Performance indicators in logistics

Abstract

Few serious studies have attempted to analyze exactly how the logistics activity can be measured as a whole - its component parts are usually studied alone. In October 1985, a seminar was held on performance indicators in logistics by the Netherlands Association for Logistics Management. A model was presented whereby performance could be measured, and over the next few years, it was put to the test in several companies. It was found that the data necessary for performance measurement needed structuring. The results of ongoing research have been collected and published in a book called Performance Indicators in Logistics. The research is based on the control cycle, which is characterized by a number of essential stages and conditions necessary for its implementation. By combining a number of elements of the models considered, it appeared possible to construct a model for logistic applications that satisfies all of the functional specifications. The logistic input-output model is a schematic representation of reality with an operational character.

For 25 years leaders in academe and industry have enthused about logistics as the great integrating corporate authority, underpinned always by information. Yet, in this quarter-of-a-century, few serious studies have attempted to analyse exactly how the logistics activity can be measured as a whole--its component parts are typically studied alone.

In October 1985 a seminar was held on "Performance Indicators in Logistics" by The Netherlands Association for Logistics Management (NEVEM). A model was presented whereby performance could be measured, and over the next few years it was put to the test in several companies.

One finding was that, while much of the data necessary for performance measurement existed in companies, the data needed structuring. During the course of further study, other aspects emerged which were found to be important. The results of this ongoing research have now been collected and published 1!.

THE CONTROL CYCLE

Management has been defined as the planning, execution and control of goal-directed activities. Control processes are implemented in this in the form of control cycles. A schematic example of a control cycle is shown in Figure 1. (Figure 1 omitted.) On the one hand, the essential stages of the control cycles are reflected, and on the other, the conditions that are necessary for implementation of the control cycle are shown.

The information flows required for logistic control must take shape in the organization in such a manner that a consensus is reached over the information flows necessary for a well functioning control cycle. Logistic performance indicators need, if well applied within the organization, to be part of such a control cycle and need linking to the control processes in the logistic management. The principle of a control cycle for controlling a process is described with reference to a number of stages that have to be completed. These are respectively:

* Description of the process. Description and identification of the process for which the control cycle is representative. This also implies the indication of the boundaries.

* Data collection. Measurement and registration of the process data; explicit registration procedures and instruments are necessary.

* Transformation to performance indicators. In this phase, the results of the measurements are changed into usable information. In order to obtain good interpretations, this implies that one relates the results to norms.

* Evaluation of performance. Evaluation of performance indicators takes place by comparing planned or desired values with the corresponding control boundaries.

* Analysis of the problems. For a good diagnosis, experience with and insight into the process is required.

* Decision to intervene. From the diagnosis made in the previous stage, it can be decided whether or not to take corrective action. This means that the data concerning the process have to be such that expected results and costs of intervention allow for corrective action.

Actual execution of control activities starts with data collection and follows the control cycle clockwise. Before control is possible, all conditions necessary for adequate control have to be fulfilled. The actions are those necessary for implementation of the control cycle. Satisfying these conditions begins with the structuring of the process to be controlled.

The conditions to be fulfilled are:

* The logistics process has to be defined and structured in the direction of the flow of goods (movement of the goods themselves) as well as the management structure which is perpendicular to the flow of goods. The organization has to be structured accordingly (Chapter 2).

* Measurement systems and procedures must be at hand in order to collect at a regular frequency the data necessary for the determination of performance indicators (Chapter 6).

* Performance indictors indicating the deviation from norms must be defined and they must give insight the total logistic flow as well as in the different parts of the flow. These performance indicators have to reduce the complexity resulting from non-comparable data (Chapter 5).

* Quantitative norms must be set, with their corresponding control boundaries (Chapter 4).

* The decision makers must have enough experience to evaluate the different signals and make a diagnosis accordingly (Chapter 3).

* It must be clear which steering instruments are available to intervene in the logistic process. This signifies that one has to have a thorough knowledge of one's own specific logistic situation, the most important steering instruments within it, and their effects. The division of the authority in the organization must also relate to the steering instruments. If a function has been given responsibility for the realization of a defined task, this function also needs the appointment of the authority to use the available steering instruments according to its own insight in order to fulfil the task (Chapter 3).

The design of a control cycle consists of the establishment of the above mentioned conditions. The order in which this is done is as follows:

(1) Define objectives and structure.

(2) Define steering instruments.

(3) Make use of knowledge and experience for problems analysis.

(4) Fix norms and control boundaries.

(5) Define performance indicators.

(6) Determine measurement tools.

The authors then go on to describe exactly what is meant by performance indicators, using a transformation model as the basis. Quoting from the book:

PERFORMANCE INDICATORS

Looking at the general transformation model may be the simplest way to understand what is meant by performance indicators (see Figure 2). (Figure 2 omitted). The transformation process in this Figure embraces the activities concerned with transformation of input into output. This may include production processes, decision processes, development processes, control processes, etc. Data related to the condition of the transformation process are called conditional variables.

Measurement in the general model can take place in three categories of process data: inputs, conditional variables and outputs. In order to give an insight into the different types of data that are ranked as one category, an example of quantitites from the logistic sector is given (see Table I). (Table I omitted.) Notice that the inputs, conditional variables and outputs can almost all be expressed in both financial and physical units. However, reporting the process data that are relevant for the company is not sufficient to calculate performance indicators. In order to gain insight into the past performance of the process it is necessary to relate the collected data to norms or other data. Only after the establishment of these relations performance indicators do emerge.

The conditional variables themselves originate by making a relation between different data and are therefore by definition performance indicators. The most common performance indicators also used in this book are:

* Efficiency = norm input/real input.

* Effectiveness = real output/norm output.

* Process parameter = real conditional variable.

* Productivity = real output/real input.

* Utilization level= real input/norm input.

These these quantities are represented schematically in Figure 3. (Figure 3 omitted.)

Efficiency concerns the efforts, costs and reception of the inputs when employed in the process in relation to the preliminary stated norms (reciproca1 occupation rate). Effectiveness concerns the amount to which the process realizes the previously stated norms (compared with the actual output).

Process parameters indicate values which the quantities have that are considered to be relevant (such as inventory level and throughput time).

Productivity reflects the relation between the achieved result (output) and the means used to obtain this (input).

Two examples of performance indicators are illustrated below:

(1) Delivery reliability to customers is defined as follows:

* Number of customer orders correctly delivered in a certain period (number and time) = real output.

* Number of customer orders planned to be delivered (number and time) in that period = norm output.

* Delivery reliability = real output/norm output (effectiveness).

The performance indicator is expressed as a percentage (e.g. 95 per cent). A norm with control boundaries could then be 97 per cent +/-2 per cent. Graphically the performance indicator can be presented as in Figure 4. (Figure 4 omitted.)

(2) The inventory level in the component warehouse is defined as follows:

* Inventory (in financial terms) at the beginning of the month divided by the supplies in that month. The performance indicator is now expressed in months (e.g. two months). The norm with control boundaries could then be two months +/-half-a-month. Graphically this performance indicator can be presented as in Figure 5. (Figure 5 omitted.)

From the enumeration of quantities as in Table I it can be derived that a large number of performance indicators can be defined. In this book we want to point out how this can be achieved structurally. As every situation is specific we shall present a method that may be used in selecting performance indicators. For clarification we conclude this book with a number of practical examples.

The ensuing chapters look at the structure of logistics, steering instruments, setting norms, determining performance indicators, measurement tools, working with performance indicators, production organization and performance indicators, and the logistic input-output model.

We are reminded in chapter six, on measurement tools, of the purpose behind all this information-gathering activity --points which are often lost on the boffins who get carried away with the flashing bright lights!:

REQUIREMENTS FOR DATA COLLECTION AND PROCESSING

The following requirements can be made with respect to data collection and processing:

(1) Validity. Measurements must reflect actual performance. This requires a good definition of the performance indicators (Chapter 5), the measurement points and the measurements from which they originate. Also the time span over which the measurement data are valid has to be known.

(2) Covering potential. Measurements must completely cover the definition of the performance indicators. The more completely the measurements of the process quantities are represented, the better these quantities can be evaluated.

(3) Comparability. Measurements are to be comparable in dimension throughout the organization and over subsequent points in time (premise 4, Chapter 5).

(4) Accuracy. Sufficient precision and reliability is required of the measurements. For example, the taking of samples for inventory reliability must be representative of the real situation besides being an accurate measurement.

(5) Utility. A decision maker must benefit from the measurements done (which is the object of this book).

(6) Comparability. If a measurement system links up well with the existing data organization, it will be easier to implement. This existing data organization will also influence the choice between manual or automated measurement (Section 6.5).

(7) Profitability. The benefits of the measurements must exceed the costs.

NB: Beware of striving for too much precision (too detailed and expensive!).

How the above-mentioned requirements affect the performance indicators can be illustrated with an example about the inventory level:

(1) Validity. The measured value must represent the real inventory level. For example a measurement of the total inventory level will be invalid if the warehouse inventory and the pipeline inventory are not measured simultaneously.

(2) Covering potential. When a performance indicator aims to describe the total inventory, one may not omit to take, e.g. the pipeline inventory or subassemblies into consideration.

(3) Comparability. The units for indicating the inventory level are not comparable if on one occasion the inventory level is expressed in days, and on another, in financial terms or number of products.

(4) Accuracy. The value determined for an inventory level must be so accurate and reliable that any exceeding of the control boundaries is always detectable. Precautions have to be taken especially if an accumulation of errors is possible. With respect to inventories, reliability of the recorded data is particularly important. Not only do unreliable data lead to wrong decisions, but they also lead to misinterpretation of the performance indicators.

(5) Utility. If a certain (part of the) inventory level is not adjustable, or very difficult to adjust (e.g. it is outside the responsibility area), then it makes no sense to take measurements for that responsibility area.

(6) Profitability. At least the costs of the installation and maintenance of an inventory level measurement system will have to be regained (in the total chain).

The final chapter concentrates on the logistic input-output model, which formed the nucleus of the working group's 1984 seminar and its subsequent research. It arose from the realization that: "all logistics activities in a company need to contribute to the realization of the company's objectives". This realization, of course, is not new but how difficult it is to define, measure and provide that "contribution" in the face of often conflicting departmental objectives and agendae. Hence, the model, described thus:

INTRODUCTION TO THE MODEL

The logistic input-output model is a schematic representation of reality, with an operational character, which is expressed in:

* The possibility to analyse the logistic activities with the help of the model and their influence on the corporate objectives.

* The possibility to increase insight into the coherence of the logistic activities by means of the model .

Concretely, this means that the model has the following functional specification:

* The different logistic activities and their relations are included.

* The different levels in the organization are considered separately.

* Both physical and financial elements are included.

* The model can be utilized as both a calculation and a simulation model.

None of the models, so far available, satisfied all of these requirements, although certainly a number of elements recur. By combining a number of elements of the models considered, it appeared possible to construct a model for logistic applications that satisfies all these functional specifications. Two examples used in this way are:

* The Du Pont model (Appendix 4): this shows the coherence of a number of financial variables together forming the return on investment. (Appendix 4 omitted.)

* The logistic process chain: in this model the activities are presented in relation to the physical flow of goods.

DESCRIPTION

GENERAL

In Figure 6 (Figure 32 in the original) the logistic-input output is presented schematically. (Figure 6 omitted.) On the horizontal axis the most important elements of the Du Pont model are presented, which are:

* Capital invested (goods, capital assets).

* Operational costs (depreciation, labour, buildings, personnel, other).

* Cash flow.

For an adequate analysis of the entire process, the physical flow is added on the horizontal axis, divided into purchasing and delivery activities. On the vertical axis the horizontal structure of the process under consideration is presented. For example (Figure 6, Figure 32 in the original): supplier, purchasing, warehouse, production, customer. The process considered and thus the schematic presentation is situation dependent, as described in Chapter 2.

ELABORATION

In order to elaborate the logistic input-output model, the different blocks have to be filled in with both physical and financial terms. Each block consists of a combination of horizontal and vertical factors (e.g. operational costs warehouse). With respect to the physical and financial units the combination could be, e.g.:

* 100 hours of labour=$2,500.

* 10 tons of goods = $10,000.

The logistic input-output model is likewise filled in for a certain period (day, week, month, year). By defining two or more periods consecutively, analyses can be made with respect to time delays, trends and cash flows. As such, the decrease of lead time at constant output will imply a decrease of the input, and also the future cash outflow.

Minimally, the following relations have to be included in the logistic input-output model:

* The output of a block in the physical delivery flow has to be the input of a subsequent block in the physical purchase flow. Modification of the output of a block influences the model.

* The physical flow, expressed in physical and financial data, and the operational costs and capital invested columns need to be mutually consistent according to the following calculation:

Delivery financial = Purchase financial + Operational benefits - Mutation capital invested in goods (ending - beginning).

The cash flow column can, if desired, be itemized to receipts and expenses. But considering that these coincide with the columns for the physical flow, this is not really necessary.

FURTHER POSSIBILITIES

If so desired, the logistic input-output model can be made more complex by addition of more relations. This also implies that the model is a better reflection of reality.

Another method to obtain better facilities for analysis with the model is to enlarge the number of columns by division of the various items. For example, an item "investment mutations" could be included and the operational costs column could be split into separate columns for fixed and variable costs.

By horizontal intersection of a process level, relations can be established between the physical flow and the matching operational costs and capital invested. Especially between physical units and operational costs, relations can be established in the following two ways:

* Calculation of the necessary means based on the nature and size of the expected physical flow and the performance norms proposed (e.g. hours/ piece).

* Calculation of the performance realized based on the real physical flow and the real means utilized.

This implies that the logistic input-output model can also be applied for normative calculation based on performance norms.

INDICATIONS FOR FILLING IN THE LOGISTIC INPUT-OUTPUT MODEL

To conclude this section, we shall give some indications for the completion of the logistic input-output model. In order to fill in the operational costs and invested capital columns, the existing data from the budgets and balance sheets can be taken over. This will ensure that the data in the model coincide with the method applied in the company. The budgets for the departments are to be split into the following items:

* Fixed costs of the department. These are to be formulated and included in the logistic input-output model as one block.

* Variable costs of the department. These include those costs that are directly related to the activity level. They are calculated normatively on basis of performance norms, or included as real costs, after which performance can be calculated.

* Allocated overheads.

For a complete representation of the company, it is necessary to include the data for those departments not considered in the process as one block under "others company". In this block also the items "allocated" as identified in the budgets are to be included. With this division, simulation becomes feasible, because all variable costs vary with quantities, and the fixed costs remain unchanged or drop out entirely (if the activity is eliminated). In order to fill in the columns, belonging to the physical flows, the following can be stated:

* The process must be indicated with all branches (e.g. work put out to contract, service deliveries).

* The flows have to be indicated in quantitites and price equivalents.

* Operations have to be indicated in quantities, capacities and price equivalents. For example, 1,000 hours with a value of $25,000, 10 tons of goods with a value of $10,000.

The remainder of the chapter provides worked examples. Useful appendices provide further insight into indicators.

In summary, this is a gem of a book--just 100 pages but packed with wisdom and practical application. Reading it and using the structures presented will do much to benefit the practice of logistics as a whole.

REFERENCE

1. NEVEM Workgroup, Performance Indicators in Logistics, IFS Ltd and Springer-Verlag. Copies of the book may be ordered from IFS Ltd, Wolseley Business Park, Kempston, Bedford MK42 7PW.