ANALYZING THE EFFECT OF ROAD PERFORMANCE INDICATORS ON PENALTIES FOR LATE FULFILLMENT IN ROAD SERVICE LEVELS

The long segment policy for national road preservation comprising routine maintenance work, both road routine and condition routine maintenance, preventive maintenance, and holding with widening, rehabilitation, and reconstruction of several road sections into a single contract had been implemented in Indonesia since 2016. Reliable road performance indicators including road pavement, road shoulders, drainage, road equipment, road complementary buildings, and clearances have an impact on-road performance. Therefore, an analysis must be carried out to obtain indicators for each road performance influencing Penalties for Late Fulfillment in Road Service Levels (PLRSL) charged to contractors if they could not meet the road performance indicators based on the specified repair response time. This study aims to improve the road service level through the compliance of road performance indicators on the success of sustainable construction projects. The Structural Equation Modeling (SEM) analysis results indicated that the road performance indicator in the road preservation program had a significant effect of 77.0% on the determination of PLRSL. In addition, road pavement, drainage, road shoulder, clearance, complementary buildings, and road equipment contributed statistically significant effect on the road performance indicator with values of 88.0%, 81.4%, 80.9%, 79.1%, 78.1%, and 51.3%, respectively. These results should be a concern for contractors and the government. With a clear understanding of the significant indicators in the context of the road maintenance project, these findings could potentially contribute to the development of comprehensive pavement handling on road maintenance programs related to the quality performance of the construction projects.


INTRODUCTION
Indonesia's infrastructure quality ranking released by the World Economic Forum (2012-2019) [1] has risen from 92 nd in 2012 to 72 nd in 2019, although in 2019 it fell compared to 2017, which was 52 nd . Meanwhile, Indonesia Global Competitiveness Index (GCI) was ranked at 50 th in 2019, a decrease from 2015's achievement which was placed at 37 th . The GCI ranking and Indonesia Infrastructure Competitiveness is shown in Fig. 1. In addition, Indonesia's road quality ranking places in the middle and below Singapore, Malaysia, China, and Thailand. Road implementation fund in the 2015-2019 Strategic Plan of the Directorate General of Highways has a total State Budget plan for roads of around Rp278.18 trillion. Nevertheless, in fact, only Rp211.97 trillion (76.20% of the plan) was obtained. A study by the National Development Planning Agency in 2014 stated that the ideal cost of road investment was around Rp1,274 trillion for the period of 2015-2019 [1]. It is indicated that there was a budget backlog in the implementation of national roads. The proportion of the 2015-2019 State Budget allocation for national road preservation programs and improvement of existing roads and bridges was 54.99%, whereas 24.55% was allocated for the construction of new roads and bridges, and 20.46% for management and technical support. The national road preservation system in Indonesia, particularly the routine maintenance of roads which was formerly conducted by self-management at each Commitment-Making Officer or Area/Project Manager as the manager of road sections, had since 2016 evolved into a long segment plan. The Directorate General of Highways through the Directorate of Road Preservation implemented a long segment policy for national road preservation, which combined routine maintenance work, both road routine and condition routine maintenance, preventive, holding work with road widening, rehabilitation, and reconstruction of many road sections into single contract [2]. The quality of road maintenance work on a long segment foundation is one of the indications of success in the strategy of attaining the longevity of a road pavement plan. Division 10 [3] stipulated that contractor should carry out the fulfillment of road service levels based on road performance indicators. The fulfillment of the road service level was applied to all work outcomes included in the scope of management comprising road pavement work, road shoulders, road drainage, road equipment, road supplementary buildings (if any), and plant control [4; 5]. If the contractors are unable to meet the road performance indicators based on the stated repair response time, they will be subjected to financial penalties in the form of withdrawals from payments per day in accordance with Equation 1.
where: D : Amount of withholding payment H : Number of days of delay in repair of road service level fulfillment Pjc : Length of the defective road (does not meet performance indicators) in the specified road segment (segment length of at least 100 meters) Pjl : Length of road in the contract based on the scope of work Nlp : Value of scope of work in contract The analysis to identify the problems and challenges that exist in the road preservation control with the long segment system is still quite new. It is necessary so that the effectiveness and efficiency of the implementation of long segment-based national road preservation can be gauged and more accurate and precise handling program can be implemented. Reliable road performance indicators including road pavement, road shoulders, drainage, road equipment, road complementary buildings, and clearances, will affect PLRSL and affect road performance. Therefore, it is necessary to carry out an analysis to determine indicators for each road performance indicator that affects PLRSL.

Research Sites and Analysis Method
The study was conducted in the work region of the National Road Implementation Center of Central Java and Special Region of Yogyakarta which involves parties who handle national road projects. This study consists of road performance factors and indicators that had the potential to affect PLRSL. The factors and indicators were derived through a literature review, field observations, and interviews with project executors. Data analysis is a quantitative analysis of the factors of road performance and their indicators that affect the PLRSL. Using the SEM model [6; 7], the correlation between the factors and their indicators regarding the PLRSL was analyzed. The correlation between road pavement, road shoulders, drainage, road equipment, road complementing buildings, clearance, and PLRSL can be modeled graphically with more flexible assumptions with this SEM. Another advantage of the SEM method is that the model testing may be conducted as a whole, allowing the weight of each variable to be estimated based on its individual weight analysis [8].

Characteristics of Respondents
There were 149 respondents who completed the questionnaire, including 74 government's staff as service user and 75 contractors. Due to the mismatch between the education level and the educational background, two samples were omitted, leaving 147 samples for further analysis. The number of samples satisfied the SEM procedure's minimum criterion of 100 samples [9].

Data Collection Technique
According to Roh et al. [10], the quality of data collection is dependent on the precision of the data collection techniques. This study was a quantitative study that collected primary data in the field using a questionnaire. According to de Souza et al. [11], in order to generate valid and reliable data, the questionnaire had to be verified for validity and reliability before being distributed to respondents. In this study, questionnaire responses were tested for validity and reliability. The validity test employed the product moment correlation technique developed by Pearson with Equation 2. This analysis was conducted by combining the scores of each variable with the total score. The rcount results were then used to test the significance of the correlation results by comparing the t-count and t-table values, as shown in Equation 3. Significant correlations between variables and the total score indicated that these variables could aid in identifying the factors that influenced PLRSL in the region of National Road Implementation Center of Central Java and Special Region of Yogyakarta. The test employed a two-part test with the significance level of 5%.
where: n : number of research subjects X : chosen score A construct or variable was declared reliable if it contributed Cronbach's alpha (α) > 0.70. Otherwise, if Cronbach's alpha (α) < 0.70, then the construct is not reliable [12].

Data Analysis Method
SEM was used to determine the relationship between the factors in this study. According to Wang and Wang [13], SEM is a combination of two distinct statistical methods: (1) the measurement section that linked observed variables to latent variables through psychological and psychometric confirmatory factor models, and (2) the structural section that linked latent variables using simultaneously produced regression equations (simultaneous equation modeling) in statistics.   Table 1 summarizes the factors associated with the variables that influence the PLRSL. SEM method cannot be utilized to create a new model without the existing theory's base [13], as Table 1 includes the references that support the selection of variables and their indicators. Based on the causal relationship, the structural model equation can be stated as Equation 6.

Journal of Applied Engineering Science
Vol. 21 [20], who stated that it should be employed with a simple model and a modest sample size.
Research on SEM testing with the composite method or model simplification was a statistical approach for parceling or aggregating items (combining indicators) and using composite results as indications of latent structures [21]. Incorporating indicators into one or more "packages" or parcels and employing these packets as indicators of the latent construct was referred to as parceling or composite methods [22]. As depicted in Fig. 3, a package could be constructed by randomly or nonrandomly combining two or more indicators into one package (subset-item-parcel) or by merging all indicators into one composite package (all-item-parcel) [23]. Combining the all-item-parcel and the subset-item-parcel approach yielded this composite method approach.

Data Quality Testing
Testing the validity and reliability of data was used to assess the quality of the data. Since the t-count value of all indicators was greater than the t-table value (1.976), all indicators can be used to measure factors that influence financial PLRSL. As indicated in Table 2, the Cronbach's alpha value for each factor > 0.70, indicating that the questionnaire's variables, which were the elements that influenced the PLRSL, are very reliable.

Analysis of the Effect of Road Performance Indicators on PLRSL
The structural equation model outlines the causative relationship among constructs. If one variable or indicator in a construct change, the causality relationship, and other variables or indicators will also change [24]. The path diagram illustrating the causal relationship among constructs and their indicators is based on the conducted theoretical research. Fig. 4 shows the causal relationship among constructs. Fig. 4 describes a highly complex correlation and causality relationship among structural equation model with many constructs and indicators. Before conducting the SEM analysis using the composite or item parceling methods, the number of measurement indicators of each construct should be minimized. It is based on the notion that the more complicated the model, the larger the sample size necessary, and the more measurement items, the more complex the model. With 147 samples, several markers, and complex models, the sample size is still relatively tiny. Reducing the number of measurement indicators was required because it simplified the model, hence minimizing the impact of sample size on the results. The composite method or simplification of the structural equation model used in this study was the all-item parcel approach [9]. This model added a new construct, namely the "Road Performance Component (RPC)" which had indicators in the form of factors that had been composited. The RPC construct was analyzed for its effect on the PLRSL construct. Modification of the model using the item parceling approach can be seen in Fig. 5 for the relationship between RPC as exogenous construct and PLRSL as endogenous construct and the relationship between the manifest variables and the RPC construct.

Construct Unidimensionality Testing using Confirmatory Factor Analysis (CFA)
CFA aims to assess indicators that define a latent variable (construct) that cannot be directly measured. CFA is designed to assess the multidimensionality of a theoretical construct. The empirical analysis aims to validate the model that has been formed and estimate the parameters that are built based on theories [25].  (2) test of the significance of factor weights or convergent validity (loading factor or λ). The CFA measurement must use factor weights because the constructed test is used to confirm that the formed variable can together with other variables explain a latent variable. According to Ab Hamid et al. [26], the convergent validity criterion which is considered to have good validity is 0.70, while the convergent validity of 0.50-0.60 is still acceptable. This study used a cut-off point of 0.50. This means if the factor weight is lower than 0.50, the construct forming indicator is considered not to have the same dimensions as other indicators to explain a construct. The results of the CFA test on exogenous and endogenous constructs that have a significant effect on PLRSL is provided in Table 3.

Full Structural Equation Modeling Testing (Full SEM Model)
After completing CFA, estimation of Full SEM Model was tested by only including indicators that had been tested with CFA. The aim of evaluating the structural relationship model is to establish the relationship between the modeled constructs (latent variables). The conceptual framework of the proposed theoretical model has a structural relationship, namely the correlation relationship between exogenous factors and the causal relationship between the component factors of RPC and PLRSL. In Fig. 6, structural relationship modeling is depicted. The structural equation model depicted in Fig. 6 illustrates a highly complex correlation and causality relationship between numerous components and indicators. In this model, the composite method or simplification of the structural equation model will combine the all-item parcel approach method and the subset-item parcel approach method [9]. In order to perform a composite or parceling item, the weight of each indicator or manifest to be composited multiplied by the weight factor acquired while estimating the CFA is added together. The model adds a new construct, namely the RPC, which includes indicators in the form of composited components. The RPC was analyzed for its effect on PLRSL. As indicated in Fig. 7, this construct became an exogenous construct with indicators in the form of composite RPC. Fig. 7 shows the result of the chi-square test for the entire re-estimation  Table 4 demonstrates that the other parameter values in the model already fulfill the necessary criteria. The convergent validity value or factor weight about standardized regression weight is shown in Table 5. The RPC construct comprises 6 variables, while the PLRSL construct comprises 5 variables. According to the results of the analysis, the relationship between the constructs of RPC and PLRSL has a significant effect with a standard parameter of 0.770. It indicates that the construction management variables consisting of pavement factors, road shoulders, drainage, road equipment, complementary buildings, and clearances have a significant effect on PLRSL within the work area of National Road Implementation Center of Central Java and Special Region of Yogyakarta. The results of the full SEM model analysis regarding the effect of road performance indicators on PLRSL in the research sites indicate that road performance indicators have a significant effect on PLRSL of 77.0%. This means that failure in road performance indicators will affect the PLRSL by 77%. First, the road pavement contributes statistically significant to the road performance indicators at 88.0%. This is in line with Morecroft [27], who asserted that inadequate road maintenance and mismanagement causing delays in management will led to an accumulation of fallacies in the road network, necessitating larger costs to restore damaged roads in particular with budgeting restrictions [28][29][30][31]. Because it impacted the degree of road service, the road pavement is, therefore, the most critical component to address [32]. Second, drainage contributes statistically significantly at 81.4% to the road performance indicators. Inadequate drainage exacerbates pavement deterioration and raises annual maintenance costs [33]. Inadequate design, construction, and maintenance of drainage structures, as well as negative community behavior, are the primary causes of drainage issues. Furthermore, the road shoulder contributes a statistically significant at 80.9% to the road performance indicators. The qualities of road shoulder have a significant impact on its performance [34]. The width of the road shoulder must be adequate to accommodate stopped cars. Wide shoulders provide temporary parking space and boost road capacity by enhancing driving convenience. Adequate shoulders can protect the pavement structure's edges from deterioration and provide the pavement with lateral support. Meanwhile, the clearance provides a statistically significant contribution of 79.1% to the road performance indicators. Division 10 [3] stated that at the time of bidding, the contractor must be deemed to have carried out a thorough onsite inspection during the bidding period and to have known the actual conditions in the field by taking into account the traffic volume, the residual strength of the existing pavement, weather conditions, level of pavement damage, road shoulders, plants in the road space, drainage system, damage to other complementary buildings, signs condition, road markings, and other road equipment for the safety of road users [35; 36]. In addition, the statistically significant contribution of complementary buildings to the road performance indicators is 78.1%. According to Indonesian of Office of Highways [3], the maintenance work on the performance of existing road auxiliary buildings along the road included in the contract, regardless of the size or type of road auxiliary buildings, should include regular inspections of the structure's main components, preparation of detailed inspection reports, and routine cleaning of the place which is susceptible to damage if abandoned. The road equipment factor gives a statistically significant contribution of 51.3% to the road performance indicators. One of the notions associated with road infrastructure was "self-explaining road," which referred to roads that could deliver explanations to road users through road equipment. First introduced in the Netherlands, the self-explaining road idea encouraged drivers to adopt behaviors that were consistent with road design (Theeuwes, 1998 in [17]). Using several techniques, including the most cost-effective road markings and road signs, the concept sought to provide drivers with information about impending situations in a straightforward and simple manner. The results of the analysis of PLRSL in the research sites can be concluded that the following indicators are the determining factors for the success of PLRSL: (1) community reports/complaints, which are statistically significant at 91.6%; (2) response time, which is statistically significant at 90.6%; (3) the value of contract, which is statistically significant at 88.4%; (4) the length of the road that does not meet the road performance indicators/defect, which is statistically significant at 84.0%; and (5) statistically significant traffic volume at 80.2%.
Performance measurement and the selection of appropriate indicators are crucial for the successful establishment and implementation of a Performance-Based Maintenance Contract since performance indicators and objectives direct contractors in obtaining the desired maintenance results [37]. In addition, Tahan et al. [38] emphasized the importance of developing suitable metrics for evaluating performance-based maintenance contracts. There are instances where road administrations use incorrect performance metrics to evaluate the efficacy and efficiency of contractors. In addition, Performance-Based Maintenance Contract performance measurement has been proposed by Abu Samra et al. [39]: (1) Level of Service (LOS): the extent to which performance targets are met; (2) Timeliness of Response: response times for service requests or maintenance needs to be evaluated; (3) Safety Procedures: implementation of safety programs by contractors; (4) Quality of Services: customer perception with respect to asset conditions and contractor performance; and (5) Cost-Efficiency: cost savings (if any) obtained because of engaging contractors to perform Performance-Based Maintenance Contract services.

CONCLUSION
The results indicated that road performance indicators, such as road pavement, road shoulders, drainage, road equipment, complementary buildings, and clearance, have a significant effect on PLRSL, which affected the level of road service. These should be concerns with both contractors and government officials. If the contractor cannot fulfill the road performance indicators within the given response time, it will be liable to financial consequences that can result in cost overruns. Therefore, this study results are to enhance the level of road service through the achievement of road performance indicators to ensure the success of sustainable construction projects. This is essential for facilitating the management of sustainable road preservation projects and encouraging project managers to attain a high level of road service.