Question

Tommy has recently graduated from SUSS and has joined a well-known retailer that operates 3 department stores in Singapore. His job function is that of a business analyst. It has been well-reported that retail business in Singapore is on the decline and his employer would like to determine if the forecast for the next few years will be equally bad. Tommy has been tasked to perform the analysis and his output will provide insights in the company's hiring and expansion plan in Singapore.

The first thing Tommy did was to download the Retail Sales Index data from the Department of Statistics Singapore website. He specifically extracted the data for "Department Stores" from Q1 2008 to Q4 2018. (2017 is set as the base year with Index = 100). Refer to data below:

Year/Quarter Retail Sales Index (Department Stores)

2008 1Q 92.4

2008 2Q 93

2008 3Q 84.9

2008 4Q 105.3

2009 1Q 87.1

2009 2Q 88.6

2009 3Q 83.2

2009 4Q 103.6

2010 1Q 94.3

2010 2Q 93.9

2010 3Q 90.7

2010 4Q 109.3

2011 1Q 99.6

2011 2Q 99

2011 3Q 95.8

2011 4Q 119.3

2012 1Q 104.9

2012 2Q 99.9

2012 3Q 95

2012 4Q 117

2013 1Q 106.1

2013 2Q 100.9

2013 3Q 97.5

2013 4Q 118.6

2014 1Q 107.1

2014 2Q 98.7

2014 3Q 95.7

2014 4Q 117.6

2015 1Q 108.2

2015 2Q 101.1

2015 3Q 98.8

2015 4Q 114.4

2016 1Q 103.8

2016 2Q 93.9

2016 3Q 93.2

2016 4Q 112.9

2017 1Q 100.3

2017 2Q 93

2017 3Q 94.6

2017 4Q 112.1

2018 1Q 102.7

2018 2Q 92.3

2018 3Q 93.6

2018 4Q 112.8

(a) Use the techniques of time series modelling and with justifications, discuss the considerations on how Tommy would forecast the quarterly Retail Sales Index in 2019 and 2020.

(b) Upon confirmation of the forecast technique to undertake in Part 3(a), determine the quarterly RSI forecast in 2019 and 2020. You are required to show the essential steps in deriving the forecast and comment on any limitations of your technique.

Answer 1

a)

Tommy can forecast the retail sales index by using the exponential smoothing forecasting technique.

Since data is more dependent on the previous index values so exponential smoothing technique is the best for it .

b)

The exponential smoothing technique is applied in the R software and the forecast are drawn.

It works on the giving more weight to most recent data and less weight to past values.

If the most past values can effect the current forecast then it is partly helpful otherwise it is best to forecast.