FIFTY FOUR DEGREES | 25 Howmuch of what you buy could be considered wasteful? There are plenty of examples. You can think about howmuch food you buy and don’t eat, or howmany clothes you buy and don’t wear at all, or wear only once or twice. There is a growing level of awareness of issues around waste at the consumer level. But you can go through the supply chain and find waste all through the process. Retail outlets stock items that are never sold, from the everyday – groceries, clothing etc – to spare parts. If you take your car in for a service, the garage will use spare parts ordered from a wholesaler or a national distribution centre. They don’t want to stock too many and have waste, but if they stock too few, then they may create a different waste because people scrap their vehicles if they cannot have them repaired. This is a logistical challenge! The implications aremultiple. There is an economic implication, because a wholesaler who buys things that don’t sell, andwho can only possibly sell them for scrap, will get back only a fraction of what they paid. The same applies up and down the supply chain. Then there is the environmental cost. Resources are expendedmaking things that never sell, and there is the transport of these goods which adds to the carbon footprint, as well as the energy andmaterials used to dispose of unwanted products. All in all, wastage is very costly, both in terms of money and the environment. UNFORECASTABLE DEMAND THAT ISN’T The basic decisions that need to be made by retailers, wholesalers and distribution centres are ‘when should I replenish my stock?’ and ‘by how much?’ Most of the work in this area has focused on popular, fast-moving items. The demands for these products are generally easier to forecast – the higher the volume, on the whole, the more predictable it is. These are also the items at least risk of obsolescence. I am interested in those at the opposite end of the scale, those most difficult to forecast, where patterns of demand are not uniform. These are common in industries such as automobile manufacturing, aerospace, or any sector where there is a wide range of components involved and spare parts are required. They are not exceptions, but rather the majority of items in some industries, and they are at the highest risk of obsolescence wastage. For a long time, it was thought this was an area where we could not make progress. Just look at a typical demand pattern – no demand at all for several weeks, then demand for a small quantity, followed by another hiatus of a few weeks, topped off with demand for a larger quantity. What are you supposed to dowith items like that? How are you supposed to predict their demand? The key is to recognise that, although the next demand is unforecastable, the distribution of demand may be forecastable. The distribution assigns probability values to future possible demands, and it allows intelligent stocking decisions to be made. For example, if you are predicting only a 5% chance that demand will exceed 20 units, then buying 20 units will ensure that there is a 95%of not running out of stock. Improving demand distribution forecasts is not easy, and has been the subject of research by the Centre for Marketing Analytics and Forecasting over the last decade. This work has found its way into commercial software packages used by blue-chip companies around the world. UNFORECASTABLE DEMAND THAT IS The distribution of ‘noisy’ demand is forecastable if the underlying demand pattern is reasonably stable, but what if the pattern itself is subject tomajor shocks? The problemwith shocks is this: if they were forecastable, they would not be shocking! Themost obvious recent examples include the petrol shortages we have experienced at forecourts up and down the UK, the shortage of HGV drivers that contributed to that problem and to other supply issues, and even the lack of chicken on themenu at Nando's. Then there has been the Covid-19 pandemic.. It sent shockwaves through the demand patterns formany products, making themhighly unpredictable. This is an example of genuinely unforecastable demand. More recently, demand has been difficult to predict 24 | because of problems caused by severe shortages of HGV drivers. In the hospitality sector, this led to a slump in demand when restaurants, such as the some in theNando’s chain, had to close. In the fuel sector, it led to an extraordinary spike of demand, and long queues at petrol stations, induced by panic buying. Events like the pandemic have been described as ‘black swans’. The nature and timing of black swans is unpredictable and limits the potential value of forecasting approaches. In such situations, there is a high risk of retailers running out of stock completely and then overcompensating with excessive stock, creating significant wastage. What can be done? I believe that the problemof major disruptions calls for a different approach to forecasting, known as scenario planning. While the nature and timing of disruptions cannot be foreseen, the occurrence of some supply chain disruption (of unknown origin and timing) can be planned for. In a scenario planning exercise, managers imaginemajor causes of disruption that could happen in the future, such as a repeat of. They then think through measures that could be put in place to mitigate the effects of any sudden change in demand or supply (or both). For example, if an organisation relies on a single supplier, then they can introduce a second supplier, whowill receive regular orders and can respond to higher volumes if there are problems with the first supplier. After theonset of amajor disruption, forecastswill still need tobemade. It is natural for organisations to relyonhuman judgment in this situation, because statisticalmethods that extrapolate the recent history are likely tobeunreliable. As a ‘newnormal’ returns, statistical methods come into their own again, but may requirehuman adjustment. Oneof my colleagues, Robert Fildes, has conductedempirical researchwithmajor companies on this subject, currently being taken forwardbyAnnaSroginis. Robert’s researchhas revealed conditions underwhich judgment improves forecast accuracy andwhen it doesn’t. IMPACTFUL RESEARCH IN FORECASTING I believe that if we want to see a less wasteful future, then we need to take forecasting and planning much more seriously. It is a big responsibility and research centres can play their part. The Centre of Marketing Analytics and Forecasting at Lancaster is deeply engaged with practical applications of forecasting. This includes work that is affecting commercial software (Sven Crone, Ivan Svetunkov), open-source software (Ivan Svetunkov, Nicos Pavlidis, Rebecca Killick), and forecasting and planning practices in major organisations (Anna-Lena Sachs and myself). The centre is also active in writing pieces for practitioner audiences (Robert Fildes, Alisa Yusupova) so that our ideas can become more widely known. Industry engagement was part of Robert Fildes’ vision in establishing the centre (please see separate panel). We are committed to furthering this vision and look forward to many fruitful industry collaborations in the years to come. This is an appropriate time to pay tribute to Distinguished Professor Robert Fildes, who has recently retired after more than 30 years’ service at Lancaster. Robert was one of the founders of the International Institute for Forecasters and the International Journal of Forecasting. He also founded the Lancaster Centre for Forecasting in 1990 and set the direction for industry engagement right from the start. His work has had a tremendous impact on the theory and practice of forecasting. We are delighted that Robert has signalled his intention to continue working with the centre in his new role as an Emeritus Professor, and we can continue to benefit from his knowledge and wisdom. Professor John Boylan is the Director of the Centre for Marketing Analytics and Forecasting in LUMS, and a former Head of the Department of Management Science. He also serves as President of the International Society for Inventory Research. Professor Boylan and Professor Aris Syntetos (Cardiff University) coauthored the book Intermittent DemandForecasting: Context,Methods andApplications. It gives practitioners access to new forecastingmethods in away that does not require in-depth academic knowledge.
RkJQdWJsaXNoZXIy NTI5NzM=