The Swedish National Road Authority (from now on ... - DiVA portal

Promoting Innovation in Transportation Infrastructure Maintenance Incentives, Contracting and Performance-Based Specifications

Innovation promotion

Contracting

Performance-Based Specifications

TORBJÖRN STENBECK Doctoral Thesis in Civil and Architectural Engineering 2007 Division of Structural Design and Bridges

TRITA-BKN. Bulletin 91,2007 ISSN 1103-4270 ISRN KTH/BKN/B--91--SE CDU:S5

Submitted with due permission of the KTH Royal Institute of Technology for public examination for the degree of Doctor of Technology in Civil and Architectural Engineering, Division of Structural Design and Bridges Friday 23 March 2007, KTH, Lindstedtsv 26, room F3 Faculty opponent: Supervisor: Co-supervisor

Professor Jan Bröchner Professor Johan Silfwerbrand Professor Folke Snickars

Promoting Innovation in Transportation Infrastructure Maintenance

1

Contents Acknowledgements ................................................................................................ 2 Abstract .................................................................................................................. 3 Content of the Thesis ............................................................................................. 4 1

Introduction..................................................................................................... 5

2

Methodology................................................................................................. 13

3

Results........................................................................................................... 17

4

Analysis ........................................................................................................ 25

5

Conclusion and Future Research .................................................................. 29

6

References..................................................................................................... 31

Swedish Summary - Sammanfattning

Paper 1

Effects of Outsourcing and Performance-Based Contracting on Innovations.

Paper 2

Life-Cycle Cost Bridge Budget Model

Paper 3

Budgeting Performance-Based Winter Maintenance.

Paper 4

A Highway Design-Build-Maintain-Warranty Case Study.

Paper 5

Quality and Life-Cycle Costs in Performance-Based Contracts – Multiple Case Study.

Paper 6

Incentives, Results and Possible Success Factors for Rail Maintenance Performance-Based Contracting – Case Study.

Paper 7

Quantifying Effects of Incentives in a Rail Maintenance PerformanceBased Contract.

Paper 8

Highway Maintenance Costs and Contracting Policy.

2

T. Stenbeck

Acknowledgements Thanks to funding by SBUF, Vägverket and Banverket, CDU and KTH this study on incentives to innovation started on 1 March 2003. SBUF, the Development Fund of the Swedish Construction Industry, aims at promoting businesslike conditions for contractors enabling them to make use of research and conduct development work (www.sbuf.se, accessed 1 October 2006). Vägverket/SRA, the Swedish Road Administration, is the owner of the public national highway network. Banverket, the rail administration, is the owner of the public national railway network. CDU, the Centre for research and education in operation and maintenance coordinated and KTH, the Royal Institute of Technology supervised the project, within its Urban Planning Unit 2003– 2005 and within CBI, the Swedish Cement and Concrete Research Institute 2006–2007. Supervising Professors Johan Silfwerbrand and Folke Snickars have bravely guided me through the stormy waters with PhD Fredrick Lekarp as co-supervisor on the ship until January 2005. In the reference group, Birgitta Törne, Per Kvick and Kjell Eriksson have represented Banverket, Hardy Wikström, Hans Kvarnlöf, Sven-Erik Hallberg and Fredrick Lekarp have represented the Swedish Road Administration, Håkan Carlevi, Jan-Olof Nordlander and Bertil Grandinson have represented the private contractors and Hans Cedermark and Håkan Westerlund have represented CDU. With the reference group members as door openers, more than 100 interviews were made possible in a positively critical, contributory, constructive, and generous athmosphere. In 2005–2006, the Valle Foundation provided a unique and valuable opportunity to pursue studies at University of Washington, Seattle, USA. Chair and Professor John Schaufelberger in the Construction Management Department attentively and humbly coached the wild Swede and supervised the Papers 4 and 5 related to design-build and performance-based contracts that later became a main part of the thesis. Mark Hallenbeck, Director of the Washington State Transportation Center, provided expertise on transportation planning and ITS (Intelligent Transportation Systems) and Professor Yinhai Wang did a great job trying to avoid the statistically oriented papers to upset statistical science more than necessary. PhD students Johan Nyström, Hans-Åke Mattsson and Richard Österberg at KTH and Assisting Professor Ahmed Abdel Aziz, Professor Saeed Daniali at University of Washington have generously shared their experience and contacts. Master’s Student Tony Woody, and PhD Student Jianyang Zheng assisted with advanced calculation work and assignments in a pleasant athmosphere of high quality mixed with humour. Many thanks also to regional and national highway maintenance directors and their staff in Sweden, France, USA and Canada for their contribution of data and experience. Special thanks to Professor Rolf Magnusson, Högskolan Dalarna, who critically reviewed a prior version of the thesis as opponent in the final seminar in December 2006. The interest and support to my research have been overwhelming and a feeling is that the expectations will not be possible to fully live up to. Hopefully the small steps taken will still contribute and otherwise stimulate researchers and practitioners to further studies and experimentation in the field. Stockholm in February 2007 Torbjörn Stenbeck


3

Abstract Knowledge of what action that is needed to drive innovation at a desired speed is in demand in civil engineering and its related maintenance. 1. What measures to stimulate innovation have been tested? 2. How much innovation has been achieved by contracting? 3. How much innovation was achieved by performance-based specifications? 4. How can cost models contribute to innovation? Methods include qualitative and quantitative methods that have been timed and mixed to optimize their merits. Sweden, France, USA and Canada have used as research ground. Technology transfer, multi-criteria evaluation, variant bidding, idea mailbox, weatherregulated payment, contests and earmarked funds for innovative projects were some of the method beside and within contracting and performance-based specifications that have been tested. Contracting as such has cut costs in Sweden but not in North America. Neither Sweden nor North America has noticed any increase of innovation, rather the contrary. The savings have primarily been achieved by cuts on staff and by using standardized, less expensive and less advanced machinery. Contracted highway maintenance provinces in Canada and Sweden on average had about 50 % higher costs than inhouse provinces and Washington State. The difference is reduced to 26 %, when corrected by weather and the higher traffic in the contracted provinces. Prestige, politics and competitivity made it difficult to extract economic data from private contractors, and even from the public owners and may explain the contradictory results in previous studies. The internally driven innovation appears small and incentives to innovation weak in inhouse systems, but contrary to expectation even less in contracted systems. Performance-based specifications (PBS), such as Design-Build (DB), have reduced delivery times and kept the budget better than traditional contracts, but quality, lifecycle cost and technical progress was rarely analyzed and even less confirmed in the literature, why a multiple case study was carried out. The result was that three out of four PBS cases delivered lower quality in the long run or showed higher costs already on the opening day, when compared to a traditional contract alternative. Cost models contribute to innovation by making regions with different conditions comparable and provide tools for rational planning and decision making. One model for how highway maintenance costs depend on snow, bridges and traffic and one model for how bridge maintenance costs depend on size and age were elaborated. Models included in contracts, e.g. to allow a contractor to reduce the weather risk, appear to have contributed to a more successful contracting rollout in Sweden than in Canada. France provides experience of how inhouse innovation contests and industry-own patent-like routines can promote innovation. After the first two years with an incentive contract, Banverket received 10 % better quality measured as train delay and 20 % better quality measured as the number of technical errors at no cost. A lesson learnt is that the success of performance-based specifications depends on how well the owner can describe and define the contracts, how compliance is measured and how deviations are handled, i.e. how the contractor is penalized for non-fulfillment or awarded for excess delivery.

4

T. Stenbeck

Content of the Thesis The following papers are formal parts of the study according to the numbering below. Chapter 6 contains a list of other references. Each paper also contains a reference list. 1. Effects of Outsourcing and Performance-Based Contracting on Innovations. Published in Transportation Research Record September 2006, No 1348, pp. 3–8. 2. Life-Cycle Cost Bridge Budget Model. Published on the TRB Annual Meeting 2007 CD-ROM, Paper 07-0110, Transportation Research Board, Washington D.C. 3. Budgeting Performance-Based Winter Maintenance. Submitted to ASCE Journal of Infrastructure Systems 27 February 2007. 4. A Highway Design-Build-Maintain-Warranty Case Study. Submitted to Journal of Performance of Constructed Facilities 27 February 2007. 5. Quality and Life-Cycle Costs in Performance-Based Contracts – Multiple Case Study. Submitted to Journal of Performance of Constructed Facilities 27 February 2007. 6. Incentives, Results and Possible Success Factors for Rail Maintenance Performance-Based Contracting – Case Study. Submitted to Construction Management and Economics 19 January 2007. 7. Quantifying Effects of Incentives in a Rail Maintenance Performance-Based Contract. Submitted to Leadership and Management in Engineering 27 February 2007. 8. Highway Maintenance Costs and Contracting Policy. Submitted to Evaluation and Program Planning 14 February 2006.


5

1 Introduction New, redefined and clarified concepts and notions may be necessary or purposeful as the research front advances. Clear and common understanding of new and old vocabulary can be a pre-requisite for innovation and development, and the act of defining the terms is a type of innovation itself. The introduction chapter will therefore begin with a term definition section, followed by background and purpose of the study.

1.1 Definition of Terms and Practices The following definitions clarify how some notions are used and citation practices in this study. Some words are new, some are deliberately redefined or clarified. There is sometimes a need to differentiate various forms of a generic word. Other times there is the opposite need, that there is no generic umbrella for differentiated notions. In some cases synonyms have been considered as available words useable after redefinition to create differentiation. AADT or ADT (Annual) Average Daily Traffic is a common measure of traffic intensity on a road, useful for planning routine maintenance. Case Study Findings derived from a study of a single case (Yin 2003). Citations If the citation appears within a sentence, only the immediately preceeding statement is referred to. If at the end of the sentence, the whole sentence is referred to. If at the end of a paragraph, preceeded by a period and not followed by a period, the whole paragraph is referred to. Absence of a source normally indicates that the previous statement is an assumption by the author, but sources are also omitted to facilitate the reading, in particular when the statement is uncontroversial or found in general literature. A source reference can also be omitted as a deliberate anonymization to protect the source, e.g. which individual at the agency that contributed with the information. Contracting, Contractor The owner of a road or a railroad asset can design, build and maintain it inhouse or buy the services from a contractor. NVF 53 Road Transport Informatics Terminology Dictionary (2002) defines contract as an agreement between two or more parties, either explicit or implicit. In this study, a contractor can be from the same legal entity, i.e. contracts can be between a public owner and a public contractor, as opposed to outsourcing below. As an example, 60 % of Swedish highway maintenance is contracted to a public contractor and the remainder is divided among four nationwide private contractors. On North American railroads, the large private owners of the rail infrastructure have inhouse maintenance staff while the smaller rely more on private contractors. Combi-Contracts (new) Combi-Contracts is proposed as a generic for vertically integrated services combined into one contract (Stenbeck 2006), such as Design-Build, Design-Build-OperateMaintain and Build-Operate-Transfer, instead of today’s commonly used but ambiguous project delivery dystems and procurement systems.

6

T. Stenbeck

Design-Bid-Build (DBB) In the traditional project delivery system, the owner first hires a designer who develops the description of the object, or service, to be contracted in detail. Contractors are invited to bid with price and qualification and the winner is awarded the contract as described. Any later deviation from the description is called a change order and normally not included in the bid price. (Collier 2001) Design-Build (DB) In Design-Build (in UK: design and build) design and build are combined into one contract. The architect or engineer is employed by, or becomes a consultant to, the contractor. Normally a limited design, a phase during which the designer is consultant to the owner, precedes the Design-Build contract. (Collier 2001) Disincentive Negative consequence, see Incentive below. Functional contracts, functional criteria Functional contracts refer to contracts specified by criteria from a user perspective, Fig. 1. Grennberg (1965, 1998) and Olsson (1993) exemplify by expressing a road as a transportation corridor defined with height, length, width and flow characteristics. The recipe of the pavement should be left to the contractor to design. Expert

Process-

Output-

Outcome-

based specifications

Function-

User

Fig. 1 Bird's eye on contract types. Performance can be described at any level. Incentive Incitamenta, incitantia (latin) means stimulus to do something (Meyers 1897). The incentive to perform or develop the own task can be positive in the form of rewards, or negative in the form of penalties. Incentive can be used both as the opposite of disincentive but also as a generic for both incentives and disincentives. A synonym for the generic meaning used in recent literature is I/D – provisions. Inhouse (simplified spelling) Building or maintaining the infrastructure with own staff is referred to as inhouse. The hyphenated form in-house has been allowed replacement without hyphenation within this study to facilitate the reading. Hyphenation is normally a transitional form as twoword-notions develop into one combined word (www.grammarmudge.cityslide.com/ articles/article/426348/2805.htm, 9 February, 2007). Invention, Inventor (redefined) Inventor is the person that first brought the new, smart, technical solution for which a patent is filed. American patent jurisdiction protects this person specifically. Inventions become innovations if adopted and commercially successful. Else they remain inventions. Witin this study, invention is allowed to represent also non-physical matters (organizational, commercial etc.), just as Schumpeter (1961) allowed innovation to include organizational and commercial matters.

7


Innovation Schumpeter (1961) defined an innovation as a countable new, physical product or process (such as organization or business method). The international patent authorities require products and processes to have a commercial value to award patent protection. Another meaning or use of the word innovation is as an uncountable, buzz word for positive attitudes to change in general. Innovation is used synonymously to technical development in this study. Salter (1966) and Jones (1975) define technical development as more value for less money. Jones identifies new products as innovations (technical development) as well. In our study, cost-savings achieved only in monetary terms by moving the production to a region with lower salaries, or by making the existing work force work more or harder are not innovations, Table 1. Table 1 The innovation matrix. I = Innovation rate. Figures in bold represent combinations fulfilling the raised demand to qualify for innovation. Service 1% poorer

Service as before

Service 1% better

Price 1% higher

I = –2 %

I = –1 %

I = ±0 %

Price as before

I = –1 %

I = ±0 %

I = +1 %

Price 1% lower

I = ±0 %

I = +1 %

I = +2 %

If services are reduced by 1 %, but costs are reduced by 2 %, it could be argued that productivity has increased by 1%. In the definitions used in this thesis however, reduction of service, no matter the saving, is not regarded as technical development. If quality consists of several parameters (the most common case), none of these must drop (the Pareto principle applied to quality). The term QQ/C below expresses the same philosophy in a different manner. In the interviews, the respondents probably had a less demanding definition, which was allowed to rule within the interview without commenting the issue. Since no or very few innovations could be mentioned, despite the less demanding definition, consensus on a stricter definition would not have changed the overall outcome of this study; on the contrary. Maintenance Maintenance can refer to activity without capital-increasing effect, such as snow and ice removal, malfunctioning signals to be repaired and vegetation to be cut (Papers 1, 3, 6 – 8). Long-lasting repairs like grinding the rail every 5 years, resurfacing the roads every 15 years, or replacing bridge decks every 30 years is also included in the maintenance analyzed in Papers 2, 4 and 5. Outcome, Output (new differentiation) Recent contract vocabulary includes outcome- and output-based contracts. The difference between the two, if any, is debated (Adams and Smith 2006). Proposed is to let outcome be the result in economic terms and output the result in technical terms. Both output- and outcome-based contracts can be regarded as intermediate forms between user-defined functional contracts and process-based contracts, detailed technical specifications defined by construction experts. Fig. 1, page 6, clarifies their relative positions.

8

T. Stenbeck

Outsourcing (new clarification) Outsourcing is proposed to be synonymous to contracting, except that the parties of the contract must be independent and not within the same legal entity. Performance-Based Specifications (PBS), Performance-Based Contracting (PBC) The definition of performance is confused in the literature and even more vague in reality. NVF 53 Road Transport Informatics Terminology Dictionary 2002 defines performance measure as a quantitative or qualitative characteristic describing the level of service provided by a transport facility or service. The wide definition hints at a user perspective but does not exclude describing performance also with technical terms. Process-Based Contracts (new) Being the opposite of a functional contract, the process-based contract represent the most common, traditional construction contract with a design defined in detail in blueprint and process. The contractor’s responsibility is limited to workmanship and honesty regarding used volumes such as time and materials. Often proof of used volumes such as bills of lading are requested. Fig. 1, page 6, clarifies its relative position to other contract types. Promotion Promotion equals stimulation and includes any measure to spur the promoted activity. QQ/C =

Quality × Quantity (new) Cost Real

Another metric to determine and define innovation. To qualify as an innovation, two conditions must be fulfilled. QQ/C after the change must exceed QQ/C before the change and the first Q, quality, must never decrease. (∆(QltyQnty/C) > 0 AND ∆Qlty ≥ 0). Cost is to be interpreted in real terms, the consumption of the original resources (labor, materials). If nominal cost differs because of interest, inflation or exchange rate, it must be corrected with interest, inflation or exchange rate before the Cbefore and the Cafter are compared. Only if real volumes are saved, expressed as labor-hours, material-volumes etc, innovation and technical development have taken place. Moving production to low income countries does not qualify unless labor and materials are saved in both countries. Moving production to a low income country where less effective machinery and methods are used, will not save labor and materials on a global scale. Qualitative Study (clarified, redefined) Qualitative studies search the factors, not the values, explaining a phenomenon, often an explorative step of study to finding variables with influence on a dependent variable. Quantitative Study/Methods (clarified, redefined) Quantitative studies search the values to attribute to the factors found by the qualitative study. Quantitative methods use regression analysis and other numeric tools usually requiring several observations of the same properties (qualities). RWIS The abbreviation RWIS stands for Road Weather Information System. SRA The abbreviation SRA stands for the Swedish Road Administration (Vägverket).

9


Transportation Infrastructure In this study, transportation infrastructure is limited to highways and railroads. Triangulation, Method Triangulation Triangulation is the geodetic technique of determining an object’s position in space by taking bearings (directions) to at least three known points, Fig. 2. The position of the object is assumed to be in the point of gravity of the error triangle or polygon. The distances between the point of gravity and the edges of the polygon provide estimates of the accuracy, i.e. the potential error, of the position. Analoguously, Method Triangulation refers to studying phenomena with several independent methods to reduce the error and get estimates of accuracy.

Ref point 1

Ref point 2 Ref point 3

Fig. 2 Positioning by triangulation.

Vägverket Vägverket is the Swedish name for the Swedish Road Administration, SRA.

1.2 Background Productivity and innovation in construction and maintenance are low in many countries (Seaden 2001), also in Sweden (Swedish Public Studies 2002). The belief that innovation is the motor for growth and sustained welfare is gaining increasing support, as existing production gradually moves to low-income countries. Few, however, actually have a recipe, based on research findings of what worked and what failed in the past. Contracting has gradually been introduced in Sweden during the 1990s for road and railroad routine maintenance. Its objective was to spur efficiency and technical development by opening up a previous monopoly market to the private sector. There is however a feeling of disappointment. Many, in particular owners, express that innovations were more frequent before the contracting reform (Österberg 2003). Vägverket, or SRA, the Swedish Road Administration, started the reform process in 1992. Since 1996, the maintenance of all roads is contracted out after a competitive bidding process every 3–7 years. The process has gradually developed towards performance-specified contracts. The philosophy with these approaches is to enable alternative methods in order to spur innovations and achieve the most value of taxpayers’ funds. Banverket, the Swedish railway administration, was separated from SJ, the state-owned company operating the trains, in 1988. An independent production unit, Banverket Produktion, was separated out of Banverket, but kept monopoly to undertake all maintenance until 2001. The owner-contractor model was introduced in1998, but contracting to private competitors did not happen until 2001. (Paper 1) Vägverket hopes this study will contribute to higher quality of the road operations, smarter methods, less time on the road for the machinery, cost savings without loosing quality, innovations in the safety and working environment for road workers and the administration itself becoming an attractive employer. Banverket hopes this study will help it to formulate a procurement strategy. At present Banverket has a policy but no official strategy. It is often locked in with the original supplier who developed the tailored technical and security systems once upon a time. Often the technical

10

T. Stenbeck

installations demand a lot of service. Why no innovations? No incentives? Insufficient rewards? Both parties must gain from being innovative. At present, the service provider maximizes his profits by not solving the problem in a sustainable manner. Banverket hopes the study will provide incentives for employees at all levels to take responsibility for the long-term development. Today there are few incentives to fix problems in the long run and too little learning from experience. Contractors express hope that the study will contribute to an innovative climate in the industry. The present strict focus on price by the principals, and inability to develop the specification and delegate responsibility, favor short-term solutions. It creates too little incentives to innovations. It reduces the attractiveness of the industry when recruiting personnel. Performance-specified procurement, partnerships between principal and agent, and appraisal of soft variables are approaches of great interest from the contractors’ point of view. (Paper 1).

1.3 Initial Reflections and Considerations

Scientific Contribution

Being initiated and funded by the industry, both owners and contractors, the study can be regarded Demand-Driven Research from their point of view and as Curiosity Research from other industries’ point of view, Fig. 3 below, top row. Some results have been applied as full-scale experiments and the interim results are briefly commented upon as Applied Reseach, Fig. 3 below, bottom row.

Curiosity Research Demand-Driven Research (Bohr) (Pasteur) Applied Research (Edison) Economic contribution Fig. 3 Pasteur’s Quadrant (Stokes 1997)

It was assumed that the purpose of research is to address at least one question and give at least one answer to each question. The answer can be that there is no answer. Any answer should contain as few restrictions as possible, allowing highest possible generalizeability, everything else equal. Since the researcher cannot know all possible uses of the research findings, the decision of the generalizeability is mainly the reader’s, based on the provided description of the research methodology. Methodological considerations and the methodology of this study will be described in the next section.

1.4 Purpose The purpose was to respond to the research questions in Table 2, based on quantitative and qualitative transportation infrastructure maintenance experience and data and in Sweden, USA, Canada and France. Table 2 Research questions elaborated in the study. 1. 2. 3. 4.

How has innovation been promoted and what were the results? How much innovation did contracting bring? How much innovation did performance-based specifications bring? How can cost models be elaborated and contribute to innovation?

11


The research questions 1–3 are superimposed on research challenges in general in Fig. 4. The windshield wiper moving to the right is to symbolize the knowledge/research front. The deeper the tone, the more settled is the knowledge. The proved-right versus proved-wrong aspect is shown as a dashed line is not equal to the degree of certainty. An innovation can with 100 % certainty be profitable today, without being it tomorrow. The dashed line wants to highlight that more information on mistakes and failures are of interest to science, since they contribute to scientific knowledge and helps mistakes in the past to become successes of tomorrow with minor modifications. If only 20 % of all inventions turn out to be successful innovations, information on the less successful 80 % may produce as valuable or more useful knowledge as the few successful cases.

Hypotheses Ideas Inventions

Validated knowledge Innovations

Innovation promotion

Contracting

Proved to be right Proved to be wrong

Performance-Based

Supposed right

Specifications

Supposed wrong

Knowledge front

Fig. 4 Bird’s eye on research in general and this study in particular.

12

T. Stenbeck


13

2 Methodology Based on the works by Stake (1995), Yin (2003), Johansson (2002) and Flyvbjerg (2004) the following generic methodology differentiation was adopted. Qualitative study searches for qualities, A, B, C, parameters or properties that have an influence on the studied phenomenon Y. Quantitave studies searches for to what extent A, B, C have influence on Y.

Y = aA + bB + cC

(1)

where Y is the dependent variable, A, B and C are the independent (qualititave) variables and a, b, c are the quantified influence of A, B and C on Y. Usually case study methodology is appropriate for qualitative studies and statistics for quantitative studies, but there are probably exceptions and intermediate forms. Often quantitative studies use a large number of observations with comparable data of a small number of parameters or properties (Johansson 2002) but quantification of the parameters’ influence may also be possible with few observations. If gathered and selected correctly, in a properly randomized manner or that the observation set contains the full population, predictions and conclusions from the sample can be generalized to the population. If selected with a known bias, generalizations should still be possible to make in a scientifically cautious, i.e. the least state-of-the-art-changing, direction. If, e.g., the percentage of the population with sympathy to Manchester United is the research question, and the sample is picked among its spectators at a match, the result of the extremely biased study can still with reasonable certainty be generalized as an estimate of the maximum value of the percentage of the UK population with sympathy to Manchester United. Many studies may start with a qualitative phase, an explorative, pre-study where surveys and similar techniques proposed by Stake and Yin are useful, since a single case can be sufficient to get a first idea of the important parameters. After having achieved A, B, C from a case study, deepening the study into the same case or looking around for other cases both should be viable approaches, depending on the circumstances. Lind (2003) lists five main types, or stages, of scientific study: 1. 2. 3. 4. 5.

Build mathematical models and verify causal relationships in these models; Run statistical analyses of existing data with econometric methods; Make laboratory experiments; Interview central actors; Participatory observation.

Depending on previous knowledge, each methodological approach above carries a marginal utility. The researcher should choose the method with the highest marginal utility. Each study using a certain method will usually reduce the marginal utility for that method for future studies. If not, further studies with the same method should continue until its marginal utility falls below one of the others (Lind, 2003). In this study, Lind’s list has been used in a different order: 4-5-3-2-1. First qualitative methods such as interviews were used to identify parameters of importance for innovations and then mathematical methods were used to quantify their importance and influence on the innovation rate. The mix of methods create a method triangulation. Results and conclusions will be strengthened by being achieved with several, independent methods, even if this will inevitably produce some, hopefully minor, variation in the result, corresponding to the error triangle (Section 1.1 Definition of Terms – Triangulation, page. 9).

14

T. Stenbeck

2.1 Methods Used in This Study Literature review and interviews were the first used methods to quickly gain knowledge and ideas of what parameters A, B, C that may promote innovation. An experiment, a seminar and participatory observations complemented the survey. In 2005 - 2006, cost data have been analyzed and compared after necessary corrections achieved by regression analysis. Interviews have been conducted to find factors for correction prior and after running the statistical analyses. Table 3 lists some used methods chronologically, which regions that were included and in what papers the methods were used. Table 3 Methods used in various steps of the study. F=France, C=Canada, S=Sweden, U=USA and WA=Washington State, USA. Research question

Methods used

Papers concerned

How to promote innovation

Literature, interviews S: Experiment, seminar WA: Regression

Papers 1–8 Paper 1 Paper 1, 2, 3, 5, 8

Innovation rate 1985 – 2005

S: Literature, interviews F: Interviews

Paper 1 Paper 1

Character of innovations 1985–2005

S: Literature, interviews F: Interviews U, C: Interviews

Paper 1 Paper 1 Paper 8

S: Literature, interviews U: Literature WA: Regression, interviews C: Literature, regression, interviews

Paper 1 Paper 1, 8 Paper 3, 8 Paper 8

Quality effects

S: Literature, interviews C: Observation

Paper 1 Paper 8

Innovation effects

S: Literature, interviews

Paper 1

S: Literature, interviews U: Literature WA: Literature, interviews

Paper 1, 4, 5, 6 Paper 4, 5 Paper 4, 5

Quality effects

S: Literature, interviews U: Literature WA: Interviews

Paper 4, 5 Paper 8 Paper 5

Innovation effects

S: Literature, interviews, experiment

Paper 1

Other effects

S: Literature, interviews

Paper 1, 4, 5

Regression, interviews

Paper 2, 3, 6–8

Contracting Cost effects

Performance-based contracts Cost effects

Cost models

Interviews was the initial method used (Paper 1) allowing a general search for knowledge and information. The first selection of respondents was provided by one of the project initiators, CDU in a non-randomized manner. Since this could have a bias, a conscious effort to find opposing opinions was carried out. Owners, contractors and consultants were all represented, and the number of interviews with time became reasonably high, approximately 50. The selection method calls for an appropriate,


15

cautious, interpretation of the results, similar to when undertaking case studies of possibly untypical cases. Most studies can be deeper and more rigorous and this study is no exception. It should be regarded as the current answer only, based on the always limited dataset that was included in the particular study. To evaluate innovative contracting versus traditional contracting, as an example, quantitative, ex post comparison of many comparable objects would have been preferable. Since highway construction is complicated and multidimensional, a case study approach (Paper 4), multiple at its best (Paper 5), was considered to be the only realistic method. Within the case study concept, the depth of the study could have been deeper. The case studies involving civil engineering project comparisons could have been deepened considerably into engineering issues, but with increasing risk of running into a dead end comparable to complex mathematics leading to multiple solution equations. Papers 2, 4 and 5 describe construction projects where maintenance comes into the picture with regard to calculate life-cycle cost. Initially uncomparable objects need to be standardized with correction factors before being comparable, but this operation to produce a useful result will in most cases meet objections from experts in some aspect. When studying the blueprints of the second case in Paper 5, as an example, it was not uncontroversial whether one material’s increased thickness compensated for reduced thickness of another. Each material has its properties and they may affect endurance and load carrying capacity differently. Deepening the study into all these considerations would gradually drift it away from the incentives to innovation focus. A quantitative study, to support qualitative results obtained by interview has obvious merits to triangulate and get supporting or non-supporting results. Also, if a respondent does not have a quantitative experience, and has not experienced the alternative asked about, what is its opinion worth? For this reason the qualitative nature of Paper 1 was supplemented by the more quantitative approach of the other papers, which all try to summarize the findings by quantifying and assign weight to the information and sum all up into a number. Each such number is an estimate with a certain error, in some cases stated, in other cases up to the reader to estimate based on the description of the methodology and the reader’s own experiences. Although the quantitative studies are considerably better than pure qualitative studies, some will still find them too shallow. A number of observations close to the minimum, or even below the minimum normally required for statistical purposes (MIT 2003) were used. Challenging the statistical limits became a research in itself with new insights and saved time, since the method was more important than the result in the particular case.

2.2 Delimitations Listed methods to stimulate innovation brainstormed in group sessions and interviews are not exhaustive. With regard to performance-based contracting two cases only have been studied. Paper 3 is methodologically generalizeable, but the weather data are sitespecific why the particular model values obtained are useful only for Washington State. The sample of reality was constrained in this case as declared in Table 4.

16

T. Stenbeck

Table 4 Space, time and scope of this study. Dimension

Samples of reality used

Space

Sweden, France, Canada and Washington State, USA. The statistics used for quantitative analyses mainly relate to North America. The study has been conducted between 2003 and 2007, and as fresh as possible data and literature have been used. Conditions back to the end of the 1980s are however represented, thanks to shared experience of interviewed experts. As many practices to spur innovation as possible have been searched. As a second step, contracting and performance-based specifications were selected for further study.

Time

Scope

17


3 Results Savings of 13 – 28 % were recorded as contracting and performance-based specifications were introduced in Swedish highway maintenance (Arnek 2002). Liljegren (2003) added that the administration had grown by less than 5 %. 21 % were gained in Massachusetts and 12 % in Virginia (Segal and Montague 2005). Another American study, however, found savings in American states and Canadian provinces debatable (Riberau 2004). Swedish statistics of user’s appreciation of the quality showed a negative trend, now stopped, Fig. 5. User satisfaction

very or fairly pleased, %

80 70

72 74 62

60

71

69 64

61 54

59

57

40

39

62

61

52

50

46

40

45

40

30 20 Private drivers

10

Professional drivers 2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

0

Year

Fig. 5

Private and professional driver satisfaction with maintenance on Swedish highways. Source: Swedish Road Administration (SRA 2007).

Whether the quality has decreased in technical terms is a controversial issue (Liljegren 2003). Do the users know what to expect for the amount allocated? Would they actually be willing to pay for additional quality? The internal discussions are similar in Canada (BC MoT, personal communication, 17 December 2004) and Sweden (SRA 15 November 2006, personal communication). Design-Build, one example of the performance-based specification philosophy, has reduced conflicts between owner and contractor, delivered constructions faster and with less cost overruns than traditional contracts. Less cost overrun is not equal to lower costs. Some studies, however, also found lower costs (Sanvido and Konchar 1999). An obvious concern arising is whether the advantages have been achieved at the expense of quality and long-term maintenance cost (life-cycle cost). Sanvido and Konchar addressed the quality issue but appear to have allowed the directly involved to judge their own quality. Surprisingly few studies address the quality/cost issue by comparing design-build (DB) and design-bid-build (DBB) with robust methodology. Only few compare the actual costs per square meter, fewer have included quality and none, it seems, has compared life-cycle costs in a quantitative manner. Quality has been measured by survey and interview of key players in the projects by Sanvido and Konchar (1999) and CII, the Construction Industry Institute in 1997. The outcome of the study was slightly but significantly positive for DB compared to DBB in the CII report and positive for DB also in Sanvido and Konchar’s report.

18

T. Stenbeck

An experiment within Vägverket and a seminar within Banverket preceded and succeeded by interviews, supported lack of incentives as an explanation of a low innovation rate. New ideas have been criticized without concern for the long-term effects on the incentives. Innovative contractors have adapted to the situation by strict adherence to bidding as proposed, but have in some cases been able to save costs by smarter methods after the contract has been signed. Many innovations are difficult to hide. The openness and keenness to copy in the construction industry reduce each actor’s economic incentives to contribute to the technical progress (Stenbeck 2004). France, with an inhouse policy similar to Sweden’s before the reform, has continued to produce technical development in the same period while it has stalled in Sweden, indicating that contracting has lowered the pace of R&D as compared to if the in-house production system would have prevailed (Paper 1).

3.1 How Can Innovation Be Promoted 1 A number of innovation promotion methods were found in Sweden, USA, France and Canada, others were brainstormed and elaborated with owners and contractors within the project. In the following, the idea is presented in italics and any experience follows in normal font (symbolizing a move towards the left in the “wiped-onion” Fig. 4, page 11). Contracting 1. The previous monopolist’s specialized machinery was made available to anybody for sale (AB, BC) or for hire at competition-neutral rental rates (S). The new contractors often kept their existing, less specialized equipment (S). Even if taking into account some loss as the equipment was sold out, the outsourcing reform has saved 3 – 5% per year (AB). The sales of the equipment make it difficult to unroll the policy change (AB, BC). Environment issues have been easy to include in the contracts, like any other parameter. Presently, an ISO 14001 certificate or its equivalent gives 1% reduction of the price to be compared with other bids (S). 2. The previous monopolist’s special knowledge was made available to all by publishing a book describing current methods and practices (S). The owner’s dream was that the new contracting policy would call for a new book filled with new technology to appear shortly, but such a book has still not appeared, which is taken as an indicator of slow development (S). Performance-Based Specifications 3. Weather-regulated payment schemes have been introduced to reduce the contractor’s weather risk (S). (Papers 1 and 5) The payment being weather regulated was an argument to convince the Swedish contractors, used to be paid per hour without any risk, of the new performancebased models. The contracting reform would probably have led to higher prices or boycott, if weather had become the contractor’s risk in full. A contradicting experience is that the public contractor at least once submitted a bid indicating a preference for a lump sum deal (weather items were unit priced to zero, Paper 3). In British Columbia the contractors bid lump sum and carries the full weather risk. Bad weather in an early year of the current contracts raised complaints and concerns 1

S =Sweden, AB = Alberta, Canada, BC = British Columbia, Canada, WA = Washington State, USA, F = France


19

(BC). In Alberta, the weather risk is carried by the owner through time and materials contracts. The cost savings since contracting was introduced are considerably higher in Sweden than in most of the USA and Canada (Papers 1 and 8). An input difference is that Sweden and Alberta, both with high savings, have offered the contractors weather risk reduction. This may have contributed to the savings or it may be a coincidence. 4. Performance contracting was developed, considered as closely related to, if not a condition for, successful contracting. Contractors and current North American trends (numerous papers at TRB Annual Meeting 2007) praise performance contracting. Some, but relatively few studies, presented actual ex post experience. Most of these were more cautious regarding the positive effects. The Swedish Road Administration’s ex post experience made them hesitate to continue with performance contracting in 2003, but it was decided to give it another chance, not the least because of its modern appeal and the lack of other methods. If successful, a third of the contracts could be Design-Build in 2010 (Procurement in Focus, Swedish Road Administration seminar 4 October 2006) (S). Other criteria than price 5. Multi-criteria evaluation. Quality systems and staff competence are other factors than price that are evaluated when awarding contracts (AB, BC, S). With 70 % weight to price and 30 % to other, so-called soft parameters, Vägverket’s contractors complained that the bid selection became too arbitrary and subjective. Soft parameter weight was reduced gradually and was only 3.5 % in 2004. Some complaints are heard that this is too low, but not by the majority. Banverket weighted price 70 % in their bid selection in 2004. They have at least 10 years less experience than SRA, and similar complaints as SRA had at the time are heard. The soft parameters in both agencies have been and still are based on documentation submitted with the offer (ISO certificates etc), not on actual performance in the field. (S) (Paper 1) 6. Variant bids were encouraged. To maintain authority and fairness in the bid evaluation, the Swedish owners see no general method to allow bids not following the rules of the tender fully to win the tender. If the variant bid follows the rules fully, it becomes a main bid, and no special treatment is necessary. (S, F) (Paper 1) Idea generation 7. Idea mailbox has been used within organizations. The number and quality of contributions are falling, both in Sweden and Washington. The Swedish public contractor proposes that rewards are too low or that the staff is too overcharged to have any time for own innovative ideas. (S). In Washington, rewards have been high (profit-share 10 %) but this has not helped. Sometimes the innovator had already convinced its superior to introduce the proposed innovation in the field, when it was finally acknowledged and rewarded through the mail box system. The owner’s hypothesis is therefore that the mail box system is too bureaucratic, too slow or simply outdated as an incentive to innovation. (WA)

20

T. Stenbeck

8. Design contests with awards for the most interesting solutions. (Paper 1) Within the public French rail company SNCF a contest was tested. 80 proposals were received and 23 were selected for further study. The evaluation of the proposals will encounter similar problems as variant bids; the apples versus pears comparison dilemma, why the problems and opportunities are similar. Rules for a contest have been elaborated by a group within SRA (S). 9. Additional funds available for innovative experiments France expresses success with their funds available for innovative experiments. They are combined with particular validation routines in a public-private partnership, leading to some years protection of the innovation for the involved companies. As France’s innovation program was discontinued all innovation ceased, which provides further evidence to its efficiency. The reason for the discontinuation was debate over its compliance with EU regulation (F). (Paper 1). It was relaunched in 2006 with revised rules, why it is too early to gather experience of the outcome. In Sweden, funds have been available without applying candidates (SRA 7 July 2005, personal communication). Maybe marketing, the protection issue or the routines guiding the cooperation between owner and contractors explain why France reports success where Sweden does not. Partnering and partnerships Nutek (2006) proposes the following further, partnering-related procurement methods to promote innovation: 10. Negotiated procurement procedure, partnerships, concessions, technical and competitive dialogue These procedures are proposed as useable to develop or alter a specification together with a selected supplier in order to take advantage of its innovativeness. The methods must not be used to favor a particular bidder and reduce competition. Internationally, these methods have been used mainly to save money or realize infrastructure investments earlier than would otherwise have been possible (PrivatePublic-Partnerships). In Sweden, the methods have been used to do the best possible out of a situation with too few bids (e.g. Design-Build procurements where only one bidder has submitted a bid). The effects on technical development has not been the main concern and no study has been found. 11. Combinatoric and split procurement The Swedish Road Administration tried splitting maintenance areas both in space (smaller areas) and scope (contracting bridges, rest areas etc separately) in 2004– 2005. The main reason was to increase competition, by making it easier for small, specialized local companies to participate. Bidders were also allowed to combine adjacent areas in space and scope and offer a volume discount per combination. Some financial savings were achieved with the method, but the administrative costs were higher than intended, since the software that was developed to evaluate all bidded combinations against each other had to be complemented by manual methods (SRA spring 2005, personal communication). Better competition and lower prices was the objective. No effects on the technical development were found.

21


3.2 Did Contracting Bring Innovation? How Much? Our definition of innovation (Section 1.1) supports an answer primarily based on the resulting economic benefit. Ideally the answer should be quantified, a percentage plus or minus per year obtained by contracting. A complementary measurement unit is a qualitative assessment of the number and importance of innovations arising within each system, obtained by asking involved to mention innovations that they had witnessed, which will however be more subjective. Contracting of previous inhouse maintenance is introduced to varying extent in all the studied countries: France, Sweden, Canada and USA. The reasons, manner and results appear contradictory at first glance. The challenge has been to find common and differing elements to explain the heterogeneity and draw conclusions regarding best practice for the future. Table 5 ranks the compared regions from top to bottom in order of costs and the parameters are ranked from left to right in order of correlation with the costs. The mathematical summary of Paper 8 is that contracted provinces had on average 26 % higher costs than inhouse provinces. Table 5 Maintenance costs 2005 per m highway across Canada, Washington State and Sweden in Swedish Kronor (SEK). 1 CAD = 6.80 SEK, 1 USD = 7.50 SEK. Italicized are intrapolated values. Area Manitoba Saskatchewan1 New Brunswick Newfoundland and Labrador Sweden Nova Scotia Alberta British Columbia Washington State Ontario Quebec ADT Lanes Winter Inhouse Note 1 Note 2 Note 3

SEK/m 20 24 30 33 35 37 48 55 69 77 91

Main policy Inhouse Inhouse Inhouse Inhouse Contracted2 Inhouse Outsourced Outsourced Inhouse Outsourced Mixed

ADT 1200 500 1200 1300 1400 1400 2100 1200 5700 2900 1700

Lanes 2 2.2 2 2.0 2 2.0 2.2 2 2.63 2.37 2.1

Winter 42 % 25 % 70 % 50 % 50 % 30 % 45 % N/A 22 % 45 % 42 %

Average Daily Traffic, both directions. (Vehicle km travelled / centerline km of highway network). Average number of lanes, both directions. (Lane km / centerline km). Percentage of the budget used for snow and ice removal. Many provinces outsource some services, e.g. peak winter maintenance. When contracting is less than 10 % of the budget, they are still labeled inhouse in this comparison. Southern Region of Saskatchewan only, including the Trans Canada Highway. Probably more traffic and less severe winters than the northern ⅔. In Sweden a public contractor has 60 % of the market. 2.8 if including special lanes and bike trails. In the Canadian figures special lanes are included but not bike trails.

The literature review and qualitative research methods employed in Paper 1, resulted in lower prices in Sweden, but probably at the expense of some quality and technical

22

T. Stenbeck

development. In North America, science is contradictory regarding the cost effects. An own, quantitative, study was undertaken to investigate the costs per km in the Canadian provinces, Washington State and Sweden, Paper 8, also including interviews and ocular inspection to receive a brief impression of quality and innovation rates. The international comparisons with USA and Canada provide examples that they use more advanced chemicals and mixes of chemicals for winter maintenance and more flexible trucks. The equipment and the chemicals are however known also in Sweden, why the different uses may be rational and dependent on local factors. Environmental concerns limit the use of sodium chloride for deicing below -6 ºC, due to increasing volumens required as the temperature falls, while Newfoundland finds sodium chloride, mixed with sand when used on minor roads, useful down to -18 ºC. The opposite is the case at temperatures around 0 ºC. Mixing salt with water is considered an innovation in Sweden and used extensively. In North America pre-wetting (anti-icing 2 , deicing at temperatures above 0 ºC) is tested scientifically, its benefits considered less evident and more dependant on the situation and in France anti-icing has not been allowed for environmental reasons. Stricter trends are emerging also in Canada since its environmental agency recently declared sodium chloride a toxic substance.

3.3 Did Performance-Based Specifications Bring Innovation? How Much? Four cases were selected to provide basis for a response to the research question. Case 1 and 2 are anonymized to protect the sources and help science focus on the lessons learnt instead, Case 3 was proposed to the be studied by Banverket and Case 4 was found in the literature. Cases 1 and 2 were Design-Build (DB). The first had a maintenance part as well and was warranted for ten years, and the second is interesting because of its negligence of the maintenance and the life-cycle cost. Both cases were pointed out as interesting by research collegues as well as by those involved in the projects, why both must be considered somewhat flagships i.e. well-known, pioneer DB projects. This selection method contains a risk that the cases are positively biased, by being loaded with success hopes, and a negative bias by being pioneer projects, based on little previous experience. Availability of data thanks to personal contacts played a major role for the selection of the three first cases. Each DB case (Case 1, 2 and 4) was comparable with a DBB (Design-Bid-Build) alternative, and the third case could be compared with previous performance in the same rail maintenance area.

3.3.1 Case 1 –Design-Build-Maintain with Ten Year Warranty Case 1 initially had an owner-provided design but a less complicated design was offered by the contractor, which was according to the tender conditions. No other bidder proposed an alternative to the owner-provided design, and their bids were 15 % higher. The tender documents contained performance criteria, such as permitted deviation in space from the corridor assigned and more precise, local deviation expressing evenness like the International Roughness Index, IRI, and cross-sectional slope. Ex post 10 years later, where the owner had insisted on the initial foundation design performance criteria were exceeded, but unfortunately, much of the rest had settled more than the maximum tolerance. One 20 m segment had settled 3 times the limit, partly because the performance criteria was less tolerant near bridges. The contract was 2

Anti-icing is when deicing is used preventively to avoid ice from forming


23

precise on the requirements, but less concise regarding the financial effects of the deviations. Bonus for exceeded requirements was mentioned but not quantified. (Papers 4 and 5)

3.3.2 Case 2 – Interchange Construction with Design-Build Case 2 was, according to experts at the time, the only completed highway DB project for its region. (A later and less known case had been better to avoid the risks of bias arising from being a flagship). A major advantage with this case was that it could, according to the owner, be compared with a nearby DBB project, constructed a year earlier. Both the DB and the DBB alternatives received the same Engineer’s estimate, 18 M. At completion, with change orders added, the DB project cost 20 % more (27M) than the DBB project (22M 3 ). When comparing the sizes of the projects, the DB appear about 25 % smaller in footprint and mass of volumes 4 , contrary to what would be expected from the cost difference. The proximity between them suggest similar ground conditions and no information (except the Engineer’s estimate being equal) has hinted at the ground conditions being significant for the comparison. When studying the two designs in detail, the DB project design is less detailed and therefore difficult to compare technically with the DBB design. A civil engineering expert was consulted to make a judgment but gave the traditional an open-ended answer of an engineer (“it depends”!). The owner did not evaluate quality other than that both constructions had to fulfill the code and life-cycle cost was not evaluated (personal communication, 7 March 2006). (Paper 5)

3.3.3 Case 3 – Performance-Based Contracting with Incentives The third case for which data was available, showed positive results of performancebased specifications. Quality improvement at no additional cost was achieved when the contract included incentives based on continuously monitored performance. For Banverket, the Swedish railway owner, train delay and registered errors in the rail system had been almost constant for a number of years despite weather and organizational turmoil. Since bonus for reduced delay and number of errors were introduced at the end of 2004, 10 % reduction of delay time and 20 % reduction of errors were registered. (Papers 6 and 7). Incentives having almost too high effects on performance have been experienced also by SRA (2005).

3.3.4 Case 4 – Motorway Design-Build in the UK Langford et al (2003) compared 11 motorway projects in the UK. They were all segments along the same highway with identical specifications but different procurement systems. Five were procured with DBB+unit price 5 , five were lump sum and the 11th was procured as a Design-Build. The result was that lump sum reduced the price per km by 11 % compared to unit price, and more if complicated structures like bridges were included in the segment. The single DB project was more expensive per km than the average of the unit price projects

3

A figure provided by the owner. According the research collegue mentioned previously, the DBB project was equally expensive, 27M. The difference may seem important, but alters the study conclusion marginally regarding the gain versus loss with DB contracting.

4

According to the researcher mentioned in footnote 3, who has wished to remain anonumous. The two projects physical layout on the Internet support her figures to be of the right magnitude.

5

Langford et al use the notion Measure and Value, which has been assumed to equal the notion Unit Price used in American literature.

24

T. Stenbeck

as well as the lump sum projects. Langford et al comment by saying that it should not be taken as representative without giving a reason. A reason might be that since it turned out opposite from expected, a single observation should be treated with caution. If so, that precaution is less valid when now added to the three DB case studies why it was included in this multiple case study without adding more detail than the original source contained. (Paper 5)

3.4 Cost Models To compare maintenance costs of regions, correction factors for varying conditions would improve precision and usefulness of the comparison. Detailed cost data as well as transportation and technical highway related data were made available by Washington State DOT and weather data was found on the internet. From this data bridge influence, weather influence, traffic influence etc on maintenance cost have been modeled.

3.4.1 The Bridge Life-Cycle Cost Model (Paper 2) The resulting equations to calculate life-cycle cost, validated on Washington State, of a bridge stock was: C0 = 1,100 x deck area + 240,000 USD

(2)

where C0 is the reconstruction cost in 2004 USD if deck area is expressed in m2

Mx =

C0 x × (1 − 1 - ( ) 2 ) L1 L2

(3)

where Mx is the maintenance cost at age x, L1 design life of the new bridge and L2 the current, effective life of the existing bridge or bridge stock, x

C x = C0 + ∫ M x

(4)

0

where Cx is the accumulated cost until age x, and x =life

LC C = C 0 +

∫M

x

(5)

x =0

is the life-cycle cost. The models can be used for repair versus replacement decisions, new highway planning and budgeting and evaluation of alternate bids obtained in tenders.

3.4.2 The Snow Influence on Maintenance Cost Model (Paper 3) Regarding snow influence of maintenance cost, the method discerned the three weather stations whose annual snowfalls were best correlated with Washington State winter maintenance costs. For each inch of snowfall accumulated at these three stations together, the state maintenance cost rose by a certain amount (44,000 in 2004 and 103,000 in 2005) The intercept, independent of snowfall, which can be interpreted as the fixed costs, amounted to 23,000 in 2004 and 18,000 in 2005. The model predicted 2004’s snow and ice control cost with 7 % accuracy and 2005’s with 2 % accuracy. The ghostly high and converging accuracy in predicting the cost was reached despite the general unpredictability of weather and that one of the two concerned winters was unusually warm.


25

4 Analysis The analysis to the first research question, how to promote innovation, was embedded in Section 3.1, but analyses of the second, third and fourth research questions will follow in this chapter. Why has contracting produced different results? Why have performancebased specifications not conquered the world long ago? How can cost models contribute to innovations?

4.1 Contracting Paper 1 supports the savings found by Arnek (2002) on some aspects. “Today half the staff operate twice as large areas”. “Our five do what the public contractor needs 15 to do”. How the leaving staff were compensated as well as the average salary of the remaining staff compared to the previous average salary would be interesting further studies, that may reduce the savings from outsourcing. The further payment of the public employees is part of the problem and the considerations at stake in Canada (Paper 8). But even if the initial saving would be less, when taking the one-time costs into account, a long term saving in Sweden seem to remain. Before the policy change, maintenance had been subject to above-average inflation (Österberg 2003).

Fig. 6

Swedish highway maintenance costs (Österberg 2003). 1980–1992 prices tripled in nominal terms and increased by 25 % in real terms. The trend line has been added to the artwork of the source.

If extrapolating ’Prices in 1999’ in Fig. 6, from 1980–1992 into the future, contracting has saved 130/88 ≈ 50% in the ten years until 2002. Were the savings achieved at the expense of quality, technical development or staff’s work pleasure? Quality has dropped somewhere between 38 % 6 and an unquantified improvement, depending on source and method. More research would be desirable what the actual quality changes have been, since without clarity regarding quality effects, economic savings are difficult to value. If the economic gain has occurred without effects on 6

From 63 % satisfaction 1995 to 39 % satisfaction 2002 (Fig. 5). 39/63 = 0.62.

26

T. Stenbeck

quality and innovation rate, the long-term benefit will be the same percentage as the cost change. Else it need be corrected for the change of quality and the innovation rate. The long-term cost saving can be formulated as $ / km year 1 $ / km year 0

×

Quality year 0 Quality year 1

×

Innovation Rate year 0 Innovation Rate year1

(6)

According to the more demanding innovation definition in Section 1.1 (QQ/C), the Qualityyear 1 must be equal or greater than Qualityyear 0. For Sweden, the change of technical development would equal the difference between technical stand now and what it would have been without the contracting reform. That we will never know. A way to approximate it is to compare with other countries’ technical stand or innovation rate, in contracted and inhouse systems. The North American studies are, as the Swedish, not methodologically perfect. The Riberau (2004) report, briefly mentioned in Paper 1, would have gained from quantification of the loss with contracting. The credibility of Riberau’s report has both gained and suffered from being published by WSDOT, since its results support the current WSDOT policy. Its conclusion, “savings were not as high as expected”, sounds political. Why not tell what the savings were? Science, politicians and the public wish to know if there were any savings, how much and how they were obtained. Riberau provides little substance useful to respond to those three questions. Otherwise, however, the findings of Paper 8 as well as personal communication (24 January 2007) with one of the interviewed from Texas in the Riberau report, supports it to be true. The economic results of outsourcing in North America show less savings than in Sweden (a 26 % loss according to Paper 8), which makes the differences between the rollouts of interest. States considering a move from inhouse to contracting policy can pick the best to increase the chances for success, both regarding costs, quality and innovation rate. Some action is too late for those that have already made the move. The reform is costly to roll back, because of the lost competence and sold out assets (BC MoT).

4.2 Performance-Based Specifications The virtual non-existance of failed cases with performance-based contracting in the literature appeared to be too good to be true. Could it be that no researcher had dared to publish critical reports? Had all the researchers really forgotten quality and life-cycle costs in their studies? Perhaps publishers and peer-reviewers or awarely delimited away the quality aspects? More overall performance has been brought forward by the contractors as a route to innovation. If this is true and honest, they have to make sure that better quality is delivered, when owners give them the chance to prove their point. Owners contribute by insisting on full performance, since without full performance, no excess performance and the professional pride from having done a good job, necessary for innovation, cannot develop within any of the parties. In Case 1 (Section 3.3.1, the highway DBM with warranty), where the project was revisited after 10 years, the legal position for the owner to insist on full fulfillment of all contract requirements appear good. The contractor would hardly escape having to either do the required remedial action or pay the owner the cost of someone else doing the same job. But a positive relation has developed between the two parties and none of the involved, owner nor contractor, seem to have any incentives to push the issue.


27

In Case 1, the technical criteria proved to be defined well enough for the final evaluation. But in Case 2, the two compared interchanges, and many other PBS cases, the owner has not defined the technical criteria and their measurement well enough. Technical acceptance limits could be found in Case 1 and 3, but not in Case 2. Interviewed owners hinted at them being part of the code, but embedded in the code, they will be understandable only by few experts. This reduces the incentives and make evaluation, control and legal procedures unnecessarily complicated. Bringing them out of the code and into the contract would have been well invested money and effort. The owner would then perhaps also have realized how insufficient the code was for PBS purposes. It will most certainly not contain the quantified penalty/bonus provisions related to each deviation, also needed as incentives for full performance. Inspections are even more important in a PBS than in an ordinary contract, since the deviations have a financial importance (most of the literature expresses the contrary; an advantage of Design-Build being less inspection necessary). Not meeting demands means money both short and long term for the contractor, an incentive that the owner need to balance with harsh and frequent inspection. The old type inspector with such integrity may not exist any more. If so, PBS may deliver as good quality as traditional contracting and competitive fairness will gradually deteriorate. In Case 1, the contract stated that the performance criteria had to be met within 10 days after inspection, assumed to be taken care of by the contractor’s self-control. Obviously this inspection never took place, since the 3 times settlement beyond the performance criteria would not have passed unnoticed and could hardly have occurred during the last 10 days of the 10 year contract. The low incentives to self-control when there is no owner or thirdparty control is part of the PBS challenge, perhaps the reason for user dissatisfaction, Fig. 5, page 17. A continuous third party control like the one in case 4 might have assured the users the specified performance continuously as well as giving the contractor incentives to rectify by action before it was too late, both economically and psychologically. The frequency of the inspections, however, need not be stated in the contract. Since the specification is valid at all time, performance should be equal no matter how often the inspections take place. The costs for the inspections themselves should normally be on the owner, but if caused by the contractor (because of poor, previous performance requiring additional inspections) there are probably cases when the costs of the extra inspections should be on the contractor. Only these situations need to be part of the contract. Deviation from the specifications is nothing new, why the absence of consequence management in all the studied cases is puzzling. Bonus and penalties could have been better quantified in Case 1, perhaps like a price list (owner 7 July 2005, personal communication). Apart from facilitating the financial compensation, such clarification will be easier to forward to subordinate staff than general principles, at their best embedded in legal practice around liquidated damages that only legal experts are familiar with. Consequences of these deviations must be addressed more explicitly in PBS contracts, in particular in lump sum PBS contracts, than in traditional and unit priced contracts. How much costs were saved in the studied PBS cases? Case 1 was 15 % cheaper when built. Rectifying the safety-affecting consequences of the simplified but less performant foundation would be possible at a lower cost than the initial saving. Rectifying the whole highway (removing it, redo the foundations and put the material – or new material- back again) would, obviously, cost a lot more than the initial saving. The latter is however the correct calculation if DB is to be compared with DBB as apples to

28

T. Stenbeck

apples. There is no uncontroversial way to quantify all these considerations into a net effect. Let’s quantify it to 0, but have in mind that the true value is considerably higher. Extensively simplified, the average PBS gain can be estimated by adding the percentage gain in each case and divide by four (to understand the numbers, the Papers 4–7 need to be consulted): (0 + (27/22+1.25) + 1.03 – 0.85 ) / 4 – 1 ≈ 1.10. Assuming the studied cases representative, PBS would be 10 % less performant than DBB. As with contracting, the history of PBS proved to be considerably more disappointing than was expected, which hopefully leads to further development with realistic expectations rather than political turbulence or return to outdated methods and owner-contractor relations. The absence of control and consequences in the first less successful cases and the presence of control and consequences in the successful case support, however, that old type authoritative inspector is an element to reinstall to assure the success of PBS. If today’s cultures cannot respect such authority, it looks as the PBS is doomed to fail.

4.3 Cost Models Another reason that PBS and contracting has not worked is that the buzz words, like life-cycle cost and performance, have been poorly defined. “Hundred studies” with lifecycle cost among the keywords turned out to be purely technical studies from which the costs could not be derived, since life-cycle cost was referred to in a vague manner without calculating it, nor providing any method or tools to do so. For the PBS and contracting studies (Papers 4–8), such models would were considered necessary tools why elaborating them was made a part of the study. For the benefit of these own studies as well as future research by others, a life-cycle-cost-model applied on bridges (also useful to determine bridge content influence on total maintenance cost) was elaborated and accepted after peer-review for the TRB Annual Meeting 2007 (Paper 2). The method elaborated in Paper 3 for weather modeling is useful for snow and ice-related risk reduction in maintenance contracts anywhere in the world, since no special weather stations are necessary. Paper 3 challenges some statistical truths by showing that good predictions can result despite disobedience to the statistical rules of thumb. The new insights obtained through this approach proved useful e.g. in Paper 8, to determine traffic influence on maintenance cost in Washington, reasonably generalizeable to Canada. Being based on accumulated snowfall only, the method described in Paper 3 may be a more efficient alternative to regulate payment to contractors anywhere in the world, than complicated models requiring RWIS (Road Weather Information Systems) stations, hourly or quarterly registrations and special weather definitions to regulate payment. A similar validation exercise as in Paper 3, to validate the precision of the Swedish existing models, could not be found, why application on Swedish data and comparisons is a future interesting study to undertake.


29

5 Conclusion and Future Research Defining performance in the different levels of contracts is a challenge with many pitfalls. Aspects embedded in the existing detailed designs will be forgotten when converted to more functional language. Attitudes from owners, superiors and colleagues probably need to be more tolerant for mistakes, so that they can be reported more openly, to increase the speed of innovation. In response to the first research question, how innovation has been promoted, a number of low cost methods were found that could be initiated with relatively low success rate probability, but where negligeable cost and external positive effects could motivate them, such as contests, seminars, summarized in Section 3.1. In response to the second research question, how much innovation that contracting has produced, the qualitative answer, when owners were asked to give examples of innovations, produced a near zero result. The quantitative answer (QQ/C) was a loss of 26 %. In response to the third research question, how much innovation that performance-based specifications have produced, the qualitative answer was as meager as for contracting and the quantitative loss 1–50 % depending on how quality and innovation is defined. Three out of the four studied cases fail to qualify as having delivered innovation according to the definition in this study. In the fourth, quality was gained at no cost, which fulfills the QQ/C criteria, even if none of the interviewed could recall any particular technical innovations. In response to the fourth research question cost models are essential to even out differences between compared cases. Absence of such practice has probably contributed to the somewhat unbalanced picture of contracting and PBS that arise when studying the literature and report of previous study. The parameters included in the models elaborated here were rationally selected by being considered the most important for a grounded reason, but additional parameters and the same parameters based on more data are welcome. The actual outcome of performance-based contracts, the effect of reduced tender documents, simplified bid submission, yearly cost evaluation, incentives and other proposals adopted recently by Vägverket 2006 are interesting areas to study in a couple of years, when the first conclusions can be drawn. Those studies will complete the validation process (Fig 1) initiated by this project on incentives to innovations. If increased control and yearly life-cycle cost as contract award criteria function longterm, are still to some extent adhoc hypotheses, since more cases are needed to validate them. On a more technically detailed level, it remains to be further researched in Papers 6 and 7 which precise factors (error codes etc) that contributed to the quality improvement, by studying other similar cases or a deeper study of the already studied cases. The appropriate amounts or percentages of bonus is another research field for the future, mainly for owners, but where science can assist with evaluation, generalization and knowledge transfer.

30

T. Stenbeck


31

6 References Adams, T. and J.A.Smith (2006). A Synthesis of Measures for Highway Maintenance Quality Assurance. TRB Annual Meeting CD-ROM, Session 690, Paper 06-0966. Arnek, M. (2002). Empirical Essays on Procurement and Regulation. Uppsala Universitet, Uppsala. Johansson, R. (2002). Ett explikativt angreppssätt (An explicative approach). In Nordic Journal of Architectural Research, 15, 2, pp.19–28. Jones, H. (1975). Modern Theories of Economic Growth. Thomas Nelson & Sons Ltd, Surrey, UK. Flyvbjerg, B. (2004). Five misunderstandings about case-study research. In Seale, C. Goto, Gubrinum, J.F. & Silverman, D. (Eds.). Qualitative Research Practice. Grennberg, T. (1965) Hur kan vägbyggnadstekniken utvecklas? (How can road engineering develop). Vägbyggaren 6, Stockholm. Grennberg, T. (1998). Ta bort krökarna i byggsvängen. (Remove detours in construction). Torsten Grennberg AB, Saltsjöbaden, Sweden. Langford, D.A., Kennedy, P., Conlin, J. and N.McKenzie (2003). Comparison of construction costs on motorway projects using measure and value and alternative tendering initiative contractual arrangements. Construction Management and Economics. December, 21, 831–840, UK. Liljegren, E. (2003). Konkurrensutsättning av Vägverkets drift- och underhållsverksamhet (Contracting SRA’s maintenance). PhD thesis, Urban Planning, KTH Royal Institute of Technology, Stockholm. Lind, H. (2003). Rigoröst eller intuitivt – är det vad vi har att välja emellan? (Rigorous or intuitive – is that what we have to choose between?). In Ekonomisk debatt (Economic Debate) 6. Nationalekonomiska Föreningen, Stockholm. Meyers (1897). Meyers Konversations-Lexikon (Meyer’s Conversation Dictionary). Bibliographisches Institut, Wien, Schoenfeldt & Co Berlin. MIT (2003) Sloan School of Management. Lecture slides on http://ocw.mit.edu/NR/ rdonlyres/Sloan-School-of-Management/15-063Communicating-With-DataSum mer2003/F904D01C-8EA7-4C93-B03C-CF3E13115044/0/lecture16.pdf, accessed 18 May 2006. Nutek/Swedish Government (2006). Offentlig upphandling som drivkraft för innovation och förnyelse (Public Procurement to Promote Innovation). Publication R 2006:21, Stockholm. Olsson, U. (1993). Funktionsentreprenad för drift och underhåll av vägar och gator (Performance-Based Contracts for Road Maintenance). PhD thesis, Tekniska Högskolan, Luleå, Sweden. Österberg, R. (2003). Contracting Out Public Services. PhD thesis, KTH Royal Institute of Technology, Stockholm. Riberau, N. (2004). Review of Highway Maintenance “Outsourcing”. Washington State Department of Transportation, Maintenance and Operations, Olympia, WA, USA. Salter, W. (1966). Productivity and Technical Change, Cambridge University Press, UK.

32

T. Stenbeck

Sanvido, V. and M.Konchar (1999). Selecting Project Delivery System, Comparing Design-Build, Design-Bid-Build and Construction Management at Risk. The Project Delivery Institute, State college, USA. Schumpeter, J.A. (1961). The Theory of Economic Development, Oxford University Press, New York. Seaden, G. (2001). Symposium organized by Institute for Research in Construction and National Research Council Canada. http://irc.nrc-cnrc.gc.ca/symposium/paper/ seadenpres.pdf, accessed 18 November 2004. Segal, G. and E.Montague (2005). Competitive Contracting for Highway Maintenance: Lessons Learned from National Experience. Washington Policy Center, Seattle, WA, USA, 2004. www.wips.org/Transportation/PBMontagueSegalReasonHighway Maintenance.htm, accessed 15 November, 2005. SRA (2005). Swedish Road Administration. Presentation at VTI Annual Convention, Linköping Sweden, 12 January 2005. SRA (2007). Swedish Road Administration. www.vv.se /filer/37257/trafikantbetyg_vin ter_2006.pdf, pages 9 and 12, accessed 10 February 2007. Stake, R. (1995). The Art of Case Study Research. Sage Publications, Thousand Oaks, CA, USA. Stenbeck, T. (2004). Incentives to Innovations in Road and Rail Maintenance and Operations. Licentiate thesis, Urban Planning, KTH Royal Institute of Technology, Stockholm. Stenbeck, T. (2006). Towards an Industry of Global Competitiveness. Paper presented at the TRA, Transportation Research Arena, conference in Göteborg 14 June 2006, Session 10.07.2. Stokes, D. E. (1997). Pasteur's Quadrant: Basic Science and Technological Innovation, Brookings Institution Press, Washington, D.C. Swedish Public Studies (2002). Skärpning gubbar! Om konkurrensen, kvaliteteten, kostnaderna och kompetensen i byggsektorn. SOU 2002:115 (Competition, Quality, Costs and Competence in Construction). Swedish Government, Stockholm. Yin, R. (2003). Case Study Research. Sage Publications, Thousand Oaks, CA, USA.

Swedish Summary – Sammanfattning Främjande av teknikutveckling inom väg- och banunderhåll - Incitament, upphandling och resultatbaserade kontrakt TORBJÖRN STENBECK * * CBI - Cement och Betong Institutet SE-100 44, Stockholm, [email protected].

Sveriges Byggindustriers Utvecklingsfond, Vägverket, Banverket, Centrum för Drift och Underhåll och Kungl Tekniska Högskolan tog 2002 - 2003 initiativ till ett doktorandprojekt syftande till att analysera och höja utvecklingstakten inom drift- och underhåll av vägar och järnvägar. Sammanläggningsavhandlingen består av åtta artiklar, varav den första sammanfattar licentiatavhandlingen (2004). Övriga sju artiklar syftar till att kvantifiera vinsterna med konkurrensutsättning och resultbaserade kontrakt. Översatt till svenska, har de åtta artiklarna följande titlar: 1. Effekter av konkurrensutsättning och resultatbaserade kontrakt på innovationer. Publicerad i Transportation Research Record september 2006, 1348, pp. 3–8. 2. Livscykelkostnadsmodell för broar. Publicerad på amerikanska TRBs årliga transportkonferens CD-ROM 2007, Paper 07-0110, Transportation Research Board, Washington D.C. 3. Budgetera resultatbaserat vinterunderhåll – snös påverkan på vägars drift- och underhållskostnad. Inskickad till ASCE Journal of Infrastructure Systems 27/2 2007. 4. Fallstudie av funktionsentreprenad för väg. Inskickad till Journal of Performance of Constructed Facilities 27/2 2007. 5. Kvalitets- och livscykelkostnader i resultatbaserade kontrakt – flerfallsstudie. Inskickad till Journal of Performance of Constructed Facilities 27/2 2007. 6. Incitament, resultat och möjliga framgångsfaktorer för resultatbaserade kontrakt för järnvägsunderhåll – fallstudie. Inskickad till Construction Management and Economics 19/1 2007. 7. Kvantifiering av effekter av incitement i ett resultatbaserat kontrakt för järnvägsunderhåll. Inskickad till Leadership and Management in Engineering 27/2 2007. 8. Vägars underhållskostnad och konkurrensutsättningspolicy Inskickad till Evaluation and Program Planning 27/12 2007.

1

Swedish Summary – Sammanfattning

1 Introduktion Behov av förtydligade och helt nya ord och begrepp uppstår då forskningsfronten avancerar. Detta visade sig gälla även för analysen av innovationsprocessen i anläggningsbranschen. Det flitigt använda, men ändå något vaga, engelska ordet performance-based contract har översatts till resultatbaserade kontrakt vilket inte är helt synonymt med det i Sverige flitigt använda ordet funktionskontrakt. En viktig avsikt med funktionskontrakt har varit att öka entreprenörens frihet och därmed utrymmet för innovationer. Rangordnade på en skala från det mest tekniska perspektivet (entreprenören, underentreprenörens) till det minst tekniska användarperspektivet (trafikantens, skattebetalarens, politikerns) kan de svenska begreppen i den här studien inordnas enligt Fig. 1. Expert

Utförande-

Resultatkontrakt

Funktions-

Användare

Fig. 1 Kontraktsformerna har i denna studie rangordnats från ett expertperspektiv till ett användarperspektiv för att söka optimum ur innovationssynpunkt. Funktionsentreprenad är en synonym för funktionskontrakt. Totalentreprenad används i Sverige i regel att betyda projektering-byggande, och kan ses som ett resultatkontrakt, eftersom det är det färdiga byggnadsverket och dess egenskaper som ofta specificeras i kontraktet, inte hur det ska åstadkommas. En annan viktig definition för studien är vad som krävs av en ny produkt, process eller organisation för att kallas innovation. Även här finns en skala att vara medveten om: från en liberal inställning där varje förändring, uppfinning eller idé är att betrakta som innovation till den mer krävande, där det krävs att innovationen måste ge ett samhälleligt mervärde. Med kvoten K1K2/K3 > 0 med bivillkoret K2 ≥ 0 åsyftas att kvantiteten×kvaliteten/kostnaden måste öka samtidigt som kvaliteten inte får sjunka. Kostnaden avser reala resurser, såsom åtgången av arbetstid och material, varför flyttning av produktionen till billigare länder inte påverkar K3 med sin nominella besparing, utan bara av sin reala besparing, vilken i de flesta fall torde vara negativ.

1.1 Bakgrund Drift och underhåll av Vägverkets vägnät konkurrensutsattes gradvis under 1990-talet, och sedan 2001 har även Banverket inlett en motsvarande process. Konkurrensutsättningen har kombinerats med resultatorienterade kontrakt som gradvis ersätter löpande räkning och liknande kontrakt baserade på åtgången av insatsvaror såsom material och arbetstid. Mätning av teknikutveckling må vara en vetenskaplig utmaning i sig och även om KK/K tillåts vara svaret på den återstår utmaningen att mäta och kvantifiera kvaliteten. I den intervjubaserade studie som redovisas i artikel 1 och licentiatavhandlingen uppmuntrades de intervjuade att använda en okritisk definition på teknikutveckling och innovation. Detta var en medveten del av metoden, dels för att bibehålla öppenhet för olika definitioner på innovation, dels för att uppmuntra till att även mindre innovationer och framsteg framfördes som exempel.

2

T. Stenbeck

1.2 Syfte och forskningsfråga Huvudsyftet, Hur främja teknikutveckling inom drift och underhåll, uppdelades i tre delfrågor: 1. 2. 3. 4.

Hur har teknikutveckling stimulerats och vad blev erfarenheten? Hur mycket innovation gav konkurrensutsättning? Hur mycket innovation gav resultatorienterade kontrakt? Hur kan kostnadsmodeller tas fram och användas innovationsfrämjande?

Frågornas relation till varandra och omgivningen sammanfattas i Fig. 2. Hypoteser Idéer Uppfinningar

Bekräftad kunskap Innovationer

Innovationsstimulantia

Konkurrensutsättning

Bevisat vara rätt Bevisat vara fel

Resultatbaserade kontrakt

Antaget att vara korrekt Antaget att vara fel

Kunskaps fronten

Fig. 2 Fågelperspektiv på forskning i allmänhet och denna studie i synnerhet.

2 Metod Kvalitativa metoder, i form av intervjuer, experiment, seminarium och deltagande observation användes i de första studierna 2003-2004, medan kvantitativa metoder i form av regressionsanalys använts 2005-2006. Då det varit mycket forskning för att få fram vart enskilt data till regressionsanalyserna, ibland med fallstudiemetodik, har regressionerna i regel genomförts med färre observationer än vad den statistiska vetenskapen rekommenderar. Rekommenderat minimiantal är 5(k+2) där k står för antalet prediktorer (MIT 2003). Denna utmaning nedåt av statistikens och matematikens gränsläge genererade intressanta effekter som diskuteras i artikel 3, artikel 8 och Stenbeck & Silfwerbrand (2006).

3 Resultat och analys Under 1980-talets andra hälft ökade kostnaderna för drift och underhåll av Vägverkets vägar mer än inflationen. Denna utveckling stoppades då konkurrensutsättning infördes, Fig. 3.

3


Fig. 3 Vägverkets kostnader för drift och underhåll i fasta och löpande priser (Källa: Österberg 2003). Arnek (2002) kvantifierade prisfallet första hälften av 1990-talet till 22–28 %. Antaget att utvecklingen i fasta priser hade fortsatt enligt trendlinjen som tillfogats Österbergs diagram i Fig. 3, skulle besparingen år 2002 kunna uppskattas till cirka 50 % (130/88 avläst ur Fig. 3) Kvantifierat enligt Vägverkets statistik över kundnöjdhet sjönk kvaliteten 38 % (1-39/62) enligt yrkeschaufförerna och 21 % (1-57/72) enligt privatbilisterna (Fig. 4) fram till 2002. Vägverkets beställare håller dock inte med om denna kvalitetsbeskrivning utan anser att kvaliteten tekniskt sett snarare ökat en aning (Liljegren 2003). Åren 2002 till 2006 har kundnöjdheten som synes öka något igen (Fig. 4) men så har också kostnaderna (personlig kommunikation). Kundnöjdhet - Trafikantbetyg

Mycket+ganska nöjda, %

80 70

72 74 62

60

71

69 64

61 54

59

57

40

39

62

61

52

50

46

40

45

40

30 20

Privatbilister

10

Yrkeschaufförer 2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

0

År

Fig. 4 Trafikantbetyget sjönk gradvis åren 1995 – 2002 för att sedan öka något 2002 -2006. Källa: Vägverket www.vv.se/filer/37257/trafikantbetyg_vinter_2006.pdf, s. 9 och 12 (2007-02-10) I Nordamerika råder oenighet kring de ekonomiska vinsterna av konkurrensutsättning även på kort sikt även utan hänsyn tagen till kvalitets- och teknikutveckling (Riberau 2004, Segal & Montague 2004).

4

T. Stenbeck

Lövmar (2000) och ett stort antal amerikanska fallstudier redovisar nöjda beställare och mycket nöjda entreprenörer under och strax efter hus- och vägprojekt som upphandlats och genomförts som funktions- eller totalentreprenader. På byggnadssidan finns även omfattande kvantitativa studier där leveranstid, budgethållning och kundnytta får högre värden i totalentreprenader än utförandeentreprenader (Sanvido & Konchar 1999). Ingen av dessa tycks dock ha undersökt de byggda objekten efter en längre tids nyttjande. En rapport finns som redovisar sämre kvalitet efter sju år (anonymiserad i denna avhandling för att uppmuntra även andra att rapportera uppriktigt om resultatbaserade kontrakt). Oron för att de kortare leveranstiderna och entreprenörens ökade frihet skulle kunna leda till lägre kvalitet och högre underhållskostnader finns därmed inte vederlagd utan i ett mycket trovärdigt och tekniskt förklarat fall tvärtom bekräftad. I de amerikanska studierna kan noteras att de flesta inte påstår att objektet blev billigare, bara att det höll sin budget bättre. Kvaliteten bedöms inte av tredje part utan av dem som själva just färdigställt objektet. Några artikelförfattare kommenterar särskilt att de inte kunnat bedöma kvalitet och livscykelkostnad. Slutsatsen av de amerikanska studierna är därför att totalentreprenader sparar tid och konflikter mellan beställare och utförare, men däremot inte nödvändigtvis kostnader (Professor John Schaufelberger, University of Washington, personlig kommunikation). Den fråga man kan ställa sig är om det verkligen är möjligt att ingen undersökt kvaliteten. Kan det månne vara så att svaret är negativt och att ingen vågat, eller ens kunnat (forskarna är ju beroende av tidskriftsredaktörer och andra forskare för sin publicering) utmana de positiva, men forskningsmetodiskt tveksamma, artiklarna genom att framföra det? För att fylla tomrummet genomfördes en flerfallstudie med fyra fall av total- eller funktionsentreprenader. Frånvaron av tidigare studier, ett relativt litet antal existerande fall och de komplexa aspekterna av kvalitet motiverade val av fallstudiemetodik.

3.1 Hur stimulera teknikutveckling Ett antal metoder för att stimulera teknikutveckling framfördes under intervjuerna och i ett seminarium kring hinder för teknikutveckling som genomfördes inom projektets ram (artikel 1). Några metoder visade sig ha testats tidigare i Sverige eller utomlands, och andra testades inom ramen för forskningsprojektet (sidoanbud). Nedan sammanfattas ett antal innovationsskapande metoder följt av de erfarenheter projektet funnit 1 . Egen primärdata med avseende på konkurrensutsättning och resultatbaserade kontrakt följer i avsnitten 3.2 och 3.3. 1. Konkurrensutsättning I och med konkurrensutsättningen av Grundpaket Drift av Vägverkets vägar föll priserna och höll sig låga i tio år. Sedan 2002 har en viss ökning skett och beställaren oroar sig för att antalet anbudsgivare sjunker. Oenighet råder om konkurrensutsättningens effekter på kvalitetsutvecklingen (Liljegren 2003). Teknikutvecklingen verkar ha tagit skada (Österberg 2003, Stenbeck 2004). Frankrike, Tyskland och de flesta nordamerikanska stater och provinser, konkurrensutsätter ej sin drift och sitt underhåll. I Nordamerika råder oenighet huruvida priser och kvalitet vunnit eller tagit skada. Ansvariga i två av fem stater/provinser, British Columbia (personlig kommunikation dec 2004) och Texas (personlig kommunikation jan 2007), som var pionjärer med konkurrensutsättning (Riberau 2004) vittnar om kvalitetsproblem och tonar ned vinsterna och förväntningarna. Effekter på teknikutveckling är relativt lite beforskat, för att inte säga outforskat, i Nordamerika. 1

S = Sverige, C = Kanada, U = USA, F = Frankrike


5

2. Resultatbaserade kontrakt Resultatbaserade kontrakt har varit populärt, särskilt bland entreprenörer, sedan konkurrensutsättningen infördes i början av 1990-talet och ännu tidigare (Grennberg 1968). Mot denna bakgrund kan viss frånvaro av forskningsrapporter som ex post bekräftar de positiva förväntningarna noteras. 10-15 år har nu gått sedan de första lanserades. Samma okritiska entusiasm anas när allt fler världen över, att döma av TRBkonferensen 2007 i Washington D.C., kastar sig över de resultatbaserade kontrakten som det som ska lösa alla ekonomiska, tekniska och politiska problem. 3. Mjuka parametrar Miljö och andra hänsyn som beställarna, på samhällets uppdrag, vill att entreprenörerna ska ta har gjorts till en del av de svenska kontrakten, från 30 % till att börja med till 3 % idag. Kritikerna har menat att bedömningen av dessa faktorer är godtycklig. Allteftersom beställarna minskat %-andelen mjuka parametrar och likriktat bedömingarna har kriterierna blivit papperstigrar som alla anbudsgivare uppfyller utan att någon förändring sker i verkligheten. 4. Sidoanbud I och med införandet av Lagen om Offentlig Upphandling (LoU) och den transparens och de möjligheter denna medfört till att överklaga beställarens bedömningar, har entreprenörernas möjligheter att få sidoanbud, anbud som inte exakt följer förfrågningsunderlagen, godkända minskat. Ytterligare bekräftat av det experiment som genomfördes 2003-2004 (Stenbeck 2004) ser den vägen till teknikutveckling för närvarande ut att vara stängd. 5. Idé-brevlådor Allt färre bidrag hittar vägen till de idébrevlådor som Vägverket Produktion och Washington State Department of Transportation satt upp. Det är oklart om det beror på incitament, samhällstrender eller något annat. Låg kostnad för stimulansåtgärden talar för att ändå fortsätta med den. 6. Teknikutvecklingstävlingar Teknikutvecklingstävlingar lider av samma problem som sidoanbuden, hur man ska jämföra äpplen med päron på ett sätt som är förenligt med LoU. Det försvårar möjligheterna att genomföra en tävling externt, i synnerhet av en offentlig myndighet, men franska järnvägen SNCF har internt genomfört en tekniktävling och fick ett 80-tal bidrag, varav ett 20-tal bedömdes intressanta att gå vidare med. Relativt låg kostnad och reklam för teknikutveckling som bieffekt talar för att stimulansåtgärden ändå kan vara värd att pröva. 7. Öronmärkta pengar Genom att anslå ett belopp som ska gå till teknikutveckling kan staten eller en myndighet se till att ett minimum av forskning sker, oberoende av dess lönsamhet. Denna studie stöder att ett visst sådant stöd fortsätter, som övergångslösning eller permanent, tills annan teknikutvecklingsstimulans visat sig fungera.

6

T. Stenbeck

8. Samarbeten Utökat samarbete mellan beställare och utförare lovordas av båda parter, men det bör observeras att den tredje parten, skattebetalaren, inte är närvarande. Om beställare och utförare bildar nära partnerskap behövs en ny part att representera både den nuvarande och den framtida skattebetalaren. Konkurrensutsättningens fördelar är svåra att förena med samarbetets fördelar, men kanske finns en balansgång. I artikel 1 och licentiatavhandlingen tilläts tyckanden komma till uttryck utan att ifrågasättas närmare, medan artiklarna 2-8 mer kritiskt och matematiskt kvantifierar vinsterna med konkurrensutsättning och resultatbaserade kontrakt på ett så vetenskapligt sätt som möjligt utifrån en begränsad mängd data. Utifrån den nationalekonomiska litteraturen (Salter 1969, Jones 1976) definierades och förenklades teknikutveckling till Innovation = K1K2/K3

(1)

(där K1 står för kvantitet, K2 för kvalitet och K3 för kostnad), med bivillkoret att K2, kvaliteten, aldrig får sjunka 2 . K3, kostnaden ska inte kunna nedbringas genom att flytta produktionen till andra länder eller sänka lönerna i egna landet, utan bara genom att mindre reala resurser (råvaruvolymer, arbetstid) åtgår till produktion av samma mängd som tidigare. Med denna definition blir den nordamerikanska forskningens kostnadsfokusering (medan den europeiska även har med en mer social dimension) inte motstridig med teknikutveckling. Vid lika löner, materialkostnader och kvalitet, blir lägre kostnader ett mått på teknikutveckling. Vid lika kostnad blir kvalitetsförbättringen ett mått på teknikutvecklingen. För att räkna ut kvalitetsförbättringen krävs dock kvantifiering av kvalitet, vilket kan vara svårare att mäta än kostnaderna vid bibehållen kvalitet. Dessa vetenskapliga utmaningar, kvantifiering av kvaliteter, gjordes till en del av forskningsprojektet, eftersom ett ytterligare tyckande utan kvantifiering till den flora av sådana som redan publicerats ansågs överflödigt.

3.2 Konkurrensutsättning Arnek (2002) har redan lämnat ett kvantifierat bidrag vad gäller kostnadsutvecklingen i Sverige, vars storleksordning överrensstämmer med Österbergs siffror, (Fig. 3). Liljegren (2003) tog sig an det som i den här avhandlingen kallas K2 och kom fram till att den (kvaliteten) varit konstant. Hon undersökte också hur mycket beställarsidans kostnader ökat med konkurrensutsättningen, dvs förtydligade Arneks svar på K3 (kostnadsresultatet). Med viss reservation men ändå, kom hon fram till att beställarsidans kostnader inte ökat med mer än 5 %. Det värsta fallet av denna siffra, alltså 5 %, räcker inte till för att förklara (reducera till noll) Arneks funna vinst (minst 22 %), eller de 50 % som trendlinjen i Fig. 3 indikerar jämfört med de faktiska kostnaderna 2002. Det kan finnas anledning att fördjupa dessa studier, eftersom motsägelserna mellan trafikanternas och beställarnas bedömning inte utretts övertygande, exempelvis genom jämförelser av vägarnas faktiska, tekniska tillstånd (en tillämpning av skillnaden mellan olika perspektiv och deras konsekvenser som åskådliggörs i Fig. 1). Den studien inkluderades dock ej i denna avhandling (och är inte heller gjord), utan Arneks, Liljegrens och Österbergs studier fick förbli och bestå, som representanter för Sveriges erfarenheter av konkurrensutsättning och dess effekter. Istället användes Nordamerika som forskningsfält. Regressionsanalys genomfördes av hur kostnaderna i olika kanadensiska provinser, Sverige och staten Washington beror på hur 2

Enligt ett Pareto (2007) - baserat synsätt att inget ska förlora på utveckling

7


vägarnas väglängd mäts, väder, trafik och konkurrensutsättning kontra egen regi. En modell härleddes i två steg för att neutralisera alla effekter utom konkurrensutsättningsfaktorn. Dessutom genomfördes besök och intervjuer i Kanada, vilkas resultat också redovisas i artikel 8. Kvalitet och teknikutvecklingseffekter utreddes i dessa samtal. Beställare i egen-regiprovinserna hade fler exempel på innovationer, men det skulle kunna bero på att de hålls bättre informerade och behöver inte bero på att teknikutvecklingstakten i sig är högre. Det kunde noteras att beställare i konkurrensutsatta provinser lämnade ut mindre information och en, Ontario, kunde efter att ha fått veta att de skulle bli jämförda med andra provinser, plötsligt inte ta emot besök alls. Även detta skulle dock möjligen kunna bero på den mindre kunskap om detaljerna som outsourcing leder till eller högre intern arbetsbelastning och behöver inte bero på att siffror och information inte skulle tåla insyn och jämförelse med andra provinsers kostnader. Med stor felmarginal pga begränsad data verkade kvalitet och teknikutveckling ungefär lika över Kanada, varför dessa faktorer slutligen sattes till 1/1 tills mer metodiskt korrekta studier finner annat i den variant av KK/K som kan uttryckas i följande: $ / kmår 1 $ / kmår 0

×

Kvalitet år 0 Innovationstakt år 0 × Kvalitet år 1 Innovationstakt år 1

(2)

Formeln är överdrivet byråkratisk i det här fallet, eftersom andra och tredje faktorn sattes till 1/1 resp 1/1 till slut. Den står ändå med för att bidra med ett exempel på hur teknikutveckling skulle kunna mätas mer noggrant i framtida studier, nämligen som en kostnadsförändring korrigerad med kvalitetsförändring och innovationstaktsförändring. Efter att ha isolerat ut konkurrensutsättningsfaktorn hade de konkurrensutsatta provinserna i genomsnitt 26 % högre kostnader än egen-regi-provinserna. Kostnaderna som analyserats i artikel 8 är relativt säkra, men felkällor som kan vara stora är kvaliteten, kvalitetsförändringen och trafikens effekter. Förlusten med konkurrensutsättning kan vara både högre och lägre efter att dessa klarlagts bättre, både i Kanada och i Sverige.

3.3 Resultatbaserade kontrakt Vad gäller ex post effekterna av resultatbaserade kontrakt i väg- och järnvägsrelaterade projekt, så fanns inte hundratals fall att välja på. Fyra valdes ut på praktiska grunder (tillgång till data tack vare personliga kontakter). Två av dessa kan förväntas vara skeva i förhållande till populationen i positiv riktning, då kontakterna som tillhandahöll dem får anses vara positiva till konkurrensutsättning och resultatkontrakt. Övriga två tillhandahölls av vetenskapen (en forskarkollega, kritisk till det mesta) och ett återfanns i litteraturen på ett sätt och i ett sammanhang som också det kan bedömas leda till neutralitet vad gäller skevhet. Likväl består stickprovet av enbart fyra fall, varför långtgående slutsatser ej bör dras. De slutsatser som kan och bör dras är kvalitativa, t.ex. vilka åtgärder som skulle kunna förbättra pris, kvalitet och teknikutveckling baserat på de fyra fallen. Det framkalkylerade mätvärdet, den genomsnittliga teknikutvecklingsprocent som de fyra fallen ger upphov till, bör tas med en nypa salt, men ändå tillåtas gälla tills framtida forskning visar annat. Två av fallen är anonymiserade för att undvika sökande efter syndabockar, vilket inte har något vetenskapligt värde. Det är mycket tack vare de s.k. syndabockarna som fallen har tillhandahållits vetenskapen att dra lärdom av.

8

T. Stenbeck

Fall 1 Första fallet är ett kombinationskontrakt Projektering-Byggande-Underhåll, där underhållstiden först var satt till sju år, men sen förlängdes till tio år. Låt oss kalla det för fall 1. I enlighet med kontraktsvillkoren hade entreprenören frihet att projektera vägen efter eget huvud, så länge ställda funktionskrav var uppfyllda. Beställaren bifogade en projektering som underlag för anbuden, men denna var inte bindande för entreprenören. Alla anbudsgivare utom en utgick ändå från beställarens projektering. Vinnande entreprenör ifrågasatte nödvändigheten av de kalkcementpelare som var projekterade och kunde vinna anbudstävlingen med 15 % marginal genom att ta bort dem. Beställaren förtydligade ansvarsfrågan bl a genom att förlänga underhållstiden till tio år och det förlängda ansvar och garanti detta medförde för entreprenören. Vidare insisterade beställaren på kalkcementpelare utefter den mest känsliga delen av sträckningen, inte pga jordförhållandena utan för att en skevt sjunkande väg utefter denna bit skulle kunna innebära stor fara för säkerheten pga kurvans dosering, hög vägbank och bro med förhöjd risk för halka. Vid inspektionen tio år senare är det endast denna del som uppfyller funktionskraven. Utefter övriga delar återfinns sättningar på upp till tre gånger funktionskravet. Med glimten i ögat, låt oss sammanfatta detta till 1-0 för beställarstyrda kontrakt. Väghållaren har i detta fall hittills generöst bjudit på avvikelsen, med de signaler till branschen detta medför. Den positiva signalen är att väghållaren är öppen för andra lösningar än sin egen och är beredd att låta entreprenören få fullfölja dem. Den negativa är att lösningen i detta fall inte var kvalitetshöjande och därför inte teknikutvecklande enligt K1K2/K3 med sidovillkoret ∆ K1 ≥ 0 som definierats att krävas för att kallas paretooptimal teknikutveckling. Den största skadan ligger dock i att entreprenören inte gavs möjlighet (dvs tvingades) göra rätt för sig, vilket ger de anbudsgivare vatten på sin kvarn som menar att ett orealistiskt underpris är det enda sättet att vinna en upphandling på. Kombinerat med underprestation utan konsekvenser blir underpris den enda möjliga överlevnadsstrategin och lönar sig i längden. En tänkbar medelväg vore att entreprenören endast anmodas att åtgärda det ställe där funktionskravet underskreds med stor marginal och där en faktisk olycksrisk finns pga närhet till bro med dess tillhörande förhöjda halkrisk (personlig kommunikation). Kontraktet ger beställaren rätt att kräva att entreprenören står för hela åtgärden (t.ex utför den utan ersättning). Enligt grova kostnadsberäkningar skulle en sådan åtgärd kosta högst hälften, mer troligt en fjärdedel, av den initiala 15%-iga vinsten. Genom att ta över anläggningen utan krav på åtgärd effektuerar beställaren i praktiken ett slags vinstdelning av den initiala vinsten, som med tillägg av generositet och god vilja kan försvaras. För samhället kan en vinst på 15 % – åtgärdskostnaden bokföras som nettovinst, dvs. 7–11 % under ovanstående antaganden. Låt oss generöst avrunda till 10 %. Teknikutvecklingsvinsten är mindre, men en viss teknikutvecklingsvinst finns i att man nu fått ytterligare empirisk erfarenhet av kalkcementpelares möjligheter och begränsningar. Med mycket generositet, ännu en gång, kanske den vinsten kan kompensera kostnaden för den felaktiga premieringen av underpris+underkvalitet och de signaler till branschen detta gav.

Fall 2 Det andra fallet är det mer klassiska kombinationskontraktet projektering-byggande, ofta benämnt Totalentreprenad i Sverige. Utfallet, slutkostnaden, av detta fall jämfördes med ett referensalternativ på ett annat sätt än fall 1, med en närliggande, liknande trafikplats. Låt oss kalla den trafikplatsen för fall 2b. Fall 2b upphandlades på traditionellt sätt, som en utförandeentreprenad, dvs med beställarrepresentanter som översta auktoritet på byggplatsen. Fall 2a och 2b var båda beställarbudgeterade till 180 miljoner (valutan ointressant och siffran förvrängd som en del av anonymiseringen) baserat på kostnadsberäkningen av respektive


9

projektör (eventuellt samma, har ej utretts närmare, då jämförelsen med budget inte tilldelats större betydelse i den analys som kommer). Okulärt bedömt är fall 2b ca 25 % större än fall 2a. I fall 2b hade entreprenören givetvis (eftersom det var en utförandeentreprenad) en projektering tillhandahållen av beställaren att utgå ifrån. Detta gällde även för fall 2a, men anbudsgivarna hade liksom i fall 1 rätt att frångå den, vilket den vinnande anbudsgivaren gjorde. I detta fall finns ännu inte det långsiktigt hållbara resultatet, eftersom bara ett par år gått sedan bygget färdigställts, men liksom i fall 1 vore det tjänstefel av anbudsgivaren att överlämna en konstruktion dyrare än nödvändigt. Med detta incitament i beaktande finns fog att anta att det levererade objektets livscykelkostnad högst kommer att tangera, förmodligen underskrida, den förprojekterade lösningens. Kopplingen incitament-kvalitet-underhållsbehov-livscykelkostnad motiverar fallets inkludering i denna studie som överordnat handlar om incitament och transportinfrastrukturens underhåll. Med detta incitamentsrelaterade resonemang (plus erfarenheterna av beställares tolerans vunna av fall 1) som stöd kan man anta att en noggrann teknisk jämförelse av den resulterande kvaliteten i den vinnande entreprenörens konstruktion på sin höjd kommer att tangera beställarens projekterade kvalitet. Då fall 2a var färdigbyggt, godkänt och överlämnat, översteg kostnaden budget med 50 % (ovanligt mycket för en totalentreprenad, överrensstämmelse med budget anses som en av totalentreprenadens fördelar, detta konstaterande dock inom parentes eftersom det är en bisak i kommande analys). Fall 2b översteg sin budget med 20 % (relativt normalt för en utförandeentreprenad, till och med i underkant i det land detta exempel är hämtat ifrån. Prishöjningarna under resans gång beror ofta på fel i ritningarna eller att entreprenören övertygar beställaren om en smartare, något dyrare, lösning). Med dessa 20 % multiplicerat med storleksskillnaden 25 % kostade totalentreprenaden 50 % mer än den jämförbara utförandeentreprenaden vid byggets slut. Med incitamenten i beaktande finns ingen anledning att anta att denna fördyring kommer att inhämtas med tiden, tvärtom. Farhågan ökar ytterligare av att beställaren ”inte tagit med livscykelkostnaden i bedömningen” (enligt telefonsamtal 7 mars 2006). Med glimten i ögat kan vi konstatera 2-0 till beställarstyrda entreprenader efter att fall 1 och 2 studerats och analyserats (på den detaljnivå som redovisats). Ytterligare fakta kan rucka slutsatserna av fall 2, men knappast fall 1. En felkälla för jämförelsen i fall 2 är grundförhållandena vid de två trafikplatserna. Risken förknippad med dessa brukar inte vara inkluderade i Totalentreprenaderna (i det landet) och kan uppgå till 25 % av hela byggets kostnad, vilket därmed skulle kunna förklara beställare-projektörens likabudgetering av de två trafikplatserna trots deras storleksskillnad. Storleksskillnaden försvinner som faktor om man antar att beställar-projekteringarnas prisberäkning är gjord lika i båda fallen. I så fall blev Totalentreprenaden bara 20 % dyrare än utförandeentreprenaden (enligt beställarens angivna siffor, 270 miljoner resp 220 miljoner) och exakt lika dyr (270 milj resp 270 milj) enligt inofficiella källor. 2-0 med stolpe in, i så fall, antaget den för entreprenören gynnsammaste tolkningen av fall 2. Förutom projektörernas lika-budgetering har dock inget framkommit som tyder på att grundförhållandena var olika i de två fallen.

Fall 3 Fall 3 är en biprodukt av en annan studie som återfanns i litteraturen. Tio totalentreprenader utefter en motorväg i Storbritannien kontrakterades med olika kontraktsformer (Langford et al 2003). Studien konstaterade, att fastpriskontrakt blev billigare än reglerbara mängder. Tio av de elva projekten var utförandeentreprenader. Studien nämner att det elfte projektet var en

10

T. Stenbeck

totalentreprenad, som var dyrare än genomsnittet av de två andra kategorierna men avdramatiserar detta utan att ge någon motivering. Baserat på frånvaron av motivering kickar fallet in 3-0 till beställarstyrda kontraktsformer, dock ej med helt klockren träff.

Fall 4 Fall 4 tillhandahölls för studium av Banverket. Banverket Produktion lämnade ett fastpris för det mesta av underhållet av ett antal bandelar under två år, med reservation för extrema väderförhållanden, i första hand vinterrelaterade, att regleras efter löpande räkning för det fall det understeg eller översteg vissa trösklar (vid understigande en viss återbetalning till beställaren, vid överstigande ett visst tillägg). Utöver detta, det intressanta med detta kontrakt, var dock en incitamentsformel. Om slutprestationen till resenärerna, antalet förseningsminuter och vibrationen i tågen, minskade skulle bonus utgå. Järnvägsanläggningens långsiktiga standard, mätt som minskning av antalet fel per tidsenhet, blev en tredje parameter som ingick i incitamentsformeln. Artiklarna 4 och 5 handlar om detta och deras motsvarighet på svenska presenterades på VTI Transportforum 2007. Slutresultatet var att kvaliteten höjdes med 10 % mätt som förseningsminuter, 20 % mätt som tekniska fel. Vibrationsdelen av incitamentet visade sig otydligt formulerat och kommer sannolikt att utgå ur den slutliga regleringen mellan parterna. Felkällor kan finnas i tekniska fel med längre planeringshorisont än kontraktets löptid ignorerats, och att de positiva effekterna i längden inte är hållbara. I takt med att kontraktsformen tillämpas i ett kontrakt med privat sektor kommer det visa sig om det positiva resultatet är generaliserbart även till kontrakt med externa parter, eller bara vid kontrakt med den offentliga entreprenören. Tills vidare kan vi dock med glädje konstatera att ställningen reducerats till 3-1 med ett relativt övertygande skott.

Sammanvägning av fallen 1 - 4 Sammanvägning av de kvantifierade vinsterna/förlusterna med lika vikt för fallen 1, 2 och 4 resulterar i (10 -20 + 15)/3 = 1,67 % vinst med resultatbaserade kontrakt. Fall 2 belastar dock medelvärdet med ett osäkert men negativt belopp liksom den urvalsmetods som ledde fram till fallen 1-4. Den positiva bias som den bör ha medfört liksom den långtgående generositet som tillämpades i fall 1 för att försvara försökets positiva bidrag till teknikutvecklingen sänker medelvärdet. Med detta invägt är det mer troligt än otroligt att resultatbaserade kontraktens bidrag till teknikutvecklingen hittills varit mer negativt än positivt. Den historiska belastningen behöver dock inte betyda att även nästkommande års seriematcher är dömda att förloras. Med ökad medvetenhet om bristerna och ytterligare målmedveten träning har konkurrensutsättningen och resultatkontrakten framtiden för sig.

3.4 Kostnadsmodeller Artiklarna 2 och 3 är biprodukter som uppstått som ett led i att kvantifiera resultaten av konkurrensutsättning i Nordamerika. Artikel 2 är en modell som indirekt gör det möjligt att uppskatta brounderhållets andel av ett drift-paket och artikel 3 uppstod i försöken att utreda klimatets påverkan på drift- och underhållskostnaden. Dessa modeller användes dock inte till slut, i den kvantifiering av konkurrensutsättningens effekter i Kanada som redogörs för i artikel 8, men finns med i avhandlingen ändå, för att stödja metoder och antaganden i artikel 8. Modellerna som utvecklades för att kunna jämföra regioner med olika förutsättningar, kan bli användbara, eventuellt med viss modifikation, för jämförelse av olika årskostnader i framtiden, exempelvis vid sidoanbudsbedöming. Vintermodellen i artikel 3 kan bidra till lägre kostnader för beställaren tack vare att entreprenören lägger in en lägre riskpremie för vädret i


11

sitt anbud, i synnerhet i Nordamerika där vintermodeller inte varit en del i konkurrensutsättningen på samma sätt som i Sverige. Vintermodellen i artikel 3 är väsentligt enklare än den som praktiseras i Sverige. Om den visar sig stämma lika bra i Sverige som i Washington, finns mycket pengar, och konflikter mellan beställare och utförare, att spara på att låta den ersätta den nuvarande modellen, då väderinstitutets standard-väderstationer kan användas. Artikel 3 är tänkt att ge nya idéer kring hur den nuvarande svenska modellens träffsäkerhet kan kvantifieras. Man kan förvänta sig en diskussion kring det begränsade statistiska underlaget som använts, då n < 5(k+2), men artikelns huvudsyfte är att presentera ett förslag på metodik och inte att lägga fram en slutgiltig modell för just det studerade området (Washington).

4 Slutsats Vad gäller första forskningsfrågan, hur teknikutveckling kan stimuleras, så utreds i avsnitt 3.1 ett antal metoder som med begränsad sannolikhet till framgång men å andra sidan till liten kostnad, kan användas för att skapa teknikutveckling. (Artikel 1) Vad gäller andra forskningsfrågan, hur mycket konkurrensutsättning stimulerar teknikutveckling, så har konkurrensutsättningen sparat ungefär 50 % under sina tio första år i Sverige om man antar att den prisutveckling som rådde tidigare hade fortsatt. Å andra sidan visade sig konkurrensutsatta regioner ha i genomsnitt 26 % högre kostnader än de egen-regiregioner som ingick i en studie bestående av Sverige, nio kanadensiska provinser och staten Washington. (Artikel 8) Vad gäller tredje forskningsfrågan, hur mycket resultatbaserade kontrakt stimulerar teknikutveckling, så visade sig tre av fyra fall vid närmare studium vara antingen dyrare eller med sämre kvalitet (Artikel 4 och 5). Det positiva fallet karaktäriserades av att incitament tillfördes som komplement till ett ordinarie expertorienterat (Fig. 1) kontrakt, med endast ett fåtal slutanvändarorienterade, mätbara och tydliga parametrar som bonuspåverkande. Dessa mättes med kontinuerlig övervakning av (tredje part betald av) beställaren och ekonomisk reglering var direkt kopplat till utfallet (Artikel 6 och 7). Kostnadsmodeller för att korrigera för olika förutsättningar krävdes för att kunna göra jämförelserna mellan konkurrensutsatta och egen-regi-regioner respektive utförande- och resultatbaserade kontrakt. Svårigheterna att skapa kostnadsmodeller med flera olikartade svårkvantifierade variabler kan vara en förklaring till att forskningsresultaten tidigare blivit så ensidigt och okritiskt positiva till konkurrensutsättning och resultatbaserade kontrakt. (Artikel 2 och 3). En annan förklaring skulle kunna vara att forskningen blivit populariserad och politiserad. En debatt pågår i Nordamerika om att alltför få forskare vågar riskera sin akademiska karriär på att kritisera tidigare forskning. I synnerhet som risken är stor att det blir just dessa kritiserade kollegor som granskar och vars välvilja är förutsättningen för att de nya rönen ska bli publicerade.

Framtida forskning Uppföljning av Lövmars (2000) studieobjekt 10+ år efter byggets färdigställande och kvalitetsjämförelser av skilda regioners väg- och järnvägsstandard i både tid och rum har redan nämnts som intressanta framtida studier. Broars livscykelkostnad, artikel 2, blev mer teoretisk än det var tänkt från början, beroende på att kostnaderna inte redovisas tydligt och enhetligt i Vägverkets BatMan-system. Löpande, bättre, mer standardiserad och mer ärlig dokumentering-av alla kostnader och inspektionsresultat bäddar för bättre forskning och kunskapsinhämtning för de direkt inblandade såväl som för de externa institutioner som vill

12

T. Stenbeck

hjälpa till. Data är en kunskapskälla och förutsättning för innovation att lägga till övriga föreslag på hur innovationstakten i byggbranschen kan ökas. Om branschen prövar några eller alla av förslagen i avsnitt 3.1 (ytterligare några finns i den engelska avhandlingen) uppstår intressanta forskningsstudier kring vart och ett som transportforskning som stora delar av världen kommer vara intresserade att ta del av.

5 Slutord Trots det i denna studie konstaterade negativa utfallet hittills för konkurrensutsättning resultatbaserade kontrakt, hoppas jag att branschen fortsätter i den nyliberala riktningen. Allt är relativt och det mesta kan vändas i något positivt. Om en mutation på tusen är positiv, och en uppfinning på tio en innovation, så är ett lyckat fall av fyra bättre än förväntat. Misstag är den handlingskraftiges privilegium, lär Ingvar Kamprad ha sagt. En förutsättning för utveckling är att även misstagen kan redovisas på ett öppet sätt, så att lärdomar kan dras. Varför har inte funktionsentreprenaderna erövrat världen för länge sen? undrar paneldebattör Susanne Lindh, regiondirektör för Vägverket Stockholm på Upphandling i fokus den 4 oktober 2006. Ett svar är att om misstagen hade delgetts och tagits emot på ett positivt sätt för 10-20 år sen hade vi idag inte varit kvar lika nära ruta ett. Många entreprenörer värjer sig mot kontroll. De många amerikanska studierna som nämnts menar att totalentreprenader är bra för att beställaren slipper kontrollera. Jag håller inte med det. Tvärtom, rent teoretiskt borde mer kontroll behövas i resultat- och funktionskontrakt, eftersom det finns naturliga ekonomiska incitament i sådana kontrakt för entreprenören att spara på allt som går. Om beställaren inte kontrollerar blir det relativt lätt att spara på kvaliteten, vilket är på beställarens och framtida nyttjares bekostnad. Detta är inte fallet i kontrakt på löpande räkning och reglerbara mängder. Där kommer entreprenören inte spara på något, vilket kan bli ett problem för ekonomin, men inte för kvaliteten. Att ekonomiska slutregleringar mellan beställare och utförare upphört i och med införandet av resultatbaserade kontrakt är ett tecken på att något inte stämmer. Eftersom ett byggobjekt aldrig kan bli exakt enligt specifikationen, bör en reglering ske i ena eller andra riktningen i stort sett i alla kontrakt. Först när kontraktsuppfyllelse blivit självklart uppstår förutsättningarna för ytterligare smarta lösningar och innovation. Resultatkontrakt behöver förmodligen vara lika detaljerade, fast på ett annat sätt, som utförandekontrakten, för att rättssäkerhet ska råda, och ingen anledning finns att sänka kraven. Om målet är teknikutveckling så är det ointressant om projekterade kalkcementpelare behövs eller inte. Borttagande av projekterade kalkcementpelare innebär sänkning av vägens motståndskraft mot naturens överraskningar, och om beställaren vill gå med på det måste upphandlingen göras om. Resultatkontrakt får inte innebära att beställare och redan utvald anbudsgivare kommer överrens om att sänka kraven och dela på vinsten sinsemellan. Det är motsatsen till teknikutveckling. Den gamle, stränge, inspektörens auktoritära återintåg på arenan är välkommet och borde gå att kombinera med övriga innovationsstimulerande metoder; från bonus till entreprenörerna för överprestation och innovationer till tekniktävlingar och andra mer spektakulära och populistiska metoder. Förvisso förkväver den gamla sortens detaljerade förfrågningsunderlag i viss mån entreprenörernas kreativitet. Å andra sidan är entreprenörernas idéer inte alltid så fantastiska, särskilt inte ur samhällets perspektiv, som de kanske tror själva, så en viss skepsis är samhällsekonomiskt lönsamt, och det är bara beställaren som kan ta det ansvaret. Realistiskt är att några enstaka nya idéer och uppfinningar blir innovationer och det gäller att hitta och belöna dem. Lite till övriga som försökt är förmodligen väl investerat spill.


13

Vägverkets strategi att till 2010 ha 1/3 totalentreprenader (sagt på Upphandling i fokus 2006) är ett exempel på en gradvis utveckling från utförande- till funktionskontrakt i den takt kompetens, politik och ekonomi medger som låter lovande. Förmodligen tar det ytterligare 10 år, men vägen är målet i ett innovativt klimat.

6 Källförteckning Lövmar, P-O (2000) Uppföljning av funktionsentreprenader. Förläggare okänd, genomförandetid ca år 2000. Internt dokument på Vägverket, Borlänge och KTH, Stockholm. Pareto, V. (1869) Principii Fondamentali della Teorie dell' Elasticità, Förläggare okänd. Stenbeck, T and Silfwerbrand, J. (2006). Scientific Method Selection. Submitted to Concrete International 4 October 2006. För källförteckning hänvisas i övrigt till den engelska avhandlingens kapitel 6