Forecasting and Logistics John H. Vande Vate Fall, 2002 1 1 Fundamental Rules • The farther in the future we must forecast, the worse the forecast • The longer we have to do something the cheaper it is to do it. • Balance these two – Long plans mean bad forecasts – Short plans mean high operational costs 2 2 Balancing Risk • News vendor problem • A single shot at a fashion market • Guess how much to order – If you order too much, you can only salvage the excess (perhaps at a loss) (s-c = net salvage value) – If you order too little, you lose the opportunity to sell (r = revenue) • Question: What value do you choose? 3 3 The Idea • Balance the risks • Look at the last item – What did it promise? – What risk did it pose? • If Promise is greater than the risk? • If the Risk is greater than the promise? 4 4 Measuring Risk and Return • Profit from the last item $p if the Outcome is greater,$0 otherwise • Expected Profit $p*Probability Outcome is greater than our choice • Risk posed by last item $r if the Outcome is smaller, $0 otherwise • Expected Risk $r*Probability Outcome is smaller than our choice Example: r = Cost – Salvage Value What if r < 0? What if Salvage Value > Cost? 5 5 Balancing Risk and Reward • Expected Profit $p*Probability Outcome is greater than our choice • Expected Risk $r*Probability Outcome is smaller than our choice • How are probabilities Related? 6 6 Risk & Reward Distribution 0.45 Prob. Outcome is smaller 0.4 Our choice 0.35 How are they related? 0.3 Prob. Outcome is larger 0.25 0.2 0.15 0.1 0.05 0 0 2 4 6 8 10 12 7 7 Balance • Expected Revenue $p*(1- Probability Outcome is smaller than our choice) • Expected Risk $r*Probability Outcome is smaller than our choice • Set these equal p*(1-P) = r*P p = (r+p)*P p/(r + p) = P = Probability Outcome is smaller than our choice 8 8 Making the Choice Distribution 0.45 0.4 Prob. Outcome is smaller Our choice 0.35 0.3 Cumulative Probability 0.25 0.2 p/(r+p) 0.15 0.1 0.05 0 0 2 4 6 8 10 12 9 9 How does this Apply? • Source in Asia – Cost is Low, Lead time is high (add 2 wks) • Source in Mexico – Cost is higher, lead time is lower • Place your first order far in advance in Asia. Asia has capacity • Place subsequent order in Mexico when you know more • What do you order from Asia? • What do you order from Mexico? 10 10 What Matters? • Ordering from Asia – Expected Demand from? – Variance in Demand from? • Ordering from Mexico – What you ordered from Asia – Expected Demand – Variance in Demand Anything Else? 11 11 What Else! How much you can learn from waiting! If the forecast is bad before and bad after you might as well order early If the forecast is bad before but good after, you gain by waiting. But, how to know…. 12 12 Example: Common Variance Order x(k) of product k Products differ only in mean demand Can’t order more than 5,000 from Asia Express x(k) = (k) + z* Probability Demand < x(k)? Normal (x(k) - (k))/) = Normal(z) So? Use News Vendor to find z 13 13 Only Means Matter Product Profit 1 2 3 4 5 Risk 110 110 110 110 110 25 25 25 25 25 Total Mean Std 1000 1000 1500 2500 1000 7000 200 200 200 200 200 P* z Order Q P(z*) Order 0.814815 0.89578 1179.156 0.02275 600 0.814815 0.89578 1179.156 600 0.814815 0.89578 1679.156 1100 0.814815 0.89578 2679.156 2100 0.814815 0.89578 1179.156 600 Total Order 7895.78 5000 Max Order 5000 Target z* -2 x(k) = (k) + z* S x(k) = S ((k) + z*) = 5,000 n*z* = 5,000 - S (k) z = (5,000 - S (k))/(n*) Then find x(k) from z. 14 14 Sourcing Decisions • What matters: – – – – – – Piece price Quality Freight Packaging Pipeline inventory Etc • And Now: – Leadtime impacts on forecast accuracy and – Forecast accuracy impacts on supply chain costs! 15 15 Where Else? • Where else do these ideas apply? • What if we are the manufacturer? • The longer it takes us to make a product, the farther forward we must forecast • What can we do to reduce this? 16 16
© Copyright 2026 Paperzz