Analyzing Tender Selection Using a Geometric Graph Analysis (GGA) Approach

Analyzing Tender Selection Using a Geometric Graph Analysis (GGA) Approach

Kwo-Wuu Wang[1]and Wen-der Yu[2]

1Department of Civil Engineering, Chung Hua University, No.707 Sec. 2 WuFu Rd. Hsinchu, 300 Taiwan R.O.C.

2Institute of Construction Management, , Chung Hua University, No.707 Sec. 2 WuFu Rd. Hsinchu, 300 Taiwan R.O.C.

Abstract

As the projects and products become more and more versatile and complex, the procurement approaches have also diversified. Traditional competitive bidding method that based on lowest price (also named Lowest Tender, LT) usually resulted in high competition and low quality. As a result, the most advantageous tender (MT) was brought into the Government Procurement Law (GPL). After several years of practice, the public procurement agencies are still wondering why and how to select LT or MT. This paper proposes a Geometric Graph Analysis (GGA) model adopted 2-D analytical graph for modeling the two most important factors in procurement, i.e., price and quality to develop an analytical tool for answering the abovementioned questions. It is found that the proposed GGA provides not only a theoretical framework for analysis of suppliers’/contractors’ behaviors during procurement process, but also an efficient tool for procurement strategy planning.

This paper proposes a Geometric Graph Analysis (GGA) approach for the following analysis of procurement process such as: (1) the price effect of the lowest tender (LT) on quality; (2) the selection of tenders and (3) the appropriate contract awarding zone. Several application examples are demonstrated to show the applicability of the proposed GGA.

Keywords: Tender selection, most advantageous tender, lowest tender, procurement policy

Introduction

As the projects and products become more and more versatile and complex, the procurement approacheshave also diversified. Traditional competitive bidding method that based on lowest price (also named Lowest Tender, LT) usually resulted in high competition and low quality, which is not sufficient for selecting the most appropriate tender in the diversified procurement requirements nowadays. As a result, the most advantageous tender (MT) was brought into the Government Procurement Law (GPL) since 1999 inTaiwan. In contrast to LT, the MT selects the tender that is the most advantageous to the project owner. After severaln years of practice, the public procurement agencies are still wondering why and how to select LT or MT, the pros and cons of LT and MT, and the reason of poor procurement? It is very desirable to develop an analytical tool for answering the abovementioned questions.

In this paper, a Geometric Graph Analysis (GGA) approach is proposed to: (1) explore the price effect of LT in the procurement; (2) to analysis of the different procurement methods such as LT and MT; and (3) to compare the efficiency of tender selection between LT and MT. The proposed GGA is an analytical approach that expresses two most important factors, price and quality, in a Cartesian Axis. The behaviors of the demander (buyer of project owner) and the supplier (contractor) are modeled and analyzed with GGA. In order to demonstrate the applicability of the proposed GGA, case studies of real world procurement examples are conducted. The rest of this paper is presented in the following manner: the related researches are reviewed first followed by the definition of terms utilized in the paper; then, the theoretical framework of GGA is proposed; application to real world procurement cases is demonstrated in the fifth section; finally, discussions and conclusions are described.

Literature Review

According to the definitions of Taiwan GPL(Article 5213), the methods of contract awarding include LT and MT [1].The LT methodassumes that plans and designs are detailed and clear enough, so that the contractors won’t mistaken during cost estimation. As a result, the contractor with best management capability can offer the service or product with lowest price and thus wins the bid.According to Gransbergand Ellicott [2], LT creates a business relationship based on a single factor, price. He also addressed that LT can simplify the process of selection, solicitation preparation and tender review. It is assumed that the quality of LT can be assured via auditing and inspection procedure [3]. Such presumption however did not work out in practical excises of GPL in the past seven years.Kashiwagiand Al-Sharmani[4] pointed out the weakness of LT that it favors contractors who can provide minimum quality and performance.In LT, no credit is given to who provides more skilled craftspeople, better construction methods, and higher quality materials. Nissenfurther the points that LT makes flawed assumptions, encourages cost-cutting underperformance and does nothing to screen out unscrupulous contractors[3].

The USFederal Acquisition Regulation (FAR) 15.605 states: “(in addition to price)…quality also shall be addressed in every source selection” [5]. In the same article, it also states: “While the lowest price or lowest total cost to the government is properly the deciding factor in many source selection, in certain acquisitions the government may select the source while proposal offers the greatest value to the government in terms of performance over and above the minimum acceptable level will enhance mission accomplishment and be worth the corresponding increase in cost [6].” The above regulations have led to the most advantageous tender (MT), which shifts the competition point from price to quality. The “quality” is defined as the merits that can improve utilities of the buyer/project owner. The items of the abovementioned merits may include technical advantages, previous performance, safety practice, local experience, worker training, and price. In many cases of MT, the price is fixed, and thus the most advantageous is determined by the best quality offered by the suppliers/contractors. Since the fixed price of the project should be developed early, it is found that change orders and litigations were reduced dramatically in MT [3]. According to Taiwan GPL (Article 52), the MT method can only be adopted when a procurement contract may result in discrepancies in technology, quality, function, efficiency or the implementation of commercial terms, etc., as it is carried out by different suppliers/contractors [17].

Kashiwagi[78] proposed a 2-D figure (as shown in Fig. 1) to model the competition and performance of contractor selection methods in construction industry. Due to the increasing competitionworldwide, the quadrant Ι and Π in Fig. 1are becoming more and more popular for construction industry. Quadrant I is related to low performance and high competition, which is closed to the LT. Quadrant II is related to high performance and high competition, which is closed to the MT. Even though subjective comments were found in literature, analytical tool for analysis of contractor selection methods is still not available.

Definition of Terms

For the convenience of model construction of the proposed GGA, the abbreviations and symbols adopted in this paper are defined in this section and summarized in Table 1. Note that the abbreviations and symbols with a superscript “*” denotes the theoretical curves or ranges that the suppliers/contractors do not participate in the bidding of public procurements; while the ones without “*” denotes the curves or ranges that the suppliers/contractors participate in the bidding of public procurements.

Fig. 1 Constructor selection in construction industry [78]

Definition of Terms

Table 1:Definition of abbreviations and symbols

Abb./symb. / Definition
point (x , y ) / Point(x, y) represents a tender with a specific set of price and quality related to the service/product provided by the supplier/contractorunder rational basis.
Price / The bid price offered by a supplier/contractor in conformity with the requirements of the buyer/project ownerunder rational basis.
Quality / A broad definition of quality/merit that improves the utilities of the buyer/project owner including everything except the price during the trade.
PB / The ceiling price of buyer//project owner, usually relates to the buyer’s/project owner’s budget.
PL / The bottom price (the lowest allowable price) set up by the buyer/project ownerin order to avoid poor quality of obtained service/product.
Pi / A certain price offered by the supplier/contractor lying between PMIN and PMAX, i =0, 1…n. Pi can also be regarded as any possible price that the buyer is willing to pay, thus, Pi will lie between PL and PB.
LT / The lowest tender
MT / Most advantageous tender
SS* / The curve of all possible tenders in the free market without constraints of GPL
SS / The curve of all possible tenders awarded by the principle of contract awarding of GPL
HQ* / The curve of the highest-quality tenders under rational pricein the free market
LQ* / The curve of the lowest-quality tenders under rational price in the free market
HQ / The curve of the highest-quality tenders awarded by the principle of contract awarding of GPL
LQ / The curve of lowest quality tenders awarded by the principle of contract awarding of GPL
Q*MAX / The highest-quality tenders of service/product in conformity with the requirements of the buyer/project owner in the free market
Q*MIN / The lowest-quality tendersof service/product in conformity with the requirements of the buyer/project owner in the free market
QMAX / The highest-quality tenders of service/product in conformity with the requirements of the buyer/project owner in the public procurement
QMIN / The lowest-quality tendersof service/product in conformity with the requirements of the buyer/project owner in the public procurement
P*MAX / The highest-rational-price tender in the pool of Q*MAX
P*MIN / The lowest-rational-price tenderin the pool ofQ*MIN
PMAX / The highest-rational-price tender in the pool of QMAX
PMIN / The lowest-rational-price tender in the pool of QMIN
W* / The pool of suppliers/contractorsin a free market, which is selected by the buyer/project owner. The region matches buyer’s/project owner’srequirements andis encircled by curve HQ*, LQ*, Q*MIN, and Q*MAX.
W / The pool of suppliers/contractorsselected by the public buyer/project owner. The region matchesthe public buyer’s/project owner’srequirements andis encircled by curve HQ, LQ, QMIN, and QMAX.

Theoretical Framework of Geometric Graph Analysis (GGA)

4.1 Basic Model

As shown in Fig. 2, the Proposed GGA model is a two-dimensional graph with the X-axis presenting the broadly-defined “quality” and Y-axis presenting the “price” of the service/product provided by the contractor/supplier. There are two curves in Fig. 2: (1) LQ*—the lowest-quality tenders under rational price in the free market; (2) HQ*—highest-quality tenders under rational price in the free market. Both of the curves are subject to the rational price set up by the contractor/supplier in conformity with the requirements of the buyer/project owner.Every point (x, y) in Fig. 2 represents a tender with a specific set of price and quality related to the service/product provided by the supplier/contractor. On the price-axis (Y-axis) there is anupper-bound on curve LQ*, P*MAX, which stands for the highest rational price tender of the Lowest-quality service/product providers; similarly, there is a lower-bound on curve HQ*, P*MIN, which stands for the lowest rational price tender of the Highest-quality service/product providers. The upper-bound and lower-bound are also found on the quality-axis (X-axis), which are Q*MAX, andQ*MIN respectively. In constructing GGA, two assumptions should be made beforehand:

Assumption 1: Rational Price

The rational price is set up by the supplier/contractor based on the requirements defined by the buyer/project owner and efforts and costs he/she must spend to meet the requirements.

Assumption 2: Continuity of curveCurve

It is assumed that in the considered range of GGA, there is at least one supplier mapping to a tender point (x, y). In real world, the tenders on the curves in GGA graph may not be continuous but discrete.

It is noted that the “Quality”in GGA is the broadly-defined quality that may improve the utilities of the buyer/project owner except price[9888]. On the other hand,the“price”isdetermined on rational basis, which generally takes into account all influential factors such as direct cost, indirect cost, profit margin, competition situation, cost and fixed price of product performance, market price and competitor, etc. It implies that the rational price is closely related to the quality; as the quality increases the rational price increases accordingly[910]. Such relationship is shown in Fig. 2, there HQ* is a concave curve, which means the marginal increase of quality is less than the increased cost. Similarly, LQ* is a convex curve, which means the marginal increase of quality is greater than that of the cost. Such definition conforms to that of Value Engineering (VE), where the marginal quality increase is less than the increasedcost in the high-quality area (meaning “low-value” product or “redundant quality”). Moreover, the HQ* curve is always right and lower to the LQ*. It means that, for thesamequality level, the tender on the HQ* is always lessexpensive than that on the LQ*.

Fig. 2 GGA graph

4.2 Supplier Area (W, W*)

The region GEF*I in Fig. 2 bounded by curves HQ*and LQ* represents the area where the potential qualified (met with the buyers’/owners’ lowest requirements on quality, Q*MIN) tenderswhich is the biggest assembling of buyer’s considered selection are distributed. This potential tender area is defined as the “Supplier area in the free market (W*)”. W represents the largest set of tenders that the buyer/project owners can select in a free market. The other region DEFHIC in Fig. 2 represents another set (W) of potential qualified tenders under the regulation of GPL. DEFHIC is bounded with the ceiling price of buyer//project owner (PB) and bottom price set up by the buyer/project owner (PL).Under special conditions that PL =P *MIN and PB =P *MAX, W is equal to W*.There are four regions in W*: (1) low-price/low-quality area (LL); (2) low-price/high-quality area (LH); (3) high price/high-quality area (HH); and (4) high-price /low-quality area (HL). The tenders located in LH are best candidates for selection in procurement. LL and HH are second choices, while tenders in HL are those should be avoided by the buyers/project owners.

4.3 Tender Classification and Supplier/Contractor Selection

Before taking into account the competitive bidding, let consider a certain price, P0, all tenders on line are potential candidates for selection. While relaxing single price point P0 to a price range [PL,PB], the best choices are tenders on curve EF and the worst choices are on curve AC. In practical application, the distance between LQ* and HQ* or their shapes on the graph will be different according to the different procurement items. The buyer/project owner often classifies the procurement itemsinto A, B, C classes (see Fig. 3) according to the amountin stock, quantity for procurement, the levels of price or value[89]. The items inClass A mean small quantity, high price, and great value products, which is the “important minority”in procurement, e.g., the high-tech equipment. On the contrast, items in Class C mean that large quantity and cheap of little money, are the insignificant majority in procurement, e.g., standard products.

Fig. 3 Classification of procurement items

In practical tender selection, Classes A and B require more attentions than those of Class C. The requirements of security, reliability, post-sale service, etc., are higher. The flexibility of Classes A and Bare also greater than that of C. As a result in procurement of tenders in Classes A and B,the ranges of [P*MIN, P*MAX] and [Q*MIN, Q*MAX] are wider than that of Class C. The shapes of curves LQ* and HQ* in procurement of tenders in Classes A and B is relatively high, wide, and steep, such as those shown inFig. 4. The typical shape of curves LQ* and HQ* in procurement of tenders in Classes Cis shown inFig. 5.

Fig. 4 Typical GGA graph of Classes A and B /
Fig. 5 Typical GGA graph of Classes C

Fig. 4 Typical GGA graph of Classes A and B

In this paper, procurement of tenders in Classes A and B are considered. It is also assumed that the curves LQ* and HQ* intersect each other on the origin.

Fig. 5 Typical GGA graph of Classes C

In this paper, procurement of tenders in Classes A and B are considered. It is also assumed that the curves LQ* and HQ* intersect each other on the origin.

In selection under the “performance specification”, the range between curves LQ* and HQ* is also relatively wide while compared with those of the technical (prescriptive or descriptive) specifications. It is because that the technicalspecifications specify the quality of procured items more precisely. Usually, the procurements are performance-basedfor tenders in Classes A and B, and technical-based for those in Class C.

4.34 Adaptation for Practical Application

The GGA model proposed in this paper is a theoretical framework for procurement analysis. It needs to be modified before application to real world situation. Before the modification, one assumption (Assumption 3) should be made first:

Assumption 3: Sequential Order of Quality

It is assumed that all tenders in the GGA graph are discriminable in the orders of their quality levels, where the Transitive Law holds. That is, any three different tenders (A, B, and C) in procurement can be ordered according to their quality levels. If the quality of tender A is better than that of tender B and the quality of tender B is better than that of tender C, then the quality of tender A is better than that of tender C.

In practical application, a GGA graph should be constructed like the one shown in Fig. 6. Based on Assumption 2, the curves in GGA are assumed to be continuous. The tender price and quality information collected from the market place is finite. The dots “*” in Fig. 6 represents the potential suppliers tenders(tenders) in the marketplace. The upper and lower frontiers of data envelope form the LQ* and HQ* in GGA graph. Interpolation and extrapolation can be employed for continuous curve construction. Sometimes, ordering of the qualityof tenders is difficult. Semi-quantitative methods, such AHP[1112], can be adopted for quality ordering. For example shown in the Fig. 6,if the buyer is willing to purchase item within budget of $100, theoretically the buyer would neverselect the supplierwith $120 offer. Only when the supply lowers down his/her priceto $100 (in other words, to cut the prices with $20) can be selected by the buyer.