FORECASTING AND THE ENTERPRISE

EXECUTIVE SUMMARY Executives must include some form of forecasting in nearly all decisions they make as most operating decisions rely on “the future” as a significant input. As a result, good forecasting is a necessity. The more management understands forecasting techniques and processes and how they should manage and organize a successful forecasting function, the more successful the firm will be. This document addresses the key organizational, management, process, and operational aspects of forecasting that allow an enterprise to use it to drive corporate decisions. The main points or the areas on which management should most heavily focus their efforts include:

FORECASTING AND THE ENTERPRISE Best practices for operating an effective forecasting function Copyright 2002 - Inforte Corporation EXECUTIVE SUMMARY Executives must include some form of forecasting in nearly all decisions they make as most operating decisions rely on “the future” as a significant input As a result, good forecasting is a necessity The more management understands forecasting techniques and processes and how they should manage and organize a successful forecasting function, the more successful the firm will be This document addresses the key organizational, management, process, and operational aspects of forecasting that allow an enterprise to use it to drive corporate decisions The main points or the areas on which management should most heavily focus their efforts include: Organizational • Full senior management commitment to forecasting as an enterprise-wide initiative is critical Management must be prepared to lead by example The behaviors and attitude that management portrays directly influences the way the rest of the organization responds to the forecasting function By eliminating forecasting politics from its behavior, management can significantly reduce the amount of politicking that takes place throughout the rest of the organization • Forecasts should drive decisions in all functions Clear guidance should be given across all parts of the organization as to the importance of forecast results and the need to respond rapidly and appropriately All functions within the organization should become more demand-driven If senior management is committed to the process, reviewing results and basing decisions on forecasts, business area leadership will follow • It is crucial that the forecasting function be centralized and objectified Although forecasting should be collaborative, across both internal departments and external partners, reporting relationships should remain independent from P&L areas to ensure objectivity Management & Process Implementation • One of the most important steps in implementing a successful management process for the forecasting function is performing a gap analysis between current capabilities and where the firm should be This should, in fact, be an ongoing process that allows management to track the effectiveness of the process on an ongoing basis • Creation of a formal plan for the forecasting function is crucial to its success within the organization During the drafting of the plan, management should help to scope and define expectations and goals for the forecasting function It is important that management fosters an environment in which bias is minimized Formalizing the process and decision rules within the forecasting function and rewarding forecasting accuracy can minimize bias and keep both the preparation and the analysis of the forecasts as objective as possible • One of the most overlooked, but also one of the most important aspects, of the forecasting process is the enterprise response process A formal enterprise response plan defines how managers should review forecasts and determine subsequent actions The creation of a formal response plan is critical for a firm to be able to respond quickly and in unison to demand conditions Best Practices for Forecasting Copyright 2002 - Inforte Corporation Operational • Forecast preparation requires that forecasters work with the users of the forecast to define and agree on the dynamics of the system being forecast It is important that users and forecasters collaboratively agree on the interrelations between variables, the constraints and risks of the forecast, the appropriate timeframe and the appropriate level of detail By working collaboratively, forecasters and users are able to establish clear lines of communication, alleviating one of the most common problems in the forecasting process – the lack of trust and understanding between preparers and users • During technique selection it is important to consider a number of factors including the characteristics of the situation being forecast, quality of available input and the type of output required It is also important to assess the known strengths and weaknesses of each technique Although they should be used selectively, judgmental techniques are occasionally the most appropriate Documented guidelines should be established and used to determine when to correctly apply judgments during the forecasting process • It is vital that forecast accuracy is carefully defined and tracked It not only forms the basis for many statistical projection models, it is also used at the corporate level to determine the level of slack to be kept in assets, capital and resources Forecast accuracy is used to determine the desired level of enterprise responsiveness – the enterprise should be able to respond fast enough to make up for the average error in the forecast Best Practices for Forecasting Copyright 2002 - Inforte Corporation TABLE OF CONTENTS 1.0 INTRODUCTION & OBJECTIVES 2.0 FORECASTING OVERVIEW .12 2.10 2.20 2.30 3.0 WHY IS FORECASTING IMPORTANT 12 FORECASTING HORIZONS 12 FORECASTING MODELS & TECHNIQUES 13 ORGANIZATION & CULTURE CONSIDERATIONS 14 3.10 LEVEL – REPEATABLE 14 3.101 Remove politics from the forecasting process 14 3.102 Limit influence of opinion on quantitative results 15 3.103 Formalize a structure for the forecasting function 15 3.104 Implement a career path for forecasters 15 3.105 Ensure the forecasting team has a comprehensive skill mix 15 3.106 Define the responsibilities of the forecasting team 16 3.20 LEVEL – DEFINED 16 3.201 Ensure full senior management commitment 16 3.202 Ensure strong leadership within the forecasting function 16 3.203 Implement a collaborative forecasting approach 17 3.204 Centralize and objectify the forecasting function 17 3.205 Ensure reporting relationships are independent 17 3.206 Rethink the training approach 17 3.207 Conduct training for management in forecasting 18 3.30 LEVEL – MANAGED 18 3.301 Measure and monitor forecasting performance 18 3.302 Implement demand-driven planning 18 3.303 Define responsiveness of the enterprise 19 3.304 Implement an executive steering committee 19 3.305 Align compensation to the firm’s demand-driven goals 20 3.306 Implement collaborative inter-firm forecasting 20 3.40 LEVEL – OPTIMIZING 20 3.401 Ensure forecasts drive decisions in all functions 20 4.0 IMPLEMENTATION CONSIDERATIONS 21 4.10 LEVEL – REPEATABLE 21 4.101 Conduct a diagnostic of current capabilities 21 4.102 Produce a gap analysis on current capabilities 21 4.103 Define the problem and needs for each forecast 22 4.104 Manage against bias 22 4.105 Produce a formal plan for the forecasting function 24 4.106 Define rules for management input 24 4.107 Plan adequate time, resources and access for data gathering and preparation 24 4.108 Objectify the data gathering process 25 4.109 Choose a standard forecasting model and supporting tool 26 4.20 LEVEL – DEFINED 26 4.201 Define and communicate expectations 26 4.202 Define, publish and communicate a methodology for the forecasting process 27 4.203 Identify critical communications points 27 4.204 Define formal rules for the interpretation of forecast results 27 4.205 Define a formal enterprise response process 27 4.30 LEVEL – MANAGED 28 4.301 Implement cross-functional performance measures 28 4.302 Implement a cross-functional performance measurement process 28 4.40 LEVEL – OPTIMIZING 29 4.401 Implement continuous forecasting 29 4.402 Identify a demand signal action map for all areas of the business 29 Best Practices for Forecasting Copyright 2002 - Inforte Corporation 4.403 5.0 Monitor and continuously improve enterprise responsiveness 29 OPERATIONAL CONSIDERATIONS - FORECASTING PRACTICES .31 5.10 LEVEL – REPEATABLE 31 5.101 Forecast set-up 31 5.101.1 Define, justify and document assumptions 31 5.101.2 Collaboratively define and agree on the forecasting problem 31 5.101.3 Define appropriate timeframe for the forecast 32 5.101.4 Document constraints and risks 32 5.101.5 Define appropriate level of detail 32 5.101.6 Establish clear lines of communication between users and forecasters 33 5.102 Technique Selection 33 5.102.1 Determine the characteristics of the situation being forecast 33 5.102.2 Determine resources requirements 33 5.102.3 Assess quality of available input 33 5.102.4 Assess type of output required 34 5.102.5 Assess known strengths and weaknesses of techniques 34 5.102.6 Account for product life-cycle 36 5.102.7 Determine when to use judgmental techniques 36 5.104 Forecast accuracy 37 5.104.1 Lower uncertainty but be realistic 37 5.104.2 Evaluate the situation to determine accuracy requirements 37 5.104.3 Account for demand stimulation activity 38 5.105 Short term forecasting considerations 38 5.105.1 Define components of the forecast 38 5.105.2 Distinguish between sales, shipment and demand 38 5.105.3 Use statistical models 39 5.105.4 Determine granularity 39 5.105.5 Objectify the process 39 5.20 LEVEL – DEFINED 39 5.201 Define aggregation and combination rules 39 5.202 Identify turning points and trends 40 5.203 Present results simply and graphically 40 5.30 LEVEL – MANAGED 40 5.301 Participate in all major business area meetings regarding forecast interpretation 40 5.302 Review with users actual results versus forecast predictions 40 5.303 Forecast collaboratively across distribution channels 41 5.40 LEVEL – OPTIMIZING 41 5.401 Long term forecasting considerations 41 5.401.1 Use judgment but support it with quantitative methods 41 5.104.2 Account for economic cycles 42 6.0 REFERENCES 42 Best Practices for Forecasting Copyright 2002 - Inforte Corporation 1.0 INTRODUCTION & OBJECTIVES Forecasting, a critical operational function for virtually all organizations, provides an organization with views of coming sales, changing patterns and long term trends It is nearly impossible to align the capacities, assets and resources of the firm appropriately without good forecasting practices This results in a failure to meet demand in robust economies and red ink in weak economies While some companies rely on bottom-up forecasts from field sales, others depend on topdown planning from a central function However, very few companies have adopted a systematic and well-organized approach to forecasting that accommodates forecasts from different parts of the organization with different levels of detail and different horizons Furthermore, few organizations have established management processes allowing all functions within the firm to systematically review and update actions based on forecast changes Consequently, most enterprises not move in unison with demand changes This document, addresses the need for a systematic approach to forecasting It identifies the key organizational, management, process, and operational aspects of forecasting that allow a forecast to be the center of enterprise planning and the driver of corporate decisions Companies that have successfully implemented these approaches are able to keep supply and demand in balance in any type of economic environment Because their profitability is less volatile than other firms, they meet their earnings projections, allowing management to focus on strategic issues instead of fighting the fires caused by surprises and losses Forecasting is at the heart of the demand-driven enterprise Forecasting is at the heart of Demand Chain Management (DCM) - the operational process of projecting, capturing, stimulating and responding to demand in an integrated, enterprise-wide fashion Companies that this well, tend to produce a consistent profit performance due to closer control over supply and demand An effective DCM program begins with effective demand forecasting - an area that has been neglected within many corporations This document focuses on providing best practices for forecasting as a first and vital step toward DCM Research reveals the following observations about forecasting in the Fortune 500: • • • • • Proven forecasting techniques are applied poorly in most organizations Forecast results are not communicated adequately across the organization Systematic processes that allow each and every department and business unit to respond properly to projected demand levels not exist or are underdeveloped Forecasting is currently static in most organizations and should be more continuous Due to poor forecasting practices, firms are missing a major opportunity to correlate costs and revenue much more closely, regardless of the prevailing demand environment Good forecasting leads directly to higher revenue-cost correlation and higher profitability In addition, good forecasting practices can help highlight specific areas of high customer demand, even in poor demand environments It can also provide valuable feedback for product design and marketing as it detects emerging buying preferences It is important to note, this document is not a comprehensive “how-to” manual on the various methods of forecasting It instead focuses on providing best practices for forecasting as a first Best Practices for Forecasting Copyright 2002 - Inforte Corporation and vital step toward DCM The objectives of these best practices are to provide senior management and forecasters with a fuller picture of how forecasting fits into the operations and strategy of the firm This document also provides guidelines for the initiatives and procedures needed to become excellent at forecasting This document also does not cover the many detailed processes required for the enterprise to effectively respond to the results of good forecasting We allude to those processes in this document However, they are described in a separate paper by Inforte entitled Best Practice for Enterprise Response to Demand Some of the processes outlined in the paper include: • • • • • • • • Supply Chain Responsiveness Revenue Management Inventory Management Sales Incentives Market Segmentation Dynamic Budgeting Procurement Financial Planning As well as these generic responsiveness procedures there are many industry specific practices for responding to demand, such as risk management within financial services - these are addressed in a series of Inforte papers on industry-specific demand chain management Capability Maturity Model The best practices contained within this document are framed within Inforte’s Demand Forecasting Capability Maturity Model (CMM) The Capability Maturity Model (CMM) for Demand Forecast & Response describes the principles and practices underlying demand forecast and response process maturity It is intended to help companies improve the maturity of their demand forecast/response processes in terms of an evolutionary path from ad hoc, chaotic processes to mature, disciplined demand forecasting and response processes Once a firm has determined the level at which it resides, it is easier to determine the processes and tools they must implement to achieve a more effective demand chain management program Inforte’s Demand Forecast & Response CMM is a top-down, assessment-based framework; it is not a bottom-up, business problem framework This is why it is important that a firm move upward from one level to the next Each level describes certain key processes that must be in place before residing on that level Additionally, for an organization to reside on a certain maturity level, they must have implemented all of the key processes for that level, and those of the lower levels The key processes are not intended to require a specific implementation or organizational structure Instead, they relate to activities that the organization must implement to reach a certain level of maturity The manner in which they are implemented can vary from firm to firm Additionally, the term key process simply means that these processes are key to reach the next level of maturity However, there may be additional “non-key” processes that are useful but not mandatory to reach the next level Each section within this document outlines the activities and best practices that should be implemented at each stage of the Demand Forecast & Response CMM It does not, however, provide best practices for the Initial level as all forecasting this maturity level is ad-hoc – a practice not recommended for any firm Best Practices for Forecasting Copyright 2002 - Inforte Corporation DEMAND FORECAST & RESPONSE CAPABILITY MATURITY MODEL – Optimizing All functional decisions enterprise-wide are continuously made and adjusted based on contextually-relevant demand forecast information – Managed Detailed measures of the forecast process accuracy and response are collected; enterprise response is primarily through scheduled dynamic departmental budget allocations throughout the quarter – Defined Standard enterprise demand forecast process for the organization; typical enterprise response is through quarterly departmental budget adjustments with business development initiatives periodically dynamically adjusted throughout the quarter – Repeatable – Initial Objective, statistically-based demand forecasts; each business area has formal forecasting processes; response processes include yearly departmental budgets updates with business development initiatives adjusted quarterly Ad-hoc, subjective forecasting 1) Initial The demand forecast/response process is characterized as ad hoc, and occasionally even chaotic Few processes are defined, and success depends on individual effort or heroics, with forecasts often including subjective or judgmental inputs Process Areas: There are no key process areas at the “Initial” level Except for Level 1, each maturity level is decomposed into several key process areas that indicate the areas an organization should focus on to improve its demand forecast and response processes Horizon: undefined (i.e changes with each forecast) Frequency: ad-hoc (i.e only run when management feels it’s necessary) Tools: Heavy reliance on subjective forecasting; Excel spreadsheets Metrics: No formal measurement of accuracy of forecasts or responsiveness of the enterprise 2) Repeatable Basic objective, statistically-based demand forecast management processes are established to track history, accuracy, and actuals The necessary process discipline is in place to repeat earlier successes with product lines/business units/divisions with similar data Typical enterprise response is primarily through yearly departmental budgets, and also with business development initiatives (sales/marketing/customer service plans) adjusted quarterly Process Areas: The key process areas at Level focus on the product line/business unit/divisional concerns related to establishing basic, objective, statistically-based demand forecasting controls They are Customer Information Capture, Departmental Forecast Creation, Forecast Review, Executive Alignment, Departmental Response, and Organization Process Definition Best Practices for Forecasting Copyright 2002 - Inforte Corporation Demand Information Capture: ability to systematically and objectively capture customer demand information in each department/business unit/product line/channel Departmental Forecast Creation: ability to create a statistically-based forecast for the department/division/etc Forecast Review: formal rules, tools, and processes are defined in each department/division/etc for the interpretation of forecast results Executive Alignment: full senior management commitment to forecasting and response as an enterprise-wide commitment Departmental Response: response process in place to adjust departmental/division budgets and plans based on forecast Organization Process Definition: a formal structure for the forecasting function is defined and sufficient resources and budget are allocated to the forecasting function Horizon: Medium to long-term forecasting; looking to determine demand for the next year and, occasionally, the upcoming quarter Frequency: Forecasts are produced on a yearly basis for use in departmental budget adjustments and on a semi-annual (or quarterly) basis for adjustment of business development plans Tools: Customer Relationship Management System, Opportunity Management System, Supply Chain Management System, Statistical Forecasting Program, Data Warehousing, Demand Planning System, Marketing Analytics System, and Order Management System Metrics: departmental forecast accuracy, departmental response time, budget variance 3) Defined The demand planning process for forecast activities is documented, standardized, and integrated into a standard enterprise demand forecast process for the organization All forecasts use an approved, tailored version of the organization's standard forecast process for developing and maintaining forecasts Typical enterprise response is primarily through quarterly departmental budget adjustments, and also with business development initiatives (sales/marketing/customer service plans) periodically dynamically adjusted throughout the quarter Process Areas: The key process areas at Level address both product line/business unit/divisional and organizational issues, as the organization establishes an infrastructure that institutionalizes effective demand forecast management processes across all product lines/business units/divisions They are Aggregated Input Collection, Standard Output Distribution, Enterprise Response, Governance Process Development, and Corporate Communication Process Aggregated Input Collection: a standard, objectified process is in place across all product lines/business units/divisions for collecting forecast inputs to help with the creation of an enterprise-wide forecast Standard Output Distribution: standard, objectified process for distribution of a unified forecast to all product lines/business units/divisions Enterprise Response: enterprise responsiveness goals are set; standardized processes for adjusting budgets, inventory policies, resources, service levels, etc to the forecast exist across the organization Governance Process Development: the development of standard forecasting meeting schedules, agendas, participants, roles and responsibilities across the organization Corporate Communication: process for communicating expectations and gathering feedback on corporate forecasting and response policies and goals Horizon: Medium-term forecasting; goal is to determine demand on a quarterly basis and, occasionally, adjust the forecast once or twice a quarter Best Practices for Forecasting Copyright 2002 - Inforte Corporation Frequency: Forecasts are prepared on a quarterly basis for departmental budget adjustments and are periodically, dynamically adjusted for use in business development plans throughout the quarter Tools: Executive Dashboard, Output Distribution System, Decision Support System, Employee Relationship Management System, Responsiveness Scorecard, and Inventory Optimization System Metrics: enterprise forecast accuracy, enterprise response time, stock-out/capacity out situations, campaign effectiveness, product introduction rate, warehousing costs, obsolete/excess inventory cost, time-to-market 4) Managed Detailed measures of the forecast process accuracy and response are collected Both the forecast process and responses are quantitatively understood and controlled Typical enterprise response is primarily through periodic scheduled dynamic departmental budget allocations throughout the quarter, requiring prioritization of functional initiatives based on contextually-relevant demand forecast information (units/headcount requirements/etc instead of recognized revenue), with most business development activities (sales/marketing/customer service decisions) tied directly into demand forecast information Process Areas: The key process areas at Level focus on establishing a quantitative understanding of both the forecast process and the enterprise response They are Forecast Performance Monitoring, Quantitative Process Management, Forecast Accuracy Assurance, and Collaborative Inter-Firm Forecasting Forecast Performance Monitoring: review process that includes continuous review through a high level of collaboration between users and forecasters during the forecast process as well as a formal monthly or quarterly review process with the forecast team and users to asses forecast performance and define improvement priorities Quantitative Process Management: control the process performance and cost of the forecast creation/distribution and response process quantitatively Forecast Accuracy Assurance: reviewing and auditing of working procedures to see that they comply with applicable standards and procedures Management is provided with the results of the reviews and audits Collaborative Inter-Firm Forecasting: define process for collecting inputs and sharing results with other value system partners Contextually-relevant Forecasting: process for turning the enterprise-wide forecast into the most relevant view for the department/division Horizon: Short-term, operational forecasts; goal is to assess near-term demand (i.e anywhere from several times a quarter to hourly/weekly for business development initiatives) Frequency: Departmental forecasts are produced and adjusted several times a quarter while business development activities utilize a continuous forecasting/response approach Tools: Accuracy Scorecard, Functional/Departmental Application Integration to Executive Dashboard, and Interconnectivity with External Partners Metrics: forecast error variability, order processing lead time, fulfillment percentage per customer, supplier lead time, resource allocation, call center response time, return-rate, customer service levels, service cost per customer, product mix effectiveness, inventory turns 5) Optimizing Continuous process improvement is enabled by quantitative feedback from the process and from piloting innovative ideas and technologies All functional decisions enterprise-wide are continuously made and adjusted based on contextually-relevant demand forecast information Best Practices for Forecasting 10 Copyright 2002 - Inforte Corporation ƒ ƒ ƒ ƒ ƒ • Are significant changes in management decisions expected? Are significant environmental changes expected? Are significant shifts expected among variable relationships? How critical are seasonal variations? How likely is it that turning points may occur or that market shifts are happening? Request that the organization starts tracking data that is commonly required but not currently tracked 5.102.4 Assess type of output required • Determine the type of output required by the user, by determining the uses for the forecast results: ƒ How much detail is required? ƒ What is the accuracy requirement? ƒ In what form is the output required? ƒ Should the forecast be capable of detecting direction changes? ƒ Are component forecasts required? ƒ Is a high level of accuracy critical? ƒ Should turning points be reflected promptly? ƒ Should turning points be identified early? ƒ Is an interval or probabilistic forecast critical? 5.102.5 Assess known strengths and weaknesses of techniques • Assess which techniques are best suited for the type of forecast being complied and how they can be combined for a better overall result Some known strengths and weaknesses include9: ƒ Overwhelming evidence proves statistical forecasting is more beneficial and can be performed at a lower cost than judgmental forecasting ƒ Statistical models, such as time series, tend to be most appropriate for operational (short-term) forecasting ƒ Weighting time series models toward more recent data produces better results (e.g exponential smoothing techniques) ƒ When randomness dominates trend cycles, as is often the case with shortterm data, single exponential smoothing is often the most accurate approach ƒ Judgmental methods work best in unique situations where precedents not exist, or for very long term forecasting ƒ Regression techniques are used best in situations that are well understood and for which plenty of historical data exists They tend to be popular for medium term (e.g 24 months) forecasts For detailed selection criteria, see: Georgoff, D & Murdick, R (1986) Manager’s Guide to Forecasting Harvard Business Review, 1/1/86 Chambers, J., Mullick, S & Smith, D (1971) How to Choose the Right Forecasting Technique Harvard Business Review, 7/1/71 Best Practices for Forecasting 34 Copyright 2002 - Inforte Corporation • Review forecaster surveys There have been more than 25 surveys among forecasting users since 197010 When assessing the strengths and weaknesses of various techniques, it is useful to review these surveys as they show the opinions forecasters hold about the various methods they have used The results can be summarized as follows11: Results of Forecast Surveys 80 70 60 50 Satisfied 40 Neutral Dissatisfied 30 20 10 ƒ ƒ ƒ ƒ ƒ ƒ ƒ Box-Jenkins Straight-line projection Life cycle analysis Simulation Classical decomposition Tend-line analysis Moving average Exponential smoothing Regression Sales force composite Customer expectations Jury of executive opinion In general, forecasters are less satisfied with subjective (jury of executive opinion, customer expectations and sales force composite) than with objective methods Highest satisfaction goes to regression although empirical studies show time series methods are more accurate than the explanatory regression methods Classical decomposition does not fare well in surveys although empirical evidence suggests that the ability to decompose a series into seasonality, trend-cycle and randomness is of high importance Among the subjective methods (jury of executive opinion, customer expectations and sales force composite) users are quite dissatisfied with sales force and customer methods Jury of executive opinion is the most widely used method, though unfortunately has been proven to be less than accurate (due to biases) Sales force composites and customer expectations are used for short-term forecasts However, over reliance introduces biases Exponential smoothing and moving averages are used most often for shortterm forecasts – consistent with empirical data indicating these methods perform best over shorter time horizons 10 Winklhofer, H., Diamantopolous, A & Witt, S (1996) Forecasting Practice: A Review of the Empirical Literature and an Agenda for Future Research International Journal of Forecasting, 12, 193-221 11 Makridakis, S., Wheelwright, C & Rob, J (1997) Forecasting: Methods & Applications New York, NY: Wiley Best Practices for Forecasting 35 Copyright 2002 - Inforte Corporation ƒ ƒ • Straight-line projection is used fairly frequently even though it is inaccurate in short and medium term forecasting Regression methods are used most often for medium term forecasts consistent with theory indicating medium and long term forecasting should focus on understanding the variables to be forecast and the factors that influence them Ensure proper examination of the benefits is made before deciding to switch from an established method 5.102.6 Account for product life-cycle • Account for the stage of the product/service in its lifecycle when selecting techniques Here are examples of techniques commonly used at various points of the lifecycle12: ƒ Product Development Delphi method (for undefined markets) Historical analysis of comparable products Priority pattern analysis (describing consumer preferences and the likelihood they will buy a product) Input-output analysis Panel consensus Judgments of expert personnel are often needed as input to forecast at this stage Comparing to similar or ancestor products is a common technique ƒ Market Testing & Early Introduction Consumer surveys Tracking and Warning Systems Market tests Experimental designs It is important to differentiate between sales to innovators and sales to mainstream adopters For product introduction, product differences measurement may be effective ƒ Rapid Growth Statistical techniques for identifying turning points Tracking and warning systems Market surveys Intention to buy surveys ƒ Steady State Time series analysis and projection Causal and econometric models Market surveys for tracking and warning Life-cycle analysis 5.102.7 Determine when to use judgmental techniques • Use judgmental forecasting techniques selectively They should be used to identify forthcoming changes and predicting the direction and extent to which these changes will influence the future In this way, statistical predictions, which can more objectively and correctly identify and extrapolate established patterns and/or existing relationships, can be appropriately modified 12 For more detail see: Chambers, J., Mullick, S & Smith, D (1971) How to Choose the Right Forecasting Technique Harvard Business Review, 7/1/71 Best Practices for Forecasting 36 Copyright 2002 - Inforte Corporation • The forecaster should incorporate subjective judgments in dynamic situations when the quantitative models not reflect significant internal and external changes Even in these cases, the forecaster should incorporate the subjective adjustments as inputs in the model rather than adjusting the model’s final outcome • Use documented guidelines to determine when to appropriately apply judgments in the forecasting process Research shows that when historical data is available, judgmental adjustments tend to reduce accuracy (and increase forecasting cost) The following are situations where judgment adjustments may, however, be appropriate: ƒ A competitor is known to be going out of business ƒ Interest rates are expected to change ƒ A natural disaster has affected economic conditions in a region In this case the forecaster might use judgment to select an appropriate precedent and then attempt to quantify affect based on this historical data ƒ Other situations where past history is expected not to be an accurate predictor of future events 5.104 • 5.104.1 Forecast accuracy It is vital that forecast accuracy is carefully tracked Not only does it form the basis for many statistical projection models, it is also used at the corporate level to determine the amount of slack to be kept in assets, capital and resources Additionally, it forms a benchmark used to determine the level of desired enterprise responsiveness The enterprise should be able to respond fast enough to make up for the average error in the forecast Lower uncertainty but be realistic • Ensure forecasting is as accurate as possible while the magnitude of forecasting errors, or the extent of the uncertainty involved when predicting the future, is as small as possible but also estimated as realistically as possible • Ensure users and managers realize that while forecasters should always strive for greater accuracy, the maximum accuracy to expect from a technique must fall within the range bounded by the average percentage error of the random component of the data series 5.104.2 Evaluate the situation to determine accuracy requirements • Different situations require different levels of accuracy For example, forecasting to determine whether to enter a line of business does not have to be as accurate as a forecast used to set budgets • Predetermine the level of inaccuracy that is tolerable For example, calculate the ROI of lower safety stocks versus accuracy • Keep in mind that historical data will be more accurate predicting near term results Due to changing business conditions and various initiatives carried out by the company, it will likely be less accurate over the longer term Best Practices for Forecasting 37 Copyright 2002 - Inforte Corporation 5.104.3 • 5.105 • 5.105.1 Account for demand stimulation activity Take into account promotions and other special actions that are meant to change historical patterns and relationships and are, therefore, part of the “performance” being evaluated Short term forecasting considerations Short-term forecasting typically involves projecting revenue streams over the next 3-6 months This is the most commonly generated forecast within most enterprises and drives most day-to-day operational decisions across the firm Careful planning, coordination and execution of short-term forecasts are required due to the various complexities involved Define components of the forecast • Realize that for most large companies, forecasting is carried out over multiple business units and within each business unit, multiple product lines A critical factor is that different techniques will be needed for the various areas and that specific aggregation rules must be applied • Consider that some components may not be under direct control of the organization, such as an acquired company, a third party product, an alliance with another firm or a franchise • The operational forecast to be generated should be broken down into its component parts Each component may require a separate technique and specific forecast to be generated For example: Business Unit A ƒ Product a (direct channel) ƒ Product a (distributor channel) ƒ Product a (Retail channel) ƒ Product b ƒ Joint alliance product ac ƒ Assortment product d ƒ Service e Business Unit B (Acquired entity) ƒ Product c ƒ Product d ƒ Service line e ƒ Resale product f 5.105.2 • Distinguish between sales, shipment and demand It is important to distinguish among sales, shipment and demand, especially when communicating with user groups Making this distinction is critical to determine demand levels and forecast accuracy ƒ Demand - what customers want to buy ƒ Sales - the ability to accept orders from customers ƒ Shipments - what operations systems can actually deliver Best Practices for Forecasting 38 Copyright 2002 - Inforte Corporation 5.105.3 • 5.105.4 Use statistical models Use statistical models to generate operational forecasts With statistical models trends tend to emerge gradually, making them excellent for predicting short-term sales Determine granularity • Determine the level of granularity of the components Often this will be a trade off between the following factors: ƒ Time required to forecast and available resources ƒ Accuracy ƒ Availability of data ƒ Comparative accuracy of higher aggregate levels • Consider that each component may well require a different forecasting method In most cases, components to be forecast must be identified before the appropriate technique can be identified 5.105.5 Objectify the process • Quantitative forecasts have been shown to be more accurate than ones based on human judgments This is because significant biases naturally influence the judgmental process • Strive to remove judgments as much as possible during the preparation of the forecast Various guidelines for achieving this include: ƒ Ensure opportunities are categorized systematically using standard rules ƒ Ensure the assessment of probabilities associated with revenue streams, opportunities, etc are rule-based and that the rules are applied consistently ƒ Ensure that sales force opinion on opportunities is be minimized in the preparation of the forecast Facts about each case are all that should be considered ƒ Ensure that opinions are not incorporated into the quantitative forecast Executives are free to interpret the results of the forecast as necessary It is at this time that opinion becomes useful Although even here, executives tend over influence forecast interpretation and therefore respond inappropriately to results ƒ Ensure that for forecasting revenue streams, a structured process is in place to categorize sales opportunities 5.20 Level – Defined 5.201 Define aggregation and combination rules • Aggregate by using detailed component forecasts and rolling up If the aggregate is determined first (through a top-down approach), the forecaster must resort to past patterns to determine component forecasts, at a higher risk of inaccuracy This is especially important, as component forecasts are typically the ones used to set the detail of the departmental quotas and budgets Best Practices for Forecasting 39 Copyright 2002 - Inforte Corporation • Combine forecasts for more accuracy Combining forecasts, especially those using different techniques, is a proven way to dramatically increase accuracy.13 In the majority of cases, combining or averaging forecasts produces more accurate results • Clearly define the division of responsibility and create a clear methodology for combining forecasts • Documentation of the aggregation process is critical to ensure forecast accuracy and performance can be tracked accurately 5.202 Identify turning points and trends • A key task of the forecasting process should be identifying turning points • Adjust results for trend, seasonal variation, cyclical variation and irregular movements This is in addition to executing the primary model In this sense, model execution may be somewhat iterative • Use the established techniques for adjusting for seasonality The simplest approach for estimating seasonality in time series analysis is the classical decomposition approach 5.203 Present results simply and graphically • Present results in a form that gives a central mean value and a range of possible outcomes • Represent the forecast results graphically and contextually 5.30 Level – Managed 5.301 Participate in all major business area meetings regarding forecast interpretation • 5.302 Attend all major discussions involving forecasts throughout the business The forecasting function should also be represented at all corporate level meetings where divisional forecasts are integrated Review with users actual results versus forecast predictions • Review formally all forecasts versus actual results with appropriate user groups Specific discussion of areas to improve should be conducted • Ensure management, forecasters and users can explain all material differences between forecast and actual results • Ensure forecast review results are communicated to the executive steering committee and feedback is solicited from the executive steering committee on forecast performance 13 For more detail on combining forecast: Georgoff, D & Murdick, R (1986) Manager’s Guide to Forecasting Harvard Business Review, 1/1/86 Best Practices for Forecasting 40 Copyright 2002 - Inforte Corporation 5.303 Forecast collaboratively across distribution channels • Special consideration is required for forecasting across the firm’s multiple distribution partners There are many types of partners - resellers, large distributors, small distributors, retail outlets, etc • Collaboration is necessary to achieve the best results when forecasting across multitiered distribution systems • Map the network dynamics so that the key points of demand are identified It may be unrealistic to forecast in detail across every channel and outlet, but a relatively small group of key points in the system may be the key to the demand picture • Dedicate a team of forecasters drawn from the key firms in the network Their responsibilities should include: ƒ Agreeing upon the methods and procedures for forecasting ƒ Keeping the goal of optimizing the system in mind rather than the individual company ƒ Applying the same standards across all entities and forecasting components ƒ Creating common forecast output which can then be contextualized for each firm in the system ƒ Ensuring senior management is committed ƒ Ensuring commitment to a “service level” of responsiveness so that the collaborative forecasters can be assured of a consistent response across all organizations ƒ Tracking accuracy and monitoring “dark spots” in the system where forecasts are less detailed or where inaccuracies persist 5.40 Level – Optimizing 5.401 Long term forecasting considerations • 5.401.1 Long term forecasting is far more difficult than short term forecasting as past data does not necessarily provide good guidance to the future and quantitative methods are limited The forecaster must incorporate changing relationships between the independent variables in the forecast Use judgment but support it with quantitative methods • Use historical precedent, judgment or counting methods for situations with extended horizons or with novel situations that have limited data and no historical precedent Applying judgment in these situations should be structured carefully • Use judgment to stimulate thought and explore new relationships but, where possible, incorporate quantitative techniques to test and support assumptions • Use scenario techniques and ensure that contingency plans are built for the range of most likely scenarios14 14 For more detail see: Two part article on scenario forecasts: Wack, P (1985) Scenarios: Uncharted Waters Ahead Harvard Business Review, 9/85 Wack, P (1985) Scenarios: Shooting the Rapids Harvard Business Review 11/85 Best Practices for Forecasting 41 Copyright 2002 - Inforte Corporation 5.104.2 6.0 Account for economic cycles • Ensure users understand key facts about economic cycles, as they can and change patterns ƒ We have not been able to predict the timing or depth of recessions or the start and strength of booms ƒ This makes medium/long term forecasting hazardous as recessions and booms can start anytime in a planning horizon of 18 months, the usual length of medium term forecasts ƒ Typically, the deeper the recession the worse the forecasting errors ƒ In booms, opposite errors can result in fulfillment issues, erosion of customer satisfaction and serious consequences to competitive position • Monitor for leading indicators in order to get a better sense for the possibility of economic downturn Firms can get significant advantage in predicting a downturn more quickly than rivals Leading indicators include: ƒ GDP line item trends for the good/service or industry ƒ Stock market weakness ƒ Global economic weaknesses ƒ Change in customer sentiment ƒ Profitability changes of key customers ƒ Trends in certain products versus others ƒ Increasing competition ƒ Increasing price pressure REFERENCES Ajinkya, B & M Gift (1984) Corporate Managers Earnings Forecasts and Symmetrical Adjustments of Market Expectations Journal of Accounting Research, Autumn, 425-444 Armstrong, J (1978) Forecasting by Extrapolation: Conclusions from 25 Years of Research Interfaces, 14 Armstrong, J (1978) Forecasting with Econometric Methods: Folklore Versus Facts Journal of Business, 51, Armstrong, J (1978) Long Range Forecasting: From Crystal Ball to Computer New York, NY: Wiley Ascher, W (1978) Forecasting: An Appraisal for Policy Makers and Planners Baltimore, MD: John Hopkins University Press Ashley, R (1988) On the Relative Worth of Recent Macroeconomic Forecasts International Journal of Forecasting, Bails, D & L Peppers (1993) Business Fluctuations: Forecasting Techniques and Applications Upper Saddle River, NJ: Prentice Hall Barnett, W (1988) Four Steps to Forecasting Total Market Demand Harvard Business Review, 7/1/88 Bass (1969) A New Product Growth Model for Consumer Durables Management Science, Jan Best Practices for Forecasting 42 Copyright 2002 - Inforte Corporation Bass, King & Pessemeier (1968) Applications of the Sciences in Marketing Management New York, NY: Wiley Bernstein, P (1996) Against the Gods: The Remarkable Story of Risk New York, NY: Wiley Bowman, E (1963) Consistency and Optimality in Managerial Decision Making Management Science, 10, 310-321 Box, G & G Jenkins (1970) Time Series Analysis, Forecasting & Control Upper Saddle River, NJ: Prentice Hall Bretschneider, S., W Gorr, G Grizzle & E Klay (1989) Political and Organizational Influences on the Accuracy of Forecasting State Government Revenues International Journal of Forecasting, 5, 307-319 Bright, J (1973) A Guide to Practical Technological Forecasting Austin, TX: Prentice Hall Brown (1959) Less Risk in Inventory Estimates Harvard Business Review, July-August Bunn, G & G Wright (1991) Interaction of Judgmental and Statistical Forecasting Methods: Issues and Analysis Management Science, 37, 501-518 Callahan, B (1988) What Price Integrity? Best’s Review, October, 22-24 Chambers, J (1974) An Executive’s Guide to Forecasting New York, NY: Wiley Chambers, J., S Mullick & D Goodman (1971) Catalytic Agent for Effective Planning Harvard Business Review, Jan-Feb, 110 Chambers, J., S Mullick & D Smith (1971) How to Choose the Right Forecasting Technique Harvard Business Review, 7/1/71 Chatfield, C (1993) Neural Networks: Forecasting Breakthrough or Passing Fad International Journal of Forecasting, Chow, G (1966) Technological Change and the Demand for Computers The American Economic Review, Dec Clark, A (2000) Profiles of the Future New York, NY: Bantam Books Cleary, J & H Levenbach (1982) The Professional Forecaster: The Forecasting Process Through Data Analysis Belmont, CA: Lifetime Learning Publications Clelland, R (1966) Basic Statistics with Business Applications New York, NY: Wiley Clemen, R (1989) Combining Forecasts: A Review and Annotated Bibliography International Journal of Forecasting, 5, 559-584 Clifton, P., H Nguyen & S Nutt (1992) Market Research Using Forecasting in Business Oxford: Butterworth & Heinemann Conroy, R & R Harris (1987) Consensus Forecasts of Corporate Earnings: Analysts' Forecasts and Time Series Methods Management Science, 33, 725-738 Crosby, J (1999) Cycles, Trends and Turning Points New York, NY: NTC Publishing Group Darymple, D (1975) Sales Forecasting Methods and Accuracy Business Horizons, December Best Practices for Forecasting 43 Copyright 2002 - Inforte Corporation Darymple, D Sales Forecasting Practices International Journal of Forecasting, Dawes, R., D Faust & P Meehl (1989) Clinical Versus Actuarial Judgment Science, 243, 1668-1674 De Gooijer, J & K Kumar Some Recent Developments in Non-linear Time Series Modeling, Testing and Forecasting International Journal of Forecasting, Diebold, F (2000) Elements of Forecasting (2nd Edition) Stamford, CT: Southwestern Thomas Learning Dreman, D (1991) Flawed Forecasts Forbes, December 9, 342 Einhorn, H & R Hogarth Prediction, Diagnosis and Causal Thinking Journal of Forecasting, Jan-Mar 1982, 23 Evans, M (1969) Macroeconomic Activity: Theory, Forecasting & Control New York, NY: HarperCollins Evans & Preston Discussion Paper #138: Economic Input-Output Model Wharton School of Finance: University of Pennsylvania Fildes, R & C Beard (1992) Forecasting Systems for Production and Inventory Control International Journal of Operations and Production Management, 12 Fildes, R & S Makridakis (1995) The Impact of Empirical Accuracy Studies on Time Series Analysis and Forecasting International Statistical Review, 63 Fisher, M., J Hammond, W Obermeyer & A Raman (1994) Making Supply Meet Demand in an Uncertain World Harvard Business Review, 5/1/94 Fox, J (1997) The Economic Outlook: Reasons to Worry Fortune, Feb Galbraith, C & G Merrill (1996) The Politics of Forecasting: Managing the Truth California Management Review, 1/1/96 Gardner, E & D Dannenbring Forecasting with Exponential Smoothing: Some Guidelines for Model Selection Decision Science, 11 Garland, L The Problem of Observer Error Bulletin of the New York Academy of Medicine 569-584 Georgoff, D & R Murdick (1986) Manager’s Guide to Forecasting Harvard Business Review, 1/1/86 Goldberg, L (1970) Man vs Model of Man: A Rationale, Plus Some Evidence, For a Method of Improving on Clinical Inferences Psychological Bulletin, 422-432 Goodwin, P & G Wright (1993) Improving Judgmental Time Series: A Review of the Guidance Provided by Research International Journal of Forecasting, 9, 147-161 Gonik, J (1978) Tie Salesmen’s Bonuses to their Forecasts Harvard Business Review, 5/1/78 Hadley (1968) Introduction to Business Statistics Hanke, J & A Reitsch (1997) Business Forecasting (6th Edition) Old Tappan, NJ: Prentice Hall Best Practices for Forecasting 44 Copyright 2002 - Inforte Corporation Harvey, N (1995) Why are Judgments Less Consistent in Less Predictable Situations? Organizational Behavior and Human Decision Processes, 63 (3), 247-263 Hastie, R & R Dawes (1988) Rational Choice in an Uncertain World San Diego, CA: International Thomson Publishing Henry, R & J Sniezek (1993) Situational Factors Affecting Judgments of Future Performance Organizational Behavior and Human Decision Processes, 54, 104-132 Hogarth, R & S Makridakis (1981) Forecasting and Planning: An Evolution Management Science, 27, 115-138 Hughes, D., S Madridakis & S Wheelwright (1987) Handbook of Forecasting: A Manager’s Guide New York, NY: Wiley Janis, L (1972) Victims of Groupthink Boston, MA: Houghton Mifflin Jennings, R (1987) Unsystematic Security Price Movements, Management Earnings Forecasts, and Revisions in Consensus Analysts Earnings Forecasts Journal of Accounting Research, Spring, 90-110 Kahneman, D & A Tversky (1979) Intuitive Prediction: Biases and Corrective Procedure TIMS Studies in Management Sciences, 12, 313-327 Kahneman, D P Slovic & A Tversky (1982) Judgment Under Uncertainty New York, NY: Cambridge University Press Kwong, K., L Cheng, V Simunek & C Jain (1994) Bibliography on Forecasting & Planning Flushing, NY: Graceway Publishing Co Lawrence, M., R Edmunsen & M O'Connor (1986) The Accuracy of Combining Judgmental and Statistical Forecasts Management Science, 32, 1521-1532 Lawrence, M & S Makridakis (1989) Factors Affecting Judgmental Forecasts and Confidence Intervals Organizational Behavior and Human Decision Processes, 54, 104-132 Lawrence, M & M O'Connor (1992) Exploring Judgmental Forecasting International Journal of Forecasting, 8, 15-26 Leontief, W (1966) Input-Output Economics Oxford: Oxford University Press Lippitt, V (1969) Statistical Sales Forecasting Ann Arbor, MI: George Wahr Publishing Co Mahmoud, E (1984) Accuracy in Forecasting: A Survey Journal of Forecasting, April-June 1984, 139 Madridakis, S (1982) The Accuracy of Extrapolation (Time Series) Methods Journal of Forecasting, April-June 1982, 111 Makridakis, S (1982) The Chronology of the Last Recessions Omega, 10 Makridakis, S (1990) Forecasting and its Role and Value for Planning and Strategy International Journal of Forecasting, 12, 513-539 Makridakis, S (1990) Forecasting, Planning and Strategy for the 21st Century New York, NY: The Free Press Best Practices for Forecasting 45 Copyright 2002 - Inforte Corporation Makridakis, S., A Andersen, R Cabone, R Fildes, M Hibon, R Lewandowski, J Newton, E Parzen & R Wrinkler The Accuracy of Extrapolation (Time Series) Methods: Results of a Forecasting Competition Journal of Forecasting, Makridakis, S., C Chatfield, M Hibron, M Lawrence, T Mills, K Ord & L Simmons (1993) The M2-Competition: A Real Time Judgmentally Based Forecasting Study International Journal of Forecasting, Makridakis, S & M Hibon (1979) Accuracy of Forecasting: An Empirical Investigation Journal of the Royal Statistical Society, Series A, 142 Madridakis, S & R Hogarth (1981) Forecasting and Planning: An Evaluation Management Science, Feb 1981, 115 Maridakis, S & S Wheelwright (1982) Forecasting: Time Studies in the Management Sciences (Vol 12) Honolulu, HI: Institute of Management Sciences Maridakis, S & S Wheelwright (1989) Forecasting Methods for Management New York, NY: Wiley Makridakis, G., S Wheelwright & J Rob (1997) Forecasting: Methods & Applications (3rd Edition) New York, NY: Wiley Madridakis, S & R Winkler Averages of Forecasts: Some Empirical Results Management Science, 29, 987 Mathews, B & A Diamantopolous (1986) Managerial Intervention in Forecasting: An Empirical Investigation of Forecast Manipulation International Journal of Research in Marketing, 3, 3-10 McLaughlin & Boyle (1962) Time Series Forecasting American Marketing Association McNees, S (1986) Forecasting Accuracy of Alternative Techniques: A Comparison of U.S Macroeconomic Forecasts Journal of Business and Economic Statistics, McNichols, M (1989) Evidence of Informational Asymetrics from Management Earnings Forecast and Stock Returns The Accounting Review, Jan, 1-27 Meehl, P (1954) Clinical Versus Statistical Prediction: A Theoretical Analysis and Review of the Literature Holmes, PA: Jason Aronson Mentzer, J & J Cox (1984) Familiarity, Application and Performance of Sales Forecasting Techniques Journal of Forecasting, North, H & D Pike (1969) Probes of the Technological Future Harvard Business Review, May-June , 68 Nystom, P & W Starbuck (1981) Handbook of Organizational Design (Vol 1) Oxford: Oxford University Press Oliver & Boyd (1964) Techniques of Production Control ICI Pyatt, G (1964) Priority Patterns and the Demand for Household Durable Goods New York, NY: Cambridge University Press Retscheider, S & W Gorr (1989) Editorial International Journal of Forecasting, 5, 305506 Best Practices for Forecasting 46 Copyright 2002 - Inforte Corporation Riise, T and D Tjostheim (1984) Theory and Practice of Multivariate ARIMA Forecasting Journal of Forecasting, Sanders, N The Accuracy of Judgmental Forecasts: A Comparison OMEGA International Journal of Management Science, 20 (3), 353-364 Sanders, N & C Manrodt (1994) Forecasting Practice in US Corporations: Survey Results Interfaces, 24/2, 92-100 Schleifer, A (1996) Forecasting with Regression Analysis Harvard Business Review, 8/2/1996 Schnaars, S (1984) Situational Factors Affecting Forecasting Accuracy Journal of Marketing Research, August 1984, 290 Schuster, R On the Periodicity of Sunspots Philosophical Transactions, Series A, 206 Silk, L & M Curley (1970) Business Forecasting: With a Guide to Sources of Business Data New York, NY: Random House Silverman, B (1992) Judgment Error and Expert Critics in Forecasting Tasks Decision Sciences, 23 (5), 1199-1219 Sjoberg, L (1982) Aided and Unaided Decision Making: Improved Intuitive Judgment Journal of Forecasting, Oct-Dec 1982, 349 Spencer, Clark & Hoguet (1961) Business & Economic Forecasting Steiner, G (1997) Strategic Planning: What Every Manager Must Know New York, NY: The Free Press Stone, J & R Rowe (1960) The Durability of Consumers' Durable Goods Econometrica, 28-2 Thomopoulos, N (1980) Applied Forecasting Methods Upper Saddle River, NJ: Prentice Hall Urresta, L (1995) Recession, What Recession? Fortune, June 12 Vatter, P (1990) Chemplan Corporation: Paint-Rite Division Harvard Business Review, 11/1/90 Wack, Pierre (1985) Scenarios: Shooting the Rapids Harvard Business School, Nov 1985 Wack, Pierre (1985) Scenarios: Uncharted Waters Ahead Harvard Business School, Sept 1985 Walker, K & L McClelland (1991) Management Forecasts and Statistical Prediction Model Forecasts in Corporate Budgeting Journal of Accounting Research, 29 (2), 373-382 Webby, R & M O'Connor (1996) Judgmental and Statistical Time Series Forecasting: A Review of the Literature International Journal of Forecasting, 12, 91-118 Wheelwright, S (1989) Forecasting Methods for Management New York, NY: Wiley Wheelwright, S & D Clarke (1976) Corporate Forecasting: Promise and Reality Harvard Business Review, Nov-Dec, 54 Best Practices for Forecasting 47 Copyright 2002 - Inforte Corporation Wilson, H (1996) IABF Members' Opinions Regarding Forecasting Methods Presented in the IABF Winklhofer, H., A Diamantopolous & S Witt (1996) Forecasting Practice: A Review of the Empirical Literature and an Agenda for Future Research International Journal of Forecasting, 12, 193-221 Zarnowitz, V (1984) The Accuracy of Individual and Group Forecasts from Business Outlook Surveys Journal of Forecasting, Jan-Mar 1984, 10 Best Practices for Forecasting 48 Copyright 2002 - Inforte Corporation