چه زمانی اهرم های فشار باعث آسیب به رشد بهره وری می شود؟تجزیه و تحلیل در سطح بنگاه ها

In the wake of the global financial crisis, several macroeconomic contributions have highlighted the risks of excessive credit expansion. In particular, too much finance can have a negative impact on growth. We examine the microeconomic foundations of this argument, positing a non-monotonic relationship between leverage and firm-level productivity growth in the spirit of the trade-off theory of capital structure. A threshold regression model estimated on a sample of Central and Eastern European countries confirms that TFP growth increases with leverage until the latter reaches a critical threshold beyond which leverage lowers TFP growth. This estimate can provide guidance to firms and policy makers on identifying “excessive” leverage. We find similar non-monotonic relationships between leverage and proxies for firm value. Our results are a first step in bridging the gap between the literature on optimal capital structure and the wider macro literature on the finance-growth nexus.

The global financial crisis has revived interest in the risks of excessive credit expansion at the macroeconomic level. In a recent paper titled “Too much finance”, Arcand et al. (2011) identify a threshold level of domestic credit to the economy beyond which output growth begins to fall. Reinhart and Rogoff (2010) identify a similar non-monotonic relationship between public debt and growth. In this paper we argue that there can be a non-monotonic relationship between leverage and productivity growth at the firm level; using insights from the macroeconomic and corporate finance literatures, we identify a threshold level of leverage beyond which further increase in leverage can lower firm-level productivity growth. Corporate leverage decisions are among the most important decisions made by firm executives and have been the focus of intense scrutiny since Modigliani and Miller (1958). Financial conditions in the corporate sector not only affect firm performance but, as macroeconomists have long recognized, they can have a powerful effect on macroeconomic outcomes. The literature on “financial accelerators” is concerned with the role of financial conditions in amplifying shocks to the economy (see e.g. Bernanke et al. (1999)) while the literature on the finance-growth nexus (e.g. see Ang (2008) for a recent survey) is concerned with their contribution to long-term growth. The present paper is a first attempt at bridging the gap between the literature on optimal capital structure and the macroeconomic literature on finance-growth linkages. We use threshold regressions to investigate the non-monotonic relationship between leverage and several indices of firm performance, and the extent to which this relationship varies across types of firms. Among all possible measures of firm performance, our analysis particularly focuses on total factor productivity (TFP) growth for several reasons. Productivity growth is generally considered to be the main driver of growth at the macroeconomic level. A number of studies have demonstrated that TFP growth is more important for income growth than other factors such as capital accumulation, and that TFP differences explain more of the variation in cross-country per capita GDP than variables like human capital, physical capital or trade.1 Productivity has also been used to gauge firm performance in the corporate finance literature,2 the management accounting literature,3 and the literature on corporate control.4 It is an important determinant of how firms react to business cycle fluctuations. In the framework of Imrohoroglu and Tüzel (2011), low TFP firms are more vulnerable to business cycles and hence are riskier than firms with high TFP. Low TFP firms have a higher implied cost of capital (ICC) and both the levels of ICC and the ICC spread between low and high TFP firms are countercyclical. Therefore, understanding the effects of leverage on productivity gains is relevant both from the perspective of capital structure theory and the macroeconomic perspective of long-term growth and business cycle fluctuations. Several papers find that productivity is positively related to firm value.5 Intuitively, productivity growth results in the efficient use of scarce inputs. This allows the firm to reduce its output prices while maintaining or increasing profit margins, and in the long-run, to survive. This enhances shareholders’ wealth. Therefore, our starting point to understand the link between leverage and productivity growth is the finance literature that relates leverage to firm value. In particular, our hypothesis is inspired by the trade-off theory of optimal capital structure, which explains firms' choice of leverage by a trade-off between the benefits and costs of debt. The second most influential theory of corporate leverage is the pecking order theory due to Myers (1984). We however focus on the trade-off theory rather than the pecking order theory for various reasons. From a policy perspective, it is important to identify firms or sectors where leverage may be excessive. Excessiveness must be defined with respect to an optimal capital structure. However, there is no optimal debt ratio in the pecking order theory, in the sense that firms do not aim at a particular target debt ratio. Instead, the theory suggests that observed leverage is the cumulative result of hierarchical financing decisions over time (Shyam-Sunder and Myers (1999)). In addition, the US empirical evidence is not very supportive of the pecking order theory.6 In a recent comprehensive review of the literature, Frank and Goyal (2009) conclude that the empirical evidence is rather consistent with some versions of the trade-off theory. This does not imply that the pecking order theory is irrelevant in our sample; but given our objective, we focus on the trade-off theory, which is simply the most natural starting point for a study of “excessive” indebtedness. The trade-off hypothesis goes back to Kraus and Litzenberger (1973), who weigh bankruptcy costs against the benefits of interest tax shields. The benefits of debt also include the mitigation of agency problems. In particular, debt has a disciplining role due to the associated reduction in free cash flow (Jensen, 1986). The costs of debt include debt overhang (Myers (1977)), risk-shifting (Jensen and Meckling 1976), bankruptcy costs (Warner (1977)), and asset fire sales (Schleifer and Vishny (1992)). Trade-off theory predicts that net benefits to debt financing rise for companies with low debt but decrease as leverage becomes high, implying that net benefits are a non-monotonic function of leverage. The empirical literature tests this hypothesis (against the competing pecking order theory) by typically running cross-sectional or panel regressions of leverage on various firm-level, industry-level and market characteristics that determine optimal leverage.7 While the literature has explored the relationship between leverage and firm value or performance,8 it has remained silent on the relationship between leverage and productivity. With both benefits and costs to leverage, we posit a hump-shaped relationship between leverage and productivity growth at the firm-level.9 At low levels of leverage, higher leverage is likely to be associated with higher TFP growth as the benefits to leverage outweigh the costs and debt is used to finance productive investment. As leverage increases, the costs of debt become larger and erode the net benefits to leverage. Highly-levered firms not only suffer from a debt overhang problem, which reduces their incentives to invest in productive investment, their attention is also diverted from productivity improvements by the need to generate cash flow in order to service their debts. While our hypothesis is clearly inspired by the trade-off theory, our analysis does not constitute a test of the theory. Trade-off theory relates the cost of capital of the firm (or its market value) to its market-value-based debt-to-equity ratio. In this paper, we relate TFP growth to the book value of leverage. The use of book values is dictated by the lack of market data for our sample countries where equity markets remain rather underdeveloped. Our sample consists of Central and Eastern European countries. Transition economies are an interesting sample for several reasons. First, the transition experience has long been described as a “natural experiment” (see for example, Eicher and Schreiber (2010)). While transition countries started the transition process from similar (though not identical) positions in terms of liberalisation, institutional reform has progressed in varying ways and to different degrees. Even after more than a decade of financial sector reforms, there is a growing feeling that the latter have failed to spur adequately the development of corporate financing opportunities. There is a striking proportion of firms in our sample with zero outstanding debt, including both short- and long-term debt. The “mystery of zero-leverage firms” (Strebulaev and Yang (2006)) is very pronounced in transition countries. Second, this puzzle is augmented by another one: among those firms with outstanding debt, many tend to have very high, potentially excessive leverage. Unlike much of the literature on developed countries, the literature on capital structure in developing and transition countries has highlighted the importance of excessive leverage (e.g. see Driffield and Pal (2010)). Many CEE countries have experienced rapid credit growth in recent years, in particular the Baltic States, Southern Eastern Europe and Ukraine. While the benefits of rapid credit growth have been recognized, the risks related to credit booms have been highlighted by the recent financial crisis, which has severely hit many CEE countries. Assessing the sustainability of firm-level credit growth and developing appropriate policy tools remains one of the priorities of policy makers and international organizations active in this region. In addition, the continued practice of soft budget constraints in this region may contribute to the negative impact of excessive leverage on TFP growth in our sample.10 Soft budget constraints (SBCs) imply that government or financial institutions are willing to provide additional financing to firms with negative NPV projects (see e.g. Dewatripont and Maskin (1995)). If firms take advantage of SBCs, borrowed funds may be used inefficiently rather than for productivity-enhancing investment. Research has for instance shown that one of the detrimental impacts of SBCs on the economy is a lack of R&D (Kornai (2001) and Brücker et al. (2005)). Our estimates confirm that TFP growth increases with leverage until the latter reaches a critical threshold beyond which leverage becomes “excessive” and lowers TFP growth. We confirm the robustness of this result by estimating an alternative threshold model using lagged leverage. All the results point to the existence of an optimal leverage ratio where the net benefits of debt in terms of productivity gains are exhausted. Our paper reaches some qualitatively similar conclusions to Korteweg (2010). Using a different methodology and a market-based assessment of the net benefits to leverage, the author finds that as leverage increases, net benefits to leverage first increase and then decrease, and finally turn negative for distressed firms. In addition, our analysis sheds light on how optimal leverage varies with firm characteristics, particularly profitability and size. Unlike existing studies that use traditional cross-sectional or panel regressions using observed leverage ratios, the threshold model allows us to determine optimal leverage despite firms' temporary deviations from the optimum.11 Again, we reach conclusions similar to those of Korteweg (2010). Since our conceptual starting point is the trade-off theory, we attempted to gauge its relevance in our sample. Data limitations permit only an indirect test of the traditional trade-off theory. We employ two indices of firm value based on earnings (return on assets defined as EBIT to total assets and return on equity defined as EBIT to book value of equity) and relate these to the book value of leverage using a threshold regression. While the exact values of the thresholds are somewhat higher, the qualitative results regarding the non-monotonic relationship described above hold when the dependent variable is a proxy for firm value rather than TFP growth (see further discussion in Section 3). Finally, our paper also contributes to the burgeoning macro literature on the finance-growth nexus. Best practice in the recent literature on finance and growth uses industry-level data to overcome endogeneity problems, typical of analyses that rely on aggregate data, and identify the channel through which finance affects growth. In their seminal contribution, Rajan and Zingales (1998) find that industries that are relatively more dependent on external finance grow disproportionately faster in countries with more developed financial markets. Our paper provides an alternative firm-level approach for studying the finance-growth nexus by directly linking firms' financial structure to TFP growth. In addition, our paper is related to the literature on the macroeconomic risks associated with lending booms. Kiyotaki and Moore (1997) show how increases in corporate leverage lead to higher costs of external financing due to a higher default probability. This could lower investment and therefore output. Kiyotaki and Moore (1997) and Bernanke et al. (1999) show how high indebtedness in the corporate sector can induce severe slowdowns by amplifying and propagating adverse shocks to the economy. Our analysis provides a tool to identify the point at which corporate sector indebtedness becomes a cause for concern. Indeed, whether a firm is below or above the threshold can be seen as a measure of “sustainability” of a firm's leverage. The recent financial crisis has highlighted the risks of excessive indebtedness. Policy makers need to be able to assess the sustainability of leverage, both in order to prevent similar crises in the future and to identify those firms or sectors of the economy that need to go through a deleveraging process following a crisis. The empirical literature on lending booms has generally focused on various aggregate measures of indebtedness such as various debt-to-GDP ratios (e.g. see Gourinchas et al. (2001)), or the growth rate of the domestic credit to GDP ratio as in the literature on banking and currency crises (e.g. see Kaminsky and Reinhart (1999)). Our paper extends this literature by looking at the sustainability of credit at the firm level. The remainder of the paper is structured as follows. In Section 2, we describe the dataset and the variables used in the empirical analysis. In Section 3, we discuss the empirical methodology and present our results. Section 4 summarizes our main results and concludes.

The paper aims to bridge the gap between the literature on optimal capital structure and the wider macroeconomic literature on finance-growth nexus. On the basis of the trade-off theory of capital structure, we posit a non-monotonic relationship between leverage and productivity growth at the firm level. TFP growth is not only the most important metric in the macroeconomic growth literature; it has attracted increasing interest from the finance literature. We provide evidence supporting our hypothesis of non-monotonicity, using a threshold regression model (Hansen, 2000). Estimates for a sample of Central and Eastern European countries confirm that TFP growth increases with book leverage until the latter reaches a critical threshold beyond which leverage becomes “excessive” and lowers TFP growth. This result points to the existence of an optimal leverage ratio where the net benefits of debt in terms of productivity gains are exhausted. Despite some variation depending on the sample and the measure of leverage, the estimates seem quite robust. The estimates of the slope coefficients for the three bands of leverage (low, intermediate and excessive) suggest that the productivity gains (costs) to leverage are substantial for underlevered (overlevered) firms. Leverage is found to have similar non-monotonic effects on return on assets and return on equity. Due to data limitations, our results are not a formal test of the trade-off theory. However, they suggest that the threshold regression approach is a promising methodology for the study of optimal capital structure and how the latter varies with firm characteristics. In contrast to existing empirical evidence based on observed leverage ratios, the threshold model allows us to endogenously determine optimal leverage despite firms' temporary deviations from the optimum. Our results suggest a positive (negative) relationship between profitability (size) and optimal leverage, unlike existing studies that use traditional cross-sectional or panel regressions. Using the leverage threshold estimates, we find evidence of a significant proportion of firms with debt ratios in excess of the upper debt threshold in our sample. Our results suggest that the proportion of firms with excessive leverage is higher among relatively less profitable firms in most sample countries, a result that extends beyond the transition sample, as shown by the study by Korteweg (2010) on the US. Finally, our results are robust to different specifications, such as instrumental variables, and to different sample periods.