DEVELOPMENT OF A COST INDEX FOR IN-HOUSE ROADWAY MAINTENANCE WORK FOR THE NORTH CAROLINA DEPARTMENT OF TRANSPORTATION

Transcription

1 DEVELOPMENT OF A COST INDEX FOR IN-HOUSE ROADWAY MAINTENANCE WORK FOR THE NORTH CAROLINA DEPARTMENT OF TRANSPORTATION John Hildreth, PhD Associate Professor Department of Engineering Technology and Construction Management UNC Charlotte 9201 University City Blvd. Charlotte, NC Tel: Fax John.Hildreth@uncc.edu Scott Capps, PE, CEM State Maintenance and Equipment Engineer North Carolina Department of Transportation Fleet and Materials Management Unit 4809 Beryl Road Raleigh, NC Tel: Fax scapps@ncdot.gov Word count: 4,692 words text + 5 tables/figures x 250 = 5,942 words November 3, 2016

2 Hildreth and Capps ABSTRACT Maintenance activities represent a significant portion of roadway lifecycle costs and a cost index representing changes in maintenance costs over time is useful to State Highway Agencies (SHA) for developing multi-year work plans and budgets. The Highway Maintenance and Operations Cost Index previously maintained by FHWA was discontinued in Since that time, SHAs have met the need for a maintenance cost index with the use of available cost indices that are not specific to roadway maintenance. Cost indices were developed for the North Carolina Department of Transportation to represent in-house maintenance work in the Roadside, Maintenance, Traffic, and Bridge categories of the Standing Maintenance budget and combined into a composite index for Roadway Maintenance. This paper presents the rationale for selecting the fixed-base Laspeyres index form and the methods for selecting the basket of tasks and calculating the indices from work order task records. The resulting index values are presented and the observed annual cost increases are compared to the annual increases reflected by the previous Highway Maintenance and Operations Cost Index. Keywords: Cost index, maintenance, budget, planning

3 Hildreth and Capps INTRODUCTION Costs incurred throughout the life cycle of transportation infrastructure include construction, rehabilitation, and maintenance costs. Maintenance activities are those activities that do not extend the service life of a roadway, but rather keep the roadway in a usable condition. The costs of these activities are significant, as they total nearly half of the capital outlay (construction) costs. In 2012, roadway maintenance expenditures by all units of government were $48 billion and capital outlay expenditures were $105.2 billion nationally (1). In North Carolina in 2012, roadway maintenance expenditures were $932 million and capital outlay expenditures were $2,835 million (1). State Highway Agencies (SHA) routinely develop multi-year plans including the estimated timing and cost of the future construction work. Recently, the North Carolina Department of Transportation (NCDOT) has started developing similar plans for preservation, rehabilitation, and maintenance work over a 3 to 5 year period. A substantial portion of the work included in these plans consists of in-house maintenance tasks performed by NCDOT personnel and equipment. While the nature and timing of the work tasks can be planned based on currently available asset performance models, estimating the cost of future work requires knowledge of current maintenance costs and trends. Maintenance may also be performed under contract, which is solicited through a competitive bidding process. Bid price indices similar to those for construction costs can be used to track and trend costs of contract maintenance work. Maintenance performed in-house differs in the tasks performed and the performance cost excludes the contractor overhead and profit. Therefore, the bid price indices are not directly applicable. The goal of this research was to develop a set of maintenance cost indices (MCI) that can be updated annually to quantify trends of in-house maintenance costs and used to estimate future budgetary needs. BACKGROUND A cost index is a ratio of the cost at one time or location to another and is used to quantify the proportionate changes in cost (2) and to adjust costs from one location and/or time period to another (3). Indices may be developed to reflect the costs of labor, operations, equipment, commodities, maintenance, and other types of costs (4). A cost index may be developed for a single item or multiple goods in a basket in order to establish a single composite cost measure representative of related goods. Cost Index Forms The three common forms of cost indices are the Laspeyres, Paasche, and Fisher indices. Each has a separate computational form, but all are based on the unit cost and quantity for goods in the defined basket of goods. Sharma et al. (5) noted the importance of applying the correct index when developing cost updates. The Laspeyres index is a fixed base index. It is based on the cost of purchasing a fixed basket of goods, in terms of type and quantity, representing the base period and then the present cost of the same basket (6) and is calculated as: ILt = 100* n j=1 p jt q j0 n j=1 p j0 q j0 Where: ILt = Laspeyres cost index in year t pjt = unit cost of good j in year t

4 Hildreth and Capps pj0 = unit cost of good j in the base year 0 qj0 = quantity of good j in the base year 0 n = the number of goods included in the basket of goods The Laspeyres index is a commonly used cost of living measure and the Consumer Price Index (CPI) is essentially a Laspeyres price index (7). It is widely noted that the Laspeyres index is a biased estimate of the cost of living and over estimates cost increases because it does not consider a change in the quantity of goods over time due to relative price changes (2,6,7,8). However, the magnitude of this bias is reduced when goods are not substituted or when the relative change in cost for all goods is equal. A noted advantage of the Laspeyres index is that implementation requires quantity data be collected only for the base year and cost data collected on an ongoing basis (2). In many applications, data regarding the quantity of goods consumed is much more difficult to obtain in a timely manner. The Paasche index is a changing base index. It is based on the cost of purchasing a fixed basket of goods, in terms of type and quantity, representing the present period and then the base year cost of the same basket (6). It is calculated as: Ipt = 100* n j=1 p jt q jt n j=1 p j0 q jt Where: Ipt = cost index in year t pjt = price of good j in year t pj0 = price of good j in the base year 0 qjt = quantity of good j in year t n = the number of goods included in the basket of goods The Paasche index is also a biased estimate of changes in cost and underestimates cost increases. While it considers changes to quantity of goods due to price changes, it does not account for consumer preferences (6). The Fisher index is the geometric mean of the Laspeyres and Paasche indices and considers the effects of both substitution and preference. It is calculated as: IFt = I Lt *I Pt Where: II FFFF = cost index in year t II LLLL = Laspeyres index in year t II PPPP = Paasche index in year t The noted advantages of the Fisher index include the consideration of both substitution and preference effects, it is a homogeneous symmetric average of the Laspeyres and Paasche indices, and it satisfies the time reversal test (2,6). However, Bowley (9) noted that when the Laspeyres and Paasche indices are approximately equal there is no difficulty, which implies there is no significant advantage to using the Fisher index in such circumstances. Construction Cost Indices There are a number of construction cost indices that focus on various industry sectors and vary in computational form. The two most common are the Building Cost Index (BCI) and Construction Cost Index (CCI) published by the Engineering News Record (ENR). These indices reflect national trends in construction labor and material costs from all industry sectors. Remer et al. (4) noted that the indices maintained by ENR are the oldest indices used by engineers, with the CCI and BCI dating to 1908 and 1915, respectively. These indices are based on the assumption that

5 Hildreth and Capps past conditions remain constant in future trends and are typically used for short term budgeting (10). The Federal Highway Administration (FHWA) has maintained an index of national roadway construction costs since The initial index, officially titled Price Trends for Federal-Aid Highway Construction, was more commonly referred to as Bid Price Index (BPI), Contract Price Index, or Composite Bid Price Index. It was maintained until 2006 when it was replaced with the National Highway Construction Cost Index (NHCCI). The BPI was initially developed as a set of four Laspeyres indices: one each for excavation, structures, and surfacing, and those combined into a composite mile of roadway (11). Stern (12) noted that the Laspeyres (fixed base) index was subsequently found to be more stable and more reliable than a Paasche (changing base) index. A five year base period (1925 to 1929) was selected to account for variations across states in terms of both prices and quantities. Harrison (11) discusses in detail the selection of five bid items for inclusion in the indices: - Common excavation to represent excavation costs, - Portland cement concrete pavement to represent surfacing costs, and - Reinforcing steel, structural steel, and structural concrete to represent structures costs. The indices remained in essentially their initial form, but were updated as follows: 1. In 1961, bituminous concrete was included in the surfacing costs (12). 2. From 1977 forward, bid price data was collected only from the award of contracts over $500k for Federal aid highway projects (13). 3. The base period was revised to 1957 to 1959 (12), and then subsequently on an approximate 10 year interval to 1967 (14), to 1977 (13), and finally to Revision of the base period revises the base period quantities, which effectively adjusts the relative weights of each item. FHWA replaced the BPI in 2006 with the National Highway Construction Cost Index (NHCCI) (15). The NHCCI is intended to track roadway construction price changes and for conversion of current dollar expenditures to constant dollar expenditures (16). It is a Fisher index based on all successful bids, not only those valued at over $500k. Some SHAs, including NCDOT, maintain state specific construction cost indices. Shrestha et al. (17) surveyed SHAs and reported 21 state indices are maintained, with two additional states planning to develop indices. Walters and Yeh (18) synthesized information regarding SHA indices and found: 1. Most state indices are similar in form to the FHWA BPI and were developed from 1987 to 1990, and many anticipated changes to reflect the NHCCI 2. Most states found that weights are largely stable over time 3. Update periods vary from monthly to annually Highway Maintenance Cost Indices FHWA developed the Highway Maintenance and Operations Cost index in 1947 and published data annually until The basket of goods was developed to include 34 items, but was never revisited or revised. In 1990, Markow et al. (19) evaluated the index to assess whether it should be maintained, revised, or discontinued. They surveyed nine SHAs and found varying levels of reliance and use. The principal conclusion drawn from the study was that the basket of goods required revision in terms of content and number to alleviate the data reporting burden. It was proposed to reduce the number of goods to between 10 and 20, update the units in which data was reported, and to provide guidance regarding which goods were included in each category.

6 Hildreth and Capps Based on the results, FHWA chose to discontinue the index. Figure 1 provides index data from 1960 to 1990 (data prior to 1960 is unavailable). 250 FHWA Highway Maintenance and Operations Cost Index FIGURE 1 Highway Operations & Maintenance Cost Index (1977 base year). Despite the discontinuation of the index over 20 years ago, the need for a cost index in estimating maintenance costs has continued. SHAs have attempted to meet this need through the use of other available indices. The CPI published by the US Bureau of Labor Statistics has been applied to highway maintenance costs (20,21) despite representing all goods and services purchased. Highway construction cost indices have also been applied (22,23). However, Ozbek noted that such an index is less than ideal because it is not specific to highway maintenance. It is clear from the literature that a cost index specific to highway maintenance is useful in developing plans and budgets. The need for such an index is also clear from the use of other less relevant indices by transportation agencies. DEFINING THE MAINTENANCE COST INDEX The MCI is defined in terms of its computational form, as previously described, and content, roadway maintenance work in this instance. The Laspeyres index form was selected for the MCI because: 1. As a fixed base index, the data collection and management requirements are most manageable. Quantity data is needed only for the base year and only cost data is required on an on-going basis. 2. It is the most common form of cost index and was the form of the BPI and Highway Maintenance and Operations Cost index formerly maintained by FHWA. 3. The intended use of the index is to aid in estimating maintenance costs 3 to 5 years in the future, for which fixed base indices have generally been applied (24).

7 Hildreth and Capps The selection of type of roadway maintenance to be performed is not heavily influenced by cost, and therefore there is little need for the index to account for substitution of tasks. Maintenance tasks in the Roadside, Traffic, Maintenance, and Bridge categories of the Standing Maintenance budget were selected to be represented by the MCI and 2014 was selected as the base year. Pavement maintenance tasks were not included because they are almost entirely performed under contract rather than by in-house forces. Individual work order records of maintenance performed in-house from 2012 to 2015 were provided and included the maintenance task, dates of performance, quantity, and costs. For each maintenance task, the annual total cost was calculated and the number of records were summarized based on the data contained. It was found that: - the annual total costs were relatively consistent across the years for each maintenance category and maintenance task - the annual number of records for individual tasks varied from less than 10 to over 10,000 - a large majority of records included both the quantity and cost of maintenance work performed, as shown in Table 1. TABLE 1 Summary of Maintenance Task Record Data Records with Quantity Data Records without Quantity Data Total With Cost Data 85% 4% 89% Without Cost Data 10% 1% 11% Total 95% 5% 100% DEVELOPMENT OF THE INDICES Development of the indices involved selecting the basket of goods (representative set of tasks) for each maintenance category and determining the base year parameters for the selected tasks. In selecting the basket, the objective was to identify a small number of tasks to represent each category. The selected tasks should be related to maintenance and represent a significant portion of costs in each category. When considered collectively, the selected tasks should also represent a significant portion of all maintenance costs. To support unit cost and index calculations, it is necessary that the quantity of work be measured. Tasks infrequently performed and for which little data is available should not be included in the basket of goods. While the indices could be developed from all tasks, this would be inefficient and create a more burdensome data and index management process. The base year parameters required to develop the Laspeyres index were the unit cost and total quantity for each task in the baskets. Selection of Basket Goods A basket of goods was selected for each maintenance category based on total recorded cost in the base year In order of total cost, starting with the task with greatest total cost, tasks were added to the basket until the total recorded cost for each basket comprised approximately 75 percent of the total recorded cost in each category. Tasks in each basket were then reviewed to ensure that:

8 Hildreth and Capps The tasks were related to roadway maintenance some tasks, such as 3240 Electricity for Traffic Control Devices are not related to maintenance 2. The task was not funded off of the top funds for some tasks are set aside at the start of the year, such as task 3100 Snow and Ice 3. The task was measured in an appropriate quantity unit some tasks, such as 3110 Beaver Control are measured as lump sum (units of dollar) 4. A sufficient volume of task records were available tasks with less than 50 work records were neglected Based on these criteria, a total of 20 tasks were selected: 3 each for Roadside and Traffic, 5 for Maintenance, and 9 for Bridge, as shown in Table 2. These tasks accounted for a significant portion of the total costs recorded for each category. In 2014, the selected tasks collectively accounted for over $82.7 million in maintenance cost, which represents 76 percent of all recorded costs in the four categories. TABLE 2 Basket Tasks in each Maintenance Category Category Roadside Maintenance Traffic Bridge Maintenance Work Task 2912-Brush and Tree Control / Mechanical/ Other 2900-Grass Mowing 2914-Vegetation Management at Stationary Objects 3112-Shoulder Maintenance / Reconstruction 3108-Drainage Ditch Maintenance 3126-Install Pipes (<=48") 3128-Maint/Repair Pipes (<=48") 3102-Removal of Hazards/Debris From ROW 3250-Install/Replace Ground Mounted Signs 3222-Long Line Painted Pavement Markings 3252-Repair Ground Mounted Signs 3300-Install/Replace NON NBIS > 48" up to NBIS Structures 3302-Maintain/Repair NON NBIS > 48" up to NBIS Structures 3352-Maint Slope Protection 3314-Maintain Steel Superstructure Components 3344-Repair/Replace Timber Substructure Components 3326-Maintain Concrete Deck Maint/Repair/Replacement of Std Bridge Exp Joints 3366-Drift and Debris Removal 3348-Maintain Concrete Substructure Components Portion of Category Cost Captured by the Tasks Base Year Parameters The parameters required for the 2014 base year were the unit cost and quantity for each task in the baskets, and were calculated from the work order records. In reviewing the work order records, it was noted that extraordinarily large quantities were occasionally recorded. Some of 80% 78% 77% 71%

9 Hildreth and Capps the records with extraordinarily large quantities appear to be an aggregate of multiple individual work activities based an associated performance period spanning several months. However, not all records had long performance periods and extraordinarily large quantities were not always associated with extraordinarily high costs. It was determined that at least a portion of these records may be the result of data entry errors. To address this issue, only records where quantity was less than the 98 th percentile of all quantities recorded for the task were used, effectively neglecting the 2 percent of records for each task with the greatest recorded quantity. The 2 percent of records with the least recorded quantity were not neglected because these records could not be clearly determined to be made in error. The unit cost was calculated for each work record that contained both cost and quantity data. The distribution of unit costs was investigated to determine an appropriate method for calculating a unit cost that represents each task. Median unit cost was more representative and selected for the index calculations because the distributions of task unit costs were heavily right skewed and not normally distributed. The distribution of unit costs for task 2912 Brush and Tree Control/Mechanical/Other was typical and is shown in Figure No. of Task Records $513 Median Unit Cost $2,514 Mean Unit Cost 0 $100 $200 $300 $400 $500 $600 $700 $800 $900 $1,000 $1,100 $1,200 $1,300 $1,400 $1,500 $1,600 $1,700 $1,800 $1,900 $2,000 $2,100 $2,200 $2,300 $2,400 $2,500 $2,600 $2,700 $2,800 $2,900 $3,000 $3,100 $3,200 $3,300 $3,400 $3, Unit Cost ($/SHM) FIGURE 2 Unit Cost Distribution for Task 2912 Brush and Tree Control/Mechanical/Other The total base year quantity for each task was found by summing the recorded quantity for each work order and adding an estimated quantity for records containing only cost data. The total cost for records without quantity data was summed and divided by the median unit cost to estimate the quantity. Estimating quantity based on median unit cost introduces some error into the total quantity parameter due to the variability in the unit cost values. However, the estimated portion of total quantity was typically less than 10 percent, with the exception of bridge tasks. The error resulting from a small estimated quantity is considered negligible.

10 Hildreth and Capps CALCULATED INDEX VALUES The cost index for each maintenance category and Roadway Maintenance composite index were calculated for each year in the period 2012 to 2015 based on the unit cost and total quantity for basket tasks calculated in the manner previously described. The results are provided in Figure Roadside Maintenance Traffic Bridge Roadway Maintenance FIGURE 3 Maintenance Cost Index Values from 2012 to 2015 The Roadside index value in 2012 was heavily skewed upward by a large unit cost for one of the three basket tasks in that category. The median unit cost in 2012 for task 2914 Vegetation Management at Stationary Objects was over three times greater than in the base year The total quantity for the task in each year was similar and there was no clear explanation for the large unit cost, such as a fundamental change in what is charged to the task. By extension, this also skewed the 2012 value for the Roadway Maintenance composite index. Neglecting these two heavily skewed data points for 2012, Figure 3 shows a general increase in maintenance costs over the period. From 2013 to 2015, the Roadway Maintenance composite index reflects 9.4 percent annual increase. Bridge maintenance had the largest cost increase over the same period at 10.9 percent annually, while Traffic maintenance had the smallest increase at 4.1 percent annually. The Roadside and Maintenance categories both had annual cost increases of approximately 9 percent.

11 Hildreth and Capps The rates of cost increase observed for NCDOT were similar to the rates reflected in the Highway Maintenance and Operations Cost Index previously maintained by FHWA. In the 20 year period prior to being discontinued in 1990, the Highway Maintenance and Operations Cost Index reflected a 7.1 percent annual increase in maintenance costs. In the period of greatest cost increase, from 1978 to 1982, maintenance costs increased 10.4 percent annually. CONCLUSIONS In-house maintenance activities represent a significant portion of roadway lifecycle costs and a cost index representing changes in maintenance costs over time is useful to SHAs for developing multi-year work plans and budgets. The available types of cost indices were evaluated and the Laspeyres index was selected because the type of roadway maintenance to be performed is not heavily influenced by cost and it provides the least burdensome data collection and management requirements. Cost indices specific to NCDOT were developed for in-house maintenance work in the Roadside, Maintenance, Traffic, and Bridge categories of the Standing Maintenance budget and a composite index for Roadway Maintenance. The maintenance work order records maintained by SHAs are sufficient for developing and annually updating the cost indices. The goal in selecting the basket tasks was to include the small number of tasks that comprised a large portion of costs for each maintenance category. The 20 basket tasks selected accounted for approximately 75 percent of all recorded in-house maintenance costs, and between 71 and 80 percent of costs in each category. It is important to examine costs over a multiple years to ensure consistency in cost from year to year. In defining the cost index, the unit cost and total quantity parameters should be accurately reflected for each basket task. The median unit cost was used because the unit costs were heavily right skewed and not normally distributed. Where work order records failed to capture the quantity of work performed, quantity was estimated based on the recorded total cost and median unit cost. The methodology presented is expected to be attractive to SHAs endeavoring to develop cost indices because the resulting indices overcome the noted limitations of the previous FHWA Highway Maintenance and Operations Cost index. The noted limitations included too many basket tasks, outdated units of measure, and a lack of guidance regarding which goods or tasks were included. While the resulting number of basket tasks depends on the data structure and spending habits of the SHA, the 20 tasks selected for NCDOT is at the high end of the range recommended by Markow et al. (19) and the objective of capturing 75 percent of the total cost could be adjusted slightly to manage the number of selected tasks. The use of maintenance cost data specific to the SHA ensures familiarity with the included tasks and that current units of measure are used. Additionally, the resulting indices are directly applicable to SHA maintenance cost forecasts because they are SHA specific rather than a national average. ACKNOWLEDGEMENT This research is sponsored by the North Carolina Department of Transportation (NCDOT) and the Federal Highway Administration (FHWA). The authors express gratitude to the NCDOT Maintenance Operations and Fleet Management Unit, NCDOT Research and Development Unit, and project steering committee for their support in this work. DISCLAIMER

12 Hildreth and Capps The contents of this report reflect the views of the authors and not necessarily the views of the University. The authors are responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official views or policies of either the North Carolina Department of Transportation or the Federal Highway Administration. This report does not constitute a standard, specification, or regulation. REFERENCES 1. FHWA [Federal Highway Administration]. (2013). Highway statistics (annual). Available at: Accessed May Hill, P. (2004). Consumer price index manual: Theory and Practice. edited by TP Hill: International Labor Organization. 3. Qasim, S. R., Siang W. (Daniel) Lim, Motley, E. M., & Heung, K. G. (1992). Estimating Costs for Treatment Plant Construction. Journal (American Water Works Association), 84(8), Remer, D. S., Lin, S., Yu, N., and Hsin, K. (2008). An update on cost and scale-up factors, international inflation indexes and location factors. International Journal of Production Economics, 114(1), Sharma, J. R., Najafi, M., and Qasim, S. R. (2013). Preliminary Cost Estimation Models for Construction, Operation, and Maintenance of Water Treatment Plants. Journal of Infrastructure Systems, 19(4), Diewert, W. E. (1998). Index number issues in the consumer price index. The Journal of Economic Perspectives, Braithwait, S. D. (1980). The substitution bias of the Laspeyres price index: An analysis using estimated cost-of-living indexes. The American Economic Review, Fisher, F. M., and Shell, K. (1972). The Economic Theory of Price Indices; two essays on the effects of taste, quality, and technological change. Academic Press, New York. 9. Bowley, A. L. (1901). Elements of Statistics. Westminster, Orchard House. 10. Wilmot, C. G., and Cheng, G. (2003). Estimating future highway construction costs. Journal of Construction Engineering and Management, 129(3), Harrison, J. L. (1933). An Index of the Cost of Highway Construction. Public Roads, 14(5), Stern, E. L. (1961). A New Base for The Highway Construction Bid Price Index Compiled by the Bureau of Public Roads. Public Roads, 31(10), Mirack, F. (1981) Becomes the New Base Year for FHWA's Bid Price Index. Public Roads, 45(1), Stern, E. L. (1970). The Third Base for the Federal Highway Administration's Contract Price Index. Public Roads, 36(4), White, K., and Erickson, R. (2011). New Cost Estimating Tool. Public Roads, 75(1). 16. FHWA: National highway construction cost index. (2015). Federal Highway Administration, Washington. Accessed May Shrestha, K. J., Jeong, H. D., and Gransberg, D. D. (2016). Current Practices of Highway Construction Cost Index Calculation and Utilization. Construction Research Congress 2016, Walters, J., and Yeh, D. (2012). Transportation Literature Search & Synthesis Report: Research and State DOT Practice on Construction Cost Indices. Wisconsin Department of Transportation (WisDOT).

13 Hildreth and Capps Markow, M. J., E. L. Seguin, E. F. Ireland, and D. M. Freund (1990). Feasibility Study of Changes to the Highway Maintenance and Operations Cost Index. Transportation Research Record, (1268), Litman T. (2011). Transportation Cost and Benefit Analysis: Techniques, Estimates, and Implications. Victoria Transportation Policy Institute, 2 nd Ed. 21. Adams, T. (2011). Estimating Cost per Lane Mile for Routine Highway Operations and Maintenance. Midwest Regional University Transportation Center Project Final Report. 22. CalTans (2011) Ten-year State Highway Operation and Protection Program Plan. California Department of Transportation; Business, Transportation, and Housing Agency. 23. Ozbek, M. (2007). Development of a Comprehensive Framework for the Efficiency Measurement of Road Maintenance Strategies using Data Envelope Analysis. a dissertation submitted to Virginia Polytechnic Institute and State University (Virginia Tech). 24. Wilmot, C. G., and Mei, B. (2005). Neural network modeling of highway construction costs. Journal of Construction Engineering and Management, 131(7),