공공조직의 순위 평가 방안: 확률적 변경분석과 자료포락분석의 적용

정수관,원두환 한국자료분석학회 2014 Journal of the Korean Data Analysis Society Vol.16 No.6

Stochastic frontier analysis and data envelopment analysis are widely used for ranking evaluating in efficiency of public institute performance, and most previous research have measured efficiency and ranking evaluation by an analysis method. However, decision-makers are interested in efficiency ranking in order to support selective parts, but they can be confused with the differences in ranking evaluation by analysis methods. This study shed light on the complimentariness in stochastic frontier analysis and data envelopment analysis in ranking evaluation. Return to scale in efficiency of public institute is tested by stochastic frontier analysis. Then, a type of data envelopment analysis is determined by the result of return to scale. The differences of evaluation ranking between two analysis are examined. We find that the ranking order in efficiency is significant differences between the analysis and suggest a weighted average of two ranking evaluation as an alternative efficiency ranking. The method can be applied to other public areas. 공공조직의 성과를 평가하기 위해 확률적 변경분석과 자료포락분석이 많이 활용되고 있는데, 대부분의 연구는 각각의 분석법을 이용하여 효율성 측정에만 집중하였다. 그러나 두 가지 분석법은 장단점이 있기 때문에 각 방법의 특성을 이해하고 보완한다면 효율성 측정의 정확성을 향상시킬 수 있다. 본 연구는 확률적 변경분석과 자료포락분석을 이용하여 효율성을 각각 평가하고 두 결과를 보완하고자 한다. 이를 위해 부산시 28개 공공도서관을 사례로 효율성을 측정하였다. 먼저, 확률적 변경분석에서는 공공도서관의 투입-산출 변수를 이용하여 규모수익을 검증하고, 도서관의 효율성을 측정한다. 둘째, 확률적 변경분석의 결과를 통해 규모수익여부에 따라 적합한 자료포락분석을 선택한 후 개별 도서관의 효율성을 측정한다. 셋째, 자료포락분석의 모형 선택이 잘못되었을 경우에 효율성을 측정하고 비교하여, 확률적 변경분석을 통해 규모수익 확인의 중요성을 보인다. 마지막으로 분석방법의 차이점을 이해하고 각 분석을 통해 추정된 효율성에 가중평균을 이용하여 새로운 효율성 지표를 만들고자 한다. 분석결과 확률적 변경분석을 통해 부산시 공공도서관의 효율성은 규모수익체증이 나타남을 알 수 있었고, 규모수익을 고려하지 않은 자료포락분석은 효율성과는 측정에 왜곡이 발생하였다. 또한 확률적 변경분석과 자료포락분석에서 추정된 각각의 효율성을 가중평균하여 새로운 효율적 지표로 사용할 수 있다. 이러한 평가방안은 공공도서관을 포함한 다양한 공공조직의 효율성 측정에 응용할 수 있을 것이다.

R&D투자와 수출의 생산효율성 제고 효과-SFA기법을 활용한 실증분석-

신범철,이의영 한국기업경영학회 2010 기업경영연구 Vol.17 No.1

This paper empirically examines determinants of technical efficiency of Korean manufacturing firms over the period 2000~2005, focusing on the effects of R&D expenditure and exports. In particular, the paper comparatively analyzes the technical efficiency of large firms and SMEs (small medium sized enterprises) by using the micro panel data. Two main approaches to the systematic measurement of production technical efficiency have been developed in the economic literature: DEA (data envelopment analysis) and SFA (stochastic frontier analysis). The former involves the application of the linear programming technique to trace the efficiency frontier, while the latter applies the econometric technique to estimating the various production frontiers. However, the usefulness of DEA for modeling technical efficiencies has been controversial, specially among econometricians. The main drawback of DEA is that any deviation of an observation from the production frontier must be attributed to inefficiency. Therefore, there is no provision for statistical noise or measurement error in the model. For this reason, we employ parametric SFA to measure the technical efficiencies. In analyzing the technical production efficiency, many studies employ two-step estimation method. In the first step, they estimate a stochastic production function and predict firm level efficiencies using various forms of production functions. Then, in the second step, the predicted efficiencies are regressed on firm-specific variables in identifying any sources for differences in predicted efficiencies across firms. However, the two-step method can suffer an inconsistency problem caused by assuming individual independence of the inefficiency effects in both estimation stages. To alleviate this inconsistency problem, we estimate simultaneously a stochastic frontier production function and the technical inefficiency model by the maximum likelihood estimation. The empirical results of the paper can be summarized as followings. First, the empirical results suggest that ignoring firm-specific heterogeneity in estimation of production function allowing for technical inefficiencies may lead to biased estimates of technical inefficiency. Second, production technology of manufacturing firms during the sample period is found to be progressive and translog production function can represent firm technology better than Cobb-Douglas production function could. Given the translog production function, the hypothesis that manufacturing firms have no technical inefficiency can be rejected statistically. This implies that the stochastic fron tier approach might be more appropriate rather than the standard production function approach. Third, exports expansion would not lead to an increase in technical efficiency of manufacturing firms over the period after Korea's financial crisis. Fourth the empirical results suggest that there is an inverse U shape of non-linear relationship between technical inefficiencies and the size of the firm. This implies that an expansion of firm size may worsen production technical efficiency over some level of firm size. Finally, the empirical results show that R&D expenditure would be positively associated with technical efficiency in small and medium sized firms but not in large firms. This suggests that government subsidy for enhancing productivity would be more effective in small and medium sized firms than in large firms. 이 논문은 2000~2005년 기간 동안 한국 제조업체의 R&D투자와 수출의 확대가 생산의 기술적 효율성에 미치는 효과를 확률변경분석(SFA) 기법을 활용하여 실증·분석하였다. 특히, R&D집중도와 수출비중이 생산효율성에 미치는 효과를 중소기업과 대기업을 구분하여 비교하였다. 이 연구의 핵심적인 결과를 요약하면, 우선 수출과 R&D, 그리고 생산성간의 관계를 분석함에 있어서 기업 간 기술적 이질성을 고려하는 것이 중요하다는 것이다. 따라서 산업별 기술적 효율성에 관한 분석만으로 기업의 고유한 특수적 상이성을 파악하는데 한계가 있음을 보이고 있다. 둘째, 고려된 표본기간 내에서 국내 제조업체의 생산기술은 기술진보적이며 콥-더글라스 형태보다는 초월대수 생산함수가 보다 적합함이 통계적으로 나타났다. 또한 고려된 모든 유형의 생산함수에서 기술적 비효율성이 없다는 귀무가설이 통계적으로 기각될 수 있었다. 이는 국내 기업들이 비효율적으로 생산하기 때문에 일반적인 생산함수를 사용하는 것보다 확률변경 생산함수를 사용하여 추정하는 것이 보다 적합함을 의미한다. 셋째, 전체적으로 수출비중은 통계적으로 기술적 효율성에 부정적인 영향을 미치는 것으로 나타났다. 즉, 기업의 매출액 대비 수출 증가가 기업의 생산 효율성을 오히려 떨어뜨린다는 것이다. 이 결과는 한국의 수출 증대가 생산성 향상과 고용 증가를 유발한다고 하더라도 기술적 효율성에는 오히려 부정적인 영향을 미치고 있음을 시사한다. 넷째, 생산의 기술적 효율성과 기업규모는 逆U자형의 비선형관계를 갖고 있는 것으로 나타났다. 이는 기업규모가 확대됨에 따라 생산의 효율성이 증대되지만 일정 수준 이상으로 확대될 경우 오히려 효율성은 증가된다는 것을 의미한다. 마지막으로, 제조업체 전반적으로 R&D집중도는 생산효율성을 향상시키는 것으로 나타났다. 기업규모에 따라 구분해서 보면 R&D집중도 증가가 중소기업에서는 생산효율성을 향상시키는 것으로 나타나지만 대기업에서는 통계적으로 유의한 영향을 미치지 못하는 것으로 나타났다. 이러한 결과는 정부의 R&D 지원이 대기업 보다는 중소기업에서 생산효율성 제고 효과가 크게 발생할 수 있음을 시사한다

지역별 의료 자원의 성과 효율성 분석

장인수,권대영,Hong Sok(Brian) Kim 한국보건사회연구원 2017 保健社會硏究 Vol.37 No.4

In order to propose a method to measure the performance of medical resources, this study empirically analyzed the impact of regional medical resource input on age standardized mortality rate through probability frontier analysis. If prior studies so far focused on the operational efficiency of institutions that use medical resources, this study focused on performance efficiency in the macro perspective. As is already well known, the probability frontier method measures the inefficiency of the analysis object in consideration of the error variation, overcoming the disadvantages of the existing data envelope analysis method. In our empirical analysis, we examined16 metropolitan cities and provinces in Korea for the period 2006-2013, and employed the panel stochastic frontier method and Malmquist Productivity Index Analysis by putting the number of doctors, the number of specialists, the number of beds, the number of operations and the age standardized mortality rate. The result showed that all the medical resources allocated to each region decreased the age-standardized mortality rate. Regional inequality in medical resources distribution should be dealt with in depth from various perspectives. In this respect, the method of applying the stochastic frontier analysis to measure the performance of regional medical resources has the advantage that the inefficiency of each input element can be examined more in depth, and this study is meaningful in this respect.

확률적 생산프런티어를 이용한 국내 화력발전부문의 효율성 평가

유상열 한국자료분석학회 2012 Journal of the Korean Data Analysis Society Vol.14 No.1

The purpose of this study is to evaluate technical efficiencies of Korean thermoelectric power companies for the 2003-2010 period. We estimate a stochastic frontier production function with an exponential specification of time-varying efficiencies which incorporates panel data associate with observations on the sample of 5 thermoelectric power companies over 8 years. We also compare efficiencies of stochastic frontier analysis (SFA) with those of data envelopment analysis. Total number of employees and fuel usages (coal, oil, LNG) are used as input variables, and electricity is used as output variable. The main results of our paper are summarized as follows. First, maximum likelihood estimates for the stochastic frontier function show that there exist technical inefficiencies during 2003-2010, and efficiencies increase in a monotone over time. Second, technical efficiency measures based on SFA range between 0.723 and 0.821 during the above period. Third, the average efficiencies of total period are 0.774, 0.845, 0.938 based on SFA, CCR model and BCC model, respectively. Finally, S Power Co. is the most efficient in comparison with others based on SFA. The main contribution of this study is that we estimate a stochastic frontier production function and test time variant technical efficiencies of Korean thermoelectric power companies. 본 연구는 확률적 생산프런티어를 이용하여 국내 5개 화력발전회사의 연료 및 노동 효율성을 평가하기 위해 수행되었다. 분석기간은 2003년부터 2010년까지 8년간이며, 화력발전부문의 산출물은 전력판매량, 투입물로는 석탄, 유류, LNG, 생산직 종업원 수를 선택하였다. 연구결과를 요약하면 다음과 같다. 첫째, 확률적 생산프런티어를 최우법으로 추정한 결과 기술적 비효율성이 존재함을 검정하였다. 둘째, 국내 화력발전부문의 기술적 효율성은 시간 가변적이며, 시간이 경과함에 따라 효율성이 증가하였음을 알 수 있었다. 셋째, 추정된 확률적 생산프런티어를 이용하여 효율성을 평가한 결과 연도별 평균값은 2003년 0.723에서 2010년 0.821까지 단조 증가하였다. 개별 DMU 중에는 S발전이 가장 효율적인 것으로 나타났다. 넷째, 전체 표본에 대한 평균 효율성은 확률적 생산프런티어에 의한 값이 0.774이었고, 투입기준 CCR 효율성은 0.845, 투입기준 BCC 효율성은 0.938이었다. 본 연구는 확률적 생산프런티어를 적용하여 우리나라 화력발전부문의 연료 및 노동 효율성을 평가하고 효율성 값이 시간의 흐름에 따라 증가하였음을 통계적으로 검정하였다는데 공헌점이 있다.

Stochastic Frontier Analysis를 이용한 제약회사의 효율성과 그 결정요인분석

사공진 ( Jin Sakong ),김정규 ( Jeongkyu Kim ) 한국보건행정학회 2015 보건행정학회지 Vol.25 No.2

Background: In 2012, Ministry of Health and Welfare announced 5-year plan of the development of the pharmaceutical industry, which pointed out the problems of the pharmaceutical industry as the insufficient investment in research and development (R&D) and the unfair business practice, and suggested the technological innovation, market transparency and nurturing global enterprises as solutions. It is quite interesting to study which factors determine the efficiency of the pharmaceutical firms. Methods: To estimate the efficiency of the pharmaceutical firms and the determinants of their efficiency, methods of the stochastic frontier analysis and the panel study are applied to the data of 60 domestic pharmaceutical firms from 2006 to 2012. The efficiency is derived by the distance between the decision making units and the most efficient frontier curve. Results: First, the result of the stochastic frontier analysis shows that overall efficiency of the pharmaceutical firms is increasing as time goes by. If classified by the scale, the larger firms show higher efficiency and if classified by the degree of innovation, the innovative firms show higher efficiency compared to the less-innovative ones. Second, the result of estimation of the determinants of efficiency shows that the firms with larger expenses of the sales promotion and the entertainment, show lower efficiency. Conclusion: The first result implies that the scale of the firm and the investment in R&D explain the efficiency of the pharmaceutical firms. Therefore, it is necessary to increase the investment on the fundamental researches to vitalize R&D of the new drugs. The second result can be explained by the structural characteristics of the small pharmaceutical firms which produce generic drugs. Therefore, the government tries to make the pharmaceutical firms to optimize of the expenses of the sales promotion and the entertainment, and to increase R&D with the policies such as to boost up the R&D investment and to prohibit rebates, etc.

Measuring efficiency and ICT ecosystem impact: Hardware vs. software industry

Lee, Kyoungsun,Park, Yuri,Lee, Daeho Elsevier 2018 TelecommunicationsPolicy Vol.42 No.2

<P><B>Abstract</B></P> <P>The ICT industry is reshaping itself through convergence and coopetition across the ecosystem. In particular, many ICT companies have been trying to achieve both hardware and software capabilities to accelerate innovation. To understand the changes in the ICT industry, we empirically study the hardware and software industries using the ecosystem framework. We estimate the efficiency levels of the hardware and software industries in China, Japan, South Korea, and the United States (US) using stochastic frontier analysis and meta frontier analysis, and we examine the effects of the ICT ecosystem structure on efficiency levels using a Tobit regression. We find that the US, which is the global ICT leader, performs significantly better than other countries, with the biggest technology gap ratios in both hardware and software, and South Korea, which has the most hardware-centric industrial structure among the four countries, has the biggest efficiency gap between the hardware and software industries. Tobit results further reveal that a well-balanced ecosystem is important in achieving high efficiency in both hardware and software, and the effect is much higher in the software industry than in the hardware industry.</P> <P><B>Highlights</B></P> <P> <UL> <LI> We study the hardware and software industries using the ecosystem framework. </LI> <LI> We estimate the efficiencies of the hardware and software industries using SFA. </LI> <LI> We compare the efficiency results using meta-frontier analysis. </LI> <LI> We examine the effects of the ICT ecosystem structure on efficiency levels. </LI> <LI> A well-balanced ecosystem is important in achieving high efficiency. </LI> </UL> </P>

The Effects of Human Resource Factors on Firm Efficiency: A Bayesian Stochastic Frontier Analysis

신상우,장혜정 국제문화기술진흥원 2018 International Journal of Advanced Culture Technolo Vol.6 No.4

This study proposes a Bayesian stochastic frontier model that is well-suited to productivity/efficiency analysis particularly using panel data. A unique feature of our proposal is that both production frontier and efficiency are estimable for each individual firm and their linkage to various firm characteristics enriches our understanding of the source of productivity/efficiency. Empirical application of the proposed analysis to Human Capital Corporate Panel data enables identification and quantification of the effects of Human Resource factors on firm efficiency in tandem with those of firm types on production frontier. A comprehensive description of the Markov Chain Monte Carlo estimation procedure is forwarded to facilitate the use of our proposed stochastic frontier analysis.

The Effects of Human Resource Factors on Firm Efficiency: A Bayesian Stochastic Frontier Analysis

Shin, Sangwoo,Chang, Hyejung The International Promotion Agency of Culture Tech 2018 International Journal of Advanced Culture Technolo Vol.6 No.4

This study proposes a Bayesian stochastic frontier model that is well-suited to productivity/efficiency analysis particularly using panel data. A unique feature of our proposal is that both production frontier and efficiency are estimable for each individual firm and their linkage to various firm characteristics enriches our understanding of the source of productivity/efficiency. Empirical application of the proposed analysis to Human Capital Corporate Panel data enables identification and quantification of the effects of Human Resource factors on firm efficiency in tandem with those of firm types on production frontier. A comprehensive description of the Markov Chain Monte Carlo estimation procedure is forwarded to facilitate the use of our proposed stochastic frontier analysis.

The Effects of Human Resource Factors on Firm Efficiency : A Bayesian Stochastic Frontier Analysis

Sangwoo Shin,Hyejung Chang 국제문화기술진흥원 2018 International Journal of Advanced Culture Technolo Vol.6 No.4

This study proposes a Bayesian stochastic frontier model that is well-suited to productivity/efficiency analysis particularly using panel data. A unique feature of our proposal is that both production frontier and efficiency are estimable for each individual firm and their linkage to various firm characteristics enriches our understanding of the source of productivity/efficiency. Empirical application of the proposed analysis to Human Capital Corporate Panel data enables identification and quantification of the effects of Human Resource factors on firm efficiency in tandem with those of firm types on production frontier. A comprehensive description of the Markov Chain Monte Carlo estimation procedure is forwarded to facilitate the use of our proposed stochastic frontier analysis.

국세징수의 효율성 평가와 그 결정요인 탐색-확률적 프런티어 생산함수의 적용-

유상열 한국세무학회 2013 세무학 연구 Vol.30 No.1

The purpose of this study is to measure technical efficiency and explore determinants of the efficiency in Korean regional tax offices (RTO) from 2000 to 2010. We estimate a stochastic frontier production function with an exponential specification of time varying efficiency which incorporates panel data associated with observations on the sample of 6 RTOs over 11 years. The amounts of tax collection (Tax) are used as a output variable. We measure inputs by 4 indicators:Tx payers (Taxpayer), gross regional domestic products (GRDP), number of employees (L), and district offices of regional tax jurisdiction (K). The contextual variables determining the efficiency of RTOs are direct tax ratio (Dirtax), indirect tax ratio (Indirtax), direct tax payer ratio (Dirtaxpayer), indirect tax payer ratio (Indirtaxpayer), number of tax agents (Agent), and tax concentration ratio (Tax concentration). The main results of this paper are summarized as follows. First, we found that Taxpayer, GRDP, K are positively related with Tax, where as L has a negative influence on Tax. The efficiency of the RTOs is time invariant, however there is a possibility the efficiencies would have decreased during the period. Second, the average technical efficiencies of each RTO range between 0.358 and 0.518. The efficiency rankings of RTOs are different from each other depending on the assumption of inefficiency distribution, implying that we must consider the alternative efficiency measures in holistic fashion when the efficiencies of RTO are evaluated using a stochastic frontier analysis. Third, we found out that the determinants of efficiency are Dirtax (-), Agent (+), and Tax concentration (+) under the assumption that technical inefficiencies are distributed non negative truncation, more over, Indirtax (+) and Dirtaxpayer (+) are added in the determinants under the assumption of half normal distribution for the inefficiency. The main contribution of this study is that we attempt to measure the efficiency of RTOs using a stochastic frontier production function and discover the determinants of efficiencies in Korean RTOs for the first time. The insights gleaned from this paper can be helpful when RTOs handle the combinations of output and inputs in order to upgrade their efficiencies. 본 연구는 2000년부터 2010년까지 국내 6개 지방국세청(RTO)의 국세징수의 효율성을 평가하고, 효율성 결정요인을 탐색하기 위해 수행되었다. 산출요소는 국세징수액, 투입요소는 납세자 수, 지역내총생산, 직원수, 관할 세무서 수이다. 국세징수의 효율성에 영향을 미치는 변수로는 직접세비율, 간접세비율, 직접납세자비율, 간접납세자비율, 세무대리인비율, 과세집중도를 선택하였다. 효율성 평가를 위해 확률적 프런티어 생산함수(SFP)를 추정하였다. 주요 연구결과를 요약하면 다음과 같다. 첫째, SFP를 추정한 결과, 투입요소 중 납세자․지역내총생산․관할세무서 수가 많고 직원수가 적을수록 산출요소인 국세징수액이 많은 것으로 나타났으며, 효율성의 시간가변성은 통계적 유의성이 부족하였다. 둘째, 분석모형에 따라 개별 RTO의 효율성 평균은 0.358~0.518로 나타났다. 개별 RTO에 대한 효율성 순위는 효율성의 시간가변성보다는 비효율성의 분포에 대한 가정에 따라 차이가 있었다. 셋째, 기술적 비효율성이 절단된 정규분포(non negative truncation)를 따른다고 가정한 경우 효율성에 영향을 미치는 변수는 직접세비율(-), 세무대리인(+)과 과세집중도(+)이었고, 기술적 비효율성이 절대정규분포(half normal distribution)를 따르는 경우에는 상기의 요인 이외에 추가로 간접세비율(+)과 직접납세자비율(+)인 것으로 분석되었다. 본 연구결과는 다음과 같은 시사점을 제공해 줄 수 있다. 첫째, 상대적으로 효율성이 낮게 평가된 RTO가 효율성을 향상시킬 수 있도록 효율성에 근거한 인센티브 제도를 도입할 필요가 있다. 둘째, SFP에 의해 개별 RTO의 효율성을 평가할 때는 기술적 비효율성의 분포에 대한 가정에 따라 효율성 순위가 달라지므로, 대체적인 효율성 측정치를 모두 반영하여 평가하는 것이 바람직하다. 셋째, 연도별 평균 효율성이 2000년도 이후 지속적으로 감소했을 가능성이 있으므로, 효율성을 개선하기 위해 산출 및 투입요소를 조정하는 대책이 필요하다. 본 연구는 개별 지방국세청의 산출 및 투입요소를 이용하여 생산함수를 추정하고 투입요소가 산출요소에 미치는 영향을 추정하였으며, 효율성 결정요인을 탐색하였다는 데 공헌점이 있다. 국세징수의 효율성을 개선하기 위해 투입요소의 배합을 변경하거나 효율성에 영향을 미치는 것으로 파악된 변수를 조절하는 정책을 시도할 수 있다.