VDA 9000-2010 Measuring aftermarket forecast accuracy《测量售后市场预测精度》.pdf

《VDA 9000-2010 Measuring aftermarket forecast accuracy《测量售后市场预测精度》.pdf》由会员分享，可在线阅读，更多相关《VDA 9000-2010 Measuring aftermarket forecast accuracy《测量售后市场预测精度》.pdf（23页珍藏版）》请在麦多课文档分享上搜索。

1、VDA Measuring aftermarket forecast accuracy 9000 Version 1.0, May 2010 Working Group SCM Aftermarket Publisher: Verband der Automobilindustrie Copyright Behrenstrae 35 Reprinting and any other form 10117 Berlin of duplication is permitted only Phone +49 (0)30 897842-0 with citing of source. Web: www

2、.vda.de VDA Recommendation 9000 May 2010 Page 2 of 23 Copyright VDA Disclaimer VDA recommendations are freely available for general use. The user is responsible for ensuring correct application for the specific case. They represent the latest technology available at the time of issue. Application of

3、 VDA recommendations does not relieve the user from responsibility for his own actions. In this regard, all users act at their own risk. VDA and those involved with VDA recommendations do not accept any liability. Anyone applying VDA recommendations who identifies inaccuracies or possible incorrect

4、interpretations is invited to inform VDA immediately and any errors can thus be rectified. VDA Recommendation 9000 May 2010 Page 3 of 23 Copyright VDA Table of Contents 1 General information.4 1.1 Foreword.4 1.2 Goal of the recommendation4 1.3 Structure of the recommendation.5 2 Requirements and goa

5、ls of aftermarket forecast accuracy measurement5 2.1 Subject.5 2.2 Requirements6 2.3 Goals of forecast accuracy measurement6 2.4 Intended purpose of the measurement method6 3 Performance indicator for measuring aftermarket forecast accuracy7 3.1 Forecast Accuracy Index (FAI).7 3.2 Weighted Tracking

6、Signal (WTS).8 3.3 Example FAI and WTS calculations.9 3.4 Generalization with regard to time frame and horizon10 3.5 Generalization of the reference value.11 4 Delimitation of the performance indicator.11 5 Recommendations for application12 5.1 Selecting the time frame12 5.2 Reference value and cust

7、omers desired delivery date12 5.3 Selecting the lag forecast date.12 5.4 Selecting the measurement horizon.12 5.5 Selecting the weighting factors12 5.6 Interpretation of FAI and WTS performance indicators.14 5.7 Product groups and clustering.15 5.8 Aggregation.15 5.9 Correlation to the service level

8、.16 5.10 Priorities for performance indicator analyses17 5.11 Consistency checklist.17 6 Factors influencing measurement17 6.1 Product phase-in and phase-out17 6.2 Taking quantity scales into account.18 6.3 VMI/CMI.18 6.4 Bilateral arrangements.18 7 Derivable measures.19 7.1 Measures for improving f

9、orecast accuracy.19 7.2 Measures for improving supply chain performance despite poor forecast accuracy.19 8 Optimization flow chart20 9 Summary.20 10 Appendix.21 10.1 Abbreviations, terms and definitions21 10.2 References21 10.3 Examples of performance indicator aggregation22 VDA Recommendation 9000

10、 May 2010 Page 4 of 23 Copyright VDA 1 General information 1.1 Foreword This guideline was designed by the Working Group for Aftermarket Forecast Accuracy Measurement as part of the VDA Task Force for Supply Chain Management. The study was conducted in response to the highly dynamic development of t

11、he market for spare parts as well as increasingly volatile demand in the automotive industry. This represents additional challenges for supply chain management in terms of efficiently controlling service level and inventory, both at the OEM and the supplier. A standardized method for measuring forec

12、ast accuracy is useful for analyzing forecast situations and deriving measures for increasing supply chain performance. In developing this recommendation, the working group utilized the experience of the participating OEM and supplier companies to define a best practice performance indicator with th

13、e help of extensive statistical analyses. This new performance indicator is based on the one developed a year earlier for VDA Recommendation 5009 with regard to OEM equipment, aftermarket-specific goals for adapting to partially very long lead times for aftermarket spare parts, and the correlation b

14、etween forecast accuracy and service level. The developed performance indicator serves as a standard tool for measuring the quality of demand forecasts, normally generated by OEMs, with respect to changes over time and a realized reference value typically the ultimately ordered quantity of a spare p

15、art. The measurement time period is typically one month, but the measurement logic can also be extended to other time frames. Application of the method at other forecasting interfaces in the supply chain is also feasible, for example within an OEM organization or between tier 1 and tier 2 suppliers.

16、 Positive performance indicator characteristics are values ranging from 0% to 100% as well as positive correlation to the service level. 1.2 Goal of the recommendation The global supply chain for the automotive aftermarket is often subject to strong fluctuations in demand. Optimizing OEM and supplie

17、r inventory levels is particularly challenging whenever demand is highly volatile. If all safety stocks in the supply chain are used up, supply shortages and long wait times for material replenishment by the supplier occur. On the other hand, high inventory levels are needed to minimize this risk. B

18、ullwhip effects can therefore occur in the supply chain and cause correspondingly high escalation management expenses not only higher costs for short-term procurement, but also higher management and escalation expenses that are often not directly reported as costs. To support management efforts in t

19、hese situations and achieve the desired or maximum possible service level at acceptable inventory costs, a standardized method for measuring forecast accuracy is recommended. This method can be used to develop measures for stabilizing forecasts or directly handling supply and inventory situations. E

20、xpenses for associated exception management are thus reduced. VDA Recommendation 9000 May 2010 Page 5 of 23 Copyright VDA 1.3 Structure of the recommendation Section 1 contains general information regarding the recommendation. Section 2 describes the subject of the recommendation and defines the req

21、uirements and goals of aftermarket forecast accuracy measurement. Section 3 defines the recommended performance indicator for measuring forecast accuracy in the aftermarket: the Forecast Accuracy Index (FAI). A secondary performance indicator for measuring forecast accuracy is also defined: the Weig

22、hted Tracking Signal (WTS), which indicates whether preliminary forecast trends are too high or too low. This is followed by general information regarding the performance indicators and the reference value. Section 4 explains the advantages of the FAI and WTS performance indicators with regard to af

23、termarket requirements and compares the measurement logic to an existing performance indicator for original equipment (VDA 5009). Section 5 describes how the performance indicators are used in the aftermarket. Section 6 describes specific process-related situations affecting the measurement of forec

24、ast accuracy. Methods for deriving process improvement measures are presented in Section 7. Section 8 provides a flow chart depicting the performance indicator application process. Finally, Section 9 provides a short summary of the recommendation. 2 Requirements and goals of aftermarket forecast acc

25、uracy measurement 2.1 Subject This recommendation deals with methods for measuring the accuracy of forecasts for spare part demand at the material number level. Normally, demand data for developing forecast and reference values is best collected on a monthly basis, as this shows the relationship bet

26、ween forecast accuracy and delivery performance. Measures for improving forecast accuracy and thus potentially improving delivery performance can then be derived. VDA Recommendation 9000 May 2010 Page 6 of 23 Copyright VDA 2.2 Requirements The following requirements for an aftermarket forecast accur

27、acy performance indicator were considered: Target values ranging from 0% to 100% As with service level measurement, a higher value should indicate good forecast accuracy while a lower value should indicate poor accuracy. This should also reflect the expected positive correlation between forecast acc

28、uracy and service level. Adaptability to lead times in the supply chain (with a focus on ability to respond to changes) through weighted measurement of forecast figures against a reference value. Aggregation can be performed by product group, sales share, etc. 2.3 Goals of forecast accuracy measurem

29、ent The goals of measuring aftermarket forecast accuracy are: Standardizing forecast accuracy measurement for the purpose of increasing transparency at the interface between OEMs and suppliers in the supply chain Recognizing weaknesses in the forecasting process Improving supply chain performance th

30、rough derived actions: Improving preliminary forecasts Adjusting inventory to achieve target service levels at the OEM and the supplier Reducing expenses by avoiding escalation management in unstable supply chains Differentiating service level goals for specified forecast accuracy performance indica

31、tor levels 2.4 Intended purpose of the measurement method The performance indicator for measuring forecast accuracy as described in this recommendation is intended to be used exclusively for the purpose of increasing joint supply chain performance between OEMs and suppliers. It may be used to evalua

32、te service level goals as well as inventory optimization goals. VDA Recommendation 9000 May 2010 Page 7 of 23 Copyright VDA 3 Performance indicator for measuring aftermarket forecast accuracy The following parameters are needed to define the forecast accuracy performance indicator: Horizon n = 4 (mo

33、nthly time frame) d0: Realized value (reference value) l1: Lag 1 forecast: forecast value 1 month prior to realization l2: Lag 2 forecast: forecast value 2 months prior to realization l3: Lag 3 forecast: forecast value 3 months prior to realization l4: Lag 4 forecast: forecast value 4 months prior t

34、o realization Deviations (signed values): 044 033022 011: dl dl dl dl= = = = Notes: Positive value: lag forecast too high Negative value: lag forecast too low Weights and weighting factors: 4321 , ; 0i where 141 =i i 3.1 Forecast Accuracy Index (FAI) Definition of Forecast Accuracy Index (FAI) (for

35、horizon n = 4) VDA Recommendation 9000 May 2010 Page 8 of 23 Copyright VDA If: 00d + + + =044033022011 |1;0max|1;0max |1;0max|1;0max: dd ddFAI If: 00=d 0: 4,.,1;0: = = FAIIfor iiIwhereFAI iIi i Notes: 1) 0% FAI 100% 2) 100% represents the best value 3) If all weighting factors have the same value (i

36、.e. 4,3,2,1;41=ii ), then all lag forecasts l1, , l4 have the same impact on forecast accuracy. 4) To determine whether (lag) forecast values are too high or too low compared to the reference value, a tracking signal (TS) is required. A suitable weighted tracking signal is defined below. 3.2 Weighte

37、d Tracking Signal (WTS) Definition of Weighted Tracking Signal (WTS) (horizon n=4) If: = 41 0|i ii |:44332211 44332211 + += WTS If: = =41 0|i ii 0:=WTS Notes: 1) -1 WTS 1 2) WTS = 1: all deviations are positive 3) WTS = -1: all deviations are negative VDA Recommendation 9000 May 2010 Page 9 of 23 Co

38、pyright VDA 3.3 Example FAI and WTS calculations The following demand forecast values were transmitted by the OEM to the supplier for a particular material number for the reference month January 2009. Time of transmittal Transmitted demand forecast value for Jan. 2009 September 2008 October 2008 Nov

39、ember 2008 December 2008 7000 5100 2780 3400 Demand as indicated for the (OEM) customers desired delivery date of January 2009 was met by a quantity of 3400. Using the following weighting factors: i i 1 2 3 4 0.1 0.3 0.3 0.3 the FAI and WTS can be calculated as follows. Reference value d0 = 3400 (Ja

40、nuary 2009) Lag i Point in time Lag forecast Deviation from reference value d0 (+/-) i 1 December 2008 3400 0 2 November 2008 2780 -620 3 October 2008 5100 1700 4 September 2008 7000 3600 Relative absolute deviation from d0 0di 01di 01;0maxdi i 0 1 1 0.1 0.182 0.818 0.818 0.3 0.5 0.5 0.5 0.3 1.059 -

41、0.059 0 0.3 VDA Recommendation 9000 May 2010 Page 10 of 23 Copyright VDA 791.036003.017003.06203.001.0 36003.017003.0)620(3.001.0 %50495.003.05.03.0818.03.011.0 =+ += =+=WTSFAI Graphic representation of sample calculation: 3.4 Generalization with regard to time frame and horizon These formulas may b

42、e used for various time periods and horizons. For example, a weekly time period may also be selected. Generalized formulas for use with general time periods are given below. 3.4.1 Forecast Accuracy Index (FAI) For any horizon where I n the following is true: If: 00d = 01 |1;0max: dFAI ini in where 0

43、,.,21 n and 11 =ni i Sept. 08 Oct. 08 Nov. 08 Dec. 08 Jan. 09 d0=3400 l1=3400 l2=2780 l3=5100 l4=7000 Lag 1 %101=36004= 17003=6202= 01=Lag 4 Lag 3 Lag 2 %304= %303= %302=1000 3000 5000 7000 VDA Recommendation 9000 May 2010 Page 11 of 23 Copyright VDA If: 00=d =ni inFAI1: where niiI in ,.,1;0| = and

44、= IwhereFAIn 0: 3.4.2 Weighted Tracking Signal (WTS) For any horizon where I n the following is true: = =ni iini iiWTS11 |: where 0:=WTS if 0|1 =ni ii 3.5 Generalization of the reference value The reference value d0 normally represents the realized sales value and lies in the (recent) past. The refe

45、rence value may also be generalized as a future realized value. In this case, the lag forecasts represent past forecast values for the reference value. The defined formulas are then the same: 4 Delimitation of the performance indicator The FAI and WTS performance indicators were developed specifical

46、ly for aftermarket requirements and goals. The FAI performance indicator differs from the performance indicator of VDA Recommendation 5009, which is primarily intended for original equipment, in the following respects: If the reference value is 0, FAI is limited and thus allows unlimited aggregation

47、 (particularly with FAI 0 WTS 0 scenarios in the analysis, including analyses within a cluster. 5.8 Aggregation Aggregating performance indicator data to form a total forecast accuracy value as a statistical parameter is helpful for managing customer/supplier relationships. Aggregated performance in

48、dicators for management-relevant product areas can also be calculated. However, when using an aggregated FAI performance indicator, the corresponding aggregated WTS performance indicator must always be calculated and indicated. Several types of performance indicator aggregation are described below. They are intended to be illustrative and should not be regarded as complete. Other types of performance indicator aggregation are also possible. For examples of performance indicator aggregation, see the appendix. VDA Recommendation 9000 May 2010 Page 16 of 23 Copyrig