Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.
1
Chihiro Watanabe
Platform Ecosystems Impact on GDP
Research Professor, University of Jyvaskyla, Finland
Research Scholar...
Fig. 1. Growth Rate, Competitiveness and Happiness/Welfare in ICT Advanced Countries (2013).
1. New Stream of Innovation
-...
Captured
GDP
Traditional
ICT
Un-captured
GDP
(2) Co-evolution of 3 Mega-trends – Shift from Traditional to New Co-evolutio...
4
The 21st-century economy
How to measure prosperity
GDP is a bad gauge of material well-being.
Time for a fresh approach
...
5
Fig. 3. Two-faced Nature of ICT.
T
T
Y
T
T
Y
X
X
Y
X
X
Y
Y
Y
TXFY























 ),(
Y...
2) Bi-polarization of Digital Economy -Fatality of Logistic Growth
Logistic growth function as a function of time t can be...
7
Fig. 5. Transformative Role of Co-evolutionary Acclimatization.
.
that harnesses the vigor of counterparts
enables both ...
8
IGE: ICT
growing
economies
Fig. 6. Scheme of Optimal Dynamism for Balancing Captured and Un-captured GDP.
4) Optimal Bal...
2. Measurement of Un-captured GDP
2.1 Framework of the Analysis
(1) Basic Understanding
Un-captured GDP can be traced from...
10
Internet
Online Intermediaries (OI)
Core player
Provide platforms for the exchange of goods, services or information ov...
11
2) Consumer’s Preferences Shift
(i) Measurement of Elasticity of Utility to Consumption
Un-captured GDP is non-reflecti...
1.76
0.49
1.27
0.30
0.23
0.37
0.68
0.15
0.40
0.10 0.10
0.47 0.48
0.53
1.05
0.55 0.57 0.58
Finland Singapore USA UK Germany...
CapturedGDPUn-capturedGDP
Fig. 10. Trends in Elasticity of Utility to Consumption in 6 Countries (1994-2013).
(iii) Trend ...
14
(3) Factual Observation
Captured
GDP
Traditional
ICT
Internet
Un-captured
GDP
Advancement of ICT
Paradigm
change
GDPatc...
15
(4) Measurement of the Magnitude of Un-captured GDP
1) Consumption Function
Level of consumption
C = a’ + b’ Y
H = a + ...
16
2.2 Empirical Result
Table 2 Governing Factors of Household Consumption in Finland and Singapore (1994-2013)
19158.1998...
0.95 H
0.87 S
0.83 L
0.43 H
0.35 S
0.29 L
(1) Magnitude of Un-captured GDP – Un-captured GDP Ratio
Fig. 13. Trend in the I...
18
Finland
Singapore
Captured GDP
505
302
269
407
74
105
GDP at current prices (US$ billions)
GDP
1994 2000 2005 2010 2013...
19
600
500
400
300
200
100
0
Singapore
302
407
Captured GDP74
1994 2000 2005 2010 2013
Fig. 15. Comparison of Captured and...
Finland (1996 – 2013)
1996
2013
Internet productivity of ICT
Singapore (1996 – 2013)
19962013
Fig. 16. Contrast of the Shi...
21
(2) Correspondence to People’s Preferences Shift
Economic Functionality
Fig. 17. Shift from Economic Functionality to S...
1 FIN 2 SGP 3 SWE 4 NLD 5 NOR 6 CHE 7 USA 9 UK 12 DEU 13 DNK 15 ISR 16 JPN
Gaining profits
from abroad
GNI/GDPratio
Fig. 1...
23
(4) Emergence of Disruptive Business Model
Fig. 19. Consumer Surplus of Music and Audio
- visual Services (Person per m...
24
IS LM
Principle of effective
demand
Consumption function
(a, b: coefficient)
Investment function
( r : interest rate)
M...
25
(6) Optimal Dynamism Harnessing the Vigor of ICT Growing Economies
Co-evolutionary Acclimatization
Past experiences Cur...
National level
Elucidate
Trap in ICT advancement
Two-faced nature of ICT
Bi-polarization
Conceptualize
Operationalize
Busi...
27
0. Overview
1. New Stream of Innovation
-Spinoff to Un-captured GDP-driven Co-evolutionary 3 Mega-trends
2. Uber’s Ride...
28
2. Uber’s Ridesharing Revolution
2.1 ICT-driven Disruptive Business Model
(1) Uber: Hero of Spin-off
Fig. 22. Scheme of...
Fig. 23. Uber’s Astounding Rise in Trips and Continuous Decline in Prices in NYC.
$0
$200
$400
$600
$800
$1,000
$1,200
$1,...
Fig. 24. Dynamism of ICT-driven Disruptive Business Model (IDBM).
30
(2) Dynamism of ICT-driven Disruptive Business Model
...
31
2.2 Consequence of ICT-driven Disruptive Business Model
- Uber’s Expansion and Battles
(1) Rapid Expansion
Fig. 25. Ube...
32
2.3 Specific Features of ICT
- Sources of Uber’s Expansion and Battles
(1) Specific Features
Since Uber is seen as the ...
33
(2) ICT’s Indigenous Functions Derived from its Specific Features
1) Self-propagating Nature
Diffusion trajectory of in...
Comparison of Spinoff State in Finland vs Singapore and Taxi vs Uber
Simple logistic growth Carrying capacity enhance Logi...
2) Bi-polarization Fatality of Logistic Growth
Logistic growth function as a function of time t can be depicted as follows...
36Fig. 28. Transformative Role of Co-evolutionary Acclimatization
.
that harnesses the vigor of counterparts
enables both ...
37
IGE: ICT
growing
economies
Fig. 29. Scheme of Optimal Dynamism for Balancing Captured and Un-captured GDP.
4) Optimal B...
38
2.4 Structural Sources of Legal Battles
Given that Uber’s system success depends on the advancement of ICT, its conside...
39
0.00
10.00
20.00
30.00
40.00
50.00
60.00
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28
Tri...
40
Fig. 31. Correlation between Centralization of Wage Setting and Union and CBA Density
in 19 Countries in the Late 1990s...
Functional Development level
Rapid growth Steady growth
Elastic nation/city
Non-elastic nation/city
Time
Level of
Function...
42
1.Contrast between countries with and without legal battles can be attributed to with or without CCSD
(Consolidated Cha...
(4) Co-evolutionary Acclimatization
-ICT-driven Disruptive Business Model with Consolidated Challenge to Social Demand
Hos...
44
2.5 Lessons from Success Model
2.5.1 IDBM with CCSD
(1) CCSD in Success
Table 5 Structure in CCSD in Success Countries/...
45
NTUC this Week, 26 Jan. 2007 (NTUC: National Trades Union Congress).
(2) Lessons from Tripartism
2.5.2 Significance of Shift to IDBM with CCSD
(i) Nowadays, a key factor in obtaining business opportunities is the abilit...
47
3. Trust-based Education toward Digitally-rich Learning
Environments
3.1 Co-evolution with Trust as a Consequence of ID...
Fig. 37. Scheme of Spinoff Dynamism.
Uber
Ride sharing
evolution
DILE
(Digitally-rich Innovative
Learning Environments)
48...
49
3.2 Co-evolution between ICT, Trust and Higher Education
(1) Comparison in 20 Countries
Fig. 38-2. Higher Education (20...
50
(2) ICT-driven Education Development
Fig. 39. ICT-driven Education Development in 120 Countries (2013).
at
be
N
Y
N
Y
a...
51
(3) Correlations
Table 6 Co-evolution and Disengagement between ICT, Educational Level, and Trust
in Teachers
Fig. 40-1...
52
3.3 Structural Sources of Trust Decrease as ICT Advance
(1) Bi-polarization
Fatality
(2) ICT Elasticity
to Trust
(3) Co...
53
(4) Blended Learning and Teacher’s Resistance - Transition to ICT Advanced Countries
Fig. 41. State of Hybrid Developme...
54
Fig. 42. Stages of ICT Integration in Education.
(5) Stages of ICT Integration in Education
1. ICT has been integrated ...
55
)
1
1(
FD
aY
dZ
dY

Overdrawing past
information
3.4 Co-evolutionary Acclimatization toward Digitally-rich
Learning ...
56
Current Past Future
National Economy
Uber Sleeping resources
Digital learning Time
Music/Game Memory Dream
(2) Transfor...
57Fig. 45. Optimal Dynamism Harnessing the Vigor of Time, Memory, Dream.
ICT advance T (Z)
)(
dY
dZ
dY
dT
(3) Optimal Dyna...
58
4. Optimization through Commodification of Experiences
4.1 Commodification of Experiences
(1) Transfer the Anger into a...
59
(2) From Invisible Hand of God to Voiceless Voice of Consumers
Fig. 47. Scheme in Conceptualizing Invisible Voice of Co...
60
Fig. 48. Scheme of Facial Temperature Feedback Hypothesis.
(3) Facial Temperature Feedback Hypothesis
Confronting
unfor...
(1) Methodology
Monitor the consumers’ facial temperature by the thermography: novel
psychophisiologal measuring technique...
Event wagon on which sweetened
ban is displayed (15 February)
Installation of PC for data recording
(15 February)
Hanging ...
2) Finland
Cosmetics corner of Sokos Supermarket (7 March 2012)
Cosmetics corner Thermography
connected to PC
Shoppers acc...
31.0
31.5
32.0
32.5
33.0
33.5
34.0
1 2 3 4 5 6 7
Accessed to the
event corner
Perceived
Recognized
Decided
to purchase
31....
2/19 15:00-16:00
2/19 16:00-17:00
2/19 17:00-18:00
2/20 15:00-16:00
2/20 16:00-17:00
2/20 17:00-18:00
2/20 18:00-19:00
0
2...
Roman holiday Sputnik No. 1 Tokyo Olympic Game Beatles Apollo
(1953 ) (1957) (1964) (1966) (1969)
4.3 Commodification of E...
69
(2) Sublimation of 20s Experiences into Supra-functionality beyond Economic Value
Social value
Cultural value
Aspiratio...
(3) Platform for Commodification of Experiences
for Innovation-Consumption Co-emergence
Utmost gratification of consumptio...
69
5. Harness the Vigor of Memory and Dream
5.1 Digital Music as a Platform for Retrieving Music Information
– Bridging Di...
Adoption of Digital music
Enable customers to enjoy music
ubiquitously by own initiatives
Music information retrieval (MIR...
71
(3) Empirical Analyses
On the basis of the empirical analysis of the development trajectory of the music industry in th...
72
5.2 Social Gaming as a Platform for Harnessing the Vigor
of Dream
6. Conclusion
75
1. Spinoff to Un-captured GDP-driven Co-evolution of 3 Mega-trends
(1) Advancement of ICT incorporates a ...
6. Conclusion (2)
76
2. ICT-driven Disruptive Business Model - Uber’s Ridesharing Revolution
(1) Under such circumstances,...
6. Conclusion (3)
2.-2 Consolidated Challenge to Social Demand for Resilient Platforms
(1) While Uber has expanded rapidly...
6. Conclusion (4)
78
3. Trust-based Education toward Digitally-rich Learning Environments
(1) Lessons of IDBM with CCSD su...
6. Conclusion (5)
79
4. Optimization through Coomodification of Experiences
(1) Shift of consumers preferences to supra-fu...
78
References
1. C. Watanabe, K. Naveed, P. Neittaanmaki and B. Fox, “Consolidated Challenge to Social Demand for
Resilien...
Appendix I. Measurement of Un-captured GDP
81
Un-captured GDP ratio
(1) Spin-off State
Simple logistic growth Carrying capacity enhance Logistic growth within a dynamic...
81
Institutional factors Finland Singapore References
Small income disparity
Inequality (GINI index: 2010)
Capacity for in...
82
(3) Strategic Actions for Spin-off
Fig. A1.1. Strategic Actions for Spin-off.
By transferring strong engine obtained by...
83
Appendix II. Uber’s Ride-sharing Revolution
0
10
20
30
40
50
60
Jun-13
Aug-13
Oct-13
Dec-13
Feb-14
Apr-14
Jun-14
Aug-14
Oct-14
Dec-14
Feb-15
Apr-15
Jun-15
Aug-15
AII....
AII.2 Institutional Enablers Creating Platform Ecosystems
AII.2.1 Preference Shift to Sharing Economy
1. In line with peop...
PU decrease
rate
Contribution by
Period
SP increase rate UT increase rate Miscellaneous
-2.58 -0.911 x 1.34 = -1.22 -0.173...
Fig. AII.4. Virtuous Cycle between Uber’s Trips and Its Prices (Jun. 2013– Sep. 2015).
Figures in parenthesis indicate t-s...
$0
$200
$400
$600
$800
$1,000
$1,200
$1,400
$1,600
Jan-04
May-04
Sep-04
Jan-05
May-05
Sep-05
Jan-06
May-06
Sep-06
Jan-07
M...
Fig. AII.7. Scheme of the Measurement of the Emergence of Uncaptured GDP in case of Uber in NYC.
$0
$200
$400
$600
$800
$1...
(3) Correlation between Uber’s Dependency and Medallion Prices
Fig. AII.9. Virtuous Cycle between Uber’s Dependency and Me...
TT
TUTT
A
UT
UPTP
P



13.5
14.0
14.5
15.0
15.5
16.0
16.5
17.0
17.5
18.0
Jun-13
July-13
Aug-13
Sep-13
Oct-13
Nov-13
D...
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Jun-13
July-13
Aug-13
Sep-13
Oct-13
Nov-13
Dec-13
Jan-14
Feb-14
Mar-14
Apr-14
May-14...
AII.3.3 Spin-off to Sharing Economy
(1) Uber’s Self-propagating Function
93
Table AII.3 Estimates of Taxis’ and Uber’s Dev...
AII.4 Conclusion
AII.4.1 Secret of the Success of Uber’s System
In light of the disruptive digital-technology-driven busin...
AII.4.3 Implications of Uncaptured GDP
The emergence of un-captured GDP in case of Uber can be attributed to
(i) People’s ...
Source: NYC Taxi and Limousine Commission.
0
250
500
750
1,000
1,250
1,500
Jan-04
Oct-04
July-05
Apr-06
Jan-07
Oct-07
July...
S1.2 ICT Prices Trajectory and Two-faced Nature
(1) Modified Bi-logistic Growth
Ia
i
Ja
j
I ij
eb
N
eb
N
p



 
11
ji...
98


































...
99
0)1(
0)1(
)1(
1
),1(
1






















 
IJ
Ia
N
p
Ia
p
I
I
p
IJ
Ja
N
p
Ja
...
$0
$200
$400
$600
$800
$1,000
$1,200
$1,400
$1,600
Jan-04
May-04
Sep-04
Jan-05
May-05
Sep-05
Jan-06
May-06
Sep-06
Jan-07
M...
Supplement 2 . Correlation between Medallion Prices and Taxi/Uber Prices
Figures in parenthesis indicate t-statistics: all...
102
0.00
10.00
20.00
30.00
40.00
50.00
60.00
1 2 3 4 5 6 7 8 9 10111213141516171819202122232425262728
UT
1
3.12 2.75
3.37 ...
13.5
14.0
14.5
15.0
15.5
16.0
16.5
17.0
17.5
18.0
Jun-13
Aug-13
Oct-13
Dec-13
Feb-14
Apr-14
Jun-14
Aug-14
Oct-14
Dec-14
Fe...
-1.0
-0.5
0.0
0.5
1.0
1.5
2.0
2.5
3.0
Jun-13
July-13
Aug-13
Sep-13
Oct-13
Nov-13
Dec-13
Jan-14
Feb-14
Mar-14
Apr-14
May-14...
S3.3 Effects of Uber’s Development Trajectory Estimate
105
Table S5 Estimates of Taxi and Uber’s Development Trajectories ...
106
Fig. S12. Correlation between Centralization of Wage Setting and Union and CBA Density
in 19 Countries in the Late 199...
107
1 Success
(1) Singapore [Legality is pending but operating actively]
1. Taxi drivers and passengers in Singapore are g...
108
(2) Tokyo [Seems illegal but operating]
1. Uber has had tremendous difficulties in making inroads into the Japanese ma...
109
1. Black taxis have been the kings of the British capital's roads for over a century but now they are
battling a high-...
110
(4) USA [Generally Positive]
Uber is operating in 75% of US locations although banned in Nevada and Oregon, and there ...
111
(5) Saudi Arabia
1. Saudi Arabia's discriminatory automotive policies against women have allowed Uber to achieve
great...
112
(6) Russia [No ban, but difficult to offer service]
1. Regulations in Russia are comparatively simple in comparison to...
113
(7) Canada [Changing to Support]
1. Uber drivers in Canada are required to register, collect and remit HST/GST from th...
114
(8) Philippines [Developed Nationwide Regulations Making Legal]
1. The Philippines became the first country to develop...
115
(9) China [Generally positive]
Although there was some raids and fines against Uber in some of locations, there is no ...
116
(1) France [Partial ban as illegal]
1. France government initially started to suppress the service with their policy a...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Fin...
Upcoming SlideShare
Loading in …5
×

Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Finland's Future

585 views

Published on

Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Finland's Future

Published in: Economy & Finance
  • ★★ How Long Does She Want You to Last? ★★ A recent study proved that the average man lasts just 2-5 minutes in bed (during intercourse). The study also showed that many women need at least 7-10 minutes of intercourse to reach "The Big O" - and, worse still... 30% of women never get there during intercourse. Clearly, most men are NOT fulfilling there women's needs in bed. Now, as I've said many times - how long you can last is no guarantee of being a GREAT LOVER. But, not being able to last 20, 30 minutes or more, is definitely a sign that you're not going to "set your woman's world on fire" between the sheets. Question is: "What can you do to last longer?" Well, one of the best recommendations I can give you today is to read THIS report. In it, you'll discover a detailed guide to an Ancient Taoist Thrusting Technique that can help any man to last much longer in bed. I can vouch 100% for the technique because my husband has been using it for years :) Here's the link to the report ★★★ https://tinyurl.com/rockhardxx
       Reply 
    Are you sure you want to  Yes  No
    Your message goes here
  • Be the first to like this

Chihiro Watanabe: Platform Ecosystems Impact on GDP - Increasing Dependency on Un-captured GDP and Its Consequences to Finland's Future

  1. 1. 1 Chihiro Watanabe Platform Ecosystems Impact on GDP Research Professor, University of Jyvaskyla, Finland Research Scholar, International Institute for Applied Systems Analysis (IIASA) Professor Emeritus, Tokyo Institute of Technology watanabe.c.pqr@gmail.com watanabe.foxcj@rhythm.ocn.ne.jp Watanabe Foresight 20161123 Foresight Meeting Helsinki 23 November 2016 WP 7 Platform Ecosystems Impact on GDP Increasing Dependency on Un-captured GDP and Its Consequences to Finland Future
  2. 2. Fig. 1. Growth Rate, Competitiveness and Happiness/Welfare in ICT Advanced Countries (2013). 1. New Stream of Innovation - Spinoff to Un-captured GDP driven Co-evolutionary 3 Mega-trends (1) Dilemma of World ICT Leaders Figures in red and blue indicate top and lowest level in 12 countries. Sources: IMF, World Economic Forum (WEF), The Earth Institute, ILO, WHO. 2
  3. 3. Captured GDP Traditional ICT Un-captured GDP (2) Co-evolution of 3 Mega-trends – Shift from Traditional to New Co-evolution Advancement of ICT Paradigm change 1. Internet has changed the computer- initiated ICT world significantly. 2. Internet provides utility and happiness to people but cannot be captured by GDP. 5. Consequently, current ICT driven economy depends on shifting two co- evolutional mega-trends: (1) Traditional co-evolution between ICT, captured GDP and economic functionality, (Singapore) (2) New co-evolution between Internet, un-captured GDP and supra-functionality (Finland). 6. While Finland has shifted to new co-evolution, Singapore has maintained traditional growth – oriented co-evolution. Fig. 2. Co-evolution between the Internet, Un-captured GDP and Supra-functionality. 3. Un-captured GDP has become the major source of consumers utility . 4. This corresponds to people’s preferences shift and induces further Internet advancement. Singapore Finland Internet 3 IoT
  4. 4. 4 The 21st-century economy How to measure prosperity GDP is a bad gauge of material well-being. Time for a fresh approach April 30, 2016 第1章 ICTによるイノベーションと経済成長 第4節 経済社会に対するICTの多面的な貢献 2.ICT化による経済社会の非貨幣的側面の変化 White Paper on Telecommunications (Ministry of Internal Affairs and Communications) August 1, 2016 Beyond GDP IoT, Big data, AI New values that network and data create
  5. 5. 5 Fig. 3. Two-faced Nature of ICT. T T Y T T Y X X Y X X Y Y Y TXFY                         ),( Y R p p Y R T Y Y T    decrease Y Y decline T Y decreasepT    Y: GDP, X: Traditional production factors (Labor, Capital), T: Technology stock ( ) , R: R&D investment (Marginal productivity) (Relative prices) RT  Tyler Cowen Professor of economics at George Mason University Tp T Internet dependency ICT’s contribution to growth (3) Un-captured GDP 1) Trap in ICT Advancement – Two-faced Nature of ICT Prices of ICTprices ICT advancement Trend in ICT Prices Prices increase by new functionality Prices decline due to freebies, easy copying, standardization Growth stagnation Un-captured GDP Decline in marginal productivity of technology. Increase in ICT stock Search engine Social networks Online advertising Socio-cultural value Internet provides new utility and happiness but cannot be captured by GDP data (Un-captured GDP)
  6. 6. 2) Bi-polarization of Digital Economy -Fatality of Logistic Growth Logistic growth function as a function of time t can be depicted as follows: This function can be developed to the following bi-polarization function: at be N Y N Y aY dt dY    1 )1( Inflection point Fig. 4. Bi-polarization Fatal to Logistic Growth Function. 6 Excessive increase changes to a vicious cycle.
  7. 7. 7 Fig. 5. Transformative Role of Co-evolutionary Acclimatization. . that harnesses the vigor of counterparts enables both economies maintain sustainable growth. While the advancement of ICT contributes to enhancing its prices by increasing new functionality development, dramatic advancement of the Internet tends to decrease ICT prices due to freebies, easy copying and mass standardization, among other things. ICT advanced economies suffer a vicious cycle between ICT advance- ment and marginal productivity decline. ICT growing economies expect growth as ICT increases but they cannot afford Co-evolutionary Acclimatization 3) Consequence of Two-faced nature of ICT
  8. 8. 8 IGE: ICT growing economies Fig. 6. Scheme of Optimal Dynamism for Balancing Captured and Un-captured GDP. 4) Optimal Balance between Captured GDP and Un-captured GDP ICT advanced economies ICT growing economies
  9. 9. 2. Measurement of Un-captured GDP 2.1 Framework of the Analysis (1) Basic Understanding Un-captured GDP can be traced from both sides as: a. New functions of online intermediaries such as e-commerce, online advertising and search engines, and b. Consumer’s preferences shift from economic functionality to supra-functionality beyond economic value. (2) Both Sides of Un-captured GDP Emergence 1) New Services provided by Online Intermediaries 2) Consumers Preferences Shift 9 Locomotive of the Spin-off Emerging Un-captured GDP.
  10. 10. 10 Internet Online Intermediaries (OI) Core player Provide platforms for the exchange of goods, services or information over the Internet Substantially change the way that goods, services and information distribution Social networks Twitter, Facebook, LinkedIN, YouTube Search engine Google, Yahoo, Wikipedia e-commerce platform eBay, Amazon, Alibaba, Rakuten, Priceminister Cloud computing Skype, Viber, Watspp, Yahho Messenger 1. Direct GDP contribution 2. Indirect GDP contribution by productivity increase 3. Beyond captured GDP Services provided by OI by (1) Private consumption (2) Government consumption (3) Investment involving OI* (4) Exported or imported Efficiency/cost improve by (1) Search provider (Reduce costs for information search) (2) Social networks (Find/exchange information efficiently) (3) e-commerce (Sell efficiently/purchase cheaper) (4) Cloud computing (Turning fixed costs into marginal costs) Services not counted GDP as (1) B2B platform by e-commerce (not final cons. but input to others) (2) Online advertising (Similar to B2B) (3) Consumer benefits of free services as Google search (4) Socio cultural value induced by social networks * 220 bil. Euro (1.7% of GDP) 210 bil. Euro (1.65%) 640 bil. Euro (5.0%) EU 27 in 2012 (GDP = 12,900 bil. Euro) Original Source: The Impact of Online Intermediaries on the EU Economy (Copenhagen Economics, 2013). Substantial total values including un-captured GDP change current 2% p.a GDP growth to 3-8% growth society. Fig. 7. Beyond Captured GDP generated by Online Intermediaries. Un-c.GDP measurement is critical * Not included in the estimate. In addition, “Underground resources” (e.g., Un-licensed software, online piracy), “Un-used potential” due to organizational reform delay cannot be overlooked.
  11. 11. 11 2) Consumer’s Preferences Shift (i) Measurement of Elasticity of Utility to Consumption Un-captured GDP is non-reflection of utility to consumption (measured by captured GDP) and its magnitude can be measured by elasticity of utility to consumption. Utility is governed by ICT stock (I) and Internet dependency (J) in the digital economy, its elasticity to consumption can be decomposed to elasticity of ICT to consumption and elasticity of the Internet to consumption as follows: )()()( ),( ),(),,(),( cjcicu C J J C C I I C C U U C J J C I I C C U J J U I I U JIUU JIQQJIVVQVUU                                 U: Utility, C: Traditional consumption V: Economic functionality, Q: Supra-functionality J: Internet dependency, I: ICT stock Thus, elasticity of utility to consumption can be estimated by a sum of elasticity of I to C and J to C. Utility is particularly induced by J. Fig. 8. Governing Factor of Utility in the Digital Economy.
  12. 12. 1.76 0.49 1.27 0.30 0.23 0.37 0.68 0.15 0.40 0.10 0.10 0.47 0.48 0.53 1.05 0.55 0.57 0.58 Finland Singapore USA UK Germany Japan Internet to Consumption ICT to Consumption 12 Fig. 9. Elasticity of Utility to Consumption in 6 Countries (2013). Finland Singapore USA UK Germany Japan Table 1 Elasticity of Utility to Consumption in 6 Countries (2013) 2/1 (ii) Elasticity of Utility to Consumption in 6 Countries 0.23 0.30 0.39 0.49 1.27 1.76 0.37 0.68 0.15 0.40 0.10 0.47 0.10 0.48 Under the digital economy, Elasticity of U to C = Elasticity of I to C + Elasticity of J to C
  13. 13. CapturedGDPUn-capturedGDP Fig. 10. Trends in Elasticity of Utility to Consumption in 6 Countries (1994-2013). (iii) Trend in the Elasticity in 6 Countries Singapore USA Japan Germany UK Finland 13 Elasticityofutilitytoconsumption 1994 95 96 97 98 99 2000 01 02 03 04 2005 06 07 08 09 2010 11 12 2013 Singapore Conspicuously high U reflects to C High dependency on Captured GDP Finland Extremely low U does not reflect to C High dependency on Un-captured GDP Internet commercialization Net bubble bursting Lehman shock cu cu cu
  14. 14. 14 (3) Factual Observation Captured GDP Traditional ICT Internet Un-captured GDP Advancement of ICT Paradigm change GDPatcurrent prices(US$billions) Stimulation from ICT advancement Inducement by people’s preferences shift Significance difference of consumption level between Finland and Singapore notwithstanding the similar level of GDP can be attributed to difference of un-captured GDP based consumption. This difference can be attributed to the difference of the state of spin-off in the shifting co-evolution between 3 mega-trends. Locomotive of such spin-off impacting on un-captured GDP can be: (i)Stimulation from ICT advancement, and (ii)Inducement by people’s preferences shift. Both maintain equilibrium leading to lifting power which can be depicted as follows: )1( nY YXAeXAZ n   Scale factor Primary impacts Secondary impacts Singapore Finland 300 200 100 Consumptionindex Fig. 11. Utility of Consumption Measured by Un-captured GDP. Fig. 12. The Locomotive for the Spin-off Impacting Un-captured GDP.
  15. 15. 15 (4) Measurement of the Magnitude of Un-captured GDP 1) Consumption Function Level of consumption C = a’ + b’ Y H = a + bY Y Y + uY Consumption measured by captured GDP Estimated gross consumption measured by both captured and un-captured GDP Gross income level motivating consumption Additional consumption measured by un-captured GDP (W) Captured GDP un-captured GDP aBase consumption 2) Discrepancy between two consumption functions (i) ICT advancement stimulation (Attributed to the Internet (J) with secondary impacts of consumer’s preferences)). : adjusting factor (= a - a’) (4) [ see Note] (ii) Consumer’s preferences inducement (Represented by elasticity of utility to consumption (E) with secondary impacts of J) (5)   buYaaYbbaabYaYbaHCW  )'()'()'()(''   n J eEAbuY Since (4) and (5) maintain equilibrium, leading to lifting power. (6) (7) (9) Un-captured GDP ratio (10) nm JE eEAeJHbuY   mn m n EJ E J eeJEA eJ eEAH        1 nn EJJEAH   lnlnlnln b eEe uY n JA    ln Y uY     m E eJHYb (1) (2) (3) 3) Identification of un-captured GDP a, a’: base consumption, b, b’: marginal propensity to consume.          m E eJHHuY H Y Y H Y H uY Y H buY H eJeJ Y uY H Y Y H mm EE        ,, (8) Note (Y elasticity to H) 0 H  
  16. 16. 16 2.2 Empirical Result Table 2 Governing Factors of Household Consumption in Finland and Singapore (1994-2013) 19158.1998.0.917.01003.1ln208.0ln185.1435.6ln 20.26.14   AICDWRadjEJJEH 19061.1998.0.887.01070.1ln214.0ln230.1520.6ln 22.25.14   AICDWRadjEJJEH 19060.1998.0.939.01003.1ln209.0ln158.1380.6ln 28.16.14   AICDWRadjEJJEH 12263.1982.0.575.11000.1ln198.0ln740.0721.3ln 222.000.12   AICDWRadjEJJEH 12163.1981.0.484.11070.1ln229.0ln735.0856.3ln 222.090.02   AICDWRadjEJJEH 12262.1982.0.658.11060.0ln172.0ln745.0593.3ln 222.010.12   AICDWRadjEJJEH (7.24*) (2.23*2) (-3.10*) (1.60*4) (0.69*5) (7.52*) (2.35*2) (-3.29*) (1.58*4) (0.64*5) (6.69*) (2.10*2) (-3.63*) (1.61*4) (0.69*5) (3.28*) (2.14*2) (-3.32*) (5.41*) (1.37*4) (3.27*) (2.08*2) (-3.64*) (5.31*) (1.25*4) (3.27*) (2.18*2) (-2.99*) (5.49*) (1.48*4) Finland S H L Singapore S H L S: Standard estimate, H: higher possible estimate, L: lower possible estimate. H: Household consumption (Index: 1994=100), E: Elasticity of utility to consumption, J: Internet dependency. Singapore’s E for 1994 -1996 are adjusted taking backward trend between 1997-2000. Figures in parenthesis indicate t-statistics (*, *2, *4, *5 means significant at the 1%, 5%, 20% and 50% level, respectively).
  17. 17. 0.95 H 0.87 S 0.83 L 0.43 H 0.35 S 0.29 L (1) Magnitude of Un-captured GDP – Un-captured GDP Ratio Fig. 13. Trend in the Internet-driven Un-captured GDP Ratio in Finland and Singapore (1994-2012). Un-captured GDP ratio = Un-captured GDP/ Captured GDP. S: Standard estimate, H: Higher possible estimate, L: Lower possible estimate. 17
  18. 18. 18 Finland Singapore Captured GDP 505 302 269 407 74 105 GDP at current prices (US$ billions) GDP 1994 2000 2005 2010 2013 600 500 400 300 200 100 0 Fig. 14. Trends in Captured and Un-captured GDP in Finland and Singapore (1994-2013). (2) Trends in Captured and Un-captured GDP
  19. 19. 19 600 500 400 300 200 100 0 Singapore 302 407 Captured GDP74 1994 2000 2005 2010 2013 Fig. 15. Comparison of Captured and Un-captured GDP in Finland and Singapore (1994-2013). While (captured) GDP is lower than Singapore, Finland depends largely on un-captured GDP. (3) Dependency on Un-captured GDP GDP at current prices (US$ billions) Captured GDP 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 300 250 200 150 100 50 Finland Singapore Internet dependency 93 75 Experience ratio 48 30 Share of retail sales 9 2 Clothing/footwear purchase Popular 4 B2B Internet use 1 14(world rank) Comparison of the Internet Use (2013) % Sources: ITU, WEF, Statistics Finland, Singapore DOS. Online shopping 600 500 400 300 200 100 0 Finland Un-captured GDP 269 505 Captured GDP105 1994 2000 2005 2010 2013 19 302 269
  20. 20. Finland (1996 – 2013) 1996 2013 Internet productivity of ICT Singapore (1996 – 2013) 19962013 Fig. 16. Contrast of the Shift to New Co-evolution between Finland and Singapore (1996-2013). 3. Consequence of Increasing Dependency on Un-captured GDP (1) Shift to New Mega Trend 2003 2000 20 Un-capturedGDPratio
  21. 21. 21 (2) Correspondence to People’s Preferences Shift Economic Functionality Fig. 17. Shift from Economic Functionality to Supra-functionality beyond Economic Value. People’s preferences shift in Japan Source: Japan Cabinet Office.
  22. 22. 1 FIN 2 SGP 3 SWE 4 NLD 5 NOR 6 CHE 7 USA 9 UK 12 DEU 13 DNK 15 ISR 16 JPN Gaining profits from abroad GNI/GDPratio Fig. 18. Comparison of Interactive Return Gain Structure in 12 ICT Advanced Countries (2012, 2013 average). Figures on the country indicates the ICT competitiveness rank in 2013. Sources: World Economic Outlook Database (IMF 2013, 2014), World Health Statistics 2014 (WHO 2014),. GNI / GDP Ratio 1 FIN 2 SGP 3 SWE 4 NLD 5 NOR 6 CHE 7 USA 9 UK 12 DEU 13 DNK 15 ISR 16 JPN GDP Growth Rate (2006-2013) % p.a. at fixed price GDPgrowthrate(%p.a.)(3) Change in Interactive Return Gain Structure Losing domestic gains Singapore’s Losing Structure While Singapore enjoys growth, it loses domestic gains. Finland gains profits from abroad under the great stagnation. 22 GNI – GDP = Income balance + balance derived from favorable terms of trade (higher exports prices with lower imports prices). This irony can be attributed to the consequences of the strategic option in shifting to new co-evolution or clinging to traditional co-evolution.
  23. 23. 23 (4) Emergence of Disruptive Business Model Fig. 19. Consumer Surplus of Music and Audio - visual Services (Person per month: Aged 20s.). Source: White Paper of Japan’s ICT (Min. of Internal affairs and Communication, 2016). Fig. 20. Un-captured GDP Emerged by Uber (US$/trip, NYC). Source: Uber's Ride-sharing Revolution (Watanabe et al., 2016). Un-captured GDP 1) Music and Audiovisual Services 2) Ride Sharing Service: Uber % of cumulative answer Yen Willing to pay Actual payment (146 Yen/m)
  24. 24. 24 IS LM Principle of effective demand Consumption function (a, b: coefficient) Investment function ( r : interest rate) Multiplier theory GDP VbaC  )(rII  Substitute C in V IGVbaIGCV  )( I b G bb a V        1 1 1 1 1 Fiscal policy r Monetary policy MrI  CV (Money supply) Price x Interest rate = Interest (Fixed) (Bond yields) Securing government fund IS: Investment and Saving LM: Liquidity of Money Tax revenue + National bond TaxV  Income Tax ratio r IS-LM analysis (Integration of effective demand and money supply and interest) Growth Private consumption Government consumption IGCExIGCV  Invest -ment Net exports (Exp – Imp) (5) Reconstruction of Taxation System Good economic condition (Maximize synergy effects as a consolidated system) Lower r I increase V increase Tax increase C increase V increase G increase Handled sole by European Central Bank (ECB) and each member country is allowed only Fiscal Policy initiative. 100% 61.1 20.6 21.1 - 2.8 (16.2-19.0) [Japan 2013] 68.5 18.2 16.4 -3.1(13.4 -16.5) [US 2014] 55.2 24.9 20.8 - 0.9 (38.2 -39.1) [Finland 2013] 35.0 9.8 29.1 26.1 (194.1-168.0) [S’pore 2013] Fig. 20. Captured GDP Flow in EMU. Imposed to ICT phase, Used for supra-functionality
  25. 25. 25 (6) Optimal Dynamism Harnessing the Vigor of ICT Growing Economies Co-evolutionary Acclimatization Past experiences Current resources Future dream Un-captured Captured GDP GDP Fig. 21. Optimal Dynamism Harnessing the Vigor of ICT Growing Economies.
  26. 26. National level Elucidate Trap in ICT advancement Two-faced nature of ICT Bi-polarization Conceptualize Operationalize Business/Individual level Un-captured GDP Ride-sharing revolution: Uber Trust-based education toward digitally-rich learning environments Commodification of experiences Digital music industry Game industry Printing/publishing ind. Uber’s ride-sharing revolution ICT-driven Disruptive Business Model (IDBM) Uber’s worldwide expansion 0. Overview 0.1 Journal Papers Legal battles Co-evolution IDBM without CCSD IDBM with CCSD Co-evolution between Trust in Teachers and Higher Education toward Digitally-rich Learning Environments CCSD: Consolidated Challenge to Social Demand Trust (Overdrawing past information) Commodification of experiences 26
  27. 27. 27 0. Overview 1. New Stream of Innovation -Spinoff to Un-captured GDP-driven Co-evolutionary 3 Mega-trends 2. Uber’s Ridesharing Revolution - ICT-driven Disruptive Business Model (IDBM) 3. Trust-based Education toward Digitally-rich Learning Environments 4. Optimization through Commodification of Experiences 5. Harnessing the Vigor of Memory and Dream 6. Conclusion Platform Ecosystems Impact on GDP ICT-driven Disruptive Business Model with Consolidated Challenge to Social Demand
  28. 28. 28 2. Uber’s Ridesharing Revolution 2.1 ICT-driven Disruptive Business Model (1) Uber: Hero of Spin-off Fig. 22. Scheme of Uber’s Spinoff Dynamism. 1. Customer: Convenient, cheep, time saving 2. Driver: No invest, use time, to be a boss 3. Uber: Market creation, optimal price, utilize sleeping resources 4. Government: Accelerate disruptive innovation People’s preferences shift Advancement of ICT Un-captured GDP emergence Digital economy Uber
  29. 29. Fig. 23. Uber’s Astounding Rise in Trips and Continuous Decline in Prices in NYC. $0 $200 $400 $600 $800 $1,000 $1,200 $1,400 $1,600 Mar. 2009 Uber established Jun 2013: Medallion prices stagnate May 2011: Uber launched in NYC 2013/6 2015/1 2015/92014/52011/5 PT PT PU UT TT MP Un-captured GDP per trip 31 2004/1 2009/3 2011/5 2013/6 2014/5 2015/9 Magnitude of Un-captured GDP te e MP 02.0 36.61 2247    TT TUTT A UT UPTP P    Trips Prices Estimated medallion prices without Uber Aggregated prices (2) Uber’s Conspicuous Development Taxi price Taxi price Uber price Uber trips Taxi trips Uber Uber Taxi Taxi NYC corporate medallion prices MP (1,000 US$) 1. Astounding rise in trips and prices decline. 2. Provide striking value to all stakeholders as (1) Better services, with cost and time saving for passengers, (2) High efficient operation without new investment for drivers, (3) Optimal price-setting and market making for company (4) Advancement of nation’s disruptive innovation for government.
  30. 30. Fig. 24. Dynamism of ICT-driven Disruptive Business Model (IDBM). 30 (2) Dynamism of ICT-driven Disruptive Business Model Uber’s system success 1. Co-existing development trajectory with taxi corresponds to two-faced nature of ICT, 2. This can be attributed to a virtuous cycle between price decline and trips increase, 3. This virtuous cycle can be attributed to ICT’s self-propagating function, and 4. This function plays a vital role in spinoffs from traditional co-evolution to new co-evolution. Computer initiated ICT Taxi Internet Uber Demand
  31. 31. 31 2.2 Consequence of ICT-driven Disruptive Business Model - Uber’s Expansion and Battles (1) Rapid Expansion Fig. 25. Uber’s Expansion in 479 Cities in the World (as of June 2016). Source: Uber. Uber expanded rapidly: 479 cities in more than 75 countries by June of 2016. (2) Emergence of Legal Battles Proportional to such rapid expansion legal battles emerged significantly. Operating notwithstanding legality Ban/Partial ban Fig. 26. Contrasting Features of Uber’s Global Expansion in 16 Countries (as of June 2016). Sources: Authors classification based on, NY Times, HuffPo, Reuters, WSI, CNN and local news reports.
  32. 32. 32 2.3 Specific Features of ICT - Sources of Uber’s Expansion and Battles (1) Specific Features Since Uber is seen as the jewel of ICT, its system success (expansion) and failure (battles) can be attributed to the following ICT’s indigenous function:
  33. 33. 33 (2) ICT’s Indigenous Functions Derived from its Specific Features 1) Self-propagating Nature Diffusion trajectory of innovative goods Y Simple Logistic Growth (SLG) with fixed carrying capacity (N) Particular innovation which create new N during Logistic Growth within a Dynamic Carrying Capacity the process of diffusion. (LGDCC) Carrying capacity increases as Y increases. Functionality spirally increases as Y increases. Self-pr ) )( 1)(( )( N tY taY dt tdY  at be N tY    1 )( ) )( )( 1)(( )( tN tY taY dt tdY  ta aa bat k k k k ebe N Y      /11            )( )(1 1 1 )()( tY tY a tYtN dt tdY tY tY a tY tY tN FD )( )( )(1 )( 1 1 )( )(     Self-propagation a. Spinoff b. Institutional elasticity
  34. 34. Comparison of Spinoff State in Finland vs Singapore and Taxi vs Uber Simple logistic growth Carrying capacity enhance Logistic growth within a dynamic carrying capacity Self-propagating dynamism by spinning off to higher functionality level Finland Singapore 36 N a b ak bk c adj.R2 0.815 (31.73) 0.311 (8.50) 1.833 (9.28) 0.965 1.000 (8.47) 1.123 (2.35) 23.519 (1.19)** 0.149 (3.75) 2.734 (4.80) 0.047 (20.43) 0.985 N a b ak bk c adj.R2 0.344 (63.61)* 0.591 (10.57)* 16.58 (3.71)* 0.982 0.344 (63.46)* 0.591 (10.59)* 16.58 (3.71)* 1.00*10-9 (-) 1.00*10-9 (-) 0.982 Un-captured GDP ratio (1994-2013) SLC LGDCC SLC LGDCC Table 3 Comparison of Spin-off State between Finland vs Singapore and Taxi vs Uber Self- propagating Self- propagating
  35. 35. 2) Bi-polarization Fatality of Logistic Growth Logistic growth function as a function of time t can be depicted as follows: This function can be developed to the following bi-polarization function: at be N Y N Y aY dt dY    1 )1( Inflection point Fig. 27. Bi-polarization Fatal to Logistic Growth Function. 37
  36. 36. 36Fig. 28. Transformative Role of Co-evolutionary Acclimatization . that harnesses the vigor of counterparts enables both economies maintain sustainable growth. 3) Two-faced nature of ICT While the advancement of ICT contributes to enhancing its prices by increasing new functionality development, dramatic advancement of the Internet tends to decrease ICT prices due to freebies, easy copying and mass standardization, among other things. ICT advanced economies suffer a vicious cycle between ICT advance- ment and marginal productivity decline. ICT growing economies expect growth as ICT increases but they cannot afford Co-evolutionary Acclimatization
  37. 37. 37 IGE: ICT growing economies Fig. 29. Scheme of Optimal Dynamism for Balancing Captured and Un-captured GDP. 4) Optimal Balance between Captured GDP and Un-captured GDP ICT advanced economies ICT growing economies Uber Uber introducing countries
  38. 38. 38 2.4 Structural Sources of Legal Battles Given that Uber’s system success depends on the advancement of ICT, its considerable legal battles proportional to its rapid expansion can be attributed to ICT’s indigenous functions characterized as (1) Spin-off, (2) Institutional elasticity, and (3) Co-evolutionary acclimatization.
  39. 39. 39 0.00 10.00 20.00 30.00 40.00 50.00 60.00 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 Trips per day 2 1 T T U U (1) Optimal Growth Rate for Spin-off Fig. 30. Comparison of Uber Trips Estimate (Jun. 2013 – Sep. 2015). 2013/6 2014/1 2014/5 2015/3 2015/9 Table 4 Comparison of Adaptability of Uber’s Development Trajectory by Growth Rate to LGDCC (NYC, Jun. 2013-Sep. 2015) Rapid growth 11% p.a (2015/3 – 2015/9) Steady growth 9% p.a Rapid growth Non self-propagation Steady growth Self-propagation Optimal velocity of growth would be crucial to self-propagating functionality development that leads to spin-off. While self-propagation can be expected in steady growth by fitting to LGDCC, it cannot be expected to have rapid growth. Figures in parenthesis indicate t-statistics: all significant at the 1% level except *3: 5%, *5: 15%, *6: 20%, x: non-significant.
  40. 40. 40 Fig. 31. Correlation between Centralization of Wage Setting and Union and CBA Density in 19 Countries in the Late 1990s. CBA: Collective bargaining agreements. Union and CBA density = (Union density + CBA coverage)/2 Sources: Warner (2002), The Global Competitiveness Report 2015-2016 (World Economic Forum, 2015). (2) Institutional Elasticity of the Host Flexibility of wage setting (not centralized ratio) UnionandCBAdensity Ranking of flexibility of wage determination
  41. 41. Functional Development level Rapid growth Steady growth Elastic nation/city Non-elastic nation/city Time Level of Functionality Development, Institutional elasticity Co-evolution withinstitution Disengagement frominstitution Germany France Singapore Saudi Arabia Tokyo Finland Non-adaptive level Fig. 32. Scheme of Adaption of Uber in Institutional Systems . )( )(1 1 1 tY tY a FD    Uber Adaption in Countries/Cities depending on Growth Rate and Institutional Elasticity 43 Growth rate
  42. 42. 42 1.Contrast between countries with and without legal battles can be attributed to with or without CCSD (Consolidated Challenge to Social Demand). 2.While the former develops co-evolutionary dynamism, the latter results in disengagement. Fig. 33. Co-evolution and Disengagement between Uber-driven IDBM and Institutional Systems. (3) Co-evolution and Disengagement Countries without legal battle Countries with legal battle Uber induced CCSD leading to a co-evolution between ride-sharing revolution and advancement of the institutional systems. Singapore: Induced incorporating users requirements into the tripartism framework (company, employee, government) by stimulating social demand (transport, job, productivity). Saudi Arabia: Enabled women’s social participation by providing the reliable transportation leading to a co-evolution. Tokyo: Stimulated better service seeking competitive market, broader stakeholders involvement for social demand solution. Traditional quasi-monopolistic market protected by non-innovative government impeded Uber’s revolution resulting in disengagement from the institutional systems. Germany: Government non-innovative policy urging traditional legal requirements in response to taxi companies’ requirement to preserve existing profit impeded Uber’s disruptive innovation resulting in failing CCSD construction. France follows the similar results. Finland: Suspicious to illegal, but connive to operate.
  43. 43. (4) Co-evolutionary Acclimatization -ICT-driven Disruptive Business Model with Consolidated Challenge to Social Demand Host: Uber introducing countries/cities Harness the vigor of counterparts (CCSD) Co-evolutionary acclimatization Bi-polarization between ICT advanced and growing group Fig. 34. Scheme for ICT-Driven Disruptive Business Model with Consolidated Challenge to Social Demand (IDBM – CCSD). Thus, ICT-Driven Disruptive Business Model with Consolidated Challenge to Social Demand (IDBM – CCSD) by harnessing the vigor of counterparts would be decisive for resilient IDBM. 45 Computer initiated ICT Taxi Internet Uber
  44. 44. 44 2.5 Lessons from Success Model 2.5.1 IDBM with CCSD (1) CCSD in Success Table 5 Structure in CCSD in Success Countries/Cities Consolidated challenge by all stakeholders Social demand Co-evolutionary acclimatization Singapore Tripartism framework User involvement Company, employee, user, government consolidation Traffic service, Job opportunity, Overall productivity enhance, Digital innovation Tripartism framework, Well developed infrastructure, Innovation seeking spirit Saudi Arabia Women (user, employee) Company involvement Government involvement Women’s social participation, Education, Industrial structure Strong inertia to women’s social participation, Affluent financial base Japan User welcome Company, employee concern Government involvement Traffic service, ICT advancement, e-commerce, Depopulation, Aging society High potential demand, Demanding nature, Matured competitive environment
  45. 45. 45 NTUC this Week, 26 Jan. 2007 (NTUC: National Trades Union Congress). (2) Lessons from Tripartism
  46. 46. 2.5.2 Significance of Shift to IDBM with CCSD (i) Nowadays, a key factor in obtaining business opportunities is the ability to solve social demand. [SD] (ii) A company to gain a profit must consolidating all stakeholders: company, employee, user, and government with respective heterogeneous objectives. (iii) Developing systems that address all stakeholders’ demands in society as a whole can allow these disparate groups to successfully function together. [CC] (iv) The company that can attain such system success is required following abilities: (a) Penetrate the social demand that can be its business opportunity, (b) Organize and orchestrate all stakeholders, and (c) Attain the system success thereby gain profit. (v) Development and utilization of ICT enable such endeavor. [IDBM] (vi) Thus, shifting to IDBM with CCSD has been becoming crucial. (vii) Function of trust-based tripartism framework suggests the significance of ICT and trust toward IDBM with CCSD in the digitally-rich environment. Company Employee User Government Social demand Profit seeking Job after retire To be a boss Utility, eco, health, comfort Income, job, welfare, happiness, eco, aging, health, safety ICT advancement (Internet, smartphone, big data) 48 Fig. 35. Consolidated Challenge to Social Demand.
  47. 47. 47 3. Trust-based Education toward Digitally-rich Learning Environments 3.1 Co-evolution with Trust as a Consequence of IDBM with CCSD (1) Consequence of IDBM with CCSD Fig. 36. Consequence of IDBM with CCSD. Own experiences + Past information Decrease risk and uncertainty Enhance trust
  48. 48. Fig. 37. Scheme of Spinoff Dynamism. Uber Ride sharing evolution DILE (Digitally-rich Innovative Learning Environments) 48 (2) Spinoff from Traditional Co-evolution to Un-captured GDP oriented New Co-evolution Transforming Infusing Applying Emerging Digital economy Shifting from trust in personality to trust in system of augmented reality (AR)
  49. 49. 49 3.2 Co-evolution between ICT, Trust and Higher Education (1) Comparison in 20 Countries Fig. 38-2. Higher Education (2013). Fig. 38-3. Trust in Teachers (2013). Fig. 38-1. ICT Advancement (2012-2015 average).
  50. 50. 50 (2) ICT-driven Education Development Fig. 39. ICT-driven Education Development in 120 Countries (2013). at be N Y N Y aY dt dY    1 )1( IAC ISC IGC ICT advancement (Z) Highereducation(Y)
  51. 51. 51 (3) Correlations Table 6 Co-evolution and Disengagement between ICT, Educational Level, and Trust in Teachers Fig. 40-1. Correlation between Higher Education and Trust (2013). Fig. 40-2. Correlation between ICT and Trust (2013). +: Co-evolution - : Disengagement Y X X Z Why trust decrease as ICT advance? (1) (2) (3) ( < 0) IAC ISC IGC IAC ISC IGC 872.0. 2 RadjIAC ISC IGC IAC ISC IGC 848.0. 2 Radj D: Dummy (D1: IAC, D2: ISC, D3: IGC, D: Jpan, Kor, Isr, Chz =1, Others = 0). Figures in parenthesis: t-statistics (all significant at the 1 % level, except **: 3%)
  52. 52. 52 3.3 Structural Sources of Trust Decrease as ICT Advance (1) Bi-polarization Fatality (2) ICT Elasticity to Trust (3) Composition of ICT Elasticity to Trust (4) ICT Elasticity to Higher Education IGC ISC IAC IAC: ICT advanced countries ISC: ICT semi-advanced countries IGC: ICT growing countries 1. Trust decrease ( ) ) depends on stage of ICT advancement. 2. ISC suffers a vicious cycle between Z and Y.
  53. 53. 53 (4) Blended Learning and Teacher’s Resistance - Transition to ICT Advanced Countries Fig. 41. State of Hybrid Development in 20 Countries. 1. ICT’s contribution to higher education can be developed in a hybrid manner through traditional teaching practice and blended learning (introduction of digital and online media in educational system). 2. While strong resistance by teachers impedes blended learning, once a certain higher education level has been attained, increasing dependency on blended learning exceeds such resistance leading to co- evolution between ICT and higher education. 3. While IAC has attained, ISC is in transition from traditional to blended learning overcoming resistance.
  54. 54. 54 Fig. 42. Stages of ICT Integration in Education. (5) Stages of ICT Integration in Education 1. ICT has been integrated to an education system by 4 stages as (i) Emerging, (ii) Applying, (iii) Infusing, and (iv) Transforming. 2. While the 1st 3 stages are advancing toward more digitally rich, the 4th stage transforms learning environments into digitally-rich new environments that can absorb and apply ICT-driven new services to higher education. 3. Transforming stage accelerates over- drawing of past information by introducing such ICTs as augmented reality (AR), simulations and digital games thereby increasing trust. 4. While IAC has shifted to transforming stage and constructed a virtuous cycle between ICT and trust by overdrawing past information. ISC and IGC have remained non-transforming stage thereby suffered a vicious cycle between ICT and trust. Trust decrease depends on stage of ICT advancement
  55. 55. 55 ) 1 1( FD aY dZ dY  Overdrawing past information 3.4 Co-evolutionary Acclimatization toward Digitally-rich Learning Environments – Harness the vigor of Past Information (1) Reconstruction of a Virtuous Cycle ISC, IGC IAC Shifting to higher productivity level by harnessing the vigor of past information Fig. 43. Co-evolutionary Acclimatization Harnessing the Vigor of Past Information. dY dZ  dY dZ 2 ,0 0)2( )( 1 )1( 1 1 2 N Yand dZ dY when dZ dY NY YYN aYdZ d dZ d dY dZ dZ d N Y dZ dY                                  Z Enables absorption and application of ICT-driven new services for higher education, which hitherto could not be afforded (Un-captured GDP) Reconstruct a virtuous cycle by overcoming trap in ICT advancement Transformative ICT
  56. 56. 56 Current Past Future National Economy Uber Sleeping resources Digital learning Time Music/Game Memory Dream (2) Transformative Direction of Co-evolutionary Acclimatization Target Fig. 44. Transformative Direction of Co-evolutionary Acclimatization Target in ICT-driven Disruptive Business Model. Increasing significance of co-evolutionary acclimatization for ICT-driven disruptive business model. Its target has been shifting from current economy (e.g., economic growth in growing economies) to sleeping resources (e.g., Uber’s ride sharing revolution), past time (e.g., trust-based higher education), and to past memory/future dream (e.g., digital music, game).
  57. 57. 57Fig. 45. Optimal Dynamism Harnessing the Vigor of Time, Memory, Dream. ICT advance T (Z) )( dY dZ dY dT (3) Optimal Dynamism Harnessing the Vigor of Time, Memory, Dream Co-evolutionary Acclimatization Enables absorption and application of ICT- driven new services, which hitherto could not be afforded (Un-captured GDP) Past experiences Current resources Future dream
  58. 58. 58 4. Optimization through Commodification of Experiences 4.1 Commodification of Experiences (1) Transfer the Anger into a Springboard for New Innovation 1. Shift of consumers preference from economic functionality to supra-functionality beyond economic value emerges conflict in the transition leading to growing anger of consumers. 2. Innovation-consumption co-emergence has thus become crucial. Commodification of experiences may provide significant solution to this problem. Fig. 46. Dynamism in Transferring the Anger into a Springboard for New Innovation.
  59. 59. 59 (2) From Invisible Hand of God to Voiceless Voice of Consumers Fig. 47. Scheme in Conceptualizing Invisible Voice of Consumers. Optimization
  60. 60. 60 Fig. 48. Scheme of Facial Temperature Feedback Hypothesis. (3) Facial Temperature Feedback Hypothesis Confronting unforgettable memory
  61. 61. (1) Methodology Monitor the consumers’ facial temperature by the thermography: novel psychophisiologal measuring technique enables observation in the objective circumstances without providing any cautions to examinees. Record in a PC Analyze the recorded data by the exclusive software “FLIR Research IR” (able to identify a pin-point temperature) 33.4 4.2 Demonstration by Experiment (a) With the measurement of the relationship between attractive goods and consumers’ facial temperature elevations at the leading supermarkets in Japan and Finland. (b) Demonstrate facial temperature feedback hypothesis. 63
  62. 62. Event wagon on which sweetened ban is displayed (15 February) Installation of PC for data recording (15 February) Hanging situation of the thermography (15 February) Shelf on which PC is stalled (15 February) Positions of event wagon and the thermography (15 February) Angle of the thermography and target of monitoring (21 February) 64 (2) Pilot Experiment at a Leading Supermarket 1) Tokyo Fig. 49. Pilot Experiment at a Japanese Leading Supermarket in Tokyo (February 16-21, 2011). Melon-bread with reasonable price, attractive enough to empting shoppers appetite
  63. 63. 2) Finland Cosmetics corner of Sokos Supermarket (7 March 2012) Cosmetics corner Thermography connected to PC Shoppers accessing to cosmetics corner Examining anticipating cosmetics Inducement by sales promoter Trial makeup 65 Fig. 50. Pilot Experiment at a Finish Leading Supermarket in Jyvaskyla (March 6, 7 2012).
  64. 64. 31.0 31.5 32.0 32.5 33.0 33.5 34.0 1 2 3 4 5 6 7 Accessed to the event corner Perceived Recognized Decided to purchase 31.4 31.7 31.7 31.6 31.9 31.7 33.3 Further perceived Further recognizedAccessed to the event corner Perceived Further perceived Recognized Further recognized Resemble past learning Surprise Remember past learning Gratification Correspondence with gratification experienced Facetemperature MCS SNS MCS SNS MCS (3) Empirical Results – Tokyo 1) Facial temperature 31.4℃ 31.7 31.7 31.6 31.9 31.7 33.3 MCS: Metabolic Control System (elevate temperature), SNS: Sympathetic Nervous System (descend temperature) Fig. 51. Standard Pattern of Facial Temperature Change in Purchased (Tokyo). Decided to purchase Facial temperatures elevate recalling gratification ever experienced. 66
  65. 65. 2/19 15:00-16:00 2/19 16:00-17:00 2/19 17:00-18:00 2/20 15:00-16:00 2/20 16:00-17:00 2/20 17:00-18:00 2/20 18:00-19:00 0 2 4 6 8 10 12 14 16 18 20 0 100 200 300 400 500 600 700 S X Sales volume: S 2) Correlation between facial temperature and sales Attractive goods elevate customer’s facial temperature. Facial temperature index: X 67 Fig. 52. Correlation between Facial Temperature and Sales.
  66. 66. Roman holiday Sputnik No. 1 Tokyo Olympic Game Beatles Apollo (1953 ) (1957) (1964) (1966) (1969) 4.3 Commodification of Experiences (1) Unforgettable Impressive Memory Experienced in the 20s 68 Fig. 53. Major Impressive Memory Never Forget Experienced in their 20s.
  67. 67. 69 (2) Sublimation of 20s Experiences into Supra-functionality beyond Economic Value Social value Cultural value Aspirational value Tribal value Emotional value Energy saving Small is beautiful Eco Cool Sensitivity Private brand Voluntary social service Aesthetic sense Quiet simplicity Distance between firms and society, Creation of social value Social welfare, Support of social disability Creation of social communication, Contribution to social platform Contribution to social needs (e.g. PV) Shifting from tough agreement to loose agreement Brand value, Private brand (PB) Solicitation of saving mind Aesthetic sense, Quiet simplicity, Cool, Cute, J-sense Voluntary participation Sensitivity of Japanese products (Steve Jobs aspired) Authentic goods, Goods with high-grade sense Professional, Rich Sense of recognizing own position Fellow feeling, Patriotism Re-recognition of Made in Japan Symbolic meaning Five senses, Sensitivity Only one Casio-mini Cassette-tape Apple II Carry-compo Walkman PC Card calculator (Casio) (Sony) (Apple) (Apple) (Sony) (NEC) (Sharp) Cup noodle Bulgaria yogurt Chipstar Surprise chocolate
  68. 68. (3) Platform for Commodification of Experiences for Innovation-Consumption Co-emergence Utmost gratification of consumption ever experienced Memorize in the brain Similar innovative goods/services Collate with memory of utmost gratification Facial temperature change Elevate Descend Awake sleeping experiences on utmost gratification once experienced Real Pseudo Commodification of experiences corresponding to utmost gratification of consumption ever experienced Fig. 54. Platform for Commodification of Experiences. Induce resonance between innovative goods and consumers Trigger innovation-consumption co- emergence 70 This dynamism leads a way to Optimal Dynamism Harnessing the Vigor of Time, Memory, Dream
  69. 69. 69 5. Harness the Vigor of Memory and Dream 5.1 Digital Music as a Platform for Retrieving Music Information – Bridging Digital Innovation to the World of Memory Retrieval (1) Approach 1. The advancement of the Internet has changed the way of business and daily life dramatically. Evolutional change in the music industry over the last decade can be one of the most striking example. 2. Fundamental source of this evolution can be attributed to the dramatic advancement of the Internet- mediated technologies that enabled constructing a platform for retrieving music information thereby customers can enjoy music ubiquitously by their own initiative. 3. This platform provides significant insight in bridging digital innovation to the world of memory retrieval. 4. Given the increasing significance of memory retrieval in creating customer initiated business model in the digital economy, dynamism of this platform should be analyzed.
  70. 70. Adoption of Digital music Enable customers to enjoy music ubiquitously by own initiatives Music information retrieval (MIR) MIR enabled applications 70 Fig. 55. Platform of Digital Music Industry. Platform Internet-mediated technologies Digital innovation Musical Moment Listen Music Cognitive itch Stuck in head Appreciate & Memorize Striking / intriguing Hooked Catchy partInvoke Past memories Emotionally, structurally, perceptually Catchy part of the song Mediation tools Description Search music by Singing or humming, natural language search, keyboard search, similarity search, text search etc. Music service providers, music search engines, music generators etc. Enable to categorize, manipulate and even create music (Recommender systems, track separation, instrument recognition, auto transcription, categorization and music generation etc.) Key features Substitution for physical music New style of music treat (search, deep discovery, listening, personalization and sharing) Music intelligence enabled services Transformation to music information (2) Platform of Digital Music Industry
  71. 71. 71 (3) Empirical Analyses On the basis of the empirical analysis of the development trajectory of the music industry in the world over the last 4 decades, unique dynamism of the music industry enabling to retrieve information identical to music is elucidated. Conceptualization of this dynamism applicable to business model construction in the digital economy is also attempted. 1) Substitution Dynamism 1. Digital music for physical music (1974-2014) monthly 2. Within digital music (2004-2014) monthly Logistic growth within a dynamic carrying capacity (self-propagation function) Bi-logistic growth, tri-logistic growth (substitution dynamism) 2) Institutional Factors Contrasting Heterogeneous Dependency on Digital Music 1. 50 countries in 2007 – 2014 (comparison between 2007, 2010 and 2014?) 2. Music: Subscription vs streaming ? Institutional factors: Global Competitiveness Report, Global Information Technology Report Principal Component Analysis Factor Analysis Note Demand pattern of digital music Almost all data are supply side (e.g., sales, revenues)
  72. 72. 72 5.2 Social Gaming as a Platform for Harnessing the Vigor of Dream
  73. 73. 6. Conclusion 75 1. Spinoff to Un-captured GDP-driven Co-evolution of 3 Mega-trends (1) Advancement of ICT incorporates a two-faced nature as ICT advancement has resulted in price decrease due to freebies, easy copying and standardization. (2) The Internet provides incredible services to people but they cannot be captured by GDP. (3) Un-captured GDP can largely be attributed to consumer utility which corresponds to their preferences shift from economic functionality to supra- functionality beyond economic value. (4) This shift, in turn, induces further advancement of the Internet, leading to co- evolution of the 3 mega-trends: advancement of ICT, a shift to un-captured GDP and also a shift to people’s preferences. (5) Consequently, today’s global digital innovation can be identified as spining-off co-evolution toward un-captured GDP-driven new mega-trends.
  74. 74. 6. Conclusion (2) 76 2. ICT-driven Disruptive Business Model - Uber’s Ridesharing Revolution (1) Under such circumstances, we are in the midst of transformative shift in business design. (2) Business models move from pipes to platforms which allow external producers and users to exchange value with each other, leading to users acting as producers and creating value for other users. (3) The ridesharing revolution initiated by Uber can be seen as the jewel of such an ICT-driven platform ecosystems. (4) It creates better services and higher value to all stakeholders: passengers, drivers, company, and government. (5) The success of Uber system suggests the significance of an ICT-driven disruptive business model (IDBM).
  75. 75. 6. Conclusion (3) 2.-2 Consolidated Challenge to Social Demand for Resilient Platforms (1) While Uber has expanded rapidly worldwide, the contrast between co- evolutionary success and legal battles with host countries/cities has become distinct. (2) This can be attributed to the bi-polar nature of ICT’s rapid advancement, suggesting the significance of harnessing the vigor of counterparts. (3) Lessons from successful co-evolution suggest the significance of IDBM with CCSD (consolidated challenge to social demand). (4) Thus, IDBM with CCSD suggest the significance of resilient platform ecosystems that incorporate contingency. 77
  76. 76. 6. Conclusion (4) 78 3. Trust-based Education toward Digitally-rich Learning Environments (1) Lessons of IDBM with CCSD suggest the significance of ICT and trust toward the digitally-rich learning environments. (2) While ICT advanced countries have embarked on co-evolution between ICT, higher education and trust, ICT growing countries have not been successful in this due to a vicious cycle between ICT and trust. (3) This co-evolution can be attributed to co-evolutionary acclimatization by harnessing the vigor of past information thereby absorption and application of ICT-driven new services for higher education, which hitherto could not be afforded, were enabled. This is equivalent to un-captured GDP emergence. (4) With the notion that trust depends on overdrawing of past information, such business model as constructing co-evolutionary acclimatization by harnessing the vigor of time should be envisioned.
  77. 77. 6. Conclusion (5) 79 4. Optimization through Coomodification of Experiences (1) Shift of consumers preferences to supra-functionality beyond economic value emerges conflict in the transition leading to growing anger of consumers. (2) This anger can be transformed into a springboard for new innovation. (3) This highlights the significance of voiceless voice of consumers leading to the increasing significance of commodification of experiences. (4) This commodification enables to pave a way to identifying the optimal dynamism harnessing the vigor of time, memory and dream.
  78. 78. 78 References 1. C. Watanabe, K. Naveed, P. Neittaanmaki and B. Fox, “Consolidated Challenge to Social Demand for Resilient Platforms: Lessons from Uber’s Global Expansion,” Technology in Society (2016) in print. 2. C. Watanabe, K. Naveed, P. Neittaanmaki, “Co-evolution of Three Mega-trends Natures Un-Captured GDP: Uber’s Ride-sharing Revolution,” Technology in Society 46 (2016) 164-185. 3. C. Watanabe, K. Naveed, P. Neittaanmaki and Y. Tou, “Operationalization of Un-captured GDP: Innovation Stream under New Global Mega-trends,” Technology in Society 45 (2016) 58-77. 4. C. Watanabe, K. Naveed and P. Neittaanmaki, “Dependency on Un-captured GDP as a Source of Resilience Beyond Economic Value in Countries with Advanced ICT Infrastructure: Similarity and Disparities between Finland and Singapore,” Technology in Society 42 (2015) 104-122. 5. C. Watanabe, K. Naveed and W. Zhao, “New Paradigm of ICT Productivity: Increasing Role of Un- captured GDP and Growing Anger of Consumers,” Technology in Society 41 (2015) 21-44. 6. C. Watanabe, “Innovation-Consumption Co-emergence Leads a Resilience Business,” Innovation and Supply Chain Management 7, No. 3 (2013) 92-104. 7. C. Watanabe, W. Zhao and M. Nasuno, “Resonance between Innovation and Consumers: Suggestions to Emerging Market Customers,” Journal of Technology Management for Growing Economies 3, No. 1 (2012) 7-31.
  79. 79. Appendix I. Measurement of Un-captured GDP 81
  80. 80. Un-captured GDP ratio (1) Spin-off State Simple logistic growth Carrying capacity enhance Logistic growth within a dynamic carrying capacity Self-propagating dynamism by spinning off to higher functionality level Finland Singapore Finland Singapore Finland has spun-off to new co-evolution while Singapore remains traditional co-evolution. 82 D: Dummy variable (2002-2005 = 1, other years= 0; During the period of stagnation due to the bursting of the Net bubble.) *, *** : Significant at the 1% and the 10% level, respectively. N a b ak bk c adj.R2 0.815 (31.73)* 0.311 (8.50)* 1.833 (9.28)* 0.965 1.000 (8.47)* 1.123 (2.35)* 23.519 (1.19)*** 0.149 (3.75)* 2.734 (4.80)* 0.047 (20.43)* 0.985 N a b ak bk c adj.R2 0.344 (63.61)* 0.591 (10.57)* 16.58 (3.71)* 0.982 0.344 (63.46)* 0.591 (10.59)* 16.58 (3.71)* 1.00*10-9 (-) 1.00*10-9 (-) 0.982 Finland Singapore 1994 2000 2005 2010 2013 1.0 0.8 0.6 0.4 0.2 0.0 Un-captured GDP ratio
  81. 81. 81 Institutional factors Finland Singapore References Small income disparity Inequality (GINI index: 2010) Capacity for innovation Capacity for industry innovation Association capacity and collaborative practices University-industry collaboration in R&D Absorptive capacity Firm-level technology absorption Trusting relationship Willingness to delegate authority Generosity (score) Freedom to make life choices (score) Women in working life Women in labor force Government ICT usage Importance of ICT to government vision Government online service Government success in ICT promotion Labor-employer relations Cooperation in labor-employer relations 19 2 2 7 4 0.33 0.52 12 16 7 16 21 45 18 4 13 23 0.19 0.43 76 3 1 4 2 Distribution of Household Income by Source (ILO, 2012) The Global Information Technology Report 2014 (WEF, 2014) The Global Competitiveness Report 2013-2014 (WEF, 2014) The Global Information Technology Report 2014 (WEF, 2014) The Global Competitiveness Report 2013-2014 (WEF, 2014) World Happiness Report (The Earth Institute, Columbia Univ. et al., 2013) idem The Global Competitiveness Report 2013-2014 (WEF, 2014) The Global Information Technology Report 2014 (WEF, 2014) idem idem The Global Competitiveness Report 2013-2014 (WEF, 2014) Table A1.1 Engine and Brake in Finland and Singapore (2013) Figures indicate world rank otherwise indicated. (2) Engine and Brake for Spin-off to the External Link Finland’s advancement in shifting to new co-evolution can be attributed to its powerful engine. While Singapore’s strong government initiatives in accelerating its ICT advancement has played a significant role as an engine for the nation to be a world ICT leader, it resulted in the delay in transferring business initiative which is essential in shifting from traditional co-evolution to new co- evolution.
  82. 82. 82 (3) Strategic Actions for Spin-off Fig. A1.1. Strategic Actions for Spin-off. By transferring strong engine obtained by government to industry, “muscular” economic environment should be created, which enables high competitive home companies (HCs) to exploring global markets leading to increasing returns on foreign investment. This ultimately contributes to increase “muscularity” of HCs which in turn develops “muscular” economic environment, thus constructing a virtuous cycle.
  83. 83. 83 Appendix II. Uber’s Ride-sharing Revolution
  84. 84. 0 10 20 30 40 50 60 Jun-13 Aug-13 Oct-13 Dec-13 Feb-14 Apr-14 Jun-14 Aug-14 Oct-14 Dec-14 Feb-15 Apr-15 Jun-15 Aug-15 AII.1 Uber’s Conspicuous Launch Fig. AII.1. Trends in Uber and Taxi Trips in the US (Jun. 2013 – Sep. 2015). Sources: Taxi: Fig. 3-2; Uber: authors’ estimate based on , where UT: Uber trip, TT: Taxi trip, UD: Uber share (Fig. 2) (See Appendix 1). Trips per day Uber Taxi TT UD UD UT    1 84 UT TT $10 $12 $14 $16 $18 $20 $22 $24 Jun-13 Aug-13 Oct-13 Dec-13 Feb-14 Apr-14 Jun-14 Aug-14 Oct-14 Dec-14 Feb-15 Apr-15 Jun-15 Aug-15 Taxi Uber Average prices per trip PT PU Fig. AII.2. Trends in Uber and Taxi Prices in NYC (Jun. 2013 – Sep. 2015). Sources: Taxi: Fig. 3-3; Uber – Jun. 2013 - Nov. 2014: Lunden (2014): other period: authors ‘estimate based on TLC, Uber, Stone (2015) and Silverstein (2014) (See Appendix 1). 1. Corresponding to its astounding success, Uber’s prices continued to decline and reached in May 2014 the same level as taxis, The prices further declined with the introduction of UberPool in August 2014. 2. This decline in prices was reversed as a consequence of Uber’s surge pricing, which resulted in an “F” (flunk) rating from the Better Business Bureau (BBB) in Oct. 2014. BBB cited complaints over unexpectedly high charges. 3. In response to such complaints and also to competition from competitors such as Lyft, Uber managed to decrease prices by introducing Uber Go in Nov. 2014. 4. This move, together with the establishment of the Uber Advanced Technology Center in Feb. 2015, led to lower prices again in 2015. 2014/5 14/8 14/10 14/11 2015/1
  85. 85. AII.2 Institutional Enablers Creating Platform Ecosystems AII.2.1 Preference Shift to Sharing Economy 1. In line with people’s preferences shift from economic functionality to supra-functionality beyond economic value, sharing economy in physical products (i.e., rooms and cars) has been gaining momentum. 2. The underlining paradigm of the original sharing economy is that users aim at increasing resource-use efficiency. This is done either to lower costs or to create new value. The resources are offered to others at times when the owners do not use them themselves. 3. Online trading platforms such as Napster and eMula were amongst the first to provide users with a shared access to digital music and videos. 4. It was possible to download these digital products from lenders on the platform for free, and uploading and downloading happened simultaneously. This constitutes the very essence of sharing economy (Winterhalter et al., 2015). 5. People’s preference shift to supra-functionality has led to requests for a similar platform also for physical products. People wish to use such products (which were provided passively, primarily for their economic functionality) in a more sophisticated manner and by their own initiative (Adner, 2012). 6. Sharing economy for physical products (initiated by Uber and AirbnB: car transport and rooms, respectively) is needed by the market with such underlining paradigm. 85
  86. 86. PU decrease rate Contribution by Period SP increase rate UT increase rate Miscellaneous -2.58 -0.911 x 1.34 = -1.22 -0.173 x 11.28 = -1.95 0.59 2013/6 - 2014/9 -2.34 -0.737 x 0.61 = -0.45 -0.359 x 11.14 = -4.00 2.11 2014/10 - 2015/9 1. Uber’s prices have been governed by the advancement of smartphones, learning and economy of scale effects. The prices continued to decline before serious complaints over unexpectedly high charges due to surge pricing in Oct. 2014. 2. While this upward shifting factor remains, the price decline trend was maintained by introducing Uber Go in Nov. 2014. 3. This demonstrated high elasticity of trips to prices and compensated the stagnation of smartphones’ share increase in 2015. PU is governed by SP advancement, learning and economy of scale effects. SP: monthly trend in smartphone share of mobile subscriber market (comScore) TU TU USPAP USPAP lnlnlnln      Table AII.2 Contribution of PU decrease (Jun. 2013/6 – Sep. 2015) - % p.a 88 )58.4()97.10()88.3(32.2()71.2()33.5( ) 01.1979.0.164.0ln359.0ln173.0ln737.0ln911.0105.7ln * 2 2121   DWRadjDUDUDSPDSPDP TTU D1:2013.6 – 2014.9 = 1, rest = 0. D2:2014.10 – 2015.9 = 1, rest = 0, D:2014.10 = 1, rest = 0. U U P P Table AII.1 Governing Factors of Uber Prices in the US (Jun. 2013 – Sep. 2015) 2014. 10 PU: Uber’s prices, SP: Smartphone subscriber market share (%), UT: Uber trips, and D1, D2 , D3: Dummy variables. AII.2.2 ICT’s Self-propagating Virtuous Cycle (1) Governing Factors of Uber Prices Decline Uber was given an “F” (flunk) rating from the Better Business Bureau (BBB) in response to complaints over unexpectedly high charges due to surge pricing.
  87. 87. Fig. AII.4. Virtuous Cycle between Uber’s Trips and Its Prices (Jun. 2013– Sep. 2015). Figures in parenthesis indicate t-statistics: all significant at the 1% level except #: 15% level. )16.3()01.18()44.1()34.12()17.68( # 2 2131211 08.1957.0.573.1ln289.0ln270.0ln398.0639.3ln    DWRadjDUTDUTDUTDPU tttt )63.4()45.16()52.3()27.20()09.24( 00.2991.0.887.33ln782.2ln275.9ln055.3979.10ln 2 111 2321    DWRadjDPUDPUDPUDUT tttt Fig. AII.3. Correlation between Uber’s Trips and Their Prices (Jun. 2013 – Sep. 2015). Uber’s trips Uber’s prices D1:2013.6– 2014.6 = 1, rest = 0. D2:2014.7 – 2014.11 = 1, rest = 0. D3:2014.12 – 2015.9 = 1, rest = 0. PU: Uber’s prices, UT: Uber trips D1, D2 , D3 Dummy variables. D1:2013.6– 2014.6 = 1, rest = 0. D2:2014.7 – 2014.11 = 1, rest = 0. D3:2014.12 – 2015.9 = 1, rest = 0. Source: Certify. 87 1. Uber’s prices demonstrated sharp decline as smartphones advanced. 2. This decline induced more Uber trips, which in turn further accelerated the decline of Uber’s prices. 3. Thus, a self-propagating virtuous cycle has been created in Uber’s development as demonstrated in Fig. AII.4. Jun. 2013 Sep. 2015 $10 $12 $14 $16 $18 $20 $22 $24 $26 0 10 20 30 40 50 Uber’sprices/trip Uber’s trips 2014/7 2014/1 1 (2) Virtuous Cycle between Uber Trips Increase and Their Price Decline
  88. 88. $0 $200 $400 $600 $800 $1,000 $1,200 $1,400 $1,600 Jan-04 May-04 Sep-04 Jan-05 May-05 Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09 Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11 Jan-12 May-12 Sep-12 Jan-13 May-13 Sep-13 Jan-14 May-14 Sep-14 Jan-15 May-15 Sep-15 NYC Corporate Medallion Prices (1,000 US$) Actual Logistic Growth May 2011: Uber launched in NYC Mar. 2009 Uber established May 2014 Jun 2013: Medallion prices stagnate Fig. AII.5. Trend in Corporate Medallion Prices and their Estimate without Uber in NYC (Jan. 2004 – Sep. 2015). Source: NYC Taxi and Limousine Commision (TLC). 90 (1) Medallion Prices as a Proxy of the Trend in Taxi Demand AII.3 Co-evolution of 3 Mega-trends Leading to a Spinoff to Sharing Economy AII.3.1 Emergence of Un-captured GDP * While advancement of ICT generally contributes to the enhancement of prices by increasing new functionality development, the dramatic advancement of the Internet tends to decrease ICT prices due to its freebies, easy copying and mass standardization, among other things. Estimated medallion prices without Uber Decline due to Uber Magnitude of un-captured GDP Fig. AII.6. Two-faced Nature of ICT*. Uncaptured GDP
  89. 89. Fig. AII.7. Scheme of the Measurement of the Emergence of Uncaptured GDP in case of Uber in NYC. $0 $200 $400 $600 $800 $1,000 $1,200 $1,400 $1,600 Mar. 2009 Uber established Jun 2013: Medallion prices stagnate May 2011: Uber launched in NYC 2013/6 2015/1 2015/92014/52011/5 PT PT PU UT TT MP Uncaptured GDP per trip* 91 2004/1 2009/5 2011/5 2013/6 2014/5 2015/9 Magnitude of Un-captured GDP )( 1 1 1 1 UT UT T TT UTTT TAT PP PP P UT PUPT PPP            Uncaptured GDP where PT: Taxi prices, PU: Uber prices, PA: Aggregated prices, TT: Taxi trip, UT: Uber trip, : UT/TT ratio. te e MP 02.0 36.61 2247    TT TUTT A UT UPTP P     (2) Magnitude of the Emergence of Un-captured GDP The foregoing observation on the trend in medallion prices inspires the following analogy with respect to discrepancy of prices between taxis and Uber, as illustrated in Fig. 11: Medallion prices Actual Medallion Un-captured GDP without Uber MPe prices MP MPe - MP Taxi prices PT Aggregated prices PA Un-captured GDP PT - PA Based on this, the emergence of un-captured GDP was estimated as follows: Trips (Fig. 4) Prices (Fig. 5) Estimated medallion prices without Uber Aggregated prices *PT vs Mpe and PA vs MP demonstrates significant parallel correlation. (Appendix 3).
  90. 90. (3) Correlation between Uber’s Dependency and Medallion Prices Fig. AII.9. Virtuous Cycle between Uber’s Dependency and Medallion Prices (Jun. 2013 – Sep. 2015). Figures in parenthesis indicate t-statistics: all significant at the 1% level except #: 30% level. )07.5()31.3()22.14()06.1()50.11()43.77( # 2 2131211 77.1948.0.131.0895.1ln420.0ln220.0ln554.0435.8ln    DWRadjDDUDDUDDUDDMP tttt )62.3()88.11()06.3()11.13()22.15( 46.1980.0.770.92ln910.1ln030.11ln051.2022.17ln 2 111 2321    DWRadjDMPDMPDMPDUD tttt Fig. AII.8. Correlation between Uber’s Dependency and Medallion Prices (Jun. 2013.6 – Sep. 2015). Uber’s dependency Medallion prices D1:2013.6 – 2013.12 = 1, rest = 0. D2:2014.1 – 2014.5 = 1, rest = 0. D3:2014.6 – 2015.9 = 1, rest = 0. D: 2013.7-8, 2015.9 = 1, other months = 0. MP: Medallion prices, UD: Uber dependency D1, D2, D3, D: Dummy variables. D1:2013.6 – 2013.12 = 1, rest = 0. D2:2014.1 – 2014.5 = 1, rest = 0. D3:2014.6 – 2015.9 = 1, rest = 0. Medallion prices in NYC, while Uber dependency (% of Uber) is nationwide. Sources: NYC Taxi and Limousine Commission (TLC) and Certify. 90 1. Uber’s astounding success brought its charges below the taxi charges in May 2014. 2. This resulted in a significant decrease in medallion prices. 3. This in turn induced further dependency on Uber, leading to a virtuous cycle between medallion price decline and increase in the dependency, as demonstrated in Fig. . 4. In addition to the foregoing significant parallel correlation between PT vs Mpe and PA vs MP, the correlation among Uber, taxi and medallion demonstrates the significance of the foregoing analogy with respect to the emergence of un-captured GDP and endorses the view that the balance between taxi prices and aggregated prices represents the emergence of un-captured GDP by Uber. Jun. 2013 May. 2014 Sep. 2015 $600 $700 $800 $900 $1,000 $1,100 $1,200 $1,300 $1,400 $1,500 0 10 20 30 40 50 60 70 Medallionprices(1000US$) Uber’s dependency % Jan. 2014
  91. 91. TT TUTT A UT UPTP P    13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0 17.5 18.0 Jun-13 July-13 Aug-13 Sep-13 Oct-13 Nov-13 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Jun-14 July-14 Aug-14 Sep-14 Oct-14 Nov-14 Dec-14 Jan-15 Feb-15 Mar-15 Apr-15 May-15 Jun-15 July-15 Aug-15 Sep-15 PT PA Uncaptured GDP Average prices (US$/trip) Fig. AII.10. Trends in Taxi Prices and Aggregated Prices in NYC (Jun. 2013 – Sep. 2015). Aggregated prices PA are measured by the following equation: 93 AII.3.2 Emergence of Uber-Driven Un-captured GDP (1) Concept of the Emergence of Un-captured GDP High-quality services with lower cost and shorter time. Increasing initiative of passangers and the company’s systematic market strategy of continuous reduction of costs and time in search and matching, eliminating information asymmetries and compiling a massive database. Supported by the foregoing endorsement, Fig. demonstrates the magnitude of un-captured GDP per trip by Uber. The substance of this un-captured GDP can be summed up as follows: Taxi prices Aggregated prices 2014/6 2015/1 Fig. demonstrates that, while Uber nurtured “negative un-captured GDP value” (its services were unable to catch up with those of taxi accumulated over the last 120 years) by June 2014, it succeeded in nurturing increasing un-captured GDP from the beginning of 2015 corresponding to its success in sustainable decline in prices from the end of 2014.
  92. 92. -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Jun-13 July-13 Aug-13 Sep-13 Oct-13 Nov-13 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Jun-14 July-14 Aug-14 Sep-14 Oct-14 Nov-14 Dec-14 Jan-15 Feb-15 Mar-15 Apr-15 May-15 Jun-15 July-15 Aug-15 Sep-15 Fig. AII.11. Trend in the Emergence of Uncaptured GDP by Uber in NYC ( Jun. 2013 – Sep. 2015). ratio T U T T  94 (2) Increase in the Emergence of Uncaptured GDP )( 1 1 1 1 UT UT T TT UTTT TAT PP PP P UT PUPT PPP            Un-captured GDP (US$/trip) Uncaptured GDP where On the basis of the foregoing review, the trend in the value of un-captured GDP per trip by Uber in NYC was measured as illustrated in Fig. This figure demonstrates that Uber-induced un- captured GDP has been increasing significantly from the beginning of 2015. As emulated in the following equation, this can be attributed to a virtuous cycle between prices (PU) decline and trips (UT) increase, as demonstrated in Fig.. 2015/1
  93. 93. AII.3.3 Spin-off to Sharing Economy (1) Uber’s Self-propagating Function 93 Table AII.3 Estimates of Taxis’ and Uber’s Development Trajectories in NYC From equation (3), dynamic carrying capacity can be expressed as follows: (5) This demonstrate that N(t) increases together with the increase of Y(t) and its growth rate as time goes by. This implies that the LGDCC function demonstrates functionality development in the context of the self-propagating behavior (Watanabe et al. (2004).         )(/1 1 )()( )(1 tY tYtN dt tdY b Figures in parenthesis indicate t-statistics: all significant at the 1% level except *: 5 %, **: 15 %, #: non-significant. Table demonstrates that while taxis depend on SLG Uber depends on LGDCC. This demonstrates that Uber has developed with the self- propagating function.
  94. 94. AII.4 Conclusion AII.4.1 Secret of the Success of Uber’s System In light of the disruptive digital-technology-driven business model that Uber has used to trigger a ride-sharing revolution, the institutional sources of the company’s platform ecosystem architecture were analyzed. Aiming at elucidating institutional enablers creating Uber’s platform ecosystem, an empirical analysis of its co- existing development trajectory with taxi was attempted. Noteworthy findings include: (i) This co-existing development trajectory corresponds to two-faced nature of ICT that is behind the emergence of uncaptured GDP, (ii) This emergence can be attributed to a virtuous cycle between prices decline and trips increase, (iii) This virtuous cycle can be attributed to its self-propagating function, and (iv) This self-propagating function plays a vital role in spin-offs from traditional co-evolution to new co-evolution ICT advancement, paradigm change and people’s preference shift. AII.4.2 Noteworthy Elements Essential to Well-Functioning Platform Ecosystem Architecture These findings form the base for the following suggestions supportive of constructing a well-functioning platform ecosystem: (i) Penetrate the current demand and meet its challenge, (sharing economy, saturation of taxi business, popularity of smartphone) (ii) Fully utilize the advancement of ICT, particularly of the Internet, (smartphone, digital payment, big data analysis) (iii) Construct a co-evolution between sophisticated platform ecosystems and consolidation of stakeholders. (mutual rating system among the company, its drivers and their passengers) (iv) Take care of the platform orchestration for efficiency, development and innovation, (Successive innovation for novel services as competitor like Lyft boosting and also as against movement emerging) (v) Thereby, create a novel business model which has never been conceived before. 94
  95. 95. AII.4.3 Implications of Uncaptured GDP The emergence of un-captured GDP in case of Uber can be attributed to (i) People’s preference shift to sharing economy and advancement of ICT, particularly of the Internet and later on smartphones, (ii) Better services, with cost and time saving for passengers, high efficient operation without additional investment and license fees for drivers, and optimal price-setting and market making beyond marginal cost for the company through a massive database on driver and passenger behavior. (iii) Paradigm shift from resources to ecosystem that corresponds to the shift from captured GDP to uncaptured GDP. Thus, Uber’s un-captured GDP can be considered as a consequence of the co-evolution between people’s preference shift, advancement of ICT and this paradigm shift. This co-evolution has been leveraged by Uber to create new business, to create services through interactions between the stakeholders: the company, drivers and passengers. All this can be attributed to systems success: platform ecosystem architecture under the contemporary digital economy. AII.4.4 Criticism to be Solved However, as a consequence of transition to new dynamism, there remains the following areas of criticism: (i) Discrimination (e.g., equivalence of services for remote areas with low population density), (ii) Safety issues, (iii) Treatment of privacy issues, and (iv) Compliance with labor standards. AII.4.5 Future Works This analysis has explored a prototype of the analysis of co-evolution of three mega-trends that nurtures uncaptured GDP. Further analyses applying this approach are expected to be undertaken for similar disruptive business models in (i) music industry, (ii) game industry, (iii) printing and publishing industry, and (iv) education industry. 95
  96. 96. Source: NYC Taxi and Limousine Commission. 0 250 500 750 1,000 1,250 1,500 Jan-04 Oct-04 July-05 Apr-06 Jan-07 Oct-07 July-08 Apr-09 Jan-10 Oct-10 July-11 Apr-12 Jan-13 Oct-13 July-14 Apr-15 N May 2011: Uber launched in NYC Uncaptured GDP Medallion prices (1,000 US$) Two-faced nature of ICT Un-captured GDP Fig. S2. Two-faced Nature of ICT Fig. S3. Anticipating Un-captured GDP. Supplement 1. Two-faced Nature of ICT and Uncaptured GDP S1.1 Two-faced Nature of ICT and Subsequent Uncaptured GDP Fig. S1. Trend in Corporate Medallion Prices in NYC and Contributors (2004-2015). The trend in medallion prices as a consequence of co-existing diffusion trajectory of taxi with prices increase and that of Uber with prices decrease suggests that this trajectory is subject to the two-faced nature of ICT that is behind the emergence of un-captured GDP. 96 $0 $250 $500 $750 $1,000 $1,250 $1,500 Jan-04 May-04 Sep-04 Jan-05 May-05 Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09 Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11 Jan-12 May-12 Sep-12 Jan-13 May-13 Sep-13 Jan-14 May-14 Sep-14 Jan-15 May-15 Sep-15 2011/5 Uber launched in NYC 2009/3 Uber established 2014/5 Start to fall 2013/6 Peak MP Uber Taxi Contributors to medallion prices level Phase 1 Phase 2 Uncaptured GDP MP MP without Uber
  97. 97. S1.2 ICT Prices Trajectory and Two-faced Nature (1) Modified Bi-logistic Growth Ia i Ja j I ij eb N eb N p      11 jiji bbandaa ,, 0 1 ,0 1 ,0 1 )1)(1( 1 1 111 1 )1( 1 )1( 11 1 111 1 1 1 1 1 1 1 1 1 1 2 1 2 )1)(1( 11                                                             ji ii ji jj ji ji ji jiji ji ii ji jj ji jiji i ji i j ji j ji ji ji Ia i ji Ja j ji ji N p I Ia i Ja jji jiI Ia i Ja jji ji Ia i Ja jji Ia i Ja j JaIa ji Ia i Ja j Ia i Ja j Ja j Ia i Ia i Ja jI bb ba bb ba bb bb bb bbbb where IJI bb ba J bb ba bb bbbb Ia bb b Ja bb b bb bb bb eb bb eb bb bb pN N ebebbb bb N p ebebbb bb ebebbb ebeb eebbebeb ebeb ebeb ebeb N p ij I ij ijij ij jiij ij ji ij   JaandIa ji .,.0,., 1 ,,0 .,.1, 1 ),0(,0 jiIj j J Ii i I bbThereforepJWhen N bN b N NpJWhen NpIWhen N b b N pIWhen              (A2) (A3) ICT prices can be depicted by the following modified bi-logistic growth as illustrated in Fig. S4: where I: ICT stock, J: dependency on the Internet, N: carrying capacity, : diffusion velocity of I and J. [1] Since the Internet has been playing a leading role in the whole ICT and providing significant impacts on the diffusion trajectory of ICT, the carrying capacity of logistic growth in I and reverse logistic growth in J as well as their diffusion tempo ( ) were treated as behaving in the similar way (a i I=a jJ). Equation (A1) can be developed as follows: (A1) Fig. S4. Modified Bi-logistic Growth due to Two-faced Nature of ICT. Ia i Ja j I ij eb N eb N p      11 Uber Taxi 97 PI
  98. 98. 98                                    jijii j jj ii ji i ji iji ii i ii ijiijji i j ii jj bbbbb b ab bb bb a bbfornecessaryisas bbb bb b bb bbas I J bb I J bThereforeJaIa b b J I ba ba , 1 ,1)0() 1 ()0() 1 ( 1 .0,1.0 )0( )1(2 )1)(1(4)1()1( )0( )1(2 )1)(1(4)1()1( 0)1()1()1( 1 )1(1 )1(,)( 22 2 2 2  35.1970.0.178.0103.0005.0355.1 2 DWRadjDTU MPN N TT   (-3.12) (-2.96) (8.54) (5.42) Table S1 Co-existing Trajectory of Taxis and Uber in NYC (Jun. 2313 – Sep. 2015) where N (carrying capacity) = 2247,(Table 5) MP: medallion prices, D: 2014. May., Aug., Sep. = 1. This demonstrates that coexistence of taxis and Uber is subject to two-faced nature of ICT. In case of a co-existing diffusion of taxis and Uber, J and I correspond to UT (Uber trips) and TT (taxi trips) and Eq. (A2) can be represented as Table S1. (2) Diffusion Coefficient Coefficients governing modified bi-logistic growth in Eq. (A1) can be identified as follows (here J and I correspond to UT and TT ): (A4) (A5)
  99. 99. 99 0)1( 0)1( )1( 1 ),1( 1                         IJ Ia N p Ia p I I p IJ Ja N p Ja p J J p N p pa I p eb N p N p pa J p eb N p iI i I I i jJ j J J j I Ii I Ia i I I Ij I Ja j I ij     Thus, co-existing trajectory of taxis and Uber as demonstrated in Table A1 can be demonstrated as follows: TT TUI ee P 33.020.0 31.01 2247 03.01 2247      (A6)* * Demonstrating the state in Sep. 2015 when = 0.08. This modified bi-logistic growth demonstrates contributors to medallion prices level illustrated in Fig.A2.  (3) Trip Elasticity to Prices The marginal contribution of Uber and taxis dependency to medallion prices change can be depicted as follows: Thus, the elasticity of Uber and taxi dependency to prices elasticity can be depicted as follows: This demonstrates that, contrary to taxis’ prices increasing as their trips increase, Uber prices decrease as its trips increase leading to a virtuous cycle for Uber. All this support the analysis of institutional sources being behind the emergence of un-captured GDP. (A7) (A8)
  100. 100. $0 $200 $400 $600 $800 $1,000 $1,200 $1,400 $1,600 Jan-04 May-04 Sep-04 Jan-05 May-05 Sep-05 Jan-06 May-06 Sep-06 Jan-07 May-07 Sep-07 Jan-08 May-08 Sep-08 Jan-09 May-09 Sep-09 Jan-10 May-10 Sep-10 Jan-11 May-11 Sep-11 Jan-12 May-12 Sep-12 Jan-13 May-13 Sep-13 Jan-14 May-14 Sep-14 Jan-15 May-15 Sep-15 NYC Corporate Medallion Prices (1000 US$ ) Actual Logistic Growth Parabola May 2011: Uber launched in NYC Mar. 2009 Uber established Uncaptured GDP May 2014 Jun 2013: Medallion Prices stagnate Logistic growth Estimate t-value adj. R2 N 2247.11 7.23 0.976 a 0.02 14.21 b 6.36 7.21 Parabolic growth Estimat e t-value adj. R2 a 288.30 25.80 0.977 b 5.31 11.91 c 0.02 5.42 at be N Y    1 S1.3 Prospect of Uncaptured GDP Nurtured by Uber Fig. S5. Estimate of Uber’s Impact on Medallion Prices Decline (Jan. 2004 - Sep. 2015). Table S2 Estimates of Medallion Prices (Jan. 2004 - Jun. 2013) Y: Medallion prices, N: Carrying capacity, t: Monthly trend, a,b,c: Coefficients 0 500 1000 1500 2000 2500 Jan-04 Feb-05 Mar-06 Apr-07 May-08 Jun-09 July-10 Aug-11 Sep-12 Oct-13 Nov-14 Dec-15 Jan-17 Feb-18 Mar-19 Apr-20 May-21 Jun-22 July-23 Aug-24 Sep-25 Oct-26 Nov-27 Dec-28 Jan-30 Feb-31 Mar-32 NYC Corporate Medallion Prices (1000 US$) Uncaptured GDP 2247 Fig. S6. Estimate of un-captured GDP anticipated by Uber (May. 2014 - May. 2032). 1. As reviewed in Fig. 9, the magnitude of un-captured GDP can be measured by the balance between actual medallion prices and medallion prices without Uber. 2. The former can be estimated by Eq. A6 and the latter by Table A2. 3. Table S2 demonstrates how the trend in medallion prices without Uber can be estimated both by logisti growth and parabolic growth. The latter provides higher estimate. 4. Fig. S6 demonstrates the prospect of un-captured GDP emerging in case of Uber as estimated by the foregoing approach. 100 2 ctbtaY 
  101. 101. Supplement 2 . Correlation between Medallion Prices and Taxi/Uber Prices Figures in parenthesis indicate t-statistics: all significant at the 1% level )81.5()24.5()81.4()38.5( 73.1931.0.247.19ln227.1ln607.53441.3ln 2 121   DWRadjDPADPADMP Fig. S7. Correlation between Taxi/Uber Aggregated Prices (PA) and Medallion Prices (MP) (2014.5– 2015.9). D1:2014.5 – 2014.11 = 1, rest = 0. D2:2014.12 – 2015.9 = 1, rest = 0. MP: Medallion prices, PA: Aggregated prices per trip D1, D2: Dummy variables. D1:2014.5– 2014.8 = 1, rest = 0. D2:2014.9– 2015.1 = 1, rest = 0. D3:2015.2– 2015.9 = 1, rest = 0. MPe: Estimated Medallion prices, PT: Taxi prices per trip D1, D2, D3 : Dummy variables. 101 May. 2014 Sep. 2014 Feb. 2015 Sep. 2015 $1,200 $1,250 $1,300 $1,350 $1,400 $1,450 $1,500 $15.6 $15.8 $16.0 $16.2 $16.4 $16.6 $16.8 MedallianpriceswithoutUber (1000US$) Taxi Prices (PT) May. 2014 Nov. 2014 Sep. 2015 $600 $700 $800 $900 $1,000 $1,100 $1,200 $1,300 $1,400 $1,500 $13.0 $14.0 $15.0 $16.0 $17.0 Medallionprices(1000US$) Aggregated price (US $) )84.3()90.3()84.2()80.3()88.4( 26.1945.0.813.10ln144.1ln731.2ln127.1018.4ln 2 2321   DWRadjDPTDPTDPTDMPe Fig. S8. Correlation between Taxi Prices (PT) and Medallion Prices without Uber (Mpe) (2014.5 – 2015.9) . Table S3 Correlation between Taxi/Uber Prices and Medallion Prices (2014.5 – 2015.9) PT vs Mpe and PA vs MP demonstrates significant parallel correlation as far as 2015 is concerned. This supports the significance of un- captured GDP measurement depending on the balance between PT and PA during the above period.
  102. 102. 102 0.00 10.00 20.00 30.00 40.00 50.00 60.00 1 2 3 4 5 6 7 8 9 10111213141516171819202122232425262728 UT 1 3.12 2.75 3.37 3.09 3.62 3.47 3.96 3.9 4.34 4.38 4.82 4.92 5.38 5.51 5.96 6.18 6.65 6.92 6.18 7.74 7.19 8.66 9.96 9.67 12.40 10.79 12.80 12.03 14.92 13.39 14.56 14.89 15.86 16.52 20.46 18.31 20.40 20.25 26.18 22.25 25.06 24.62 26.83 27.05 30.96 29.66 33.95 32.43 37.27 35.36 40.94 38.44 45.03 41.67 49.64 45.03 1 Jun-13 2 July-13 3 Aug-13 4 Sep-13 5 Oct-13 6 Nov-13 7 Dec-13 8 Jan-14 9 Feb-14 10 Mar-14 11 Apr-14 12 May-14 13 Jun-14 14 July-14 15 Aug-14 16 Sep-14 17 Oct-14 18 Nov-14 19 Dec-14 20 Jan-15 21 Feb-15 22 Mar-15 23 Apr-15 24 May-15 25 Jun-15 26 July-15 27 Aug-15 28 Sep-15 2TT UU Period Trips per day Trips per day 2013/6 2014/1 2014/5 2015/3 2015/9 2T T U U Supplement 3. Sensitivity of Uber Trips Estimate S3.1 Estimate without and with Spline Interpolation Fig. S9. Comparison of Uber Trips Estimate (Jun. 2013 – Sep. 2015). Table S4 Comparison of Uber Trips Estimate (Jun. 2013 – Sep. 2015). UT: Uber trips estimated by taxi trips and Uber dependency (Appendix 1) UT2: Uber trips estimated with spline interpolation In analyzing Uber’s diffusion trajectory (3.3 (1) and (2)), given the sensitive impacts of fluctuation on the trajectory formation within the limited samples, comparative analysis was attempted by comparing Uber trips estimate with and without spline interpolation, as shown in Fig. S9 and Table S4. The function used for spline interpolation was based on the logistic growth function .
  103. 103. 13.5 14.0 14.5 15.0 15.5 16.0 16.5 17.0 17.5 18.0 Jun-13 Aug-13 Oct-13 Dec-13 Feb-14 Apr-14 Jun-14 Aug-14 Oct-14 Dec-14 Feb-15 Apr-15 Jun-15 Aug-15 TT TUTT A UT UPTP P    Uncaptured GDP Average prices (US$/trip) Fig. S10. Trends in Taxi Prices and Aggregated Prices in NYC (Jun. 2013 – Sep. 2015). Aggregated prices PA are measured by the following equation: 105 S3.2 Effects of the Estimates of Uber-Driven Un-captured GDP (1) Emergence of Un-captured GDP in case of Uber Taxi prices Aggregated prices 2014/6 2015/1 PT PA2 PA With spline interpolation Without spline interpolation No substantial differences between the interpolations with and without spline.
  104. 104. -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 Jun-13 July-13 Aug-13 Sep-13 Oct-13 Nov-13 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Jun-14 July-14 Aug-14 Sep-14 Oct-14 Nov-14 Dec-14 Jan-15 Feb-15 Mar-15 Apr-15 May-15 Jun-15 July-15 Aug-15 Sep-15 Fig. S11. Trend in the Emergence of Un-captured GDP in case of Uber in NYC ( Jun. 2013 – Sep. 2015). ratio T U T T  106 (2) Increase in the Emergence of Uncaptured GDP in case of Uber )( 1 1 1 1 UT UT T TT UTTT TAT PP PP P UT PUPT PPP            Un-captured GDP (US$/trip) Uncaptured GDP where Un-captured GDP Un-captured GDP2 without spline interpolation with spline interpolation No substantial differences between the interpolations with and without spline.
  105. 105. S3.3 Effects of Uber’s Development Trajectory Estimate 105 Table S5 Estimates of Taxi and Uber’s Development Trajectories in NYC Taxi: based on medallion prices (Fig. 9), Uber: based on trips (Fig. 4) with spline interpolation in case 2, without spline interpolation in case 1. Figures in parenthesis indicate t-statistics: all significant at the 1% level except *: 5 %, **: 15 %, ***: 20%, #: non-significant. While Uber’s development trajectory estimate using the trips trend without spline interpolation demonstrates a slight possibility of self-propagation by LGDCC, the estimate with spline interpolation demonstrates explicit self- propagation and the significance of LGDCC. This difference does not have any significant effects on aggregated prices and un-captured GDP estimates. The effects on self-propagation can be attributed to a slightly higher pace (1-9%) of trips after March 2015. This suggests that an optimal and not too rapid development pace seems essential for incorporating the self-propagating function.
  106. 106. 106 Fig. S12. Correlation between Centralization of Wage Setting and Union and CBA Density in 19 Countries in the Late 1990s. CBA: Collective bargaining agreements. Union and CBA density = (Union density + CBA coverage)/2 Source: Warner (2002) [22] Appendix III. Institutional Elasticity against Uber
  107. 107. 107 1 Success (1) Singapore [Legality is pending but operating actively] 1. Taxi drivers and passengers in Singapore are generally welcoming taxi app services. 2. This has led to a highly competitive taxi app market in Singapore, and existing taxi companies as ComfortDelgro and Trans-Cab endeavoured to improved their services by introducing their own mobile app services. 3. COE (Certificate of Entitlement) scheme based on the tripartism framework (consists of Ministry of Manpower, National Trades Union Congress, and Singapore National Employers Federation) plays a decisive role in Singaporean’s efficient utilization of ride-sharing. 4. Requirements and complaints can be solved through dialogues with the regulators, employers and employees (drivers) under the tripartism framework. 5. Uber induced incorporating users (passengers) requirements into the tripartism framework by stimulating better services, thereby consolidation of all stakeholders: company, employee, user and government was constructed. 6. Government agile reaction to complains from incumbent through open dialogue with all stakeholders by acknowledging new stream of innovation, not resisting played a key role. 7. The government is secretly* welcoming the taxi app services because (i) young people enjoy using services like Uber, and the government must not resist innovation, (ii) they provide job opportunity to Singapore citizen (toward aging society) and increase the overall productivity, (iii) the ride- sharing can be an approach to tackle problems of traffic clog and achieve efficient road usage. * Transport Minister urged that “we must always be fair to players, whether incumbent or insurgents, and strike a balanced approach.
  108. 108. 108 (2) Tokyo [Seems illegal but operating] 1. Uber has had tremendous difficulties in making inroads into the Japanese market due to “Byzantine” and complicated regulations. 2. Uber was ordered to suspend its pilot project in Fukuoka city in Feb. 2015 because it violate the laws. Uber stopped the project in Mar. 2015 3. Tokyo has a rather tranquil market so far due to its qualified service seeking competitive market with 50,000 taxi (20% of the total in Japan and 4 times the number in NYC). Nov. 2013 Uber started in Tokyo (limited launch. Expanded whole Tokyo area from Aug. 2014). Jan. 2014 Tokyo Hire-Taxi Association also introduced a mobile app service. Jan. 2015 Japan’s largest taxi company, Nihon Kotsu launched a mobile app Line Taxi. Mar. 2015 Japanese government stated Uber probably violates laws (unlicensed, safety). Uber reacted continued to talk. Mar. 2015 Japan’s e-commerce giant Rakuten entered the ride-sharing industry by purchasing 11.9% in Lyft. Oct. 2015 Prime Minister Abe instructed relaxing regulation for ride-share in isolated areas. 4. Although the legal framework in Japan does not allow private cars or ordinary person to operate as a paid taxi, taxi companies in Tokyo recognized Uber as a business competitor and worked towards improving their services by developing new functions. 5. With CCSD (government, broader industries involvement for social demand (traffic, aging, isolated rural)) co- evolution emerged between IDBM (existing taxi companies also improved their services by introducing their own mobile app services) and advancement of institutional systems by solving social demand.
  109. 109. 109 1. Black taxis have been the kings of the British capital's roads for over a century but now they are battling a high-technology rival that threatens their dominance. Uber is active in three cities (London, Manchester and Leeds) in the UK. 2. Uber has won a significant legal victory in the UK, with London's high court ruling that Uber’s app does not constitute a taximeter. 3. The legal challenge was brought by London's transport agency Transport for London (TfL), following pressure from the city's black cab and taxi drivers. 4. While taximeters devices which record distance travelled and are used to calculate fares are only allowed for licensed taxis, the judge ruled that the legal definition of a taximeter doesn't include "smart phones which rely on data from a server outside the vehicle." 5. Uber hailed the decision as a "victory for common sense," adding that the ruling means the company won't have to change how its app works in London. 6. London Mayor reported that “The technological innovation should not banned unnecessarily that will serve a good purpose to the Londoners”. This really showed the positive impact of the service in the country. He also added that some solution needs to be sorted out that the growth of Uber services does not impact the traditional black taxi drivers anyway. 7. London's Licensed Taxi Drivers Association, described the outcome as unbelievable. The transport authority has also asked the court to determine if the service is in fact legal. 7. Notwithstanding the above victory, Uber still faces ongoing legal challenges in London, including proposals to introduce compulsory five-minute wait times and the removal of car icons from the map in the Uber app. (3) London [Legality is pending but operating with expectation]
  110. 110. 110 (4) USA [Generally Positive] Uber is operating in 75% of US locations although banned in Nevada and Oregon, and there was multiple on-going lawsuits. State legislators in Ohio and Florida are moving ahead with regulations governing Uber and other ride services that would designate all drivers as independent contractors, bolstering a critical but much- disputed aspect of Uber's business model.
  111. 111. 111 (5) Saudi Arabia 1. Saudi Arabia's discriminatory automotive policies against women have allowed Uber to achieve great success, due to females having limited options in transportation. 2. Women are not allowed to drive as it is feared to damage their ovaries leading to children born with clinical problems. 3. Since women cannot keep their jobs in Saudi Arabia because they have trouble in finding reliable transportation to get to work, Uber triggered institutional revolution for women’s social participation as demonstrated by the fact that women make up 70% of Uber's customers. 4. With such expectation, Uber operates in the holy Islamic cities of Mecca and Medina, as well as the capital city of Riyadh, and the port cities of Jeddah and Dammam. The service is expected to be available in several more cities in the near future. 5. While there remains the issue of compliance with traditional government regulations, negotiation with the institutional regulators in Saudi Arabia have been extremely positive compared to other countries reception towards the app business. 6. Thus, Uber is expected to grow 50-60 % in trips per months in Saudi Arabia in 2016, which in turn accelerates social innovation in the country leading to a co-evolution between ICT-driven disruptive innovation and change in institutional systems triggered by women’s social participation. [Legality is pending but operating actively]
  112. 112. 112 (6) Russia [No ban, but difficult to offer service] 1. Regulations in Russia are comparatively simple in comparison to other countries. 2. Moscow has already a culture of unlicensed taxis that makes Uber’s expansion there difficult. 3. Citizens can often hail one by standing on the street corner or via a number of apps that have existed for years before Uber arrived. 4. Since 2011, Russia’s main search engine company, Yandex, has been running an taxi-app that most now simply known as Russia’s Uber. 5. Uber also trails Gett, known as the Uber of Israel, which operates 10,000 cars in Moscow.
  113. 113. 113 (7) Canada [Changing to Support] 1. Uber drivers in Canada are required to register, collect and remit HST/GST from their fares to the government, regardless of their income. 2. In December 2012, officials in the city of Toronto charged Uber with 25 municipal licensing infractions. A Toronto city councilor has warned that passengers using UberX may be fined up to $20,000. 3. Uber was made legal in the city of Edmonton by passing by-law. However, Uber ceased its operations in Edmonton in March 2016 citing inability to obtain the necessary insurance. The City of Calgary, Alberta has charged at least 17 drivers illegally driving for Uber. 4. These drivers were operating without legally mandated insurance. Uber continues to operate illegally in the other regions of Canada. 5. Toronto Mayor expressed his support for Uber in 2014 and other cities are slowly beginning to look at regulatory options.
  114. 114. 114 (8) Philippines [Developed Nationwide Regulations Making Legal] 1. The Philippines became the first country to develop nationwide ride-hailing regulations, making it legal for app-based transportation services like Uber to operate anywhere in the nation in Nov. 2015. 2. "We view technological innovation as a driver for progress, especially in transportation where it can provide safer and more convenient commuting options to the public,” Jun Abaya, the Philippines’ Department of Transportation and Communications secretary, said. “App-based transport services help address the increasing demand for mobility spurred by rapid urbanization.” 3. All the specifics of the regulations have yet to be finalized, but in general, the DOTC says cars that operate on these services must have a GPS system; must be sedans, Asian Utility Vehicles (AUVs), SUVs, or vans; and can’t be more than seven years old. Operators will also be required to obtain certificates for each vehicle on the service, and drivers must be screened and accredited by Uber (or other ride-hailing services) and registered with the local transportation regulatory board. 4. But Uber still faces challenges unique to the Phillipines. For one, Uber’s routing algorithm doesn’t work as well in Manila, which has some of the world’s worst traffic. And as one writer in the Philippines points out, Uber has been operating in the country a bit like a condo rental service; operators are buying small fleets of brand new cars and hiring individual drivers - essentially layering a new middleman on top of Uber itself. As the incentives Uber has put into place to spur growth are being phased out, drivers’ salaries are apparently taking a hit so that these fleet owners can break even. 5. Many locals do say that the service is often cheaper and more convenient than local cab services. But Uber drivers, regulators and the company itself still have work to do to find the right fit if Uber expects to keep growing in the Philippines.
  115. 115. 115 (9) China [Generally positive] Although there was some raids and fines against Uber in some of locations, there is no official and legal banning to Uber services. Major challenge Uber facing in China is from the market rivals and competitors. Such lack of “legal banning” can be attributed to (i) lack of strict regulations in transportation market, (ii) lack of legal protection to the taxi drivers, and (iii) current lousy services provided by taxi industry.
  116. 116. 116 (1) France [Partial ban as illegal] 1. France government initially started to suppress the service with their policy and later started allowing the Uber services in certain case, not all the services. 2. UberX is the low cost service that allows only the licensed drivers to operate the cabs and UberPop is also the service but allows even the drivers without the driving license to operate the cab. 3. Government allowed the former but not the latter stating that it would severely affect the regular taxi drivers drastically. 4. Uber did not accept the decision and filed against government which led to huge violent protests by taxi drivers. Finally, Uber has suspended their UberPop services until hearing the final judicial result. 5. Uber announced that it will re-launch its Pop services if the government considers Uber to be legal. But that seems to be highly unlikely. 6. Currently, UberPop is banned from functioning in France. Uber was facing equal protests from traditional Taxi drivers stating that it is not a fair competition as the taxi drivers are exempted from the taxes paid by them. 2 Failure

×