Abstract
Abstract
This study analyzed the macroeconomic and institutional determinants of total factor productivity (TFP) in the MINT (Mexico, Indonesia, Nigeria, and Turkey) countries during the period 1980–2014. Annual data covering the period between 1980 and 2014 were used. Data on real gross domestic product (real GDP), labor force, gross fixed capital formation, foreign direct investment (FDI), human capital, and inflation were sourced from the World Development Indicators published by the World Bank. Also, data on corruption, government stability, and law and order were obtained from the database of International Country Risk Guide. Panel autoregressive distributed lag (PARDL) regression technique was used to estimate the model. Results showed that TFP growth rate declined on average by 1.4 per cent and 1.8 per cent in Mexico and Turkey, respectively, while Indonesia and Nigeria did not experience productivity growth on the average. Results also showed that in the long run, human capital and government stability had positive and significant effects on TFP, while FDI and corruption had negative but significant effects on TFP. In the short run, there existed a significant negative relationship between TFP and inflation. However, the effects of human capital and corruption on TFP were positive and significant. The study concluded that human capital and corruption were key drivers of TFP in the MINT countries both in the long run and short run.
Keywords
Introduction
The macroeconomic growth literature identifies both factor accumulation and total factor productivity (TFP) as principal determinants of growth. This is because labor and capital can be employed efficiently when there is adequate stock of technology in the economy (Nwosu, Njoku, Akunya, Ihekweme, & Marcus, 2013). Indeed, in the economic growth literature, labor, and capital can be employed efficiently when there is adequate stock of technology in the economy. This assertion can be verified by the differing economic performance of country groups around the globe (Nwosu et al., 2013). It can be concluded from the aforementioned argument that there are two sources of growth. The first is input-driven, that is, by adding more and more resources into the same production function. Such growth is hard work and, by the law of diminishing returns, cannot be sustained indefinitely. The second is technology-driven which invokes increasing returns and can be sustained. Technology-driven source is difficult to measure directly, but what is not input-driven is by default technology-driven, which has come to be called TFP growth (Jajri, 2007). For this reason, various researchers have supported TFP growth as the main source of long-run economic performance (see Hall & Jones, 1999; Solow, 1956; Swan, 1956).
According to Van Ark (2014), TFP is the growth in output after taking into account the contribution from other inputs. The TFP is defined by Dhyne and Fuss (2014) as the efficiency with which goods and services are produced using a given technology and taking into account the quantity of available inputs. According to Shackleton (2013), TFP growth is the unexplained (or residual) portion that is conventionally attributed to all of the production factors together. This presumably reflects advances in production technologies and processes. Therefore, TFP in a way helps to understand economic growth of any country since it also means the growth in output that cannot be explained by increases in factor inputs (say labor and capital).
Empirically, several studies have been carried out to examine the possible factors that drive TFP since it is known to contribute greatly to the growth of any economy. These factors include human capital, innovations, foreign direct investment (FDI), inflation rate, exchange rate, trade openness, unemployment, value-added manufacturing, government consumption, financial development, information and communication technology (ICT), wage rate, interest rate, research and development (R&D) expenditures, infrastructures, lending rate, liquid liabilities, competition, credit to private sector, imports, exports, external debt, absorptive capacity, and population growth rate among others (see Adejumo, 2012; Akanbi, 2011; Akinlo, 2006; Ali, Mushtaq, Ashfaq, Abdullah, & Dawson, 2012; Anders & Thiam, 2006; Khan, 2006; Kolawole, 2015). This study, therefore, intends to make use of some of the variables used by previous authors in determining TFP in the MINT (Mexico, Indonesia, Nigeria, and Turkey) countries. The MINT countries are therefore classified as emerging economies according to Jim O’Neill of Goldman Sachs based on the similar characteristics they share. Moreover, they are expected to show strong growth rates and provide high returns for investors over the coming decades. Consequently, if the MINT countries are to do this, then they are expected to have sustainable economic growth and development. Sustainable growth has been found to depend not only on factor accumulation but also on TFP.
Meanwhile, evidence shows that the average growth rates for the MINT economies have not been smooth over time even for the so-called high growth rate performing country, Nigeria (The World Bank, 2014). It, therefore, shows that the four economies had grown at an average annual rate of 16.06 per cent with wide fluctuations across the past three and half decades. For instance, it is shown that Indonesia had the highest average growth rate of 5.49 per cent followed by Turkey with 4.16 per cent. Next to Turkey was Nigeria showing 3.72 per cent average growth rate and finally, Mexico had the lowest average growth rate of 2.69 per cent.
Nigeria and Indonesia lend themselves naturally to comparison since they started together but today Indonesia has overtaken Nigeria in terms of growth and development (Bevan, Collier, & Gunning, 1999). Figure 1 shows that there was no evidence of growth sustainability in Nigeria as revealed by a negative average growth rate between 1985 and 1989, and a sudden increase between 2000 and 2004, while Indonesia was said to show a long period of sustained growth with the highest average growth rate. But, there was a sudden decline between 1995 and 1999. An explanation for Indonesia success is that the country created an environment conducive to growth in the non-oil economy, whereas Nigeria did not provide such environment for growth to spur. Turkey and Mexico are said to show partially smooth and positive growth rates for all the sub-periods. Meanwhile, the growth rates of labor for the four economies had been consistent over the years for most of the study period, while growth rates of capital and TFP had been oscillating over the years—even at times negative (The World Bank, 2014).

For that reason, if TFP is crucial for growth to take place, an understanding of the dynamics and the determinants of TFP is desirable. While studies have examined TFP and its determinants among the newly industrialized countries (NICs) (see Nirvikar & Hung, 1996); European countries (see Gehringer, Martinez-Zarzoso, & Danziger, 2013) and sub-Saharan African countries (see Akinlo, 2006), studies that have compared TFP and its determinant among the MINT countries as emerging economies are sparse. In view of the fact that the MINT countries are classified to show strong growth rates in the future, an examination of the factors that determine TFP in these countries becomes crucial.
This article is structured as follows: the second section reviews the existing literatures both theoretically and empirically, while the third section presents the methodology employed and data used to achieve the objective of the study. In the fourth section, we analyze the determinants of TFP by using an economic model. Finally, the fifth and the sixth sections discusses the findings of the results and concludes the article together with the policy implications, respectively.
Related Literature
Both theoretical and empirical studies have documented the importance of TFP for long-term growth (Solow, 1956). Basically, there are five theories of growth that readily come to mind in this analysis. They include stages of growth theories, the classical growth theory, the Harrod–Domar growth model, the neoclassical growth model and the new growth/endogenous growth theory. As it were, postulations by those theories have strategies import in the attempt to establish how those factors determine TFP. Although there are different ways of measuring TFP, Solow approach has been one of the commonest ways of measuring the TFP, and this is through calculating the growth accounting equation (see neoclassical theory upon which the study is based). This is called residual approach and TFP is obtained after the contribution of the physical inputs, say labor and capital, is determined.
Interestingly, the first empirical study on productivity growth was carried out by Solow (1957). He used aggregate production function model on the United States (US) economy during 1909–1949. His results showed that output per labor hour increased by 200 per cent during the study period and about 88 per cent of the increase in output was accounted for by technical change. Thus, based on Solow’s study, it was technological progress that brings about long-run growth in output. However, data on output per labor hour as used by him to capture labor input were not always available for most countries.
Looking at the developed countries, Botirjan (2011) used system GMM estimator method in a panel of 46 countries during 1974–2008 and found that FDI increased TFP growth. Gehringer et al. (2013) conducted a research on the determinants of TFP in the 17 European Union (EU) countries using panel data estimation techniques for the period 1995–2007 with the measurement of TFP under two approaches: value-added approach which is used to estimate sector-specific TFP levels over time and across countries and output-based approach, where TFP is obtained as a residual from the output-based production function and intermediate production factors. It has therefore been argued that the latter approach is theoretically more appropriate, as it permits the explicit consideration of intermediate production factors in the technological-driven sector-level growth (Jorgenson & Stiroh, 2000).
Additionally, Danquah, Enrique, and Quattara (2014) studied the determinants of TFP growth in 67 countries for the period 1960–2000. The TFP growth was computed by using the data envelopment analysis (DEA) frontier approach together with the Malmquist index method. The explanatory variables used in the study included initial GDP, consumption share, trade openness, government share, investment share, labor force, population growth, secondary education, life expectancy, investment price, primary education, and health. Their results affirmed that it was only the degree of trade openness that determined TFP while other variables showed no significant effect. Meanwhile, Baltabaev (2013) investigated the relationship between FDI and TFP growth in a panel of 49 countries between 1974 and 2008. The author made use of FDI, R&D expenditure, human capital, trade openness, inflation, and population growth as the independent variables. For estimation, the study employed the system GMM technique. Results from the estimations indicated that FDI and R&D expenditure had positive and significant impacts on TFP, while a negatively significant relationship was established between TFP and each of human capital, trade openness, inflation, and population growth in the sample countries. In another related study, Espinoza (2012) looked at how factor accumulation and certain macroeconomic variables affect TFP in Bahrain, Qatar, Kuwait, Omar, Saudi Arabia, and the UAE for the period 1990–2009 by employing the growth accounting methodology and static panel data technique. The author used government size for all the countries, value-added agriculture, and growth volatility as independent variables. The results showed that oversized governments (in Kuwait and the UAE) and volatile growth in the region would have contributed negatively to TFP. However, value-added agriculture had positive effect on TFP in all the countries.
Exploring the developing countries, Khan (2006) examined the macro determinants of TFP in Pakistan from 1960 to 2003 by utilizing the conventional growth accounting framework and the method of analysis was ordinary least square (OLS) linear regression. The explanatory variables used in the study included inflation, FDI, financial sector depth, private credit, budget deficit, population growth, investment, employment, and government consumption. The results showed that openness, budget deficit, population growth, and education expenditure had negative effect on TFP while FDI, domestic investment, inflation, financial sector depth, private credit, employment, and government consumption were significant and positively related to Pakistan’s TFP.
In addition, Akinlo (2006) examined the impact of macroeconomic factors on TFP in 34 sub-Saharan African countries during the period 1980–2002. Using fixed effect estimation technique, the results showed that human capital, export–GDP ratio, credit to private sector as a percentage of GDP, manufacturing development, FDI as a percentage of GDP and liquidity liabilities as a percentage of GDP had significant positive effects on TFP. Meanwhile, inflation rate, lending rate, population growth, external debt, value-added agriculture as a percentage of GDP ratio and local price deviation from purchasing power parity had significant negative effects on TFP. Saha (2012) adopted translog-based growth accounting methodology to investigate the relationship between trade openness and TFP growth in India for the period 1961–2008. The author found out that there is a one-way relationship between trade openness and TFP growth for Indian economy. The study revealed that trade openness in India had affected TFP growth positively and significantly.
Meanwhile, Park (2010) conducted a research on the TFP growth for 12 Asian economies using the neoclassical Solow growth model to calculate TFP and considered the following variables as the determinants of TFP growth: initial per capita income, initial population, openness, initial life expectancy, geographical factors, R&D stock, FDI flows, inflation rate, budget deficit, and current account deficit. Among other potential determinants, initial life expectancy and openness were the two variables that consistently influenced TFP growth, while initial per capita income, initial population, geographical factors, R&D stock, FDI flows, inflation rate, budget deficit and current account deficit were found to be insignificant in these countries. Using stochastic frontier analysis (SFA) to the time-series data of 44 Asian countries from 2000 to 2010, Shahabinejad, Mehrjerdi, and Yaghoubi (2013) analyzed TFP growth and its components in Asian countries. Only in 11 countries, technical change had a positive role in productivity growth. In an attempt to develop a simple theoretical model for TFP in a typical developing country with testable hypotheses, Kamaly (2011) made use of an actual estimation of a production function for Egypt to obtain TFP series. It is obvious from the study that human capital, captured by literacy rate, was found to have a significant positive effect on TFP. However, capital stock was found to have a negative but not very significant effect on productivity. Meanwhile, imported capital had a spillover positive effect on productivity as imported capital has superior technology which bears productivity gains. Relatedly, Aravena, Hofman, Fernández de Guevara, and Mas (2014) developed a framework to analyze the potential of different variables to increase TFP growth in countries with poor productivity performance. Results showed that the top priorities for these countries were to improve the labor market, to reduce the share of self-employed people and to modernize the functioning of their economic systems. The results also indicated that the intensification of investment in ICT and R&D activities was a key instrument for promoting growth. Public policies should also aim to encourage a higher endowment of Internet infrastructures and their use.
Looking at some sectors in an economy, Ray (2012) recognized and investigated the dynamic factors responsible for TFP growth of India’s aluminium industry during 1979–1980 to 2008–2009 using the OLS methodology. Variables used included export–output ratio, import penetration, growth in output, tariff rate, real effective exchange rate, terms of trade, inflation rate, capacity utilization, investments in fixed assets, and gross markup. Export and import were found to have negative impacts on TFP growth while capacity utilization and growth in output were positively related to TFP growth in aluminium industry in India. It is also discovered from the study that there existed insignificant positive association between tariff variable and TFP growth but real effective exchange rate had a significant negative impact on productivity as it is expected. Gross markup also had a positive impact on TFP growth which means that profit is a driving force of TFP growth. Hwang and Wang (2012) coupled with the studies earlier studied the effectiveness of export, FDI, and R&D on TFP growth in 22 Taiwanese and seven Korean manufacturing industries over the period 1981–2001. The study used factor decomposition method of growth accounting with Tornqvist expression of TFP growth rate. The determinants of TFP employed in the study were FDI, R&D, export, import penetration ratio and total trade ratio to production. Empirical findings showed that neither growth of FDI nor R&D showed positive effect on TFP growth. However, the results indicated that output growth had a robust and positive relationship with TFP growth for both countries. Export, import penetration ratio and total trade ratio to production were not significant in the estimation.
The MINT countries are not left out in the review. In Mexico, Alvarez-Ayuso, Becerril-Torres, and Moral-Barrera (2011) carried out an empirical study on the effect of infrastructures on TFP and its component (technical change and efficiency change). The DEA was used to obtain TFP and its components. In an attempt to determine the effect of infrastructures, they used panel data econometrics through the estimation of a model of fixed effects. The results indicated that infrastructures positively affected private productive factors and the components of technical change. For Indonesia, Ikhsan-Modjo (2006) looked at TFP in Indonesia manufacturing sector using a stochastic frontier approach between 1988 and 2000. The results showed that TFP grows by 1.55 per cent between the study period. It was found that technical progress was the utmost important factor in explaining TFP growth in the Indonesian manufacturing sector from 1998 to 2000. Concerning Nigeria, Adenikinju (2005) studied productivity performance in Nigeria. According to him, the government has an important role to play in creating conducive environment for productivity initiatives by the private economic agents. It must ensure the efficient and effective provision of public goods, support the provision of infrastructures and address various forms of market failures. More to the point, Adejumo (2012) investigated the effect of macroeconomic factors on productivity in Nigeria between 1970 and 2009 using cointegration and error correction method (ECM). The long-run results indicated that FDI was positively related to productivity growth while economic openness, human capital, and inflation rate were negatively related to productivity growth. Meanwhile, the short-run results revealed that FDI had significant negative effect on productivity while human capital had significant positive effect on productivity growth.
Still on Nigerian economy, Akanbi (2011) examined the macroeconomic determinants of technological progress (TFP) in Nigeria that is consistent with the endogenous growth theory during 1970–2006. The results showed that macroeconomic instability, the level of financial development, and the level of human development were highly significant determinants of technological progress in Nigeria. While Hassan, Abdullah, Ismail, and Mohamed (2014) analyzed the maize TFP growth using DEA based on Malmquist Index in Nigeria between 1971 and 2010. For the determinants of maize TFP growth, R&D spending, net value of production, fertilizer price, and labor were identified to have a significant influence on TFP growth.
Ismihan and Metin-Özcan (2009) studied productivity and growth in an unstable emerging market economy in Turkey between 1960 and 2004 using cointegration and impulse response analyses. The results showed that TFP and capital accumulation were major sources of growth during the study period and that TFP was positively affected by imports and public infrastructure investment and negatively affected by macroeconomic instability.
On a final note, Berument, Dincer, and Mustafaoglu (2011) studied relationship between TFP and macroeconomic instability in Turkey from the first quarter of 1987 to the third quarter of 2007 using VAR-GARCH models. The measures of macroeconomic instability explored include volatility in inflation, openness of an economy and financial market deepness. The empirical results showed that the degree of openness and financial market deepness reduced TFP growth, whereas volatility of inflation increased TFP growth. They concluded that lowering macroeconomic instability might have helped in increasing Turkey’s TFP and catch up with EU countries. The impact of FDI on aggregate growth and TFP in Turkey for the period 1960–2005 was examined by Arisoy (2012). Using FDI and physical capital accumulation as the explanatory variables, the results indicated that FDI contributed positively to TFP growth via capital accumulation and technological spillovers.
Conclusively, it is observed from the literature that there are still few studies on the determinants of TFP in the MINT countries. Based on the author’s knowledge, no single study has been carried out on the MINT countries (as a whole) as far as total productivity is concerned, and since the aim of bringing these countries together according to Jim O’Neil of Goldman Sachs is to ensure that they show strong growth rates in the future, it is imperative to examine the factors that drive productivity in these economies as this will go a long way in explaining their further growth prospects. Conclusively, most of the studies (see Adejumo, 2012; Akanbi, 2011; Akinlo, 2006; Baltabaev, 2013; Espinoza, 2012; Khan, 2006) examined the determinants, particularly macroeconomic determinants of TFP without looking at the possibility of some other variables that can affect TFP. Hence, this study contributed to this area of economic literature by analyzing the determinants of TFP in the MINT countries with the inclusion of corruption, government stability, and law and order (as the institutional factors) as possible determinants.
Methodology and Data
This section contains the research methods used to achieve the stated objectives.
Estimating TFP
In this section, TFP for the four countries are estimated by subtracting capital and labor inputs proxied by gross fixed capital formation and total labor force, respectively, from output proxied by real GDP. This is known as Solow residual where TFP is obtained as a residual from the valued added-based Cobb–Douglas production.
Model Specification
The neoclassical production function of the representative firm in the business sector which is assumed to be Cobb–Douglas production function can be used to generate the TFP for the study period. Considering a Cobb–Douglas production function:
Making lnAt (or In TFPit ) the subject of the formula yields:
By differentiating Equation (2) with respect to time, the rate of TFP growth can be expressed as:
where yit, kit, and lit denote the growth rates of output, capital, and labor, respectively, and TFPit is the rate of TFP growth.
Thus, to analyze the factors that determine TFP as computed in Equation (2), the following model by Gehringer et al. (2013) and Khan (2006) was adopted.
where
TFPit: Total factor productivity
Xit the vector of determinants of TFP
μit:the error term.
Note that subscripts i and t are used to capture the countries involved in this study (i.e., the MINT countries) and the time frame for the study (i.e., 1980–2014), respectively.
The determinants (Xit) of TFP to be used in this study include foreign direct investment (FDI), inflation (INF), human capital (HUMC), corruption (COR), government stability (GOS), and law and order (LAO). The choice of the variables is based on their importance to determine productivity and, also on the fact that they have been used in the previous studies as discussed further.
One of the key characteristics of emerging economies like MINT is large inflows of FDI and this explains its inclusion as one of the determinants of TFP. Also, FDI as used by Akinlo (2006), Khan (2006), Park (2010), Baltabaev (2013), and so on is another major vehicle in transferring foreign technology (see Akinlo, 2006), creating employment, reducing the barriers in adoption of technology and bringing improvement in the quality of labor and capital inputs in the host economy (see Khan, 2006) with positive impact on TFP. Human capital is added to the model because it possesses necessary abilities not only to become familiar with and efficiently use existing innovations, but also to contribute to the generation of brand new innovative outcome, which leads to increased productivity. Based on this, Jajri (2007) and Mohammad and Jalil (2011) asserted that there is need for education and quality investment in human capital for productivity to be enhanced. In the same line, Akinlo (2006) affirmed increasing an economy’s skill base can have a positive impact on TFP growth by facilitating structural change and technological improvements. It is observed that there is a mixed feeling about the role of inflation in TFP among scholars (see Adejumo, 2012; Akinlo, 2006; Baltabaev, 2013; Khan, 2006). However, inflation is used as a regressor in the model of this study to capture the stability of economy, which is hypothesized as necessary for TFP growth. Also, developing economies signal the impact of money illusion and that is why inflation is included as a macroeconomic determinant of TFP.
For the institutional variables, corruption is included in the model because of its indirect effect on productivity through incentives, innovations, and investments (either domestic or foreign which is the key characteristic of the MINT countries). According to Baumol (1990) and Murphy, Shleifer, and Vishny (1991), corruption can distort the allocation of human resources, generating incentives that lead the best qualified people to devote themselves to rent-seeking activities instead of productive or innovative activities, thus negatively affecting productivity growth. In addition, corruption is seen to be the major problem faced by the MINT countries which made them to score badly in Transparency International’s Corruption Perceptions Index (2014); Turkey ranks 64 out of 175, while the rest are closer to the other end of the scale (Matsangou, 2014). Based on the study of Nachega and Fontaine (2006), political (government) stability makes stable economic growth by assisting the business environment and economic activity, creating economic certainty and therefore increasing incentives to invest. Since the MINT countries are created to attract the needed capital flows, government stability is said to affect TFP and hence, its inclusion in the model.
Finally, law and order is included in the model because TFP is said to include rule of law according to Hulten (2001). Also, law and order means state of society where vast majority of population respects the rule of law, and where the law enforcement agencies observe laws that limit their powers and rule of law which represents the extent to which citizens have confidence and abide by the rules of society such as the effectiveness and predictability of the judiciary, or the enforceability of contracts, is said to have great impact on labor productivity and therefore translates to improved overall productivity.
Implicitly, Equation (4) can be written as:
Explicitly, Equation (5) can be written as:
where αi, i = 1, 2, 3, and 4 are the intercepts, βk for k = 1, 2, 3, 4, 5, and 6 are the slopes or coefficients of the determinants and μit are error terms.
Rewriting Equation (6) in its explicit form becomes:
Estimation Techniques
To achieve the stated objective, first, TFP was computed using Equation (2) with 0.6 (α) and 0.4 (β) as the coefficients of labor and capital, respectively, following previous studies by Chen (1997), Easterly and Levine (2001), Akinlo (2006), and Adejumo (2012). This is based on the fact that the techniques of production in most economies of developing countries are more of labor intensive than capital intensive. Hence, the appraisal of TFP growth rates computed for the MINT countries was done using descriptive statistics (that involves the use of graphs). Second, a country-specific and panel analyses were conducted to analyze the factors that determine TFP in the MINT countries in order to compare the results so as to show if the countries have the same peculiarities.
To empirically investigate the factors that drive TFP, Equation (7) was estimated using the bounds testing (or autoregressive distributed lag [ARDL]) cointegration procedure and extending the techniques of panel autoregressive distributed lag (PARDL) analyses which include the pooled mean group (PMG) estimator (Blackburne & Frank, 2007; Pesaran, Shin, & Smith, 1997). The country-specific and panel results were obtained from the model for comparative analysis in order to show if the countries have the same peculiarities or not.
The procedure was adopted for the following: first, the bounds test procedure is simple. Contrary to other multivariate cointegration techniques, for instance, Johansen and Juselius (1990) cointegration test, bounds test allows the cointegration relationship to be estimated by OLS once the lag order of the model is identified. Second, the bounds testing procedure does not have need of the pretesting of the variables incorporated in the model for unit roots unlike other techniques. It is valid regardless of whether the regressors in the model are purely I(0), purely I(1) or mutually cointegrated. Third, the test is comparatively more efficient in small or finite sample data sizes. The procedure will, however, crash in the presence of I(2) series. Fourth, PMG estimator helps to relax the condition of n > T for panel study. Also, its results include the long-run estimates and the averaged short-run parameter estimates.
The PARDL procedure involves two stages. One, the existence of the long-run relation between the variables under investigation is tested by computing the F-statistic for testing the significance of the lagged levels of the variables in the error correction form of the underlying PARDL model. Two, the analysis is to estimate the coefficients of the long-run relations and make inferences about their values.
The first step assumes that the (asymptotic) distribution of this F-statistic is non-standard, irrespective of whether the regressors are I(0) or I(1). Pesaran et al. (2001) tabulated the appropriate critical values for different numbers of regressors (k) and determined whether the PARDL model contains an intercept and/or trend. To obtain the vital F-statistic, Equation (7) was estimated by the OLS method, and the related variable addition test was subsequently conducted to obtain the related F-statistic from the results.
Thus, the hypothesis tested relates to the null of the non-existence of a long-run relationship which is defined as follows:
H0:β1= β2= β3= β4= β5= β6= 0
Against
H0:β1≠ β2≠ β3≠ β4≠ β5≠ β6≠ 0
The existence of a long-run relationship gave the ticket to proceed with the analysis. This means that one can only proceed to the second stage only if there is a satisfactory long-run relationship existing between the variables to be estimated, in that it is not spurious. The error correction model associated with the long-run estimates may be estimated to determine the stability of the long-run relationship.
Since there was evidence in support of a long-run relationship or cointegration among the variables included in Equation (7), the following long-run dynamic model was estimated:
Thus, the PARDL specification of the short-run dynamics may be derived from the error correction representation of the form:
The symbol ∆ is the difference operator and the error correction tern, in this case is defined as:
All coefficients of the short-run equation are coefficients relating to the short-run dynamics indicating the model’s convergence to equilibrium following a shock to the system and the symbol is the speed of adjustment parameter measuring how fast errors generated in one period are corrected in the following period. The lag length in the aforementioned equations was determined by using Schwarz (Bayesian) information criterion (SIC). Due to the small size of the data and the consistency of this criterion, SIC is preferable to the Akaike information criterion (AIC) because of its parsimonious in lag length selection, to avoid losing a lot of degrees of freedom.
Data Measurement and Source
Annual data covering the period between 1980 and 2014 were used for this study and the data were secondary data. The macroeconomic variables (FDI, HUMC, and INF) were sourced from the World Development Indicator (The World Bank, 2014) database while the institutional variables (corruption, government stability, and law and order) were sourced from International Country Risk Guide (ICRG, 2014) database as shown in Table 1
Results and Discussion
This section contains the analysis carried to determine TFP in the MINT countries.
Unit Root Tests
This section reveals the various unit root tests conducted on the selected variables so as to ascertain whether the variables are stationary or not. The augmented Dickey–Fuller (ADF) and Phillips–Perron (PP) tests (with intercept only) were used for the individual variables while the Levin, Lin, and Chu (2002) and Im, Pesaran, and Shin (1997) tests were conducted for the panel variables. For the individual variables as shown in Table 2a, the ADF and PP tests showed that all the variables were stationary at 5 per cent level of significance except Indonesia’s LAO and Nigeria’s HUMC that were I(1) at 10 per cent level of significance using PP test. That is, the variables were of different orders of integration (i.e., I(0) and I(1)) with intercept only. More importantly, there was no I(2) variable that will make the ARDL procedure to crash. Therefore, the ARDL procedure is said to be appropriate for this study since the variables were both I(0) and I(1) and no I(2) variable. Also, in Table 2b, the panel unit root tests (with individual intercept only) conducted revealed that all the variables were stationary at both level and first difference, that is, they were I(0) and I(1) variables.
Analysis of Macroeconomic and Institutional Determinants of TFP
Presentation of Variable Measurement and Source
Country-specific Unit Root Tests
Panel Unit Root Tests
Lag Order Selection for the ARDL Model
Before carrying out the ARDL models, it becomes very important to include an optimal lag length so as to select the model with the best fit among the various alternatives. Therefore, for this study, the optimal lag length was obtained through the SIC. Due to the small size of the data and the consistency of this criterion, SIC is preferable to the AIC because of its parsimonious in lag length selection, to avoid losing a lot of degrees of freedom. As shown in Table 3, the optimal lag length based on SIC was given to be one (1) in all the countries. However, the ARDL models of this lag in Mexico and Turkey were affected by serial correlation which called for increasing the lag length to two (2) in Mexico; and three (3) for dependent variable and two (2) for independent variables in Turkey in order to correct for serial correlation.
ARDL Bounds Test
A long-run relationship is said to exit among the variables if they are cointegrated, and this is obtainable only if the F-statistic is greater than the upper bound value of the Pesaran et al. (2001) critical table at various levels of significance. If otherwise, the test is inconclusive (if the F-statistic falls between the lower and upper bound limits) or has no long-run relationship (if the F-statistic falls below the lower bound value). However, the bounds test results for the MINT countries are presented in Tables 4a and 4b. For Mexico only, time was an important factor in determining TFP before any significant result could be achieved and this caused Mexico’s variables to be analyzed with trend.
The F-statistic for Mexico (13.25) approximately fell above the upper critical bound (with unrestricted intercept and unrestricted trend) at 5 per cent level of significance (4.00) while the F-statistics for Nigeria (7.23), Turkey (11.93), and Indonesia (3.34), approximately fell above the upper critical bound (with unrestricted intercept and no trend) at 5 per cent level of significance (3.61) for Nigeria and Turkey and only at 10 per cent level of significance (3.23) for Indonesia. The results, therefore, from Tables 4a and 4b indicated that there existed a long-run relationship among the variables in Mexico, Nigeria and Turkey since the null hypotheses of no long-run relationships exist were rejected at 5 per cent significance level in each country while the variables showed only long-run relationships at 10 per cent and not at 5 per cent significance level in Indonesia.
Long-run Relationships—Country-specific
Having established that a long-run equilibrium relationship existed among the variables in each country through the bounds tests in Tables 4a and 4b, it becomes necessary to estimate the long-run form of the models in order to determine any significant long-run equilibrium relationship. Table 5 presents the results of long-run estimations for each country.
Short-run Relationships and Error Correction Term—Country-specific
Lag Order Selection Criteria
LR: sequential modified LR test statistic (each test at 5% level); FPE: final prediction error; AIC: Akaike information criterion; SC: Schwarz information criterion; HQ: Hannan–Quinn information criterion.
Bounds Test and Pesaran et al. Critical Table for Mexico
Bounds Test and Pesaran et al. critical table for Indonesia, Nigeria, and Turkey
Long-run and Short-run Relationships for Panel Study
Long-run Relationships
H0: TFP and its determinants have no long-run relationships.
For the panel study (refer Equations [8]–[10]), the PMG estimation technique was employed. The result as presented in Table 7b showed the long-run and short-run estimates of the panel study as well as the ECT which showed whether the variables move together in the long run.
Discussion of Findings
Short-run Parsimonious Regression Estimates for Mexico and Indonesia
H0: FDI, HUMC, INF, COR, GOS, and LAO have no short-run effect on TFP.
Short-run Parsimonious Regression Estimates for Nigeria and Turkey
H0: FDI, HUMC, INF, COR, GOS, and LAO have no short-run effect on TFP.
Kao Residual Cointegration Test
Panel ARDL (3, 4, 4, 4, 4, 4, 4) Result
Meanwhile, the insignificant effect of FDI in Nigeria and Indonesia was due to lack of improved investment climate (i.e., good business environment) such as good governance, effective institutions, security, accountability, absence of political and religion violence, bureaucratic inertia among other reasons. In addition, most FDI goes to oil and gas sector and not productive sectors like manufacturing industry in Nigeria. This is also in line with the findings of Akinlo (2004) that extractive FDI might not be growth enhancing as much as manufacturing FDI. While Indonesia has been an outlier within the Asian region and even also among the MINT countries, with lower inflows of FDI than other (MINT) countries, especially in manufacturing, and with lower inflows than could be expected from its size and other country characteristics (see also Lipsey & Sjoholm, 2010).
An interesting result was found for human capital as human capital for all the four countries was statistically significant at 5 per cent level both in the long run and short run. A positively relationship, which is in conformity with the theory, was established between human capital and TFP in all the countries except Mexico that showed otherwise in the long run. The short-run analysis was different as human capital had a significant negative effect, which does not support the theory, on TFP in all the countries except Indonesia that showed positive relationship. Another interesting thing in the short run was that the previous year (lag of 1 year) of human capital was said to be determining the current year TFP in all the four countries of our focus, except in Turkey. The positive effect of human capital, as expected, is consistent with some studies (Adejumo, 2012; Ahmed & Bukhari, 2007; Aiyar & Feyrer, 2002; Akinlo, 2006; Ascari & Di Cosmo, 2004; Jajri, 2007; Kamaly, 2011; Khan, 2006; Lucas, 1986; Miller & Upadhyay, 2002; Park, 2010; Romer, 1986) where increased productivity is as a result of good education and quality investment in human capital while the negative impact of human capital is in line with the study of Baltabaev (2013).
Hence, the negative and positive association of human capital with TFP in Mexico and Indonesia, respectively, can be attributed to many reasons. In Mexico, it reflects that educational system was highly inefficient, incentives for improvement were weak, and the quality of educational provision was well below OECD standards (Hopkins, Ahtaridou, Matthews, Posner, & Figueroa, 2007), thus reducing TFP, while Indonesian government paid attention to the educational system by devoting 20 per cent of government expenditure to education which helps in increasing efficiency, allocating resources efficiently, better tailoring of provision of local needs and circumstances, stronger performance management, and requiring further efforts to build capacity at regional and district levels to implement and monitor education reform (OECD, 2015), thus making a positive impact on TFP. Also, in Nigeria and Turkey, the negative effect of human capital on TFP confirmed the lack of skill oriented education (the accepted phenomenon to raise TFP) in the short run, while the later effect showed that TFP will increase due to high level of skilled personnel in the two countries.
Similarly, inflation was said to be insignificant (see Bosworth & Collins, 2003; Edwards, 1998; Freeman & Yerger, 2000; Nachega & Fontaine, 2006; Park, 2010) in all the four countries except in Mexico where it was positively significant (see Hercowitz, Lavi, & Melnick, 1999) at 5 per cent level in the long run. This is not so in the short-run analysis since inflation was found to be significant with a mixed result, as expected, in all the four countries. Inflation had a significant negative effect at 5 per cent on TFP in Mexico (but its lag of 1 year) and Turkey which aligns with Hercowitz et al. (1999), Miller and Upadhyay (2000), Akinlo (2006), Loko and Diouf (2009), Adejumo (2012), Baltabaev (2013), and Kolawole (2015) who found out that there was a negative relationship between inflation and TFP. The inverse relationship is expected because according to Adejumo (2012), inflation reduces the demand for real money balances, and thus if money serves as a factor of production, it reduces productivity.
In addition, the negative impact of inflation on TFP could be as a result of high and unstable prices that lead to a lot of economic uncertainties that discourage investors (both foreign and domestic) from investing in projects that will improve productivity (Akinlo, 2006). However, a significant positive relationship was established between inflation (but its lag of year in Indonesia) and TFP in Indonesia and Nigeria at 5 per cent and 10 per cent levels, respectively. This result is consistent with the findings of earlier empirical studies (see Berument et al., 2011; Khan, 2006) that a predictable and stable or gently rising price level provides the best climate for healthy economic growth which in turn enhance productivity. Moreover, Khan (2006) explained that inflation adds to economic growth by generating employment or merely increasing the working hours of employed labor in a sense that the positive relationship of inflation rate and TFP can be expected.
As expected, corruption was found to have an insignificant long-run negative effect on TFP in Mexico, Indonesia, and Nigeria, while it had a significant positive effect on TFP in Turkey at 5 per cent. In the short run, the effect of corruption on TFP was negative and significant at 5 per cent level in all the countries except in Indonesia where the effect was negative but insignificant at 5 per cent level. This, therefore, follows the a priori expectation of an indirect relationship between corruption and TFP. The negative effect confirms the findings of Salinas-Jiménez and Salinas-Jiménez (n.d.) that corruption affects negatively the efficiency levels (productivity) at which an economy performs.
Given the fact that corruption is one of the major problems faced by the MINT countries and the level of corruption in these countries as confirmed by International Country Risk Guide (ICRG, 2014) corruption index, corruption was seen to affect all the countries negatively in the short run and long run, thereby leading to reduced productivity due to capital flight, bribery, abuse of office and ineffective anti-corruption laws, except in Turkey where its effect in the long run was positive.
Specifically, the negative impact of corruption on TFP shows that corruption would disrupt allocation of human resources and generate incentives that lead the best qualified people to devote themselves to rent-seeking activities instead of productive or innovative activities, especially in Nigeria where the effect was high, thereby leading to reduced productivity growth. On the other hand, arguments that stress the positive effects of corruption are based on ideas coming from ‘second best’ theory—given a set of distortions created by governmental procedures or policies, corruption would permit agents to evade those regulations that hinder economic activity, acting to ‘grease the wheels’ of the economy (Huntington, 1968; Leff, 1964).
Government stability had both positive, as expected, and negative effects on TFP in Nigeria (at 5% level) and Turkey (at 10% level), respectively, in the long run. The long-run effect of government stability on TFP in Mexico and Indonesia was positive, as expected, but insignificant at 5 per cent. Meanwhile, the short-run results showed that government stability had a significant negative impact on TFP in Mexico and Turkey with insignificant effect in Indonesia while its impact on TFP in Nigeria was significantly positive, as expected at 5 per cent level. The negative effect was against the a priori expectation that government stability makes stable economic growth by assisting the business environment and economic activity, creating economic certainty and therefore increasing incentives to invest so as to enhance productivity (see Nachega & Fontaine, 2006).
Consequently, government stability was seen to have only immediate effect of decreasing TFP in Mexico due to the political unrest experienced in the economy but its effect was not felt in the long run while Indonesian economy did not feel the impact of government stability because of occurrences like labor unrest at the Grasberg mine and numerous strikes elsewhere in the economy. In Nigeria, government stability affected TFP positively due to the democratic dispensation experienced in the economy since 1999 which gives room for conducive business environment to enhance productivity while Turkey experienced negative effect of government stability but the effect was later positive. This implies that the positive effect overrode the immediate negative effect with the help of large inflows of FDI and a successful macroeconomic stabilization program implemented in the country to improve productivity.
In conformity to the a priori expectation, the relationship between law and order in the long run was positive in Mexico, Indonesia, and Turkey, but it was only significant at 5 per cent in Turkey. In Nigeria, law and order was said to be negative and insignificant. Also as expected, law and order (but its lag of one year in Turkey) impacted positively and significantly at 5 per cent on TFP in Mexico and Turkey while its impact on TFP in Indonesia and Nigeria was negative and insignificant in the short run. This result of positive relationship aligns with Hulten (2001) who says that rule of law which is part of law and order is necessary to improve TFP. By implication, the impact of law and order on TFP was found to be insignificant in Nigeria and Indonesia as a result of violence and unrest (such as Boko Haram and Niger Delta Avengers in Nigeria; labor unrest and numerous protests in Indonesia, particularly) that disrupt government and economic activities in the economies. Mexico’s productivity would increase as a result of the immediate effect of law and order in the economy while the effect in the long run was affected by the assassinations of ruling party’s presidential candidate, the secretary general of the party and brother of the assistant attorney general in 1994. However, Turkey’s law and order showed positive association with TFP both in the long and short run. This means that maintaining law and order would increase productivity in Turkey but the recent ISIS crisis in Turkey would definitely cause reduced productivity.
Furthermore, time (trend) was said to be an important variable in determining TFP in Mexico only because no significant result was achieved without trend and as a result, time was said to be positively and negatively related to TFP in the long run and short run, respectively.
For the panel study, FDI, human capital, corruption and government stability were said to be affecting TFP significantly at 5 per cent level in the long run while inflation and law and order had no significant long-run effects on TFP in the MINT countries. There was a positive relationship between TFP and each of human capital and government stability and a negative association between each of FDI, corruption and TFP. On the other hand, the short-run result indicated that there was a significant direct relationship between TFP and each of human capital and corruption with an indirect relationship between TFP and inflation in the MINT countries. All other determinants (i.e., FDI, government stability and law and order) had no significant short-run effects on TFP in the MINT countries. This implies that governments of the MINT countries pay attention to educational system since a positive relationship of human capital with TFP was recorded both in the long run and short run. In spite of this, there is still a high level of risk in each country, particularly in terms of foreign investment. This is because FDI contributes to the effects of other regressors either through positive or negative externalities. Also, there are still even more obstacles and further to go in terms of development in several sectors, notably, infrastructure, corruption and social capital that can reduce productivity.
Concluding Remarks
The study examined the determinants of TFP in MINT during the period 1980–2014. The motivation for this study was as a result of what Jim O’Neil of Goldman Sachs did in 2014. He classified these countries as emerging economies that are expected to show strong growth and provide high returns for investors over the coming decades. Consequently, there is need for sustainable economic growth and development if they are to show strong growth. Sustainable growth has been found to depend not only on capital accumulation but also on TFP and this brings about examining those factors that can drive it in these emerging economies.
Using a sample of annual data on real gross domestic product (real GDP), labor force (LF), gross fixed capital formation (GFCF), foreign direct investment (FDI), human capital (HUMC), inflation (INF), corruption (COR), government stability (GOS), and law and order (LAO) covering the period from 1980 to 2014 which were sourced from the databases of The World Bank (2014) and International Country Risk Guide (2014). Data collected were analyzed using appropriate descriptive statistics and econometric techniques.
The study therefore concluded that human capital and corruption were seen to be key drivers of TFP in the MINT countries both in the long run and short run and if the MINT countries want to improve their TFP, the governments should create and promote conducive business environment to attract FDI (especially manufacturing FDI) that is necessary to improve TFP, pay more attention to educational system (primary, secondary, and tertiary) of their respective countries in order to build and/or boost human capacity, make sure that the inflation rate is moderate and stable in order to increase the demand for final goods and services which will in turn lead to increased production, and as a result, improved productivity, corruption which is seen as the use of public gain for private gain should be tackled and fought by the appropriate agencies, should ensure stability in form of good governance, fairness, honesty, justice, and a careful nurture of democracy through good education and maintaining law and order because this helps dealing with occurrences of theft, violence, and disturbance of peace.
Further research should therefore look into other variables that can affect TFP in the MINT countries apart from the variables used in this study.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors received no financial support for the research, authorship, and/or publication of this article.
