Local Multipliers in New Firm Creation: Inter-Sectoral Spillovers of Entrepreneurship and Startup Agglomeration

ABSTRACT

This paper investigates the local multiplier effects in new firm creation by analysing a unique panel dataset of new firms and patents at the regional level in Australia. It finds that new firms in the tradable sector such as low-technology manufacturing and knowledge-intensive services industries have local multiplier effects on new firm creation in the non-tradable sector such as other services industries. It also shows that the local multiplier effect in new firm creation exists between manufacturing and knowledge-intensive services industries in the tradable sector. The local multipliers are greater in regions with more patents than in other regions.

KEYWORDS:

• Local multiplier effect
• new firm creation
• regional entrepreneurship
• local spillovers

JEL CLASSIFICATIONS:

• M13
• R11
• L26
• O30

Acknowledgement

The author thanks Beth Webster and Alfons Palangkaraya for helpful comments on an earlier version of this paper.

Disclosure Statement

No potential conflict of interest was reported by the author(s).

Notes

1 For a summary of the main findings and research agenda on regional entrepreneurship, refer to Audretsch and Keilbach (2004), Qian, Acs, and Stough (2013), Fritsch and Storey (2014), and Chatterji, Glaeser, and Kerr (2014).

2 For example, when a new job is created in the manufacturing industry, there is a possibility that additional jobs could be created in the local services industry through increased demand for local goods and services in the region. In general, the local multiplier effect stems from an increase in the local demand of the non-tradable sector (e.g. local services industry) that is caused by an increase in the total earnings of the tradable sector (e.g. manufacturing industry). For an overview of the concept, causes, and empirical studies regarding the local multiplier effect, refer to Moretti (2010, 2012), Moretti and Thulin (2013), Thulin (2015), and Van Dijk (2017).

3 The tradable sector consists of the industries, including all manufacturing and some services industries (e.g. knowledge-intensive services) that provide goods or services that are sold outside a particular region such as a city, and the non-tradable sector (e.g. local services) consist of the industries that sell goods or services in the region that are produced mainly based on local demand (Van Dijk 2017; Lee and Clarke 2017). For example, the electronics and financial industries are a part of the tradable sector, and retail industries, local health services, and restaurants are a part of the non-tradable sector. Porter (2003) referred to the non-tradable sector as local industries.

4 Knowledge-intensive services sector consists of high-technology services (e.g. information and telecommunication (IT) services or computer-related service activity, and research and development (R&D)-related services), business services (e.g. legal and accounting services and consulting services), financial services (e.g. financial and insurance services, and activities related to financial intermediation), and other knowledge-intensive services (e.g. education and training services). For details of the OECD classification of the knowledge-intensive services sector, corresponding industry classification codes, and related explanations, refer to Hatzichronoglou (1997), Miles (2008) and Muller and Doloreux (2009).

5 A MB is the smallest geographical area defined by the ABS in Australia. MBs are used to form the building blocks for the larger regions of the Australian Statistical Geography Standard (ASGS). MBs can have between 30 and 60 dwellings or no dwellings. I collected the correspondence from the MB code to the LGA code from public information provided by the ABS (1270.0.55.003 – Australian Statistical Geography Standard (ASGS): Volume 3 – Non ABS Structures, July 2011).

6 To create this link, I used the correspondence from the postcode to the LGA code in public information provided by the ABS because the IPGOD does not provide the applicant's MB code. I checked the reliability of using the correspondence from the postcode to the LGA code as an alternative method of using the correspondence from the MB code to the LGA code. 95% of the LGA codes matched with the postcodes are the same as those with the MB codes in the ABR database. Therefore, the aggregation method for patents at the regional level is acceptable.

7 Existing studies on new firm formation and the local employment multiplier controlled for the effect of incumbent firms (Colombelli and Quatraro 2018; Goos, Konings, and Vandeweyer 2018). Using both ΔN(HTM)_it and I(HTM)_it−1 as explanatory variables together could cause the multicollinearity problem because I(HTM)_it−1 is correlated with ΔN(HTM)_it. In this respect, when the dependent variable is an increase in the number of new firms in a sector (e.g. ΔN(OTS)_it), only the number of incumbents in that sector (e.g. I(OTS)_it−1) is employed as the control variable. This study also controls for the effect of the resource-dependent sector due to its significant impact on the Australian economy and job creation (Fleming and Measham 2014).

8 The Hausman test statistics (Hausman 1978) for the estimation of Equations (1) to (6) reject the null hypothesis that the random-effects estimator is consistent, which means that the fixed-effects model is appropriate.

9 This study uses the xtscc command with the fixed-effects (fe) option of STATA to estimate the robust standard errors based on Driscoll and Kraay’s (1998) covariance matrix (Hoechle 2007). This command allows the error structure to be heteroskedastic, autocorrelated, and cross-sectionally dependent between the groups (i.e. spatially dependent between the regions in this study).