Financial dependence and growth: The role of input-output linkages

https://doi.org/10.1016/j.jebo.2018.11.024Get rights and content

Abstract

We widen the understanding of the finance-growth nexus by accounting for the indirect effect of financial development through input-output (IO) linkages in determining the growth of industries across countries. If financial development is expected to promote disproportionately more the growth of industrial sectors that are more in need of external finance, it also favours more the industries that are linked by IO relations to more financially dependent industries. We explore this new channel in a sample of countries at different development stages over the period 1995–2007. Our results highlight that financial development, besides easing the growth of industries highly dependent on external finance, also fosters the growth of industries strongly linked to highly financially dependent upstream industries. Moreover, the indirect effect - propagated through IO linkages - of finance has a higher and non-negligible role compared to the direct effect and its omission leads to a biased and underestimated perception of the role of finance for industries’ growth.

Introduction

A general consensus in the growth literature exists on the positive repercussions that a well developed financial sector has on aggregate output (Levine, 2005).1 In particular, banking industry is pivotal to channel financial resources towards more innovative firms and sectors which are plagued by asymmetric information problems, and whose investment expenditures are strongly constrained by the availability of external finance. In this view, there would be a link between financial development and the growth of a non-financial industry S going through the support that banks provide directly to firms in that sector for seizing growth opportunities and responding to global shocks (Beck, Levine, Loayza, 2000, Fisman, Love, 2007, Rajan, Zingales, 1998).

However, quantitative macroeconomic models and empirical studies have clearly documented that credit supply shocks propagate in the economy through the network of IO linkages across firms and industries (Alfaro, Garcia-Santana, 2017, Bigio, La’O, 2016, Dewachter, Tielens, Van Hove, 2017). This means that financial development can be expected to foster the growth of a non-financial sector S also indirectly, through the support to investments of financially dependent firms in downstream and upstream sectors, which buy outputs from and supply inputs to sector S. In this paper we analyze the role of input-output (IO) linkages in transmitting and amplifying the effects of financial development on the growth of industrial sectors. More specifically, we explore this conjecture by extending the cross-country industry approach initially proposed by Rajan and Zingales (1998) to include IO linkages among sectors.2

The idea that IO linkages are at the heart of the process of economic development has a long tradition in the economic literature, dating back to Scitovsky (1954), Fleming (1955) and Hirschman (1958). In a nutshell, the development of an industry activates sizable positive effects on firms in other industries, which provide inputs to the former one (backward linkage effects) and use its outputs as inputs in the production process (forward linkage effects).

Recently, literature has refocused the attention on the importance of IO linkages for productivity improvements and economic growth. Ciccone (2002) develops a model of industrialisation in which, consistently with empirical regularities, increasing-return technologies are highly intensive in the use of intermediate inputs and are adopted throughout the chain of intermediate inputs. In this context, the introduction of new industrial technologies generate a large increase in aggregate productivity and income, even if the productivity improvements at the firm level are small. By the same token, however, minor frictions in IO linkages (due, for example, to imperfections in financial markets) can also be expected to cause great differences in productivity and income levels across countries. This view is corroborated by Acemoglu et al. (2007) and Jones (2011). The former show that greater contractual incompleteness leads to the adoption of less advanced technologies, and that this effect is more pronounced when there is strong complementarity among intermediate inputs. Jones (2011) explores the role of input linkages and complementarity and shows that frictions along the production chain can sharply reduce aggregate output.

On the empirical side, Bartelme and Gorodnichenko (2015) at the country level document the “Hirschman conjecture”, that the strength of IO linkages is positively associated to output per worker and total factor productivity.3 In the same vein, Fadinger et al. (2016) find that a multi-sector model with IO linkages explains cross-country income differences much better than a model abstracting from IO linkages.4 In addition, a number of studies have shown the role of IO linkages among industrial sectors in generating aggregate fluctuations, finding that the chain of input-output relations contributes to spread out and amplify the effects of idiosyncratic individual or sectoral shocks over the entire economy (Acemoglu, Carvalho, Ozdaglar, Tahbaz-Salehi, 2012, Carvalho, 2014, Di Giovanni, Levchenko, Mejean, 2014). In particular, Acemoglu et al. (2016) find that in the United States industries’ value added, employment and labor productivity respond more to indirect supply and demand shocks affecting the IO chain than to direct shocks hitting the same industry. Moreover, when they distinguish between upstream and downstream IO linkages, they show that demand shocks propagate upwards to input-supplying industries, while supply shocks propagate downwards to customer industries.5

Relatively unexplored are the factors that contribute to explain the aggregate impact of IO linkages, and in particular the role of financial markets in the propagation of shocks in the presence of inter-sectoral IO linkages. As partial exceptions, Acemoglu et al. (2009) document that countries characterised by high contracting costs and low financial development are concentrated in industries where firms tend to be more vertically integrated, relying less on IO linkages. Furthermore, within each industry, they show that financial development helps firms to circumvent contracting costs by providing them with financial resources necessary to grow in size and vertically integrate activities. Bigio and La’O (2016) introduce financial frictions in a multi-sector network framework à la Acemoglu et al. (2012). They calibrate the theoretical model to the IO structure of the U.S. economy during the 2007–2008 Great Recession, and show that IO linkages amplify the effect of financial shocks on aggregate output by a factor between two and six, relative to the case of an hypothetical industrial structure with no interactions across sectors. Finally, Alfaro and Garcia-Santana (2017) and Dewachter et al. (2017), using very detailed firm level data on Spanish and Belgian firms respectively, find that individual credit supply shocks strongly propagate to other firms in the value chain by affecting their capital investments, export sales and output.

In this paper, we depart from this business cycle perspective and for the first time, to the best of our knowledge, we investigate whether and to what extent, in a long run perspective, the impact of financial development on the growth rate of an industry S is amplified by IO linkages connecting that industry to other industries which are in need of external finance.

Our intuition is simple and extends the same argument advanced by Rajan and Zingales (1998) for financial dependent sectors to the whole IO chain. If financial intermediaries mitigate asymmetric information problems that hamper firms’ access to credit, sectors buying from and selling to highly financially dependent sectors should grow at a disproportionately faster pace in countries with a well developed and functioning financial sector. To illustrate this point, consider a number of sectors connected by IO linkages. Each sector produces an output which is used in downstream industries as an input, and buy inputs from upstream sectors. If financial development allows firms in a sector S1 along this IO chain to have a larger access to credit, capital accumulation in this sector increases, as well as productivity. This possibly causes an increase in the demand of inputs by S1 produced by firms in an upstream sector S2, and a decrease in the price of output in S1 used as an input by firms in a downstream sector S3. As a result, the output of S2 and S3 increase, as well as their investment opportunities. If firms in these sectors are financially dependent, the growth opportunities produced by the higher investments in sector S1 can be better exploited where they can rely on a well developed financial sector. Beyond these first order interconnections, the increase in investments and productivity in sectors S2 and S3 may create, in turn, opportunities of growth in their upstream and downstream sectors that a developed financial sector allows to actually take, and so on.6 In this way, the impact of financial development on the growth of an industry S reflects the financial support that banks provide along the whole IO chain linking this industry to other industries, and it is the greater the more financial dependent are these industries.

Our empirical analysis considers a sample of countries at different development stages over the period 1995–2007. Following Rajan and Zingales (1998) and Kroszner et al. (2007), the measure of financial dependence varies by sector and is calculated on the basis of U.S. firms’ cumulated capital expenditures and cash flows over the period 1990–2007. IO linkages, instead, refer to the first year - 1995 - of our sample and are retrieved from the OECD IO database. For every industry, we distinguish between upstream linkages with industries supplying intermediate goods and downstream linkages with customer industries. Then, we compute two indicators reflecting the overall financial dependence of upstream and downstream industries where the financial dependence of each upstream/downstream sector is weighted by its share in the industry’s total purchases/sales. In the baseline model, financial development is proxied by the standard ratio of domestic credit over GDP (WDI, 2015).

We model the growth of real value added of an industry as dependent on the interactions between a country’s financial development and the financial dependence of the same industry and of its upstream and downstream industries. Our results confirm the finding by Rajan and Zingales (1998) of a positive contribution of financial development to the growth of financially dependent sectors. However, we also find that, quantitatively, the direct effect of financial development on the industry’s growth rate is smaller than the indirect effect that financial development exerts by relaxing financial constraints of the sectors linked to the industry by IO relations. More specifically, we find that the latter effects operate through the financial support that a developed financial sector warrants to upstream industries selling intermediate inputs. The sustain of well developed financial intermediaries to downstream sectors, instead, is not significantly associated with the growth of supplying sectors. This is consistent with the hypothesis that, if we focus the analysis on intermediate goods, the positive role of financial development for the investments of financially dependent firms in upstream sectors fosters productivity improvements (Beck et al., 2000), thus reducing the prices of their goods and opening new opportunities of investment for downstream customer sectors (Acemoglu, Akcigit, Kerr, 2016, Hirschman, 1958). The search for the channels behind the baseline evidence corroborates this interpretation. By contrast, the benefits of financial development are not significantly transmitted across sectors by its potential effects on the demand for intermediate inputs.

Our results are robust to several checks. In order to account for the potential endogeneity of financial development, we implement an instrumental variable (IV) approach based on the close relation between the quality of a country’s legal system and its financial development (Beck, Demirguc-Kunt, Levine, 2003, La Porta, de Silanes, Shleifer, Vishny, 1998). Furthermore, we use alternative measures of financial development, industry’s growth, financial dependence and IO linkages. Also, we explore competing and potentially confounding factors which could affect the growth of industries within a country. More precisely, we consider the impact of countries’ development stage, as a proxy of the extent of maturity of a country’s industries, human capital endowment, as a further alternative growth determinant of more skill intensive industries, and foreign direct investment (FDI) as an additional source of finance in recipient economies, especially for capital intensive industries. We also confirm our baseline evidence on the original sample used by Rajan and Zingales (1998).

In our country-industry framework, we further inspect the existence of some potential heterogeneity in the finance-growth nexus. We, first, test whether the documented non-linearity of the nexus is also valid for financial development working through IO linkages (Arcand, Berkes, Panizza, 2015, Cecchetti, Kharroubi, 2012, Demirguc-Kunt, Feyen, Levine, 2013, Easterly, Islam, Stiglitz, 2000). Finally, we separately test the role of finance in the 90s, when a relevant number of banking crises occurred, and in the 2000s, during the productivity slowdown decade.

The rest of the paper is organised as follows. The next Section presents the empirical model, while Section 3 presents the data and describes the computation of financial dependence for a sector, for its upstream and downstream sectors. Section 4 shows the results from the estimation of the empirical model, all the array of robustness checks and the test for non-linearity in the finance-growth nexus through IO linkages. Finally, Section 6 concludes the work.

Section snippets

Empirical model

Our main testing hypothesis is that industrial sectors that are linked in the IO chain to industries that are more dependent on external finance grow at relatively higher rates in countries whose financial sector is more developed. To test this hypothesis we need a model of industry growth where the effect of financial development is heterogeneous across sectors according to the financial dependence of their upstream suppliers and downstream buyers. Therefore, we estimate the following

Industry level data

Data on countries’ value added by industry are retrieved from the UNIDO Industrial Statistics Database for the period 1995–2007 at the 2-digit level of the ISIC revision 3.7

Baseline results

Table 1 shows baseline estimation results from model (1) on the 1996–2007 cross-section of country-sector growth rates. Financial development not only favours directly the growth of a financially dependent sector i, but it also matters indirectly through the support to financially dependent upstream industries supplying inputs to sector i. More specifically, in columns [2]-[4] financial development seems to promote economic growth of sectors through both output and input linkages. However, when

Channels and heterogeneity of the finance-growth nexus

After proving the robustness of our findings, in this section we investigate the channels through which financial development foster industries’ growth. In addition, we analyze whether the positive direct and indirect average effects of financial development on the industry growth hide heterogeneity in the finance-growth nexus in two dimensions: the size of the financial sector and the time period under analysis.

Conclusion

For the first time, in this paper we have studied the role of IO linkages in amplifying the positive effect of countries’ financial development on the growth of manufacturing sectors. We have extended the Rajan and Zingales’s (1998) empirical country-sector growth model by including the interaction of upstream and downstream sectors’ financial dependence with countries’ financial development. In a cross-section of countries at different development stages, observed in the time span 1995-2007,

References (51)

  • P.L. Rousseau et al.

    What is happening to the impact of financial deepening on economic growth?

    Econ Inq

    (2011)
  • D. Acemoglu et al.

    Networks and the macroeconomy: an empirical exploration

    NBER Macroecon Annu

    (2016)
  • D. Acemoglu et al.

    Contracts and technology adoption

    Am Econ Rev

    (2007)
  • D. Acemoglu et al.

    The network origins of aggregate fluctuations

    Econometrica

    (2012)
  • D. Acemoglu et al.

    Determinants of vertical integration: financial development and contracting costs

    Journal of Finance

    (2009)
  • L. Alfaro et al.

    Credit Supply Shocks, Network Effects,and the Real Economy

    Working Paper

    (2017)
  • J. Arcand et al.

    Too much finance?

    Journal of Economic Growth

    (2015)
  • J.-N. Barrot et al.

    Input specificity and the propagation of idiosyncratic shocks in production networks

    Quarterly Journal of Economics

    (2016)
  • D. Bartelme et al.

    Linkages and Economic Development

    NBER Working Papers, 21251

    (2015)
  • M. Berlemann et al.

    Estimating aggregate capital stocks using the perpetual inventory method. a survey of previous implementations and new empirical evidence for 103 countries

    Review of Economics

    (2014)
  • S. Bigio et al.

    Financial frictions in production networks

    NBER Working Papers, 22212

    (2016)
  • C. Borio et al.

    Labour reallocation and productivity dynamics: financial causes, real consequences

    BIS Working Papers, 534

    (2015)
  • V.M. Carvalho

    From micro to macro via production networks

    Journal of Economic Perspectives

    (2014)
  • S. Cecchetti et al.

    Reassessing the impact of finance on growth

    BIS Working Papers, 381

    (2012)
  • S. Cecchetti et al.

    Why does financial sector growth crowd out real economic growth?

    BIS Working Papers, 490

    (2015)
  • Cited by (11)

    • Industrial linkage and clustered regional business cycles in China

      2023, International Review of Economics and Finance
    • How do VAT reforms in the service sectors impact TFP in the manufacturing sector: Firm-level evidence from China

      2021, Economic Modelling
      Citation Excerpt :

      The effect of the VAT program is smaller for SOEs than for non-SOEs, and is larger for labor-intensive firms than for capital-intensive firms; manufacturing firms with high intensities of exporting activities are affected only through BLs. Our definitions of FLs and BLs are consistent with those provided in the literature (e.g., Javorcik, 2004; Debaere et al., 2010; Turco et al., 2019). For example, Debaere et al. (2010) use the Chinese and South Korean input-output tables to measure the proportion of each sector’s output internationally supplied to or purchased from another sector.

    • Structural evolution of real estate industry in China: 2002-2017

      2021, Structural Change and Economic Dynamics
      Citation Excerpt :

      They believe that the industrial asymmetry is the origin of economic crisis because even though a seemingly weak shock from a tiny industry is likely to significantly impact the whole economic system due to structural asymmetry. This important finding is also confirmed by subsequent input-output network studies (e.g. Tsekeris, 2017; Zhang, 2018; Turco, 2019; Contreras, 2020). But in terms of the studies focusing on sector level, the number of researches is not abundant (e.g. Chan et al., 2016; Wang and Wang, 2019a; Wang and Wang, 2019b).

    • Global networks, local specialisation and regional patterns of innovation

      2020, Research Policy
      Citation Excerpt :

      However, using total employment in manufacturing as standardising factor would have excessively penalised regions where services prevail. Indeed, it is well recognized that knowledge-intensive services sectors play a crucial role in the production and diffusion of knowledge in the economic system (Muller and Zenker, 2001; Lo Turco et al., 2019).3 Moreover, patents have long been considered as the best, though not perfect, output for innovation activity at the regional level (Acs et al., 2002; Bottazzi and Peri, 2003; Miguelez and Moreno, 2015; Marrocu et al., 2013).

    • Does going tough on banks make the going get tough? Bank liquidity regulations, capital requirements, and sectoral activity

      2020, Journal of Economic Behavior and Organization
      Citation Excerpt :

      Dropping FinDev and its interaction with ExtDep from the specification could instead result in omitted variable bias. Relying on a vast literature that emphasizes a significant effect of a country's legal system on the development of the financial sector (La Porta et al., 1998) and following Beutler and Grobéty (2019) and Turco et al. (2019), we use legal origins and rule of law as instruments. Exact variable definitions and sources are in Appendix Table A2.

    View all citing articles on Scopus

    We thank participants of the conference “Finance and Growth in the Aftermath of the Crisis” and seminar participants at the University of Messina, the University of Trento and the Univesrity of Florence. We are particularly grateful to Giovanni Busetta, Matteo Lanzafame, Fabio Pieri, Giorgio Ricchiuti and Chiara Tomasi for their helpful comments and suggestions. Finally, we thank Alex Borisov for providing us the COMPUSTAT data useful for the calculation of the financial dependence indicators and Luc Laeven for sharing the do file with the calculation of the financial dependence indicator for the 80s and the 90s.

    View full text