Entry Determinants of the Software and Video game firms in Barcelona - - PowerPoint PPT Presentation

Place the ‘Candy’ and ‘Crush’ it: Entry Determinants of the Software and Video game firms in Barcelona

Carles Méndez-Ortega – carles.mendez@urv.cat Universitat Rovira i Virgili (CREIP-QURE) Seminario ECONRES March 29th, 2019 Facultad de Ciencias Económicas y Empresariales (UAM)

Outline:

• 1. Introduction
• 2. Data and Area of Study
• 3. Methodology
• 4. Results
• 5. Robustness Analysis
• 6. Conclusion

Méndez-Ortega, Carles ECONRES SEMINARS (UAM) 2/37

Introduction (I): Definitions

Software and Video game Industry

ICT Industry Creative industry

• Exponential growth
• Technical procedures
• Business features
• Creative components
• Abstact skills

5.4 % of the World GDP in 2010. (Dutta and Mia, 2010) Méndez-Ortega, Carles 3/37 ECONRES SEMINARS (UAM)

Introduction (II): Motivation

• The Computer revolution has caused the appearance and rise of high-tech industries,

which are considered key drivers of economic growth in developed countries due to their capacity in knowledge-generation, creativity and innovation. (Berger and Frey, 2016).

• Concretely, the impact of Software and Video games is huge and growing over time.
• European Union (2014)
• 900 billion euros (7.9% of EU28 GDP).
• 11.6 million jobs (5.3% of EU28 jobs).
• The average wage per worker is 34% higher than the EU wage average and 80% higher

than the service sector average).

Méndez-Ortega, Carles 4/37 ECONRES SEMINARS (UAM)

Motivation (III): Previous Work

• Méndez-Ortega, Carles and Arauzo-Carod, JM. (2017):
• Location patterns of the Software and Video game Industry in Barcelona at micropolitan

level:

• Located in urban areas.
• Polycentric location:

22@ Urban project Eixample district

• Tended to be collocated with some others specific creative sectors.
• VFI, ADV, RTV.
• Young and small firms more concentrated than big and old ones.

Méndez-Ortega, Carles 5/37 ECONRES SEMINARS (UAM)

Introduction (IV): Aim and Contribution

The aim of this paper is to determine which reasons lead Software and Video games firms to locate in certain areas of Barcelona.

• Despite being an industry located in urban areas, most of previous empirical

research in location determinants of high-tech firms has been done at country and/or regional level. For this reason: This paper contributes to the literature filling the lack of empirical studies that analyze location determinants of SVE industry at urban level, dealing with factors that either had not been taken into account, or had not been analyzed together at this scale.

Méndez-Ortega, Carles 6/37 ECONRES SEMINARS (UAM)

Introduction (V): Why Barcelona?

Méndez-Ortega, Carles

• The province of Barcelona (NUTS3) accounts

more than the 80% of the Software and Video game firms (SVE) of Catalonia.

• The city of Barcelona accounts more than the

60% of SVE firms in Catalonia.

• Urban renewal projects (e.g. 22@), university

degrees and cultural environment among others, have made Barcelona one of the most attractive cities for these firms in Europe and worldwide. SVE firms’ entries in Barcelona between 2011 and 2013 7/37 ECONRES SEMINARS (UAM)

Introduction (VI): Theoretical Model and Literature Review

𝐷 = 𝐺(𝐵𝐹, 𝐻, 𝐼𝐷, 𝑢, 𝑀𝑄, 𝑇

• The cost for a firm selecting a location has been represented in the literature with the following

function (Brülhart et al., 2012; Gómez-Antonio and Sweeney, 2018):

• AE: Agglomeration Economies.
• G: Public goods ( e.g. Transport services, Wi-Fi Public services, Public centers and Urban

renewal areas by public iniciative).

• HC: Human Capital or Labour.
• t: Taxes (are constant inside the city).
• LP: Land Price.
• S: Other site characteristics (e.g. Technological parks, Universities, Creative diversity, Crime,

among others).

Méndez-Ortega, Carles 8/37 ECONRES SEMINARS (UAM)

Introduction (VII): Theoretical Model and Literature Review

Méndez-Ortega, Carles REAL SEMINARS – FALL 2018

Studies LAE HTA CCD HC LPT C Abramovsky and Simpson (2011) X X X Acosta et al. (2011) X X X X Audretsch and Lehmann (2005) X X X Audretsch and Keilbach (2004) X X Chatman and Noland (2011) X X X Marra et al. (2017) X Florida and Mellander (2016) X X X X Florida and Mellander (2009) X X Goetz and Rupasingha (2002) X X X X Hackler (2003) X X X X Kinne and Resch (2017) X X X X X Li and Zhu (2017) X X X Li et al. (2016) X X X Méndez-Ortega and Arauzo-Carod (2018) X X X Viladecans-Marsal and Arauzo-Carod (2012) X X X Wang et al. (2017) X X X Wood and Dovey (2015) X X X X Woodward et al. (2006) X X X X Zandiatashbar and Hamidi (2018) X X X X

Méndez-Ortega, Carles 9/37 ECONRES SEMINARS (UAM)

Introduction (VIII): Hypotheses

Méndez-Ortega, Carles Hypothesis 1: The impact of high-tech amenities, Cultural and creative diversity and Human Capital will have a positive impact on the SVE firms’ entries while this impact will be different across type of entries (i.e. Creative and All entries). The impact of High-tech amenities and Human Capital will be higher for SVE firms’entries than for Creative and All firms’entries. Hypothesis 2: The impact of Agglomeration economies, High-tech amenities, Human Capital, Creative and Cultural Diversity and Crime on the SVE firms’entries will go beyond neighborhood borders. 1/21 Méndez-Ortega, Carles 10/37 ECONRES SEMINARS (UAM)

Data (I)

• The Data used in this paper come from different sources:
• Data about firms comes from SABI (Sistema de Análisis de Balances Ibéricos).
• Data about amenities and social characteristics comes from Barcelona Statistical Service (Open

Data Barcelona).

• Some variables by Own elaboration (Scientific parks, entropy index, Co-working spaces, among
• thers).
• Area Study: City of Barcelona, at Neighborhood level (73).
• Period of Study: Stock of firms and amenities (2010) and entry of firms (2011-2013).

Méndez-Ortega, Carles REAL SEMINARS – FALL 2018 Méndez-Ortega, Carles 11/37 ECONRES SEMINARS (UAM)

Data (II): Own created variables

• Agglomeration (Aggl_10): Stock of Video and Film, Graphic arts and Radio and TV

firms (Méndez-Ortega and Arauzo-Carod, 2017, 2018).

• Entropy index (Entro): Indicator of inequality (Theil, 1972). It takes values between 0

and 1 and is typically used to detect whether a spatial unit (i.e., neighbourhood) is homogenous or diverse.

• Measures the Creative diversity: 17 Creative sectors (Boix and Lazzeretti, 2012)

Méndez-Ortega, Carles REAL SEMINARS – FALL 2018 Méndez-Ortega, Carles 12/37 ECONRES SEMINARS (UAM)

Data(III): Descriptive analysis

Méndez-Ortega, Carles 13/37 ECONRES SEMINARS (UAM)

Methodology (I): Empirical Model

𝐺𝑗𝑠𝑛 𝑓𝑜𝑢𝑠𝑗𝑓𝑡

𝑗(2011−2013 = 𝛾0 + 𝛾1𝑜𝐵𝐹𝑗𝑜 + 𝛾2𝑙𝐼𝑈𝐵𝑗𝑙 + 𝛾3𝑘𝐷𝐷𝐸𝑗𝑘 + 𝛾4ℎ𝐼𝐷𝑗ℎ + 𝛾5𝐷𝑠𝑗𝑛𝑓𝑗 + 𝜈𝑗

The empirical model is the following*: Count data model (CDM):

• Software and video games firms: Poisson Model.
• Creative and All fims: Negative Binomial Model.

(*) Taxes are constant in the city and land price effect is captured by other variables, as population density or agglomeration

• economies. Therefore, land price and taxes are not included in the empirical specification (Figueiredo et al., 2002).

Méndez-Ortega, Carles 14/37 ECONRES SEMINARS (UAM)

Methodology (III): Spatial Aproach (SVE firms)

• SLX Model
• P-SAR model

Méndez-Ortega, Carles REAL SEMINARS – FALL 2018 Méndez-Ortega, Carles 15/37 ECONRES SEMINARS (UAM)

Methodology (IV): SLX Model

SLX: Spatial lag of the covariates. 𝑧 = 𝑌𝛾 + 𝑋𝑌𝜄 + 𝜁

• W: First Order Queen contiguity.
• Variable Selection:
• Correlation X vs WX
• Significance in the Aspatial Model
• Moran’s I (Moran 1948)
• LISA (Anselin 1995)

Méndez-Ortega, Carles 16/37 ECONRES SEMINARS (UAM)

Variable Correlation with WX Moran I

• Sig. As

patial SLX Model SVE_10 0.574* 0.406 Yes Yes cowork2 0.680* 0.538 No No wifi 0.618* 0.447 Yes No ctp 0.134 0.060 Yes Yes dis t_22 0.472* 0.343 Yes Yes ent_f 0.739* 0.566 Yes No markets 0.089* 0.046 Yes Yes cc

• 0.020
• 0.011

No No dis t_centre

• No

Pol_rat 0.668* 0.478 Yes Yes uni 0.291* 0.143 Yes Yes edu_2010 0.841* 0.681 Yes No popd_2010 0.385* 0.240 Yes No

Source: Author. Note: Sig. Aspatial indicates whether this variable was significant in the aspatial model.

• Methodology (III): SLX Model

Méndez-Ortega, Carles 17/37 ECONRES SEMINARS (UAM)

Methodology (III): SLX Model

Sve_10 CTP Dis t_22 Markets Pol_rat Uni

Source: Author.

Méndez-Ortega, Carles Source: Author’s calculations 18/37 ECONRES SEMINARS (UAM)

Methodology (III): P-SAR

Méndez-Ortega, Carles

Developed by Lambert et al., (2010)

• The function 𝑕(𝑧𝑘 represents the logged-transformed values approximating

neighboring counts (Burbidge et al., 1988). Log-likelihood function of the first-stage estimator is:

𝑚𝑜𝑀1 =

𝑗=1 𝑜

𝑔

1 𝑋 · 𝑕 𝑧𝑘 𝑅𝑗; 𝜀

With the instruments:

19/37 ECONRES SEMINARS (UAM)

𝑅 = [𝑌, 𝑋𝑌, 𝑋𝑋𝑌]

Methodology (III): P-SAR

Méndez-Ortega, Carles

The instruments regressed on the transformation yields the vector of predicted values: In the second step, the first-stage predicted values enter in the Poisson probability density function as:

𝑅𝜀 with 𝜀 = 𝑅(𝑅′𝑅 −1𝑅′𝑋 · 𝑕 𝑧𝑘

∗

𝑔 𝑧 𝑦, 𝑋, 𝑅𝑗𝜀′; 𝛾, 𝜍 = ex p( 𝛾′𝑦𝑗 + 𝜍 · 𝑅𝑗

′𝜀 𝑧𝑗 ex p( − exp 𝛾′𝑦𝑗 + 𝜍 · 𝑅𝑗 ′𝜀

𝑧𝑗!

20/37 ECONRES SEMINARS (UAM)

SVE_ent (0.373) Source: Author. Note: Moran index in brackets. Red colour means neighbourhoods with a high value surrounded by neighbourhoods with high value, red off means neighbourhoods with a high value surrounded by neighbourhood with low value, blue off means neighbourhoods with a low value surrounded by neighbourhood with high value and blue means neighbourhoods with a low value surrounded by neighbourhoods with low value. Results after 999 permutations.

Méndez-Ortega, Carles

Methodology (III): P-SAR

21/37 ECONRES SEMINARS (UAM)

Méndez-Ortega, Carles REAL SEMINARS – FALL 2018

RESULTS: ASPATIAL APPROACH

Méndez-Ortega, Carles Software and Video games firms Creative Firms All firms PM NBM NBM (1) (2) (3) (4) Agglomeration Economies Loc_10 0.00787*** 0.00193*** 0.000293*** (0.00148) (0.000367) (6.11e-05) Aggl_10 0.00592*** (0.00102) Cowork 0.0475

• 0.0159

0.125 0.119 (0.147) (0.147) (0.113) (0.128) High-Tech Amenities Wifi 0.0211** 0.0296*** 0.00986 0.0231** (0.00943) (0.00850) (0.00772) (0.0111) CTP 0.410* 0.620*** 0.137 0.959*** (0.235) (0.236) (0.221) (0.270) Dis t_22 0.883*** 0.853*** 0.287

• 0.343

(0.301) (0.301) (0.270) (0.330)

• 23/37

ECONRES SEMINARS (UAM)

Cultural and Creative Divers ity Entropy 3.483*** 3.476*** 6.977*** 2.881*** (1.351) (1.306) (1.271) (0.554) Markets 0.106*** 0.135*** 0.0759*** 0.0889** (0.0328) (0.0342) (0.0274) (0.0400) CC

• 0.0381
• 0.131

0.0405

• 0.0571

(0.0976) (0.0955) (0.0759) (0.0907) Dis t_centre 3.79e-05 1.77e-05

• 2.94e-06

5.14e-05 (8.74e-05) (8.64e-05) (7.40e-05) (7.06e-05) Human Capital Uni 0.0607** 0.0563** 0.0374

• 0.0221

(0.0274) (0.0274) (0.0287) (0.0419) Edu_2010 5.214*** 4.574*** 4.854*** 2.559*** (1.264) (1.267) (1.029) (0.903) PopD_2010 0.00104*** 0.00102*** 0.000751** 0.000655** (0.000393) (0.000389) (0.000334) (0.000309)

• Méndez-Ortega, Carles

24/37 ECONRES SEMINARS (UAM)

Méndez-Ortega, Carles

Crime Pol_rat

• 0.112**
• 0.140**
• 0.0799*
• 0.0135

(0.0544) (0.0548) (0.0415) (0.0536) Constant

• 4.237***
• 3.924***
• 5.608***
• 0.297

(1.286) (1.244) (1.186) (0.734) Observations 73 73 73 73 Non-zero observations 49 49 59 71 LR chi2 570 576.2 154.4 152.8 Log likelihood

• 137.3
• 134.2
• 159.3
• 309.4

Pseudo R-squared 0.675 0.682 0.327 0.198 /ln alpha

• 3.124***
• 1.501***

(0.676) (0.207) alpha 0.0440 0.223 VIF 2.92 2.87 2.91 2.89

Note: Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1. Note: Loc_10 refers to stock of the current type of firms in

• 2010. Poisson Model (PM), Negative Binomial Model (NBM).

25/37 ECONRES SEMINARS (UAM)

Méndez-Ortega, Carles REAL SEMINARS – FALL 2018

RESULTS: SPATIAL APPROACH

Robustness Analysis

• Multicollinearity:
• Variance Inflaction Factor (VIF)
• Correlation
• Land Price
• Test: we found a positive and statistically

significant effect of population density and education over rent prices.

• Model Selection (Cameron and Trivedi 2013):
• Akaike Information Criteria (AIC)
• Bayesian Information Criteria (BIC)
• Vuong test

Méndez-Ortega, Carles

• Variable SLX selection:
• Aspatial Significance, Moran I and Local

indicator of Spatial Association (LISA).

• P-SAR
• Test for different W matrices (First

Order Queen Contiguity, Second Order Queen Contiguity, 5 k-nearest neighbours, Rook contiguity and median distance between neighborhoods).

28/37 ECONRES SEMINARS (UAM)

Robustness Analysis

• Multicollinearity:
• Variance Inflaction Factor (VIF)
• Correlation
• Land Price
• Test: we found a positive and statistically

significant effect of population density and education over rent prices.

• Model Selection (Cameron and Trivedi 2013):
• Akaike Information Criteria (AIC)
• Bayesian Information Criteria (BIC)
• Vuong test

Méndez-Ortega, Carles

• Variable SLX selection:
• Aspatial Significance, Moran I and Local

indicator of Spatial Association (LISA).

• P-SAR
• Test for different W matrices (First

Order Queen Contiguity, Second Order Queen Contiguity, 5 k-nearest neighbours, Rook contiguity and median distance between neighborhoods).

29/37 ECONRES SEMINARS (UAM)

Rent Price SVE Firms Creative Firms All Firms (1) (2) (3) (4) Loc_10

• 0.00926*
• 0.00197
• 0.000209

(0.00475) (0.00119) (0.000129) Edu_2010 15.79*** 15.88*** 16.04*** 15.91*** (2.024) (2.054) (2.063) (2.054) PopD_2010 0.00146** 0.00145** 0.00148** 0.00148** (0.000568) (0.000576) (0.000573) (0.000573) Cons tant 6.288*** 6.241*** 6.150*** 6.152*** (1.306) (1.320) (1.318) (1.320) LAE var. Yes Yes Yes Yes HTA var. Yes Yes Yes Yes CCD var. Yes Yes Yes Yes HC var. Yes Yes Yes Yes Crime var. Yes Yes Yes Yes Obs ervations 73 73 73 73 R-s quared 0.768 0.763 0.764 0.763 Standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

• Méndez-Ortega, Carles

30/37 ECONRES SEMINARS (UAM)

Robustness Analysis

• Multicollinearity:
• Variance Inflaction Factor (VIF)
• Correlation
• Land Price
• Test: we found a positive and statistically

significant effect of population density and education over rent prices.

• Model Selection (Cameron and Trivedi 2013):
• Akaike Information Criteria (AIC)
• Bayesian Information Criteria (BIC)
• Vuong test

Méndez-Ortega, Carles

• Variable SLX selection:
• Aspatial Significance, Moran I and Local

indicator of Spatial Association (LISA).

• P-SAR
• Test for different W matrices (First

Order Queen Contiguity, Second Order Queen Contiguity, 5 k-nearest neighbours, Rook contiguity and median distance between neighborhoods).

31/37 ECONRES SEMINARS (UAM)

Robustness Analysis

• Multicollinearity:
• Variance Inflaction Factor (VIF)
• Correlation
• Land Price
• Test: we found a positive and statistically

significant effect of population density and education over rent prices.

• Model Selection (Cameron and Trivedi 2013):
• Akaike Information Criteria (AIC)
• Bayesian Information Criteria (BIC)
• Vuong test

Méndez-Ortega, Carles

• Variable SLX selection:
• Aspatial Significance, Moran I and

Local indicator of Spatial Association (LISA).

• P-SAR
• Test for different W matrices (First

Order Queen Contiguity, Second Order Queen Contiguity, 5 k-nearest neighbours, Rook contiguity and median distance between neighborhoods).

32/37 ECONRES SEMINARS (UAM)

Robustness Analysis

• Multicollinearity:
• Variance Inflaction Factor (VIF)
• Correlation
• Land Price
• Test: we found a positive and statistically

significant effect of population density and education over rent prices.

• Model Selection (Cameron and Trivedi 2013):
• Akaike Information Criteria (AIC)
• Bayesian Information Criteria (BIC)
• Vuong test

Méndez-Ortega, Carles

• Variable SLX selection:
• Aspatial

Significance, Moran I and Local indicator of Spatial Association (LISA).

• P-SAR
• Test for different W matrices (First Order

Queen Contiguity, Second Order Queen Contiguity, 5 k-nearest neighbours, Rook contiguity and median distance between neighborhoods).

33/37 ECONRES SEMINARS (UAM)

Méndez-Ortega, Carles 34/37 ECONRES SEMINARS (UAM)

Conclusion (I)

• This paper has contributed to the literature filling the lack of empirical evidence having

analyzed location determinants of this industry at urban level, dealing with factors that either had not been taken into account, or had not been analyzed together at this scale.

• SVE firms tend to choose locations with good high-tech amenities, high diversity of creative

firms and places with presence of SVE firms and other similar type of firms (i.e. VFI, RTV and ADV).

• Our hypothesis were met except for the spatial effect of some covariates, since lagged

variables of University and Crime are not significant.

Méndez-Ortega, Carles 35/37 ECONRES SEMINARS (UAM)

Conclusion (II)

• Policy Implications:
• Until now, it was mainly taken into account that SVE firms were located in places

with technological facilities, human capital and good infrastructures in general.

• This paper has shown that not only these characteristics are important, but also

cultural and creative diversity are very important for the location and development of this industry inside a city.

• Promotion and attraction of

creative activities, jointly with previous factors mentioned above, will contribute to the location of SVE activities.

• Limitations:
• Specific Period of time (2010-2013) and place (Barcelona).
• MAUP Problem.

Méndez-Ortega, Carles 36/37 ECONRES SEMINARS (UAM)

Place the ‘Candy’ and ‘Crush’ it: Entry Determinants of the Software and Video game firms in Barcelona

Carles Méndez-Ortega – carles.mendez@urv.cat Universitat Rovira i Virgili (CREIP-QURE) Seminario ECONRES March 29th, 2019 Facultad de Ciencias Económicas y Empresariales (UAM) Méndez-Ortega, Carles ECONRES SEMINARS (UAM)