Risk management in road construction: The case of Sri Lanka

Risk is an unavoidable phenomenon in construction projects. Proper risk allocation in construction contracts has therefore come to assume prominence because risk identification and risk allocation have a clear bearing on risk handling decisions. The proper management of risks requires that they be identified and allocated in a well-defined manner. This can only be achieved if contracting parties comprehend their risk responsibilities, risk event conditions, and risk handling capabilities. This research aims at identifying the risk responsibilities of contractual parties in order to improve their risk handling strategies with regard to Sri Lankan road projects. Semi-structured interviews were used for the primary data collection. This was complemented with documentary evidence. The results show that road construction projects in Sri Lanka are exposed to many risk sources while most risks are borne by parties who were assigned with risks via contract clauses. However, parties not allocated with risks too happened to bear the consequences of such risks. Therefore, it is concluded that there is no one best way to respond to a risk and that different risk handling strategies should be adopted in order to deal effectively with risks.


BACK GROUND
Every human endeavor involves risk (Dey and Ogunlana, 2004;Poh and Tah, 2006). The success or failure therefore of any venture depends crucially on how we deal with it (Dey, 2001). The construction industry is more prone to risk and uncertainty than most other industries (Flanagan and Norman, 1993;Kim and Bajaj, 2000;Tah and Car, 2000), the element of uncertainty having to do with its inherent characteristics (Hayes et al., 1986;Bunni, 1997;Kangari and Riggs, 1989;Bing et al., 1999). But these risks are not always dealt with properly by the industry (Thomson and Perry, 1992;Mills, 2001).
According to Mills (2001), the productivity, performance, quality and cost of the project are affected by the risk. Edward and Bowen (1998) identifi ed risk management as an im-portant tool to cope with construction risks and to overcome above problems of a project. Dey (2002) also shows that there are many examples of non-achievement of time, cost and quality of projects due to the absence of risk management techniques in project management. Therefore, the success parameters of a construction project, namely, the timely completion, staying within the specifi ed budget, and achieving requisite performance would depend upon the capability of each party in risk management. Baker et al. (1999a) argued that risk management is also useful in maximizing profi ts. The construction industry however has been very slow in moving towards understanding the benefi ts of risk management (Flanagan and Norman, 1993;Raftery, 1994;and Ward et al., 1999).
The Road Development Authority (RDA) of Sri Lanka, due to the ever-increasing traffi c volume, is planning for the future development of a national highway network (RDA, 2006). Road projects however often confront many uncertainties due to factors such as the presence of interest groups, resource availability, the physical, economic and political environments, statutory regulations, etc. According to Wang and Chou (2003), such risks have a signifi cant effect on the outcome of a road construction process.
Proper risk allocation in construction contracts will reduce the impacts of adverse conditions, and increase effi ciency and effectiveness in management (Barnes, 1983;Abrahamson, 1984;Thompson and Perry, 1992;McCallum, 2000;Rahman and Kumaraswamy, 2002). Risk allocation upon risk handling of road projects in Sri Lanka has not been satisfactorily established because of different interpretations of risk allocation between owners and contractors. This research highlights the signifi cance of understanding proper risk allocation between contractual parties in Sri Lankan road projects. It aims at assisting Sri Lankan road contractors and employers to a) identify the risk sources inherent in road projects, b) understand their risk responsibilities, and c) improve their risk handling strategies so that they would optimize the scarce resources and enhance the socio-economic value of Sri Lankan road projects. Section 2 of this paper discusses the literature pertaining to risk management in construction highlighting risk identifi cation, risk allocation and risk handling. Section 3 gives the research methodology followed by results in section 4. Section 5 concludes the paper with a discussion on risk handling techniques to be followed in road projects.

THE LITERATURE ON RISK MANAGEMENT IN CONSTRUCTION
Bufaied (1987 cited in Akintoye and Macleod, 1997) has described risk in relation to construction as "a variable in the process of a construction project whose variation results in uncertainty as to the fi nal cost, duration and quality of the project". According to Dey (2001), such variation is due to the absence of risk management techniques in project management. Hence, risk management, as defi ned by Toakley (1989 cited in Uher andToakley, 1999) describes a procedure which controls the level of risk and mitigates its effects. A number of scholars have come up with definitions of risk management (Boehm, 1991;Edwards, 1995;Jaafari et al., 1995;Kerzner, 2001;Chapman and Ward, 1997;Edwards and Bowen, 1998;Hastak and Shaked, 2000;Lyons and Skitmore, 2004;Project Risk Management Handbook, 2003;Gray and Larson, 2005). The proposed defi nitions divide the risk management process into a number of steps which varies from three steps to more. However, the defi nitions are consistent in recognizing risk identifi cation, risk analysis, and risk handling/ risk response as the key steps of the risk management process. Only the important elements of this procedure are discussed in terms of their relevance to the stipulated objectives of this research. Hayes et al. (1986), Williams (1995), and Godfrey (1996) have seen risk identifi cation as the fi rst important step in the risk management process. Dawood (1998) has shown that systematic risk management enables the early detection of risks. This eliminates the need for contingency plans to cover almost every eventuality. Risk identifi cation involves identifying the source and type of risks. According to Flanagan and Norman (1993), an identifi ed risk is no longer a risk but a management problem. It has also been pointed out that a bad defi nition of a risk may precipitate other risks. Therefore, obtaining a clear view of the risk event is the fi rst step when focusing on the sources of risk and their potential effects.

Risk identifi cation and classifi cation
Classifi cation of risks entails identifying the type, consequence and impact of risk. Wiguna and Scott (2006) have derived a risk hierarchy under four risk categories: external and site condition risks, economic and fi nancial risks, technical and contractual risks, and managerial risks. This classifi cation of risks adopted in this study. According to Bunni (1997), when a risk has been identifi ed, assessed and analyzed, it must be allocated to various parties in order to keep it under control and to prevent the occurrence of harmful consequences. Andi (2006) has argued that "construction risks, can hardly ever be eliminated. They can merely be transferred or shared from one party to another through contract clauses". This is supported by Mak and Picken (2000) who emphasize the fact that contractors should be ready to accept a certain level of risk due to unforeseen costs they incur during construction and that risk is also an issue for clients. Such allocation of risk becomes part of the risk management process. Thompson and Perry (1992) suggest that a carefully drawn up contract will ensure the right allocation of responsibilities in the same way as the procedure which determines the type of contract and the tendering procedure for a project. It will defi ne the role of each constituent in the contract, such as the contract agreement, conditions of contract, specifi cations, preamble notes, bills of quantities and drawings, etc., which determine the allocation of risks. Although risks can be transferred beyond the limits of contract clauses that can only be with the concurrence of both parties as seen in the study by Wang and Chou (2003).

Risk allocation
A party to whom a risk is allocated is considered to have the "ownership of risk," which according to Uff (1995) and Godfrey (1996) has several meanings: a) having a stake in the benefi t or harm that may arise from the activity that leads to the risk; b) responsibility for the risk; c) accountability for the control of risk; and d) fi nancial responsibility for the whole or part of the harm arising from the risk should it materialize. Kartam and Kartam (2001) have argued that all the risks should rightfully reside with the owner and transfer to another party should entail fair compensation. However, the common understanding on risk allocation has it that the receiving party has both the competence and expertise to fairly assess the risk and to control or minimize it (Hartman, 1996;Fisk, 1997;Godfrey, 1996;Perry and Hayes, 1985).

Risk handling / risk response
Risk handling by lessening their impact is a critical component of risk management. Managers need to realize the contents and effects of all alternatives before making decisions about an appropriate strategy for risk handling (Wang and Chou, 2003). Risk handling is the choice of a proper strategy to reduce the negative impact of the risk (Miller and Lessard, 2001). It is defi ned as the fi rst step in risk control by Baker et al. (1999a). But Kim and Bajaj (2000) defi ne risk handling/response as the way risk issues are dealt with. According to Flanagan and Norman (1993), risk response refers to how the risk should be managed either by transferring it to another party or by retaining it. Further, risk handling principles are classifi ed mainly into four categories, i.e. risk retention, risk reduction, risk transfer and risk avoidance (Carter and Doherty, 1974;Flanagan and Norman, 1993;Raftery, 1994;Baker et al., 1999b;Dey, 2001;Wang and Chou, 2003). Wang and Chou (2003) see risk handling strategies as consisting of one, or a combination, of the above methods. Studies have proved the validity of various strategies chosen on the basis of individual projects. However, the study by Fan et al. (2008) has established that the risk-handling decisions of a project are determined by project characteristics (e.g. project size, slack, unit prevention cost, risk situation, etc.).

METHODOLOGY
The research adopted the Multiple Case Studies approach. According to Yin (1994), multiple case studies validate results through replication as the approach uses different cases. Further Yin stresses that the criteria for selecting cases is a matter of discretion and judgment, convenience, access and to be those which are subjective for purpose of the research. Therefore this research focused on two mega foreign-funded road projects which were near completion to avoid complexities which may arise in evaluating different types of road projects simultaneously. Projects which adopt traditional procurement method with ad-measurement were selected as it is the most widely used procurement method used in Sri Lanka. The cases selected on the basis of having a project duration of about twenty four months or more, as researchers believe that a longer period is necessary to get risk related information. The Table 1 gives the details of the two cases.
Multiple sources of evidence comprising semi-structured interviews, documents such as letters, weather records, bill of quantities, claim reports, non-conformity reports, variation orders, project programme, public complaint reports, certifi ed monthly bills and monthly progress reports, and archival records such as past weather records were used in this study for data collection. Triangulation, which is the rationale behind the use of multiple sources of evidence, has been addressed here. According to Love et al. (2002), the triangulation approach is useful since it enables both qualitative and quantitative data to be used in generalizing the fi ndings.
The data was analyzed through the content analysis method. The software QSR NVivo1.0 was used to codify interview transcripts. The results were arrived at after a cross case analysis.

Risk sources associated with road projects
The study began with 26 risk sources which were gathered through the literature review and through interview transcripts. However, it was found that only 23 risk sources were pertinent to the two cases. The 23 risk sources have been classifi ed in Figure 1 under four types of risk sources in order to formulate a risk classifi cation framework based on the literature review.
There were only two risk sources that were not common to the two cases out of identifi ed 23 factors. They were delayed payments and insuffi ciency in the preliminaries bill. There had been a delay in two interim payments and the contractor's facilities had not been included in the preliminaries bill in one case.
All other risks were common to both the cases. Construction activities had to be halted for a few days after the Tsunami disaster. With regard to the other, the reason had to do with earth slips. This could be classifi ed under Acts of God. The impact of adverse weather conditions was such that materials had been washed away and critical work affected by the unexpected rainfall. In addition, changes imposed by the Engineer, dealings with utility agencies, late handing over of the site and late approvals too caused signifi cant diffi culties to the Contractor.
The risk of defective design and scope change cannot be underestimated because this would lead to poor performance of the completed road. The dependence on foreign funds too was a risk as the contract sum in both projects had exceeded the forecasted sum while the amount of funds was limited. Insuffi cient estimation was a risk mainly because price escalation had not been considered for recurrent preliminary items. The increase in the contract sum by more than fi fty percent was due to infl ation. Legislative changes were also signifi cant since there had been a change in the labour act which required the salaries of labourers to be increased and fuel adjustment charges on electricity bills.
Special attention was also paid to the risks of low labour and equipment productivity and procurement of resources. With regard to relations with neighborhood, many complaints had been received from the neighbourhood such as house damage due to cracks, damages to boundary walls and access paths, the problem of land fi ll, endangerment of houses due to land cutting, accumulation of waste in paddy  lands, etc. The risk of public security and safety was also high in this type of infrastructure project. Both the regulations and the diffi culty in obtaining permits was also a risk as the projects were required to obtain permits for the use of explosives in road works and in the quarry. The contractor had to pay royalty too for the quarry because it had been forest land. Scope change and tentative drawings were the risks that contributed the most to the cost and time overrun in both cases. Increase in the road width, change in the road surface from Double Bituminous Surface Treatment (DBST) to asphalt paving in one case, and the addition of a binder course layer and the introduction of a hard shoulder instead of the earth shoulder in the other case were due to the change in scope. Finally, the risk of unforeseen site ground conditions was signifi cant due to the diffi culty in identifying underground cables, or due to changes in sub-grade requiring the use of rock fi ll or type-1 soil in areas where the water table was high and extra excavations for places where soil conditions were weak. Having identifi ed risk sources, their proper allocation becomes necessary to apportion the risk responsibilities of the parties.

Actual risk allocation vs. risk allocation through contract clauses
The Conditions of Contract that had been used in both projects was FIDIC the one published by the International Federation of Consulting Engineers (FIDIC, 1987). Since risks are allocated to contracting parties through contract clauses, the administration of construction risks was fi rst analysed with the aid of the Conditions of Contract which had been used in the two cases. Since the Conditions of Contract used in the two cases was the same, there was an identical basis for the analysis. Secondly, the actual allocation which is the allocation of risks beyond the contract clauses but within the consensus of contracting parties is identifi ed and these risks were also analysed in line with the views of respondents and in particular the archival sources (refer Table 2).
It was found that actual risk allocation and risk allocation through contract clauses were the same for majority of risk factors, although the risk of Acts of God becomes a risk to the Employer through the sub-clause 20.4 (h) [Employer's Risks], it was revealed that the Contractor too had to share in this risk. Similarly, though the risk of late handing over of the site had been allocated to the Employer under the sub-clause 42.2 [Failure to Give Possession], the Contractor too had to share this risk because of irrecoverable diffi culties he had to face. The Employer had borne the risk of scope change in both projects as in sub-clause 52.3 [Variations Exceeding 15 per cent], while the Contractor too had to carry a certain risk due to delays in the completion of the project. In other words, in the three instances cited, the risks had been shared by both parties though the allocation was only to the Employer according to contract clauses.
According to sub-clause 20.4 (g) [Employer's Risks], Defective Design, that is, loss or damage to the extent that it is due to the design of the Works rather than any part of the design provided by the Contractor or for which the Contractor is responsible, constitutes a risk to the Employer. However, the risk also lies with the Contractor according to the sub-clause 8.1 [Contractor's General Responsibilities]. However in both cases this risk had been transferred to the Consultant through a separate agreement between the Employer and the Consultant since the Consultant had been appointed as an independent party. Moreover, it could be seen that the Employer in one case had taken measures before the start of the project to eliminate design defects. Accordingly, this risk had been shared by all three parties in actual fact. In accordance with sub-paragraphs (a), and (b) of sub-clause 19.   Table 2 shown above compares the actual risk allocation (denoted by the dark circle -•) against risk allocation through contract clauses (denoted by the light circle -○).

Proposed risk handling framework
Having considered the allocation of risks between contracting parties all actual handling of those risks, a risk handling framework was developed. Semi-structured interviews with documentary evidence were used for this task. This allowed the researcher to ask for the facts as well as for the respondent's opinions about an event. Respondents were allowed to disclose the current handling methods and also propose their own views regarding possible handling of these risks. These results are summarized in Table 3. In developing this framework, the risk sources were categorized into four types initially. The actual allocation of risks, which was identifi ed using the case study approach is shown against each risk source along with the risk response that could be used in dealing with it. The last column provides the risk handling actions that could be adopted for the relevant risk response.

CONCLUSIONS
The research was initially begun with twenty six risk sources. However, during the analysis it was found that there were a few risks that were not relevant to the two cases under study. The observance of real cases, as the literature on the subject demonstrates reveals the extent to which the environment determines construction work and the many risks to which they are exposed throughout the entire process.
With regard to Acts of God, it is evident that they are specifi c to the geographical location of the construction project, so that any party to a contract is expected to identify the probability of occurrence of such events. Simi-larly, the effect of the dependence on foreign funds was also specifi c to each project as the terms of the funding arrangement are not the same.
Risks of defective design, late approvals, late handing over of the site, tentative drawings and unforeseen site ground conditions had thwarted the Contractor on many occasions. Moreover, relations with neighbourhood and public security and safety were also very important in pursuing these social capital development projects. Infl ation and scope change can also be cited as factors that determine the failure of the parties concerned to confi ne themselves to the cost and time limits of the two cases. Therefore, these risks are identifi ed as very vital.
Another important aspect in risk identification is that the contractual parties should adopt a continuous learning approach. Past projects and past events are real-life scenarios from which to gain experience that might stand the parties in good stead in the future so that probable risks that might be encountered in a new project can be identifi ed beforehand and measures taken in order to avoid triggering those risk events. Thus, it was felt that early identifi cation of a risk source was essential for its proper allocation.
It is a fact that the employer allocates construction risks through contract clauses before the contract is awarded. This should encourage Contractors to obtain a clear understanding of the risks they are allocated with. Disagreements may also occur from the absence of related contract clauses, unclear stipulations, and queries on the fairness of risk allocation. In such situations, though certain risks had been allocated specifi cally to a party through contract clauses, it might transpire that the other parties too might have to bear consequences that have arisen because of those risks. Risk sharing by the contract team has proven to be more effective in dealing with such risks. It could also be observed that some risks could be effectively reduced with corporation among the parties concerned, for example, risks arising from Acts of God, with utility agencies, in relations to neighbourhood, in public security and safety issues, pertaining to regulations and the diffi culty of obtaining permits. Intervention of the Employer, who represents the government, is essential when dealing in particular with neighbourhood, public security and safety issues.
Transfer of risks is also an important risk response method because it could act as a defense against certain losses and to achieve the organizational objectives of each party. Contractors however were reluctant to forward claims for losses in order not to harm the good rapport with the Employer. In this instance the involvement of insurers is signifi cant.
Risk avoidance, which is considered the most effective risk handling method, can be achieved with the early detection of events, possibly at the estimating stage, and by keeping written records and giving written notices of possible negative events to the Engineer. It is also evident that the effects of risks that were to be retained by a particular party would be minimal if that party were to handle it using other methods. It is also to be noted that there is no one best way to deal with risk and different handling methods would have to be employed depending on the type and nature of the risk.