Understanding Entropy Generation during the Execution of Business Process Instantiations: An Illustration from Cost Accounting Peter De Bruyn, Philip Huysmans, Herwig Mannaert and Jan Verelst University of Antwerp Faculty of Applied Economics Department of Management Information Systems Normalized Systems Institute (NSI) 3rd Enterprise Engineering Working Conference (EEWC 2013) Luxembourg, 14 May 2013 Outline • Introduction • Theoretical framework: Entropy • Entropy generation within a business process context - the business process run-time instantiation space defining microstates and macrostates possible (cost) information aggregation dimensions understanding business process entropy generation by (cost) information aggregation - towards controlling business process entropy • Impact on (cost) accounting information systems in practice • Limitations and conclusions 1 Outline • Introduction • Theoretical framework: Entropy • Entropy generation within a business process context - the business process run-time instantiation space defining microstates and macrostates possible (cost) information aggregation dimensions understanding business process entropy generation by (cost) information aggregation - towards controlling business process entropy • Impact on (cost) accounting information systems in practice • Limitations and conclusions 2 Introduction • Appropriate business decisions require accurate information of the organization (e.g., to diagnose and trace problematic situations) - much of this data is delivered by the execution of business processes e.g., cost information as collected by cost accounting however, often gathered at different aggregation levels and therefore complex to understand or insufficient to decide • Sound theoretical basis regarding the structuring of information from executed business processes seems lacking - e.g., theoretical basis for cost accounting frameworks? • This paper: leveraging the concept of entropy to study complexity at the business process level - what does this imply for – for instance – cost accounting? requires run-time perspective 3 Outline • Introduction • Theoretical framework: Entropy • Entropy generation within a business process context - the business process run-time instantiation space defining microstates and macrostates possible (cost) information aggregation dimensions understanding business process entropy generation by (cost) information aggregation - towards controlling business process entropy • Impact on (cost) accounting information systems in practice • Limitations and conclusions 4 Theoretical framework: Entropy • Statistical thermodynamics - entropy proportional to number of microstates consistent with one macrostate (i.e., multiplicity ) • microstate = the whole of microscopic properties of the particles of a system • macrostate = the whole of externally observable and measurable properties of a system 5 Head Tail 1 2 3 4 Macrostate 8 tail, 0 head 5 6 7 8 Microstate 1 Multiplicity 1 6 Head Tail 1 2 3 4 Macrostate 7 tail, 1 head 5 6 7 8 Microstate 8 Multiplicity 8 7 Head Tail 1 2 3 4 Macrostate 4 tail, 4 head 5 6 7 8 Microstate 70 Multiplicity 70 8 Outline • Introduction • Theoretical framework: Entropy • Entropy generation within a business process context - the business process run-time instantiation space defining microstates and macrostates possible (cost) information aggregation dimensions understanding business process entropy generation by (cost) information aggregation - towards controlling business process entropy • Impact on (cost) accounting information systems in practice • Limitations and conclusions 9 Entropy generation and aggregation dimensions in a business process context: a general business process BP1 𝐵𝑃1 = 𝑡1 , 𝑡2 , 𝑡3 , 𝑡4 , 𝑡5 11 The run-time instantiation space task instantiation: 𝑡𝑘,𝑚 business process instantation: 𝐵𝑃𝑖,𝑗 𝐵𝑃1,1 = 𝑡1,1 , 𝑡2,1 , 𝑡3,1 , 𝑡5,1 𝐵𝑃2,1 = 𝑡1,2 , 𝑡2,2 , 𝑡4,1 , 𝑡5,2 𝐵𝑃3,1 = 𝑡1,3 , 𝑡2,3 , 𝑡3,2 , 𝑡5,3 12 Interpreting macrostates and microstates • Here: particles are tasks or “information units” • Hence: - microstate = the union of the values of the properties (e.g., costs) for each individual information unit (i.e., task instantiation): 𝐶 𝑡𝑘,𝑚 𝑘,𝑚 - macrostate = the aggregated information available for the observer, generally entailing unrecoverable loss of information • How easily can we solve typical management questions? - situations in which low entropy occurs, seem desirable 14 Possible (cost) information aggregation dimensions during business process instantiation execution • Approach in this paper = cost-accounting perspective • In an instantiated business process, each of the steps can be associated with some costs • Each process owner might be interested in the components which contribute to the costs of products or services - what does a product or service cost? what does a particular part of a production process cost? where does an extremely high cost originates from? (cf. BPR, TQM, etcetera) • These questions require detailed and structured information of each of the process steps - however, for plenty of reasons, several aggregations of these information units may occur in practice 16 Aggregation dimension 1 • Information is gathered at its most fine grained level: 𝑡𝑘,𝑚 • No aggregation or interaction with any other information units occurs 17 Aggregation dimension 2 • Information regarding two or more “information units” k is aggregated for each business process instance j • e.g., initial interest in only a few major phases of the process 18 Aggregation dimension 3 • Information is aggregated over all tasks k for each business process instance j • e.g., cost-based price setting 19 Aggregation dimension 4 • Information units among all instances m of a particular task k within a business process BPi are aggregated • e.g., one operator solely put in charge of checking the completeness of an insurance imbursement request 20 Aggregation dimension 5 • Information units are aggregated according to the time elapsed • e.g., a “counter” registering electricity consumption which can be inspected at every point in time t 21 Aggregation dimension 6 • Information units regarding all (task) instances of the considered business process type become aggregated • e.g., only incoming and outgoing cash flows are deemed of interest, KPI’s 22 Overview (1) Microstate Macrostate 24 Overview (2) Suppose task instance t1,1 is extremely high, how easily can we trace this within the several aggregation dimensions? 25 Towards controlling business process entropy by increasing the structure of the systems • Entropy reduction in business processes can be reduced by strict partitioning (e.g., regarding the cost structure) 1. 2. 3. 4. • introducing states: “measuring points” for intermediate registration identifying the right concerns, here: information units data instance traceability: which were the characteristics of the data the business process was operating on? task instance traceability: to which business process (instance) does a particular task instance belong? General guidelines, consistent with the entropy rationale within Normalized Systems theory - however, more business related and even domain specific guidelines for identifying the right concern may be necessary 28 Outline • Introduction • Theoretical framework: Entropy • Entropy generation within a business process context - the business process run-time instantiation space defining microstates and macrostates possible (cost) information aggregation dimensions understanding business process entropy generation by (cost) information aggregation - towards controlling business process entropy • Impact on (cost) accounting information systems in practice • Limitations and conclusions 29 Impact on (cost) accounting systems in practice (1) • Towards explaining criticisms on traditional cost accounting approaches - suppose two products A and B (where B has a more complex assembly step: task 4) - suppose aggregation dimension 4 - traditional cost-accounting: attribution of costs based on volume related measures → costs of product A will be underestimated and v.v. 30 Impact on (cost) accounting systems in practice (2) • Activity-Based costing as a solution? - finer-grained way of allocating (indirect) costs to products - however, tasks are considered as “composed of the aggregation of units of tasks”. • Current software applications? ? - case: budgeting tool at government agency 31 Outline • Introduction • Theoretical framework: Entropy • Entropy generation within a business process context - the business process run-time instantiation space defining microstates and macrostates possible (cost) information aggregation dimensions understanding business process entropy generation by (cost) information aggregation - towards controlling business process entropy • Impact on (cost) accounting information systems in practice • Limitations and conclusions 32 Limitations and conclusion • Contributions: - entropy unambiguously defined in a business process state space 6 aggregation dimensions presented some initial principles to reduce entropy some initial implications for practice • Limitations: - cost aspects are only one perspective to be considered - uniform cost assumption simplified example (e.g., all costs around 5€) only one business process type, extra aggregation dimensions possible case studies needed trade-off: benefits of low entropy vs. costs for setting up and gathering this fine-grained information. • other perspectives might suggest other concerns (e.g., financial reporting, throughput optimization, etcetera) • domain-dependent concerns? 33 Questions / remarks? e-mail peter.debruyn@ua.ac.be phone +32 3 265 40 21 office Stadscampus – S.B. 304 Prinsstraat 13 2000, Antwerpen 34