.
                        Estimate gives the country's score on the
                        aggregate indicator, in units of a standard
                        normal distribution, i.e. ranging from
                        approximately -2.5 to 2.5.

CorruptionControl       Control of Corruption captures perceptions of
                        the extent to which public power is exercised
                        for private gain, including both petty and grand
                        forms of corruption, as well as "capture" of the
                        state by elites and private interests. Estimate
                        gives the country's score on the aggregate
                        indicator, in units of a standard normal
                        distribution, i.e. ranging from approximately
                        -2.5 to 2.5.

PrimarySchoolEnrolment  Primary school enrolment (gross). Gross
                        enrollment ratio is the ratio of total
                        enrollment, regardless of age, to the population
                        of the age group that officially corresponds to
                        the level of education shown. Primary education
                        provides children with basic reading, writing,
                        and mathematics skills along with an elementary
                        understanding of such subjects as history,
                        geography, natural science, social science, art,
                        and music.

TradeToGDP              Trade (the sum of exports and imports of goods
                        and services) measured as a share of gross
                        domestic product.

AccessToElectricity     Access to electricity is the percentage of
                        population with access to electricity.
                        Electrification data are collected from
                        industry, national surveys and international
                        sources.

FinDepositsToGDP        Financial system deposits to GDP (%).Demand,
                        time and saving deposits in deposit money banks
                        and other financial institutions as a share of
                        GDP. Raw data are from the electronic version of
                        the IMF's International Financial Statistics.
                        Financial system deposits (IFS lines 24, 25, 44,
                        45, FOST and FOSD); GDP in local currency (IFS
                        line NGDP); end-of period CPI (IFS line PCPI);
                        and average annual CPI is calculated using the
                        monthly CPI values (IFS line PCPI).

DepositMoneyToGDP       Deposit money banks' assets to GDP (%).Claims on
                        domestic real nonfinancial sector by deposit
                        money banks as a share of GDP. Raw data are from
                        the electronic version of the IMF's
                        International Financial Statistics. Deposit
                        money bank assets (IFS lines 22, a-d, FOSAG,
                        FOSAOG, FOSAON and FOSAOP); GDP in local
                        currency (IFS line NGDP); end-of period CPI (IFS
                        line PCPI); and average annual CPI is calculated
                        using the monthly CPI values (IFS line PCPI).

CreditToPrivateSector   Domestic credit to private sector (% of GDP).
                        Domestic credit to private sector refers to
                        financial resources provided to the private
                        sector, such as through loans, purchases of
                        nonequity securities, and trade credits and
                        other accounts receivable, that establish a
                        claim for repayment. For some countries these
                        claims include credit to public enterprises.

VARIABLE                           SOURCE

ctfp                    PennWorldTables

population              PennWorldTables

techODApc
                        OECD

netODApc
                        OECD

netLoansPc
                        OECD

               Control variables

PoliticalStability      World Government Indicators through the World
                        Bank database.
                        The WGI are produced by Daniel Kaufmann (Natural
                        Resource Governance Institute and Brookings
                        Institution) and Aart Kraay (World Bank
                        Development Research Group). See also:
                        www.govindicators.org and Kaufman et al. (2010)

CorruptionControl       World Government Indicators through the World
                        Bank database.

PrimarySchoolEnrolment  United Nations Educational, Scientific, and
                        Cultural Organization (UNESCO) Institute for
                        Statistics.

TradeToGDP              World Bank national accounts data, and OECD
                        National Accounts data files.

AccessToElectricity     World Bank, Sustainable Energy for All (SE4ALL)
                        database from the SE4ALL Global Tracking
                        Framework led jointly by the World Bank,
                        International Energy Agency, and the Energy
                        Sector Management Assistance Program.

FinDepositsToGDP        International Financial Statistics (IFS),
                        International Monetary Fund (IMF)

DepositMoneyToGDP       International Financial Statistics (IFS),
                        International Monetary Fund (IMF)

CreditToPrivateSector   World Development Indicators (WDI), World Bank

Source: Own elaboration.


Katarzyna Swierczynska, (1) Agata Kliber (2)

(1) Faculty of Economics, Poznan University of Economics and Business, Poland

(2) Faculty of Informatics and Electronic Economy, Poznan University of Economics and Business, Poland

corresponding e-mail: kata(dot)swierczynska[at]gmail(dot)com

address: Faculty of Economics, Poznan University of Economics and Business, 61-875, aleja Niepodleglosci 10, Poznan, Poland

(*) In sub-Saharan Africa, the level and the pace of economic growth differ from other continents, but also inside the region (Fosu, 2013; Anyanwu, 2014). Heterogeneity of the region is reflected in a number of socioeconomic features, including human resources, physical infrastructure, private sector development, quality of governance, and institutional capacity (Acemoglu, Johnson, & Robinson, 2001; Birdsal & Rodrik, 2005; Fosu, 2013; Anyanwu, 2014; Rodrik, 2014; Swierczynska, 2019). Despite these differences, if compared to other groups of countries and world regions, Sub-Saharan Africa reserves some shared characteristics, especially in the context of the level of industrialization and technological advancement (Easterly, 2009). Economies in the region are generally commodity exporters with domination of traditional sector and relatively low rates of high technology exports (Lipton, 2012). Their industry is dominated by the extractive sector, particularly in energy and minerals, with limited role of manufacturing. Despite the fact that industrial upgrading based on technological changes is deemed crucial for economic development in the region (Lin & Monga, 2010), the pace of structural change is far too moderate. In terms of institutional development, Fosu (2013) presented evidence of "policy-syndromes" which hinder expected growth rates. Hence, the countries from the region still find themselves in the OECD "developing countries" list, largely in the low and lower middle-income groups; their societies face deep inequality issues and exclusion of vulnerable groups.

DOI: http://dx.doi.org/10.15208/beh.2019.10
TABLE 1. DESCRIPTIVE STATISTICS OF TOTAL PRODUCTIVITY FACTORS AND
FOREIGN DISBURSEMENTS

VARIABLE                ctfp  techODApc  netODApc  LoansPC

Mean                    0.45     9.05      32.57     -1.49
Median                  0.38     6.56      27.55     -0.48
Minimum                 0.11     0.34       0.50   -134.84
Maximum                 1.56    52.01     338.08     31.14
Std. deviation          0.25     8.30      25.12     10.07
Volatility coefficient  0.55     0.92       0.77      6.75
skewness                0.98     2.17       4.56     -6.61
kurtosis                0.89     5.10      45.59     74.32
5%-percentile           0.14     1.20       6.81    -11.72
95%-percentile          0.90    30.25      79.13      6.31
missing observations    0        0          0        11

Source: Own elaboration based on data from Penn World Tables, 2018 and
OECDstat, 2018.
Note: ctfp - total productivity factor, techODApc - technical ODA in
2015 US dollars received per 1 mln citizens, netODApc - net ODA
received per 1 mln citizens, LOANSpc - value of loans received per 1
mln citizens.

TABLE 2. ECONOMIES INCLUDED IN THE STUDY - WITH GROUP NUMBERS

Number             COUNTRY        Number    COUNTRY

 1      Benin                     13      Namibia
 2      Botswana                  14      Niger
 3      Burkina Faso              15      Nigeria
 4      Burundi                   16      Rwanda
 5      Cameroon                  17      Senegal
 6      Central African Republic  18      Sierra Leone
 7      Cote d'lvoire             19      South Africa
 8      Gabon                     20      Sudan (Former)
 9      Kenya                     21      Swaziland
10      Lesotho                   22      Tanzania
11      Mauritania                23      Togo
12      Mozambique                24      Zimbabwe

Source: Own elaboration.

TABLE 3. RESULTS OF THE ESTIMATION OF PANEL MODELS INVESTIGATING
INFLUENCE OF TECHNOLOGICAL ODA ON TOTAL FACTOR PRODUCTIVITY

Variable   Estimate           Std.error       Student t  p-value

    SPECIFICATION 1: 2-STEP GMM-SYSTEM, ASYMPTOTIC ERRORS
ctfp(-1)    0.5893             0.0114          51.5600   0.0000
const       0.0002             0.0003           0.6449   0.5190
LOANSpc     0.0002             0.0000          11.0400   0.0000
techODApc   0.0030             0.0004           7.7460   0.0000
netODApc    0.0000             0.0000           1.2590   0.2079
          SPECIFICATION2: 2-STEP, ASYMPTOTIC ERRORS
ctfp(-1)    0.7247             0.0120          60.2300   0.0000
const       0.0802             0.0064          12.5700   0.0000
LOANSpc    -3.35*[10.sup.-6]   0.0000          -0.1410   0.8878
techODApc   0.0034             0.0003          13.5600   0.0000
netODApc   -9.49*[10.sup.-5]   0.0000          -4.382    0.0000
                     SPECIFICATION 3: 2-STEP
ctfp(-1)    0.5893             0.0816           7.2230   0.0000
const       0.0002             0.0007           0.2713   0.7861
LOANSpc     0.0002             0.0004           0.4062   0.6846
techODApc   0.0030             0.0013           2.3630   0.0181
netODApc    0.0000             0.0001           0.1490   0.8815
                SPECIFICATION 4: 2-STEP GMM-SYSTEM
ctfp(-1)    0.7247             0.0828           8.7490   0.0000
const       0.0802             0.0245           3.2730   0.0011
LOANSpc    -3.35*[10.sup.-6]   0.0003          -0.01243  0.9901
techODApc   0.0034             0.0018           1.8670   0.0620
netODApc   -9.49*[10.sup.-5]   0.0001          -0.6678   0.5043
                SPECIFICATION 5: FIXED-EFFECT MODEL
const       0.0994             0.0311           3.1950   0.0015
ctfp(-1)    0.7296             0.0736           9.9190   0.0000
LOANSpc     0.0003             0.0004           0.6863   0.4929
techODApc   0.0024             0.0012           1.9200   0.0556
netODApc   -7.29*[10.sup.-5]   0.0001          -1.163    0.2454
        SPECIFICATION 6: FIXED-EFFECT MODEL - FGLS ESTIMATOR
const
ctfp(-1)    0.7220             0.0079          91.7551  <0.0001
LOANSpc     0.0002             2*[10.sup.5]     9.0399  <0.0001
techODApc   0.0029             0.0001          18.7503  <0.0001
netODApc   -4*[10.sup.5]       2.8*[10.sup.5]  -1.5644   0.1177

Source: Own calculations.
Note: Insignificant variables are put in italics. Models 1-5 were
estimated using GRETL, while model 6 - using package plm in RCran.
Small values of the coefficients are the results of the differences in
magnitude.

TABLE 4. PANEL DIAGNOSTIC TESTS - SPECIFICATIONS 1-4.

                SPECIFICATION 1  SPECIFICATION 2   SPECIFICATION 3

Instruments no   175               193             175
AR(1) test        -1.97             -1.87           -1.86
                  (0.049)           (0.06)          (0.06)
AR(2) test        -1.22             -1.21           -1.21
                  (0.22)            (0.22)          (0.23)
Sargan test       17.23 (>0.99)     18.04 (>0.99)   17.23
                                                   (>0.99)
Wald test       3837.72 (<0.01)  10175.6 (<0.01)   111.067 (<0.01)

                SPECIFICATION 4

Instruments no     193
AR(1) test          -1.80
                    (0.07)
AR(2) test          -1.21
                    (0.23)
Sargan test         18.03
                   (>0.99)
Wald test          754.193
                    (0.01)

Source: Own calculations.

TABLE 5. RESULTS OF THE ESTIMATION OF THE MODEL FOR DATA AVERAGED OVER
5-YEARS INTERVALS

Variable     Estimate  Std error  Stat. t  p-value  Robust errors

(Intercept)   0.3626    0.0499     7.2616  <0.0001      0.0501
TechODApc     0.0095    0.0021     4.4446  <0.0001      0.0030
LoansPC      -0.0006    0.0015    -0.3906   0.6970      0.0013
netODApc     -0.0001    0.0007    -0.1199   0.9048      0.0007

Variable     p-value

(Intercept)   0.0000
TechODApc     0.0019
LoansPC       0.6412
netODApc      0.9014

Source: Own calculations.
Note: Model was estimated in RCran using package plm. Robust errors
computed using robust covariance matrix of parameters for a panel model
according to the Beck and Katz (1995).

TABLE 6. CONTROL VARIABLES - WITH LITERATURE REFERENCES

CATEGORY                               CONTROL VARIABLE

Political stability      Index of corruption control; Index of political
                         stability;

Level of infrastructure  Access to electricity;
development

Human capital            Primary school enrollment - gross;

Financial development    Domestic credit to private sector (% of GDP);

Trade openness           Trade to GDP ratio;

CATEGORY                           LITERATURE REFERENCE

Political stability      Chanda and Dalgaard, 2008;

Level of infrastructure  Frija et al. 2015;
development

Human capital            Mankiw, Romer and Weil, 1992; Easterly and
                         Levine, 1997; Barro and Sala-i-Martin, 2004;
                         Klenow and Rodriguez (1997) and Krueger and
                         Lindahl (2001);

Financial development    Comin, 2006; Delechat et al., 2010

Trade openness           Chanda and Dalgaard, 2008; Erken et al. (2016);
                         Edwards (1998);

Source: Own elaboration.

TABLE 7. RESULTS OF THE ESTIMATION OF PANEL MODELS INVESTIGATING
INFLUENCE OF TECHNOLOGICAL ODA ON TOTAL FACTORS PRODUCTIVITY. MODEL
WITH CONTROL VARIABLES

Variable               Estimate  Std.error  Student t  p-value

    SPECIFICATION 1: 2-STEP GMM-SYSTEM, ASYMPTOTIC ERRORS
ctfp(-1)                0.7246     0.0130    55.7100    <0.001
const                   0.1422     0.0173     8.2000    <0.001
TradeToGDP              0.0008     0.0002     5.3280    <0.001
TechODApc               0.0015     0.0001     9.9230    <0.001
ElectricityLack        -0.0013     0.0002    -5.473     <0.001
CorruptionControl       0.0125     0.0056     2.2200     0.0264
CreditToPrivateSector   0.0001     0.0001     1.4060     0.1597
lackOfPrimaryEdu        0.0003     0.0001     4.4550    <0.001
         SPECIFICATION2: 2-STEP, ASYMPTOTIC ERRORS
ctfp(-1)                0.3402     0.0105    32.5500    <0.001
const                  -0.0055     0.0006    -9.826     <0.001
TradeToGDP              0.0008     0.0001     6.3190    <0.001
TechODApc               0.0020     0.0003     7.0270     0.0000
ElectricityLack        -0.0020     0.0008    -2.529      0.0114
CorruptionControl       0.0238     0.0077     3.0780     0.0021
CreditToPrivateSector   0.0011     0.0002     4.8710     0.0000
lackOfPrimaryEdu       -0.0010     0.0002    -4.413      0.0000
                  SPECIFICATION 3: 2-STEP
ctfp(-1)                0.3402     0.0828     4.1080     0.0000
const                  -0.0055     0.0026    -2.117      0.0343
TradeToGDP              0.0008     0.0004     2.1940     0.0282
TechODApc               0.0020     0.0008     2.5040     0.0123
ElectricityLack        -0.0020     0.0019    -1.013      0.3112
CorruptionControl       0.0238     0.0210     1.1340     0.2567
CreditToPrivateSector   0.0011     0.0014     0.8037     0.4216
lackOfPrimaryEdu       -0.0010     0.0008    -1.250      0.2114
                 SPECIFICATION 3: 2-STEP, GMM
ctfp(-1)                0.7246     0.0987     7.3400     0.0000
const                   0.1422     0.0883     1.6110     0.1072
TradeToGDP              0.0008     0.0006     1.2850     0.1988
TechODApc               0.0015     0.0009     1.6590     0.0970
ElectricityLack        -0.0013     0.0009    -1.413      0.1577
CorruptionControl       0.0125     0.0148     0.8446     0.3983
CreditToPrivateSector   0.0001     0.0003     0.4300     0.6672
lackOfPrimaryEdu        0.0003     0.0002     1.2270     0.2199
               SPECIFICATION 5: FIXED-EFFECT MODEL
ctfp(-1)                0.5586     0.0551    10.1300     0.0000
const                   0.0948     0.0391     2.4220     0.0161
TradeToGDP              0.0006     0.0003     2.1420     0.0331
TechODApc               0.0021     0.0010     2.0460     0.0418
ElectricityLack         0.0006     0.0005     1.1260     0.2612
CorruptionControl      -0.004      0.0165    -0.2360     0.8136
CreditToPrivateSector   0.0000     0.0008     0.0072     0.9943
lackOfPrimaryEdu       -0.0004     0.0003    -1.198      0.2322

Source: Own calculations.
Note: LackOfElectricity = 100--AccessToElectricity; lackOfPrimaryEdu =
100--EnrollmentToPrimarySchool. All variables are described in Table 9
in Appendix. Insignificant variables are put in italics. Small values
of the coefficients are the results of the differences in magnitude.

TABLE 8. ESTIMATES OF THE MODEL WITH CONTROL VARIABLES AND USING ROBUST
COVARIANCE MATRIX

Variable               Estimate  Std error  Stat. t  p-value

(Intercept)             0.3465    0.1417     2.4449   0.0169
TradeToGDP              0.0019    0.0008     2.3944   0.0193
TechODApc               0.0093    0.0025     3.7356   0.0004
ElectricityLack        -0.0017    0.0015    -1.1211   0.2660
CorruptionControl       0.0028    0.0405     0.0687   0.9454
CreditToPrivateSector   0.0016    0.0014     1.1326   0.2612
lackOfPrimaryEdu        0.0005    0.0006     0.8666   0.3891

Source: Own calculations.
COPYRIGHT 2019 Prague Development Center
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Swierczynska, Katarzyna; Kliber, Agata
Publication:Business and Economic Horizons
Geographic Code:60SUB
Date:Jan 1, 2019
Words:11332
Previous Article:Estimation of the world economic system stability from 1963 to 2013 by using a discrete dynamic model.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters