Tuesday, 26 April 2011 11:34

**Introduction **

This chapter deals with general aspects of forecasting or estimating of construction cost. Definitions are first given. Roles of the estimators are described. Classes of estimates and accuracy are addressed. Next, the importance and preparation of pre-design estimating are explained. Following that, pre-design estimating models are reviewed in detail. They are grouped into three main categories: 1) empirical; 2) regression; and 3) probabilistic models. Factors affecting construction costs and causes of errors in estimating are also explained. Then, adjustment of the historical cost data, e.g. using cost indices is amended. Project budgeting is briefly reviewed. Finally, the roles of computer in estimating are discussed. **Definitions**

Estimating may be defined as one or a combination of the following.

1) Estimating is an information processing approach to form construction cost estimate during the design process (Birrell, 1980).

2) Estimating is the compilation of all the costs of the elements of a project or effort included within an agreed upon project scope. To a contractor, this is the cost that will most likely be incurred to complete the project as defined in the contract documents and to turn it over to the owner (Nelson et al., 1982).

3) Estimating is a combination of a mechanical process and subjective expertise. It consists of application of appropriate unit rates to the measures finished quantities of a proposed building (Ashworth, 1988).

4) Estimating is the production of a statement of the approximate quantity of materials, time, and costs to perform construction decisions (Carr, 1989).

5) Estimating plays important roles in forecasting future events in construction process. It consists of two distinct tasks;

1) to determine the probable real cost; and

2) to determine the probable real time to build a project (Peurifoy and Oberlender, 1989).

6) Estimating is a technical function undertaken to access and predict the total cost of executing item(s) of work in a given time using all project information and resource (Kwaku, 1994).

7) Estimating is the fundamental process of the construction industry that answers the question "how much is the project expected to cost?" (Schuette and Liska, 1994).

8) Estimating the cost of building and civil engineering works takes into account the three major groups:

1) material, labour, and plant;

2) dayworks, provisional sums, and prime cost; and

3) preliminaries (Geddes, 1996).

As this research deals with pre-design cost estimating, estimating is defined as forecasting of building cost done at a stage when the design of building has not yet been performed. The process uses basic information of the building, e.g. height, structural system, foundation, exterior finishing, interior decoration, and site accessibility. This information is used in conjunction with price indices to account for changes in value of money. The term "estimating" and "forecasting" are used interchangeably in this dissertation.

**Estimators **

An estimator (or quantity surveyor, or cost engineer) is the person who prepares estimates (Schuette and Liska, 1994). In the planning, design, and perhaps, construction stages, an estimator is always involved for studies requiring thorough understanding of the principles and methods of engineering economics. He or she must often work closely with managers, accountants, financial analysts, and engineers to forecast the cash or borrowing needs for the project. As major decision is made from information contained in the conceptual or preliminary estimate, this places a responsibility and liability on the estimator. He or she will risk reputation when insufficiently accurate estimate is prepared for a bid but the owner or the contractor will risk money (ASCE, 1985; Schuette and Liska, 1994).

A good estimator must conceptualize the complete building before it is fully designed. He or she must be able to think, see holistically, and then perceive the details of the project (Carr, 1989; Raftery, 1991; Bledsoe, 1992; Schuette and Liska, 1994). The estimator must also have all the qualities and skills of a detailed estimator plus the ability to anticipate design decisions and communicate those assumptions made during the conceptual estimating process. He or she must also be knowledgeable of the expected life of construction materials, accounting, taxation, law, economics, and awareness of engineering design. Ahuja (1976) summarized qualifications for a good estimator which are: 1) patience of detail; 2) technical knowledge; 3) good memory; 4) knowledge of construction process; 5) able to plan the works; 6) have an idea of relative costs; and 7) good judgement.

Good estimator should not select information simply for its convenience or its appearance of objectively, e.g. using cost data from projects or published estimating guides without sufficient knowledge (Carr, 1989). Consequently, he or she must not spend so much time and effort to analyze unnecessary details too precise in determining the costs of insignificant items as the estimating will take time and be expensive (Carr, 1989; Diamant and Tumblin, 1990). In a bill of quantities for civil engineering project, 80% of the costs can be attributed to 20% of the items, and vice versa (Ashworth, 1988; Poh and Horner, 1995).

The estimator who has many years of experience in estimating is called an expert. Based on very little project information, the expert can exhibit expertise in performing cost estimating which are: 1) more relax but confident; 2) more concerned with maintaining familiarity with the market and overall price levels than others who are more concerned with careful analysis: and 3) able to recall the overall price of the projects undertaken (O'brien, 1994). Skitmore and Wilcock (1994) found that smaller builders price only half of the items in bill of quantities for competitive tender by the detailed methods prescribed in the standard texts. The remaining items are rated mainly by experience. The more experienced the estimator, the more the likelihood that subjective decisions will be based on objective experience and personal bias (Bladey et al., 1990).

**Classes of Estimates**

Estimates can be categorized into several classes according to purposes, budget limitation, time, and accuracy (Antill, 1970; ASCE, 1985; Peurifoy and Oberlender, 1989).

Antill (1970) classified three basic types of estimates. First, approximate estimates (or rough, or pre-design estimate) is generally applied to assess the magnitude of a project prior to its design. It can also be performed with minimum preparation of design and related information. Therefore, it is based on a sound intelligent guess, judgement and experience of similar projects. Second, preliminary or definitive estimate is used for accessing the relative costs of alternative designs or comparing alternative schemes in economic studies. It needs designs and preparation of drawings. When more details and a reasonable quantity can be obtained, the accuracy can possibly be lowered. Third, detailed estimate provides the final assessment of the cost of a project. It is compiled from; 1) drawings; 2) specification; and 3) bill of quantity. Ranges of accuracy are depended on the methods of obtaining the costs, i.e. ฑ8 to ฑ12% for unit rates, and ฑ5 to ฑ8% for basic cost method.

The AACE (1979) classified three classes of estimates; 1) order of magnitude; 2) budget; and 3) definitive estimates. The order of magnitude can be made any detail but it is often developed by using cost-capacity curves, scale-up or down factors or ratio estimating techniques. The budget estimate is referred to the owner's budget which needed flow sheets, layouts and equipment details. The definitive estimate is developed from very defined engineering data including plans, elevations, system and instrument diagrams, equipment data and quotations, soil data and major foundation, and completed specifications. The AACE stated general fact that estimates are more likely to be low than high due to numerous factors, including inflationary trends and incomplete data among others.

Nelson et al. (1983) grouped cost estimates into two main categories; 1) preliminary estimate; and 2) detailed estimate. The former can further be divided into two sub-categories; 1) order of magnitude; and 2) conceptual estimate. The latter could be divided into three sub-categories which were; 1) semi-definitive estimate; 2) definitive; and 3) check estimate

Bowen and Edwards (1985) verified the historical records obtained from the architects, and quantity surveyors. They classified error margins in estimating according to the stages of works: 1) conceptual or cost target; 2) sketch design; 3) design investigation; 4) drawings submission; and 5) bill rate.

Hendrickson and Tung (1989) classified cost estimates into three main groups. First, the design cost estimate is used for the owner or designers. It runs parallel with the planning and design, and can further be divided into four subgroups: 1) screening estimate (or order of magnitude estimate); 2) preliminary estimate (or conceptual estimate); 3) detailed estimate (or definitive estimates); and 4) engineering's estimate. Second, the bid estimate consists of direct cost plus markup, the direct cost is generally derived from: 1) sub-contractor quotations; 2) quantity takeoffs; and 3) construction procedures. Third, the control estimate is used for monitoring the project during construction. It can be derived from available information to establish: 1) budget estimate for financing; 2) budgeted cost after contracting but prior to construction; and 3) estimate cost to completion during the progress of construction.

Peurifoy and Oberlender (1989) divided cost estimates into two classes: 1) approximate (or preliminary or conceptual or budget) estimates; and 2) detailed (or final or definitive) estimates. They explained that estimating consists of two main tasks; 1) to determine the probable real cost; and 2) to determine probable real time to build a project.

Ogunlana (1989) summarized the nature and characteristics of estimating which are: 1) accuracy improves with the development of the project such that the distribution of errors narrows from feasibility to settlement; 2) underestimates are more likely than overestimates; and 3) the final cost of a project cannot be established until the settlement of project accounts.

Barrie and Paulson (1992) classified cost estimates into three classes. First, conceptual or preliminary estimate is used for budgeting purposes. It is prepared to prior bidding. Second, fair-cost estimate is normally prepared by the contractor based on the bid documents. It is used for bidding. Third, definitive estimate is prepared to fix anticipated cost of the project within little margin of error.

Poh and Horner (1995) stated that reduction in the number of items for which prices are estimated might encourage an improvement in accuracy of the unit rates derived by the estimator. So, the total accuracy of cost-significant models might not be very different from the values of 5 to 15%, typically achieved at the tender stage.

**Pre-design Cost Estimating **

1. Other names of pre-design estimating Pre-design estimate means estimating done at early stage when full design of the project has not yet been completed. The purposes of pre-design estimate are to make a final decision on investment, budgeting, and financing the project. Pre-design estimate may involve several iterations of estimate in parallel with conceptual or preliminary design. The pre-design estimate can be called by other names, i.e. rough estimate (Antill, 1970), preliminary estimate (Antill, 1970; Neilson et al., 1983; Peurifoy and Oberlender, 1989; Barrie and Paulson, 1992; Bledsoe, 1992), Order of magnitude estimates (AACE, 1979), approximate (Antill, 1970; Peurifoy and Oberlender, 1989), budget estimate (AACE, 1979; peurifoy and Oberlender, 1989), conceptual or cost target estimate (Bowen and Edwards, 1985; Peurifoy and Oberlender, 1989; Barrie and Paulson, 1992; Bledsoe, 1992; O'brien, 1994; Schuette and Liska, 1994; Elazouni et al., 1997), and feasibility estimate (Schuette and Liska, 1994).

2. Importance of pre-design estimatesConstruction projects always start with the preparation of estimates of probable cost for budget, financing, bidding, accounting, and cost control (Kouskoulas and Koehn, 1974; Rurkpuritat, 1978; ASCE, 1985). It provides the frameworks for evaluating different alternatives at the conceptual design stage (Elazouni et al., 1997). Preliminary estimating is important for projects that utilize scarce resources because of limited funding and rapid increase of investment costs to develop the project (Sanders et al., 1992). Vergara and Boyer (1974) explained the processes of estimates. Before any civil engineering project starts, estimates need to be prepared. The owner has to engage the services of a designer to design a building that satisfies his or her desires within budget constraints. Thus, continuous estimates have to be made to inform the owner, or simply to assure an economical design. The estimates are very crude at beginning stage as they are prepared on the basis of rough designs. Using these estimates, continuous modification of the design is also made until the objectives are met. In preparing an estimate, a number of assumptions are required such that no method might be considered perfect. Since the process of accepting or rejecting the various proposal is based on the estimated costs, they must be as accurate as possible so that the results and conclusion will be as realistic as practicable (Neilson et al., 1983; Ashworth, 1988).

Barrie and Paulson (1992) explained that the preliminary estimate assists the overall cost-control program by serving as the first check against the budget. It will indicate the cost overruns early enough for the project team to review the design for possible alternates. Since preliminary estimate is made prior to the completion of detailed design, the margin of error will be relatively large. Then, the larger contingency should be applied. The contingency varies with the amount of design information available and the extent of cost information obtainable from similar projects.

3. Preparation of pre-design estimate Typically, before a final decision is made, several iterations of conceptual design and estimate has been accomplished. At the pre-design phase, only basic design information is available, e.g. architectural and basic engineering systems, and design criteria (O'brien, 1994). A set of schematic drawings such as plans, elevations, sections, and perhaps, perspective sketches or renderings are prepared. These are sufficient for pre-design estimates (Peurifoy and Oberlender, 1989; Schuette and Liska, 1994). The pre-design estimates are prepared in practices primarily based on analogy with previous similar cases (Park, 1963; Antill, 1970; Kouskoulas and Koehn, 1974; Rahman, 1977; Rurkpuritat, 1978; Sadashiv, 1979; Karshenas, 1984; Ashworth, 1988; Ogunlana, 1989; Peurifoy and Oberlender, 1989; Raftery, 1991; Barrie and Paulson, 1992; Bledsoe, 1992; O'brien, 1994; Schuette and Liska, 1994; Taksana, 1994; Poh and Horner, 1995; Elazouni et al., 1997).

**Cost Modeling**

1. Stages of cost modeling Cost models are the techniques used for forecasting the estimated cost of proposed construction projects (Ashworth, 1988). They may be mathematical models or formula that best describe the data collected. Ashworth (1988) summarized eight conventional stages of cost modeling:

1) formulating the problem; 2) collection of data; 3) analysis of data; 4) model building; 5) optimum model; 6) evaluation; 7) testing; and 8) application. In addition, the new data would be added to the collected data, i.e. in the database. It can be used for calibrating the model in the next updating. Raftery (1991) suggested four criteria for assessing the performance of the models when discussing the internal cognitive processes of human decision making: 1) data; 2) data/model interface; 3) model technique; and 4) interpretation of output. Figure 2.3 shows his conceptual framework for the assessment of model performance (Raftery, 1991).

2. Requirements of good models Ashworth (1988) summarized that a good cost model should incorporate five criteria. First, the data required for the model should be freely available in the form and amount necessary. Second, the model should allow for continuous updating by incorporating new data that become available. Third, the model should account for the changing situation in the construction industry. Forth, the model should provide for quick, cheap, and efficient processing. Fifth, the model should be accurate and reliable.

Seeley (1996) suggested that a good model should be simple enough for manipulation and understanding by those who use it. It should be representative enough in the total range of the implications it may have. It should also be complex enough to accurately represent the system. Further, the better model should be able to function with input data outside the original data used to formulate the model.

Development of cost models is a lengthy process, and they will or will not be successfully practiced or applied in the construction industry. However, necessary expertise accompanied with sufficient research time will possibly make the models become valuable, i.e. the required results will be achieved (Ashworth, 1988).

3. Types of modelsAshworth (1988) classified models into two main types: 1) deterministic; and 2) probabilistic models. In the deterministic models, the outputs are assumed to be predicted exactly as they are attributed by the known input variables. The probabilistic models recognize the uncertainty of some variables which can only be estimated using concepts based on the probability theory. In construction industry practice, the majority of models fall into the second category

**Generations and Applications of Cost Estimating Models **

1. Estimating methods and modelsVarious methods of estimation are based on different factors, e.g. purpose of estimate, amount of design information available, time and budget to perform estimate, and historical records (Antill, 1970; ASCE, 1985; Peurifoy and Oberlender, 1989). There are three main characteristics of pre-design estimating models as below.

1) All pre-design estimating models use historical records to derive mathematical equations that define the relationship between the independent variables and construction costs (Park, 1963; Kouskoulas and Koehn, 1974; Rahman, 1977; Rurkpuritat, 1978; Sadashiv, 1979; Karshenas, 1984; Lim, 1988; Kiattikomol et al., 1992; Sanders et al., 1992; Maroun and Alkass, 1993; Skitmore and Wilcock, 1994; Taksana, 1994). In a sense, all the cost estimating methods are cost models.

2) They can provide reasonably accurate, easily computed, and inexpensive prediction method to the key people, i.e. the owners, designers and consultants and the contractors who had to be involved in construction (Rurkpuritat, 1978; Karshenas, 1984; Poh and Horner, 1985; Peurifoy and Oberlender, 1989; Schuette and Liska, 1994).

3) They serve decision making, economic feasibility, financing, and budget controlling for the project in the absence of complete design (Kouskoulas and Koehn, 1974; Rurkpuritat, 1978; Barrie and Paulson, 1992; Maroun and Alkass, 1993).

Cost models can be at least classified into three main generations: 1) empirical; 2) regression or factor; and 3) probabilistic or simulation models (Ashworth, 1988; Raftery, 1991; Elazouni et al., 1997). They are described in details as follows.

2. Empirical models These types of model are based upon observation, experience, and intuition. Physical appearance of the building and the methods of constructions are modelled in terms of descriptions and dimensions. Bill of quantities are example of an empirical model (Ashworth, 1988). Empirical models may use different ways of calculating, e.g. unit method, cubic method, floor area method, storey enclosure method, approximate quantities method, elementary cost analysis, comparative estimate, factor estimate, range estimate, parameter estimate, and percentage estimate (Ashworth, 1988; Ogunlana, 1989; Raftery, 1991; Taksana, 1994).

This method is simple, easy to understand, and the calculation can be performed quickly. Therefore, the people at management level in construction industry are familiar with these methods or models. Construction costs can be represented in forms of the units of facilities, e.g. cost/car space, cost/bed, cost/seat, cost/pupil (Park, 1963, Ogunlana, 1989, Taksana, 1994) or in other ways.

The American Society of Civil Engineering - ASCE (1962) categorized three main methods of cost estimating: 1) product unit method; 2) physical size method; and 3) work unit method. The product unit method is suitable for pre-design phase of the buildings for which the product unit can be counted easily, e.g. the number of guest rooms for a hotel, the number of rooms for an apartment or a school, and the number of patient beds for a hospital. The physical size method derives the cost from size of project and unit cost per area which is suitable for master planning phase of the project. The work unit method needs completed detailed design to derive the cost from the cost per unit of materials used in the project, e.g. cost per cubic meter of concrete, cost per ton of reinforcement.

Park (1963) developed a preliminary estimate by using end-product unit curves for four different types of civil engineering projects: 1) sewage treatment plants; 2) steam-electric generating plants; 3) school construction; and 4) public housing project. However, the curves are no longer being used due to changes of construction materials and techniques.

Rahman (1977) used historical price data to develop simple methods for pricing the construction projects applicable for use in Asian countries. He summarized three forecasting methods: 1) component method; 2) simplified component method; and 3) unit method.

Lim (1988) developed a method for pre-design estimate with adjustment for alternative specifications and price escalation. He considered breakdown of individual work item. He described that a single rate method as a function of total building cost seemed not adequate today because of significant changes in materials, labour, construction techniques and effects of price escalation.

Kiattikomol et al. (1992) developed knowledge base for building valuation for various business and industry sectors, e.g. construction, finance, banking and insurance. The models cover multi-stories buildings, medium size factories, and rice mills.

Taksana (1994) derived parametric model for pre-design cost estimating of high-rise office building in Bangkok. The model consists of parameters, area of trade, physical estimating ratios and unit price. Total cost can be derived from all the costs of each area of trade, i.e. multiplying parameter and physical estimating ratio by unit price. The unit price could occasionally be updated according to the changes of market conditions.

However, the weakness of empirical models are lack of precision. The drawback derives from the inability to arrive at suitable relationships between elements and project on which cost forecasts can be based (Ashworth, 1988; Ogunlana, 1989). It is also difficult to make allowance for a whole range of factors, i.e. shape and size of the buildings, materials, construction methods, preliminaries, and market consideration (Ashworth, 1988; Ogunlana, 1989; Raftery, 1991).

3. Regression modelsRegression is a technique of determining a formula or mathematical model which best described the data (Ashworth, 1988; Elazouni et al., 1997). The output (or dependent variable) may be expressed in terms of a single or a set of independent variables. Thus, the relationships may be called linear or multi-linear regression models. However, non-linear regression models also exist, e.g. in forms of polynomial, exponential, logarithmic, and power functions (karshenas, 1984; Carpenter and Barthelemy, 1994; Elazouni et al., 1997).

Kouskoulas and Koehn (1974) derived pre-design cost estimating function from historical cost data for an arbitrary building, e.g. school, apartment, store, and office. They said that criterion for the selection of the variables is the availability of data on such variables from completed projects.

Rurkpuritat (1978) used historical costs to examine four parameters affecting cost of buildings in Bangkok: 1) price index; 2) building type; 3) building height; and 4) building quality. The study covered several types of buildings, i.e. school, hospital, residence, telephone building, shopping center, factory, flat and warehouse.

Sadashiv (1979) used historical costs and multiple regression to develop two probabilistic models for estimating project duration and cost of buildings in Bangkok. He found that the best four independent variables: 1) height in meter or number of floor; 2) number of types of major equipment used; 3) type index; and 4) quality index.

Karshenas (1984) used historical costs of buildings to investigate the relationship among cost for typical floor area, and building height for multi-story buildings. The best explanatory model was expressed in forms of power function. The costs of the buildings were not include landscaping, roads, open parking space, waste treatment, facilities, and special equipment.

Kaka and Price (1991) developed a reliable net cash flow model to be used by contractors at the tendering stage. They said that the cost commitment model is based on historical cost data of projects. They can be adjusted for inflation by either expected value of inflation index input by the users or randomly generated to simulate the effect of actual rates on cash flow.

Sanders (1992) presented models for preliminary cost estimating of bridge widening projects. The models are based on statistical analysis of past bid. Numbers of linear regression for each work item which constitute more than one percent of the total cost were developed. Total cost derived from the sum of these models took 70% of the total cost of the project.

Maroun and Alkass (1993) used regression analysis to derive simplified parametric price revision models based on two sensitive variables; 1) exchange rate; and 2) the labour minimum monthly wage. The models were built for price escalation to adjust the change in the prices of construction contracts for works executed in a high inflated economy. They said that parametric estimating relies upon statistical equations that relate cost to some other variables, this method is sometimes called statistical estimating. They also emphasized that the development of cost models is heavily dependent on the basic cost estimating relationships that describe the situation being modeled.

Poh and Horner (1995) explained that in the UK, a variety of cost models incorporating a wide range of levels of detail has been used and resulted in parametric models. Mostly, the models which commonly express the cost of building in terms of a rate per floor area have been applied at the feasibility study stage. They said that the unit rate provides a convenient method of valuing the works for interim payments. An ideal model for cost-significance should be: 1) simple; 2) sufficiently accurate; 3) able to provide rapid feedback; 4) made up only of elements which are easy to measure and which reflect site operations usable from inception to operation. Mostly, cost-significant models rely on the well documented finding that 80% of a bill value is contained within the 20% of the items which are most expensive. The total price of the project can be calculated simply by multiplying the total price of cost-significant work packages by appropriate factor, close to 1.25. The developed model then can be used for estimating and controlling the building cost of future similar projects.

The reliability of regression models is based on a sound knowledge of previously achieved performance, i.e. historical records (Ashworth, 1988; Elazouni et al., 1997). In the construction industry, recording of the performance or data is difficult (Ashworth, 1988). Insufficient numbers of similar projects can lead to low capability of the models.

4. Probabilistic cost modelsBowen and Edwards (1985) described needs for shift of existing paradigm in quantitative cost modelling and price forecasting for construction projects. The existing traditional cost models and price forecasting techniques are dependent upon data derived from historical sources. Hence they could be called historical-deterministic. Four main reasons for shifting to the new paradigm are: 1) computer revolution that induced change and availability of cheap calculating power; 2) the potential problems of the regression models that only represent reality but do not thereby, become reality; 3) substantial database is necessary to support the formulation and maintenance of mathematical cost models; and 4) many users are baffled by the statistics.

Simulation means imitation. It is not real, but only pretending to be. A simulation model duplicates the behavior of the system, i.e. under investigation, by carefully collected data over a long period, and studying the interaction among the components. The probabilistic or simulation models allow the users much flexibility in representing complex systems that are normally difficult to analyze by standard mathematical models. Therefore, developing a model may be a very costly procedure and time-consuming particularly in optimizing the models (Ashworth, 1988).

Simulation or probabilistic modeling started with a popular method called Monte Carlo technique, i.e. used for determining a proper unit rate and productivity (Ashworth, 1988; Raftery, 1991). Probabilistic or simulation models can be used in the area of tender bidding and cost forecasting which are indeterminate in practice (Ashworth, 1988; Raftery, 1991).

Vergara and Boyer (1974) proposed probabilistic approach to cost estimating and control which are done at different stages of a project. They said that construction cost estimating is normally approached in what appears to be a deterministic manner. However, implicit in the approach is the awareness that estimating is probabilistic. Therefore, the objective would rather be trying to measure the variability of the estimates and associating a probability of occurrence with each of them than seeking an exact number. Then, a probabilistic distribution of the estimate should be generated. Having this possible variability of costs, deeper insight into the cost of the item analyzed could be gained. This was reported to have led to better judgement and weighing the risk involved.

Touran (1993) proposed probabilistic approaches for estimating the probability of a budget overrun and determining the amount of contingency funds to ensure that the project costs remains within the budget.

Bradley et al. (1990) stated that traditional methods of estimating project costs do not attempt to assess the magnitude of the variation inherent in the estimate. Therefore, there is a risk that decisions on strategy selection will be based on a high degree of uncertainty.

Seeley (1996) summarized that the probabilistic approach of modelling centers around two primary characteristic: 1) acceptance of uncertainty and imprecision; and 2) interest in artificial intelligence. However, he remarked probability models have some disadvantages. The distribution of the occurrence or variable must be known or pre-determined so that the probability can be assigned to the experiments or events. Raftery (1991) observed that the decision maker has been more and more removed from the model technique because of the increasingly sophisticated computer software. He hope that the four criteria for assessing the performance of the models may help the decision maker to gain a sufficiently systematic overall view of the models in order to interpret the output in some meaningful ways.

5. Further classification Ogunlana (1989) classified two additional generations of cost modelling. The fourth generation may not qualify as cost modelling but represents a departure from traditional practice. This approach makes use of the on-line system information provided by the Building Construction Information Service; BCIS in UK. On the other hand, Asworth (1988) and Raftery (1991) said that the system introduced and provided by the BCIS enables the estimators who use the empirical models to prepare estimates. Seeley (1996) used the name "BCIS on-line approximate estimating package" instead of model. The fifth generation is called resource based estimating. This model makes use of the available resources (i.e. materials, manpower), necessary assumptions, tentative schedule, equipment and method of construction. Costs can be derived (or measured) in the way in which they arise. Therefore, this kind of model is best prepared by the contractor at the design phase.

Factors Affecting Building CostsElazouni et al. (1997) grouped factors that determine project resource requirement into two main categories: 1) design factors; and 2) construction factors. The former mainly consists of structural features, e.g. surface area, and height. The latter deals with construction methods and productivity. However, it is sometimes difficult to take into account the construction factors at the pre-design stage. These factors usually need and make use of resources available to the constructors, i.e. the available materials, manpower, necessary assumptions, tentative schedule, equipment and method of construction (Ogunlana, 1989). In other words, these factors can be well understood only by the contractor at the design phase. In preparing pre-design estimates, only the design factors are useful. Factors affecting construction costs can be summarized as below.

1. Building function One important facet in designing of construction project is the function (Ashwoth, 1988). The function of a building implies the business target that the building serves. Nkado (1992) categorized non-interval level of measurement (qualitative) of end use or function of building into three main groups: 1) office; 2) retail and 3) other buildings.

2. Structural system Structural system deals with construction materials and techniques by which the buildings are built. Nkado (1992) considered three different predominant frames or types of structures, i.e. concrete, steel, and others, which reflected resources and time for construction.

3. Height of buildingIn general, the height results in needs of technology and presents difficulties. During construction, the height effects vertical transportation and material storage. It effects structural and system engineering, e.g. foundation, circulation, plumbing, fire fighting, and lighting (Ashworth, 1988). The height of structure can be considered either in number of storeys or in unit of length, i.e. feet or meters (Kouskoulas and Koehn, 1974; Rurkpuritat, 1978; Sadashiv, 1979; Kaeshenas, 1984; Ashworth, 1988; Nkado, 1992). Using numbers of floor or storey reflect better understanding for both the difficulties and the degree of repetition.

4. Complexity of foundation work Special features of foundation, e.g. basement, mat foundation, retaining or diaphragm wall make foundation work more complex. This variable reflects the level of technology, and time to complete the construction. Sadashiv (1979) considered this variable but Nkado (1992) used number of atrium floors instead of foundation and basement.

5. Exterior finishingExterior finishing presents the appearance of the building. It also reflects construction cost and time, and perhaps, marketing target. This variable is often included within the quality index (Kouskoulas and Koehn, 1974; Rurkpuritat, 1978; Sadashiv, 1979; Ashworth, 1988). Only the work done by Nkado (1992) considered this variable independently under the heading "cladding" (which comprised prefabricated panel, curtain wall, and brick wall).

6. Quality of interior decorating Quality can be classified by variables or attributes, i.e. appearance, strength, stability, materials used, performance finish (Ashworth, 1988). In general, basic finishing inside the building includes flooring, plastering, rendering, wall tiling, mosaics, and thin surface finishing (Burgess and White, 1979). On the other hand, decorating quality reflects additional cost and time to provide various items of interior decorating, e.g. sanitary fixture and accessories, luxury electrical accessories, carpet, floor tile, wall paper or special decorated wall, built-in furniture. This variable is included in some previous works (Kouskoulas and Koehn, 1974; Rurkpuritat, 1978; Ireland, 1985; Nkado, 1992). Ireland (1985) called this index an architectural quality instead of building quality or index.

7. Accessibility to the site At the early stage of design and estimating, information about the site is very important, e.g. restrictions or easements that exist, and availability of public services (Burgess and White, 1979). Eventhough it appeared only in the work done by Nkado (1992), this independent variable seems very important to explain the difficulties in mobilizing construction equipment, providing working space, and delivery of materials. In practice, accessibility to each construction site also affects the design of buildings.

There are other factors included in the previous researches. Sadashiv (1979) included types of major equipment as a variable in his regression model for cost forecasting. Ireland (1985) summarized that construction costs may be affected by variation to contract, use of nominated sub-contractors, and construction planning.

**Errors in Estimating **

Apart from the major factors affecting the accuracy of estimating, there are other kinds and sources of errors which are found in a number of researches. They can be summarized as follows.

1. Procedural errors Rahman (1977) summarized the factors influencing the pricing of construction projects: 1) accuracy in pricing; 2) cost involved in pricing; 3) time required for pricing; 4) availability of methods of pricing; and 5) availability of price data.

Raftery (1991) explained two major problems of data transformations in cost modeling. First, when the sum of the resources costs is spread over the unit rates to produce the priced bill of quantities with the attendant loading and tactical decision, the use of sub-contractors renders unit rates less appropriate. The other problem occurs when the unit rates are sub-divided and clustered into element costs. Raftery (1991) recommended that the elementary rates should be used by the estimators or quantities surveyors for cost planning at an early stage. Consequently, at sketch design stage, the unit rates based obtained from the comparable projects are used in conjunction with approximate quantities to produce detailed cost plans.

2. Human errorsAhuja (1976) pointed out four types of human error. First, changes in design, and incompleted information are the external errors caused by circumstances beyond the control of estimators. Second, accidental errors caused by procedural mistakes, e.g. omitting items, using wrong dimensions. Third, judgmental errors caused by poor or wrong judgment on the part of the estimator, e.g. overlooking, poor pricing, and not allowing wastage. McNulty (1982) stated that this kind of error can result from: 1) overlooking a significant item; 2) missing the impact of small detail that change an element from standard to custom; 3) unawareness of a market change. fourth, deliberate errors caused by the estimator or management. He recommended that the most important step in analyzing errors is to develop a systematic understanding of how errors develop and where to concentrate preventive efforts. He recommended some techniques to minimize and control the errors: 1) use standard procedures; 2) study available plans and specifications thoroughly; 3) analyze construction methods; 4) visit the site; 5) understand overhead items; 6) use mathematical shortcuts; 7) analyze past performance; 8) check the final estimate for gross errors; 9) compare the portions of each element to the total cost with similar portions from other projects; and 10) avoid a last minute rush. Carr (1989) said that good estimates should be well documented or systematic. So that they can be understood, verified, and corrected when necessary.

3. Uncertain nature of the projectVergara and Boyer (1974) concluded that the factors which complicate the preparation of estimates are their dynamic and uncertain nature. Costs then, may continually be updated. Examples of the uncertain nature of the project itself are: 1) weather conditions; 2) construction delays; 3) supervision policies; 4) construction methods; 5) political and economic variations; 6) changing nature of construction technology and costs; 7) difference in maintenance technology; 8) differences in labour productivity; and 9) materials and equipment availability. O'brien (1994) described factors affecting estimates which are: 1) labour productivity; 2) material availability; 3) equipment availability; 4) financial market; 5) weather; 6) constructability issues and technology needed; 7) contract types; 8) quality issues; 9) control systems; 10) supervision and management; 11) land cost; 12) tax and depreciation; 13) inflation; and 14) insurance.

4. Significance of the factorsOgunlana (1989) reported on a survey of factors affecting estimating performance: 1) type of project; 2) size of the project; 3) geographical location of the project; 4) number of bidders; 5) state of the market; 6) level of information available; 7) ability of the estimator; and 8) project duration. He summarized that historical cost data, estimator's experience (or expertise) and design information are the most highly rated factors affecting the accuracy of cost estimates. Market conditions, project type and project complexity are moderately rated factors. Lowly rated factors are: 1) project size; 2) project location; 3) number of bidders; and 4) project duration.

**Adjustments to Historical Data **

1. Factors to be adjustedA major problem in estimating is to project the future escalation which is often anyone's best guess (O'brien, 1994). Generally, the estimated costs of a building originates from the historical cost records. These are based on the knowledge and experience of the estimator. The estimator therefore, must be careful when using historical cost records from the completed projects as the proposed project may have significantly different features. It is also important to clearly define the elements of the historical records so that there will be no mis-understanding on what information is included, e.g. land cost, overhead, tax, and delays (O'brien, 1994). ASCE concluded that historical cost records is the most valuable references for new work with similar nature include; 1) weather conditions; 2) mobilization cost data; 3) man-hours and equipment usage; 4) basic manual labour information (availability, skills, turnover, transportation, and special requirements); 5) equipment productivity records. The use of cost information from previous projects to forecast the cost of the proposed project will not be reliable unless an adjustment is made that; 1) is proportional to the difference in time; and 2) represents the difference in costs between the locations. Ahuja (1976) mentioned that estimates must be adjusted for many factors affecting productivity; 1) geographic location; 2) local employment conditions; 3) season; 4) weather; 5) site conditions; 6) supervision; 7) scheduling; and 8) project size.

2. Inflation The cost information which has been collected and recorded over a period of time must be converted to a current date or future time scale according to the market condition and inflation (Ashworth, 1988). Inflation is caused by increase in the stock of money that is available for spending while the quantity of goods available for purchasing does not properly increase (Pilcher, 1992). Inflation has been a common and unsolved problem for a couple of decades because of some bad effects, i.e. energy crisis, and wars. Cost estimating of a construction project is improper if computation of the inflation effects are not included (Shih, 1985). There are three kinds of inflation. First, creeping inflation occurs when inflation rate falls in between 1 to 6%; raising of income to withstand the inflation is possible. Second, rapid inflation occurs when the inflation rate is greater than 6% by which raising of income to withstand the inflation may not be possible. Third, hyper inflation is a situation in which price accelerate increased to the detriment of convertability of local currency, at the extreme end, the local currency would become valueless.

2. Inflation The cost information which has been collected and recorded over a period of time must be converted to a current date or future time scale according to the market condition and inflation (Ashworth, 1988). Inflation is caused by increase in the stock of money that is available for spending while the quantity of goods available for purchasing does not properly increase (Pilcher, 1992). Inflation has been a common and unsolved problem for a couple of decades because of some bad effects, i.e. energy crisis, and wars. Cost estimating of a construction project is improper if computation of the inflation effects are not included (Shih, 1985). There are three kinds of inflation. First, creeping inflation occurs when inflation rate falls in between 1 to 6%; raising of income to withstand the inflation is possible. Second, rapid inflation occurs when the inflation rate is greater than 6% by which raising of income to withstand the inflation may not be possible. Third, hyper inflation is a situation in which price accelerate increased to the detriment of convertability of local currency, at the extreme end, the local currency would become valueless.

Price index can be defined against a base period that is a range or point in time in which the cost of various constant quantities of representative items is usually set to 100. The values of each index for other points in time are the ratio of the new cost divided by the base period cost, expressed as a percentage (Erikson and Boyer, 1976). Price indices reflect the escalation of materials prices when compared to those prices in the base year.

Price indices are used for converting historical cost data to forms in which they can be reasonably used for estimating and forecasting (Ashworth, 1988; O'brien, 1994). An index can be constructed for the categories: 1) total construction costs of a building or building type; 2) an element or trade within building process, e.g. brickwork; and 3) a single material such as cement (Ashworth, 1988). It measures changes from one period or location to another. Thus a datum is necessary for the index. The datum is usually set to a particular number, and at a particular point, i.e. year, place, that can easily be handled to show increasing and decreasing cost. Specifically, the price index can be defined as the ratio of the cost of an item at a specified point in time to a datum established in the past. The known past costs for items are related them to the present or future time. A location index represents differences in costs when the location being considered is a distance remote from the established datum. The use of cost information from previous projects to forecast the cost of the proposed project will not be reliable unless adjustment is also made to account for the difference in location (O'brien, 1994).

Erikson and Boyer (1976) said that the price indices can be used to update construction costs without requiring detailed takeoff and pricing of the project's components. They concluded that selection of an appropriate index to upgrade an estimate involved judgement on the part of the estimator. Some factors affecting estimating are: 1) type of construction project on which the index was based; 2) geographic area on which the index was based; and 3) individual components in the project which the index does not consider.

Taylor and Bowen (1987) studied building price indices that provide a means of updating cost estimating data by extrapolating from existing cost trends to forecast price levels. A basic assumption underlying the index model is that total building costs will react correspondingly with the cost of certain items selected as representative of the movement of building costs.

Sittiwannarak and Chowichien (1990) studied processes, structures, and components of tender construction price indices derived from tender prices of different types of buildings in Bangkok. The indices comprise nine components of construction materials price indices (1976's price = 100) provided by the Department of Commercial Economics, Ministry of Commerce, Thailand, accompanied with labour index and mark-up index (overhead, profit, and tax).

William (1993) used weighted aggregate indices of the prices of constant quantities of structural steel, portland cement, lumber, and common labour. He defined cost index as a method of comparing cost changes from period to period for a fixed quantity of goods or services. More accurate forecasts of changes in construction costs can have several beneficial affects on the estimating practices of the owners and contractors. There are several factors affecting variation in cost indexes, e.g. interest rate and level of demand for new construction. The construction cost index exhibited a long term increase due to inflation. In several years, the index would be at a higher level than it is now.

Various organizations publish indices showing the differences and/or trends of the construction industry with respect to time and geographic location, e.g. Building Cost Information Service (BCIS) by Royal Institute of Chartered Surveyors (RICS) in UK, Engineering News-Record (ENR) in USA. The ENR's location index sets New York = 1.0, and carries past escalation factors in annual index, based on 1913 = 100. In Thailand, monthly and annual price indices of 13 major items of construction materials were established and have been recorded continuously since 1987 (1987's price = 100), by the Department of Business Economics. The indices are also periodically published in Thai Contractors News as they are basically used for adjusting contract prices of government projects due to various escalations.

It is seen that price index for other materials tends to increase gradually over time. The index for steel fluctuates in between a band, ranging from 110 to 140, approximately. The index for cement had a sharp peak during 1990 to 1991 which conforms to the peak of inflation rate. However, the index for cement has fluctuated within a band range from 120 to 140 since 1992.

Shih (1985) remarked that indices play important role in indicating economic activities, i.e. the expansion of an economy's capacity to produce goods and services, called economic growth, or otherwise, economic stagnation or recession. In cost estimating, the criteria for selecting an index are: 1) it should be of major importance; 2) it should be a reasonably amenable to analysis; and 3) it should have a fairly close relationship to the construction industry. The followings are some other difficulties in selecting and applying indices (Ashworth, 1988).

1) Only general indications of costs the changes in value of commodity are provided, i.e. the indices are usually produced for typical products so that problem of finding a typical construction project exist. The project will usually have some characteristics which is not provided for in the indices.

2) From time to time, products may disappear from the market. When other products are substituted, errors then, are introduced into the indices.

3) Changes in fashion, especially for materials and forms of construction, may make an index constructed many years back unreasonable for predictions.

4) Errors in sources of data can never be eliminated. For commercial or business reasons, false data may be supplied.

5) Indices attempt to measure changes in the overall pricing level thus changes in individual prices may be hidden.

Project Budgeting Erikson and Boyer (1976) explained that good estimates are necessary for all the people involved in construction projects. The owners can allocate resources to projects that promise the most benefit per unit investment while the designers need good estimating capabilities to facilitate designs within the owner's cost guidelines and to compare possible design alternatives.

Ahuja (1976) stated that economic and financial feasibility are related. They can be considered together in planning sequence. The economic feasibility study is essentially the determination of project duration at minimum cost.

Hendrickson and Tung (1989) categorized costs associated with constructed facilities of a project into two main categories: 1) capital cost; and 2) operation and maintenance cost. The former consists of the expenses related to the initial investment which were: 1) land acquisition (includes assembly, holding, and improvement); 2) planning and feasibility study; 3) architectural and engineering design; 4) construction (materials, equipment, and labour); 5) field supervision of construction; 6) constructing financing; 7) insurance and taxes during construction; 8) owner's general office overhead; 9) equipment and furnishings which was not included in construction; and 10) inspection and testing. The latter means the costs incurs in subsequent years over the project cycle: 1) land rent (if applicable); 2) operating costs, i.e. staff payroll, bonus, and benefits; 3) labour and materials for maintenance and repairs; 4) periodic renovations; 5) insurance and taxes; 6) financing costs; 7) utilities, i.e. electricity, water supply, and telephone; and 8) owner's other expenses. The magnitude of each component depends on the nature, size, and location of the project as well as the management organization. The owner is interested in achieving the lowest possible overall project cost that is consistent with its investment objectives. Even though the construction cost is normally the single largest component of the capital cost, other cost components are also significant, e.g. land cost is the major expenditure for building construction in high-density urban areas, and financing cost in large projects such as the construction of nuclear power plants.

Lo (1982) studied the importance of budgeting. He suggested six steps in the planning process: 1) determine defined and quantified objective and effectiveness; 2) do forward-looking forecasting; 3) list all rational and feasible options of execution; 4) determine amount of resources needed with consideration of budget constraints; 5) determine the optimal option of execution and test its actual feasibility; and 6) provide indexes of evaluation.

Nelson et al. (1983) emphasized the characteristic of capital investment cost which was a one-time cost, as opposed to another kind of costs called operating costs which is continuous or repeated. They divided capital investment into two parts: 1) fixed capital investment which is the capital for providing all the facilities needed for the project; and 2) working capital investment which is a revolving fund to keep the facilities operating. Capital cost estimating for the project is divided into two major elements: 1) direct costs for materials, labour, and other resources involved in the construction and installation of permanent facilities; and 2) indirect cost. Capital-cost estimating is an intuitive process to predict the final outcome of a future capital expenditure program, eventhough not all parameters and conditions concerning a project are known or were fully defined when the estimate is prepared.

Khanduri et al. (1993) categorized the life cycle costs into four main items: 1) capital costs of land and construction; 2) annual operation costs, i.e. energy, cleaning, security, ground maintenance, administrative, taxes and insurance; 3) annual repair and maintenance costs; and 4) salvage value form resale land and scrap value of building. They stressed that the operation and maintenance costs which are incurred either annually or at periodic intervals over the life of the facilities can far outweigh initial capital cost which are incurred at the initial construction stage.

This dissertation is concerned with the cost of construction only. It does not account for other life cycle costs.

Roles of Computer in EstimatingToday, estimators are responsible to prepare work at high productivity levels and desire quality estimates because modern projects are becoming larger and more complex (Kwaku, 1994). Construction information for large and complex projects may increase in quantity but time for estimating may not increase proportionally. The high productivity implies fast processing based on minimum cost while high quality deals with accuracy and reliability. Using computer as a tool can improve the effectiveness of estimating dramatically. Bowen and Edwards (1985) described needs for shift of existing paradigm in quantitative cost modeling and price forecasting for construction projects. The existing traditional cost models and price forecasting techniques are depended upon data derived from historical sources, called historical-deterministic. One reason for shifting to the new paradigm is the computer revolution that induced change and availability of cheap calculating power. Furthermore, substantial database is necessary to support the formulation and maintenance of mathematical cost models.

Bledsoe (1992) stated that estimating technology has advanced significantly in four areas: 1) quantity survey or takeoff; 2) estimating software; 3) computerized database; and 4) unit cost data on computer media. Kwaku (1994) summarized that the computer provides an expedition of the repetitive aspects of estimator's works with better productivity. It provides efficient means for recording and storing information. It can access and/or retrieve stored information rapidly. It can process calculation and update the stored information rapidly. It is possible and flexible to examine and manipulate the effects of adjustment to the cost data.

Computer technology can be successfully used in conjunction with various methods of estimating. It provides rapid retrieval of cost data, calculating an appropriate estimate using pre-determined parameter (Bowen and Edwards, 1985; Ashworth, 1988; Callahan et al., 1992). Ashworth (1988) also summarized four advantages of using the computer accompanied with cost models in the aspects of construction pricing: 1) more information leads to better decisions; 2) the information will be more reliable and giving greater confidence into decision making; 3) the cost information can be provided more quickly; and 4) suitable cost information can be produced at an earlier stage within the design process.

However, the computer should not replace the skills of estimators. Instead, it should be regarded as a tool which assist the estimators to make the optimum use of their experience and skill in preparing estimates (Kwaku, 1994). Ashworth (1988) summarized current application of computers in two main areas. First, the computerized traditional methods allow the user using known techniques to speed up the preparation of estimates as well as updating or accessing to a wider database. Second, the computerized statistical techniques require the estimator to use new methods of cost modeling.

Summary This chapter reviewed cost estimating in detail, i.e. importance, classes, errors, cost modeling, historical records and adjustments. Accuracy of an estimate is based on three main factors: 1) knowledge and experience of the estimator; and 2) available information (which categorizes cost estimating into several stages); 3) methods of estimating. Knowledge of the estimator is strongly based on the historical records of completed built projects. Consequently, the estimator may use experience in conjunction with the available information and appropriate method to prepare the estimate. Time for preparing the estimate is also important. Nowadays, building projects are becoming more complex. The estimator needs to do more work but time is limited. Need for developing of a new approach and tool (i.e. the model) which can help the estimator is an interesting research topic. The model should need only simple and basic information but yield the same or better accuracy when compared to results obtained from the existing methods. The model should not be time-consuming and costly. Building the new cost model should also make use of historical records and computer technology.

Last Updated on Friday, 16 December 2011 14:44