for any budgetary quarter. Finally, general and admin- istrative costs are the cost of top executives' salaries and centralized plant computing facilities. The labor/ overhead rate for any cost center may be described by LOH CC L D E R B G8A 2 where LOH CC The labor=overhead rate for a specific cost center in dollars=hr L D The direct labor rate; dollars=hr E R The expense rate; dollars=hr B Burden in dollars=hr G&A General and administrative cost rate in dollars=hr The dollar amounts for burden and general adminis- trative costs expected to be expended during any quarter are divided by the total number of direct labor hours to be expended to determine the burden and G&A cost rates in Eq. (2). The mechanics of the capital budgeting process are best illustrated with an example. This example is the subject of the following section and has been adapted from Ref. 5. 2.5.2 A Capital Budgeting Example Consider a manufacturing plant with a total of 1000 employees. Suppose it is desired to determine the labor/overhead rate for the machining cost center 22. For example purposes we will assume that cost center 22 has a total of 200 employees. Of this total we will further assume that 150 are involved with direct labor activities. To begin our analysis, we need to know the rest of the cost centers that exist and the respective number of employees in the plant that are associated with them. ThesestanglevelsareillustratedinTable5.InTable 5, there are a total of four cost centers on which direct labor is performed. These include cost centers 21, 22, 23, and 42 which are Manufacturing Engineering, Machining, Assembly, and Inspection respectively. 856 Malstrom and Collins Figure 3 Revised cost estimate grid. Copyright © 2000 Marcel Dekker, Inc. labor activities. The total indirect costs to be recovered are summarized in Table 7. Column 6 in Table 7 includes no entries for cost centers 21, 23, and 42 because these indirect costs will be captured in their entirety when the budgeting process is repeated for these cost centers. There is no entry in the same column for cost center 22 because the procedure assumes that cost center 22 must pay for all of its own indirect costs. This is applicable for all cost centers that include direct labor activities. The expense rate, E R , for the cost center may now be determined. The total of the entries in column 6 of Table 7 is $1,599,916. The indirect costs on cost center 22 ($350,000) are added to this total. This sum is divided by the total number of direct labor hours expected to be worked in cost center 22 during the quarter (72,000 hr. The result is E R 6350,000 61,599,916 72,000 hr 627:08=hr 2.5.5 Burden Determination There are two types of burden rates to be considered. The ®rst is the burden for cost center 22 itself. The second portion is for all of the pure indirect labor cost centers. As with the expense rate, the burden on cost center 22 must be recovered in its entirety during the quarter. The burden for the rest of the indirect labor cost centers is charged proportionally to cost center 22 on the basis of the total number of employees on each of the direct labor cost centers. We assume that the burden on cost center 22 is $500,000 and that the total burden for the quarter on all of the indirect labor cost centers is $1,000,000. The burden rate then is calculated as Total Burden $500,000 $1,000,000 Â 200=550 6863; 636 Burden Rate B 6863,636=72,000 hr 611:99=hr 2.5.6 G&A Determination We assume that the total dollar value of G&A to be recovered for the budgetary any quarter is $400,000. This amount is prorated on the basis of total direct employees and is charged to cost center 22 as 6400,000 Â 200=550 6145,440 Therefore, the G&A rate per hour is G8A 6145,440=72,000 hr 62:02 hr 2.5.7 Hourly Rate Determination and Adjustment The four components of the labor/overhead rate for cost center 22 have now been determined. These com- ponents may now be summed as speci®ed by Eq. (2) 858 Malstrom and Collins Table 7 Proration of Indirect Costs for Cost Center 22 24 1 Average 3 Number of 5 6 Cost indirect Column 2 indirect Column 3 Â Column 5 Â center salary ($) Â1=4 ($) employees column 4 ($) 200/500 ($) 21 32,000 8,000 50 a 400,000 22 28,000 7,000 50 a 350,000 b 23 28,000 7,000 50 a 350,000 30 32,000 8,000 50 400,000 145,440 41 36,000 9,000 50 450,000 163,620 42 28,000 7,000 20 a 140,000 43 32,000 8,000 50 400,000 145,440 50 36,000 9,000 50 450,000 163,620 60 40,000 10,000 100 1,000,000 363,636 71 44,000 11,000 100 1,100,000 400,000 72 48,000 12,000 50 600,000 218,160 Total 1,599,916 a Number of indirect employees out of total on cost center. b Total must be recovered in its entirety by cost center 22. Copyright © 2000 Marcel Dekker, Inc. to determine the labor/overhead (LOH) rate for cost center 22: LOH 22 610:00=hr 27:08=hr 11:99=hr 62:02=hr 51:09=hr This procedure is repeated for the remaining direct labor cost centers, 21, 23, and 42. Each LOH rate obtained is an estimate or forecast. This is because the dollar values of indirect labor, burden, and G&A, are estimated values. Likewise, it was estimated that 72,000 hr would be worked during the quarter on cost center 22. As the budgetary quarter is completed, actual values of indirect labor, burden, G&A, and the number of direct labor hours worked will dier from the estimated values used in the LOH rate computa- tion. This is also true of the average direct labor wage rate used in Eq. (2). Cost engineering personnel will collect actual values for each of these parameters at the end of the quarter. This will permit an actual LOH rate to be determined. This actual rate will be either greater or less than the budgeted rate prior to the quarter that was calculated above. If the budgeted rate is greater than the actual rate, excess costs will be recovered. If the budgeted rate is less than the actual rate, insucient costs will be recovered. A common practice is to compare these budgeted and actual rates for all direct labor cost centers over several quarters. Subsequent estimates for the para- meters in Eq. (2) may be adjusted upward or down- ward, as appropriate, in subsequent budgetary periods to make the actual and budgeted LOH values more closely approximate one another. 2.6 ACTIVITY BASED COSTING Activity-based costing (ABC) has gained increased popularity in recent years as a more accurate alterna- tive to conventional careful budgeting methods for some manufacturing activities [14]. Some cost analysts have questioned the accuracy and validity of the use of direct labor hours as a proration basis variable in cost recovery. Some organizations have found that conventional capital budgeting cost recovery methods are not ver- nier enough to accurately estimate overhead costs and indirect labor amortization. Implementers of ABC ®nd that their organizations have often been grossly under- charging their customers for small-volume production end items. High-volume items often have in¯ated pro- duction cost values with conventional cost recovery methods. 2.6.1 Cost Center De®nition With ABC, the number of cost centers is usually expanded to encompass small work cells, or even indi- vidual machines. Consider a work cell that contains one new and highly automated ®ve-axis machining center. Suppose this cell contains ®ve other manual, conventional machine tools. With this work cell de®nition, parts processed by the work cell might not be accurately charged for the manufacturing procedures they require during the fabrication process. The overhead rate for the cell will seek to recover the costs of all of its machines. Parts entering the cell that require processing only by the cheaper, conventional machines, might be unfairly charged overhead for the automated machining center, even though they did not require processing by this piece of equipment. A better cell de®nition might be to de®ne the auto- mated machining center as its own work cell. The older conventional machines might be grouped together in a second work cell. This philosophy can result in more than one hundred separate cost centers, many of which may consist of only one machine. This type of cost center de®nition enables the cost of manufactured end items to be determined as a function of those manufacturing procedures arid services they consume or require during their manufacturing process sequence. This is true for both direct and indirect labor and materials. The cost recovery methodology for ABC is the subject of the subsection that follows. 2.6.2 ABC Cost Recovery Methodology Activity-based costing de®nes a cost rate per unit time that is associated with each de®ned work center or center. This rate has been described by Ramachandran et al. [7, 8] and is illustrated by R D;i C DL C D C U C FS C T C I C IL C O H B 3 where R D;I Hourly operation cost of direct labor work center i in dollars per hour C DL Cost of direct labor in the work center over the budgetary period in dollars C D Cost of depreciation for equipment in the Manufacturing-Cost Recovery and Estimating Systems 859 Copyright © 2000 Marcel Dekker, Inc. work center during the budgetary period in dollars C U Cost of utilities attributable to the work center during the budgetary period in dollars C FS Cost of building ¯oor space attributable to the work center during the budgetary period in dollars C T Cost of taxes attributable to the work center during the budgetary i period in dollars C I Cost of insuring the work center's equipment and ¯oor space during the budgetary period in dollars C IL Cost of indirect labor required to support the work center during the budgetary period in dollars C O Cost of fringe bene®ts, other overhead, and any supplementary indirect labor wages required by the work center during the budgetary period in dollars H B Estimated or capable number of hours during the budgetary period that the work center is expected to operate Some of the parameters that make up Eq. (3) require interpretation. Suppose a total of m direct labor employees work in a given work center, i, during a budgetary period. The direct labor cost can then be described by C DL m j1 L j N j;i 4 where: L j Hourly pay rate including beneEts for direct labor employee j in dollars per hour N j;i Number of hours worked by employee j in work center i during the budgetary period 2.6.3 ABC Operation on a Quarterly Basis The process of assessing the accuracy of work center rates using Eq. (3) is similar to the capital budgeting process previously described. Prior to a given budget- ary quarter, estimates of all parameters is Eq. (3) must be compiled to,determine a budgetary or forecast cost rate for all direct labor work centers. At the end of each quarter, estimated parameters in Eq. (3) are com- pared with actual values at the end of the quarterly budgeting period. This enables an actual work center cost rate to be determined. If the actual rate is less than the estimated or bud- geted rate, the plant has overcharged for the services of the work center. If the reverse is true, the plant has undercharged products for the work center's use. Comparing actual and projected work center rates on a quarter by quarter basis gives the cost analysis some basis for adjusting follow-on estimates for each of the parameters in Eq. (3). This facilitates the compilation of more accurate budgeted work center rates in future budgetary quarters. 2.7 QUICK-RESPONSE ESTIMATING AND ESTIMATING SOFTWARE Manufacturers often ®nd it necessary to supply cost estimates with extremely short lead times. Quick- response estimating systems rely on parametric esti- mating techniques. Parametric estimating entails the development of mathematical relationships that relate the product's manufacturing cost to salient, identi®- able features of the product being manufactured. Other factors considered in the estimating equation are production quantity and the number of times the end product has been previously manufactured. 2.7.1 Development of Estimating Equations Multivariate regression analysis is a tool that can be used to develop parametric estimating relationships. Both linear and nonlinear equations can be ®tted to historical cost data using this approach. Many spreadsheet packages currently permit multivariate regression analysis to readily be completed on a perso- nal computer. It is important to assess the quality of obtained curve ®ts that relate product cost to features of a the end item being produced. A coecient of determina- tion (r 2 ) of at least 80% is recommended. It is also important that developed estimating relationships be validated before being used. This may be accomplished by dividing historical cost data bases into two parts. The parametric equations can be developed using the ®rst half of the historical data. The developed relation- ships can then be ``tested'' on the second half, compar- ing projected cost from the obtained equations, with actual historical costs that were incurred in practice. 2.7.2 De®nition of Part Families Estimating relationships should be developed by part family. For example, cost data used to make an FM 860 Malstrom and Collins Copyright © 2000 Marcel Dekker, Inc. radio should obviously not be used to estimate the cost of an automobile drive shaft. Group technology (GT) de®nes manufacturing production cells that are dedi- cated to producing parts with geometrically similar attributes. Coding and classi®cation methods are used to de®ne families of parts that are manufactured by GT cells. Often, parts within de®ned GT families are candidates for the use of the same parametric cost relationship. Group technology lends itself most readily to metal- working operations. For electronics manufacturing, part family de®nition must be more intuitive. Example part families for electronics manufacturing might be as de®ned below. Point-to-point wired cables Printed circuit board assemblies Wired chassis assemblies to interface printed circuit boards Interface wiring between separate chassis assemblies Metalworking operations to punch and machine the chassis prior to interfacing wiring. 2.7.3 Production Mix Parametric estimating systems are much easier to implement in high volume, low mix types of produc- tion environments. Organizations should expect to make an investment on the order of person-months to person-years in terms of statistician and cost engi- neering time to develop and validate such relationships prior to their being used. 2.7.4 Development and Use of Estimating Software In many cases, software is developed to assist in expediting the completion of cost estimates and order quotations to customers. In most cases, it is more expe- ditious to generate such software ``in-house'' as opposed to procuring commercially available software packages. Such software must have the organization's product line ``embedded'' within its architecture. This require- ment often makes it necessary to modify commercial software packages. Required modi®cation costs can be extensive enough to justify ``company-generated'' soft- ware. This justi®cation is enhanced by the fact that many commercially available estimating software packages are extremely expensive. Commercially avail- able software is also usually di cult to interface with the user organization's historical cost records. Parametric estimating relationships, if previously developed by the organization, are easily incorporated into estimating software that is written ``in-house.'' In- house software is not necessarily less expensive, but is almost always more useful to the organization in the long run. 2.8 IMPACT OF AUTOMATION ON ESTIMATING PROCEDURES Manufacturing automation has, and continues to have a dramatic eect on conventional estimating proce- dures. High levels of automation reduce the amount of human direct labor associated with the manufacture of parts. Overhead costs are increased, usually due to the procurement of expensive automated machining centers, industrial robots, and assembly equipment. Indirect labor costs tend to increase due to program- ming and maintenance costs. Product quality is generally enhanced as a result of automation. This has a positive eect in reducing war- ranty and return costs. Cost variation between parts and groups of parts is greatly reduced. Machine cycle times are more constant, and except for unscheduled maintenance, are functions of programs for the machines and transfer mechanisms. This makes the incorporation of learning curves (previously described), of less importance than for manufacturing procedures that incorporate low automation levels. Production, inspection, and part transfer times that are machine dependent need to be incorporated in cost estimates that will use these manufacturing facilities and machines. The large capital investments tied up in automated manufacturing facilities increasingly mandate the use of activity based costing. The indivi- dual work center de®nition and cost rates provided by ABC more accurately associate automation costs with end products that use or consume automated facilities and resources during the manufacturing process. Automated equipment is almost always more expen- sive than its conventional counterpart. Manufacturing automation results in recurring costs associated with more expensive maintenance, programming costs (gen- eration, debugging, and storage). When a plant invests heavily in automation, there is extreme pressure to have high equipment utilization levels. This may prompt the use of such equipment for some parts that might have otherwise been manufactured by cheaper, more conventional methods. The cost increases and savings associated with man- ufacturing automation have been described in detail by Manufacturing-Cost Recovery and Estimating Systems 861 Copyright © 2000 Marcel Dekker, Inc. Butler et al. [7, 9, 10]. These authors have developed software that enables prospective users to assess the net cost increase or savings associated with manufac- turing automation alternatives. Readers desiring more information on this subject are urged to consult these references. 2.9 SUMMARY This chapter has overviewed manufacturing cost esti- mating and recovery techniques and methods. There have been many books that have been devoted exclu- sively to the extensive coverage of these topics. Readers desiring more information on any of the subjects of this chapter are urged to consult the list of references that follows. REFERENCES 1. M Malstrom. What Every Engineer Should Know About Manufacturing Cost Estimating. New York: Marcel Dekker, 1981, pp 1±48. 2. PF Oswald. Cost Estimating for Engineering and Management. Englewood Clis, NJ: Prentice-Hall, p 1±4. 3. R Vernon, ed. Realistic Cost Estimating for Manufacturing. Dearborn, MI: Society of Manufacturing Engineers, 1968, p 1. 4. EM Malstrom, ed. Manufacturing Cost Engineering Handbook. New York: Marcel Dekker, 1984. 5. EM Malstrom. Cost estimating and control. In: R Veilleux, ed. Tool and Manufacturing Engineers Handbook, vol. 5. Dearborn, MI: Society of Manufacturing Engineers, 1988, pp 4.1±12. 6. EM Malstrom, RL Shell. A review of product improve- ment curves. Manuf Eng 82(5): 1979, pp 70±76. 7. DP Butler, EM Malstrom, K Ramachandran. A com- puterized ABC model for a job shop environment. AACE Trans, June: 1995, p 9.1±3. 8. K Ramachandran, DP Butler, EM Malstrom. A com- puter assisted model for activity based costing. Proceedings of the 22nd International Conference on Computers and Industrial Engineering. Cairo, Egypt, December 1997. 9. DP Butler, EM Malstrom, SC Parker. A tutorial model for the economic evaluation of automated manufactur- ing systems. Cost Eng 38(6): 1996, pp 25±32. 10. DP Butler, EM Malstrom, SC Parker. Assessing the true cost savings associated with the procurement of automated systems. Technical paper MM93-384. Dearborn, MI: Society of Manufacturing Engineers, 1993. 11. CM Creese, M Adithan, BS Pabla. Estimating and Costing for the Metal Manufacturing Industries. New York: Marcel Dekker, 1992, p 12. 12. EM Malstrom, W Beatty. A cost recovery methodology for automated manufacturing workcells. Cost Eng 34(5): 1992, pp 15±20. 13. PE Ostwald, ed. Manufacturing Cost Estimating. Society of Manufacturing Engineers, 1980, p 9. 14. ER Sims, Jr. Precision Manufacturing Costing. New York: Marcel Dekker, 1995. 862 Malstrom and Collins Copyright © 2000 Marcel Dekker, Inc. . cells that are dedi- cated to producing parts with geometrically similar attributes. Coding and classi®cation methods are used to de®ne families of parts that are manufactured by GT cells. Often,. commercially available estimating software packages are extremely expensive. Commercially avail- able software is also usually di cult to interface with the user organization's historical cost records. Parametric. completed, actual values of indirect labor, burden, G&A, and the number of direct labor hours worked will dier from the estimated values used in the LOH rate computa- tion. This is also