Performance Evaluating Model for Construction Companies: Egyptian Case Study

Transcription

1 Performance Evaluating Model for Construction Companies: Egyptian Case Study Ahmed Elyamany 1 ; Ismail Basha, M.ASCE 2 ; and Tarek Zayed, M.ASCE 3 Abstract: The dynamic nature of today s construction industry compels construction partners to seek strategies in order to improve performance. Current research introduces a performance evaluation model for construction companies in order to provide a proper tool for a company s managers, owners, shareholders, and funding agencies to evaluate the performance of construction companies. The model developed helps a company s management to make the right decisions. Financial, economical, and industrial data are collected from Egyptian construction companies for nine consecutive years Five indices models are developed: company performance score, economy performance score, industry performance score, performance index, and performance grade. The models developed consider companies in four construction sectors: general building, heavy, special trade, and real estate. These models accommodate the effect of macroeconomic and industry related factors and company size on the performance evaluation. The final outcome of current research is a performance grade, which provides the performance of a construction company. The developed model is validated, which shows robust results. DOI: / ASCE :8 574 CE Database subject headings: Construction industry; Performance characteristics; Egypt; Construction firms; Evaluation; Case reports. Introduction Performance evaluation of construction companies gains its importance from the fact that today s world is moving rapidly toward globalization, which means the international General Agreement on Tariffs and Trade GATT. The concept of GATT is to conduct business anywhere, any time, and anyhow. In this environment, many multinational companies are awarded business in other countries in which they are competing with local companies. Both multinational and local construction companies should seriously look forward to improving their performance in order to maintain their international reputation. The current study develops performance evaluation of construction companies that shows the position of such a company among others. This evaluation is deemed essential for owners, shareholders, and funding agencies of a company because it clearly draws the correct position of the company. If the company position is good, this will increase the agencies interest in the company and vice versa. Many models 1 Ph.D. Student, Dept. of Civil Engineering and Construction, North Dakota State Univ., CIE 201, Fargo, ND ahmed. elyamany@ndsu.edu 2 Professor, Construction Management Program, Georgia Southern Univ., P.O. Box 8047, GA. ismailbasha@georgiasouthern.edu 3 Assistant Professor, Building, Civil, and Environmental Engineering Dept., Concordia Univ., 1455 De Maisonneuve West, Montreal QC, Canada H3G 1M7 corresponding author. zayed@bcee. concordia.ca Note. Discussion open until January 1, Separate discussions must be submitted for individual papers. To extend the closing date by one month, a written request must be filed with the ASCE Managing Editor. The manuscript for this paper was submitted for review and possible publication on February 9, 2005; approved on February 22, This paper is part of the Journal of Construction Engineering and Management, Vol. 133, No. 8, August 1, ASCE, ISSN / 2007/ /$ were developed to evaluate companies performance, but few of them consider economical and industrial changes in their models. Therefore, the major objective of the current study is to design a performance index that evaluates a company s financial situation within the construction industry. This model will consider the economical and industrial changes and their effect on the company. The current study presents a performance evaluation model based on financial ratios as well as economic and industry factors. The model developed considers four construction categories based on the Egyptian contractor union : 1 general building; 2 heavy; 3 special trade; and 4 real estate. It considers the effect of company size and economical and industrial variables on its performance. Companies that perform business across categories are not considered in the current study. Background A number of construction companies performance evaluation models have been developed along the previous 5 decades. They are dealing with this problem at three different levels: 1 construction industry; 2 company; and 3 project. Models at the construction industry level are used to measure the effect of economical, political, and social changes on the performance of the construction industry as a whole. Kangari 1988 relates the changes in construction industry failure rate to some macrocosmic factors: average prime interest rates, amount of construction activity, inflation, and new business entering the construction industry. Most performance evaluation models for construction companies are based on their annual financial statements or reports. Different analytical techniques have been used to develop these ratios: 1 financial statement trend analysis; 2 financial statement structural analysis; and 3 financial statement ratio 574 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT ASCE / AUGUST 2007

2 analysis. The most important variables that could be used in financial statement trend analysis to differentiate between failed and nonfailed companies are: accounts receivable, under-belling, accounts payable, notes payable, total long-term debts, stock and retained earnings, cost of sales, and gross profit Basha and Hassanein 1988; Severson et al Financial statement structural analysis determines the proportion that each company s group or subgroup represents in the financial statement Hasabo A decomposition ratio is used to determine changes in the percentage of the company s asset components in two consecutive years Hasabo The first statistical evaluation models to predict financial failure were developed by Beaver 1966 but they were not targeting the construction industry. Companies were divided into two main groups: failed and nonfailed companies with similar asset size. Statistical models that are based on financial ratios of failed and nonfailed companies represent financial statement ratio analysis. Beaver 1966 compared these two groups based upon 30 financial ratios in which there were differences in the mean values of these two groups for at least 5 consecutive years before failure, with differences increasing as the year of failure approached. Many studies followed Beaver 1966 that investigated multivariate techniques to distinguish failed from nonfailed companies. The Z score discriminate model dealt with companies that became legally bankrupt Altman Deakin 1972 used accounting data to perform multivariate discriminate analysis on bankrupt and nonbankrupt companies. This study concluded that most financial ratios showed discriminatory ability. The model of Kangari et al used multiple regression analysis to evaluate the performance of construction companies. This study developed a performance grade G curve in which the relative financial situation of any construction company, satisfying model limitations, could be determined. Severson et al based their model on discrete choice modeling to predict the probability of failure of a construction company in meeting its obligations in a construction contract. Russell and Zhai 1996 also used multiple regression analysis to predict construction companies failure and considered economic and financial variables together with their stochastic dynamics. Another quantitative model based on financial ratios was developed by Goda The model objective was to develop standard financial ratios that reflect the performance of the construction industry in Egypt. These standards could be used to compare the performance of the Egyptian construction industry with the international one. According to this study, regression analysis had provided more reliable results than that produced using the supervised neural network. Previous models that were developed by Kangari et al and Goda 1999 focused mainly on evaluating the performance using financial ratios without considering the effect of macroeconomic and industry related factors on the performance of construction companies. In addition, the model developed by Goda 1999 did not consider the effect of company size on its performance. Therefore, the newly developed model in current research accommodated the effect of macroeconomics and industry related factors and company size on the company performance. Data Collection for Case Study The financial data for construction companies were collected in the time period , from the Authority of Money Market, Egyptian Government in the form of: annual balance sheets Table 1. Values of Macroeconomic and Industry-Related Variables Macroeconomic variables Industry-related variables Year of record IFN ITR AWD AWS AWF and/or annual income statements. Out of thousands of construction companies working in Egypt, only 122 companies are registered in the Egyptian money market. Approximately 415 financial statements from 9 consecutive years were collected. Based upon the annual balance sheets and annual income statements, financial ratios of the Egyptian construction companies were determined. Only six financial ratios were considered in the model developed: 1 current ratio CR ; 2 total debt to net worth ratio TD/NW ; 3 fixed assets to net worth ratio FA/ NW ; 4 revenue to working capital ratio RV/WC ; 5 net profit to total assets ratio NP/TA ; and 6 net profit to net worth ratio NP/NW. According to the importance of each ratio in evaluating the performance of the financial references, the six ratios are chosen to represent the four ratio groups. Choosing only six ratios in the model was made in order to increase the reliability of the model using the most related ratios to the performance of the company. Economy data were collected from the Egyptian Ministry of Foreign Trade 2003a,b Egyptian government reports published quarterly. Data, as shown in Table 1, include: inflation rate IFN and average annual interest rate ITR as macroeconomic variables. On the other hand, average work capacity AWC, average work demand AWD, and their difference AWF are used as industry related variables. Data were collected for 9 consecutive years Model Development and Application to Case Study The model development process passes through various steps: data preparation, mathematical formulation, model building, and model validation. The flow of these steps is illustrated in Fig. 1. Financial Data Preparation The collected data should be prepared for mathematical models formulation usage. The process by which the data are prepared for these models is illustrated in the flow chart shown in Fig. 2. Normalizing Financial Data In fact, some financial ratios are calculated in terms of time while others are calculated in terms of percentages. Mathematical formulation of such types of data will result in a bias to the larger values of ratios Kangari et al In order to overcome this problem, normalizing the values of different ratios would make JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT ASCE / AUGUST 2007 / 575

3 Table 2. Normalization Coefficient fn for Various Construction Sectors Construction sector CR X 1 TD/NW X 2 Financial ratio FA/NW X 3 RV/WC X 4 NP/TA X 5 NP/NW X 6 General building Heavy construction Special trade Real estate Fig. 1. Flow chart of model development process them nonbiased and considered smaller value. Table 1 shows the normalization coefficient f ni of the general building construction companies sector. The normalized value X ni of standard ratio X si can be calculated by Eq. 1 as follows: X ni = S i fn i X si 1 Company-Size Effect Comparing the performance of a small company with the overall industry average is inappropriate because financial structure and characteristics of small companies are different from those of well-established, large companies Kangari et al This problem can be overcome by adjusting the normalized ratio value by size factor Z i. The size factor Z i for any single ratio X i is obtained by dividing the median of that ratio, for the whole construction sector, by its median in a similar subgroup size. Table 2 shows size factor Z i values for all ratios based on the company s total assets TA. Mathematical Formulation Regression analysis is used to develop the current model for the following reasons Goda 1999 : Its simplicity, reliability, and suitability for the problem under study; and Discriminate analysis is used to discriminate between failed and nonfailed companies. Although the unsupervised neural network technique seems to be suitable to the current problem, it needs a quite large data set for each single year. Mathematical Formulation of Company Performance Score Sc The company performance score Sc, according to Kangari et al. 1992, is defined as a performance grading system for assessing the position of a company within the overall construction industry and which is very difficult to be assigned a certain value. The Sc method is applied by Goda 1999 and the current study. Preliminary values of 100, 0, and 100 are assigned to the company performance score Sc for the upper, median, and lower quartiles of the previously prepared financial ratios as shown in Table 3, respectively. By using the regression analysis technique, both assigned values and the quartiles of the Sc for the previously prepared data are considered. The model is developed using multiple linear regression as shown in Eq. 2 6 Sc = C 0 + i=1 C i X ni 2 Company performance score Sc is assumed to have a value of 100, 0, and 100 for the upper, median, and lower quartiles, respectively see Table 4. The value of X ni can be calculated using Eq. 3 as follows: Fig. 2. Flow chart of data preparation process X ni = S i Zi f ni X si 3 Substituting Eq. 2 into Eq. 3, the final regression Eq. 4 is determined 6 Sc = C 0 + C i S i Zi f ni X si i= / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT ASCE / AUGUST 2007

4 Table 3. Size Factor Z i Value Based on Company s TA Ranges of Financial ratio Construction sector total assets millions CR TD/NW FA/NW RV/WC NP/TA NP/NW General building TA TA TA TA TA Heavy construction TA TA TA TA TA Special trade TA TA TA TA TA Real estate TA TA TA TA TA Table 5 shows the values of regression constant C 0 and coefficient C is of regression variables X is for each construction sector. The value of C 2, C 4, and C 6 except for the real estate sector has a negative sign. This can be interpreted as an inverse relation between these financial ratios and the company performance score Sc. Mathematical Formulation of Se and Si The economy performance score Se and industry performance score Si are developed using regression analysis. The available macroeconomic and industry related variables for 9 consecutive years of record are shown in Table 1. Economy Performance Score Se The development of economy performance score Se passes through five steps as follows: 1. Sort the economical variables: inflation and interest rate in ascending order from the best to the worst; 2. Assign nine values equal to 100, 75, 50, 25, 0, 25, 50, 75, and 100 for economy performance score Se within the 9 available years, respectively; 3. Use regression analysis to develop the regression Eq. 5 for Se Se = C 0 + C 1 X 1 + C 2 X 2 5 where, C 1, C 2 regression coefficient of variables X 1, X 2, respectively; X 1 inflation; and X 2 interest rate. The MINITAB statistical package is used to develop the required models. Statistical analysis showed that excluding the X 1 variable from Eq. 5 generates the best results Neter and Kutner 1996; Lapin 1983; Little Then, Eq. 6 is developed as follows: Se = ITR 6 Table 4. Quartiles of Financial Ratios for General Building Sector Year CR TD/NW a FA/NW RV/WC a NP/TA NP/NW a Upper quartile b Median quartile c Lower quartile a Ratios in percentages. JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT ASCE / AUGUST 2007 / 577

5 Table 5. Regression Constants and Coefficients Construction sector Regression constant C0 CR C1 TD/NW C2 FA/NW C3 RV/WC C4 General building Heavy construction Special trade Real estate NP/TA C5 NP/NW C6 4. Apply Eq. 8 on the values of economy variables; and 5. Normalize the calculated values of Se in Eq. 6 using Eq. 7 to be within the range as follows: Se mod = 200 Se Se min / Se max Se min 100 Industry Performance Score Si Similar to the economy performance score Se, the development of the industry performance score Si also passes through of five steps: 1. Sort the AWD in descending order and AWC in ascending order. This is because the best situation for a company occurs when market demand exceeds the supply i.e., AWD AWC become positive. 2. Assign nine values equal to 100, 75, 50, 25, 0, 25, 50, 75, and 100 for industry performance score Si within the 9 available years, respectively. 3. Using regression analysis, develop the regression Eq. 8 for Si as follows: Si = C 0 + C 1 X 1 8 Using regression analysis yields to Eq. 9 as shown below Si = AWF 9 4. Apply Eq. 11 on the values of economy variables. 5. Normalize the calculated values Si based on Eq. 9 using Eq. 10 to be within the range as shown below Si mod = 200 Si Si min / Si max Si min After normalization, the two models were developed as shown in Eqs. 11 and 12 Se = ITR Si = AWF Performance Index Model Building The performance index PI is developed by combining the effect of company, economy, and industry related factors. These factors are represented in the model using Sc, Se, and Si, respectively. The combination process was performed based on Hasabo 1996, which reported that the responsibility of company failure was carried out by three major factors: macroeconomic 35 40%, industry 10 15%, and company related factors 40 45%. These factors are used to formulate the PI. Macroeconomic, industry, and company related factors are represented by the normalized values of Se, Si, and Sc, respectively. The PI value can be determined from Eq. 13 as follows: PI = 0.5 Sc Se Si 13 When a company has the best Sc value Sc= +100 during a year that has the worst values of both Se and Si Se= 100 and Si= 100, it will be assigned the best value for performance index PI= This company might have a good financial performance during a fiscal year that has bad economical and industrial circumstances. In such a case, this company has a good financial and managerial performance; however, it is worth surviving in business. On the other hand, a company might have the worst Sc value Sc= 100 during a year that has the best values of both Se and Si Se= +100 and Si= This company will be assigned the worst value of performance index PI= 100. Therefore, the company had bad financial performance during a fiscal year that had good economical and industrial circumstances. Therefore, the company has weak financial and managerial performance that needs suitable remedial actions to survive in business. Development of Company Performance Grade The PI of a construction company should be compared to other companies in the same construction sector in order to know the relative situation of such a company within the industry. Performance grade G is the percentage of companies that have PI below that of the company under consideration Kangari et al Therefore, the G index is equivalent to the cumulative distribution function of the PI for all construction companies in the same construction sector. In order to calculate the distribution function for various construction sectors, a series of steps must be followed: 1. The result from the application of the new data sets on the Sc developed equation are considered. 2. Assume that the company performance score Sc values for the whole construction sector under consideration follow a normal probability distribution. This normal distribution has 99.7% of the area under curve falls within a distance =±3 from the mean. Assuming the arithmetic mean and standard deviation of Sc values for the whole construction sector are and, respectively, the prediction interval for the company performance score Sc would be calculated using Eq Sc p Values of company performance score Sc falling outside the interval are excluded, while the other values are then modified to be within a range of as shown in Eq. 15 Sc mod = 200 Sc Sc min / Sc max Sc min / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT ASCE / AUGUST 2007

6 Fig. 3. Values of performance grades G for different construction sectors Fig. 4. Company performance grade G for Elyasmin International for trade and contracting company 4. Using Eq. 13 and modified values of company performance score Sc mod, the PI values for various companies within the sector are computed. 5. The recent step is to calculate the probability distribution and then the cumulative probability distribution for the construction sector s companies. The cumulative probability distribution is known as performance grade G. In Fig. 3, a comparison between the G index values for different construction sectors is shown. According to G values, the pioneer position in the Egyptian construction industry belongs to the heavy construction sector with only 65% of its companies under PI=0. The models developed were applied to the Elyasmin International for Trade and Contracting Company. In Table 6, company performance score Sc, company performance index PI, and finally the performance grade G were determined for the 9 consecutive years. The G index values for the 9 consecutive years are shown in Fig. 4. The list of corrective actions related to various G index values is shown in Table 7. The company looked fine in the early 1990s; however, the G index curve started to decline in the mid- 1990s. It reached the lowest level bad situation between 1995 and 1997 G 20%. The situation started to enhance from 1996 to its high value in However, the G index value for the year 2000 was within the average performance range that needs considerable changes in management policies. This shows the power of the G index developed in assessing the company s performance for its management to take remedial actions. Table 6. Performance Grade G for Jasmine International for Trade and Contracting Year Sc Se Si PI G Model Significance and Validation The developed model has to be validated to test its prediction capabilities. The validation process is mainly concerned with the equation developed for the company performance score Sc. The collected data set is divided into model building 70% and validation 30% subsets. The validation data subset consists of approximately 139, 80, 74, and 142 observations for general building construction, heavy construction, special trade, and real estate Table 7. Management Courses of Action Suggested by Performance Grade Adapted from Kangari et al Performance grade G range % Management action 80 G G G G 40 0 G 20 Total management satisfaction; company policy is set on the ideal track; no adjustment actions required. No danger is anticipated in the near future, management policy is quite satisfactory, may need minor adjustment action. Company s performance is within the average performance range; management policy needs considerable changes; may be difficult to complete and keep the business afloat; financial trends are to be watched continuously. Company is in critical condition; typically due to inadequate financial management; immediate changes are required in company s policies; management should be changed if it fails to take quick recovery measures; if this situation continues for the next year, the company will fail. Company has reached lowest performance level in industry; very low probability that management can succeed in salvaging the company in this competitive business; the company has a high probability of bankruptcy in the near future; should consider going out of business. JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT ASCE / AUGUST 2007 / 579

7 Table 8. Validation of Developed Models Model building data subset results Validation data subset results Construction sector Mean Standard deviation Variance Mean Standard deviation Variance General building , , Heavy construction , , Special trade , , Real estate , , Average of sectors , , companies sectors, respectively. Results from the application of the validation data subset are compared to those of the application of the model building subset. Table 8 shows that the mean of the model results is around ; however the mean of the validation data subset is This means that the models developed are robust in representing various construction sectors with a validation of 93.18% /1.76. The average standard deviation for the model results is 79.13; however, it is for validation results, which shows more than 50% enhancement in the validation results. This also shows that the models developed are robust in representing construction sectors. In conclusion, based upon the results in Table 8, the models developed show acceptable results in general; however, the deviation is almost within the 10% range in various sectors. Summary and Conclusions Current research developed a performance evaluation model for construction companies Egyptian case study. A PI is developed using three performance scores: company financial Sc, economical Se, and industrial Si. The PI developed did not provide proper evaluation of the company performance relative to other competitors within the industry. Therefore, a company grade G index is developed using cumulative distribution of the PI values. The G index shows percentage of companies below the industry average and the situation of a specified company under consideration. According to regression analysis, inflation has no effect on the economy score Se. In addition, depending only on the volume of demand or supply on construction did not reflect the proper evaluation of industry performance score Si. The G index for the Egyptian construction industry shows that the pioneer position belongs to the heavy construction sector with only 65% of its companies under PI=0. The model developed is validated, which shows robust results. Notation The following symbols are used in this paper: AWC average work capacity; AWD average work demand; AWF AWD AWC; C 0 regression constant; C i regression coefficient of variable i; FA fixed assets; F ni normalization coefficient; IFN inflation; ITR interest; i integer subscript equal to 1 6 referring to the chosen financial ratio; NP net profit; NW net worth; PI performance index; RV revenue; Sc company performance score; Sc max maximum value of company performance score in sector; Sc min minimum value of company performance score in sector; Sc mod normalized value of company performance score; Sc p predicted value for company performance score Sc ; Se economy performance score; Se mod modified value of economy performance score; S i sign correction factor set equal to 1 if X si negative, and +1 otherwise ; Si industry performance score; Si mod modified value of industry performance score; TA total assets; TD total debt; WC working capital; X i regression variable of ratio i; X ni normalized value of ratio i; X si standard value of ratio I; and company-size factor. Z i References Altman, E. I Financial ratios analysis discriminate analysis and the prediction of corporate bankruptcy. J. Financ., 23 4, Basha, I., and Hassanein, M Measurement of construction projects profitability, Arab Contractor Press, Cairo, Egypt. Beaver, W. H Financial ratios as predictors of failure. J. of Accounting Research, Supplement 4, Deakin, E. B A discriminate analysis of predictors of business failure. J. Accounting Research, Supplement 10, Egyptian Ministry of Foreign Trade. 2003a. Quarterly economic digest, Egyptian Ministry of Foreign Trade publication, VII 4, Egypt. Egyptian Ministry of Foreign Trade. 2003b. Quarterly economic digest, Egyptian Ministry of Foreign Trade publication, IX 2, Egypt. Goda, A Assessment of construction contracting companies performance in Egypt. Ph.D. thesis, Faculty of Engineering Library, Zagazig Univ., Egypt. Hasabo, H. A Modern directions in financial and accounting analysis, Chicago University Press, Chicago. Kangari, R Business failure in construction industry. J. Constr. Eng. Manage., 114 2, Kangari, R., Farid, F., and Elgharib, H. M Financial perfor- 580 / JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT ASCE / AUGUST 2007

8 mance analysis for construction industry. J. Constr. Eng. Manage., 118 2, Lapin, K. W Statistical analysis for modern engineers, McGraw Hill, New York. Little, R. E Probability and statistics for engineers, Matrix Publishers Inc., Portland, Ore. Neter, J., and Kutner, M. H Applied linear statistical models, McGraw-Hill, New York. Russell, J. S., and Zhai, H Predicting contractor failure using stochastic dynamics of economic and financial variables. J. Constr. Eng. Manage., 122 2, Severson, G. D., Russell, J. S., and Jaselskis, E. J Predicting construction contract surety bond claims using contractor financial data. J. Constr. Eng. Manage., 120 2, JOURNAL OF CONSTRUCTION ENGINEERING AND MANAGEMENT ASCE / AUGUST 2007 / 581

9