1.1 Background and Purpose of Research

The average age of construction workers is rising, and the number of new hires is declining, leading to a shortage of skilled laborers in the field. This shortage has caused an inadequate supply and inefficient allocation of skilled workers across construction sites. To address this, a skill grade system was introduced to attract new workers and enhance their job prospects. The system aims to stabilize the employment structure of skilled workers while systematically fostering and managing their development.

The current method for calculating skill grades in the construction industry relies on a combination of on-the-job experience, qualifications, education and training, and recognition days earned through construction skills competitions, which are converted into years of experience. The average number of days reported per year for conversion varies by occupation. A key issue with this approach is its heavy reliance on years of experience as the primary indicator of a worker's abilities, which can undermine trust within the field. This lack of reliability in the grade criteria erodes confidence in the system, ultimately reducing efficiency in the construction industry. The average number of reported working days across 60 different occupations is approximately 71 days per year. However, when excluding weekends and public holidays, the potential maximum working days in a year is estimated to be around 250. This discrepancy suggests that the current standard for working days is set too low. For example, rebar workers are allocated 110 days annually, while demolition workers are only given 31 days, resulting in a significant gap of approximately 79 days between occupations.

Furthermore, as the skill grading relies solely on converted working days, it is essential to establish a more accurate and appropriate standard for the average working days used in the grade system. This study seeks to enhance the accuracy and fairness of the grade system by gathering feedback from construction workers regarding actual working days, analyzing climatic conditions, and deriving a suitable legal standard for annual working days.

1.2 Research Methods and Scope

This study evaluates the validity of the construction skill grading system by analyzing the classification and management criteria for workers. The analysis employs three primary approaches: a survey of construction workers, data analysis using the integrated database (DB), and an assessment of climatic conditions affecting annual working days. Additionally, a simulation is conducted to verify the appropriate distribution ratio of skill grade.

The current formula used for calculating on-the-job experience is represented by the following equation, as shown in equation (1).

Chapter 2 reviews previous studies on the construction skill grade system and analyzes the classification and management criteria for skill grades. To enhance the skill grade classification criteria for workers, Chapter 3 presents a survey targeting workers and outlines the methodology for analyzing trends in working days using the integrated database (DB) and climatic conditions to derive annual workable days. Chapter 4 interprets the results from Chapter 3, determines the appropriate number of working days by occupation, and verifies the optimal distribution of skill grades through simulation analysis using the Exponential Smoothing (ETS) algorithm in Excel. Finally, Chapter 5 concludes the study. The research methodology is illustrated in <Fig. 1>.

Fig. 1. DIagram of Research methodology

2.1 Prior Research on Operation of The Construction Skill Grading System

The necessity for a grading system for skilled workers has been a recurring theme in research, with several studies focusing on its introduction and design. One such study is that of the ^{Korea Vocational Training Center (2019)}, titled "Construction Skilled Workers' Career Calculation [Integrated DB Model Construction]." This study highlights that data pertaining to skilled workers is currently managed by seven separate organizations, each overseeing a distinct database by field.

The research primarily analyzed data mapping between the employment insurance DB and the retirement deduction DB, identifying two key limitations in the data management process. Firstly, the cumulative matching rate for primary salaries is 73.79%, while for secondary salaries it is only 13.89%, indicating that the DB management is not systematic. The study recommends the unification and systematic management of the retirement deduction and employment insurance DB to address these issues, thus emphasizing the importance of integrated data management.

In the field of education and training for construction workers, a notable study by ^{Son (2019)}, titled "Analysis of the Actual Situation of Skills Training for Production Construction Workers and Recommendations for Improvement," employed a comprehensive survey approach that covered 25 sites and 24 occupations. The study analyzed 573 questionnaires and found that production construction workers were most likely to receive apprenticeship-style skills training from their seniors or coworkers on construction sites. Despite the availability of government-funded training programs, many workers were reluctant to participate due to a lack of tangible benefits and the inability to generate income during the training period. The study also identified a significant limitation of apprenticeship training: the quality of the training largely depends on the skill of the trainer. It recommended implementing periodic skills training on construction sites and linking such training to career management systems like the construction skill grading system.

A related study conducted by the ^{Korea Construction Industry Research Institute and the Korea Association of Building and Construction Craftsmen (2022)}, titled "Education and Training System for Construction Worker Skills Classification," highlighted that the current system relies solely on converted experience for skill grade classification. The study proposed incorporating additional criteria, such as certification acquisition and job-related training programs, to better assess skill grade promotion.

^{Kim & Kim (2023)} evaluated the practical effects of the construction skill grading system, revealing that only 2,000 workers had registered in the system within the first year following its introduction in 2021. To enhance market acceptance and stabilize the system, they suggested several solutions. These included establishing four independent grading criteria-'experience,' 'qualifications,' 'education and training,' and 'rewards'-and analyzing the relationship between these factors and the system’s policy goals, such as 'influx of new workers,' 'improvement of job performance,' and 'reduction of attrition' through regression analysis. The study concluded with four recommendations: first, the integration of the e-card system with the relevant wage structure; second, proactive use of the e-card system to reduce discrepancies between ratings and actual competencies; third, the establishment of an educational framework to enhance job performance; and fourth, the need for policy adjustments to ensure job stability. This study identifies key challenges and offers forward-looking recommendations for improving the system.

In 2021, the Korea Institute of Vocational Proficiency conducted a study on the classification and management standards for skill grades, titled "Introduction-based Design Study of the [Construction Skilled Person Classification System]." The study proposed a skill classification framework that considers field experience, education, qualifications, and awards. It categorizes over 60 occupations in construction and suggests how to set skill grades, ranging from beginner to special grades, based on years of experience. The study identified a limitation in the current system, where the sole criterion for distinguishing skill grades is years of experience. This can result in an overrepresentation of higher grades (advanced and special), which may not accurately reflect workers' actual skills, complicating the recruitment of skilled labor.

Since the early 2000s, the use of exponential smoothing for workforce forecasting has been explored in various academic fields. While direct research on construction workforce simulation is limited, studies like ^{Bang & Park (2022)} have successfully applied time-series analysis methods such as exponential smoothing and ARIMA models. These methods, which give more weight to recent data, are ideal for short-term forecasting, including predicting labor supply and demand.

The current skill grading system was implemented on May 27, 2021. However, the system’s limitations include its reliance on the integrated database (DB) for determining reference days for grade classification, without considering workers’ actual opinions or regional climate conditions. This study aims to address these gaps by collecting input from construction workers and factoring in regional climate conditions to derive more realistic estimates of the average annual working days, improving the accuracy of the skill grading system.

3.1 Designing a Questionnaire to Improve the Construction Worker Skills Grading System

After establishing the objectives, a questionnaire was developed to gather insights into the construction skill grading system and the appropriate number of working days. A structured survey was created and administered via Google Forms. The survey was designed as follows <Table 1>.

The main contents of the questionnaire included basic information about the workers (such as gender, age, region, work field, and department), career history (including average monthly and annual working days, total working period, certificates, education and training completed, and awards), and career-related perceptions (such as length of field experience, acquisition of certificates, completion of education and training, and awards in skill competitions).

To analyze the relative importance of these factors, the Analytical Hierarchy Process (AHP) was employed, using a seven-point Likert scale for comparative evaluation.

The respondents' fields of work included 198 in construction, 1,041 in plants, and 552 in other sectors such as dredging, diving, cultural property construction, substations, power distribution, internal and external electricity, information and communication, general welding, and specialty welding.

3.2 Analyzing Retirement Deduction and Employment Insurance Data to Analyze the Trend for Construction Workers

In the study to determine the criteria for the functional class, a design study was conducted to establish the classification criteria based on the integrated database (DB), retirement deduction, and employment insurance data. To improve the analysis, this research focuses on data from 2015 onward, when occupational classifications were systematically recorded, and incorporates recent data from 2021 and 2022 to reflect trends following the implementation of the construction skill grading system. Each DB was compared and integrated by considering overlapping dates, and the number of working days per year for each worker was calculated. The data analysis process is illustrated in <Fig. 2>.

Fig. 2. DB Analysis Process

As an initial step, a dataset containing retirement deduction and employment insurance data was prepared. The dataset was organized by year using Python and the Pandas library to facilitate structured analysis.

In the subsequent process, employment-related data—including new deduction subscriber numbers, occupation categories, reference year, work month, and workdays—were extracted from the prepared employment insurance and retirement deduction Excel files. Each sheet was converted into CSV format to ensure compatibility for further processing. The converted retirement deduction and employment insurance CSV files were then merged into a single CSV file, with duplicate data removed. The overall process is illustrated in <Fig. 2>.

However, raw data entries that were incomplete or did not adhere to the prescribed format were excluded from the analysis. The new deduction subscriber numbers and primary occupations were referenced based on 2022 data.

During the process of deriving the average annual working days from the reported employment records, it was observed that the presence of inactive workers significantly lowered the overall average. To enhance the efficiency of the analysis and identify areas for improvement, a criterion was established to define and exclude inactive workers from the dataset.

Inactive workers were classified based on working days not covered by employment insurance. Specifically, a worker was classified as inactive for a given year if they had fewer than 96 working days annually (i.e., an average of fewer than 8 days per month). Additionally, workers were considered inactive if they had fewer than four years of active employment within the eight-year dataset.

3.3 Analyzing Climatic Conditions to Derive Annual Workable Days

The currently implemented standard for average annual working days varies across construction trades. However, the existing calculation methodology, which relies on aggregating post-collected employment data, presents limitations that make deriving an objective and standardized metric challenging. To address this issue, this study utilizes climatic data—an objective and reliable source—to establish an integrated standard for annual working days.

Before applying climatic data to determine the appropriate number of working days, this study considers climate-related restrictions that affect construction activities. Additionally, regional climatic conditions are incorporated to ensure fairness in the classification criteria. Since most major construction processes involve outdoor work, they are highly susceptible to meteorological conditions. Non-working days due to public holidays and scheduled leave were first taken into account, followed by an assessment of work restrictions arising from extreme seasonal conditions, such as winter cold waves and summer heat waves, to estimate the number of weather-induced work stoppages.

A comprehensive analysis of weather-related work restrictions identified four primary climatic factors that significantly impact construction activities: low temperatures, high temperatures, wind speed, and snow accumulation. To establish standardized workability criteria, this study references national construction regulations and technical specifications, including the Korean Standard Specifications (KCS), Expressway Construction Specifications (EXCS), LH Corporation’s Standard Specifications (LHCS), and Seoul Metropolitan Government Construction Standards (SMCS). By synthesizing the meteorological restrictions stipulated within these regulatory frameworks, a unified climate-based work restriction standard was formulated. This standard prioritizes key meteorological conditions that have a direct influence on major construction processes. The final classification criteria are presented in <Table 2>.

3.4 Determine the Appropriate Number of Working Days and Simulation Intervals

To establish the appropriate number of working days, a Focus Group Interview (FGI) was conducted with representatives from the Ministry of Employment and Labor, the Construction Workers Mutual Aid Association, relevant associations and organizations, and research institutes. The number of working days was reviewed and adjusted to a suitable grade, considering public holidays, weather-related non-working days, and redundancy days.

Previous research suggests that skill grading systems should be reviewed for validity every three years. Additionally, among the 97 nationally recognized private qualifications, 28 require renewal every five years. Based on these qualification renewal criteria, an analysis of the skill grade distribution ratio was conducted, and simulations were performed to project the distribution ratio five and ten years into the future. The baseline number of working days used in the simulation was determined by comparing the number of working days derived from the FGI meeting with the current system.

The dataset used for the simulation was obtained from the retirement deduction DB and employment insurance DB, covering the period from 2015 to 2022, with projected working days extended to 2032. The total number of workers reported during this period amounted to 3,684,210.

However, as previously mentioned, individuals with duplicate records, missing entries, or formatting errors were excluded from the analysis. The dataset included 60 representative occupations classified under the Construction Workers Mutual Aid Association, excluding the "no occupation" category. For the simulation, the criteria for classifying Inactive Workers used in the DB analysis were also applied. Specifically, the simulation incorporated the average inflow and outflow of workers, excluding those with fewer than 96 working days per year. Calculations were performed by occupation to ensure accurate projections.

4.1 Analyzing Survey Response Results to Improve Construction Worker Skill Classification System

To analyze the survey responses, the 60 occupations classified by the Construction Workers Mutual Aid Association were grouped into 12 occupational categories, and an Analytical Hierarchy Process (AHP) analysis was conducted. The rationale for grouping was based on the Construction Skill grade System Classification and Management report, which indicated that as of the third quarter of 2023, 18 occupations had fewer than 1,000 workers. Given the limited number of survey respondents from these smaller occupational categories, it was necessary to consolidate occupations with similar working environments to enhance the reliability of the analysis.

The occupational classification was conducted in two stages. In the first stage, the 60 occupations were categorized into 16 groups in accordance with Article 7 of the Enforcement Decree of the Framework Act on the Construction Industry. In the second stage, these 16 groups were further consolidated into 12 groups, based on the Ministry of Land, Infrastructure, and Transport's (MOLIT) Professional Construction Industry Classification data. The final classification structure is presented in Table 3.The occupational classification was conducted in two stages. In the first stage, the 60 occupations were categorized into 16 groups in accordance with Article 7 of the Enforcement Decree of the Framework Act on the Construction Industry. In the second stage, these 16 groups were further consolidated into 12 groups, based on the Ministry of Land, Infrastructure, and Transport's (MOLIT) Professional Construction Industry Classification data. The final classification structure is presented in <Table 3>.

The survey results for construction workers were analyzed based on groupings. In terms of the average days worked per month, more than half of the respondents worked 20 or more days. In Group 11, welding, 91% reported working 20 or more days.

Like the average monthly working days, the average annual working days indicate that more than half of the workers in each group reported working over 210 days per year. Among the occupational groups, Group 11 (welding occupations) exhibited the highest average number of working days, with 85% of respondents reporting annual working days exceeding 210 days.

Fig. 3. Survey Response Results (Monthly Average)

Fig. 4. Survey Response Results (Annual Average)

When calculating the average annual working days for each occupational group, the overall average number of working days per year across all groups was found to be 178 days. Applying this to the current advancement criteria—entry-grade (0–3 years), intermediate (3–9 years), advanced (9–21 years), and extraordinary (21+ years)—the distribution of workers was as follows: 52.66% at the entry-grade, 28.36% at the intermediate grade, 17.55% at the advanced grade, and 1.43% at the extraordinary grade.

Furthermore, a survey was conducted to assess the perceived importance of on-the-job experience, qualifications, education and training, and recognition days awarded as prizes for placing in construction skills competitions in skill classification. The findings are presented in <Fig. 5>.

Fig. 5. 7-Point Scale During the Survey

The AHP analysis, which was consistent across different age groups, provided valuable insights into the priorities for skill classification in the construction industry. The findings showed that on-the-job experience was by far the most important factor, with 53.5% of respondents ranking it highest, followed by qualifications at 21.5%, education and training at 14.3%, and recognition days for achievements in skills competitions at 10.7%. This highlights the strong emphasis placed on practical experience, suggesting that the current criteria for skill progression based on years of experience-specifically, the 10.5 years for craftsmen and 4.9 years for industrial engineers-might be overly stringent. A reduction in these requirements could potentially better reflect the real-world skill grades of workers.

The analysis also reveals that education and training are currently underrepresented in the experience-based classification system. This calls for a recalibration of how training and educational milestones are factored into skill classification, so that they can play a more significant role in worker advancement.

Additionally, the low percentage (only 4%) of workers who had received recognition for achievements in skills competitions indicates a need for further inclusivity in the system. Since 37 occupations currently lack skills competitions, expanding the criteria for recognition would ensure that workers in all areas of construction have opportunities for acknowledgment and advancement, ensuring equity across different roles within the industry.

Overall, the findings suggest that the current system could benefit from updates to better align with workers' actual skills and achievements, particularly by integrating education and training as key components of the skill classification system.

4.2 Results of the Integrated Data to Analyze the Trends of Labor Days for Construction Workers

The analysis of the annual average number of working days revealed a significant difference based on whether Inactive Workers were included or excluded. When Inactive Workers were excluded, the average number of working days per year was 241 days, compared to just 162 days when they were included. This shows how the presence of Inactive Workers, who contribute fewer working days, distorts the overall average and the classification outcomes.

When examining the distribution of workers across skill grades, the data showed a marked difference depending on whether Inactive Workers were considered. Including Inactive Workers, 27% of workers were classified at higher skill grades. However, when these workers were excluded, this proportion dropped to less than 13%. This highlights a clear disparity in the skill grade classification based on whether Inactive Workers were factored in, suggesting that the current system might not be accurately reflecting the true distribution of skill grades among workers. The low percentage of workers in the highest skill category is largely attributed to the inclusion of workers who do not work primarily in construction, often leading to an inflated classification at lower skill grades.

Furthermore, the findings indicate that the current classification criteria are insufficient for distinguishing workers at higher skill grades, pointing to the need for a more nuanced and systematic classification framework. The present system seems to overlook important factors like the consistency and nature of employment, which are essential for determining skill grade more accurately. This suggests that refining the classification standards to better reflect these elements would create a more accurate and fair system for categorizing construction workers.

4.3 Derive Annual Workable Days Using Analyzed Climate Conditions Results

In estimating the annual average number of working days, the analysis took into account various factors, including weather-related non-working days, official leave days, and overlapping days. It was found that except for Jeju Island, most regions experienced an average of 80 days impacted by adverse weather conditions, such as temperature fluctuations, precipitation, snowfall, and wind speed. Taking these factors into account, it was estimated that climatic conditions contributed approximately 84 non-working days per year, assuming seven non-working days each month due to weather.

Official leave days were calculated based on the “Regulations on Public Holidays for Government Offices,” which indicated a maximum of 68 public holidays per year. When 52 Saturdays were included, the total number of leave days amounted to 120. However, overlapping days between public holidays and Saturdays were not factored into this initial calculation.

To enhance accuracy, overlapping days between weather-related non-working days and official leave days were considered. The calculation followed the method outlined in Article 7 of the Ministry of Land, Infrastructure, and Transport Ordinance No. 1140, which sets standards for determining the construction period of public construction projects. By incorporating these overlapping days, along with monthly leave days and weather-related non-working days, the total number of overlapping days was calculated to be approximately 25. Consequently, the estimated annual average number of working days, factoring in climatic conditions, was concluded to be 186 days.

Using the 186-day average, the distribution of workers across various skill grades was analyzed. The results showed that 56.04% of workers were classified as Beginner, 26.21% as Intermediate, 16.85% as Advanced, and 0.90% as Expert. The combined percentage of workers in the Advanced and Expert grades was 17.75%. These findings emphasize a skewed distribution, with a majority of workers classified at the entry-grade skill grade, suggesting room for improvement in both skill development and the system’s ability to accurately reflect higher-grade expertise.

4.4 Simulation Comparison to Check Appropriateness of Newly Defined Workdays

The dataset used in this analysis, derived from the retirement deduction DB and employment insurance DB from 2015 to 2022, originally contained a total of 3,684,210 unique deduction subscriber numbers. However, to ensure the accuracy and reliability of the results, records with formatting errors, missing data, or inconsistencies were excluded. This step was essential to maintain the integrity of the analysis and ensure that only valid and complete records were included for further processing and calculations. By refining the dataset in this way, the study was able to provide more reliable insights into the annual working days and skill grade distributions.

4.4.5 Analyzing Simulation Results

The Exponential Smoothing (ETS) algorithm, a time-series forecasting method in Excel, was utilized for the simulation analysis. The ETS model assigns greater weight to more recent data while progressively decreasing the weight of older data, making it a suitable technique for forecasting future trends based on time-series data. Exponential smoothing is commonly used in forecasting applications, such as predicting tourism trends and workforce demand, and was selected in this study due to its ability to emphasize recent working day trends when projecting future workforce estimates.

Based on the number of working days from 15 to 22 years, the model was used to forecast working data up to 32 years. The predicted results were then summed to reflect the inflow and outflow of workers by occupation, as shown in equation (3).

However, for occupations that were not classified under a specific trade, such as general laborers without an occupational designation, the data was excluded from the skill classification ratio calculations.

Table. 10 Simulation of Expected Values When Maintaining Current Working Days

Year Range	Basic	Intermediate	Advanced	Special
2015 ~ 2022	32%	27%	22%	19%
2015 ~ 2027	27%	24%	21%	28%
2015 ~ 2032	23%	22%	21%	34%

Table. 11 Simulation of Expected Values When Changing to 186 Working Days

Year Range	Basic	Intermediate	Advanced	Special
2015 ~ 2022	54%	27%	18%	1%
2015 ~ 2027	49%	25%	18%	8%
2015 ~ 2032	44%	23%	19%	14%

The decline in the proportion of workers at the Beginner grade and the increase in the proportion at the Expert grade exhibit similar trends in both <Table 10> and <Table 11>. However, when maintaining the current classification system, as presented in <Table 10>, the Advanced and Expert categories collectively account for more than 50% of the total workforce, resulting in an inverse pyramid structure that deviates from the conventional pyramid-shaped skill distribution. This overrepresentation in higher skill grades may lead to diminished credibility in the classification system and reduced differentiation among skill grades, as illustrated in <Fig. 6>.

Fig. 6. 2032 Expected Graph

By standardizing the number of working days to 186 days, the proportion of workers in the Advanced and Expert categories remains below 20%, allowing for greater control over skill grade distribution. Moreover, simulations projecting the distribution five and ten years into the future under this criterion indicate a pyramid-shaped skill structure, aligning with expected workforce trends. The findings suggest that unifying the standard number of working days to 186 days enhances the distinction between skill grades, facilitating more effective classification management and ensuring a balanced distribution of skilled workers over time.