The role of regulation and regional government quality for high-growth firms: the good, the bad and the ugly

ABSTRACT

High-growth firms (HGFs) are important for job creation and productivity growth. We investigate the relationship between product and labour market regulations, as well as the quality of regional governments that implement these regulations, and the development of HGFs across European regions. Using data from Eurostat, the Organisation for Economic Co-operation and Development (OECD), World Economic Forum (WEF), and Gothenburg University, we show that both regulatory stringency and the quality of the regional government relate to the regional shares of HGFs. In particular, we find that the effect of labour and product market regulations is moderated by the quality of regional government. Depending on the quality of regional governments, regulations may have a ‘good, bad or ugly’ influence on the development of HGFs. Our findings contribute to the debate on the effects of regulations and offer important building blocks to develop tailored policy measures that may influence the development of HGFs in a region.

KEYWORDS:

• high-growth firms
• regulation
• quality of regional governments
• regions

JEL:

• H11
• L25
• L50
• O43
• R11
• R50

1. INTRODUCTION

Regulations and administrative burdens affect the capacity of firms to innovate, grow and create new jobs. The effectiveness of regulations is linked to the quality of regional public administration and governance. In particular, poor quality of governance and institutions appear to be one of the main obstacles to business development in low-growth regions (Committee of the Regions (CoR), 2019a). While there is some theoretical and empirical evidence for the effect of institutional quality on economic growth at the country (Acemoglu et al., 2014; Djankov et al., 2006) and regional (Charron et al., 2014; Charron et al., 2019; Rodríguez-Pose & Garcilazo, 2015) levels, much less is known about the relationship between institutions and high-growth firms (HGFs) at the regional level.¹

HGFs are receiving substantial attention from policymakers because of their potential to contribute to productivity growth² and job creation at the regional level (Coad et al., 2014; Henrekson & Johansson, 2007). In Europe, these firms alone are responsible for 50% of employment growth, even though HGFs represent only a very small share of all firms (Bravo-Biosca et al., 2016; Coad et al., 2022). These firms, in their crucial stage of development and growth, may be sensitive to how effectively regional public administrations implement regulations. Despite their small share, HGFs are economically relevant as job creators, tend to have positive spillovers on non-HGFs (Du & Vanino, 2021), and have been more resilient during crises in terms of avoiding negative employment effects (Teruel et al., 2022). Yet, HGFs are unequally distributed. The Organisation for Economic Co-operation and Development (OECD) and European Commission report evidence of stark differences in the shares of HGFs across countries (European Commission, 2016; Schreyer, 2000).

A growing number of studies investigate factors explaining the distribution of HGFs across countries. Among macroeconomic factors, the role of regulation in the economic performance of countries is central in the political economy literature. One channel through which regulation can impact growth and employment is via its effect on firm dynamics. Accordingly, there is considerable literature regarding the impact of regulations on entry (Ardagna & Lusardi, 2010; Ciccone & Papaioannou, 2007; Djankov et al., 2002). Only a few studies address the influence of regulation on HGFs (Bravo-Biosca et al., 2016; Flachenecker et al., 2020), and these examine cross-country differences, leaving the heterogeneity in the distribution of HGFs across regions largely unexplored.

Regional heterogeneity should matter when exploring the link between regulations and HGFs in Europe. The CoR states that the majority of European Union (EU) legislation is implemented by regional and local authorities as they ‘have valuable first-hand experience of applying EU legislation, in close contact with local businesses, social partners, civil society and citizens’.³ In this regard, the quality of the regional governance may play an important role alongside the level of regulation. Acemoglu et al. (2001) are among the first to highlight the role of institutional quality in explaining the differences in growth rates across countries. Further, several papers empirically identify the causal effects of a better administration on higher income per capita, returns to public policy and local development (Acemoglu et al., 2001; Crescenzi et al., 2016, 2022; Hall & Jones, 1999; Jalilian et al., 2007).

Despite the key role of government quality at the regional level, no previous study investigates how heterogeneity in the quality of regional governments affects the relationship between regulations and HGFs. Therefore, the main aim of this paper is to examine the mechanisms through which the heterogeneity of government quality affects the interplay between regulations and HGFs. In this context, we hypothesise that regulations might be conducive to prosperity, increasing the regional share of HGFs, when implemented by high-quality regional governments.

To investigate our research question, we match several open-source datasets at the NUTS-2 and one-digit Statistical Classification of Economic Activities in the European Community (NACE) levels from various sources including Eurostat, Gothenburg University, the OECD, the World Economic Forum (WEF), and the European Quality of Government index (QoG). The latter indicator summarises citizens’ and businesses’ experiences with public services in their regions while remaining agnostic on whether these services are provided by local or regional self-governments, or by local and regional branches of national governments (Charron et al., 2014). To estimate the direct and moderating role of government quality on the relationship between regulations and the share of HGFs, we use a series of linear regression models. In particular, to observe how the empirical relationship between HGFs and regulation varies with the quality of regional government, we use interaction effects between regulations and quality of government alongside separate regressions for regions with different levels of government quality and for regions that improved or worsened their quality of government. This approach allows us to reveal under what conditions a change in regulatory stringency may influence the shares of HGFs at the regional level.

This paper contributes to the literature on the relationship between HGFs and regional institutional quality and regulation. First, it provides empirical evidence on the direct relation between labour and product market regulations, government quality, and the regional shares of HGFs in Europe. Second, and more importantly, it investigates the interaction between national regulations and regional government quality, exploring the heterogeneity in relationship between regulations and HGFs. We find that the empirical relation between labour and product market regulations and the regional share of HGFs is moderated by the quality of regional government; thus, it is the institutional quality that is the decisive factor turning regulations good, bad, or ugly for the regional development of HGFs shares.

The rest of the paper is structured as follows. Section 2 summarises the literature related to institutions and economic growth as well as HGFs. Section 3 describes the datasets. Section 4 reports the estimation strategy and results. Finally, section 5 discusses the results and concludes.

2. REVIEW OF THE LITERATURE: INSTITUTIONS AND HIGH-GROWTH FIRMS

Given its economic relevance, research on HGFs has gained momentum, with three main lines of research developing in parallel. The first strand looks at the micro-level characteristics explaining the emergence and growth of firms. In a literature survey on firm growth, Storey (2016, p. 122) identifies and classifies 35 characteristics into three categories. The starting resources of the entrepreneur(s), for example, motivation and education; the firm, for example, age and size; and strategy, for example, management and personnel training and market positioning. Generally, HGFs tend to be younger, more international, more innovative and heterogeneously distributed across countries and sectors (Audretsch, 2012; Konon et al., 2018; Teruel & de Wit, 2017). Some specific firm characteristics, such as rapid past employment and export growth, as well as recruiting and training qualified personnel, seem to be good predictors for identifying firms that have a higher likelihood of becoming HGFs (Lopez-Garcia & Puente, 2012).

The second line of enquiry considers the local enabling environment. Minghao Li et al. (2016) investigate which local factors are important for stimulating HGFs at the county level in the United States. They find that HGFs are more often located in counties where firms are – on average – larger, educational attainment is higher, and where the location itself is attractive. They also find that HGFs are not necessarily restricted to urban areas, but can also be found in more rural counties. Yet, more peripheral regions may have disadvantages when compared with central regions (Mason et al., 2015). Friesenbichler and Hölzl (2020) also examine whether structural regional factors explain differences in HGFs shares across Austrian regions. They observe that higher shares of HGFs are found within regions with a higher degree of digitalisation, industrial specialisation and lower mobility barriers.⁴

The third strand of research, to which this paper contributes, investigates regulatory and framework conditions for firm dynamics and economic growth. The role of institutions (e.g., regulations, property rights, statutory laws, etc.) for economic development is generating considerable interest among scholars of political sciences and economics since the 1970s. Public interest theory of regulation, fathered by Pigou and Aslanbeigui (1938), holds that regulation seeks the protection and benefit of the public at large. In contrast, public choice theorists see regulation as socially inefficient and as a tool to create rents for bureaucrats or incumbent firms (McChesney, 1987; Niskanen, 1971; Shleifer & Vishny, 1997). The empirical evidence on the link between regulatory institutions and economic growth is polarising, mirroring these contrasting approaches. While some work finds a positive relationship between legislation and economic development (Ash et al., 2022; Fukumoto, 2008), others provide evidence of a negative relationship between regulatory institutions and indicators of economic development, that is, technology diffusion, productivity and employment growth, as well as firm entry and firm growth (Acemoglu et al., 2005; Blanchard, 2004; Buccirossi et al., 2013; Ciccone & Papaioannou, 2007; Lynn et al., 1996; Nickell, 1997; Nicoletti & Scarpetta, 2003).

Conversely, there is unanimous consensus on the positive role of institutional quality. Hall and Jones (1999) and Acemoglu et al. (2001) highlight the role of institutional quality in explaining the differences in economic development across and within countries. Empirical evidence shows that ‘effective regulation’ is needed for businesses and the whole economy to develop (Jalilian et al., 2007). The availability of the regional index of quality of government developed by Charron et al. (2014) has spurred several studies on the role of the institutional endowment, with research reporting a strong link between the quality of regional government and the innovation performance of European regions. Ketterer and Rodríguez-Pose (2018) find, with a subnational analysis, that government effectiveness and the fight against corruption are conducive to regional economic development. Cortinovis et al. (2017) show that regional government quality matters for industry diversification. Agostino et al. (2020) also report strong evidence of the positive relationship between regional institutional quality and total factor productivity of European small and medium-sized firms in the manufacturing industries.

Specific to HGFs, several innovation scholars investigate the relationship between institutions and HGFs. For example, Davidsson and Henrekson (2002) examine the relationship between Swedish institutional setups and entrepreneurial activity, such as firm entry and growth. Their findings suggest that high rates of taxation (of labour and entrepreneurial income) weakens the start-up and scale-up incentives. Further, strict labour security legislation and wage-setting institutions disadvantage smaller firms with good growth prospects. These firms need the flexibility to adjust the number of employees in response to the changes in demand. Flexibility and freedom of contracting may further help these firms achieve their high-growth potential. Henrekson and Johansson (2007) corroborate this hypothesis by finding that labour regulations, especially those that restrict contracting flexibility, are harmful to enterprises that would like to grow rapidly. More recently, Bravo-Biosca et al. (2016) show that stringent employment protection legislation is related to lower shares of both growing and shrinking firms in sectors that are more labour intensive and more innovative, in turn, reducing the speed of the resource reallocation.

Moreover, restrictive product market regulation may affect firms’ investment decisions as well as the decisions on whether to enter or leave the market. High transaction and entry costs may discourage small, young and innovative companies, which are usually unable to obtain sufficient capital or to overcome cost- and non-cost-related barriers to entry and growth. Djankov et al. (2002) analyse data on entry regulation in 85 countries and find that heavy entry regulation benefits only politicians and bureaucrats; further, it is associated with lower quality public and private goods. Falkenhall and Junkka (2009) claim that low barriers to entry and contestable markets are key for the development of HGFs in Sweden. Therefore, the predominant opinion on the effect of regulation on HGFs is that a stringent regulatory framework hampers business growth because high taxes, alongside entry and labour market regulations, create obstacles that start-ups and small firms are neither financially nor organisationally equipped to overcome (Audretsch et al., 2022; Belitski et al., 2016).

To some extent, in contradiction to this picture, and thus important for our further analysis, the review of Kitching (2006) on regulation and business performance, while also reporting evidence of mainly negative effects of regulation and taxes for business growth and performance, finds occasional evidence of the potential benefits of regulation. For example, statutory audits impose financial discipline on firms and protect society from business malpractice. Further, environmental regulations can stimulate firms to find innovative product and process solutions. In a follow-up review, Kitching et al. (2015) argue and show, with selected case studies, that regulation generates contradictory effects, ‘enabling as well as constraining performance’. A study by Lee et al. (2015) of UK firms finds that regulation is actually less likely to be a problem for HGFs, because ‘presumably in order to grow, the businesses must have succeeded in overcoming some regulatory difficulties or have been lucky enough not to face them’ (p. 189). Lucidi (2012) and Amoroso and Martino (2020) also find that too much flexibility in the labour market may reduce a firm’s ability to attract and retain talented employees.

While the role of formal institutions (e.g., rule of law, regulations, etc.) in firm growth is already studied, there is very little research on the link between informal institutions (e.g., culture, religion, corruption, etc.) and HGFs, let alone the nexus between formal institutions and the implementation of these regulations and their joint impact on HGFs. We identify only two studies. Estrin et al. (2013) study the effect of formal and informal institutions on business growth aspiration and find that higher levels of corruption, weaker property rights, and a larger government significantly constrain entrepreneurial employment growth aspirations. Krasniqi and Desai (2016), exploring the effects of a set of informal institutions and tax rates, trade and customs regulations, and business permits in transition economies, find no robust evidence of the effect of any of these factors on the share of HGFs.

After having reviewed the relevant literature, we think that the quality of regional government may be the missing piece of the puzzle in the understanding of the nexus between regulation and economic outcomes such as the share of HGFs. Local governments, such as regional authorities, are real political and legislative actors in the context of European multilevel governance. They are tasked to implement many regulations as independent or as delegated administrations. Hence, the quality of regional government should be particularly important for HGFs, where the efficiency in processing lengthy administrative procedures is key to avoid the loss of opportunities for firm growth. Therefore, in this paper, we investigate how the quality of regional administrations might be a key factor shaping the growth performance of HGFs and hence regional development, when considering an array of regulatory environments that spans from stringent to flexible.

3. DATA AND DESCRIPTIVE RESULTS

For the empirical analysis, data on business dynamics (entry, exit and share of HGFs) at the region- and one-digit aggregate sector-level are extracted from Eurostat’s Structural Business Statistics (SBS).⁵ Regional SBS data are available at the NUTS-2 level for 21 of the 27 EU member states (data for regions in Belgium, Cyprus, Germany, Greece, Ireland and Luxembourg are not available) and for the 2008–20 period, although some regions only report a few years (e.g., data for Polish, Slovenian and Swedish regions are only available between 2011 and 2015, after 2010, or before 2018, respectively). HGFs are defined using the OECD’s definition according to which HGFs are firms with a minimum of 10 employees that realise an employment or turnover growth of more than 20% over three consecutive years. To calculate regional shares of HGFs, we divide the number of HGFs measured in employment by the total population of all active firms.⁶

SBS data are matched to the regional European QoG (available for the survey years 2010, 2013, 2017) at the NUTS-2 level, which was developed by the Quality of Government Institute, Gothenburg University (Charron et al., 2014, 2015, 2019), and is the only measure of institutional quality available at the regional level in the EU. This index hinges on experiences with public sector corruption, as well as the extent to which EU citizens believe various public sector services are impartially allocated and of good quality.⁷ Maps of Europe shown in Figure 1 display considerable subnational variation, but also show a sizeable change over time and a steady deterioration of performance in old EU member states (Fazekas & Czibik, 2021). Figure 4 goes one step further in showing, for each country included in the final sample, the QoG levels in 2010 and 2017. If the observations are above (below) the 45° line, the quality of government has worsened (improved). In the majority of cases, the quality of government institutions has decreased. Notable exceptions are Czechia and Poland.

Figure 1. European Quality of Government index (QoG), (a) 2010 and (b) 2017.

Note: Darker hues indicate higher quality of government. The legend is based on the quartile calculation for 2010.

Source: Authors’ on calculations based on the European Quality of Government index (QoG) from the QoG Institute, Gothenburg University.

Two maps of Europe displaying sub-national variation in the quality of government in 2010 (left) and 2017 (right).

Figure 2 reveals a positive relationship between the QoG index and the average share of HGFs in 2017, where southern Italian regions, such as Abruzzo and Calabria (ITF1 and ITF6, respectively), have low QoG and a lower share of HGFs (less than 0.5%); in contrast, Finnish regions, such as Helsinki-Uusimaa (FI1B) and North & East Finland (FI1D), have some of the highest QoG index scores and higher shares of HGFs (0.73% and 0.85%, respectively). Generally, the average shares of HGFs are very small, ranging from 0% to 3%, with an average of just 0.5%.

Figure 2. Share of high-growth firms and quality of regional government, 2017.

Sources: Authors’ on calculations based on the European Quality of Government index (QoG) from the QoG Institute, Gothenburg University; and Eurostat–SBS data.

Scatter plot of regional QoG index and the average regional shares of HGFs in 2017.

A clear picture of the within-country and sectoral distribution of HGFs is shown in Figure 3. For the majority of countries, the bulk of HGF shares lies between 0.2% and 0.7%. Exceptions are Estonia and the Netherlands, where, in almost 50% of their regions, more than 1% of firms are HGFs. Variation across sectors is more pronounced. Sectors with the highest concentration of HGFs are Industry (B–E), Transportation and storage (H), and Information and communication (J). Incidentally, these are also the most product market regulated sectors (with the highest averages of product market regulation indicator values of 3.6, 26.0 and 3.2, respectively).⁸

Figure 3. Share of high-growth firms by country and sector, average for 2008–18.

Note: The box plots display the first to the third quartiles. The vertical line is the median. Whiskers represent the minimum and maximum.

Countries are: AT = Austria; CZ = Czechia; DK = Denmark; EE = Estonia; ES = Spain; FI = Finland; FR = France; HU = Hungary; IT = Italy; NL = Netherlands; PL = Poland; PT = Portugal; SI = Slovenia; and SK = Slovakia.

Sectors are: B–E: Industry (B = Mining and quarrying; C = Manufacturing; D = Electricity, gas, steam and air-conditioning supply; E = Water supply; sewerage, waste management and remediation activities). F = Construction. G = Wholesale and retail trade; repair of motor vehicles and motorcycles. H = Transport and storage. I = Accommodation and food service activities. J = Information and communication. K–L: K = Financial and insurance activities; L = Real estate activities. M–N: M = Professional, scientific and technical activities; N = Administrative and support service activities. P–Q: P = Education; Q = Human health and social work activities. R–S: R = Arts, entertainment and recreation; and S = Other service activities.

Source: Authors’ calculations based on Eurostat–SBS data.

Two box plots reporting the distribution of HGFs shares by country (within-country variation) and by sector (within-sector variation) of our sample.

To measure product market regulation, we use the regulatory impact (REGIMPACT) indicator developed by Égert and Wanner (2016) for the OECD (1975–2018) as a proxy. The indicator follows the same rationale of the product market regulation (PMR) indicator developed by the OECD itself,⁹ but has the advantage of being disaggregated by sector (NACE rev.2, two-digit). The indicator measures the indirect impact of regulatory barriers to firm entry, business activities and competition in the Energy, Transport, and Communication sectors on all other sectors in the economy (via intermediate trade networks), covering the 1975–2018 period. The rationale is that sectors using intermediate inputs from regulated sectors are indirectly affected. The indicator has normalised values between 0 (low regulation) and 100 (high regulation). Unfortunately, the indicator is not available for six countries (Bulgaria, Croatia, Latvia, Lithuania, Malta and Romania), leaving our sample with a total of 15 countries. This sample gives us an unbalanced panel of 150 regions and 10 macro-sectors across 11 years.

It is worth noticing that regions with high QoG index scores, for example, Finnish regions, despite their higher-than-EU average level of PMR (stringent product market regulation), have high shares of HGFs. By contrast, Italian regions have a low quality of government, a low PMR level (below the EU average) and a low share of HGFs.

To measure the extent of regulation in the labour market, we use the country-level indicator ‘hiring and firing practices’ from The Global Competitiveness Index Historical Dataset 2007–17 of the WEF (Version 20180712).¹⁰ The original indicator ranges from 1 to 7 (‘In your country, to what extent do regulations allow flexible hiring and firing of workers?’ [1 = not at all; 7 = to a great extent]), but we converted it to a 0–100 scale to have comparable marginal effects. In addition to government quality, and product and labour market regulations, we control for variables that influence the emergence and growth of HGFs (as mentioned in section 2.2), such as access to finance, churn rate, absorptive capacity (measured as regional share of engineers and scientists), share of high-tech employment and growth of GDP per capita.

A description of all variables used in the analysis, alongside their sources and descriptions, is given in Table 1. The final sample is a balanced panel dataset of 1450 region–sector pairs (145 regions and 10 macro-sectors) for 11 years (2008–18).

Table 1. Description of the variables.

Variable	Description	Sources	Sample
HGF (%)	Number of high-growth enterprises measured in employment (growth by 20% or more for at least three consecutive years) divided by the population of active enterprises	Eurostat-SBS (BD_HGNACE2_R3)	$\mathrm{rjt}$ (2008–18)
QoG	The European Quality of Government index is based on a large citizen survey where respondents are asked about perceptions and experiences with public sector corruption, along with the extent to which citizens believe various public sector services are impartially allocated and of good quality. The index ranges from 0 to 100 (highest quality). It is available only for the survey years 2010, 2013, 2017 and 2021. We use either interpolated data for the missing years, or assume the index to be constant between survey years	The Quality of Government (QoG) Institute, Gothenburg University	$\mathrm{rt}$ (2008–17)
PMR	Impact of the regulatory set-up in energy, e-communications and transport sectors ($k$) on 37 industries ($j$) that use the output of these sectors as intermediate inputs ($w_{\mathrm{jk}}$):	OECD REGIMPACT	$\mathrm{cjt}$ (2008–18)
	$\mathrm{REGIMPAC}{T}_{\mathrm{jt}}={\sum }_{k=1}^K\mathrm{PM}{R}_{\mathrm{kt}}{w}_{\mathrm{jk}}$
L flex	Hiring and firing practice index based on the calculated score from the answers of a representative sample of business leaders in their respective countries to the following question: ‘In your country, to what extent do regulations allow flexible hiring and firing of workers? [1 = not at all; 7 = to a great extent]’. The index has been rescaled to 0–100	The Global Competitiveness Index Historical Dataset – World Economic Forum (WEF)	$\mathrm{ct}$ (2007–17)
Access2K	We built a principal component-based weighted index from three separate scores^a to the survey questions related to (1) ease of access to bank loans, (2) access to equity funding for financing innovative and risky projects and (3) access to finance by issuing bonds or shares on the capital market. The index has been rescaled to 0–100	The Global Competitiveness Index Historical Dataset – WEF	$\mathrm{ct}$ (2008–17)
Churn	The sum of the entry and exit rates, measured as the share of the number of firms’ births and deaths over the total population of active enterprises	Eurostat–SBS (BD_HGNACE2_R3)	$\mathrm{rjt}$ (2008–18)
HT empl	Employment in technology and knowledge-intensive sectors as a percentage of total employment	Eurostat (HTEC_EMP_REG2)	$\mathrm{rt}$ (2008–18)
$\mathrm{\Delta }$GDP	Growth rate of gross domestic product (GDP) per inhabitant at purchasing power standard	Eurostat (NAMA_10R_2GDP)	$\mathrm{rt}$ (2008–18)
literacy1930	Share of the population able to read and write in 1930. Data are available only for 100 European regions	Diebolt and Hippe’s (2019) based on based on Kirk’s (1969) data	$r$ (1930)
HC	Number of scientists and engineers aged 15–74 years as a share of the active population	Eurostat–Human Resources in Science and Technology (HRST)	$\mathrm{rt}$ (2008–18)

Note: $r$, Region; $c$, country; $j$, sector; $t$, year.

^aOur preliminary analysis uses the three indicators separately. However, they all yield similar results, which are available from the authors upon request.

Table 2 reports the mean, median, interquartile range (IQR),¹¹ minimums and maximums for our variables of interest. All variables, except the growth rate of GDP per capita, range from 0 to 100. The average share (across all sectors, regions and years) of HGFs shows that only 5.3 in 1000 firms are HGFs, with some sectors in some regions reporting no HGFs. Given that the median is only 3.7 HGF per 1000 firms, it clarifies that there are some outliers in the upper tail of the distribution. The QoG seems to be normally distributed, with a mean and median of 58 and 57, respectively. Its dispersion (IQR/median) is similar to that of the share of scientists and engineers (HC), albeit much greater than that of the churn and literacy rates.

Table 2. Summary statistics.

Variable	Mean	Median	IQR	Minimum	Maximum	N
HGF (%)	0.53	0.37	0.54	0	3.3	9254
QoG	58	57	31	1.8	100	13,820
PMR	4.3	1.8	1.6	0.13	48	14,500
L flex	27	26	19	0	100	14,500
Access2K	42	42	33	0	100	14,500
HC	5	4.7	2.6	1.6	14	11,760
Churn	18	17	7.4	5.1	52	8935
HT empl	3.3	2.9	1.9	0.8	11	12,870
$\mathrm{\Delta }$GDP	1.5	1.9	3.6	–14	12	14,300
literacy1930	79	88	32	29	100	7800

Note: Averages across years, regions (or countries) and sectors.

IQR, interquartile range.

Table 3 reports the correlation coefficients for all the variables used in the analysis. The only two pairs of variables to exhibit a strong correlation ($>\left|0.7\right|$) are access to capital and quality of government, and human capital and high-tech employment. Table 4 gives more insights on the averages of our main variables of interest (HGFs and regulations) by groups of regions. We divide regions into three groups according to their level of government quality (high, medium and low) and into two groups according to whether their level of QoG increased or decreased over the 2010–17 period.

Table 3. Cross-correlation table.

Variables	HGF	QoG	PMR	L flex	Access2K	HC	Churn	HT empl	$\mathrm{\Delta }$gdp
HGF (%)	1.000
QoG	0.192	1.000
PMR	0.332	0.079	1.000
L flex	0.197	0.524	0.051	1.000
Access2K	0.293	0.711	0.082	0.452	1.000
HC	0.168	0.325	0.055	0.432	0.400	1.000
Churn	–0.117	–0.033	–0.203	0.004	0.088	0.080	1.000
HT empl	0.184	0.264	0.039	0.252	0.340	0.750	0.018	1.000
$\mathrm{\Delta }$gdp	0.119	–0.020	0.002	0.299	0.094	0.087	–0.064	0.044	1.000
literacy1930	0.132	0.423	0.012	0.314	0.278	0.292	–0.412	0.466	0.125

Table 4. Averages of high-growth firms (HGFs) (%), product and labour market regulation, by regions group.

QoG	Variable	$\mathrm{\Delta }\mathrm{Qo}{G}_{10-17}<0$	$\mathrm{\Delta }\mathrm{Qo}{G}_{10-17}\ge 0$	Row average	N
High	HGF (%)	0.71	0.27^a	0.67	2315
	PMR	3.76	4.62	3.83
	L flex	38.67	27.90	37.77
Medium	HGF (%)	0.51	0.45	0.50	4421
	PMR	4.63	4.89	4.65
	L flex	21.08	32.04	21.86
Low	HGF (%)	0.42	0.59	0.43	2438
	PMR	4.08	4.87	4.21
	L flex	23.68	33.02	25.28
Column average	HGF (%)	0.53	0.42	0.53
	PMR	4.23	4.82	4.33
	L flex	26.24	31.57	26.77
	N	8721	453	9174

Note: The three levels of the European Quality of Government index (QoG) are defined using the regional average 25th and 75th percentiles. Variables’ averages are across all sectors, years and within groups of regions. The number of observations refers to the minimum available for estimations.

^aOnly three regions in the category high quality of government experienced an increase in quality over the period 2010–17: the small Finnish island Åland and the Spanish Autonomous cities of Ceuta and Melilla. These averages cannot be compared with those of other groups of regions.

The share of HGFs increases with the level of government quality (column ‘row average’). In terms of regulation, while high QoG regions are, on average, less regulated than the low QoG ones, regions with a medium quality of government are the most regulated, with the highest average product market regulation (4.65) and the lowest labour market flexibility (21.86). There are also noticeable differences between regions where the QoG deteriorated versus those where the QoG did not change or even improved (i.e., as mentioned above, these are mainly regions in the Czechia and Poland). In particular, regions where the QoG improved, or at least did not worsen, are more subject to product market regulation while having a higher degree of labour market flexibility (measured as flexibility of hiring and firing practices). The share of HGFs is larger in regions where QoG worsened by 0.07 percentage points (ppt) compared with those regions where it improved. However, it is important to note that the majority of regions where QoG improved started from a ‘low’ QoG level. The HGFs share is much larger in regions with low but positive long-term QoG growth compared with those with low QoG and negative QoG growth (0.59% versus 0.42%, respectively). All differences between these averages are statistically significant at the 0.01 level.

4. EMPIRICAL ANALYSIS

4.1. Estimation strategy

Using open-source data, we analyse whether government quality and regulations affect the share of HGFs in sector $j$ of region $r$ at time $t$ with a linear regression model. More specifically, we estimate both the direct and indirect effects of the quality of institutions. As argued in section 2, the effect of labour and product market regulations on the share of HGFs may depend on the quality of regional institutions. Therefore, our estimating equation is: (1) $\mathrm{HG}{F}_{\mathrm{jrt}}={\text{β}}_0+{\text{β}}_1\mathrm{Qo}{G}_{\mathrm{rt}-1}+{\text{β}}_2^{\mathrm{\prime }}\mathrm{Re}{g}_{\mathrm{jrt}-1}+{\text{β}}_4^{\mathrm{\prime }}{X}_{\mathrm{jrct}-1}+{ϵ}_{\mathrm{jrt}}$(1) The panel indicator is the region–sector pair $\mathrm{rj}$; $\mathrm{HGF}$ is the share of HGFs; $\mathrm{QoG}$ and $\mathrm{Reg}$ are the Quality of Government index as well as product and labour market regulations (PMR and L flex), respectively; $X=(\mathrm{Access}2K,\mathrm{HC},\mathrm{Churn},\mathrm{HTempl},\mathrm{\Delta }\mathrm{GDP},\mathrm{Dummies})$ is a set of control variables, including access to capital, availability of human capital, churn rate, share of high-tech employment and past economic development, measured as growth rate of GDP per capita, as well as industry $j$, region $r$, country $c$ and year $t$ dummies. The control variables are considered to be predetermined variables and we assume their current and lagged values are uncorrelated with the current error term ${ϵ}_{\mathrm{jrt}}$. Additionally, all explanatory variables are lagged by one year.¹² Although we are aware that lagging the explanatory variables does not overcome the problem of reverse causality (Bellemare et al., 2017), it helps to remove the strong and untestable strict exogeneity assumption $E\left({ϵ}_{\mathrm{jrs}}\right|X_{\mathrm{jrt}})=0,\mathrm{\forall s},t$. Indeed, when dealing with institutional variables and economic outcomes, issues of endogeneity may arise (Acemoglu et al., 2014; Eicher & Leukert, 2009; Tabellini, 2010). The main endogeneity issues are reverse causality – institutions can influence the growth of firms in a region and, in turn, institutions can be shaped by the ability of a region to have a higher share of HGFs – and omitted variables that could influence both the dependent variable and the control variables (Cinelli et al., 2022).

To account for endogeneity, we adopt different strategies to select instrumental variables (IV). The first method – generalized method of moments (GMM) estimation of the panel data model with fixed-effects – exploits past values of the endogenous variables. However, the validity of the time lags as instruments may be limited by the high degree of persistence of institutions. Second, among the established instruments for economic institutions, immediate candidates relate to historical variables. Tabellini (2010) argues and shows that historical variables, such as the literacy rate in 1880 and early political constraints on executives, isolate the exogenous variation in culture from the possibly endogenous variation in culture due to the unobserved error term. We adopt the same strategy and use the literacy rate in 1930 as an exogenous determinant of current government quality. Our exclusion restriction assumption implies that the regional literacy rate in the 1930s does not have any direct causal link with the current share of HGFs, but it is positively related to governments and their quality (Porta et al., 1999).

While Tabellini’s (2010) dataset is freely available, it only contains 69 regions, many of which belong to countries that are excluded from our sample. Therefore, we use the literacy rate in 1930 from Diebolt and Hippe (2019),¹³ which covers 81 regions of six countries available in our sample (Austria, Spain, Finland, France, Italy and Portugal), giving a subsample that represents Southern, Western and Northern Europe.

Finally, to estimate the moderating role of the quality of institutions in the relationship between regulations and the share of HGFs, we use both interaction effects ($\mathrm{Qo}{G}_{\mathrm{rt}-1}\times \mathrm{Re}{g}_{\mathrm{jrt}-1}$) (Table 5) and separate regressions for different levels of the index $\mathrm{QoG}$ (three levels chosen according to the 25th and 75th percentiles) (Table 7) and different levels of the growth of the index $\mathrm{QoG}$ ($\mathrm{\Delta }\mathrm{Qo}{G}_{2010-2017}<0$ and $\mathrm{\Delta }\mathrm{Qo}{G}_{2010-2017}\ge 0$) (Table 8).

Table 5. Instrumental variables (IV) estimation (two-step generalized method of moments – GMM).

Dep. var.: HGFs (%)	(GMM)	(GMMNI)	(GMM)	(GMM)
QoG	0.003***	0.003***	0.005***	0.002***
	(0.000)	(0.000)	(0.000)	(0.000)
PMR	–0.007***	–0.007***	–0.011***	–0.012***
	(0.003)	(0.003)	(0.003)	(0.004)
L flex	0.001**	0.001**	0.009***	0.001**
	(0.000)	(0.000)	(0.001)	(0.000)
HC	0.017***	0.016***	0.028***	0.017***
	(0.004)	(0.004)	(0.004)	(0.004)
Access2K	0.001***	0.001***	0.001***	0.001***
	(0.000)	(0.000)	(0.000)	(0.000)
Churn	0.003***	0.003***	0.002**	0.004***
	(0.001)	(0.001)	(0.001)	(0.001)
HT empl	0.007*	0.007*	–0.000	0.007*
	(0.004)	(0.004)	(0.004)	(0.004)
$\mathrm{\Delta }$GDP	0.013***	0.013***	0.009***	0.013***
	(0.002)	(0.002)	(0.002)	(0.002)
L flex#QoG			–0.001***
			(0.000)
PMR#QoG				0.001*
				(0.000)
R^2	0.572	0.571	0.578	0.572
N	5578	5578	5578	5578
Sargan	163.181***	167.984***	138.565***	163.889***
Weak identification F	10,839.445	6041.122	1146.245	6059.654
Endogeneity test	11.143**	13.962***	13.587***	10.834**

Note: Significance: p < 0.01***, p < 0.05**, p < 0.1*. Robust standard errors are shown in parentheses. All right-hand-side variables are lagged by one year. All econometric specifications include year, sector and region dummies.

N, number of observations.

4.2. Estimation results

Table 5 presents results from GMM estimation exploiting past values of the endogenous variables. The direct effect of government quality on the share of HGFs is positive and statistically significant: an increase of 10 points in the QoG score corresponds to an increase of 0.03 ppt in the share of HGFs. Given that the shares of HGFs are very small, with an average across years, sectors and regions of only 0.53%, an increase of 0.03 ppt corresponds to a relative increase of 5.7%. Vice versa, this also means that a decrease in the quality of government corresponds to a reduction of HGFs. In line with previous evidence on the effect of regulation on productivity (Nicoletti & Scarpetta, 2003; Scarpetta & Tressel, 2002), product market regulation is negatively related to the share of HGFs. The interpretation of all regression coefficients is similar, as all variables are scaled 0–100 (except the growth of GDP per capita): an increase in the stringency of product market regulation score of 10 points corresponds to a decrease in the share of HGFs of 0.07 ppt. A 10-point increase in labour market flexibility corresponds to an increase in the share of HGFs of 0.01 ppt.

The estimated coefficients of the control variables are consistent with prior research. More human capital, easier access to capital, faster resource reallocation via firms’ churn, more employment in high-tech sectors, and past economic growth are all positively related to the regional share of HGFs. Human capital and labour market flexibility have the strongest positive relationship with the share of HGFs. An increase of 10 ppt in the regional share of scientists and engineers (labour market flexibility) corresponds to an increase of 0.17 (0.18) ppt in the share of HGFs, more than a 30% increase over the mean value of 0.53%.

As a robustness check, column 2 of Table 5 (GMM${}_{\mathrm{NI}}$, not imputed) reports the results using a different method of imputation for the quality of government. Rather than using interpolation, data for the missing years of 2011, 2012, 2014, 2015 and 2016 are replaced with the value of the previous non-missing observation, while data for 2008 and 2009 are replaced with data from the first survey year 2010. Results are almost identical. Columns 3 and 4 of Table 5 display results from adding an interaction term between the quality of government and either labour market flexibility or product market regulation. The coefficient for the first interaction term is negative and statistically significant, indicating that, as the quality of regional government increases, the positive influence of labour market flexibility on HGFs diminishes. The interaction with product market regulation yields a comparable picture, where the negative influence of such regulation is offset by the higher government quality. The coefficient of the interaction is, however, small and only statistically significant at the 10% level.

As an additional robustness check, we use the literacy rate in 1930 to instrument the quality of government. Table 6 reports the results of the first stage¹⁴ regression in column 1. To test the robustness of our results in Table 5, we run a GMM estimation for the subsample of regions where the literacy rate is available, but without using the literacy rate as an instrument (column 2). Column 3 reports the second stage GMM regression including the fitted value of QoG. The literacy rate is confirmed to be a significant predictor of QoG, along with all other control variables. Columns 2 and 3 look almost identical except for the influence of the quality of government. The coefficient is significant only in the GMM that controls for the endogeneity with a historical variable. As the results for the smaller sample in column 3 are very similar to those in Table 5, our use of the ‘GMM-style’ instruments is supported. Finally, Table 9 reports results from a stepwise panel fixed-effects estimation for comparison.

Table 6. Literacy rate as an instrumental variable (IV).

Dep. var.:	QoG	HGFs (%)	HGFs (%)
literacy1930	0.101***
	(0.010)
QoG		–0.001
		(0.001)
$Q\hat{o}G$			0.002**
			(0.001)
PMR	–0.018	–0.177***	–0.187***
	(0.034)	(0.053)	(0.039)
L flex	0.091***	0.003*	0.004**
	(0.010)	(0.002)	(0.002)
HC	0.295***	0.013*	0.015**
	(0.081)	(0.007)	(0.006)
Access2K	0.076***	0.007***	0.006***
	(0.009)	(0.000)	(0.000)
Churn	–0.018	0.028***	0.025***
	(0.015)	(0.002)	(0.002)
HT empl	0.177	0.016*	0.022**
	(0.143)	(0.009)	(0.010)
$\mathrm{\Delta }$GDP	–0.241***	–0.000	0.001
	(0.029)	(0.002)	(0.002)
R^2	0.969	0.190	0.161
N	5618	4411	3848

Note: The sample used for these estimations includes only regions available in both the Eurostat and the literacy rate datasets (Diebolt & Hippe, 2019), which are regions of Austria, Spain, Finland, France, Italy and Portugal. The first column reports the results obtained from a first regression where we isolate the source of variation of institutional quality explained by literacy rate in 1930. Although not reported, the first regression includes region, sector and year dummies.

Significance: p < 0.01***, p < 0.05**, p < 0.1*. Robust standard errors are shown in parentheses. All right-hand-side variables are lagged by one year. All econometric specifications include year, sector and region dummies.

N, number of observations.

4.2.1. Heterogeneous quality of government

We further explore the heterogeneity of the influence of regulation on HGFs by estimating three separate regressions according to the average regional level of quality of government: high, medium and low. The values that define the levels are chosen using the 25th and 75th percentiles. Results are reported in Table 7. The influence of regulation on the regional share of HGFs greatly varies between the three groups of regions.

Table 7. Instrumental variables (IV) generalized method of moments (GMM) estimation, by level of the European Quality of Government index (QoG).

Dep. var.: HGFs (%)	High QoG	Medium QoG	Low QoG
PMR	–0.015	–0.111*	–0.050***
	(0.054)	(0.065)	(0.015)
L flex	0.003	0.006***	0.001
	(0.005)	(0.002)	(0.001)
HC	0.067***	0.000	–0.012
	(0.008)	(0.010)	(0.012)
Access2K	0.003***	0.007***	0.002***
	(0.001)	(0.001)	(0.000)
Churn	0.030***	0.029***	0.003**
	(0.004)	(0.003)	(0.001)
HT empl	–0.020	0.054***	–0.025
	(0.014)	(0.019)	(0.015)
$\mathrm{\Delta }$GDP	–0.007***	0.002	0.009***
	(0.003)	(0.003)	(0.002)
RMSE	0.209	0.192	0.148
R^2	0.142	0.306	0.075
N	1396	2779	1480
Cragg–Donald Wald F	26.904	8.640	8.828

Note: Significance: p < 0.01***, p < 0.05**, p < 0.1*. Robust standard errors are shown in parentheses. All right-hand-side variables are lagged by one year. All econometric specifications include year, sector and region dummies.

N, number of observations; RMSE, root mean square error.

The share of HGFs in regions with high quality of government is neither affected by the level of product market regulation, nor by more or less flexibility in hiring and firing practices. Among the control variables, access to capital, the churn rate, and human capital have a strong positive association with HGFs, with the latter variable being only significant in regions with a high quality of government. Unlike regions with high-quality administration, in regions with a medium quality of government (which is 50% of our sample), both product and labour market regulation regression coefficients have the signs commonly found in the literature: a negative relationship between the level of product market regulation and the share of HGFs, alongside a positive relationship between labour market flexibility and HGFs. In addition, in regions with medium quality of government, while human capital has no influence in such regions for HGFs, the churn rate and access to capital continue to be important factors, as well as the employment share in technology intensive sectors. Thus, the well-established relationship that less regulation and more flexibility are linked to indicators of economic growth holds for regions where the quality of government is neither high nor low.

Regions with low government quality again differ from regions with a medium quality of government. There, only product market regulation is found to be statistically significant and negative: lower regulation of product markets is associated with a higher share of HGFs. More labour market flexibility is not associated with higher share of HGFs. Moreover, in these regions, past economic development (growth of GDP per capita) is, in contrast to all other regions, a strong factor influencing the increase in HGFs (a 10 ppt increase in the GDP growth rate corresponds to almost a 0.10 ppt increase in the HGFs share, which is a 23% increase over the average for low QoG regions, 0.43%). Again, in contrast to the other two groups of regions, neither the availability of human capital nor the employment share in technology and knowledge intensive sectors correlate with higher shares of HGFs, indicating that the kind of HGFs developed there, may differ with respect to the firm strategy and organisation. Most importantly, however, increases in labour market flexibility do not correspond to higher HGFs shares if the implementation is in the hands of a poorly functioning public administration.

4.2.2. Changes in the quality of government

In the final part of our analysis, we divide the sample into two groups of regions, those experiencing a deterioration in the quality of public institutions between 2010 and 2017, $\mathrm{\Delta }\mathrm{Qo}{G}_{10-17}<0$ (which as can be seen in Figure 4 corresponds to the vast majority), and those which maintained or even bettered their quality, $\mathrm{\Delta }\mathrm{Qo}{G}_{10-17}\ge 0$. Table 8 displays the corresponding results. In regions where the quality of institutions decreased, rigidities in both product and labour markets have a negative influence on HGFs. In other words, the expected associations between the different kinds of regulations and the number of HGFs in a region are in place.

Figure 4. European Quality of Government index (QoG), by country, 2010 (vertical axis) and 2017 (horizontal axis).

Source: Authors’ calculations based on the European Quality of Government index (QoG) from the QoG Institute, Gothenburg University.

For each country, a scatter plot of regional QoG indices in 2010 (vertical axis) and in 2017 (horizontal axis), and a 45 degree line across each plot.

Table 8. Instrumental variables (IV) generalized method of moments (GMM) estimation, by European Quality of Government index (QoG) dynamics.

Dep. var.: HGFs (%)	$\mathrm{\Delta }\mathrm{Qo}{G}_{2010-2017}<0$	$\mathrm{\Delta }\mathrm{Qo}{G}_{2010-2017}\ge 0$
PMR	–0.048***	–0.044
	(0.016)	(0.077)
L flex	0.004***	0.005
	(0.001)	(0.004)
HC	0.035***	0.084***
	(0.004)	(0.029)
Access2K	0.004***	–0.001
	(0.000)	(0.002)
Churn	0.017***	0.011**
	(0.001)	(0.005)
HT empl	–0.001	0.063***
	(0.008)	(0.022)
$\mathrm{\Delta }$GDP	0.002**	0.012**
	(0.001)	(0.006)
R^2	0.127	0.493
N	5535	120

Note: Significance: p < 0.01***, p < 0.05**, p < 0.1*. Robust standard errors are shown in parentheses. All right-hand-side variables are lagged by one year. All econometric specifications include year, sector and region dummies.

N, number of observations.

Table 9. Fixed effects estimation results.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
QoG	–0.004***	–0.003***	–0.003***	–0.000	–0.001**	–0.001*	–0.001	–0.001
	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
PMR		–0.029***	–0.029***	–0.022***	–0.024***	–0.031***	–0.024***	–0.024***
		(0.006)	(0.006)	(0.006)	(0.006)	(0.007)	(0.006)	(0.007)
L flex			0.001**	0.006***	0.006***	0.005***	0.005***	0.005***
			(0.001)	(0.001)	(0.001)	(0.001)	(0.001)	(0.001)
HC				0.025***	0.028***	0.036***	0.037***	0.037***
				(0.006)	(0.006)	(0.006)	(0.006)	(0.006)
Access2K					0.002***	0.003***	0.002***	0.002***
					(0.000)	(0.000)	(0.000)	(0.000)
Churn						0.014***	0.014***	0.014***
						(0.001)	(0.002)	(0.002)
HT empl							–0.004	–0.005
							(0.009)	(0.010)
$\mathrm{\Delta }$GDP								–0.000
								(0.001)
R^2	0.024	0.030	0.031	0.053	0.070	0.097	0.093	0.093
N	9302	9302	9302	7916	7916	7111	6688	6648

Note: Significance: p < 0.01***, p < 0.05**, p < 0.1*. Robust standard errors are shown in parentheses. All right-hand-side variables are lagged by one year. All econometric specifications include a constant, not reported.

N, number of observations.

When we focus on regions where institutional quality improved, we observe that the level of regulations does not matter. We interpret this in the sense that, in these regions, where regional governments made efforts to increase the quality of public services, regulation is perceived less as an obstacle by firms, but rather as a necessary part of a well-functioning institutional framework. Compared with the first group, the regional share of HGFs is mainly influenced by human capital, employment in high technology sectors, and past economic growth.

5. DISCUSSION AND CONCLUSIONS

One of the most important steps in the history of the EU was the envisioning and creation of the single market. The free movement of capital, goods, services and people has brought many advantages to its member states, as businesses benefit from easier access to markets across the EU (European Commission, 1996). In this context, it is often argued that a common regulatory framework at the European level with harmonised procedures for starting, running, and ending a business would be beneficial for all member states. Yet, leading European business associations lament the uneven integration and enforcement of EU rules across the trading bloc. In a joint statement, BusinessEurope, DIGITALEUROPE, ERT, Eurochambres and EuroCommerce¹⁵ claim that, ‘EU legislation too often allows for differentiated transposition in EU member states and the European Commission’s enforcement policy is lacking teeth against member states which introduce national rules or administrative requirements leading to further market fragmentation’. In this paper, we posit that it is not only the geographical heterogeneity of regulations, but also how efficiently the responsible regional governments implement these regulations that have a direct influence on the regional economic performance and may explain the heterogeneity in the shares of HGFs across European regions. Primarily using open-source data, we analyse how the heterogeneity in the quality of government influences the relationship between regulation and the share of HGFs at the regional level across the EU.

Exploiting the heterogeneity of regulations and of quality of public administration across regions, we show that the relationship between regulations and the share of HGFs depends on the quality of government administrating these regulations. Our results allow for the interpretation that, in regions with a high quality of government, labour and product market regulations are absorbed in a fast, efficient, and impartial way, providing firms with market stability and legal certainty without imposing unnecessary costs on businesses. In these regions, the regional administration creates an environment where high levels of regulation are not harmful to the formation of HGFs, as less flexible hiring and firing regulations are not related to lower shares of HGFs.¹⁶ As mentioned in section 2, one of the essential conditions for firm growth is the ability to recruit and train qualified personnel. HGFs and firms in general are more likely to attract skilled workers when they are able to guarantee a stable job and are willing to invest in upgrading their skills. Similarly, regulation of the product market does not hinder the formation of HGFs when in the hands of a regional administration of high quality, as it is able to efficiently process the number of bureaucratic steps needed to manage HGFs. When in the hands of a high quality of government, regulation can be considered as good for the regional development of HGF.

In regions with medium and, to a lesser extent, in regions with low quality of government, we find that our results converge with the majority of empirical studies analysing regulations and various indicators of economic performance, that is, the less regulation the better. In these regions, more regulated markets are associated with a lower share of HGFs. The less efficient implementation of laws and regulations may increase costs for firms, stunting their growth, thus more regulation is bad for the formation of HGFs. Still, more flexibility in hiring and firing is related to higher shares of HGFs only in regions with medium government quality. This positive relationship disappears in regions with low quality of government, where things could turn ugly. In such regions, regional governments may be implementing regulations in such an unfair or inefficient way that businesses fail to perceive the more flexible labour market regulations as an improvement of the regulatory environment.

Hence, our results confirm the assumptions present in both the public interest theory and the public choice theory regarding the influence of bureaucracies and regulations on firm growth. The findings of our paper suggest that, for firms, unnecessary delays due to time consuming procedures may result in foregone chances to become HGFs, connected to risks of losing to competitors. Delays caused by the need to process many regulations by slow moving public administration, corruption and personal interests, erroneously implemented regulatory requirements, and related procedures can cause substantial costs to growing business. Thus, owners of potential HGFs are likely to respond directly to regulatory barriers caused by low quality public administrations, ultimately not turning their firms into HGFs or maybe moving their firms to other regions. In that sense, it is the quality of regional government that determines whether regulations unfold good, bad or ugly influence on the regional share of HGFs.

Our study has its own limitations. First, we are limited by the quality and availability of data. Without proper IV, it is nearly impossible to identify our regression coefficients and make claims of causality. We used the historical literacy rate as an instrument for the quality of institutions, but this is available only for a subset of regions. Moreover, we are aware that we are not able to capture and solve all endogeneity issues, preventing us from making causal interpretations of our results. One way forward could be the use of synthetic variables generated by the spatial filtering of latent patterns embedded in the spatial structure of regional data (Le Gallo & Páez, 2013). Another point of discussion derives from the fact that we use our own definition of HGFs and not the one presented by Eurostat (see section 3). It would be instructive to measure the differences in the regression coefficients using these two different measures of HGFs. This exercise, however, goes beyond the scope of our research question and it is left for future research.

Nonetheless, our results provide new insights into the relationship between regulation, quality of government and HGFs. The quality of regional governments is one important ingredient when evaluating the influence of regulation. A key question for future research is to better understand the enabling factors triggering the creation of a high-quality government. So far, a solid understanding and empirical evidence of the background factors or long-run policies that may lead to a sustainable improvement of government quality is missing. A future avenue for research in this regard could be examining the role of financial independence of municipalities and local budgets. In Nordic countries, income taxes are paid to municipalities, triggering competition between local governments for tax-generating businesses while avoiding costly competition between municipalities for transfers from central government, which can potentially reduce social efficiency, as shown in experimental work on intergroup contests for rents (Abbink et al., 2010) between municipalities for grants. As Herrmann (2022) argues, the competition between local governments for tax-generating businesses could induce economic growth by providing better services to local businesses. This suggests that market-preserving federalism (Weingast, 1995) could be the mechanism behind the sustainable transition from public choice type of governments, typically characterised by low and medium quality of government, to high-quality public interest governments.

Our results also offer insights for European, national and regional policymakers. To achieve convergence in the shares of HGFs across the EU, firms should operate on a level playing field, with fair and equal conditions for competition, and with high-quality governments. Our results confirm that efforts to increase the quality of government across the EU are needed to further increase economic cohesion across the Union (Rodríguez-Pose & Ketterer, 2020). We show that the quality of regional government interacts with how regulations might be perceived by HGFs. At the same time, our finding that, in regions with high-quality government, labour market flexibility is not positively associated with the growth of HGFs indicates that the success of policies aiming to create an ‘economy that works for people’ in the EU (European Commission et al., 2020) also needs high-quality government across the EU. Our paper provides further insights for those policymakers seeking to generate firm growth in regions with low government quality. The deregulation of labour markets might lead to no increase in the share of HGFs unless they start putting effort into improving the quality of regional governments. In that regard, policymakers have to make strategic decisions on whether to invest into efforts that reduce the burdens for firms connected to their regulatory environment or into the quality of their administrative processes, as these investments may also have an impact on what kind of HGFs will be attracted in their regions.

ACKNOWLEDGEMENTS

For their valuable comments, we thank Federico Biagi, Clemens Domnick, Tomaso Duso, Paolo Pasimeni, Francesco Quatraro, Alexander Schiersch and Caroline Stiel, as well as the participants at the DIW Berlin, University of Potsdam seminars, and the University of Economics in Bratislava (EUBA), the 5th Workshop on the Creative Spark in Berlin, the 25th INFER Annual Conference in Valencia, the Concordi 2023 Conference in Seville, and the 2024 Conference of the Verein für Socialpolitik in Berlin. The views expressed are purely those of the authors and may not in any circumstances be regarded as stating an official position of the European Commission or of DIW Berlin.

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1. HGFs are companies with a minimum of 10 employees that are able to sustain a growth of employment or turnover of 20% for three consecutive years.

2. Haltiwanger et al. (2013) provide evidence that HGFs contributed significantly to productivity and job growth in the 1990s and 2000s. Decker et al. (2016) argue that the reported decrease in the rates of HGFs is one of the main contributing factors to the slowdown in US productivity since 2010.

3. See the CoR publication from the 2010 Lisbon Treaty (CoR, 2010, p. 3) as well as CoR (2019b).

4. Matthias Duschl et al. (2015) report results in a similar direction for Germany.

5. The data on regional business demography are freely available at https://ec.europa.eu/eurostat/web/regions/data/database/.

6. This calculation of regional shares of HGFs differs from Eurostat’s. Eurostat’s figures on HGFs shares are defined as the number of HGFs divided by the (smaller) number of active enterprises with at least 10 employees (source: https://ec.europa.eu/eurostat/web/products-eurostat-news/-/ddn-20171019-1). Excluding micro-enterprises (firms with nine employees or fewer), however, inflates the shares of HGFs in countries with large proportions of micro-enterprises, that is, $\mathrm{HGF}{s}_{\mathrm{rt}}/\mathrm{Active}\mathrm{Firms}<\mathrm{HGF}{s}_{\mathrm{rt}}/\mathrm{Active}\mathrm{Firm}{s}_{\ge 10\mathrm{empl}}$, because the denominator $\mathrm{Active}\mathrm{Firms}$ is larger than $\mathrm{Active}\mathrm{Firm}{s}_{\ge 10\mathrm{empl}}$. Therefore, to obtain a realistic and unbiased representation of the share of HGFs, we use the total number of active firms as a denominator.

7. The data are freely available at https://www.gu.se/en/quality-government/qog-data/data-downloads/european-quality-of-government-index/.

8. This is not a typo. Transport and warehousing segments are the most affected by entry regulation and public ownership.

9. The sector PMR indicators measure the regulatory barriers to firm entry and competition at the level of individual sectors, with a focus on network industries, professional services and retail distribution (https://stats.oecd.org/index.aspx?DataSetCode=PMR).

10. The dataset is available by searching online for ‘The Global Competitiveness Index Historical Dataset 2007–2017 World Economic Forum’.

11. The IQR measures the middle spread of the data. The ‘Quartile coefficient of variation’ – IQR/median – is a useful alternative to the more standardly used coefficient of variation (standard deviation/mean), because it gives a dimensionless, thus comparable among variables, measure of relative dispersion for non-normally distributed variables.

12. We estimated the same regression model without lagging the explanatory variables and there are no major differences in the sign or the statistical significance of the coefficients. The additional estimates are available from the authors upon request.

13. This dataset is not open access, but available from the authors upon request.

14. The literacy rate is a cross-sectional variable, so we cannot include it in a standard panel fixed effects estimation, as it would drop.

15. See https://www.eurocommerce.eu/2022/06/businesses-call-for-fresh-political-engagement-to-renew-economic-integration-in-the-single-market/.

16. These results are in contrast, yet not comparable, with Davidsson and Henrekson’s (2002) findings for Sweden, because they use firm-level data and specifically look at small firms, which unfortunately we cannot do due to the level of aggregation of our data. Further, our sample of regions with a high quality of government includes many Nordic regions of Sweden and Denmark, but also regions of Austria, France, the Netherlands and Spain. Perhaps an analysis of Swedish start-ups and HGFs, accounting for the regional government quality, could yield results comparable with ours.