Research on technology-driven polarization in Korean labor market

학위논문(박사)--서울대학교 대학원 :공과대학 협동과정 기술경영·경제·정책전공,2019. 8. 황준석.기술 진보는 경제 성장의 주요 성장 동력이지만, 중숙련 노동자에 대한 수요 감소와 함께 고숙련 및 저숙련 노동 수요를 증가시켜 고용 양극화를 주도해왔다. 장기적인 구조적 저성장, 투자 및 고용 부진, 고령화 등 한국 경제가 당면한 과제 해결을 위해 지속적인 기술 개발과 동시에 기술로 인한 충격과 변화의 속도를 통제하고 이에 적응하기 위한 대비책 마련을 간과해서는 안 되며, 정책 입안자들은 기술이 노동시장에 미치는 충격의 정도와 깊이를 이해하고 기술의 양면을 고려해야 한다. 노동시장 내 심화되는 불평등의 원인을 규명하기 위해 기술 진보가 숙련된 기술자를 선호하는 방향으로 일어나 발생한 저학력과 고학력 근로자 간 임금격차에 관한 숙련편향적 기술변화 연구가 시작되었다. 그러나 이 가설은 고숙련 및 저숙련 노동 수요가 함께 증가하고 중숙련 노동 수요만 감소하는 양극화 현상을 설명하지 못한다. 정형대체적 기술변화 가설은 반복적이며 잘 정의되고 쉽게 코드화 될 수 있는 직무들이 컴퓨터나 기계와 같은 기술로 쉽게 대체될 수 있다는 점에 주목한다. 한국의 경우, 기존 연구들은 2000년 중반까지 숙련편향적 기술변화에 의한 숙련프리미엄이 유효하게 나타난다는 증거를 찾지 못했으며, 학력프리미엄이 뚜렷하게 나타나지 않는 특성을 보인다. 양극화에 대한 많은 연구가 있지만, 기술의 정형적 직무 대체로 이를 확인한 국내 연구는 거의 존재하지 않는다. 직무 중심 분석은 한 직종 내에서 실제로 더 많이 사용되거나 더는 사용되지 않을 직무를 구분해 앞으로 근로자들이 어떤 기능을 더 강화해야 할지 파악할 수 있는 중요한 분석 도구이다. 본 연구는 직종이 여러 직무로 구성되어 있다고 가정하고, 비정형적 분석 업무, 비정형적 육체 업무, 정형적 인지 업무, 정형적 육체 업무로 각 직무를 분류하고 한국의 정형대체적 기술 변화를 분석하였다. 정형적 업무는 쉽게 프로그래밍할 수 있는 간단하고 구체적이며 반복 가능한 작업으로 기술 진보에 따라 대체 가능성이 증가한다. 직종의 업무 강도 측정을 위해 미국의 직업분류 사전과 ONET 직업정보 네트워크 데이터베이스를 사용해, 각 직종에서 공통으로 활용되는 직무들로 직무 네트워크를 구성해 직무 분류의 적합성을 확인하였다. 첫 번째 연구에서는 기술이 발전의 결과로 ICT 가격이 하락할 때 정형적, 비정형적 업무의 노동수요 변화가 어떻게 일어나는지 확인한다. 연구 결과 ICT 자본은 비정형적 분석 직종과 보완적 관계에 있지만, 정형적 및 비정형적 육체적 직종과는 대체 관계에 있음을 확인하였다. 두 번째 연구는 한국의 정형대체적 기술변화 현상을 분석한다. 지난 20여 년간 중간 소득 및 중숙련 노동자들의 수요 성장은 다른 직종에 비해 느리게 증가하고, 중간 소득층에 대한 노동 수요 감소와 더불어 나타나는 근로 소득의 변화는 추후 임금 양극화로 이어질 가능성이 있음을 시사한다. 마지막 연구는 한국노동패널 데이터를 사용해 중숙련 개별 노동자의 직종 이동 방향을 분석한다. 중간 소득 근로자는 주로 저임금, 비정형적 육체적 직종으로 이동해 그 결과 임금이 감소하는 반면, 원래의 직종을 유지한 근로자들은 임금이 지속적으로 상승하였다. 본 연구는 기술이 노동 시장에 미치는 영향을 확인하기 위하여 측정 가능한 방법론을 제안하고, 기술과 인간이 공존하는 사회에서 인간에게 요구되는 소양이 무엇인지 명확하게 가르쳐준다. 가까운 미래에 노동시장에서는 독창성, 대인 소통 및 상호 작용, 문제 해결, 정보 분석 등의 능력이 점차 더 많이 요구될 것이고, 이런 능력을 갖춘 인재를 키워내기 위한 교육과 투자를 위한 시의성 있는 새로운 방안 수립을 위해 본 연구의 결과가 활용될 수 있을 것이다.Technological advancement is the primary growth engine for economic development. At the same time, it has caused growing disparity in the labor market, by increasing the proportions of high and low skilled or high and low paying occupations to the detriment of the middle. To boost economic growth by enhancing productivity and overcoming the reduction of Koreas economically active population, further investment in technology should be consistently promoted. However, to control and stabilize the repercussions of technology implementation, it is necessary to bolster existing institutions and prepare people for this digital transformation. Therefore, policy makers should consider both the positives and negatives of technology to truly, understand the extent and depth of its potential disruption in the labor market. Studies of polarization have focused on the increasing differential between the earning potential of high school and college graduates based on wage premiums for higher skills. This has previously been explained as skill-biased technological change (SBTC) wherein technology development has traditionally favored workers with better skills and paid higher wages accordingly. However, recent demand for low-skilled employees has also increased, while occupations requiring middle-level skills have decreased. This polarization is accounted for by routine-biased technological change (RBTC) that is now prevalent in many developed economies. RBTC assumes that technology can substitute routine tasks within occupations, tasks that are simple, repeatable, well-defined, and codifiable. Yet to be determined is whether technology-driven polarization exists in Korea. To date, the literature has failed to find evidence of significant wage premiums based on educational attainment or SBTC. Studies seldom manage to present empirical results demonstrating the effects of technology on inequality until the mid-2000s. There have been various attempts to explain polarization in Korea, but they rarely considered RBTC based on a task-oriented approach. Task-oriented analysis allows us to determine which tasks have increased demand and those that have become obsolete in the workplace, thereby suggesting to workers the skills they must strengthen or acquire. In this study, occupations are defined as a set of tasks performed by the employee holding that position. There are three kinds of tasks: analytic, routine (cognitive or manual), and non-routine manual tasks. Among them, routine or repetitive tasks are more vulnerable to replacement by technology as computers or robotics can perform most simple, specific, and repeatable tasks that are easily programmed. To measure the combined task intensity of each occupation, the Dictionary of Occupational Titles (DOT) and the Occupational Information Network (ONET) database developed in the United States were used to construct cross references between the Korean Standard Classification of Occupations (KSCO) and the U.S. Standard Occupational Classification (SOC). The measurement system for task intensity was tested against the co-occurrence of effectively utilized tasks within occupations using network community analysis. In this study, rather than try to guess the impact of technology on the labor market, the employment changes of routine and non-routine labor were first tested to see whether they were affected by the reduced costs of information and communications technology (ICT) through different stages of technological advancement at the industrial level. With this theoretical framework indicating the relationship between the decline of ICT costs and labor demand, it can be empirically estimated that ICT capital is complementary with analytic and high-skill occupations but serves as a substitute for routine and manual workers. Next, RBTC related to industrial occupation levels in Korea was explored. The results imply that over the last two decades there has been a collapse in the middle-income job market while employment for more routine occupations has grown slower than others. Finally, Korean labor panel data was used to trace the career changes of mid-level or routine workers and the consequential impact on their wages after these individuals were displaced from their jobs. This study indicates that middle-income workers usually transition to lower paying occupations and have no choice but to accept reduced wages in their new workplaces, while workers who continue in routine occupations tend to experience wage growth. This research offers a measurable and quantitative index to capture the technological impact on the labor market by bridging the tasks required in the workplace (demand side) with the skills that workers should possess (supply side). It clarifies the importance of human capital investment through education and stresses the demand for enhanced skills related to creativity, interpersonal communication, problem solving, information analysis, and various soft skills. It also supports actions to develop new ways of investing and educating people to excel in new skills rather than continue to generate human capital with skills that are either already obsolete or will be in the near future. Polarization in skill demand is commonly witnessed in developed countries but this is not predestined to raise labor income inequality. Countries differ in level of inequality depending on their policies. The depth and width of technological advancement and the following socioeconomic changes are determined by not only technological side like technological feasibility, regulation on technology or cost but demand and supply of skills from the labor market. Based on this study, a meticulous consideration on the labor market is expected help solving problems of both polarization and cultivating human capital.Chapter 1. Introduction 1 1.1 Research background and problem description 1 1.1.1 Socioeconomic changes brought by technological advances 1 1.1.2 Studies on rising inequality in developed countries 5 1.2 Research objective and outline 10 1.3 Contribution of this study 13 Chapter 2. Literature review: Technology and the labor market 17 2.1 Technological advances and their impact on the labor market 17 2.2 Literature on skill-biased technological change (SBTC) and Canonical model 21 2.3 Routine-biased technology change (RBTC) and task-based approach 27 2.3.1 Routine-biased technology change (RBTC) hypothesis 27 2.3.2 Task-based view on an occupation 34 2.4 Literature on the Korean labor market 36 2.4.1 SBTC studies of Korea 36 2.4.2 Polarization studies of Korea 38 Chapter 3. Measurement of task intensity 40 3.1 Construction of measurement variables 40 3.1.1 Literature using task intensity 40 3.1.2 Task intensity measurement using ONET variables 43 3.2 Confirmation of task classification 54 Chapter 4. Substitution or complementarity? The role of ICT in the labor market 60 4.1 Introduction 60 4.2 Literature review on effect of ICT on the labor market 63 4.2.1 Cases of developed countries 63 4.2.2 Literature on Korea 65 4.3 Empirical estimation model: Task wage premium 67 4.3.1 Framework of previous studies 67 4.3.2 Alternative framework for theoretical prediction 69 4.3.3 Estimation of employment share using industrial wage premium 73 4.4 Data and summary statistics 75 4.4.1 Data source of ICT capital 75 4.4.2 Data of employment and wage 80 4.4.3 Characteristics of the occupations 81 4.5 Empirical analysis results 82 4.5.1 The labor demand change across industries 82 4.5.2 The result of estimation 86 4.5.3 Estimated wage premium of industry 90 4.5.3.1 Industry specific wage premium 90 4.5.3.2 Industry-Occupation specific wage premium 92 4.6 Conclusion 93 Chapter 5. Polarized labor demand due to routine task- biased technological change 96 5.1 Introduction 96 5.2 Data 98 5.2.1 Employment, wage, and industrial data 98 5.2.2 Routineness of occupations 99 5.3 Understanding and estimating the employment structure of Korea 102 5.3.1 Polarization of the employment share in the Korean labor market 102 5.3.2 Estimating the job polarization: The structure of model 109 5.3.3 The result of estimation 112 5.4 Conclusion 117 Chapter 6. The afterwards of hollowing-out routine workers 120 6.1 Introduction 120 6.1.1 Polarization of employment 120 6.1.2 The mismatching between skills and education at workplace in Korea 121 6.1.3 The limitation of previous studies 124 6.1.4 The purpose and findings of this study 126 6.2 Data 127 6.2.1 Task intensity of occupations 127 6.2.2 Employment and wage: KLIPS 129 6.3 Empirical analysis of wage and task intensity 132 6.3.1 Explaining the wage in relation to task intensity 132 6.3.2 Relationship between task intensity and human capital 136 6.4 Hollow-out of the middle-skilled workers 140 6.4.1 Change in employment share by mean wage 140 6.4.2 Employment share changes by means of task intensity 142 6.4.2.1 Changes in employment share by task occupational groups 142 6.4.2.2 Changes in wage after the job shift 145 6.4.3 After job shifts of routine task-intensive workers 148 6.5 Conclusion 152 Chapter 7. Conclusion 155 7.1 Summary 155 7.2 Implications 160 7.3 Limitations and future study 162 Bibliography 165 Appendix 1 (Chapter 4) 177 Appendix 2 (Chapter 5) 180 Appendix 3 (Chapter 6) 182 Abstract (Korean) 186Docto