Journal Archives

The first legal effort into technology development in renewables was discovered in 1987, enacting “Alternative Energy Development Promotion Act” to carry forward renewables’ technology development as well as start commercializing and diffusing solar-thermal and waste energies after oil crises occurred twice in the 1970s. Currently, the government specifies eight renewable energies (photovoltaic, solar-thermal, wind energy, bioenergy [including combustible renewables], ocean energy, geothermal, hydropower, and wastes [including industrial waste]) and three new energies (fuel cell, hydrogen, and integrated gasification combined cycle (IGCC; hereafter, referred to as renewable energy]) to promote based on the “Act on the promotion of development, use and diffusion of new and renewable energy” since 1997, and the act has been revised three times until now. The current 4th act, announced in September 2014, is aiming at increasing the use of renewable energies to 11% of total primary energy supply (TPES).

Financial support for R&D activities in renewables has been given since 1988, and its total R&D investment combined by government and private sectors until 2013 is KRW 3.71 trillion (USD 3.71 billion if USD 1 = KRW 1,000), which increased especially sharply in the last decade from 2004 to 2013 at a 25% average annual growth rate. The support in 2010 and 2011 accounted for 6.07% and 4.92% respectively out of IEA countries’ renewable energy R&D expenditure.^[1,2] Despite the current bold renewable energy R&D investment policy, absolute total investment during the last few decades is regarded as far behind those of other countries like the United States, Germany, France, Japan, and the United Kingdom.

The R&D tasks are strategically divided into short-term tasks for technologies with potential to be commercialized in the near future and medium- and long-term tasks performed in order to acquire future core technology. In the short-term tasks, the pragmatic technologies, like photovoltaic (PV), wind energy, and fuel cell, which can be utilized for current early diffusion by lowering electricity generation cost, supporting commercialization and overseas market expansion, and linking R&D with diffusion policy, are subjected preferentially to a heavy investment. PV received the most R&D investment among the eleven renewable energies both in number of projects and in amount of financial investment: 275 projects accounting for KRW 777.2 billion in R&D investment for new and continuous projects. Fuel cell and wind power are the next, representing 120 and 127 new projects, respectively, and KRW 747.38 billion and KRW 513.36 billion, respectively, in R&D investment for new and continuous projects.^[3] These three energies, which are designated as priority supporting energies, account for 54.95% of total renewable R&D investment.

R&D activity can be also classified according to performer (university, research institute, and large and smaller firm) or its technical progress phases (basic, applied, and development). By technical progress phase, as shown in Fig. 1, 759 new and continuous R&D projects took place on development phase, which accounts for nearly 70% of total R&D projects, gaining the most financial support, of KRW 556,431.7 million, between 2008–È2012.^[4] The research on the development phase was mostly carried out by large and smaller firms and basic and applied researches by universities and institutes.

Fig. 1. 
Renewable energy R&D support by technical progress phase from 2008-2012 (Source; NTIS www.ntis.go.kr)

Fig. 2. shows a trend in R&D investment from 2008–2012 by performer, and the large firms are taking the most advantage of R&D investment among beneficiaries, representing nearly half of the R&D of KRW 243,223.38 (54.01%) from 2008 to 2012, while others like smaller firms, institutes, and universities received KRW 147,461.80 (32.75%), KRW 43,313.50 (9.62%), and KRW 16,311.64 (3.62%) respectively during the same period.^[5]1)

Fig. 2. 
Renewable energy R&D support by perfroermes from 2008-2012 (Source; NTIS www.ntis.go.kr)

While public R&D investment in renewable energy technology has decreased in recent years, with KRW 277,304 million in 2012, KRW 271,963 million in 2013, and KRW 249,305 million in 2014,^[6] the investment trend in terms of average amount per project is getting larger, and considerable R&D investment is still being delivered into large firms whose research is mainly big and on the development phase that can generate economical profits in the short term and focus on a system field that includes activities such as standardization, planning, demonstration, operation, performance evaluation, and so on.

In this context, the approach of delivering heavy investment to large firms may need to be reconsidered. As firms grow large, efficiency in R&D is weakened due to the loss of managerial control or excessive bureaucratic control as well as the incentives of individual researchers being weakened due to decreasing their ability to capture the benefits from their individual efforts or conservative characteristics of the hierarchies of large firms that may frustrate their creativity.^[7]

However, there are distinctive benefits that favor the large firms over the smaller firms.^[8] For instance, the large firms are able to conduct risky R&D in an imperfect capital market by utilizing their internally generated funds, which are more available and stable as firms grow and provide economies of scale to reduce the risk associated with the prospective return from R&D activity. In addition, there are economies of scale in R&D activity and the returns from R&D are higher in firms that have a larger volume of sales that can be used to spread out the fixed cost generated during R&D activity.

In Korea, the selectively concentrated R&D investment on large firms that conduct systematic researches mostly on development phase in order to achieve a short-term diffusion goal may slow development in core technology development on components, equipment, and materials in which smaller firms are interested.^[9] For instance, the industrial import dependency in the PV module, a core component for PV, is highly dependent on import (46.6% in 2008) because of the lack of reliability, the shortage of skill, low price competitiveness, and so on.^[10] Moreover, the import dependency in wind energy power plant installed in the country was 99.6% in 2008, since the generating firms prefer the foreign product due to the low technological competitiveness in supporting core components and materials in charge of domestic smaller firms.^[11,12]

Despite extensive efforts in R&D since 2008 to catch up with advanced technology, the level of commercialization derived from technology supported by Korean R&D activity and export of domestic goods (that is considerably important for Korea whose domestic renewable energy market was only 0.5% of the global market in 2011) remain sluggish. This is largely due to the lack of core technology needed to lead in the international market. The Korean market share in the international renewable energy technology was 0.27% in 2012, increased slightly compared to 0.18% in 2008.^[13] Sales in renewables were KRW 3,268 billion in 2008 and increased to KRW 7,515 in 2013; exports were USD 1,706 million in 2008 and increased to USD 4,770 million in 2011; this trend then entered a downturn, with USD 2,523 million in 2012.^[14]

There have been several quantitative and qualitative analyses on renewable energy R&D productivities in Korea. Korea still has a lower interest in R&D investment in renewable energy technology compared to other OECD countries under the condition that GDP per capita is excluded due to R&D increasing as GDP increases.^[15] It seems that return on R&D is becoming more visible in sales and export along with recent R&D increases, especially in PV and wind energy technology; however, the technologies are not yet sufficient to compete in the international market, and the requisite R&D has been lacking, partly due to a lack of national R&D support.^[16]

As results of quantitative research, the extended R&D support in renewable energy technology would be desirable in Korea to increase GDP.^[17] The efficiencies of renewable energy R&D in terms of paper, patent, and engineering fee increased yearly as well as being shown to be comparatively more efficient than other energy programs such as nuclear energy.^[18] By energy source, wind power is shown to be the most efficient in terms of the government support including R&D and promotion compared to other renewables such as fuel cells and PV.^[19] On the other hand, it has been also shown that if renewable energy R&D performance described representatively as paper and patent are relative to the economical value, stability, profitability, growth, and innovation in firms who received R&D support to spend for renewable energy technology are not increased significantly compared to the firms that did not receive R&D support. This indicates that R&D support in renewable energies is not being connected to the firms’ performance by technical commercialization.^[20]

The results of research may be interpreted variously according to analysis methodology and research purpose, but there is not doubt that the aim for R&D activity is to achieve profits in the future. The concept of productivity is naturally valid in the renewable energy technology that is regarded to be significant as a future growth engine for Korea, and it is important to assess and enhance the R&D productivity in a qualitative way beyond quantitative growth. However, there have been no empirical studies to estimate Korea R&D productivity in renewable energy technology by performers (larger and smaller firms as well as institutes and universities) that are playing an essential role in conducting public R&D activity as well as by technical phases (basic, applied and development) whose purpose to perform are dissimilar one another.

DEA is developed as one of nonparametric production frontier methodologies to analyze efficiency for like public projects or non-profit firms that the price information on input or/and output is normally not given or units of measure to be estimated are different or difficult to synthesize as one index. In addition, it allows to handle multiple inputs and multiple outputs generated sporadically through out the process, namely the methodology is useful for estimating R&D productivity since it owns intrinsically various inputs and output. It assumes that a condition of Pareto-Koopman efficiency that a unit’s efficiency cannot be increased unless other’s efficiency decreases.^[33]

Farrell (1957) firstly introduced the efficiency analysis using multiple inputs and multiple outputs to measure a firm and it was developed by Charnes, Cooper and Rhodes (1978) who proposed DEA for the first time, which is input-oriented DEA model on a constant return to scale (CRS); CCR model, and by Banker, Charnes and Cooper (1984) who distinguishes technical efficiency (TE) and scale efficiency (SE) since the firms’ R&D activity is not possible to operate at optimal scale in practical; BCC model.^[34,35,36]

According to Golany et al. (1989), DMUs used for DEA analysis should satisfy some homogeneity conditions in order to have the result with economic significance.^[37] The conditions are the projects are performed under similar purpose, the DMUs are existing in the homogeneity market, and all input and output data are in every DMUs and the data are different each other. In addition, the proper number of DMUs should be existed since it is less plausible that majority of DMUs would be efficient if the number of DMUs are less than the number of input and output variables. There is not a unified standard as to the number of DMU but most of papers use the standard suggested by Fitzsimmons et al. (1994) described as below

K stands for number of DMU and N and M are the number of variables of input and output data.^[38]

When n decision-making units (DMUs) are to be evaluated, each DMU_j(j = 1, ..., n) consumes m inputs (i = 1, 2 ..., m) in order to produce s outputs (r = 1, 2 ..., s). Clearly, DMU_j(j = 1, ..., n) uses amounts X_j = x_ij inputs (i = 1, 2 ..., m and x_ij > 0) and produces amounts Y_j = y_rj of outputs (r = 1, 2 ..., s and y_rj > 0). In addition, there are two properties to ensure a piecewise linear approximation to the efficient frontier and the area dominated y the frontier; convexity and inefficiency. That is,∑j=1nλjxiji=1,2,…,m and ∑j=1nλjyrjr=1,2,…,s are possible inputs and outputs attainable by the DMU_j where λ_j(j = 1, ... n) are positive scalars and the same outputs can be produced by using more inputs; the same inputs can be used to produce less outputs.

The CRS assumption is suitable when all firms are operating at an optimal scale, but it is not possible in practical due to external factors like imperfect market condition, government regulation, etc.. Therefore, BCC model based on variable returns to scale (VRS) conditions, which can divide TE and SE simply by adding the convexity constraint; λ_j=1. The input-oriented model estimates the inputs in each DMU that can be minimized while the outputs are maintained; on the contrary the output-oriented model finds the outputs in each DMU that can be maximized at the current inputs.

Either input or output oriented DEA model is optionally selectable to use for the analysis for R&D productivity according to its research purpose. Input oriented model is suitable to the case in order to estimate minimum input variables at the current output maintained, that is the model is capable to acquire minimum R&D investment or/and human resources retaining current output level like number of patent, paper, or volume of sales. In this paper, the efficiency is estimated based on output-oriented BCC model due to the researches and public opinion that the R&D performance is not sufficient to the current level of R&D investment on the increase and the qualitative improvement in R&D productivity is necessary bringing up maximum outputs under the present R&D support.

The efficiency score of output-oriented DEA based on VRS condition (output-oriented BCC model), φ^*, is calculated as below

subject to

If ∅^*≠ 1 input and outputs slacks can be expressed as

DMU_o is less efficient not only if ∅^*≠ 1 but also if ∅^* = 1 and si- and/or sr+ are non-zero for all i. Then, the input and output slacks are estimated, which make ∅^* optimize, are also estimated

subject to

Finally the two-state output-oriented BCC model can be evaluated as

subject to

Next, the VRS model is able to separate TE and SE, which may be calculated as the ratio of TE on the assumption of CRS to TE on the assumption of VRS. The technical efficiencies on the basis of VRS, θ^* for the input-oriented model and ∅^* for the output-oriented model, are given already by calculations above and the TE under the CRS can be estimated without the convexity constraint; λ_j=1. Therefore, TE under input-oriented based on CRS assumption is estimated as

subject to

and TE under output-oriented based on CRS assumption

is described as

subject to

Finally, SE(X_j, Y_j ) and TE (X_j, Y_j) for each DMU is

for input-oriented BCC model,

for output-oriented BCC model,

and for TE.

The descriptive statistic of data from 2008 to 2012 is shown in Table 1 as below. During the period, 1340 projects were performed for renewable energy technologies under the supervision of Korea Energy Technology Evaluation and Planning (KETEP): 280 for large firms, 380 for medium-sized firms, 343 for institutes, 316 for universities, and 21 for others.

Data are classified according to technical progress phase and performers. In basic research, a large amount of R&D investment and workers are dedicated to universities and institutes, accounted for 71.3% of total R&D investment and 76.5% in total number of workers. On the other hand, average R&D investment and number of R&D workers are the higher in the large firms, which performed only 10 basic research programs, compared to the universities and institutes, which conducted 82 and 60, respectively, during the same period. R&D investment and workers in smaller firms are the least among the performers, but the average is higher than those for universities and institutes, though lower than that for large firms. In terms of the outputs, the direct outputs are more produced in universities and institutes both in total and average, but less in economical outputs presented as engineering fee and commercialization The outputs as applied research accomplished in number gives weight to institutes.

In applied research, a share of R&D investment and workers in the institutes are the most between performers but the large firms are still the largest beneficiary per project. Similarly for basic research, the institutes beside the inputs, and the economical outputs are lower than the firms’. It is notable that the smaller firms produce more economic outputs than large firms compared with the inputs, the least in total amount among performers.

The researches on development phase has the largest share of R&D investment and workers in total compared to other research phases since they are believed to generate economical returns on the R&D spend. The large firms account for the largest proportion of R&D investment and workers, with 67.6% and 65.7%, respectively, and are assumed to perform larger projects relative to those of other performers. Smaller firms have the next highest proportions for both figures. The absolute total figures in outputs are superior to those of other research in different technical phases, but it is noticeable that the averages are not distinctly different among performers.

The result gained using the output-oriented VRS model is described in Table 2 as below. First of all, efficiency of the 1,227 DMUs out of the 1340 DMUs including the researches that did not belong to any other technical progress phase is estimated by basic, applied, and development and then classified by performer as large and smaller firms, institutes, and universities. Thus, the results of efficiency and slack analysis shown in Table 2 are the average figures for each group by technical progress phase and by performer.

The overall average efficiency was 0.72 for basic, 0.60 for applied, and 0.50 for development research, which indicates the lowest efficiency score. Distinguishing the efficiencies by performer, universities’ efficiency in basic research shows the highest 0.78 more than average, and the efficiencies in other performers are nearly similar, though the large firms’ are slightly lower. In DMUs in applied research, the efficiency in smaller firms is exceptionally higher than other performers’, with 0.75, followed by large firms with 0.69; thus, the average efficiency scores of applied research in the firms exceed total average efficiency score of 0.60. The level of efficiencies in development research is almost alike among performers and it is also conspicuous that they are somewhat lower than those of other technical progress phases.

According to the result allowing to show the amount of input and output slacks that would be increased or decreased for improving efficiency, the estimated numerical values vary considerably among performers and technical progress phases. Regardless of performer, the results of efficient target that maximizes efficiency by manipulating inputs in R&D investment and labor to have maximum output level show that the current level of inputs are not required to achieve the current level of outputs. That is, outputs could be increased without increasing the current amount of inputs.

Moreover, DEA can provide information to show the DMUs whose efficiency score is 1. The DMUs in basic research phases are the most efficient, while almost two thirds of DMUs in development phase, whose project accounts for nearly 70% of R&D in number, are inefficient. In other words, the projects in the development phases show the lowest efficiency in terms of number with score ‘1’ compared to basic and applied researches. Looking at the results more in detail by performer, the combined smaller and large firms seem to perform more efficient R&D projects on average, with relatively more efficient value in applied research than universities and institutes. The number of DMUs with efficiency with score ‘1’ in development research in the firms is slightly higher than in the universities and institutes. In addition, the efficiencies between small and large firms are also not notably different, and the efficiency in smaller firm is rather higher than large firms, at 35.5% and 32.5%, respectively.

It is necessary to consider for this study if there are efficiency of scale between DMUs since the efficiency can be increased or decreased as scale increases. The scale of R&D investment and labor as input variables, as well as their outputs produced in various forms, vary by performer and technical progress phase. As a result, regardless of characteristics divided by performer and technical progress phase, there is more in efficiency of scale than technical efficiency, meaning the efficiency is increased as of the size of the DMU is increased.