Modelling and forecasting the dynamics of mobile devices market - - PowerPoint PPT Presentation

▶
modelling and forecasting the dynamics of mobile devices
SMART_READER_LITE
LIVE PREVIEW

Modelling and forecasting the dynamics of mobile devices market - - PowerPoint PPT Presentation

Apr 28, 2023 211 likes •669 views

Introduction Data Methodology Forecasting Relations Finale References Modelling and forecasting the dynamics of mobile devices market shares Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost ISMS 2018 14th June 2018

slide-1
SLIDE 1

Introduction Data Methodology Forecasting Relations Finale References

Modelling and forecasting the dynamics of mobile devices market shares

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost

ISMS 2018

14th June 2018

Marketing Analytics and Forecasting

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-2
SLIDE 2

Introduction Data Methodology Forecasting Relations Finale References

Introduction

Competition on the market of computer technologies is dense... The winner of technological competitions is often ‘who has the best platform strategy and the best ecosystem to back it up’ (Cusumano, 2010).

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-3
SLIDE 3

Introduction Data Methodology Forecasting Relations Finale References

Market structure

The market has several levels...

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-4
SLIDE 4

Introduction Data Methodology Forecasting Relations Finale References

Introduction

The wisest strategy is to create the ecosystem. But can it be distinguished from monopoly?

  • 1. Microsoft bundling its browser to its operating system

(Winkler, 2014);

  • 2. Google services as main on the mobile devices (Edelman,

2015);

  • 3. Investigating Google’s tactics on mobile devices market

(Kendall and Barr, 2015).

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-5
SLIDE 5

Introduction Data Methodology Forecasting Relations Finale References

Introduction

Rezitis (2010) examines whether it is the concentration in the market that causes the firms to mutually collude to enhance market power, or there are some other factors responsible for it. Claessens & Laeven (2004) observe that when the size of the firm increases, its share in market also increases and provides an

  • pportunity for that firm to earn higher profits.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-6
SLIDE 6

Introduction Data Methodology Forecasting Relations Finale References

Introduction

Analysing market shares helps in determining concentration on the market:

  • Herfindahl-Hirschman index (HHI) for average amount of

competition: HHI = k

j=1 s2 j

  • Coefficient of variation of market shares:

v = k

  • 1

k

k

j=1

  • sj − 1

k

2 sj is the market share of the j-th company on the whole market. k is the number of companies on the market.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-7
SLIDE 7

Introduction Data Methodology Forecasting Relations Finale References

Introduction

In addition:

  • Coefficient of segment concentration (Salihova, 2006).

◮ For each company: SCj =

m

i=1 |si,j−sj|

1+(m−2)sj

◮ For the whole market: SC = 1

k

k

j=1 SCj

where m is the number of segments on the market. si,j is the market share of jth company on the segment i.

  • SC = 0 - uniform distribution of market shares over all

segments,

  • SC = 1 - high concentration of one company on all the

segments.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-8
SLIDE 8

Introduction Data Methodology Forecasting Relations Finale References

Motivation

All these coefficients are static. But market is dynamic. If there is a connection between the segments over time, then this is probably an ecosystem. If we can forecast market shares, we can diagnose the expected situation on the market.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-9
SLIDE 9

Data

slide-10
SLIDE 10

Introduction Data Methodology Forecasting Relations Finale References

Data

Three segments in Europe: PCs, smartphones and tablets. Several platforms:

  • Windows,
  • Apple,
  • Android,
  • Other (Linux, Chrome OS, Symbian, etc).

Usage of platforms on different devices. Monthly shares of each platform from StatCounter (http://gs.statcounter.com) from 2012 to 2018.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-11
SLIDE 11

Introduction Data Methodology Forecasting Relations Finale References

Phones segment market shares

Time 2012 2013 2014 2015 2016 2017 2018 0.0 0.2 0.4 0.6

  • Windows

Apple Android Other

Data from http://gs.statcounter.com

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-12
SLIDE 12

Introduction Data Methodology Forecasting Relations Finale References

Tablets segment market shares

Time 2012 2013 2014 2015 2016 2017 0.0 0.2 0.4 0.6 0.8

  • Apple

Android Other

Data from http://gs.statcounter.com

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-13
SLIDE 13

Introduction Data Methodology Forecasting Relations Finale References

PCs segment market shares

Time 2012 2013 2014 2015 2016 2017 2018 0.0 0.2 0.4 0.6 0.8

  • Windows

Apple Other

Data from http://gs.statcounter.com

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-14
SLIDE 14

Introduction Data Methodology Forecasting Relations Finale References

Devices market shares

Time 2012 2013 2014 2015 2016 2017 0.1 0.2 0.3 0.4 0.5 0.6

  • Phone

Tablet PC

Based on the sales in millions USD from https://statista.com

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-15
SLIDE 15

Introduction Data Methodology Forecasting Relations Finale References

Shares for each OS in each segment

Combining everything, we end up with the following mess:

Time Shares 2012 2013 2014 2015 2016 2017 0.0 0.2 0.4 0.6 0.8 1.0 1.2 DWindows DApple DOther TApple TAndroid TOther PWindows PApple PAndroid POther

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-16
SLIDE 16

Introduction Data Methodology Forecasting Relations Finale References

Shares for each OS in each segment

Some platforms have died out over the years, The others have just appeared, but don’t have a big share (less than 1%). We removed those that don’t have large share at the end of series...

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-17
SLIDE 17

Introduction Data Methodology Forecasting Relations Finale References

Shares for each OS in each segment

Time Shares 2012 2013 2014 2015 2016 2017 0.0 0.2 0.4 0.6 0.8 1.0 1.2 DWindows DApple TApple TAndroid PApple PAndroid

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-18
SLIDE 18

Introduction Data Methodology Forecasting Relations Finale References

Shares for each OS in each segment

Time Shares 2012 2013 2014 2015 2016 2017 0.0 0.2 0.4 0.6 0.8 1.0 1.2 DWindows DApple TApple TAndroid PApple PAndroid

Observations:

  • Android phones are dominating.
  • Apple phones maintain the high share.
  • Windows PCs are loosing shares.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-19
SLIDE 19

Methodology

slide-20
SLIDE 20

Introduction Data Methodology Forecasting Relations Finale References

Methodology

Modelling shares should take several aspects into account (Terui, 2000):

  • Each share should be in (0, 1);
  • Shares should add up to one.

Terui (2000) formulates BVAR and models shares directly, making sure that the logical consistency is satisfied. Ribeiro Ramos (2003) uses VAR and BVAR models directly, ignoring the limitations.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-21
SLIDE 21

Introduction Data Methodology Forecasting Relations Finale References

Methodology

Agrawal and Schorling (1996) compare forecasting performance of Multinomial Logistic Regression (MNL) with Neural Networks. Fok and Franses (2001) use attraction model in order to obtain shares and acknowledge both limitations. They use regression in

  • rder to produce forecasts of shares.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-22
SLIDE 22

Introduction Data Methodology Forecasting Relations Finale References

Methodology

We use MNL to transform the data. Then we apply statistical model with additive errors. Finally we produce forecasts and return to the original scale. We use Vector Exponential Smoothing (VES from de Silva et al., 2010) for forecasting. We use VAR for the analysis of the connections.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-23
SLIDE 23

Analysis and forecasting

slide-24
SLIDE 24

Introduction Data Methodology Forecasting Relations Finale References

Shares for each OS in each segment

Transform the data using logit:

Shares

Time x[, 1] 2012 2013 2014 2015 2016 2017 −2 2 4

DWindows DApple TApple TAndroid PApple PAndroid

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-25
SLIDE 25

Introduction Data Methodology Forecasting Relations Finale References

Shares for each OS in each segment

Shares Time x[, 1] 2012 2013 2014 2015 2016 2017 −2 2 4

DWindows DApple TApple TAndroid PApple PAndroid

Time series have similar dynamics (correlated). Use VES, which captures that. We use local-trend model, produce forecasts and then transform to the original scale.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-26
SLIDE 26

Introduction Data Methodology Forecasting Relations Finale References

Overall dynamics

Time Shares 2012 2013 2014 2015 2016 2017 2018 2019 0.0 0.2 0.4 0.6 0.8 1.0 DWindows DApple TApple TAndroid PApple PAndroid Other

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-27
SLIDE 27

Introduction Data Methodology Forecasting Relations Finale References

Overall dynamics

Time Shares 2012 2013 2014 2015 2016 2017 2018 2019 0.0 0.2 0.4 0.6 0.8 1.0 DWindows DApple TApple TAndroid PApple PAndroid Other

In short:

  • The share of Windows for desktops is slowly decreasing;
  • The share of Apple for desktops is slowly increasing;
  • Shares of tablets are decreasing;
  • Share of Apple phones is expected to increase at the expense
  • f the share of Android phones.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-28
SLIDE 28

Introduction Data Methodology Forecasting Relations Finale References

Herfindahl–Hirschman Index

Time HHI 2012 2014 2016 2018 0.0 0.2 0.4

Normalised HHI tells us that this is moderately concentrated market. The concentration has been increasing lately.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-29
SLIDE 29

Introduction Data Methodology Forecasting Relations Finale References

Segment concentration coefficient

Time Segment concentration 2012 2013 2014 2015 2016 2017 2018 2019 0.0 0.2 0.4 0.6 0.8 1.0 Overall Windows Apple Android Other

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-30
SLIDE 30

Introduction Data Methodology Forecasting Relations Finale References

Segment concentration coefficient

Time Segment concentration 2012 2013 2014 2015 2016 2017 2018 2019 0.0 0.2 0.4 0.6 0.8 1.0 Overall Windows Apple Android Other

Segment concentration shows:

  • Microsoft is loosing position, because it looses on phones and

tablets segments;

  • Android is dominating, mainly because of phones;
  • Apple preserves its position;
  • Others are almost non-existent;
  • Overall, the market is moderately concentrated.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-31
SLIDE 31

Analysis of the relations

slide-32
SLIDE 32

Introduction Data Methodology Forecasting Relations Finale References

Analysis of the dynamics

Finally we fit VAR in order to see if there is relation between different segments. The optimal order is VAR(1) according to AIC. Then we can analyse Impulse Response Functions.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-33
SLIDE 33

Introduction Data Methodology Forecasting Relations Finale References

IRF, Desktops, Windows

DWindows −0.002 0.001 DApple −0.002 0.001 TApple −0.002 0.001 TAndroid −0.002 0.001 PApple −0.002 0.001 1 2 3 4 5 6 7 8 9 10 PAndroid −0.002 0.001 1 2 3 4 5 6 7 8 9 10

Orthogonal Impulse Response from DWindows 95 % Bootstrap CI, 100 runs

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-34
SLIDE 34

Introduction Data Methodology Forecasting Relations Finale References

IRF, Desktops, Apple

DWindows −0.001 0.002 DApple −0.001 0.002 TApple −0.001 0.002 TAndroid −0.001 0.002 PApple −0.001 0.002 1 2 3 4 5 6 7 8 9 10 PAndroid −0.001 0.002 1 2 3 4 5 6 7 8 9 10

Orthogonal Impulse Response from DApple 95 % Bootstrap CI, 100 runs

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-35
SLIDE 35

Introduction Data Methodology Forecasting Relations Finale References

IRF, Tablets, Apple

DWindows −0.002 0.001 DApple −0.002 0.001 TApple −0.002 0.001 TAndroid −0.002 0.001 PApple −0.002 0.001 1 2 3 4 5 6 7 8 9 10 PAndroid −0.002 0.001 1 2 3 4 5 6 7 8 9 10

Orthogonal Impulse Response from TApple 95 % Bootstrap CI, 100 runs

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-36
SLIDE 36

Introduction Data Methodology Forecasting Relations Finale References

IRF, Tablets, Android

DWindows −0.0015 0.0005 DApple −0.0015 0.0005 TApple −0.0015 0.0005 TAndroid −0.0015 0.0005 PApple −0.0015 0.0005 1 2 3 4 5 6 7 8 9 10 PAndroid −0.0015 0.0005 1 2 3 4 5 6 7 8 9 10

Orthogonal Impulse Response from TAndroid 95 % Bootstrap CI, 100 runs

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-37
SLIDE 37

Introduction Data Methodology Forecasting Relations Finale References

IRF, Phones, Apple

DWindows −0.004 0.002 DApple −0.004 0.002 TApple −0.004 0.002 TAndroid −0.004 0.002 PApple −0.004 0.002 1 2 3 4 5 6 7 8 9 10 PAndroid −0.004 0.002 1 2 3 4 5 6 7 8 9 10

Orthogonal Impulse Response from PApple 95 % Bootstrap CI, 100 runs

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-38
SLIDE 38

Introduction Data Methodology Forecasting Relations Finale References

IRF, Phones, Android

DWindows −0.001 0.002 0.005 DApple −0.001 0.002 0.005 TApple −0.001 0.002 0.005 TAndroid −0.001 0.002 0.005 PApple −0.001 0.002 0.005 1 2 3 4 5 6 7 8 9 10 PAndroid −0.001 0.002 0.005 1 2 3 4 5 6 7 8 9 10

Orthogonal Impulse Response from PAndroid 95 % Bootstrap CI, 100 runs

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-39
SLIDE 39

Introduction Data Methodology Forecasting Relations Finale References

IRF, Conclusions

Overall, there are connections in dynamics between platforms for Apple and Android devices. e.g. Apple tablets share ↑, the share of Apple phones ↑. Android tablets share ↑, the share of Android phones ↑. There are features of ecosystems for both. Phones and tablets segments are competitive, desktop is not.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-40
SLIDE 40

Conclusions

slide-41
SLIDE 41

CMAF Introduction Data Methodology Forecasting Relations Finale References

Conclusions

Analysing the platforms we find:

  • Segments of smartphones and tablets are relatively

competitive;

  • Apple dominates tablets;
  • Apple maintains the high share over several segments;
  • Apple has ecosystem, where shares on different segments are

interconnected;

  • Android dominates the phones segment;
  • Android has ecosystem for tablets and phones;

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-42
SLIDE 42

CMAF Introduction Data Methodology Forecasting Relations Finale References

Conclusions

And also:

  • Microsoft is monopoly on PCs segment;
  • But MS does not have ecosystem, so it looses overall over

time;

  • Overall market is moderately concentrated;
  • And the concentration has been increasing lately;
  • But the companies do not have equal shares in segments.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares

slide-43
SLIDE 43

Finale

Thank you for your attention!

Questions? Ivan Svetunkov i.svetunkov@lancaster.ac.uk

Marketing Analytics and Forecasting

slide-44
SLIDE 44

Introduction Data Methodology Forecasting Relations Finale References

Agrawal, D., Schorling, C., 1996. Market share forecasting: An empirical comparison of artificial neural networks and multinomial logit model. Journal of Retailing 72 (4), 383–407. Cusumano, M., 2010. The Evolution of Platform Thinking. Communications of the ACM, 53 (1), 32–34. de Silva, A., Hyndman, R. J., Snyder, R., dec 2010. The vector innovations structural time series framework. Statistical Modelling: An International Journal 10 (4), 353–374. Fok, D., Franses, P. H., 2001. Forecasting market shares from models for sales. International Journal of Forecasting 17 (1), 121–128. Ribeiro Ramos, F. F., 2003. Forecasts of market shares from VAR and BVAR models: A comparison of their accuracy. International Journal of Forecasting 19 (1), 95–110. Terui, N., apr 2000. Forecasting dynamic market share

  • relationships. Marketing Intelligence & Planning 18 (2), 67–77.

Ivan Svetunkov, Victoria Grigorieva, Yana Salihova and Florian Dost CMAF Modelling and forecasting the dynamics of mobile devices market shares