Boosting Business in Regions

Easing Administrative Burdens and Financing Constraints

Report

OECD Regional Development Studies

24 December 2025

Download PDF

2. Boosting business in regions by improving access to finance

Copy link to 2. Boosting business in regions by improving access to finance

Abstract

The chapter examines the critical role of firms’ access to finance in boosting business in regions. It first provides an overview of the financing conditions across OECD regions, with a particular focus on SMEs, which are often the most constrained by local financing conditions. It then presents evidence on how competitive banking markets influence firms’ economic performance, including firm entry, survival and growth, as well as labour productivity and labour productivity growth and overall economic outcomes at the regional level.

In Brief

Copy link to In Brief

Boosting business by improving access to finance

Despite its crucial role in fostering regional economic development, access to finance remains a significant challenge for many firms, particularly SMEs. In half of the 229 large (TL2) regions with available data across OECD member and non-member countries, at least one in ten firms faces difficulties obtaining finance. One in three firms struggles in the 10% of regions with the poorest access to finance. Common barriers include providing financial information, challenges in assessing a project’s or firm’s potential and inadequate tangible assets to use as collateral. These challenges disproportionately affect young firms and SMEs owing to limited tax filings, shorter credit histories, and fewer assets. The issue is further exacerbated in less familiar industries, where financial institutions may find it difficult to evaluate risks and revenue streams effectively.

SMEs face additional challenges in accessing finance due to their greater dependence on local financing conditions, as they typically operate within a single or a limited number of regions and obtain funding locally. Debt finance, the main financing source for SMEs, is often constrained by limited competition in the banking sector. In turn, large differences in the degree of competition in the local banking markets contribute to the significant variation in SME lending across OECD regions, which ranges more than 10 percentage points between the minimum and maximum regional value in each of the five countries with available data. In 46% of the 1 009 small OECD regions with available data, concentration (a commonly used proxy for competition) exceeds the threshold set by the Horizontal Merger Guidelines of the US Department of Justice, indicating high levels of concentration in local banking markets or low levels of competition. A country like Austria, for example, with local banking systems organised around many local banks, exhibits lower levels of local bank concentration than Greece, with a few nationwide banks operating across all regions. However, the level of bank concentration in a region is not solely determined by whether banks are local or nationwide.

Even within countries, regional differences in SME lending can be stark. For example, in metropolitan France, the share of SME lending varies by up to 18 percentage points across regions. This pattern reflects a broader OECD trend in which metropolitan regions generally offer more favourable financing conditions than non-metropolitan areas. In metropolitan regions, lower levels of bank concentration together with shorter distances to lenders, which facilitate the sharing of soft information that SMEs heavily rely on, allow SMEs to access finance more easily. By contrast, non-metropolitan regions face more concentrated banking sectors, which can limit SMEs' ability to find lenders familiar with their business models, especially for innovative or non-traditional firms.

To boost business in regions, governments should ensure all viable firms can access the external finance they need. Regions with lower levels of bank concentration, where banking activity is more evenly distributed across multiple institutions, tend to benefit from easier access to finance, leading to greater firm dynamism, higher employment and faster labour productivity growth. Comparing the Spanish region of Valencia with Lower Silesia in Poland highlights the role of competitive banking markets. Although both regions have a similar number of banks (51 in Valencia and 49 in Lower Silesia), the share of firms banking with the top three banks is significantly higher in Valencia (56%) than in Lower Silesia (less than one-third). Reducing bank concentration in Valencia to levels observed in Lower Silesia could translate into substantial economic benefits, including an estimated 11 000 additional businesses, 120 000 new jobs (0.2 additional employees per establishment on average) and a 1.3 percentage point increase in labour productivity growth.

Boosting business by improving access to finance in regions

Copy link to Boosting business by improving access to finance in regions

Investment plays a crucial role in enhancing economic activity by allowing firms to scale up their production capacity and expand their operations to other sectors. By facilitating innovation and the adoption of new technologies, investment enables firms to produce new products, reduce costs or increase quality. Over time, investment enables the accumulation of productive capital, which enhances competitiveness and drives productivity improvements, fundamental for achieving sustained economic growth and rising living standards.

Nearly half the firms need external finance to invest but not all firms in need obtain it

Between 40% and 50% of firms relied on external finance to support their investment activities in 2023.1 Specifically, the investment of 42% of firms in the European Union (European Investment Bank, 2024[1]), 46% in the UK (British Business Bank, 2024[2]) and 50% in the United States (European Investment Bank, 2024[1]) depended on external finance. Accessing external financing is, however, challenging for firms in many countries and regions. In half of the 229 large (TL2) regions of the 31 OECD Member and accession candidate countries with available data, at least one in ten firms (11% or more) find that access to finance is a severe obstacle to their activity. In the 10% of regions where firms are most affected by obstacles in accessing finance, this number grows to at least 35% of firms (Figure 2.1).

Figure 2.1. In half of OECD regions at least 11% of firms face challenges with access to finance
Copy link to Figure 2.1. In half of OECD regions at least 11% of firms face challenges with access to finance

The regional share of firms that find access to finance a severe obstacle (%), 2018-21

The limited availability of financial information and the hard-to-assess potential of a project are some of the most common examples preventing viable firms from accessing their desired borrowing levels (Crawford, Pavanini and Schivardi, 2018[3]; Stiglitz and Weiss, 1981[4]). The first affects SMEs and young firms more than large and established ones due to their limited tax filing requirements and typically shorter credit histories.2 The second affects firms in certain industries more, as financial institutions may not be aware of the prevailing business model, the revenue streams or the potential risks, thus limiting their willingness to lend to firms in these industries. The use of collateral, where borrowers provide assets as guarantees to lenders, might alleviate credit constraints. Using collateral, however, might not be feasible for the firms that find the most difficulties in accessing external finance, such as firms with insufficient assets, common for SMEs, or firms with assets whose value is uncertain, such as firms owning mostly intangible assets (e.g. brand reputation, strong customer relationships or licenses and permits) perpetuating their limitations in accessing finance.3

The volume of bank loans to SMEs varies across OECD regions. Within OECD countries, bank loans, the primary source of SME finance (Box 2.1), represent between 6% to 22% of regional GDP in the median regions of Czechia and Belgium or Portugal, respectively.4 The regional differences in SME access to finance are even larger when considering non-OECD member countries such as Indonesia and Peru, where in the median region the share of outstanding SME loans is 3% and 2%, respectively. The large cross-country regional differences are paralleled by significant within-country regional variation, with the largest regional variation displayed in France (with 18 percentage points) and Italy (with 14 percentage points) among the countries with available data (Figure 2.2).

Box 2.1. Bank loans remain by far the largest source of external finance

Copy link to Box 2.1. Bank loans remain by far the largest source of external finance

Finance provided through bank loans is 5 to 50 times more than finance provided through non-bank debt and equity finance across OECD countries. In France, outstanding SME loans were EUR 264 billion in 2019, which represented a fifth of all outstanding business loans and more than 40 times the amount of VC funding in the country. In Spain, SME lending was EUR 219 billion in 2019, or half of all outstanding business loans and 50 times the amount of VC lending. Even in countries with the largest equity markets, bank loans remain a very important source of finance. In the United States, total SME loans were USD 645 billion in 2019, less than a fifth of total outstanding business loans but close to five times more than total VC and growth capital. In the United Kingdom, outstanding SME loans were GBP 167 billion, representing a third of total outstanding business loans but 20 times more than VC and growth funding, which stood at GBP 8 billion.

Source: OECD (2022[5]), Financing_SMEsand Entrepreneurs 2022: An OECD Scoreboard

Figure 2.2. SME lending amounts to at least 5% of regional income in OECD regions and differences across regions are substantial
Copy link to Figure 2.2. SME lending amounts to at least 5% of regional income in OECD regions and differences across regions are substantial

Regional outstanding SME loans as a percentage of regional GDP, 2021 (%)

SME access to finance is about local access to finance

Firms tend to have very stable relationships with their banks over time (Kalemli-Özcan, Laeven and Moreno, 2022[6]; Chodorow-Reich, 2013[7]), with firms typically relying on banks that operate within their geographical area. This means that availability of banks in different regions and competition among them can affect lending conditions. Compared to SMEs, larger firms are less affected by the local financing conditions, given their ability to draw on internal resources and to access diverse external financing options such as bonds and equity. Furthermore, larger firms are typically active in different jurisdictions, which allows them to optimise their borrowing decisions across multiple locations. In turn, SMEs are more geographically bound than large firms, limiting their ability to diversify funding sources across regions and countries, making them more exposed to local conditions (OECD, 2019[8]; OECD, 2022[5]).

Across the OECD, the number of active banks with whom firms work greatly differs among the 128 large TL2 regions with available data, ranging from 5 in the Greek regions of Western Macedonia, Western Greece and Peloponnese to 909 in the German region of Bavaria.5 New digital technologies and the significant advancements in Information and Communication Technologies (ICT) over the past decades were expected to reduce the impact of local conditions by facilitating remote communication and standardised information sharing. However, there is no strong evidence that firms rely more on banks that operate online only rather than traditional branch-based banking relationships. Online-only banks served as the main bank for at most 0.3% of firms across the 128 TL2 regions within the 13 OECD countries with available data in 2019. In 75% of regions, it was fewer than 0.01% of firms for which their main bank was an online-only bank.6

The degree of competition in local banking markets shapes access to finance in regions

Copy link to The degree of competition in local banking markets shapes access to finance in regions

Market power concentrated in a few banks means worse access to finance for firms. Competition prompts banks to broaden their range of services, lower their cost and accept more risk in their portfolio, all of which have positive effects on small business lending (Vives, 2001[9]). Greater market power in the banking sector reduced investments by SMEs due to finance constraints in Europe between 2005-08 and increased the share of firms struggling to access finance across 53 countries, including 11 OECD countries between 2002-10 (Ryan, O’Toole and McCann, 2014[10]; Love and Martínez Pería, 2014[11]).

Bank concentration is a proxy measure of bank competition. It is calculated based on the share of firms in a region that use the same bank, using the Herfindahl-Hirschmann Index (HHI),7 a common measure for concentration (see Annex 2.A). Bank concentration, therefore, captures the degree to which a relatively small number of banks control a large share of the banking market.8 A high bank concentration is typically associated with low levels of competition among banks, which translates into fewer and worse borrowing options for firms. Across OECD regions, the greater the bank concentration, the greater the share of firms that find accessing finance a severe obstacle to their activity (Figure 2.3).

Figure 2.3. Greater bank concentration is associated with more firms with difficulties accessing finance
Copy link to Figure 2.3. Greater bank concentration is associated with more firms with difficulties accessing finance

Share of firms (%) finding access to finance a severe obstacle to their activity and bank concentration in TL2 regions, 2019

Bank concentration is substantial across OECD regions

In half of the 1 009 small (TL39) regions, the banking market is substantially concentrated at the top, with at least two-thirds of firms using one of the three largest banks in each region as their main bank (Annex Figure 2.A.3). Using the HHI, a broader measure of concentration that considers the whole distribution instead of just the top or bottom part of it, indicates high concentration10 (above 0.18) in close to half (46%) of the 1 009 OECD TL3 regions with available data with significant regional variation across and within countries. High levels of bank concentration (HHI above 0.18) are exhibited in all regions of Ireland, Latvia and Greece and in around half the regions in Germany (56%), Japan (47%) and the United Kingdom (46%). Followed by Spain, with 42% of its regions displaying high levels of bank concentration, Slovenia (42%), Poland (27%), Portugal (24%), France (18%), Austria (6%) and Hungary (0%) with no region with a highly concentrated banking sector. The largest difference between regions with high or low concentration is found in Germany (Figure 2.4). This is partly due to Germany’s large number of TL3 regions compared to other countries, making regions smaller and more heterogeneous.

Figure 2.4. Most OECD regions show highly concentrated banking sectors
Copy link to Figure 2.4. Most OECD regions show highly concentrated banking sectors

Regional bank concentration as measured by the Herfindahl-Hirschman index

A competitive regional banking sector does not necessarily require a large number of local banks, a sufficient presence of nationwide banks can also ensure competition. In other words, whether a region’s banking sector is dominated by local or nationwide banks (Box 2.2) does not, in itself, determine the level of bank concentration. While Austria, which has the lowest bank concentration, also has a high number of local banks, and Ireland, with the highest concentration, has none, this pattern is not uniform across countries. In Germany, for example, despite the strong presence of local banks, concentration remains higher than in Poland, where local banks are less prevalent. The median region in Germany has a Herfindahl-Hirschman Index (HHI) above 0.18 (0.19), compared to 0.14 in the median Polish region. For instance, in Poland, the moderate degree of bank concentration is owed to the existence of both local and nationwide banks, which have played a role in private sector development. The ensuing development of a strong network of small local banks led to better financing conditions for SMEs in the 2000s, however, new business creation was only supported by the growing number of nationwide banks in the country (Hasan et al., 2018[12]).

Box 2.2. OECD countries differ in how the banking market is organised

Copy link to Box 2.2. OECD countries differ in how the banking market is organised

Across OECD TL3 regions, the banking sectors are organised around local or nationwide banks (Figure 2.5). Nationwide banks, defined as banks present11 in at least 90% of the TL3 regions in a country, either represent less than one in four banks in the median region of Japan, Germany and Austria; or more than 80% of the total number of banks in the median region of the other 10 countries for which data are available (i.e. Poland, United Kingdom, Spain, France, Greece, Ireland, Latvia, Portugal, Hungary and Slovenia).

Figure 2.5. Banking sectors in regions organise around local or nationwide banks
Copy link to Figure 2.5. Banking sectors in regions organise around local or nationwide banks

Number of local and nationwide banks in the median region of each country

Firms in metropolitan regions have more banking options

Metropolitan regions have relatively low levels of bank concentration, on average, with an HHI of 0.17 and 0.16, depending on whether the banking sector is organised around nationwide or local banks, respectively. In contrast, in regions without a metropolitan area, the banking sectors are more highly concentrated with an average HHI of 0.2. Firms in metropolitan regions bank with close to twice as many banks as firms in regions without a metropolitan area (Table 2.1), partly explaining the lower bank concentration in metropolitan regions. A wider choice of financial institutions, familiar with a broader range of business models, makes external finance accessible to a more diverse set of firms. Conversely, in regions with fewer financial institutions, the chances of finding one aware of certain business models are lower. This explains why innovative firms are more likely to have their finance applications rejected and to become “discouraged borrowers” in remote regions, even after accounting for credit scores (Lee and Brown, 2016[13]).

	Group 1 (Austria, Germany and Japan)		Group 2 (France, Greece, Hungary, Ireland, Latvia, Slovenia, Spain, Poland, Portugal and the United Kingdom)
	Herfindahl Hirschman Index (HHI)	Number of banks	Herfindahl Hirschman Index (HHI)	Number of banks
Metropolitan region	0.17	93	0.16	30
Large metropolitan region	0.15	132	0.17	36
Metropolitan region	0.18	76	0.16	26
Region without a metropolitan area	0.20	59	0.20	17
Region near a metropolitan area	0.20	60	0.18	21
Region with/near a small-medium city	0.19	53	0.18	18
Remote region	0.21	60	0.22	12

Note: Group A: Information refers to 482 small (TL3) regions from AUT (35), JPN (47) and DEU (400), excluding regions for which there are less than 14 observations i.e. Landau in der Pfalz, Kreisfreie Stadt (DEB33) in Germany. Group B: Information refers to 527 small (TL3) regions from POL (73), HUN (20), FRA (96), SVN (12), PRT (25), ESP (57), GBR (179), LVA (6), GRC (51) and IRL (8), excluding regions for which there are less than 14 observations i.e. Evrytania (EL643) in Greece. Banking information referring to 2019.

Source: OECD calculations based on Orbis.

Being close to lenders improves borrowing conditions when banking sectors are competitive

Shorter distances facilitate the sharing of qualitative (soft) information on which SMEs rely relatively more than large firms to secure loans, strengthening the advantage of metropolitan areas in providing access to finance (Box 2.3). This challenge is further exacerbated when the distance between a bank branch and its headquarters is significant, particularly for large financial institutions with centralised decision-making at the headquarters12 (Baker, 1999[14]). The greater the disparity between local conditions and those at the headquarters, the harder it is for the headquarters to understand the local economy in which the firm operates (Alessandrini, Cocci and Zazzaro, 2005[15]). As a result, centralised financial institutions are less inclined to lend to SMEs (Zhao, Luintel and Matthews, 2020[16]). Small local banks can thus have a comparative advantage in SME lending, especially in remote regions, by having fewer hierarchical layers between headquarters and local branches and holding a closer relationship with firms.

Box 2.3. The distance to the lender shapes the nature of the lending-borrowing relationship

Copy link to Box 2.3. The distance to the lender shapes the nature of the lending-borrowing relationship

Distance affects the type of information that can be collected about potential borrowers. The shorter the distance between lenders (bank officers) and borrowers (businesses), the easier it is to collect qualitative information, also known as “soft information”, about the repayment capacity of the borrower, which is the foundation of relationship lending. Conversely, the greater the distance, the stronger the reliance on hard (or quantitative) information and the more transactional the lending relationship becomes. In transactional lending, the lending decision is based only or mostly on hard information that can be standardised, for example, using credit scoring methodologies. Since hard information is not as readily available for many SMEs (given their shorter credit histories and limited filing requirements), geographical distance might be more damaging to the credit access of SMEs rather than large firms and during downturns, when relationship lending can be pivotal to avoid credit crunches.¹

In practice, lending to SMEs relies on both relationship and transactional lending (Berger and Udell, 2006[17]). In the context of the United States, for example, Berger and Frame (2007[18]) find that credit scoring methods support small business lending, although the effect is more pronounced when this technology is combined with soft information stemming from closer interactions between lenders and borrowers. Flögel (2018[19]) also finds that both approaches are commonly used for SME lending in Germany, with the balance between the two changing based on the type of bank and the phase of the economic cycle. Partly, the evidence pointing to the mixed use of relationship and transactional lending by SMEs might be explained by the increased use of ICT, which reduced the importance of relationship lending in favour of transactional lending by making it easier to share and rely on “hard information” in the lending decision-making process. For this reason, ICT tools might have made the overall SME lending process more efficient (Petersen and Rajan, 2002[20]).²

¹ Relationship lending is especially important during economic downturns, when it can help stave off credit crunches by extending loans to viable firms, possibly with better conditions. Research from France and Italy reveals some nuances. Banks that rely on relationship lending charge higher interest rates in good times, as relationship lending is more human-resource-intensive than transactional lending. However, they also charge lower interest rates in bad times, when lending decisions based solely on statistical models are more likely to result in loan rejections and higher interest rates (Beatriz, Coffinet and Nicolas, 2022[21]; Bolton et al., 2013[22]).

² Petersen and Rajan, (2002[20]) for example, argue that greater availability of public information on creditworthiness (i.e. hard information) – e.g. if clients are on time with their trade credit payments or other existing loans – should make soft information and relationship lending less important and the overall SME lending process more efficient, including through the reach of a greater number of SMEs that it should be easier to monitor.

Proximity and competition among banks interact. Evidence from the United States suggests that proximity improves the chances of a firm to secure a loan but it also raises interest rate compared to firms that are located further from a branch with the same risk profile. If the firm is, in contrast, located closer to a branch of a competitor, they are less likely to receive a loan but if they do the interest rates are lower, e.g. because banks attempt to increase their market share (Agarwal and Hauswald, 2010[23]).13

Lower levels of bank concentration can boost business in regions

Copy link to Lower levels of bank concentration can boost business in regions

Lower levels of bank concentration positively impact firms across regions. Lower levels of bank concentration allow firms to access the external finance they need to start their businesses and grow, by expanding their operations and creating new jobs. As a consequence, regions with lower bank concentration have more and larger establishments. First, lower bank concentration leads to more firm creation (Aghion, Fally and Scarpetta, 2007[24]) and to less firm exits, as firms are more likely to survive when they are not credit-constrained (Mach and Wolken, 2012[25]; Musso and Schiavo, 2008[26]) and when facing low borrowing costs (Guariglia, Spaliara and Tsoukas, 2015[27]).14 Second, firms in regions with low levels of bank concentration have more establishments as they are more able to grow, as also evidenced by the larger establishment sizes, further adding to the aggregate number of establishments.15

Comparing the Spanish region of Valencia with Lower Silesia in Poland highlights the importance of competitive banking markets. Although both regions have a similar number of active banks (51 in Valencia and 49 in Lower Silesia), the share of firms banking with the top three banks is significantly higher in Valencia (56%) than in Lower Silesia (less than 33%). In Valencia, where bank concentration is similar to the average region in the analysis, a decrease by one standard deviation16 to bank concentration levels around those of Lower Silesia in Poland would result in 10 801 additional establishments and an average increase of 0.2 employees per establishment.17

The effect of bank concentration is stronger in industries that rely heavily on intangible assets. These industries are expected to rely more on external finance but have less collateral than firms relying on tangible assets and are therefore more exposed to banking sector competition, as banks only turn to them for profit when competition for clients is high (Box 2.4). The benefits from lower bank concentration are indeed higher for firms with greater need for external finance (Figure 2.6). For firms among the 25% of industries with the lowest reliance on intangible assets, one standard deviation lower bank concentration is associated with at most an increase of 1.3% in the number of establishments and a 2.6% increase in the average size of the establishments. In high-exposure industries, i.e. those among the top 75%, an increase of the same magnitude in bank concentration is associated with at least 4.3% more establishments and an increase in the average establishment size of at least 8.7%.18

Figure 2.6. Lower bank concentration is associated with more and larger establishments
Copy link to Figure 2.6. Lower bank concentration is associated with more and larger establishments

Change in 2019 industry-level number of establishments (left) and establishment size (right) associated with one percentage point lower bank concentration (measured by HHI), by the degree of exposure or reliance on intangibles (RIA) (%)

Box 2.4. Measuring the effect of bank concentration on regional economic performance (I)

Copy link to Box 2.4. Measuring the effect of bank concentration on regional economic performance (I)

In a nutshell, the analysis recovers the effect of bank concentration by differentiating between industries that are differently exposed to the competition in the banking market across regions and then isolating the impact of an increase in bank concentration on firm performance in each of these industries and regions.

The analysis employs econometric techniques to measure the effect of bank concentration (a variable which approximates the ease of firms to access external financing) on firm performance. Ideally, this effect would be estimated in an experimental setting, where two otherwise identical regions would differ in the degree of concentration of their banking sectors. In this setting, the differences in economic performance across the two regions could be attributed solely to differences in access to finance. Even in the absence of an experimental setting, it is still possible to estimate the impact of access to finance on regional economic performance by adapting the empirical strategy pioneered by Rajan and Zingales (1998[28]) to leverage differences both across (TL2) regions and industries. On the one hand, regions vary in the extent of concentration of their banking markets (and thus in their access to finance). On the other hand, industries vary in their “exposure” to bank concentration, owing to their distinctive business models and technological requirements. By considering the level of exposure of each industry to each region's access to finance, it is possible to incorporate the characteristics of both the region and the industry (using both industry and region controls), which if neglected could bias the estimated coefficient. The baseline approach can be summarised by the following regression equation:

$Y_{r i} = {c + γ}_{r} + v_{i} + β {B a n k C o n c e n t r a t i o n}_{r} {E x p o s u r e}_{i} + e_{r i}$ (1),

where:

The dependent variable ( $Y_{r i}$ ) is the measure of economic performance in TL2 region r and industry i in 2019.¹ It can be employment, the number of establishments, the average size of establishments, or the GVA per worker growth between 2018 and 2019.
$γ_{r}$ and $v_{i}$ refer to region and industry controls, i.e. vectors of dummy variables capturing the constant specificities of any particular region (across all industries) and any particular industry (across all regions), respectively.
The key term in the regression is the interaction between i) the measure of exposure of industry i ( ${E x p o s u r e}_{i}$ ) to the ii) measure of access to external finance in the region r, proxied by the degree of bank concentration ( ${B a n k C o n c e n t r a t i o n}_{r}$ ). The greater the bank concentration, the lower the degree of banking competition and thus the harder it is to access external financing. In the baseline analysis, the bank concentration measured by the HHI is presented from 0 to 100 (instead of 0 to 1) for illustrative purposes.
The error term is denoted by $e_{r i}$ .

Measuring an industry’s exposure to access to finance by its reliance on intangible assets

To measure the exposure to access to finance, the main analysis uses the sectoral reliance on intangible assets, following Demmou, Franco and Ștefănescu (2020[29]). Sectors relying heavily on intangible assets are more likely to be disadvantaged by greater bank concentration and face tighter credit constraints since intangible assets are typically less likely than physical assets to be accepted as collateral as their valuations are more complex to perform and more uncertain. Furthermore, banks need to turn for profits to firms with harder-to-value collateral only when competition among financial institutions is high.

An industry’s reliance on intangible assets is measured by the intangible share of fixed assets (tangible and intangible) of the top 10% firms in the industry (by the share of intangible assets). The measure of exposure in each industry is calculated for the economy with the most developed financial system and the least stringent policy interventions, as that value is the closest to that in an ideal, frictionless world, with no policy-induced distortions. In the present empirical exercise, the benchmark country is the United Kingdom, as it has the most developed financial system of the countries for which data are available, as well as the least stringent banking regulation¹(Rajan and Zingales, 1998[28]) (see Annex 2.B for further details).

¹ The Rajan-Zingales approach has been criticised in the literature (Ciccone and Papaioannou, 2006[30]; 2022[31]) especially because it uses a country to measure the different industries’ exposures likely introducing bias. The bias resulting from the approach, however, is small (Bravo-Biosca, Criscuolo and Menon, 2016[32]). The sign of the bias depends on whether technologically similar countries tend to be similar in other country characteristics. Using the United Kingdom as the benchmark economy in a regional analysis likely leads to a smaller benchmarking bias than in analyses at the national level.

¹The effect of bank concentration is qualitatively equivalent when using 2020 economic performance measures instead of 2019. In 2020, the bank concentration level is pre-set and considerations of reversed causality are less relevant. However, due to the COVID-19 pandemic in 2020, the baseline analysis presents the evidence for 2019.

Additionally, regions with lower bank concentration see faster labour productivity growth, evidencing that access to finance enables continued investment across time. Across industries, decreasing bank concentration by one standard deviation is associated with an increase of 0.4 percentage points in labour productivity growth in the industry with a low reliance on intangibles and an increase of 1.5 percentage points in the industry with a higher reliance on intangibles, which is more exposed (Figure 2.7). Combining all sectors, a decrease in bank concentration by one standard deviation would raise productivity growth by 1.3 percentage points and lead to a 0.1% labour productivity growth, instead of a decline by -1.2% experienced from 2018 to 2019 in Valencia.

Figure 2.7. Lower bank concentration is associated with faster labour productivity growth
Copy link to Figure 2.7. Lower bank concentration is associated with faster labour productivity growth

Percentage points change in 2018-2019 industry-level GVA per worker growth rate associated with one percentage point lower HHI, by the degree of exposure or reliance on intangibles (RIA)

These positive firm-level effects translate into improved economic performance at the regional level. Regions with lower bank concentration have greater aggregate employment levels. In a region like Valencia, decreasing bank concentration by one standard deviation could add as much as 120 000 employees. For industries with low reliance on intangible assets, decreasing bank concentration by one standard deviation (7.2%) would translate to up to a 3.4% increase in total employment. The magnitude of the effects is even larger in the highly exposed industry, where a decrease of the same magnitude in bank concentration would translate into an increase of at least 11.4% in employment (Figure 2.8).

Figure 2.8. Lower bank concentration is associated with higher levels of employment
Copy link to Figure 2.8. Lower bank concentration is associated with higher levels of employment

Change in 2019 industry-level employment associated with one percentage point lower bank concentration (measured by HHI), by the degree of exposure or reliance on intangibles (RIA) (%)

Annex 2.A. Bank concentration measures

Copy link to Annex 2.A. Bank concentration measures

The comparable subnational indicators of local bank concentration created for 13 OECD countries, leverage 1.8 million firm-bank relationships collected within “Orbis”, a commercial database with financial information for millions of firms, previously used to study different implications of bank health on firm performance across different countries (Annex Box 2.A.1). The indicators rely on the share of firms that identify a given bank as their main bank and on the characteristics of the banks, such as whether they are active nationwide.

Annex Box 2.A.1. Using firm-level data to obtain regional indicators of bank concentration

Copy link to Annex Box 2.A.1. Using firm-level data to obtain regional indicators of bank concentration

In this report, the data about the regional measure of concentration in local banking markets comes from Orbis, a commercial database developed by Bureau van Dijk (Moody’s Analytics). The main database includes financial and address information of millions of firms in dozens of countries. An additional dataset, called “Bankers current”, includes the banking relationships of a subset of firms in the main Orbis data.

The coverage of “Bankers current” varies significantly by country, so only the countries with a substantial number of firms (both in absolute and relative to the OECD Structural and Demographic Business Statistics) were kept in the final sample. This meant that some countries with sparse coverage in “Bankers current”, such as Korea and Sweden, were dropped from the sample. For the 13 countries included in the final sample, a majority of firms contained in Orbis also report their banks. The selection was necessary to ensure that the indicators are representative of the business reality of each country.

Constructing the regional measures of bank concentration required several steps. The first step was to assign each firm in Orbis a region. The data includes some address information on firms, but this is not standardised and can include anything from the firms’ address information in free text to only zip codes or only the firm name. Thus, each firm with some address information was first geolocated and assigned a region using an array of methods, including crosswalks between zip codes and regions as well as geolocation algorithms from Open Street Maps and Bing. Although the address of the headquarters is often not the address where firms have their main activity, especially for large firms, it is expected that it is the address where most financing decisions would take place. Second, bank names had to be homogenized using a mix of statistical and manual techniques. A third step pooled the bank names provided by firms in each of the regions, calculating the market share of each bank brand in the region and the country. Note that firms were equally weighted when calculating market shares. Since Orbis has been shown to overstate the weight of larger firms in some countries (Kalemli-Özcan, Laeven and Moreno, 2022[6]), weighing firms by their employment or revenue would have compounded this bias.

Several assumptions underpin the construction of the indicators. First, firm-bank relationships observed in the dataset were assumed to be lending relationships. Second, if a firm has more bank relationships, only the first one is considered. This assumption is in line with current practice (Andrews and Petroulakis, 2019[33]). Third, the relationships are assumed to be constant for the period of observation. This is in line with past empirical evidence (D’Ignazio and Menon, 2019[34]; Chodorow-Reich, 2013[7]) and has been confirmed using firm-bank relationships in Orbis (Kalemli-Özcan, Laeven and Moreno, 2022[6]; Andrews and Petroulakis, 2019[33]; Giannetti and Ongena, 2012[35]). Fourth, 2019 was taken as the benchmark year, so banks no longer operating in 2019 were dropped, while those taken over by others before 2019 were identified with their new ownership.

Bank concentration measures derived using firm-bank relationships have several advantages compared to other indirect measures of access to finance found in the academic literature, such as those based on the number of local bank branches. First, they are more precise than indicators which rely on information from all types of banks, as they exclude consumer-only banks. Second, indicators derived from firm-bank relationships include online banks, as opposed to indicators based on physical distance which ignore the digital banking alternatives. Third, they are more appropriate for cross-country comparisons, especially when some regions are sparsely populated and have scattered services.

Measures

Copy link to Measures

The specific values of the bank concentration descriptive statistics are presented in Annex Table 2.A.1.

	HHI	Firms using top 3 local banks, share (%)	Firms using nationwide banks, share (%)	Number of banks	Banks in country
AUT	0.10 (0.04)	44.87 (10.8)	13 (12.93)	68.11 (42.79)	595
DEU	0.19 (0.07)	63.23 (9.67)	19.79 (9.62)	73.21 (51.13)	1 537
ESP	0.19 (0.04)	65.73 (8.85)	83.66 (12.64)	25.21 (13.91)	139
FRA	0.16 (0.05)	57.88 (8.9)	86.72 (8.35)	25.01 (11.64)	142
GBR	0.19 (0.05)	66.95 (8.41)	80.83 (19.47)	25.82 (15.38)	243
GRC	0.26 (0.03)	80.32 (5.14)	94.6 (4.76)	6.40 (2.61)	26
HUN	0.14 (0.02)	56.32 (5.79)	98.22 (2.11)	24.80 (3.83)	34
IRL	0.34 (0.04)	91.58 (3.47)	97.39 (2.39)	21.25 (16.46)	65
JPN	0.18 (0.08)	59.37 (12.29)	14.37 (9.21)	111.36 (76.56)	1 044
LVA	0.30 (0.04)	80.12 (4.7)	98.49 (0.7)	20.00 (5.87)	35
POL	0.15 (0.05)	53.67 (11.26)	80.4 (7.02)	22.36 (6.73)	202
PRT	0.17 (0.02)	61.36 (6.2)	97.84 (2.46)	13.12 (4.00)	35
SVN	0.18 (0.04)	60.86 (6.25)	99.98 (0.03)	15.58 (0.51)	16

Note: The Herfindahl–Hirschman index runs from 0 (not concentrated market, with lots of small firms) to 1 (very concentrated market – only one firm holds the entire market share). A nationwide bank is a bank present in at least 90% of the TL3 regions in a country.

Source: OECD calculations based on Orbis data.

Herfindahl-Hirschman Index

The Herfindahl-Hirschman Index, or HHI, ranges between 0 and 1. An HHI of 0 indicates a perfectly competitive banking sector, with lots of small banks, each with a very small market share. The closer the HHI is to 1, i.e. absolute monopoly, the greater the market concentration. For the regressions, the HHI was multiplied by 100 to facilitate interpretation as percentage points.

The regional HHI is computed as the sum of the square of the market shares of each bank active in a specific region. The market share of a bank is defined as the share of firms in a region that have declared a relationship with that bank.

The HHI shows some countries, such as Latvia, Ireland and Greece, tend to have concentrated banking markets across all regions (Annex Figure 2.A.1). Others, such as France and Hungary, tend to have low concentration in most regions, with some local exceptions. Germany, Austria and Japan, the three countries that rely heavily on local rather than nationwide markets, have the most regional heterogeneity in banking concentration.

Annex Figure 2.A.1. Regional bank concentration in TL3 regions: Herfindahl-Hirschman Index (I)
Copy link to Annex Figure 2.A.1. Regional bank concentration in TL3 regions: Herfindahl-Hirschman Index (I)

The share of firms using the largest three banks in the region

The next indicator is the share of firms using the largest three local banks. They are calculated as the ratio between the number of firms using the top 3 banks and the total number of firm-bank relationships reported in each region. The higher the share of firms using one of the largest local banks, the more concentrated the banking sector. This measure of banking concentration displays more regional polarisation compared to the HHI (Annex Figure 2.A.2, Annex Figure 2.A.3). Across groups of regions based on their access to metropolitan areas, similarly than when using the HHI, metropolitan regions see lower levels of bank concentration among the largest three banks (Annex Table 2.A.2).

Annex Figure 2.A.2. Regional bank concentration in TL3 regions: Market share of top three banks (I)
Copy link to Annex Figure 2.A.2. Regional bank concentration in TL3 regions: Market share of top three banks (I)

Annex Figure 2.A.3. Regional bank concentration in TL3 regions: Market share of top three banks (II)
Copy link to Annex Figure 2.A.3. Regional bank concentration in TL3 regions: Market share of top three banks (II)

Regional bank concentration as measured by the share of firms using any of the top three banks in the region (%)

	Regions in AUT, DEU and JPN based on their access to metropolitan areas	Regions in 10 other countries based on their access to metropolitan areas
Metropolitan region	59	61
Large metropolitan region	55	63
Metropolitan region	61	59
Region without a metropolitan area	64	67
Region near a metropolitan area	64	65
Region with/near a small-medium city	63	72
Remote region	64	64

Note: First column: Information refers to 482 small (TL3) regions from AUT (35), JPN (47) and DEU (400), excluding regions for which there are less than 14 observations i.e. Landau in der Pfalz, Kreisfreie Stadt (DEB33) in Germany. Second column: Information refers to 527 small (TL3) regions from POL (73), HUN (20), FRA (96), SVN (12), PRT (25), ESP (57), GBR (179), LVA (6), GRC (51) and IRL (8), excluding regions for which there are less than 14 observations i.e. Evrytania (EL643) in Greece. Banking information referring to 2019.

Source: OECD calculations based on Orbis.

The share of firms using nationwide banks

Another indicator is the share of firms using nationwide banks, i.e. banks present in at least 90% of the TL3 regions in the country. It is calculated as the ratio between the number of firms using nationwide banks and the total number of firm-bank relationships reported in each region.

The relationship between nationwide banks and local access to finance is unclear. While nationwide banks potentially have greater market power than regional banks and thus a greater share of firms using the banks can be associated with more concentrated banking sectors and thus worse access to finance; nationwide banks could provide access to finance to firms which regional banks could not finance, since they likely have more diversified portfolios and can take more risks. However, the banking system based on local banks in Japan, Germany and Austria is different from the one based on nationwide banks as seen in Greece of Hungary (Annex Figure 2.A.4).

Annex Figure 2.A.4. Share of firms banking with a nationwide bank
Copy link to Annex Figure 2.A.4. Share of firms banking with a nationwide bank

The share of nationwide banks is greater in metropolitan regions than in other regions when banking systems are organised around local banks, while on average where banking systems are organised around nationwide banks, in other regions firms use in greater shares nationwide banks (Annex Table 2.A.3).

	Regions in AUT, DEU and JPN based on their access to metropolitan areas	Regions in 10 other countries based on their access to metropolitan areas
Metropolitan region	21	83
Large metropolitan region	27	86
Metropolitan region	19	82
Region without a metropolitan area	15	88
Region near a metropolitan area	16	85
Region with/near a small-medium city	16	87
Remote region	13	91

Source: OECD calculations based on Orbis.

Annex 2.B. Measuring an industry’s exposure to bank concentration

Copy link to Annex 2.B. Measuring an industry’s exposure to bank concentration

Using the United Kingdom as the benchmark economy

Copy link to Using the United Kingdom as the benchmark economy

The exposure to bank concentration of different industries is calculated using Orbis firms’ balance sheet information from firms in the United Kingdom. The United States is usually the country for which these exposure measures are computed, as it is considered to be the economy with the least stringent banking regulation. However, given the low coverage of US firms in Orbis (Bajgar et al., 2020[36]), the United Kingdom is used as the benchmark economy in this report.

According to De Serres et al. (2006[37]), in 2002-2003 the United Kingdom had the second least regulated banking sector across the 30 OECD member countries at that time, with New Zealand being first. The authors compute the banking regulation index as an unweighted average of four different indicators: a government ownership index, proxying for the extent to which competition might be distorted by the existence of government-owned entities; a domestic entry index, measuring the licensing requirement of setting up a bank in each country; a bank activity index, measuring the level of regulatory restrictiveness for bank participation in securities activity (the ability of banks to engage in the business of securities underwriting, brokering, dealing and mutual fund operations) and insurance activity; and an index of restrictions on foreign entry in banking based on earlier OECD work on foreign direct investment (FDI) restrictions (Golub, 2003[38]).

A similar exercise with the latest available iteration of the World Bank’s Bank Regulation and Supervision Survey, with surveys run in 2016/17, shows the United Kingdom has the second least regulated banking sector, only surpassed by the United States among the countries in the sample, i.e. 13 countries for which we have concentration indicators available and the United States (Annex Table 2.B.1). However, this excludes the FDI-based indicator.

	USA	GBR	LVA	GRC	ESP	IRL	AUT	FRA	HUN	PRT	SVN	POL	DEU	JPN
Government ownership index	0	8	0	0	2	15.3	7.5	7.5	9.8	37.0	40.3	25.2	37.1
Domestic entry index	10	0	100	100	100	90	100	100	100	80	80	100	90	100
Bank activity index	33.3	50.0	0	16.7	16.7	16.7	16.7	16.7	16.7	16.7	16.7	16.7	16.7	33.3
Foreign entry restriction index
Banking sector regulation index	10.8	14.5	25	29.2	29.7	30.5	31	31	31.6	33.4	34.2	35.5	35.9	44.4

Note: The banking sector regulation index and its components, range from 0 (least restrictive) to 100 (most restrictive).

Source: OECD calculations based on the latest available iteration of the World Bank’s Bank Regulation and Supervision Survey (surveys run in 2016/17).

Using an economy as a benchmark might lead to biased estimates of the effect of bank concentration on an industry’s outcome. As summarised in Bravo-Biosca, Criscuolo and Menon (2016[32]), the overall sign of this bias, referred to as ‘benchmarking bias’, depends on two specific biases.

First is the attenuation bias, which arises due to classical measurement error. The size of this bias depends on the extent to which the benchmark country differs from the ideal frictionless economy. The further away from the ideal frictionless economy the benchmark country is, the stronger the attenuation bias and the closer to zero the estimated coefficients will be. There might be reasons to believe (as shown in the previous section of the present Annex) that the United Kingdom has a slightly more regulated banking sector than the United States. For this reason, there is likely more attenuation bias when using the United Kingdom than when using the United States (i.e. the estimated effects are closer to 0 than the true effects).

Second is the amplification bias, which arises due to a systematic error component. The amplification bias might lead to inconsistent estimates of the policy effects in a particular sector of a given country, depending on the similarity in both the industry structure and characteristics between a given country and the benchmark country (i.e. whether a particular sector has a similar weight in the economy and whether a particular country has similar features in the policies of interest with the benchmark country). This correlation would spuriously amplify the true effect of the policy in the industry of those countries that are more similar to the benchmark and it would lead to an underestimation of the effect of banking concentration in the industry of the more dissimilar countries. The amplification bias is likely small in the context of the current analysis, as it focuses on regions and regional economies tend to be more specialised than country economies (e.g. capital city regions might be more oriented towards services to a larger extent than the UK’s overall economy, while other regions might be more specialised manufacturing and much more so than the UK’s overall economy).

Previously, alternative empirical strategies have found similar results to the ones obtained via the traditional Rajan-Zingales methodology. Specifically, Bravo-Biosca, Criscuolo and Menon (2016[32]) compute the IV strategy proposed in the literature (Ciccone and Papaioannou, 2022[31]) and find that the differences between the results from the IV strategy and those from the Rajan-Zingales methodology are small, i.e. the benchmarking bias is small. For this reason, in the context of the current analysis in which the two biases leading to the benchmarking bias are likely to be even smaller, for the reasons previously highlighted in the paragraphs above, the results from this IV strategy are likely to be even closer to the ones estimated with the Rajan-Zingales approach.

Measuring an industry’s exposure to bank concentration by its reliance on intangibles

Copy link to Measuring an industry’s exposure to bank concentration by its reliance on intangibles

One measure of an industry’s exposure to bank concentration refers to its reliance on intangible assets. Previous OECD work based on previous literature (Peters and Taylor, 2017[39]) used a more encompassing measure of reliance on intangible capital than the one used in the present exercise. This measure includes both knowledge-based capital (on which the measure used in the present exercise uniquely relies) and organizational-based capital (Demmou, Franco and Stefanescu, 2020[29]). Specifically, knowledge-based capital includes a firm’s capitalised spending to develop knowledge, patents, or software and is calculated as the sum of intangible assets over the sum of total assets over the 1990-2006 period for each firm. Organisational capital of a firm is calculated by capitalizing a fraction of selling, general and administrative expense (SG&A), which includes advertising to build brand capital, human capital, customer relationships and distribution systems over the same period (1990-2006). To obtain an industry’s value for each of these two capital measures (i.e. the knowledge-based capital and the organizational-based capital), the authors take the median firm within each industry. Then, to compute each industry’s reliance on intangibles they sum both values.

However, this more encompassing measure of reliance on intangible assets, computed using Compustat data about United States’ listed companies, is only available for 28 industries. In the current exercise, to cover the 68 industries in the analysis, the measures of reliance on intangibles rely on knowledge-based capital only. The measures computations rely on Orbis balance sheet information data for the period 2009-2019. Compared to Compustat, which contains information on US listed firms, Orbis includes smaller firms. To avoid having many industries with a zero reliance on intangibles instead of taking the median firm to measure an industry’s exposure the analysis takes the firm at the 90^th percentile. Despite the different calculations, both measures of an industry’s reliance on intangibles are highly correlated (Annex Figure 2.B.1).

Annex Figure 2.B.1. Comparing DFS’s RIA measures with Orbis-computed RIA measure
Copy link to Annex Figure 2.B.1. Comparing DFS’s RIA measures with Orbis-computed RIA measure

Scatterplots show a high correlation between both reliance on intangibles measures

Annex 2.C. Descriptive statistics

Copy link to Annex 2.C. Descriptive statistics

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
Log of number of establishments	6417	5.9	2.1	0	11.9	5.9	4.5	7.4	2.2	9.2
Log of employment (number employed people)	5993	8.0	1.9	0	13.8	8.1	6.8	9.3	4.6	10.9
Log of firm size	5945	1.8	1.1	-2.9	7.1	1.7	1.0	2.6	0.3	3.9

Note: The variables units of observation are the TL2 regions of AUT, POL, HUN, FRA, SVN, PRT, DEU, ESP, LVA, GRC and IRL and 68 2-digit NACE (Rev. 2) industries.

Source: OECD calculations based on Eurostat’s Structural Business Statistics (SBS) by NUTS 2 regions and NACE Rev. 2 (from 2008 onwards).

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
Growth rate of GVA per worker	954	1.6	8.2	-34.8	85.1	0.5	-1.6	3.5	-7.6	11.6

Note: The variables units of observation are the TL2 regions of AUT, POL, HUN, FRA, SVN, PRT, DEU, ESP, LVA, GRC and IRL and 9 sections of NACE (Rev. 2) industry classification.

Source: OECD calculations based on GVA per worker from the OECD Regional Economy Database.

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
RIA	68	0.3	0.3	0.0	1.0	0.3	0.1	0.5	0.0	0.9

RIA	9	0.3	0.3	0.0	0.9	0.3	0.1	0.4	0.0	0.9

Note: The first row contains the descriptive statistics of the RIA for the 68 industries present in the employment, number of establishments and establishment size regressions. The second row contains the descriptive statistics of the RIA for the 9 industries present in the GVA per worker growth regressions. For details on how the RIA is computed refer to Annex 2.B.

Source: OECD calculations based on Orbis.

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
HHI	106	14.4	7.2	1.6	37.0	13.9	10.1	17.6	3.0	27.9
SB3	106	54.1	17.7	13.5	93.1	55.7	44.1	62.9	22.1	81.9

Note: The descriptive statistics of the bank concentration indicators refer to the TL2 regions of AUT, POL, HUN, FRA, SVN, PRT, DEU, ESP, LVA, GRC and IRL. The HHI, originally ranging from 0-1, is adjusted in the regression analysis to range from 0-100 for illustrative purposes.

Source: OECD calculations based on Orbis.

Annex 2.D. Robustness

Copy link to Annex 2.D. Robustness

This annex discusses the robustness of the results to changes in the variables used in the analysis. None of the findings are driven by the regions of any specific country. Furthermore, the findings are robust to using the market share of the top three banks in the region (SB3) instead of the baseline HHI, as a measure of bank concentration and, as in the baseline, the results do not seem to be driven by a particular country. The regression results are also quantitatively equivalent to the ones obtained by excluding from HHI calculations the industry to which the economic performance indicator refers.

Furthermore, this annex also discusses the empirical results using alternative measures of exposure, which suggest that the effect of regional bank concentration on an industry’s economic performance depends on how exposure is measured. The alternative measures of exposure rely on the measure originally used by Rajan and Zingales (1998[28]), reflecting an industry’s external financing requirement (EFR). The external financing requirement captures the extent to which an industry can cover its capital expenditures with internal resources. If negative, the industry generates more internal resources than it has capital expenditures; if positive, the industry generates fewer internal resources than it has capital expenditures, thus relying on external financing.

However, there are reasons to believe the EFR measures are less appropriate measures of exposure to bank concentration than the baseline RIA measure. When computed for the median firm in each industry, as commonly done in the literature, the EFR is negative for most industries, suggesting industries generate more internal resources than they have capital expenditures. The EFR is also negatively correlated with the median loans-to-assets ratio (-0.11), which is not in line with the EFR measuring external financing (since loans are one of the main instruments of financing). The EFR measure could thus be capturing to what extent industries can secure external funding rather than to what extent they need it. In that case, it is industries with more internal resources (i.e. lower and more negative EFR) and better balance sheets and financial positions that have lower debt ratios. Alternatively, when computed for the firm at the 90^th percentile of the distribution in the industry rather than for the median, the EFR measure is positive for most industries. By construction, the measure refers to the firms that are the least constrained within each industry when it comes to securing external financing. These are likely bigger, older, more established firms and the arguments that apply to them might be less relevant to the industry as a whole. As expected, this measure is positively correlated with the median loans-to-assets ratio (0.52) across the 68 2-digit NACE (Rev. 2) industries used in the analysis.

Leave-one-country-out regressions

Copy link to Leave-one-country-out regressions

The results from the regression analysis are robust to dropping the regions of one country at a time. Specifically, the results on employment, GVA per worker growth rate and establishment size do not change (Annex Table 2.D.1, Annex Table 2.D.2 and Annex Table 2.D.4, respectively). The estimated coefficient of the effect of banking concentration on an industry’s number of establishments becomes smaller and statistically insignificant when dropping Germany (DEU) and slightly increases when dropping GRC (Annex Table 2.D.3).

	Baseline	AUT	DEU	ESP	FRA	GRC	HUN	IRL	LVA	POL	PRT	SVN
Coef.	-.03***	-.04***	-.02***	-.03***	-.03***	-.03***	-.03***	-.04***	-.03***	-.04***	-.03***	-.03***
SE	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)
Obs.	5993	5482	5087	5007	5302	5279	5483	5902	5927	4978	5604	5879
R²	.81	.81	.8	.81	.81	.8	.81	.81	.81	.82	.81	.81

Note: The baseline presents the baseline regression results (i.e. including the 11 countries such as AUT, POL, HUN, FRA, SVN, PRT, DEU, ESP, LVA, GRC and IRL). Each of the other columns presents the regression results when omitting the country in the column’s name. All regressions include region and industry fixed effects and a constant term. The 1% and 5% significance levels are *** p<.01 and ** p<.05.

Source: OECD calculations based on GVA per worker from the OECD Regional Economy Database and Orbis.

	Baseline	AUT	DEU	ESP	FRA	GRC	HUN	IRL	LVA	POL	PRT	SVN
Coef.	-.52***	-.5***	-.69***	-.53***	-.54***	-.42**	-.53***	-.44***	-.52***	-.47***	-.53***	-.52***
SE	(.15)	(.16)	(.19)	(.16)	(.15)	(.2)	(.15)	(.11)	(.15)	(.15)	(.15)	(.15)
Obs.	954	873	810	801	837	837	882	927	945	801	891	936
R²	.35	.35	.36	.35	.36	.37	.36	.35	.35	.26	.36	.35

Source: OECD calculations based on GVA per worker from the OECD Regional Economy Database and Orbis.

	Baseline	AUT	DEU	ESP	FRA	GRC	HUN	IRL	LVA	POL	PRT	SVN
Coef.	-.01**	-.01**	-.003	-.01**	-.01**	-.02***	-.01**	-.01*	-.01**	-.02***	-.01*	-.01**
SE	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)
Obs.	6417	5842	5480	5430	5572	5703	5907	6326	6351	5300	5975	6284
R²	.89	.89	.9	.89	.88	.9	.89	.89	.89	.89	.89	.89

Source: OECD calculations based on Eurostat’s Structural Business Statistics (SBS) by NUTS 2 regions and NACE Rev. 2 (from 2008 onwards) and Orbis.

	Baseline	AUT	DEU	ESP	FRA	GRC	HUN	IRL	LVA	POL	PRT	SVN
Coef.	-.03***	-.03***	-.03***	-.02***	-.02***	-.02***	-.03***	-.03***	-.03***	-.03***	-.03***	-.03***
SE	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)	(.01)
Obs.	5945	5434	5087	4959	5254	5231	5435	5854	5879	4930	5556	5831
R²	.71	.71	.67	.71	.7	.72	.71	.71	.71	.72	.71	.7

Source: OECD calculations based on Eurostat’s Structural Business Statistics (SBS) by NUTS 2 regions and NACE Rev. 2 (from 2008 onwards) and Orbis.

Measuring regional bank concentration as the share of firms using one of the top three banks

Copy link to Measuring regional bank concentration as the share of firms using one of the top three banks

The results of the regression analysis when using the share of firms using one of the top 3 banks (SB3) instead of HHI to measure regional bank concentration are presented next (Annex Table 2.D.5, Annex Table 2.D.6, Annex Table 2.D.7 and Annex Table 2.D.8). As in the baseline regressions, the measure of an industry’s exposure used is the RIA (Annex 2.B). Refer to Annex 2.C for the descriptive statistics of the variables used.

	Baseline	AUT	DEU	ESP	FRA	GRC	HUN	IRL	LVA	POL	PRT	SVN
Coef.	-.01***	-.02***	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***	-.02***	-.01***	-.01***
SE	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)
Obs.	5993	5482	5087	5007	5302	5279	5483	5902	5927	4978	5604	5879
R²	.81	.81	.8	.81	.81	.8	.81	.81	.81	.82	.81	.81

Source: OECD calculations based on Eurostat’s Structural Business Statistics (SBS) by NUTS 2 regions and NACE Rev. 2 (from 2008 onwards) and Orbis.

	Baseline	AUT	DEU	ESP	FRA	GRC	HUN	IRL	LVA	POL	PRT	SVN
Coef.	-.18***	-.17***	-.27***	-.18***	-.18***	-.13*	-.2***	-.15***	-.18***	-.17***	-.18***	-.18***
SE	(.05)	(.06)	(.08)	(.06)	(.06)	(.07)	(.05)	(.04)	(.05)	(.05)	(.06)	(.05)
Obs.	954	873	810	801	837	837	882	927	945	801	891	936
R²	.35	.35	.35	.34	.35	.36	.36	.35	.35	.25	.35	.35

Source: OECD calculations based on GVA per worker from the OECD Regional Economy Database and Orbis.

	Baseline	AUT	DEU	ESP	FRA	GRC	HUN	IRL	LVA	POL	PRT	SVN
Coef.	-.01**	-.01**	-.001	-.01**	-.005**	-.01***	-.01**	-.004**	-.005**	-.01***	-.004*	-.005**
SE	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)
Obs.	6417	5842	5480	5430	5572	5703	5907	6326	6351	5300	5975	6284
R²	.89	.89	.9	.89	.88	.9	.89	.89	.89	.89	.89	.89

Source: OECD calculations based on Eurostat’s Structural Business Statistics (SBS) by NUTS 2 regions and NACE Rev. 2 (from 2008 onwards) and Orbis.

	Baseline	AUT	DEU	ESP	FRA	GRC	HUN	IRL	LVA	POL	PRT	SVN
Coef.	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***	-.01***
SE	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)	(0)
Obs.	5945	5434	5087	4959	5254	5231	5435	5854	5879	4930	5556	5831
R²	.71	.71	.67	.71	.7	.72	.71	.71	.71	.72	.71	.71

Source: OECD calculations based on Eurostat’s Structural Business Statistics (SBS) by NUTS 2 regions and NACE Rev. 2 (from 2008 onwards) and Orbis.

Measuring bank concentration excluding the industry in the analysis

Copy link to Measuring bank concentration excluding the industry in the analysis

In the regression analysis ${B a n k C o n c e n t r a t i o n}_{r}$ is substituted by ${B a n k C o n c e n t r a t i o n}_{r, - i}$ which excludes any firm-bank relationships reported by firms in industry i in the computations:

$Y_{r i} = {c + γ}_{r} + v_{i} + β {B a n k C o n c e n t r a t i o n}_{r, - i} {E x p o s u r e}_{i} + e_{r i}$

Annex Table 2.D.9 presents the descriptive statistics for these alternative HHI measures in the first two rows and for the baseline HHI measure in the third. The differences between the alternative HHI and the baseline HHI are negligible.

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
HHIr,-i (68 industries)	8692	14.4	7.2	1.6	37.6	13.9	10.1	17.6	3.0	27.9
HHIr,-i (9 industries)	954	14.4	7.2	1.5	38.0	13.9	10.1	17.6	3.0	27.8
HHI (baseline)	106	14.4	7.2	1.6	37.0	13.9	10.1	17.6	3.0	27.9

Note: The descriptive statistics of the alternative HHI (HHI_r,-i) refer to the TL2 regions of AUT, POL, HUN, FRA, SVN, PRT, DEU, ESP, LVA, GRC and IRL and the 9 or 68 industries in the analysis. The descriptive statistics of the HHI refer to the TL2 regions of AUT, POL, HUN, FRA, SVN, PRT, DEU, ESP, LVA, GRC and IRL. HHI measures, originally ranging from 0-1, are adjusted in the regression analysis to range from 0-100 for illustrative purposes.

Source: OECD calculations based Orbis.

Annex Table 2.D.10 summarises the estimated coefficients of the regression analysis using these alternative HHI measure. Differences in the estimated regression coefficients are negligible.

	log employment		log No. establishments		log establishment size		growth rate GVA per worker
	HHI (baseline)	HHI_r,-i	HHI (baseline)	HHI_r,-i	HHI (baseline)	HHI_r,-i	HHI (baseline)	HHI_r,-i
Est. Coefficient	-.034*** (-0.007)	-.034*** (-0.007)	-.013** (-0.005)	-.013** (-0.005)	-.026*** (-0.005)	-.026*** (-0.005)	-.52*** (-0.147)	-.522*** (-0.146)
Obs.	5993	5993	6417	6417	5945	5945	954	954
R²	0.81	0.81	0.891	0.891	0.706	0.706	0.351	0.351

Note: All regressions include a constant, region and industry fixed effects, The estimated coefficients of the effects of bank concentration on the logarithmic of employment are statistically significant at the 1% level. The estimated coefficients of the effect of bank concentration on the growth rate of GVA per worker are statistically significant at the 1% level. The estimated coefficients of the effect of bank concentration on the logarithmic of the number of establishments are statistically significant at the 5% level. The estimated coefficients of the effect of bank concentration on the logarithmic of the establishment size are statistically significant at the 1% level. All regressions rely on data from 11 countries (AUT, POL, HUN, FRA, SVN, PRT, DEU, ESP, LVA, GRC and IRL). GBR is used to compute the measures of exposure, so it is removed from the regression analysis to avoid endogeneity issues as it is standard in the literature. The regressions of the logarithmic of employment and of establishment's number and size rely on data for 68 2-digit NACE Rev. 2 industries. The regression of the growth of the GVA per worker relies on data for 9 pooled NACE Rev. 2 sections (i.e. B-E, F, G-I, J, K, L, M-N, O-Q, R-U). The units of the HHI measure are transformed to ease interpretation (0-100 instead of 0-1).

Source: OECD calculations based on Eurostat’s Structural Business Statistics (SBS) by NUTS 2 regions and NACE Rev. 2 (from 2008 onwards), GVA per worker from the OECD Regional Economy Database and Orbis.

Measuring an industry’s exposure to bank concentration by its external financing dependence

Copy link to Measuring an industry’s exposure to bank concentration by its external financing dependence

An industry’s external financing dependence (EFR) is also a commonly used measure of an industry’s exposure to banking concentration (Rajan and Zingales, 1998[28]). An industry’s EFR is computed as the fraction of capital expenditures not financed with cash flow from operations. Intuitively, capital-intensive industries such as heavy manufacturing are more likely to require external financing than labour-intensive industries such as hospitality.

Computing the EFR for the median firm in each industry

When as in the literature (Rajan and Zingales, 1998[28]), the external financing requirement is computed for the median firm in each industry, unlike what is found in the literature, most industries are cash flow positive (i.e. the median firm generates more cash flow than capital expenditures in most industries) (Annex Table 2.D.11).

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
EFR, median firm	68	-0.6	0.9	-3.4	1.3	-0.4	-1	-0.1	-2.7	0.8

EFR, median firm	9	-0.7	0.8	-1.8	0.9	-0.5	-1.4	-0.3	-1.8	0.9

Note: The first row contains the descriptive statistics of the EFR (computed for the median firm) for the 68 industries present in the employment, number of establishments and establishment size regressions. The second row contains the descriptive statistics of the EFR (computed for the median firm) for the 9 industries present in the GVA per worker growth regressions.

Source: OECD calculations based on Orbis.

Specifically, the EFR is computed in each industry as a firm’s difference between the sum of its capital expenditure and cash flow, divided by the sum of its capital expenditure for the period 2009-2019. The calculations use information on the United Kingdom’s firms contained in Orbis. While this is a slight deviation from the literature, as most studies use Compustat data for United States’ listed firms to measure EFR (Rajan and Zingales, 1998[28]), using Orbis data from the United Kingdom as a benchmark economy (Annex 2.B).

The measures computed over the subsample of young and mature firms separately are highly correlated with the one including all firms (Annex Figure 2.D.1), which is the one used in the regression analysis. The correlation between the baseline measure and the measure computed using the subsample of young firms is 0.77, while the correlation between the baseline measure and the measure computed using the subsample of mature firms is 0.83.

Annex Figure 2.D.1. There is a strong correlation between the external finance requirement measure and that computed over the subsample of young or mature firms
Copy link to Annex Figure 2.D.1. There is a strong correlation between the external finance requirement measure and that computed over the subsample of young or mature firms

Correlation between the external financing requirement of the median firm (all firms) and that computed with the subsample of young or mature firms in each industry

Results from the regression analysis when using this alternative exposure measure lead to similar results for most variables, suggesting that lower bank concentration is associated with greater employment and an even greater number of establishments and thus smaller average establishment sizes; as well as greater labour productivity growth in the industry with average external financing dependence (Annex Figure 2.D.2 and Annex Figure 2.D.3).

However, the regression results also show that it is industries with low external financing dependence, whose employment, number of establishments, establishments sizes and labour productivity growth are most affected by bank concentration. These are industries, in which the median firm is relatively more cash flow positive than others, thus has a strong financial position, lower apparent risks and is better perceived by creditors. For this reason, these are industries that likely have easier access to credit. This is consistent with the data as the measure of external financing dependence is negatively correlated with the median loans to assets ratio (-0.11). For this reason, firms in these industries might be most negatively affected by bank concentration. Relatedly, industries requiring greater external financing might have alternative funding sources than banks and thus for this reason might be only marginally affected by bank concentration.

Annex Figure 2.D.2. Regression results: median EFR to measure an industry’s exposure to bank concentration (part 1)
Copy link to Annex Figure 2.D.2. Regression results: median EFR to measure an industry’s exposure to bank concentration (part 1)

Percentage change in industry-level employment (left) and percentage points change in industry-level GVA per worker growth rate (upper right) associated with a one percentage point lower HHI, by the degree of external financing dependence (EFR) measured for the median firm

Annex Figure 2.D.3. Regression results: median EFR to measure an industry’s exposure to bank concentration (part 2)
Copy link to Annex Figure 2.D.3. Regression results: median EFR to measure an industry’s exposure to bank concentration (part 2)

Percentage change in number of establishments (lower left) and establishment size (right) associated with a one percentage point lower HHI, by the degree of external financing dependence (EFR) measured for the median firm

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
EFR, P90 firm	68	19.4	22.4	1.9	158.9	13.1	6.3	26.0	3.5	45.7

EFR, P90 firm	9	20.3	10.3	7.6	37.3	17.4	11.5	27.1	7.6	37.3

Note: The first row contains the descriptive statistics of the EFR (computed for the 90^th percentile firm) for the 68 industries present in the employment, number of establishments and establishment size regressions. The second row contains the descriptive statistics of the EFR (computed for the 90^th percentile firm) for the 9 industries present in the GVA per worker growth regressions.

Source: OECD calculations based on Orbis.

Annex Figure 2.D.4. The firm at the 90^th percentile of external financing requirement has an actual external financing need in all industries
Copy link to Annex Figure 2.D.4. The firm at the 90<sup>th</sup> percentile of external financing requirement has an actual external financing need in all industries

Correlation between the external financing requirement of the mean and the 90^th percentile firm in each industry

The regression results when using the external financing dependence as a measure of exposure show that greater bank concentration is associated with greater employment, greater firm size and lower GVA per worker growth. The effect of bank concentration on the number of establishments is not statistically significant. When using this measure of exposure instead, the estimated impact of bank concentration is larger for industries with greater external financing requirements, in this case, those with greater financing needs. The detailed results are presented in Annex Figure 2.D.5 and Annex Figure 2.D.6.

Annex Figure 2.D.5. Regression results: 90^th percentile EFR to measure an industry’s exposure to bank concentration (part 1)
Copy link to Annex Figure 2.D.5. Regression results: 90<sup>th</sup> percentile EFR to measure an industry’s exposure to bank concentration (part 1)

Percentage change in industry-level employment (left), and percentage points change in industry-level GVA per worker growth rate (right) associated with a one percentage point lower HHI, by the degree of external financing dependence (EFR) measured for the 90^th percentile firm

Annex Figure 2.D.6. Regression results: 90^th percentile EFR to measure an industry’s exposure to bank concentration (part 2)
Copy link to Annex Figure 2.D.6. Regression results: 90<sup>th</sup> percentile EFR to measure an industry’s exposure to bank concentration (part 2)

Percentage change in number of establishments (left) and establishment size (right)) associated with one percentage point lower HHI, by the degree of external financing dependence (EFR) measured for the 90^th percentile firm

References

[23] Agarwal, S. and R. Hauswald (2010), “Distance and Private Information in Lending”, Review of Financial Studies, Vol. 23/7, pp. 2757-2788, https://doi.org/10.1093/rfs/hhq001.

[24] Aghion, P., T. Fally and S. Scarpetta (2007), “Credit constraints as a barrier to the entry and post-entry growth of firms”, Economic Policy, Vol. 22/52, pp. 732-779, https://doi.org/10.1111/j.1468-0327.2007.00190.x.

[15] Alessandrini, P., M. Cocci and A. Zazzaro (2005), “The geography of banking power: The role of functional distance”, No. 58, Banca Nazionale del Lavoro Quarterly Review,.

[33] Andrews, D. and F. Petroulakis (2019), Breaking the shackles: Zombie firms, weak banks and depressed restructuring in Europe, https://doi.org/10.2866/192858.

[36] Bajgar, M. et al. (2020), “Coverage and representativeness of Orbis data”, OECD Science, Technology and Industry Working Papers, No. 2020/06, OECD Publishing, Paris, https://doi.org/10.1787/c7bdaa03-en.

[14] Baker, G. (1999), “Informal authority in organizations”, Journal of Law, Economics, and Organization, Vol. 15/1, pp. 56-73, https://doi.org/10.1093/jleo/15.1.56.

[21] Beatriz, M., J. Coffinet and T. Nicolas (2022), “Relationship lending and SMEs’ funding costs over the cycle: Why diversification of borrowing matters”, Journal of Banking & Finance, Vol. 138, p. 105471, https://doi.org/10.1016/j.jbankfin.2018.12.007.

[41] Beck, T., A. Demirguc-Kunt and V. Maksimovic (2004), “Bank Competition and Access to Finance: International Evidence”, Journal of Money, Credit, and Banking, Vol. 36/3b, pp. 627-648, https://doi.org/10.1353/mcb.2004.0039.

[18] Berger, A. and W. Frame (2007), “Small Business Credit Scoring and Credit Availability*”, Journal of Small Business Management, Vol. 45/1, pp. 5-22, https://doi.org/10.1111/j.1540-627x.2007.00195.x.

[45] Berger, A., W. Scott Frame and V. Ioannidou (2011), “Tests of ex ante versus ex post theories of collateral using private and public information”, Journal of Financial Economics, Vol. 100/1, pp. 85-97, https://doi.org/10.1016/j.jfineco.2010.10.014.

[17] Berger, A. and G. Udell (2006), “A more complete conceptual framework for SME finance”, Journal of Banking & Finance, Vol. 30/11, pp. 2945-2966, https://doi.org/10.1016/j.jbankfin.2006.05.008.

[22] Bolton, P. et al. (2013), Relationship and Transaction Lending in a Crisis, National Bureau of Economic Research, Cambridge, MA, https://doi.org/10.3386/w19467.

[32] Bravo-Biosca, A., C. Criscuolo and C. Menon (2016), “What drives the dynamics of business growth?”, Economic Policy, Vol. 31/88, pp. 703-742, https://doi.org/10.1093/epolic/eiw013.

[2] British Business Bank (2024), Nation and Regions Tracker 2024.

[7] Chodorow-Reich, G. (2013), “The Employment Effects of Credit Market Disruptions: Firm-level Evidence from the 2008–9 Financial Crisis *”, The Quarterly Journal of Economics, Vol. 129/1, pp. 1-59, https://doi.org/10.1093/qje/qjt031.

[31] Ciccone, A. and E. Papaioannou (2022), “Estimating Cross-Industry Cross-Country Interaction Models Using Benchmark Industry Characteristics”, The Economic Journal, Vol. 133/649, pp. 130-158, https://doi.org/10.1093/ej/ueac047.

[30] Ciccone, A. and E. Papaioannou (2006), Adjustment to Target Capital, Finance and Growth, https://papers.ssrn.com/abstract=954976 (accessed on 6 July 2023).

[3] Crawford, G., N. Pavanini and F. Schivardi (2018), “Asymmetric Information and Imperfect Competition in Lending Markets”, American Economic Review, Vol. 108/7, pp. 1659-1701, https://doi.org/10.1257/aer.20150487.

[34] D’Ignazio, A. and C. Menon (2019), “Causal Effect of Credit Guarantees for Small‐ and Medium‐Sized Enterprises: Evidence from Italy”, The Scandinavian Journal of Economics, Vol. 122/1, pp. 191-218, https://doi.org/10.1111/sjoe.12332.

[37] De Serres, A. et al. (2006), “Regulation of financial systems and economic growth in OECD countries: an empirical analysis”.

[29] Demmou, L., G. Franco and I. Stefanescu (2020), “Productivity and finance: the intangible assets channel - a firm level analysis”, OECD Economics Department Working Papers, No. 1596, OECD Publishing, Paris, https://doi.org/10.1787/d13a21b0-en.

[44] Ellul, A. (ed.) (2022), “Credit Environment and Small Business Dynamics: Evidence from Establishment-Level Data”, The Review of Corporate Finance Studies, Vol. 12/2, pp. 326-365, https://doi.org/10.1093/rcfs/cfac037.

[1] European Investment Bank (2024), EIB Investment Survey 2024: European Union overview.

[43] Fischer, K. (2000), “Acquisition of Information in Loan Markets and Bank Market Power - An Empirical Investigation”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.240921.

[19] Flögel, F. (2018), Distance, Rating Systems and Enterprise Finance, Routledge, Abingdon, Oxon ; New York, NY : Routledge, 2018. |, https://doi.org/10.4324/9781351256124.

[35] Giannetti, M. and S. Ongena (2012), ““Lending by example”: Direct and indirect effects of foreign banks in emerging markets”, Journal of International Economics, Vol. 86/1, pp. 167-180, https://doi.org/10.1016/j.jinteco.2011.08.005.

[38] Golub, S. (2003), “Measures Of Restrictions on Inward Foreign Direct Investment for OECD Countries”, OECD Economics Department Working Papers, No. 357, OECD Publishing, Paris, https://doi.org/10.1787/335043060125.

[27] Guariglia, A., M. Spaliara and S. Tsoukas (2015), “To What Extent Does the Interest Burden Affect Firm Survival? Evidence from a Panel of UK Firms during the Recent Financial Crisis”, Oxford Bulletin of Economics and Statistics, Vol. 78/4, pp. 576-594, https://doi.org/10.1111/obes.12120.

[12] Hasan, I. et al. (2018), “The economic impact of changes in local bank presence”, Regional Studies, Vol. 53/5, pp. 644-656, https://doi.org/10.1080/00343404.2018.1475729.

[6] Kalemli-Özcan, Ş., L. Laeven and D. Moreno (2022), “Debt Overhang, Rollover Risk, and Corporate Investment: Evidence from the European Crisis”, Journal of the European Economic Association, Vol. 20/6, pp. 2353-2395, https://doi.org/10.1093/jeea/jvac018.

[13] Lee, N. and R. Brown (2016), “Innovation, SMEs and the liability of distance: the demand and supply of bank funding in UK peripheral regions”, Journal of Economic Geography, Vol. 17/1, pp. 233-260, https://doi.org/10.1093/jeg/lbw011.

[40] Leitner, Y. (2006), “Using collateral to secure loans”, Business Review, Federal Reserve Bank of Philadelphia Q2, pp. 9-16.

[11] Love, I. and M. Martínez Pería (2014), “How Bank Competition Affects Firms’ Access to Finance”, The World Bank Economic Review, Vol. 29/3, pp. 413-448, https://doi.org/10.1093/wber/lhu003.

[25] Mach, T. and J. Wolken (2012), “Examining the Impact of Credit Access on Small Firm Survivability”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.2051007.

[26] Musso, P. and S. Schiavo (2008), “The impact of financial constraints on firm survival and growth”, Journal of Evolutionary Economics, Vol. 18/2, pp. 135-149, https://doi.org/10.1007/s00191-007-0087-z.

[5] OECD (2022), Financing SMEs and Entrepreneurs 2022: An OECD Scoreboard, OECD Publishing, Paris, https://doi.org/10.1787/e9073a0f-en.

[8] OECD (2019), OECD SME and Entrepreneurship Outlook 2019, OECD Publishing, Paris, https://doi.org/10.1787/34907e9c-en.

[20] Petersen, M. and R. Rajan (2002), “Does Distance Still Matter? The Information Revolution in Small Business Lending”, The Journal of Finance, Vol. 57/6, pp. 2533-2570, https://doi.org/10.1111/1540-6261.00505.

[42] Petersen, M. and R. Rajan (1995), “The Effect of Credit Market Competition on Lending Relationships”, The Quarterly Journal of Economics, Vol. 110/2, pp. 407-443, https://doi.org/10.2307/2118445.

[39] Peters, R. and L. Taylor (2017), “Intangible capital and the investment-q relation”, Journal of Financial Economics, Vol. 123/2, pp. 251-272, https://doi.org/10.1016/j.jfineco.2016.03.011.

[28] Rajan, R. and L. Zingales (1998), “Financial Dependence and Growth”, The American Economic Review, Vol. 88/3, pp. 559-586, http://www.jstor.org/stable/116849 (accessed on 30 August 2024).

[10] Ryan, R., C. O’Toole and F. McCann (2014), “Does bank market power affect SME financing constraints?”, Journal of Banking & Finance, Vol. 49, pp. 495-505, https://doi.org/10.1016/j.jbankfin.2013.12.024.

[4] Stiglitz, J. and A. Weiss (1981), “Credit Rationing in Markets with Imperfect Information”, The American Economic Review, Vol. 71/3, pp. 393-410.

[9] Vives, X. (2001), “Competition in the Changing World of Banking”, Oxford Review of Economic Policy, Vol. 17/4, pp. 535-547, https://doi.org/10.1093/oxrep/17.4.535.

[16] Zhao, T., K. Luintel and K. Matthews (2020), “Soft information and the geography of SME bank lending”, Regional Studies, Vol. 55/4, pp. 679-692, https://doi.org/10.1080/00343404.2020.1851024.

Notes

Copy link to Notes

← 1. The focus of this part of the report is on external financing, although most firms combine external and internal resources to carry out their projects. Internal and external funds, therefore, are complements rather than substitutes. For this reason, the impact of external financial resources on enterprise performance also depend on other factors, such as the amount of retained earnings invested in the project or the management skills of the business owner.

← 2. Recent evidence from the United States shows that banks receiving positive liquidity shocks (i.e. increased deposits) react by increasing lending only to large SMEs (Lin, Tai and Xie, 2022[44]). This is consistent with the theory that enhanced bank charter value reduces bank risk-taking because larger banks are subject to stricter regulatory controls by financial authorities. There might be also other costs in credit markets that can lower the amount of credit available, particularly for small firms. For example, fixed costs in the lending process (e.g. loan application screening, loan management, or debt collection) mean that small business loans are more costly for banks to manage than larger business loans, thus resulting in relatively less capital available for small companies.

← 3. Collateralisation also involves a different set of issues. Physical assets lose value over time (depreciation), which means lenders have an incentive to sell collateralised assets as quickly as possible, thus resulting in the under-pricing of such assets (Leitner, 2006[40]). The transfer of control over assets also involves legal transaction costs. The collateralisation process also involves opportunity costs. The use of collateral imposes opportunity costs on borrowers by tying up assets that might be put to better use (Berger, Scott Frame and Ioannidou, 2011[45]), including through their sale or disposal to buy new assets (e.g. equipment and machinery). Credit guarantees, acting as collateral substitutes, can mitigate lending risk and encourage lending to firms that lack collateral (see Chapter 3).

← 4. These differences not only reflect differences in SME lending volumes but also differences in SME loan definitions across countries. This means that countries in which the SME loan definition is narrower (e.g. in terms of loan size or size of the firm receiving the loan) will tend to show lower national and regional values than those where the SME loan definition is broader. SME loan definitions in the OECD Scoreboard countries are available in Table A A.2 of the 2022 edition of the Scoreboard (OECD, 2022[5]).

← 5. OECD calculations based on Orbis data for AUT, DEU, ESP, FRA, GBR, GRC, HUN, IRL, JPN, LVA, POL, PRT and SVN referring to 2019.

← 6. OECD calculations based on Orbis data for AUT, DEU, ESP, FRA, GBR, GRC, HUN, IRL, JPN, LVA, POL, PRT and SVN referring to 2019.

← 7. The HHI is computed by the sum of the squares of the market shares of all firms and ranges between 0 (perfect competition) and 1 (monopoly). For example, a given country has two regions (A and B). Region A has only two firms. One works with Bank X and the other works with Bank Y. In Region B, there is only one firm, which works with Bank X. Thus, in Region A, Bank X and Bank Y work with 50% of the firms in the region, i.e. have a market share of 50% (one firm each). In Region B, Bank X has a market share of 100% (one out of one firm). Region A has an HHI of ${0.5}^{2} + {0.5}^{2} = 0.5$ and Region B has an HHI of $0^{2} + 1^{2} = 1$ . In the example above, the share of firms in a region using the three largest local banks is 100% in both regions.

← 8. A direct measure of competition is markups as used in the cited studies (Ryan, O’Toole and McCann, 2014[10]; Love and Martínez Pería, 2014[11]). Markups measure the market power of a firm and are computed as the difference between its marginal revenue (price) and marginal cost. An alternative indirect measure of competition is market concentration. Traditionally, market concentration is captured by the Herfindahl-Hirschmann index, computed from the sum of the squared shares of sales or revenue held by each active firm in the market; or by the share of the market held by the largest firms. As these mesures of market concentration capture both the competition among local banks and the benefit of strong lending relationships arising from it (compensating the negative implications of reduced bank competition), they are likely to underestimate the degree of competition among banks as evident in a number of older studies (Petersen and Rajan, 1995[42]; Fischer, 2000[43]; Beck, Demirguc-Kunt and Maksimovic, 2004[41]). In this report this is addressed by considering concentration in terms of the number of banks rather than sales volumes as done traditionally. The measures based on the number of banks have the advantage that they do not suffer from discounted concentration due to strong lending relationships and they consider all firm-bank relationships to be equally relevant, preventing an overemphasis on large firms and providing a more balanced view of financial access and banking market structure.

← 9. Small (TL3) regions represent administrative regions in most OECD countries, except Australia, Canada and the United States. With only a few exceptions a TL3 region is contained by a single TL2 region, except United States, where the TL3 regions are defined as Economic Areas that cross state borders. For Costa Rica, Israel and New Zealand, the TL3 level is equivalent to TL2. The OECD Territorial Level 3 includes 2 414 small regions. For more details about the OECD Territorial Level Classification, see https://www.oecd.org/en/data/datasets/oecd-geographical-definitions.html.

← 10. In December 2023, the US Department of Justice (DOJ) updated its Horizontal Merger Guidelines (https://www.justice.gov/d9/2023-12/2023%20Merger%20Guidelines.pdf, accessed 22 August 2024). As per the 2010 Horizontal Merger Guidelines, the previous version, an HHI would display moderate concentration if above 0.15 and high concentration if above 0.25 (https://www.justice.gov/atr/file/810276/dl?inline, accessed 11 July 2023).

← 11. A bank is considered to have presence in a region if at least one firm in that region reports having a banking relationship with it.

← 12. While bank branches are in direct touch with the borrowers and collect all relevant information, it is often the bank headquarters that makes the final decision on a credit application.

← 13. Agarwal and Hauswald (2010[23]) consider the distance between borrowers and lenders, in the United States, building on a dataset including all small business loan applications and small business credit offers by a major US bank over fifteen months. The authors find a major trade-off in the availability and pricing of credit, after controlling for firm-level quality indicators. The closer a firm is to one of the bank’s branch offices, the more likely the bank is to offer credit but also the higher the interest rate offered, all else being equal. Conversely, the closer a firm is to a competitor’s bank branch, the less likely it is to obtain credit from the observed bank but, if it does receive a credit offer, the lower the interest rate offered, which the authors interpret as a strategy of the bank to capture market shares from competitors. Nonetheless, the inclusion of proxies for the collection of soft information at the bank branch level reduces the importance of these effects, which reinforces the relevance of relationship lending in small business lending. Interestingly, the authors also find that when bank branches try to increase the interest rates on more distant borrowers, distant borrowers are more likely to switch lenders. Overall, these findings stress the role of relationship lending for SMEs to receive better financing conditions in local credit markets, as well as the importance of local competition in the banking sector.

← 14. Evidence suggests that improving access to credit pushes many firms away from the brink, improving their survival odds. In the United States, credit-constrained firms are significantly more likely to go out of business than non-constrained firms. Credit constraints and credit access variables are the most important factors predicting which small U.S. firms failed during 2004-2008, even after accounting for an extensive set of firm, owner, and market characteristics. Similarly, in France, financial constraints increased the likelihood of manufacturing firms exiting the market (Musso and Schiavo, 2008[26]). There is also evidence that reducing borrowing costs might increase firm survival. In the United Kingdom, increases in borrowing costs (i.e. interest burden) have been associated with reduced firm survival, particularly during financial crises and for bank-dependent, younger and non-exporting firms (Guariglia, Spaliara and Tsoukas, 2015[27]).

← 15. Both explanations of why there are fewer establishments are closely intertwined as growth is key to the business survival in the long term.

← 16. A standard deviation is a statistical measure of how much bank concentration in regions typically varies from the average. A one standard deviation higher bank concentration is equivalent to a 7.2 percentage points greater HHI.

← 17. Calculations based on the aggregation of industry-level averages for Valencia and regression coefficients reported in Figure 2.6.

← 18. The low exposure industry is the industry at the 25^th percentile of the reliance on intangible assets distribution. The high exposure industry is the industry at the 75^th percentile of the reliance on intangible assets distribution. These have been chosen for illustrative purposes. The estimates imply a linear relationship in the indicator on reliance on intangible assets, which means that for industries at the 20^th percentile, for example, the impact will be lower than at the 25^th percentile and for those in the 80^th percentile, it will be higher than for the 75^th percentile.

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
RIA	68	0.3	0.3	0.0	1.0	0.3	0.1	0.5	0.0	0.9

RIA	9	0.3	0.3	0.0	0.9	0.3	0.1	0.4	0.0	0.9

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
RIA	68	0.3	0.3	0.0	1.0	0.3	0.1	0.5	0.0	0.9

RIA	9	0.3	0.3	0.0	0.9	0.3	0.1	0.4	0.0	0.9

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

Governance

Health

Industry, business and entrepreneurship

Regional, rural and urban development

Science, technology and innovation

Society

Taxation

Trade

Energy

Nuclear energy

Transport

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Countries

Countries A - C

Countries D - I

Countries J - M

Countries N - R

Countries S - T

Countries U - Z

Regional and global engagement

Publications

Publications

Featured publications

Data

Data

Featured data

News & events

News & events

Featured events

About OECD

About

Engage with us

Work with us

Featured topics

Agriculture and fisheries

Climate change

Development

Digital

Economy

Education and skills

Employment

Environment

Finance and investment

	N	Mean	SD	Min	Max	Median	P25	P75	P5	P95
RIA	68	0.3	0.3	0.0	1.0	0.3	0.1	0.5	0.0	0.9

RIA	9	0.3	0.3	0.0	0.9	0.3	0.1	0.4	0.0	0.9