|
|
|
|
Review of Benchmarking Methods: In this section we present a review of different benchmarking methods, including traditional metrics and ratios, traditional cost-based metrics, as well as the data envelopment analysis (DEA) which is popular in warehouse benchmarking. To determine a suitable technique would depend on the nature of the warehouse, which varies for different scenarios. Some vital points to be considered are:
Single-factor metrics and ratios based on two non-monetary factors Traditional evaluation of warehouse operations can be based on a long list of metrics and ratios. The most popular of these include: total units shipped per period line/labour hour order cycle time or response time fill rate fill accuracy inventory accuracy space utilization (number of spaces occupied by any merchandise) A more complete list could be download from here[pdf]. Such metrics can be valuable in assessing warehouse performance, but caution is in order: facilities should be comparable on "situation factors". These factors include the following: line per order process delay in-house cycle (IT) inventory turnover number of products stored number of products moved per year (%) % of new products per year % of obsolete products per year labour turnover, % per year We have seen too many benchmarking studies compare facilities with completely different missions, or with completely different constraints imposed by corporate headquarters. For example, a media distributor that concentrates on high-volume sellers will naturally out-perform one that sells the entire range of products. Similarly, a "push" distribution center that delivers to captive retail outlets will out-perform one that serves independent retailers using a "pull" concept If comparable facilities can be identified, then traditional benchmarking can be very valuable. It can often lead to better diagnostics than other types of evaluation. The selection of the comparison facilities with respect to the situation factors usually leads to a thought process of how to change the factors that are leading to low productivity, An example of this might be a low inventory turnover; a change in purchasing policy at corporate headquarters might be needed to improve the situation for the warehouse manager. A variation on the above approach is segmented comparison method. This is similar to the use of traditional metrics, but the comparison facilities that are "comparable" might be in different industry sectors, such as medical services, spare parts service for automotive, or electronics equipment, etc. Examples of some metrics are: response time, order cycle time within facility fill rate, % of lines requested actually delivered on time Fill or shipping accuracy These facilities might differ from the benchmarked facilities in some characteristics relating to activity, size, number of products, etc. Other performance characteristics will be compared to facilities that are deemed "comparable" with respect to lines per direct labour hour. In this manner, subsets of relevant performance characteristics of the benchmarked facilities will be compared to appropriate "comparison" facilities. One could argue that such an approach would put the benchmarked facilities at a disadvantage, since they would have to compete against better performance aspects of a different comparison facilities. This is an erroneous perspective, and it should be kept in mind that the ultimate goal of benchmarking is to identify areas for improvement. Any numerical scores resulting from a segmented comparison method must be view in this light, and not absolute measures. A general issue in benchmarking is data validity. We have seen too many instances of people misinterpreting terminology and/or entering incorrect data to reach the following recommendation: One should not too rely on data unless a knowledgeable and trusted person has visited the facility and talked with the operating personnel. Survey responses, whether paper- or web-based, are especially susceptible to errors. Cost-based factors are also popular, especially at the strategic level where decisions are made on locations of depots, third-party ware-housing, adding or dropping products, etc. The most common factors in this group include: cost of warehouse operations as % of total costs of operations cost/line shipped cost/order shipped The difficulty in interpreting cost-based factors is that two of the three major unit input prices for a warehouse operation, labour rate and unit floor space cost, are often beyond the control of the warehouse manager. One of the reasons the giant retailer Wal-Mart has controlled its costs is by locating its distribution centers in less developed areas that still have an adequate labour supply. These decisions were made at corporate headquarters, and not by any of the 30+ managers of DCs. An important use of cost-based factors is their application to portions of the warehousing operations. For example, time, cost, and productivity can be calculated for each of the operations of 1) receiving, 2) stowing or put-away, 3) replenishment of forward pick area, 4) picking, 5) order consolidation, inspection and packing, and 6) loading or shipping. Data Envelopment Analysis (DEA) To overcome the difficulties in comparing facilities with different unit cost parameters and other situations, researchers developed a multi-input, multi-output method that focused on the operational efficiency of decision making units (DMUs). This method is called data envelopment analysis (DEA). A DMU could be a manufacturing firm, a service firm, a government organization, or a sub-unit such as division of one of the above. A typical statistical technique, such as linear regression, is characterized as a central tendency approach; it evaluates producers relative to an average producer. In contrast, DEA is an extreme point method; it compares each producer with only the "best" producers. A major advantage of an extreme point method is that the results are invariant relative to the number of inefficient producers. A corresponding drawback is that results can be very sensitive to statistical outliers and noise. A fundamental assumption behind DEA is that if a given producer, A is capable of producing some outputs with a given number of units of input, then other producers should be able to if they were to operate efficiently. Similarly, if producer C is capable of producing some outputs with the same given units of input, then other producers should also be capable of the same outputs. Producers A, C, and others can then be combined to form a composite producer with composite inputs and composite outputs. Since this composite producer usually does not exist, it is called a virtual producer. This construction can be thought of a form of interpolation as shown in the graphical example in Figure 1. Producer A is able to perform 40 item-pick orders with a given set of inputs. Producer C is able to perform 10 item-pick orders and 20 case-pick orders with the same set of inputs. The evaluation of producer B, who performs 20 item-pick order and 5 case-pick orders with the same set of inputs, requires the construction of a virtual producer V. This virtual producer is assumed to be capable of performing a linear combination of outputs of A and C, using the same set of inputs. This virtual producer is useful as a target for producer C, and the distance from the origin to the performance point of B, compared to the distance from the origin to the performance point of V, is the efficiency measure of producer B.
The heart of DEA lies in constructing the "best" virtual producer to compare against each real producer. If the virtual producer is better than real producer by either making more output with the same input or making the same output with less input, then the real producer is inefficient. Most of the subtle variation of DEA are introduced in the various ways that virtual producers are constructed. The difficulty in applying DEA lies in the number of data points needed for comparison. Typical inputs include the following: labour hours, square area of facility, and investment or replacement cost of equipment. Typical outputs are item-pick lines, carton-pick lines, pallet-pick lines, etc. Even if only 3 or 4 of the outputs are selected, this still results in a total of 3 + 3 = 6 or 3 + 4 = 7 factors. This implies that a minimum of 3 x 6 = 18 or 3 x 7 = 21 comparison facilities be evaluated. If more of the output factors are to be considered, such as response time and the number of products stored, then the required number of comparison facilities would increase. There is another remaining challenge: identifying comparison facilities that have similar characteristic with the benchmarked warehouses. One of the deficiencies of DEA is that if a candidate facility excels on one input measure (consumes less than other facilities), or excels on one output measure (produces more than other facilities), then it will be evaluated at 100% efficiency. This can happen even if it is clear to a trained observer that the facility could be improved. |
[ Home ] [ Projects ] [ Events ] [ Research Papers ] [ About Us ]Send mail to
tliwwc@nus.edu.sg with
questions or comments about this web site.
|
