Operations Management
Question 1
Use the dataset 'Netflix.xls' downloadable from Moodle. The dataset contains publicly available information from Netflix quarterly earningsreports, particularly information on Netflix online subscribers. Note that there is a big change in the dataset in Sept. 2011, when Netflix forced customers to decide whether to subscribe to online and/or offline services (these services were bundled before), and thereby significantly increased the price of the bundled service.
a) Go to the 'Forecast Main Model' tab in the spreadsheet. This tab provides a double exponential smoothing forecast for the number of (domestic) subscribers to the Netflix online service. Based on this model, provide a point forecast for the December 2013 quarter, together with a prediction interval around the point forecast, such that you are 95% sure that the true number of subscribers falls into that interval. (note that you should ignore the forecast errors made Sept. 2011 - March 2012 for this analysis, since these numbers are highly influenced by the change in bundling policy)
b) Based on this model, also provide a point forecast for the March 2014 quarter, together with a similar 95% prediction interval. Compare the two prediction intervals.Hint: For a double exponential smoothing model, the n-period ahead forecast made in period t is given by Forecastt+n = Levelt + n*Trendt
c) Marketing expenses for each quarter are known in advance, and could be used to improve forecasting performance with respect to the number of subscribers. Do you believe that this approach would improve forecast accuracy? Justify your answer.
d) Go to the 'Forecast Alt. Model' tab in the spreadsheet. This tab provides you with an alternative forecasting model - a so called multiplicative trend model, which is sometimes used for companies that are growing rapidly. In this model, the trend is not modeled as an additive component (as in the previous double exponential smoothing model), but as a growth rate. Compare this model to the double exponential smoothing model in the Main Model tab. Which model do you prefer? Justify your answer!
e) Netflix business model is often seen as representing significant economies of scale. In the data you have, can you find evidence for economies of scale? Justify your answer! Do you believe that Netflix's move to separate the online from the offline business will enable it to achieve more economies of scale?
Question 2
You work as a consultant for a local chain of coffee shops. The chain has five cafes in town. One of their signatures is that they serve excellent, freshly baked croissants each day. Currently, they bake about 500 croissants each morning on Mon-Fri (900 on Sat-Sun), and throw away many of them at the end of the day. The dataset contains demand data for the past 8 weeks. Your task is to figure out how many croissants the company should bake each morning.
a) The croissants are baked at a central bakery, and then shipped to the individual cafés. The owner of the chain says she wants a service level of 99.9% at each cafe. How many croissants should be baked each morning to achieve this service level? (Round up all order quantities to the closest integer! You can assume that demand is stable - no level changes or trends; you can assume that the order quantity needs to be the same on Mon-Fri, as well as Sat-Sun - feel free to test this assumption, though.)
b) Your primary improvement proposal is to make individual cafes call in before they run out of croissants, so that additional croissants can be transferred to them from another café before they run out of stock. What would be the implications of this policy for the amount of croissants baked? By how much (in %) would your safety stock increase or decrease?
c) You know that croissants sell for $3, and cost about $0.80 to make. How can you use this information to give further advice to the company?
Question 3
It is a common practice for airlines to oversell their flights, i.e. to sell more tickets for a flight than there are spaces available in an airplane. If too many passengers arrive for a flight, some of them will be 'bumped', either voluntarily (and receive a compensation) or involuntarily (if no volunteers step up). The Table below (taken from FORBES magazine in 2008) shows how prevalent this phenomenon is within the US:
|
Airline
|
Voluntary Bumps
|
Involuntary Bumps
|
Total Bumps
|
Total Passengers
|
|
Jetblue
|
58
|
22
|
80
|
21,900,554
|
|
Hawaian
|
317
|
54
|
371
|
7,856,711
|
|
Frontier
|
4,436
|
983
|
5,419
|
10,497,522
|
|
Alaska
|
8,128
|
983
|
9,111
|
15,546,453
|
|
American Airlines
|
56,649
|
5,568
|
62,217
|
82,247,704
|
|
Southwest
|
73,403
|
10,362
|
83,765
|
102,045,003
|
|
American Eagle
|
7,103
|
2,184
|
9,287
|
8,940,543
|
|
Continental
|
37,825
|
5,671
|
43,496
|
40,283,669
|
|
Delta
|
62,243
|
10,403
|
72,646
|
65,735,090
|
|
Northwest
|
48,473
|
3,027
|
51,500
|
42,519,162
|
|
United
|
92,624
|
6,812
|
99,436
|
57,568,962
|
|
AirTran
|
41,877
|
834
|
42,711
|
24,619,120
|
|
US Airways
|
85,001
|
7,205
|
92,206
|
53,145,064
|
|
Pinnacle
|
6,572
|
540
|
7,112
|
3,160,628
|
|
SkyWest
|
34,155
|
2,090
|
36,245
|
15,572,248
|
|
Mesa
|
25,048
|
1,355
|
26,403
|
9,947,777
|
|
Comair
|
13,461
|
1,909
|
15,370
|
5,599,468
|
|
Atlantic Southeast
|
22,982
|
3,610
|
26,592
|
9,290,037
|
a) Comment on whether you believe that overselling seats for a flight is a good business practice or not.
b) Calculate a service level for each of the above airlines. What type of service level (I or II) does this correspond to?
c) The European Union has introduced a mandatory compensation of up to €650,- for a passenger that is being bumped (depending on the distance of the flight). What is the impact of this regulation for airlines and passengers? What alternative business practices could make overbooking less prevalent, and possibly increase customer service?