“Impact of trade liberalization on technical efficiency of mining sector: A case of selected SADC countries”

Productive inefficiency and lagging technology progress are major reasons behind the Southern Africa Development Community’s (SADC) continued exportation of unprocessed minerals to the world markets. The study seeks to uncover the impact of trade openness on the technical efficiency of the mining sector in selected SADC countries (Botswana, DRC, Namibia, South Africa, Zambia, and Zimbabwe). Technical efficien- cy is the ability of any production process to produce maximum output from minimum quantities of inputs. A Cobb Douglas Stochastic Frontier Approach in a single- stage maximum likelihood estimation of Green’s true fixed effects was used to compute technical efficiency (scores) and the technological progress in the mining sector of SADC. Results indicate that there is no technical efficiency gains from trade liberalization during the period under study together with positive and significant technological progress. A coefficient of 0.72 suggests that a 1% increase in trade openness increases technical inefficiency in the mining sector by 0.72%. The parameter coefficient from the truncated normal distribution of the true fixed effects model indicated that technological progress from one year to the next year would lead to a 2.6% increase in the output index of the mining. Technological progress in the mining sector should target upstream mineral value chains instead of only upgrading technology in one dimension of extraction. In addition, countries should collectively and gradually put across laws that force new investments in the extraction of minerals to erect processing plants in mining value addition of host countries to re-direct economies into a growth path.


Abstract
Productive inefficiency and lagging technology progress are major reasons behind the Southern Africa Development Community's (SADC) continued exportation of unprocessed minerals to the world markets. The study seeks to uncover the impact of trade openness on the technical efficiency of the mining sector in selected SADC countries (Botswana, DRC, Namibia, South Africa, Zambia, and Zimbabwe). Technical efficiency is the ability of any production process to produce maximum output from minimum quantities of inputs. A Cobb Douglas Stochastic Frontier Approach in a singlestage maximum likelihood estimation of Green's true fixed effects was used to compute technical efficiency (scores) and the technological progress in the mining sector of SADC. Results indicate that there is no technical efficiency gains from trade liberalization during the period under study together with positive and significant technological progress. A coefficient of 0.72 suggests that a 1% increase in trade openness increases technical inefficiency in the mining sector by 0.72%. The parameter coefficient from the truncated normal distribution of the true fixed effects model indicated that technological progress from one year to the next year would lead to a 2.6% increase in the output index of the mining. Technological progress in the mining sector should target upstream mineral value chains instead of only upgrading technology in one dimension of extraction. In addition, countries should collectively and gradually put across laws that force new investments in the extraction of minerals to erect processing plants in mining value addition of host countries to re-direct economies into a growth path.

Shylet Masunda (Zimbabwe), Ireen Choga (South Africa)
Impact of trade liberalization on technical efficiency of mining sector: A case of selected SADC countries

INTRODUCTION
The study is founded on recent regional debates and efforts that are directed towards accelerating industrialization through value addition and beneficiation in the mining sector of SADC countries to strengthen their regional integration through the production of higher value chain tradable goods on the advantage of natural resources. SADC countries have been pursuing economic and legal regional frameworks such as the SADC Treaty The study is motivated by the need to determine the nature of technological progress and technical efficiency in the sector and hence link trade liberalization. Africa, particularly SADC, is bestowed with useful minerals and mineral materials. There are vast reserves of platinum, manganese, gold, vanadium, diamonds, and chromite (African Union, 2009). If trade openness is a major facilitating agent of productivity growth in mining through its effect on technology spillovers, the study also questions the inability of this same technology transfer flow to forwarding linkages of the mining sector, as this would also boost mining value-added productivity. There is minimum mineral processing (forward linkages) within the SADC region and minerals are exported in raw and semi-finished form (World Bank, 2014). However, this region can realize increased returns from the export of value-added mining products, which could be a source for sustainable regional growth. As a result, the sector remains surrounded by very limited upstream and downstream linkages with the rest of the world. Is it poor technical efficiency leading to lack of mining forward linkages or trade openness failing to bring in the much-needed technological progress in the mining sector? The inquiry is crucial to SADC because mining production is a source of foreign exchange that directly contributes to economic development and growth.
Trade liberalization facilitates the dissemination of progressive technologies and technical expertise through two main channels. First, the local industry has a chance to secure innovative technologies and innovative production techniques through the importation of inputs, machinery, and equipment. In addition, competition and growing markets motivate industries to invest in new technology for higher quality (Grossman & Helpman, 1991cited in Saggi, 2002).

LITERATURE REVIEW
Productivity and growth are best explained by the exogenous and the endogenous growth models of Solow (1956), Rostow (1990), Romer (1994), Lucas (1988), Rebello (1991), Mankiw (1995), and Prescott (1997). The exogenous growth model alludes that capital accumulation drives productivity but eventually it succumbs to diminishing returns to a factor, and therefore long-run productivity growth is an exogenous technological progress phenomenon. Endogenous growth models denote that technology progress is an outcome of explicit and deliberate production activity. On the other hand, new growth models concur that long-run productivity growth is attained by either circumventing diminishing returns to scale or adopting internal technological progress (Stiroh, 2001). Technological progress is viewed differently in these theories, that is, either endogenously or exogenously determined, but they both establish that technological progress is key to the long-run and sustainable growth.
Technology gap theory focuses on the supply-side and explains the differences in national productivity as differences in technology stock (Elmslie & Vieira, 2002). Posner's imitation lag model analyzed the effect of technology on trade. The process of technological change leads to an imitation gap which then influences the pattern of international trade. The technology gap is explained by differences in expenditure on research and development. The theory explains the imitation lag in main four lags: foreign reaction lag, domestic reaction lag, imitation lag, and demand lag. Foreign reaction lag is the time taken by the innovating firm to start producing a new product followed by domestic reaction lag -time taken by the domestic producer to follow the correct path and establish stable production systems in the domestic market. Imitation lag is the time required by local firms to get familiarized with new product production technologies and take them to the market.
The theory represents a significant portion of reality on the world markets. Technology is not similar across countries and ceases to be not a public good, according to Solow (1956), but rather a product which nations have to commit resources such as research and development to attain technology endogenously. The study agrees that continuous investment in research and development is the sustainable way of growing economies as being seen in industrial economies like China, Japan, the USA, Italy, etc. This implies maintaining their status of always leading in world markets, always working to catch up timeously with changing consumer preferences and tastes. To some extent, this paper is bound to doubt the predictions made by the theory in that technology diffusion has occurred truly in some sectors but not all sectors because minimum technology transfer is recorded in the mining sector that has seen much of further processing taking place in industrialized countries and thus lacking imitation.
Phuong (2018) examined Total Factor Productivity (TFP) growth, technological progress, and efficiency in the Vietnamese coal mining industry in a non-parametric approach for a period 2007-2013. TFP of Vietnam's coal mining declined because of slow technological progress and poor efficiency. It, therefore, meant that enhancing human resource training, technology, and research and development would improve efficiency and productivity. The study utilized Data Envelopment Analysis (DEAP) method to measure and separate total factor productivity into technical change and technical efficiency. Malmquist index was also used to measure productivity growth. The decomposition enabled the identification of benefits technological progress and efficiency are bringing to the growth of the coal sector. Results indicated that TFP of the coal industry declined due to technology regress and the study recommended that to improve productivity, there is a need for enhancement of qualifications of management, human capital development, strengthening of technological innovation, and implementation of better marketing activities to grow into virgin markets and maintain traditional markets. The study informs the current one on the decomposition of TFP growth using the DEA method. Splitting of the components of TFP would help in determining the contributions made by each towards mining productivity, which is the avenue to link productivity growth and the missing mining value addition in SADC. Sahoo et al. (2017) investigated the output efficiency of the mining industry in India using Total Factor Productivity decomposition. They examined the TFP growth of the Indian mining industry from 1989 to 2014 based on the decomposed formulation of the stochastic production frontier. Productivity growth and its disintegrated components were matched over the study period. Results showed that TFP growth of the mining industry rose to 3.66% annually during 1989-2005 and to 8.76% during 2006-2014. Decomposition results reflected that the key source of productivi-ty growth migrated from technological progress (TP) to technical efficiency (TE) change in current years. Recommendations for Indian mining sector suggest that it should concentrate on undertaking innovation and upgrading the present technology. The study utilized output and input data of 128 firms. Output was measured by the real mining output from 128 firms and inputs, labor (deflated number of days worked), capital (gross fixed capital using Perpetual Inventory Method (PIM)), and deflated power and fuel expenditure of mining firms. Since in Asian countries like India productivity in mining is enhanced by technical efficiency in recent years, this study aims to research this issue and provide pieces of advice. A stochastic-Translog production frontier and FRONTIER 4.1 were used to estimate the parameters. An inefficiency model was simultaneously estimated with frontier production following Battese and Coelli (1995) using a maximum likelihood estimator. Trade liberalization proxies were the effective rate of protection, the nominal rate of protection, and the import ratio. The results concluded that trade liberalization has a positive and confirmed impact on technical efficiency. Thus, trade liberalization is a tool to enforce competition on firms and chances for firms to be more productive to endure the competition in the bazaars.
Literature confirms a positive relationship between total factor productivity growth and trade liberalization in most parts of the world. However, the study observed that literature is biased towards some sectors like export-oriented manufacturing, service, and agriculture, or mining sector. Furthermore, limited researches have decomposed TFP into technological progress and technical efficiency and linked that to trade liberalization in the mining sector. The study took productivity growth to explain value addition. The paper aims to investigate the relationship between trade openness and technical efficiency that prevails in the SADC mining sector and ways of improving efficiency.
H 0 : There are no technical efficiency and technology progress gains from trade openness in the mining sector of SADC.

METHODOLOGY
The study makes use of a time-varying technical inefficiency model by Battese and Coelli (1995) and Greene (2005 cited in Sunge & Ngepha, 2020). That is disregarding Schmidt and Sickles (1984) whose models reflected that technical inefficiency does not change over time. Nevertheless, Kumbhakar (1990) and Battese and Coelli (1992) provided evidence that technical efficiency does change in the real world. In the real world, there are opportunities to learn new methods of production in mining in the face of market competition, government regulations, and international policy frameworks ( (1977). The analysis of productivity is focused on the frontier of best practice production that measures and explains variations in Total Factor Productivity relative to the frontier and the amount by which TFP is less than the potential frontier. Thus, treating TFP as a technical change would misrepresent and misinform the idea, as TFP is a compound function. Estimation follows these steps, thus estimating the production function from which the study then derives the technical efficiency and technology change component. The technical efficiency component is then regressed on trade and other determinants simultaneously in a single-stage procedure.

Cobb Douglas Stochastic Frontier using maximum likelihood estimator
To predict the technical efficiency of industry, the first step in a one-stage procedure is to estimate a production function from which the study then derives the technical inefficiency component, which is built as shown in equation (1). Thus, the study confirms that besides inputs and stochastic noise (measurement errors, omission of variables in a vector of inputs, errors in choosing the function), technical efficiency plays a major role in improving productivity. Technical inefficiency may damage total factor productivity growth.
where y it is the i country's output (mining value addition); X it is i country's inputs; β 0 , β i , are parameters to be estimated and i and t represent country and time in years. The study used the mining value-added variable. Measuring technical efficiency scores will allow to identify the most efficient Decision-Making Unity (DMU) and assess those DMUs lagging. Hence, the paper investigates the relationship between trade openness and technical efficiency that prevails in SADC mining and ways of improving efficiency.

Determinants of technical inefficiency model
The study, therefore, agrees that besides inputs shown in equation (1), other exogenous factors can affect technical efficiency across industries, countries, and firms. A single-stage procedure that involves a simultaneous Maximum Likelihood Estimator regression on z it -technical inefficiency regressors and all stochastic frontier parameters was done (Kumbhakar et al., 2015). A single-stage procedure is known to reduce bias as compared to a two-step procedure. One of the critical assumptions is that v it and u it are identically and independently distributed (iid) of each other and the regressors (Kumbhakar & Lovell, 2000). The two-stage procedure has been proven to be misspecified (Battese & Coelli, 1995) 1 and inconsistent in assumptions about the distribution of inefficiencies. The study estimates time-varying technical inefficiency following Greene (2005) and Battese and Coelli (1995) mainly because the paper practically took the mining industry as the industry where efficiency cannot be fixed but rather it varies as technology and experience are gained over time and it is specific to a particular industry (heterogeneous).
Greene (2005) stochastic production frontier for single stage using MLE estimator is as follows: where i =1, 2......N, t = 1, 2....T and y it -output for the i th firm at t time; x it , is a vector of inputs associated with i th firm and β is the vector of inputs coefficients for the associated independent variables in the production function. T captures technological progress between two periods and technical inefficiency model by: where z it , is a vector (M*1) made up of factors that affect firm efficiency /country i at period t and ω it is the error term and δ is a vector of parameters to be estimated. The parameters δ v 2 and δ u 2 are replaced with δ 2 = δ v 2 + δ u 2 and λ = δ u 2 /δ 2 for estimation using the maximum likelihood method.
If calculated values of lambda are close to zero it rules out the presence of inefficiency however near a unit (one), it predicts that the production technology is inefficient.

Empirical framework
The study adopted the findings of Díaz-Mayans and Sánchez-Pérez (2014) who focused on innovation, exports, and technical efficiency in Spain. 1 For more information on technical inefficiency estimation address Jondrow et al. (1982).
The empirical model of the one-step procedure is starting with the stochastic frontier framework: where VA it value-added as the explained variable; K it -capital stock, the value of machinery and equipment excluding grounds and buildings; L it -total salaries and wages paid by the firm; TDtime trend; DINP it -dummy that takes the value one if there is product innovation and zero otherwise product to capture the impact of innovative activities in the frontier; DINPR it -dummy that takes the value of one if there is process innovation; Dummy it -dummy to distinguish between sectors (that takes the value of one when the firm belongs to the corresponding sector of activities otherwise this value is zero).
Simultaneously, the inefficiency determinants were estimated and particular attention was given to the effect of exports on efficiency as other determinants included firm size, investment over the capital, and proportion of external funds over value-added. The determinants of the technical efficiency model are calculated by: where Y it -mining value added (USD); K it -is the capital stock; L it -labor (worked hours in the mines); fdi it -foreign direct investment inflow proxying technology transfer, D t the time trend representing technical change.
Simultaneously, the study hypothesized the determinants of inefficiency using findings of Battese and Coelli (1995) The study further elaborates the model as: where µ it are the country's technical inefficiency scores, measured as a proportion of country i's actual output Y it to its best frontier output; X′ it β + ν it . These scores are generated during the estimation of the stochastic frontier model; γ 1 Dummy it -capturing country heterogeneity.

Estimation and diagnostic procedures
If the model has been estimated using ML, the paper tests for the absence of technical inefficiency using the Z test or LR test, or Wald test. It was decided to choose the LR test.
Thus, H 0 : λ = 0, there is no technical inefficiency effect, and deviations from the frontier are due to noise.

RESULTS
This section presents the results on the panel unit root tests, pre-estimation procedures, and the Cobb Douglas Stochastic Frontier true fixed effects and the technical inefficiency model check. Table 1 displays the unit-roots tests for the sample data the study used to calculate the total factor productivity (dependent variable) using the Hicks-Moorsteen Index from DPIN software. The same data is used to analyze the technical efficiency levels in the mining sector of the sample. Hence, Table 1 shows that mining value added (ΔLOGMVA), gross value added (ΔLOGTVA), governance (ΔLOGGVNC), and labor (ΔLOGL) fail to reject the null hypothesis that panel contains unit roots in levels and they got stationary after first difference 1(1) and are statistically significant at 1%. Whereas, gross fixed capital formation (LOGGFCF), labor (LOGL), trade openness (LOGTO), and foreign direct investment inflow (LOGFD) reject the null hypothesis that panels contain unit roots and are stationary in level 1(0) and statistically significant at less than 5%. Hence, the study concludes that variables for the technical inefficiency model are stationary both in level and after the first difference.

Generalized likelihood ratio test
The selection of the appropriate model is also based on the likelihood ratio test, which is defined by LR statistics.  Results in Table 2 and Table 3 show that the LR test statistic is greater than the critical value from Kodde and Palm (1986), meaning 323.31 > 16.07 thus concluding rejection of the null hypothesis at a 1% level of significance. Stochastic frontier is inappropriate or rather rejecting null of no technical inefficiency in the model. This implies that there are technical inefficiencies in the mining sector of SADC and hence further investigations will explore the levels and the determinants of technical inefficiency.  Table 4 shows the results of the log-likelihood test and the study rejects the null hypothesis that Cobb Douglas Ordinary Least Squares be used as the functional form and accept the alternative that the Cobb Douglas Stochastic Frontier modeling is appropriate. Again, the study also crosschecked the results with parameterized log-likelihood function for half normal model δ 2 = δ v 2 + δ u 2 and λ = δ u 2 /δ 2 and the half-normal true fixed effect according to Aigner et al. (1997). The study calculated the lambda value = 0.930344. The statistic must be close to one meaning that variations in the output are a result of technical inefficiency. In the study, technical inefficiency accounts for 93% Note: *, **, and *** denote significance at 1%, 5%, and 10% respectively.
variation in output which is very high and thus strongly justifies the use of a stochastic frontier model. Therefore, using the three justifications given here, the paper can confidently use a Cobb Douglas Stochastic Frontier modeling to estimate technical efficiency in the mining sector of SADC. Note: ***, **, and * imply significance levels at the 10%, 5%, and 1% levels, respectively. The value inside the parentheses is the corresponding standard error. Note: ***, **, and * imply significance levels at the 1%, 5% and 10% levels, respectively. Table 5 shows the three-frontier model under different distribution functions documenting that factor inputs are positive and statistically significant except for FDI, which fails to be significant when exponential (ex) and half normal (hn) distribution is used. A 1% increase in capital (LOGK) would increase mineral production in SADC under this period by 0.43%, 0.46%, and 0.40%. That is in a half normal, exponential, and truncated normal (tn) distribution assumption, respectively, thus confirming conventional theories of productivity. Again, the coefficient of labor (LOGL) is positive and statistically significant, such that a 1% increase in factor labor would boost mining productivity by 0.11%, 0.10%, and 0.14% under hn, ex, and tn, respectively. Both factors are crucial in mining production evidenced by a significant percentage contribution from each factor. Capital (LOGK) has a higher percentage contribution to production, which possibly can support the need for capital deepening and sustainable human capital development in the mining sector. This finding supports the results of Ali and Hamid (1996) and Sato and Mitchell (1989) who concluded that capital is a key booster in the growth of production and value-added. In other words, the study can ascertain low productivity from labor to gradual reduction of labor as an input in the mining sector or declining investment in human capital development. Thus, policymakers in selected SADC countries need to invest in labor productivity.

DISCUSSION
The coefficient of net foreign direct investment inflow (FDI) proxying technology transfer is positive and only significant under truncated normal distributional assumption hence a 1-unit increase in FDI would increase productivity in mining by 1.52%. Literature confirms that FDI is a pure input factor that can augment domestic capital as predicted by the Neoclassicals and from the endogenous growth perspective through technology transfer and knowledge spillovers. In Africa, FDI to extractive minerals has been positively correlated to productivity in the mining sector and other industries (Romer, 1994;Iddrisu et al., 2015;Tondl & Fornero, 2008). However, a 1.52% magnitude contribution in productivity is very moderate, which suggests the need to constantly attract new and value-added activities of FDI as well as review SADC mining policy that targets value addition in the mining sector.
Technological progress and technical efficiency are two different concepts that can either move in converging or diverging directions and apparently results do not converge. The study can certainly confirm that during the period from 1990 to 2017 there was a positive and significant association with the technology progress and mining production. The parameter coefficient from the truncated normal distribution true fixed effects model indicated that for the years under study consecutively an average of 2.6% increase in the output index was recorded in the mining sector in SADC ceteris paribus. The paper concludes that in the mining sector there has been a steady movement upwards of the production frontier, pointing to advancement in technology embodied in the capital. Interestingly, the region never experienced a technical regress, which probably could have worsened the mining progress in SADC. Technological progress enhances technical efficiency and consequently improves mining value addition through a reduction of cost of production and extending that into further processing of raw minerals. Similar results are documented by Ali and Hamid (1996), Parameswaran (2002), Jajri and Ismail (2006), and Abegaz (2013). They concur that technical change is an important ingredient of value addition. The study argues that exports from the mining sector are usually unprocessed raw minerals that fetch very low prices on the international market. This reality takes away the center of attention on improving production processes and as such exports of minerals are boosted only to achieve revenue targets rather than a desire to compete internationally through value addition and beneficiation. Hence, the study ascertains the limited mineral value addition productivity in the region to the gap between the technological progress and technical efficiency emanating from the failure of trade openness to bring in the much-needed technology transfer, adoption, and mastery of technology ceteris paribus. The findings conclude that technical efficiency cannot be improved by trade liberalizing in the mining industry but probably by adopting the best mining methods and sustainable technical capacity.
Governance (LOGGVNC) is proxied by the governmental effectiveness that is measured by the quality of public services, the capacity of civil service, independence from political pressure, and quality of policy formulation (Zhuang et al., 2010, p. 8). Due to the high correlation among the most six world bank governance indicators, the study hence chooses to use one indicator. The results indicated that good governance would reduce inefficiency in the mining industry by 1.06%. Thus, a 1% increase in government effectiveness would reduce inefficiency by 1.06% at a 1% significance level. It can be concluded that improved quality of policy formulation promoted foreign direct investment initiatives, and as such aided the efficiency in the mining sector. Similar findings are shown by Fayissa and Nsiah (2013) and Rafayet et al. (2017), who approve of the positive impact of good governance on different facets of the econo-my. This is also emphasized in the African Mining Vision: promoting good governance of the mineral sector through transparency and participatory governance would work great to unlock complex linkages at all levels and enable mineral-endowed countries to achieve sustainable socio-economic growth paths.

CONCLUSIONS
The paper seeks to determine the link that exists between trade openness and technical efficiency and also to ascertain the presence of technology change in the mining sector of SADC. The key findings are that trade openness positively affects mining technical inefficiency meaning trade openness increases inefficiency in mining production by 0.72%. The study concludes that there are no technical efficiency gains recouped from trade liberalization. Nevertheless, technological progress of 2.6% was positive and statistically significant, implying that total factor productivity change in the mining industry is credited to technological progress and good governance and not trade openness. Instead, technological progress can coexist with deteriorating technical inefficiency. This means that the mining sector failed to catch up with technology change that was instigated by trade openness. The result is not surprising since the mining sector of SADC and perhaps Africa is known as an imperfect competition market structure. The study recommends that technology progress in the mining sector should target upstream mineral value chains instead of only upgrading technology in one dimension of extraction. In addition, it is recommended that countries should collectively and gradually put across laws that force new investments in the extraction of minerals to erect processing plants in mining value addition of host countries as that could re-direct economies into a growth path. The emphasis is on major investors, multinational mining companies, that have well-established supply chains in the world markets and are leaders in innovation to spearhead mining processing in host countries. The study recommends frictionless entry and exit in the industry by investors as this would call more players in the mining sector and hence induce competition mood amongst giant players. To sustain technical efficiency, investment in human capital development allows mastery of technology and diffusion of best practice technology. Further investigations are expected to be targeted on mineral-specific productivity change to allow the designing of mineral-specific policies that may help improve mineral productivity and obtain a higher value.