外文翻译---不平衡报价的平衡.doc

不平衡报价的平衡
作者：Johan Nyström
翻译：杨德平
指导老师：王欲敏
摘要：基于轶事的证据，表明不平衡的报价存在一个严重的问题，在建筑行业。这一问题是基于与承包商比所述客户更明智的情况。由于信息不对称，。他可以运用接触到更多的单价而事前出价，以提高前利润，通过增加单价的数量，令单价上升，降低单价的数量，令单价减少。对于不平衡报价的研究很大程度上集中在模型上，他帮助客户和承包商拉小差距，不平衡报价的效率损失也有理论文献。模型表明一个明智的承包商提高单位价格相对低估的数量是合理的。然而，实证研究，捕捉到的问题的幅度是缺乏的。本文阐述了填补这一空白。该分析是基于一个独特的数据集的15瑞典道路投资和795 2观测。数据包括事前的单位价格和数量相关的最终（前）工程量。通过寻找这些变量之间的正相关关系和控制其他影响变量，实证检验假设不平衡报价。随着我们数据的早期研究，本文使用更多的项目具体的控制变量得出的结论是，即使是较小或不存在的影响。
关键词：公共采购; 道路建设; 不平衡报价; 单价合同

一、简介
不平衡报价是单价承包一个潜在的陷阱（UPC）。如果出现在一个建设项目中，这个概念会产生信息租金，因此效率低下，这是由客户支付太多的最终产品。不平衡报价来自承包商更好地了解然后客户端（i.e.asymmetric信息），他们的优势前使用。这种行为没有道德方面的。承包商根据客户提供的投标文件进行行为。相反，如果承包商没有对不合格的条形码然后他们必须回答本公司股东。这个概念通常被描绘成建筑业的一个大问题。专家们经常声称“这是如何在工业中做的”。这种看法主要是基于传闻。大部分学术论文都是理论上的，这说明了一个有针对性的承包商使用不平衡的投标是合理的。然而，有缺乏实证研究，支持这些索赔和理论论文。本文从实证角度出发，对不平衡报价的问题进行了实证分析。问题是，是否这种定价策略存在于瑞典公路建设市场的数据库中的15个瑞典道路项目与795个2个独特的观测是用来测试不平衡报价的存在。这是与北欧国家做统计方法的第一个大规模的研究。本文通过介绍不平衡投标的概念，其次是对概念的学术文献的一个简短的回顾开始。然后引入继续与用于测试不平衡出价是否存在的变量和数据，接着分析和结论..

二、不平衡报价的概念
文献中讨论的不平衡招标方式有2种：以“前装”的不平衡报价和基于信息租金的不平衡报价。共同的标准是，承包商是更好的通知比客户端，可以使用此信息的优势。休斯（1982）区分前者不平衡财务成本/现金流不平衡与不平衡误差的开发。现金流不包括承包商的标记数量，预计提前完成交易量完成后期价格（Arditi和chotibhongs，2009；skitmoreand卡特尔，2011）。
本文指的是不平衡的投标基于数量方面的错误，即利用信息不对称。通常的例子是采取单价合同（UPC），其中客户端提供量的法案在事前招标阶段。这与设计和建造合同（DB)，在那里承包商负责设计和施工, 而不是提供工程量清单，如何e.d.g.e.公路应建作为UPC，，客户描述在广义上讲是希望（简体，一路从A到B），其他的细节让承包商去思考。UPC仍然是建筑行业中最常见的合同。
假设有2个输入的道路，提供的碎石和路面。对工程量的事前估计100立方米的碎石和150平方米的路面。假定承包商有不同的成本和信息。承包商2表现出较高的边际成本，但有私人信息，但承包商1没有。承包商1投标她单位价格的边际成本10。承包商2可以使用她的上级信息关于事后数量和偏离单价做出应对。如表1所描述的，承包商2提交的总投标，并赢得了合同。

三、 不平衡报价的
研究对不平衡报价的最早的报纸包括盖茨（1967）和斯塔克斯（1974年），谁概念化概念？自那时以来，两家分支机构的理论模型，不平衡招标已经发展。
第一类模型旨在为客户检测提供实用的管理（Arditi和chotibhongs 2009），和承包商优化（卡特尔等人。2010；卡特尔等人。2008；者和界面1992），不平衡报价。这些都是实践模式，以帮助从业者在他们的日常工作与准备（承包商）和评估（客户）投标。第二类型的模型是针对虚拟读者与市场效率做的。这些都是典型的经济学模型，目的是预测投标行为，从而提高社会经济效益。最突出的两个模型包括他们和莱文（2001）和（2003）ewerhart和菲施勒。这两种模型都是基于客户和承包商之间的信息不对称，风险中立的承包商和最终的角落解决方案。这是指一种情况，承包商的手在零单价的所有数量，但大部分被低估的平衡。曼德尔和NyströM（2014）引入了风险厌恶模型和发现承包商投标行为的内部解决方案。这意味着，承包商标标中和极端的零中标。
本文将基于瑞典交通管理局收集的数据（Trafikverket）进行研究。数据库是基于15完成道路建设项目采购的Trafikverket。所有的数据都在所谓的MSS文件的形式收集。这是一个标准化的Excel工作表中使用所有的项目领导人Trafikverket。该表格包括所有项目中使用的数量，估计（事前）和最终（前）。它是在不同的单元定义；米，平方和立方。Trafikverket´项目领导人使用这个文件填写整个工程量。该文件还包括单位价格和其他信息，如项目的特点，附加的订单和变化。从这种材料中，我们能够创造出感兴趣的变量。变量的偏差被定义为以下方式

其中，数量qea 事后和事前，这给出了一个标准化的测量偏差这是必要的，因为单位之间的不同数量。当没有出现偏差时，该图呈现值为零。最低的可能的数字是- 1，这是当估计量没有被用来在所有。在单位价格的偏差被定义在相同的方式。
四、分析
基于上述数据，本文建立了测试的假设不平衡报价存在Trafikverket的项目。要确认不平衡投标的存在，应积极的关系之间的偏差和偏差的问题，这将表明，承包商提高单位价格低估的数量和较低的单位价格对高估量。图1给出了观测的散点图。

图不能确认或拒绝该假设肉眼。检验这一假说的方法特别是将在价格偏差与OLS回归如下数量偏差
在¨p方程1中定义的，¨q是方程2和µ定义的误差项。然而，没有显着的相关性是发现这些变量之间。
最后，包括项目的具体变量，如企业，地理位置的项目，类型的道路，项目的时间，道路和一年的项目开始。的变量是显着的，因此，在试图找到相关的偏差和偏差，在这个数据集失败。尽管与控制变量运行几个OLS，没有明显的关系。
五、结论
建筑行业专家学者的共识是：不平衡招标是一个巨大的问题。这主要是基于传闻证据。有缺乏实证研究，使用更大的数据集和统计分析方法，这一概念。Bajari et al（2014）认为，不平衡报价是不是在准备投标的主要决定因素。作为第一个瑞典研究的基础上定量分析，不平衡招标的问题，不能找到关于道路建设。重要的控制变量已被列入，但没有显着的相关性，可以发现在单位价格的偏差和数量的偏差。不平衡投标的实证支持缺乏的原因可能是几个。一个潜在的解释是，承包商没有更多的信息，客户端。这是影响投标的基本假设，这可能不会实现。另一种解释是准备倾斜投标的交易成本。这可能需要对计算机去完成所有的量和发现错估了很长一段时间。这种成本可以延长扭曲的预期利润。也使投标风险大于什么承包商愿意裸。

附原文：
The balance of unbalanced bidding
Johan Nyström

Abstract：Based on anecdotal evidence, claims are made that unbalanced bidding is a serious problem in the construction industry. This concept is based on a situation with a contractor being more informed than the client. The asymmetry in information can be used by the contactor to skew unit prices in the ex ante bid and in order to enhance the ex post profit. This is done by increasing the unit price of a quantity that is expected to go up and lower the unit price of a quantity expected to decrease. Research regarding unbalanced bidding has to a large extent focused on models to assists clients in detecting and contractors in optimising the skew. There is also theoretical literature on efficiency losses of unbalanced bidding. The latter models show that it is rational for an informed contractor to raise unit prices on relative underestimated quantities. However, empirical studies that capture the magnitude of the problem are lacking. This paper sets out to fill that void. The analysis is based on a unique dataset of 15 Swedish road investments and 2 795 observations. Data consist of ex ante unit prices and quantities that are related to the final (ex post) quantities. By looking for a positive correlation between these variables and controlling for other affecting variables, the hypothesis of unbalanced bidding can be empirically tested. Along earlier studies on US data, this paper, using more project specific control variables concluded that the effect is even smaller or non-existent.
Keywords:Public procurement; road construction; unbalanced bidding; unit price contracting
1、Introduction
Unbalanced bidding is one potential pitfall of unit price contracting(UPC). If present in a construction project,the concept generates information rents and therefore inefficiencies. This is manifested by the client paying too much for the final product. Unbalanced bidding comes from the contractor being better informed then client (i.e.asymmetric information), which the former uses to their advantage. There is no moral aspect to such a behavior. The contractor acts according to the tendering documents provided by the client. On the contrary, if the contractor were not act on substandard UPCs then they have to answer to the shareholders of the company the concept is usually portrayed as a big problem in the construction industry. Experts often claim that “this is how it is done in the industry”. This perception is mainly based on anecdotal evidence. Most of the academic papers are theoretical, showing that it is rational for an informed contractor to use unbalanced bidding. There is however a lack of empirical studies supporting these claims and theoretical papers.
This paper sets out to empirically determine the problem of unbalanced bidding. The question is whether this pricing strategy exist on the Swedish road construction market A database of 15 Swedish road projects with 2 795 unique observations is used to test the existence of unbalanced bidding. This is the first larger study with a statistical approach that is done in the Nordic countries.
The paper starts with an introduction to the concept of unbalanced bidding, followed by a short review of the academic literature on the concept. Then the introduction continues with the variables and data used to test whether unbalanced bidding exists, followed by the analysis and the conclusions.
2. The concept of unbalance bidding
There are two types of unbalanced bidding discussed in the literature; unbalanced bidding through “front loading” the bids and unbalanced bidding based on information rents regarding quantities. The common denominator is that the contractor is better informed than the client and can use this information to his advantage. Hughes (1982) distinguished the former type of unbalancing as finance cost/cash flow unbalancing in contrast to the error exploitation unbalancing. Cash flow unbalancing involves the contractors marking up prices on quantities that are scheduled for early completion trading off quantities for late completion (Arditi and Chotibhongs, 2009; Skitmoreand Cattell, 2011).
This paper refers to unbalanced bidding based on asymmetric information regarding quantities i.e. Error exploitation. The usual example is taken with a unit price contract (UPC), where the client provides a bill of quantities in the ex ante tendering stage. This in contrast to a Design and Build contract (DB), where the contractor is responsible for both design and construction. Rather than providing a list of quantities of how the e.g. road should be built as in the UPC, the client describes in broad terms what is wanted (simplified, a road from A to B) and let the contractor do the thinking. UPC is still the most common contract in the construction industry.
Assume that there are two inputs to building a road, provision of gravel and pavement. The ex ante bill of quantities for the project estimates 100 m3 of gravel and 150 m2 of pavement. Assume that the contractors differ in costs and information. Contractor 2 exhibits higher marginal costs on both inputs than Contractor 1 but has private information, which Contractor 1 does not. Contractor 1 bids her marginal cost at unit prices of 10. Contractor 2 can then use her superior information regarding the ex post quantities and skew unit prices accordingly. As depicted in table 1, Contractor 2 submits the lower total bid and wins the contract.

3. Research on unbalanced bidding
The earliest papers on unbalanced bidding includes Gates (1967) and Starks (1974), who conceptualized the concept. Since then two branches of theoretical modeling regarding unbalanced bidding have evolved. The first group of models aims at providing practical management for clients to detect (Arditi and Chotibhongs 2009), and for contractors to optimize (Cattell et al. 2010; Cattell et al. 2008; Yizhe and Youjie 1992), unbalanced bidding. These are practical models in order to help practitioners in their day-to-day job with preparing (contractors) and evaluating (clients) bids.
The second type of models are directed towards theoretical audience and has to do with market efficiency. These are typical models in economics, with the purpose of predicting bidding behaviour and thereby socio-economic efficiency. The two most prominent models include Athey and Levin (2001) and Ewerhart and Fieseler (2003). Both models are based on asymmetric information between client and contractor, risk-neutral contractors and end up in corner solutions. This refers to a situation where the contractors hand in zero-unit prices for all quantities but the
mostly underestimated in equilibrium (the example shown in table 3). Mandell and Nyström (2014) introduced risk aversion to this model and found an internal solution to the contractors bidding behaviour. This means that the contractor do not skew the bid to the extreme with zero-bids.
Hence, there are rational arguments for an informed contractor to skew the bid. This is not the same as the contractor actually doing it. There are some empirical papers on this matter. Bajari et al (2014) focus on the extent of adaptation costs in road construction but also include data on skewed bids. They find a statically significant result,that a 10 percent quantity overrun will raise the corresponding unit price with 0,5 percent. The authors conclude that unbalanced bidding is not a major determinant in preparing bids. Miller (2014) show that ex post revisions have an effect on the unit prices from subcontractors. The following sections will present and analyse unbalanced bidding in Sweden
The data and variables
To approach the empirical question whether unbalanced bidding is a problem, we want to correlate deviations in priced to deviations in quantities with control variables. This paper will be based on data gathered from The Swedish Transport Administration (Trafikverket). The database is based on 15 completed road constructions projects procured by Trafikverket. All data have been collected in the form of the so called MSS-file. This is a standardized excel sheet that all project leaders in Trafikverket use. The excel sheet include all quantities used in a project, both estimated (ex ante) and final (ex post). Quantitates are defined in different units; meter, square and cubic meters.Trafikverket´s project leaders use this file to fill in quantities throughout the projects. The file also includes unit prices and additional information such as project characteristics, additional orders and changes. From this material we were able to create variables of interest. The deviation in quantities is defined as following way

where, qep are quantities ex post and qeaare ex ante. This gives a normalized measurement of deviation, which is necessary since the units differ between quantities. When no deviations occur, the figure takes on the value zero. The lowest possible figure is -1, which is when estimated quantities has not been used at all. Deviations in unit prices are defined in the same way。
4、Analysis
Based on the data presented above, this paper set out to test the hypothesis that unbalanced bidding exist in Trafikverket’s projects. To confirm the existence of unbalanced bidding, a positive relationship should prevail between deviations in p and deviations in q. This would indicate that contractors raise unit prices on underestimated quantities and lower unit prices on overestimated quantities. The scatter diagram of the observations is presented in Fig. 1.

Fig cannot confirm or reject the hypothesis by the naked eye. The method to test this hypothesis more specifically is to correlate deviations in prices with deviations in quantities with an OLS regression as follows,

，

where ¨p is defined in eq. 1, ¨q is defined in eq.2 and µ is the error term. However, no significant correlation is found between these variables.
Finally, project specific variables such as firm, geographical placement of project, type of road, project time,length of road and year of project start were included. None of the variables were significant.Hence in trying to find correlation between deviations in p and deviations in q this dataset fails. Despite running several OLS with control variables, no significant relationship could be found.

5、 Conclusion
There is a consensus among experts in the construction industry that unbalanced bidding is a huge problem. This is mainly based on anecdotal evidence. There is a lack of empirical studies using larger datasets and statistical analysis to approach this concept. Bajari et al (2014) conclude that that unbalanced bidding is not a major determinant in preparing bids. As the first Swedish study based on quantitative analysis, the problem of unbalanced bidding cannot be found regarding road construction. Important control variables have been included but no significant correlation can be found between deviations in unit prices and deviations in quantities. Reasons for the lack of empirical support for unbalanced bidding can be several. A potential explanation is that the contractors do not have more information that the client. This is the underlying assumption for skewing bids, which might not be fulfilled. Another explanation is the transaction cost for preparing skew bids. It might take a long time for the calculator to go through all quantities and find misestimating. This cost may extend the expected profit of skewing. Also the risk of skewing bids are larger than what the contractor is willing to bare.

Available online a