Endogenous factors

Compared with large firms, SME are more exposed to problems of information asymmetry, which imply obstacles in accessing to external finance. In addition, SME have restricted access to the stock market and so their financing sources are little diversified. Various studies conclude that small firms use internal funds as far as possible, and only when these are insufficient do they typically incur debt.

Gertler and Gilchrist (1994), Gilchrist and Himmelberg (Reference Gertler and Gilchrist1995) and Moyen (Reference Moyen2004) argue that younger, smaller firms tend to invest more using their cash flow. Concerning the relationship between the size variable and firm growth, the results of empirical studies diverge. Some studies identify a positive relationship between firm size and growth (Samuels Reference Vermeulen1965; Prais Reference Sutton1976; Aces & Audretsch, Reference Aces and Audretsch1990), while others conclude a negative relationship between size and firm growth (Dunne & Hughes, Reference Dunne and Hughes1994; Becchetti & Trovato, Reference Becchetti and Trovato2002). Beck, Demirgüç-Kunt, Laeven, and Maksimovic (Reference Beck, Demirgüç-Kunt, Laeven and Maksimovic2006) defend firm age is a proxy for the financial restrictions they face in funding their investment. The authors also argue that, as firms come through the first years of their life, the barriers they face in getting finance for funding their investment become fewer.

Devereux and Schiantarelli (Reference Devereux and Schiantarelli1990) and Oliner and Rudebusch (Reference Peterson and Rajan1992) conclude that young firms show greater investment sensitivity to cash flow, suggesting that the youngest firms face greater obstacles in obtaining external finance.

Gilchrist and Himmelberg (Reference Gertler and Gilchrist1995) and Fazzari, Hubbard, and Petersen (Reference Fazzari, Hubbard and Petersen1988) argue that regardless of the high level of investment of the youngest and smallest firms, this investment very much depends on the availability of cash flow. Fagiolo and Luzzi (Reference Fagiolo and Luzzi2006) corroborate these arguments, concluding that liquidity restrictions create a negative effect on firm growth, with the smallest firms growing more after overcoming these restrictions.

Based on the previous arguments, the following research hypotheses are formulated:

Several previous studies (Evans & Jovanovic, Reference Evans and Jovanovic1989; Gertler & Gilchrist, 1994; Gilchrist & Himmelberg, Reference Gertler and Gilchrist1995; Hartarska, Reference Hartarska2001; Beck et al., Reference Beck, Demirgüç-Kunt, Laeven and Maksimovic2006; Brown, Fazzari, & Peterson, Reference Brown, Fazzari and Peterson2009) identify firm investment sensitivity of firms investment to cash flow that increases for SME, and consequently the growth of these firms will be dependent on cash flow generated through the firms’ activity.

Insufficient internal finance implies that SME will resort to external finance, which frequently refers to credit obtained from suppliers and banks. However, the strong dependence of SME on debt finance, namely bank loans, implies the payment of principal and interests, consequently a reduction of cash flow. In turn, decreasing cash flow implies greater difficulties for firms to meet their commitments with creditors and a reduction of internal finance. This reduction in the availability of financial resources has negative effects on firm investment.

SME frequently suffer from a shortage of know-how, which in turn means they lack a reputation with creditors. Consequently, lenders increase the cost of capital in order to minimize the risk of granting credit to those firms (Diamond, Reference Diamond1989; Brennan & Subrahmanyam, Reference Brennan and Subrahmanyam1996; Easley & O’Hara, Reference Easley and O’Hara2004). In turn, there is a greater likelihood of bankruptcy associated with SME, which contributes to increased risk for potential creditors (Pettit & Singer, Reference Samuels1985; Mueller & Zimmermann, Reference Mueller and Zimmermann2009).

The ECB (2013) concludes that the difficulties in obtaining bank loans restrain the investment and growth in European SME. Problems in accessing credit frequently affects SME’s survival, as these firms may not be able to come through periods of economic recession, especially when the bank system is not operating as normal (Collignon & Esposito, Reference Collignon and Esposito2013).

The financial crisis beginning at the end of 2007 led banks to frequently cancelling credit to SME, who consequently were left with a lack of funds to meet their investment and growth needs. SME especially those of micro-size, dominate the business world in the different countries of Europe, and revealed less capacity to increase their investment in the period following the financial crisis beginning in 2008 (ECB, 2013). According to Collignon and Esposito (Reference Collignon and Esposito2013), this fact is not only related to firm size, but also to the of lack economic stability, particularly the uncertainty concerning demand.

In the European context, more than 70% of external finance is provided by banks and under 30% by the financial market. This composition of financing sources has consequences for the process of adjustment to cope with financial crisis, making it slower and with less recourse to a bond issue compared with what occurs in the United States. This situation penalizes SME particularly, as they do not turn to bond loans due to the high costs and their lack of collateral, meaning reduced investment.

Furthermore, according to Collignon and Esposito (Reference Collignon and Esposito2013), the fragmentation of financial markets in peripheral European countries has been particularly negative for SME, due to the high cost of finance and “anaemic” economic growth, which has contributed to accentuating SME’s problems in terms of access to external finance. In turn, this has contributed to the drastic reduction in productive investment (International Monetary Fund [IMF], 2013).

In the Portuguese case, the collapse in domestic demand, aggravated by reduced credit and the adoption of austerity measures in a short space of time (between 2011 and 2014), contributed to the bankruptcy of a large number of Portuguese firms, particularly small ones.

Although the monetary policy implemented by the ECB has been marked by maintaining historically low-interest rates, this was not reflected in the interest rates applied to firms in Southern Europe, and in Portugal in particular (Leão, Martins, & Gonçalves, Reference Luzio2014).

The financial fragmentation of the Eurozone has not allowed monetary policy to act anti-cyclically, as would be ideal in a context of crisis and severe budget restrictions. Following the evolution of the weight of sovereign debt and financial fragmentation of the Eurozone, the bank credit granted to firms and families recorded severe contraction in the period between 2010 and 2014 (Leão, Martins, & Gonçalves, Reference Luzio2014). The interest rate spreads of new loans to Portuguese firms increased very substantially between 2010 and 2013. For that reason, Portuguese firms faced interest rates considerably above those of their European competitors (Leão, Martins, & Gonçalves, Reference Luzio2014).

Based on the above arguments, we formulate the following research hypotheses:

Exogenous factors

The macro-economic environment has a strong influence on SME growth, and for that reason exogenous factors such as GDP and the interest rate are analyzed as possible determinants of those firms’ growth (Higson, Holly, & Kattuman, Reference Higson, Holly, Kattuman and Platis2002; Higson, Holly, Kattuman, & Platis, Reference Honjo and Harada2004; Beck et al., Reference Beck, Demirgüç-Kunt, Laeven and Maksimovic2006).

Bernanke and Gertler (Reference Bernanke and Gertler1989) and Gilchrist, Himmelberg, and Huberman (Reference Gilchrist and Himmelberg2005) defend a connection between monetary policy and investment, as the effect that policy causes on the cost of capital ends up influencing firms’ investment, so that the fall in the level of investment can provoke a propagation of shocks through the credit channel.

Gilchrist, Himmelberg, and Huberman (Reference Gilchrist and Himmelberg2005) find a negative relationship between interest rates and capital. Changes in monetary policy, namely in terms of the interest rate, influence firms’ investment and debt capacity through the effect on cash flow. Indeed, the cash flow of highly indebted firms will show greater sensitivity to alterations in the interest rate than the cash flow of firms with a minimum level of debt (Mills, Morling, & Tease, Reference Moyen1995).

Therefore, we can expect the effect of interest rates to be more pronounced on SME investment, since these firms are more sensitive to changes in the cost of credit, compared with the case of large firms. In fact, according to Wagenvoort (Reference Wagenvoort2003), the interest rate tends to be considerably higher for SME, due to the greater risk of bankruptcy of this type of firm.

Pandit, Mittal, Roy, and Ghosh (Reference Pettit and Singer2006) found that the investment of the youngest SME is more vulnerable to monetary contractions. Therefore, the negative relationship between the interest rate and investment is expected to be even more pronounced, in that the information asymmetry implicit in the relationships formed between SME owners/managers and creditors is of greater relevance, contributing to the increased cost of credit compared with what is found in the context of older SME.

Authors such as Bernanke and Gertler (Reference Bernanke and Gertler1989), Gertler and Gilchrist (1994), Oliner and Rudebusch (Reference Peterson and Rajan1992), Bernanke, Gertler, and Gilchrist (Reference Bernanke, Gertler and Gilchrist1996) and Vermeulen (Reference Hyytinen and Väänänen2002) argue that firms that are restricted financially and operate in imperfect markets, as is the case of SME, are more affected by an economic recession, in this way increasing their financial limitations.

In an economic recession (as the one that led to the adoption of austerity measures) we can expect external capital markets to be more restrictive, as far as the granting of credit is concerned, which affects investment in a given economy, in general, and SME investment in particular. In the situation described above, SME are pushed towards a reliance on cash flow, which may have experienced severe reduction due to the above-mentioned economic recession (Kaufmann & Valderrama, Reference Kaufmann and Valderrama2008). In situations of economic growth, on the one hand, SME have easier access to credit, and on the other they present higher levels of cash flow, which is an element facilitating investment. In addition, the highly uncertain climate characteristic of economic recessions also has negative repercussions on investment, inasmuch as SME tend to postpone investment options, in the expectation of future economic growth (Fuss & Vermeulen, Reference Fuss and Vermeulen2004).

In periods of economic growth, characterized by greater growth opportunities, SME can take advantage of those opportunities by investing in profitable projects, resorting for this purpose to their cash flow or credit, where access will be more favourable at such times (Atanasova, Reference Atanasova2003).

Based on these arguments, the following research hypotheses are formulated:

Data set

The present study uses data gathered from the AMADEUS database, supplied by Bureau Van Dijk. This contains economic and financial information on European firms. In this study, the firms forming the research samples meet the definition of a small firm established by the European Commission (Recommendation 2003/36/CE), according to which, a business unit is considered a small firm, when: (i) it employs fewer than 50 people; and (ii) its turnover or annual balance sheet total does not exceed €10 million. Also, according to the European Commission’s recommendation, a firm is considered medium-sized when at least two of the following criteria are met: (i) having between 50 and 250 employees; ii) total assets between €10 million and €43 million; and (iii) turnover between €10 million and €50 million. Based on these criteria, 3008 SME were obtained and data was collected for the period between 2003 and 2012.

Arellano and Bond (Reference Arellano and Bond1991) conclude that the use of dynamic panel estimators implies that the units subject to analysis are included in the sample for at least 4 consecutive years. This is due to the need to include all units of analysis in the second-order autocorrelation test, which is essential to validate the results obtained with the dynamic estimators. We obtained a research sample with an unbalanced structure and composed of 1719 SME. The use of unbalanced panels, which allows for both firm entry and exit, partially mitigates potential selection and survivor bias. It should be noted that all estimations include annual dummy variables in order to measure the effects of the economic situation on variations in SME investment.

In addition, in all the estimations carried out, we also consider industry sector dummy variables to measure the impact of potentially different relationships between determinants and investment, according to SME belonging to different industry sub-sectors. We consider the following industry sector dummy variables: (1) agriculture, forestry and fishing; (2) construction; (3) manufacturing; (4) transport, storage and communication; (5) wholesale and retail; and (6) services.

Econometric method and variables

Regarding the econometric method, the GMM system (1998) dynamic estimator was used.

Using the GMM system (1998) estimator, the relationships between factor determinants and SME growth can be presented as follows:

(1) $$\eqalignno{ GROW_{{i,t}} \,& \,{\equals}\,{\rm \beta }_{0} {\plus}{\rm \beta }_{1} GROW_{{i,t{\minus}1}} {\plus}{\rm \beta }_{2} CF_{{i,t{\minus}1}} {\plus}{\rm \beta }_{3} LEV_{{i,t{\minus}1}} {\plus}{\rm \beta }_{4} SIZE_{{i,t{\minus}1}} \cr & {\plus}{\rm \beta }_{5} AGE_{{i,t{\minus}1}} {\plus}{\rm \beta }_{6} IR_{{t{\minus}1}} {\plus}{\rm \beta }_{7} GDP_{{i,t}} {\plus}S_{S} {\plus}d_{t} {\plus}v_{i} {\plus}e_{{i,t}} , $$

where $$v_{i} $$ is the non-observable individual effects; and $$e_{{i,t}} $$ the error; d _t corresponds to the year dummies; and, S _s the industry sector dummies.

The variables used in this study were measured as follows: $$GROW_{{i,t}} $$ is the growth, given by the ratio of net total assets variation in the current period to total assets in the previous period; $$GROW_{{i,t{\minus}1}} $$ is the lagged growth variable; $$CF_{{i,t{\minus}1}} _{{}} $$ is cash flow in the previous period, given by the ratio of earnings before taxes plus amortizations and depreciations in the previous period divided by total assets in the previous period; $$LEV_{{i,t{\minus}1}} $$ is debt in the previous period, given by the ratio of total debt in the previous period to total assets in the previous period; $$SIZE_{{i,t{\minus}1}} $$ is size in the previous period, given by the logarithm of total assets in the previous period; $$AGE_{{i,t{\minus}1}} $$ is firm age in the previous period, given by the logarithm of the number of years of existence of the firm in the previous period; $$IR_{{i,t{\minus}1}} $$ is the market interest rate in the current period, represented by the Euribor 3-month interest rate; and $$GDP_{{i,t}} $$ is gross domestic product in the current period, given by the logarithm of gross domestic product.

The model (1) was rewritten to allow the assessment of a hypothetical moderating effect of the financial crisis and especially in terms of SME’s financing sources:

(2) $$\eqalignno{ GROW_{{i,t}} & \,{\equals}\,{\rm \beta }_{0} {\plus}{\rm \beta }_{1} GROW_{{i,t{\minus}1}} {\plus}{\rm \beta }_{2} CF_{{i,t{\minus}1}} {\plus}{\rm \beta }_{3} LEV_{{i,t{\minus}1}} {\plus}{\rm \beta }_{4} SIZE_{{i,t{\minus}1}} \cr & {\plus}{\rm \beta }_{5} AGE_{{i,t{\minus}1}} {\plus}{\rm \beta }_{6} IR_{{i,t{\minus}1}} {\plus}{\rm \beta }_{7} GDP_{{i,t}} {\plus}{\rm \beta }_{8} D_{{aust}} {\plus}{\rm \beta }_{9} D_{{aust}} {\times}CF_{{i,t{\minus}1}} {\plus}{\rm \beta }_{{10}} D_{{aust}} {\times}LEV_{{i,t{\minus}1}} \cr & {\plus}S_{S} {\plus}v_{i} {\plus}e_{{i,t}} , $$

where the dummy variable $$D_{{aust}} $$ and interaction variables are defined as follows: $$D_{{aust}} $$ is the dummy variable representing the period that begins with the financial crisis, that is 2008 until 2012; $$D_{{aust}} $$ ×CF _i,t−1 is an interaction variable measuring the strength of the relationship between CF _i,t−1 and $$GROW_{{i,t}} $$ ; and $$D_{{aust}} $$ ×LEV _i , _t−1 is an interaction variable measuring the strength of the relationship between LEV _i,t−1 and $$GROW_{{i,t}} $$ .

Dynamic panel estimators

Given the dynamic nature of firm growth (DeMarzo & Fishman, Reference DeMarzo and Fishman2007), use of dynamic estimators is more appropriate than other estimation methods such as fixed or random effect panel models or ordinary least square regressions. In order to estimate the relationships between determinants and SME growth, dynamic panel estimators are used, namely the GMM system (1998) dynamic estimator.

Compared with static panel models, dynamic estimators present the following advantages (Arellano & Bond, Reference Arellano and Bond1991): (1) greater control of endogeneity; (2) greater control of the collinearity of explanatory variables; and (3) greater efficiency in controlling the effects of the absence of important explanatory variables in explaining the dependent variable. In addition, dynamic estimators have the advantage of correctly estimating the relationship between the dependent variable in the current and previous periods.

Estimating relationships between determinants and growth using static panel models (fixed or random effect models or ordinary least square regressions) would lead to bias of the estimated parameters as a consequence of the correlation between $$v_{i} $$ and $$GROW_{{i,t{\minus}1}} $$ , and between $$e_{{i,t}} $$ and $$GROW_{{i,t{\minus}1}} $$ . Resorting to dynamic panel estimators, as a consequence of estimating Equations 1 and 2 in first differences, the correlation between $$v_{i} $$ and $$GROW_{{i,t{\minus}1}} $$ is eliminated. Using lags of growth and the explanatory variables of growth creates orthogonal conditions between $$e_{{i,t}} $$ and $$GROW_{{i,t{\minus}1}} $$ , eliminating the correlation between those variables.

However, the results obtained with the GMM system (1998) estimator can only be considered valid on the following conditions: (1) if the restrictions, due to using the instruments, are valid; and (2) if there is no second-order autocorrelation.

To test the validity of the restrictions we use the Hansen test. The null hypothesis proposes the validity of the restrictions imposed by using the instruments. Rejecting the null hypothesis, we conclude that the restrictions are not valid, and so the results obtained cannot be considered robust.

In addition, we test for the existence of first- and second-order autocorrelation. The null hypothesis is non-existence of autocorrelation. Rejecting the null hypothesis of non-existence of second-order autocorrelation, we conclude that the results obtained cannot be considered robust.

In brief, for the results to be considered valid and consequently open to economic interpretation, the restrictions imposed by using the instruments have to be valid and there can be no second-order autocorrelation.