Data Analytics for Managers IMBA IE Business Schoo! Risk Analysis Assignment TransAtlantic Airlines Caroline is looking at the figures on the sheet in front of her. Ryan, her supervisor who is in charge of yield management at TransAtlantic Aitlines has just called her into his office and told her that the booking policy TransAtlantic has been using for the last two years is urgently in need of revision. TransAtlantic Airlines TransAtlantic operates flights on several routes, mainly connecting London Heathrow to the East Coast of the United States. The sheet contains data on the London Heathrow — New York JFK route, a popular business route that is generating more revenue than any other route in TransAtlantic’s portfolio. Every day, 4 fights depart from London Heathrow for New York JFK using a Boeing 747 which can carry up to 500 passengers, 400 in economy class and 100 in business class. The (one-way) ticket price is £450 in economy class and £1450 in business class. In contrast with other airlines that use a large number of fares with different types of restrictions (e.g. advance purchase, maximum stay, Saturday/Sunday night included, etc.), TransAtlantic has always chosen a transparent pricing policy, charging just one business class fare and one economy fare. A problem with this route, as well as with other routes, is that often people who have made a reservation do not show up for their flight. TransAtlantic’s policy is that both economy and business class passengers who do not show up are refunded ‘This case was written by Bert De Reyck, Adjunct Associate Professor of Decision Sciences at the London Business School. It is intended to be used as a basis for class discussion, rather than to ilustrate either effective or ineffective handling of an administrative situation, All names of individuals and companies are fictitious. Copyright © 2001, London Business School, London, United KingdomSince the incremental cost of carrying an extra passenger on a plane is negligible, an empty seat represents a loss equal to the ticket price. In order to make sure that the plane does not fly with too many empty seats, TransAtlantic uses an overbooking policy. However, this overbooking policy sometimes leads to more passengers showing up than can actually get on the plane, resulting in passengers being ‘bumped’ to the next flight. Bumped customers are compensated by being offered a discount voucher, a meal allowance and a hotel room if the next flight leaves the next day. Moreover, bumping customers to the next flight creates a loss of goodwill, in the sense that some customers might switch to a competing airline for their future flights. Although this cost is difficult to estimate, TransAtlantic believes the cost of bumping a passenger to be £150 for an economy class passenger, and £300 for a business class passenger, including actual costs and goodwill loss. The London Heathrow — New York JFK route Caroline looks at the figures at the bottom of the sheet. The current policy of TransAtlantic Airlines is to accept up to 420 reservations for economy class (i.e. 5% overbooking) and up to 120 for business class (i.e. 20% overbooking). Overbooking in the business class is higher than overbooking in the economy class because the no-show phenomenon is more pronounced in business class. Caroline is not sure why Ryan chose exactly this route, perhaps because it is the most important route for TransAtlantic, perhaps because it is one of the routes that are plagued by a high number of no-shows. Whatever the reason, Ryan wants her to come up with a revised booking policy for this route, and if things work out well, he would revise the booking policies for the other routes in a similar way. On the sheet it also says that TransAtlantic has not kept detailed records about the number of no-shows on the Heathrow-JFK flight. Although the number of no-shows is highly variable from flight to fight, there is a consensus that the percentage of no-shows in economy class is somewhere between 3% and 8%, For business class passengers, the no-show percentage is expected to be between 15% and 30%. Most people at TransAtlantic estimate the economy no-shows to be 5% and business no-shows to be 20%. Well, Caroline thinks, the booking policy seems to make sense then!Caroline looks at the back of the sheet. It contains what seems like a spreadsheet with some calculations. She thinks it is probably the sheet that her predecessor used a couple of years ago to come up with the booking policy that is currently operational Economy Business Capacity and Cost Data Boeing 747-400 capacity 400 100 Fare £450 £1,450 Cost of bumping passenger £150 £300, No-show % 5% 20% Overbooking Policy Booking Level (accepted reservations) 420 120 Passenger Data ‘Actual passengers showing up 399 96 Empty Seats 1 4 Bumped Passengers 0 0 Cost and Revenues Lost Revenues £450 £5,800 ‘Overbooking Cost £0 £0 Total Cost £450 £5,800 Total Cost for both Classes £6,250 Table 1. Spreadsheet Background Information Airlines were pioneers in the use of Risk Analysis and Monte Carlo Simulation and still make intensive use of these tools, with American Airlines generally considered to be the leader. An important problem that airlines are facing is no-shows: passengers who made a reservation but do not show up for their flights. This results in empty seats and (huge) losses for the airline, considering that whether a passenger actually boards a plane or not, it hardly makes any difference in the ‘operating costs. This has led to the practice of fight overbooking, which means that more reservations are accepted on a flight than the actual capacity of the flight itself, This results in higher flight utilisation rates, at the expense of customers ‘sometimes being ‘bumped’ to the next fight if more passengers show up than can actually board the plane, which calls for some kind of compensation. American Airlines has estimated that the practice of overbooking fights alone adds $200 million per year to its bottom line!Assignment You are asked to investigate whether the current overbooking policy can be improved upon Create a report of less than 4 pages (excluding appendices) summarising your results, your interpretation of the results, and your conclusions and recommendations. You can include graphs and outputs in an appendix, but make sure to summarise and interpret the results in your report as well, Please e-mail the soft copy of the report as well as your excel sheets to [email protected] before 1159 PM (2359 hrs), January 27, 2023. 1. The Model Copy the file “TransAtlantic. xis" from your e-mail to your own computer. Start Excel, but first make sure that @Risk is not running, In Excel, open the file “TransAllantic.xls’. Using the fen the current overbooking template, build a model for the problem TransAtlantic faces. policy, calculate the total costs resulting from empty seats and bumped customers. Assume that the flights are fully booked up to the accepted number of reservations (booking level), and use the most likely scenario for each of the parameters in the model (i.e. develop a base case model) TIPS = Save your model regularly, and save a backup version of your report once in a while! — Close any other applications (e.g. Intemet Explorer), because they will reduce your PC's available memory and slow it down, especially when you reach Question 5; then ensure you have only one Excel workbook open after you've launched @Risk . Interpret your results, do not just report them. = The spreadsheet model "TransAtlantic.xis” has been pre-formatted, For instance, the cells where fares are entered are defined as currency cells (£). You can change the format of cells using FormatiCells. — Think about what the decisions are in your model (variables under your control), the Uncertainties (outside your control) and the objectives. | suggest using appropriate colours (green, red, blue) to highlight them. = Take "Capacity and Cost data’, “Overbooking policy” from Table 1 ai excel model that will calculate "Passenger data’, “Costs and Revenues.” Check if the puts and create a results of the mode! match Table 1= The Max(value 1, value 2) function might be helpful. It returns the maximum of the two values. The Round(value, 0) rounds the value to the nearest integer number. 2. Improving the overbooking policy BEFORE CONTINUING, CHECK WITH ME IF YOUR MODEL IS CORRECT Based on the base-case model you developed, would you advocate a new overbooking policy to TransAtlantic? What can you conclude from the results? TIPS = Don'tuse a simulation analysis just yet. Use the base-case model to determine a new booking policy based on the most likely scenario, — Donot use any excel functions. Try to find the best booking level by trial-and-error. 3. Scenario Analysis Save your model as “TransAtlantic SA". As mentioned in the case, TransAtlantic has not kept detailed records about the number of no-shows on the Heathrow-JFK flight. However, there is a consensus that the number of no-shows in economy class is somewhere between 3% and 8% of the total number of passengers booked. For business class passengers, the no-show percentage is expected to be between 15% and 30% Using your own suggested overbooking policy (which you previously found by trial- and-error), evaluate a pessimistic and optimistic scenario, and examine the impact on the total costs incurred. What can you conclude? What is the optimistic scenario? What is the pessimistic scenario? Can you give a range of costs that might be incurred? TIPS is not easy to determine what constitutes a pessimistic and optimistic — Sometimes, scenario. Think about this carefully, and think about why a particular case constitutes a pessimistic or optimistic view.4. Sensitivity Analysis ‘Again using your own suggested overbooking policy, examine the impact of a variation in the no-show percentage on the costs for both economy and business class more closely. You can do this using data tables and graphs. What can you conclude? TIPS - You can construct a one-way data table using the Data\Table command. Suppose you want to examine different values for the Economy no-show percentage. First, in an ‘emply part of the spreadsheet, make a list with different values for the no-show percentage (in a column). In the cell above and to the right of that column, enter "=B22" (or enter *=", click on cell B22 and press [enter)). Then, select the range consisting of the values and the column to the right of that, including the cells above. Select Data\Table, In Column input cell, refer to the cell where the data table values need to be substituted into the model. Press [OK], and the data table will be completed. — Based on a data table, a graph can be constructed as follows: Select Insert\Chart\Line. Press [Next] and specify the data range by selecting, with your mouse, the column of the data table containing the results (except the first value). Select Series in Columns. Select, the Series tab on top of the window, enter “Costs in Name, and in Category (X) axis labels, select (using your mouse), the first column of values (without the label on top). — You can also construct a two-way data table using the Data\Table command. Suppose you want to examine different values for Economy and Business no-show percentages simultaneously. First, in an empty part of the spreadsheet, make a column with a lst of different values of Economy no-show percentage; for instance, in cells G11:G16. Then, in the cell above and to the right of that column, make a row with a list with different values for Business no-show percentage; for instance, from H10:M10. Then, in cell G10 enter “=B22" (or enter “=", click on cell B22 and press [enter)). Then, select the whole range of cells (G10:M15) and select Data\Table. In Column input cell, refer to the cell where the Economy no-show percentages need to be substituted into the model. Finally, in Row input cel, refer to the cell where the Business no-show percentages need to be substituted into the model. Press [OK], and the data table will be completed, You may also refer to_https://support office. com/en-us/article/calculate-multiple-results-by-using-a- data-table-e9562487-6ca6-4413-ad12-77542a5ea50b5. Simulation Save and close your model. Again open “TransAllantic.xls" (containing the model and your recommended booking policy), and save it as example “TransAtlantic Sim.xls” (new work book). Evaluate the overbooking policy TransAllantic had originally been using (i.e. not your own) Using simulation analysis, taking into account the uncertainty in the percentage of no-shows in economy and business class. Recall that the percentage of no-shows in economy class is, somewhere between 3% and 8%, although most people at TransAtlantic estimate it to be at 5%, and that for business ciass passengers, the no-show percentage is expected to be between 15% and 30%, and is thought to be 20%. What distribution would you use? ‘What is the average (Lotal) cost of this overbooking policy? What does this mean? Why is this different to the £6,250 Caroline's processor estimated? What is the probable range of total cost (95% confidence)? What is the probability that the (total) cost will exceed £17,000? Now evaluate the overbooking policy you recommended in Section 2 (using the same criteria), ‘Also evaluate the (total) costs of a policy of not overbooking at all, a practice that is employed by some airlines to guarantee a seat to a customer who has reserved in advance. Is the policy you recommended better than the other policies? Why? Will your policy be the best on every individual fight? On most flights? TIPS — Before using the automated @Risk simulation tool, simulate the model step-by-step by pressing the [F9}-key. First, click on the little “dice symbol” that you will find in the center of the menu bar and under the two boxes that are labelled “iterations” and “simulation: This will enable one iteration at a time, each time you press the [F9]-key (ust for your understanding). — Run the automatic simulation, If the simulation is very slow, decrease the number of iterations in the “simulation” box, and close other windows that be open.= You can copy the model (the two columns with the economy and business model) side by side makin possible to perform a simulation with different overbooking policies at the same time. 6. Further improving the overbooking policy Try to further improve your recommended overbooking policy by simulating different altematives and selecting the policy with the best results, TIPS. — The RiskSimtable function enables you to compare different policies without having to run a separate simulation for each policy. Suppose you want to run multiple simulations with different economy class booking levels, contained in cells X1 through X10. In the cell containing the Economy class booking level, enter “=Risksimtable(X1:X10)", and in Simulation Settings, set the number of simulations (# Simulation) equal to 10 (and iteration }00 - so that your computer won't hang). If you now run a simulation, @Risk will simulate the mode! 10 times, each time with a different level of the Economy class booking level. The results can easily be compared in the @Risk results window. If you se several RiskSimtable functions simultaneously, @Risk will take the first values of each RiskSimtable function first, then the second values eto. Look at the means of each simulation and compare = You may need to find the optimal for economy and business class booking by running the model separately (by fixing business class at current booking policy levels of company, find the optimal booking level for economy class using risksim table, then do the opposite) 7. Improving the model If economy class is overbooked but there are still emply seats in business class, economy class passengers can be upgraded to business class leading to fewer bumped customers. Try to enhance your spreadsheet model in Excel to take this into account. Would you advise a different overbooking policy? (from the optimal booking policy you found in Question 6)?TIPS Using the [F9}+key, try to simulate a situation in which economy class passengers are bumped while there is still some space in business class. Then think about how to enhance the model, Insert a new row called ‘Transferred Passengers’ below row ‘Bumped Passengers’ “Downgrading” overbooked business class passengers to economy class is usually not a viable option. Think carefully about what is the effect on the cost of upgrading an economy passenger. ‘Once you build the model, you can use Risksim table function and run simulation to find the optimal booking policy.