The seminar will present the data sets, methodology and key findings of two AUEB funded research projects on the new entrepreneurial paradigm and the startup ecosystem that emerged during the recent economic crisis in Greece.
The new enterprise model and the startup ecosystem during the crisis in Greece
1. The new enterprise model and the startup ecosystem
during the crisis in Greece
Ioanna Sapfo Pepelasisa , Aimilia Protogeroub Ioannis Besisc,
Spiros Paraskevasd
a Professor Emerita, Department of Economics, AUEB, ipepelasis@aueb.gr
b Researcher, Laboratory of Industrial and Energy Economics, NTUA, protoger@gmail.com
c Business Analyst, Athens Stock Exchange, ioannisbesis1@gmail.com
d Data Scientist, SPhears AI, spirosparaskevas@yahoo.gr
10 March 2021
2. Part 1
Ioanna Sapfo Pepelasis and Aimilia Protogerou (2018) , “A break with the past? The Shift from inward
looking to internationally competitive and born global firms.” , Managerial and Decision Economics.
Part 2
Part 2.1.
Ioannis Besis and Ioanna Sapfo Pepelasis (2020), “Incubated early stage startuppers and their initiatives in
Athens during the Crisis (2010‐2016)” Working Paper Series 01-2020, Department of Economics, AUEB.
Part 2.2
Ioannis Besis, Ioanna Sapfo Pepelasis & Spiros Paraskevas (forthcoming 2021)
In Progress
Part 3
Next Steps
General Introduction - Project Updates
3. Basic Research Questions
1.
International Literature
• Alexander (1964)
• Cassis & Pepelasis Minoglou (2004)
• Sifneos (2011)
• Bitros & Pepelasis Minoglou (2007)
• Hassid & Karayanis (1999)
• Makridakis (1997)
• Spanos (2001)
• Protogerou, Caloghirou & Lioukas (2015)
• Protogerou (2015)
• Notta & Vlachvei (2014)
• Doukidis (2015)
• Malerba & MacKelvey (2016)
Pepelasis, Protogerou (2018) - Introduction
Part 1 Part 2.1. Part 2.2. Part 3
4. Figure 1: GDP growth in Greece (1960-2014) in
comparison to the European South and the EU
average (Source: World Bank)
Figures
0
50
100
150
200
250
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Mining Construction
Manufacturing
Figure 2: Index of Greek industrial
production: mining, construction and
manufacturing (Source: ICAP HELLAS)
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
Industry Services
Figure 3: Total number of industrial
and service firms in Greece (2004-
2014) (Source: ICAP HELLAS)
Part 1 Part 2.1. Part 2.2. Part 3
5. The continuities as understood from our
personal research and that of others in the
existing literature are a mixture of the
prevalence of:
• micro-business (MB)
• family capitalism (FC)
• introversion (I)
• unproductive entrepreneurship (UE)
• low-level entrepreneurship (LLE)
• shallow entrepreneurship (SE)
• rent-seeking entrepreneurship (RSE).
Having presented above the features of
the dominant entrepreneurship paradigm,
we should acknowledge that there have
been two positive exceptions.
The Primacy of Inward-looking orientation and Rent-seeking Entrepreneurship
The traditional model of Greek entrepreneurship has displayed certain constant features, regardless of time period,
some of which have been seen as an indication of Greece’s differentness from Western Europe.
The one is intercontinental shipping,
in which Greek-owned ships have
been the number one global fleet for
many decades now.
The other is the few industrial pioneers of the past
who, on the basis of innovation and extrovert
entrepreneurial strategies, created Greek-based
multinationals. Three important cases are those of:
• Fillipou Brothers
• Aristovoulos Petzetakis
• Karelia Fine Tobaccos
Part 1 Part 2.1. Part 2.2. Part 3
6. Our sample consists of 21 “successful firms,” in
both manufacturing and services that had rising
sales during the crisis and are export oriented.
Exports varied from around 50% to 100% of sales
and were usually nearer the latter figure.
In fact, some of these firms were actually born
global i.e. they operate internationally right from
the start.
We take into account 5 features and search for
which one(s) accounted for success?
Sample and Methodology
Part 1 Part 2.1. Part 2.2. Part 3
7. Introversion Extroversion
New Entrepreneurial Ecosystem
Today.. A break with the Past?
High (technical) level
of education
Born Global
Use of innovations-
technologies in both
low and high-tech
industries
B2B
Draw elements from
the Greek tradition
Self-financing
Object: A study of 21 success stories (manufacturing and creative industries)
Part 1 Part 2.1. Part 2.2. Part 3
8. • Our sample of firms contains a body of
over 40 entrepreneurs.
• Communalities.
The protagonist entrepreneurs
• First, almost all are Greek
in origin, and the few
foreigners are present
basically in the team‐
based firms.
• Moreover, they are
mostly male who have
founded their firms in
their 30s and 40s or even
younger.
• Females appear
sporadically.
Second, entrepreneurs are
well educated (some have
studied and/ or worked
abroad), internationalized in
outlook, and surround
themselves with highly
qualified managerial
personnel.
Most importantly, we find
entrepreneurs who have
degrees in engineering or have
had other forms of technical
education, even in
low‐technology sectors such
as fashion and food.
Third, the vast body of entrepreneurs has an important
background in terms of financial and/or human capital.
• The firms in our sample are the creations of amazing
personalities: Greeks who dared to risk and who were
flexible, perhaps unwittingly resembling somewhat
the example of Greek diaspora merchants two
centuries ago.
• The new innovative and extrovert paradigm,
embodying characteristics of high‐level and
Baumolian productive entrepreneurship, does not
share the characteristics of the mainstream
traditional way of enterprising described in Section 4,
except one: The presence of family in business
although in a different context.
In a nutshell, over a wide spectrum of activities (ranging
from medicine to gaming) the protagonist entrepreneurs
have created global cutting‐edge products by combining
in a masterful way diverse elements such as high
technological expertise and managerial skills, in many
cases coupled with knowledge and practices embodied in
Greek cultural heritage.
Part 1 Part 2.1. Part 2.2. Part 3
9. THE NEW ENTREPRENEURIAL PARADIGM: WHY NOW? It is not easy to answer “Why now and not before?”
• Changing environmental conditions due to the crisis—and also to the maturation of other long‐term trends
unrelated (at least directly) to the crisis.
• Post‐2008, the following three “crowding out” conditions specifically no longer hold true:
The New Innovative and Extrovert Paradigm
Part 1 Part 2.1. Part 2.2. Part 3
10. • This paper illustrates the emerging extrovert ecosystem of re-industrialization during the crisis years in Greece
in an environment of slower growth in the global economy. It is important to underline the existence within the
new ecosystem of companies founded pre- and post-2008, both in high- and low-tech sectors.
• The new ecosystem constitutes a break with the past, but we cannot fail to note that there are certain persisting
continuities, these being the positive side of family involvement in business and traditional handicrafts
intermingling with modern technology (innovation), especially in the organization of production, marketing and
branding. Inspiration is drawn from the specificities of Greek agricultural and pastoral traditions and Greek
civilization and culture. There is a vision of “Greekness”, a Greek way of life and philosophy to be sold to the
world, drawing upon the beauty of the country and the fact that Western civilization, democracy and
philosophy were born in Greece. Suddenly “made in Greece” has become a brand name, a value added
General Conclusions
Part 1 Part 2.1. Part 2.2. Part 3
12. Startup Ecosystem during the Crisis
Part 2.1.
Part 2.2.
Ioannis Besis, Ioanna Sapfo Pepelasis & Spiros Paraskevas (forthcoming 2021)
In the above dataset (300 startups), a combination of a feature selection algorithm and
three different modeling families was tested out in order to predict the characteristics of
startups that survived in 2020.
Ioannis Besis and Ioanna Sapfo Pepelasis (2020)
This research was constructed using a unique dataset of sixteen key socio-
economic indicators for initially 255 (now 300) incubated early stage
initiatives/startups and their founders in Athens throughout the crucial years
of 2010-2016. The project maps this ecosystem, examines its drivers and
makes preliminary predictions using descriptive statistics methodology.
Part 1 Part 2.1. Part 2.2. Part 3
13. Besis, Pepelasis (2020) – International Literature
• Albort-Morant, G. & Ribeiro-Soriano, D.
(2015)
• Antoniades, V., Giakoumelos, M., Petkakis, T.
& Zacharia, Z. (2018)
• Apergis, N., & Fafaliou, I. (2014)
• Åstebro, T., & Bernhardt, I. (2003)
• Bakouros, Y. L., Mardas, D. C., & Varsakelis, N.
C. (2002)
• Bosma, N. & Kelley, D. (2019)
• Burgel, O., & Murray, G. C. (2000)
• Cassis, Y. and Pepelasis Minoglou, I. (2006)
• Cavusgil, S. T., & Knight, G. (2015)
• Cefis, E., Marsili, O., (2003)
• Coeurderoy, R., Cowling, M., Licht, G., Murray
G., (2012)
• Cook, R., Campbell, D., Kelly, C., (2012)
• Enterprise Greece (2019)
• Found.ation, EIT Digital & Velocity Partners.
(2019)
• Tsakanikas, A., Giotopoulos, I., Valavanioti, E.
& Stavraki, S. (2019)
• Kanellos S. N. (2013)
• Lambropoulos, S., (2015)
• Max Kuhn; Kjell Johnson (2013)
• Mian, S. A. (1996)
• Nicolò D., Nania I., (2017)
• Pepelasis, I.S., Besis, I., Bournakis, I. and
Papanastassiou, M. (2019)
• Pepelasis, I. S., & Protogerou, A. (2018)
• Protogerou, A., Caloghirou, Y., & Markou, F.
(2015)
• Ratinho, T. and Mitsopoulos, M, (2017)
• Sofouli, E., & Vonortas, N. S. (2007)
• Tortella, G. & Quiroga, G. (eds) (2013)
• Vlachopoulou, M., Ziakis, C. & Petridis, K.
(2017)
• Vliamos, S. J., & Tzeremes, N. G. (2012)
Part 1 Part 2.1. Part 2.2. Part 3
14. This paper is based on a unique database for Greece and it is the first to chart the features of
start-up(per)s that have been incubated and it consists of information on:
The Startup Ecosystem in the Crisis
443
Individual
Entrepreneurs
255
Incubated Young
Enterprises
These were crucial years as the economic crisis was deepening. Our purpose is to provide insights into the socio-
economic features of start-uppers and explain/discover what factor(s) determine(s) survival among new ventures /
startups in Greece. We focused on the initial stage - i.e., nascent ventures and we measure success by checking
longevity and survival.
Part 1 Part 2.1. Part 2.2. Part 3
15. This paper is based on a unique database for Greece and it is the first to chart in detail the features of start-up(per)s and their
‘enterprises’ that have been incubated We have drawn our material from the following seven incubators, the largest in
Athens (Greece) at the time :
Data & Methods
Egg
ACEin
MIT EF
Greece
Orange
Grove
Iqbility
Ekinisi
Lab
NBG
Seeds
For our analysis we have relied on written material and videos from the following supplementary sources :
1. The web sites of incubators
2. The web sites of start-ups
3. Social media sites (LinkedIn, Facebook) and
4. Articles/interviews in the press on start-uppers and their enterprises. Also, we have relied on answers of
start-upper ‘incubees’ to our questionnaire (the total number of questionnaires answered is twenty
two); .
The data set covers 443 individual entrepreneurs/founders and 255 incubated young enterprises for the years
2010-2016.
Part 1 Part 2.1. Part 2.2. Part 3
16. Socio-Economic Indicators
The first group consists of seven
startupper/founder specific indicators:
1. Age
2. Gender
3. Level of Education
4. Field(s) of education
5. Variety in skills
6. Whether the company sector related to
founder’s education
7. Whether the founder has experience
abroad
The second group consists of nine
startup/firm specific indicators:
1. Geographical Location
2. Business Sector
3. High Tech vs Low Tech in terms of Sector
4. High Tech vs Low Tech in terms of
process of production
5. Whether the good offered is a Physical
Product or Service
6. Whether the good offered is b2c or b2b
7. Whether the startup has customers
abroad
8. Number of founders per startup
9. Whether among the founders there exist
relatives
Part 1 Part 2.1. Part 2.2. Part 3
18. Entrepreneurs Characteristics (II) – Descriptive Statistics
Part 1 Part 2.1. Part 2.2. Part 3
26,02%
73,98%
Variety in skills
(digression between undergraduate-graduate
studies)
Yes No
46,95%
53,05%
Experience Abroad
Yes No
50,40%
49,60%
Company Sector related to founder's
education
Yes No
21. Startup Characteristics (III) – Descriptive Statistics
Part 1 Part 2.1. Part 2.2. Part 3
18,91%
81,09%
High Tech vs Low Tech
High Low
23,83%
76,17%
High Tech vs Low Tech
(in Terms of Process)
High Low
22. Conclusions – The profile of the incubee
Startupper
The startupper was predominantly
under thirty, male; and with
studies (in order of importance) in
the following fields: engineering;
business and economics; and
computer science.
Startup
The initiative/ startup was
predominantly established in Athens,
but it was also the case that a little
over one in ten was based abroad.
The top sectors of startups in order
of importance were: hardware-
software; the creative industries;
agriculture and leisure.
Part 1 Part 2.1. Part 2.2. Part 3
24. Univariate Analysis
Survival
Pearson's Chi-squared test (X-squared = 15.898, df = 8,
p-value = 0.04386)
Internationalization
Pearson's Chi-squared test with Yates' continuity
correction (X-squared = 17.41, df = 1, p-value = 3.013e-05)
Part 1 Part 2.1. Part 2.2. Part 3
25. Model Mean Accuracy
(10 fold Cross
Validation)
Lift from Naive
Model
# of
Independent
Variables
Naive Model* 50% 0 0
GLM + MRMR 64.41% 28.82% 9
RF + MRMR 70.09% 40.18% 27
CART + MRMR 63.35% 26.7% 8
• Founding members average educational level
• Founding members Educational Level Variance
• Founding members mixed gender
• Having achieved customers abroad
• Lack of information on headquarters existence
• Postgraduate studies on economics
• Founding members experience abroad
• Business sector in Health/Medicine, Tourism,
Creative Industries, Agriculture
• High tech processes internally
• Studies in Engineering & Computer Science
Startup Survival (log odds) =
0.097 +0.53*High_Low_Tech_Process(=Yes)
- 0.69 * Education_Level_Variance (among founding members)
+ 0.63 * Education_Digression (=Yes)
+ 1.10 * Customer_Type_B2B_B2C
+ 0.91 * Customers_Abroad (=Yes)
• Having achieved customers abroad
• Applying high tech processes internally
• Founding Members digressing from original studies
• Founding members average educational level
• Educational level variance among founding members
Startups Survival Modeling Results
In Progress
Logistic Regression + MRMR
Random Forest + MRMR
Decision Trees + MRMR
Part 1 Part 2.1. Part 2.2. Part 3
26. Surviving Startups Profiles
1. Achieved customers abroad & whose founding members
attained highest education level. 4% of total population
with 82% survival probability.
2. Achieved customers abroad & whose founding members
have similar education levels. 12% of total population with
76% survival probability
3. Achieved customers abroad, whose founding members
lower education level is compensated by applying high
tech processes internally. 11% of the total population with
65% survival probability.
Non Surviving Startups Profiles
1. No customers abroad, whose founding members
educational level varied and did not digress from original
studies. 38% of total population with 24% survival
probability.
2. No customers abroad and low tech processes applied
internally. 15% of total population with 42% survival
probability.
3. No customers abroad, lower than average educational
level and low tech internal processes. 9% of the total
population with 38% survival probability.
Startups Survival Modeling Insights
Part 1 Part 2.1. Part 2.2. Part 3
27. In Progress
Logistic Regression + MRMR
Random Forest + MRMR
Decision Trees + MRMR
Startups Extroversion Modeling Results
Model Mean Accuracy
(10 fold Cross
Validation)
Lift from Naive
Model
# of Independent
Variables
Naive Model 50% 0 0
GLM + MRMR 66.71% 33.42% 11
RF + MRMR 72.73% 45.46% 29
CART + MRMR 60.67% 21.34% 10
Startup Customers Abroad (log odds) =
- 0.47 + 0.89 * High_Tech (=Yes)
- 1.11 * Engineering_Studies (=Yes)
+ 1.13 * Experience_Abroad
• Founding members average educational level
• Founding members Educational Level Variance
• Founding members mixed gender
• Having achieved customers abroad
• Lack of information on headquarters existence
• Postgraduate studies on economics
• Founding members experience abroad
• Business sector in Health/Medicine, Tourism,
Creative Industries, Agriculture
• High-tech processes internally
• Studies in Engineering & Computer Science
• Having members with experience abroad,
• Studies in non engineering domain,
• Having digressed from original studies &
• Engaged in High-Tech business.
• Founding members with Engineering studies.
Understood as a revealing variable of business domain
Part 1 Part 2.1. Part 2.2. Part 3
28. Startups Extroversion Modeling Insights
Extrovert Startups Profiles
1. Founding members experience abroad, non engineering studies (in
the sense of non applicable business type). 8% of the total population
with 78% probability.
2. Founding members experience abroad, non engineering studies (in
the sense of non applicable business type) & digression from original
studies. 8% of the total population with 68% probability of
extroversion. Decrease in extroversion probability from previous
profile comes from the uncertainty wrt to the type of offering services
regarding the customer.
3. Founding members with experience abroad, non engineering studies
(in the sense of non applicable business type) have digressed from
original studies and doing business engaged in High Tech domains.
14% of total population ~ 60% extroversion probability.
Introvert Startups Profiles:
1. Founding members with no experience abroad and decreased
educational level. 59% of total population with 25% extroversion
probability.
2. An interesting profile: founding members with experience abroad,
engaged in Low Tech business domains, & despite displaying higher
education levels have not digressed from original studies. 6% of
overall population, with 15% extroversion probability.
Part 1 Part 2.1. Part 2.2. Part 3
30. Final Conclusions and Next Steps
Still trying to integrate the
descriptive statistics with the
Machine Learning Methods
Few differences in the results of
2 databases (255, 300)
There is no a great differences in
the results between the 2
survival dates (2018, 2020)
2021: A new Era
Part 1 Part 2.1. Part 2.2. Part 3