Leverage dynamics: Do financial development and government leverage matter? Evidence from a major developing economy

See, for example, Booth et al. (2001), Bancel and Mittoo (2004), Demirguc-Kunt and Maksimovic (1999), Antoniou et al.(2008), De Jong et al. (2008), Psillaki and Daskalakis (2009), Hanousek and Shamshur (2011), Oztekin and Flannery (2012) and Fan et al. (2012).

See, for example, Aydın et al. (2006), Sayılgan et al. (2006), Yıldız et al. (2009), Okuyan and Taşçı (2010a), Okuyan and Taşçı (2010b), Köksal and Orman (2015), Güner (2016) and Karaşahin and Küçüksaraç (2016).

Another commonly used macroeconomic control variable, namely corporate tax rate, did not exhibit significant variation in Turkey during our sample period, especially after 2000. Besides, too many tax advantages as well as unprecedented tax amnesties are given to various sectors and these make measurement impossible. Hence, tax incentive is not incorporated into models as an economic environment factor.

Please see the CBRT’s web site for detailed information on the database including data collection process. (http://​www.​tcmb.​gov.​tr/​wps/​wcm/​connect/​tcmb+en/​tcmb+en/​main+menu/​statistics/​real+sector+stat​istics/​company+accounts​).

Soytaş and Küçükkaya (2011) construct a financial development index for Turkey for the period 1991 to 2005 by using Principal Component Analysis. Using the same methodology, we reconstructed their index for the period 1991 to 2015. For robustness, this reconstructed index is also used as an alternative measure of financial development in addition to the index created by Svirydzenka (2016). Since the results obtained by using this alternative index are in line with those in Table 5, they are not reported to conserve space but available from authors upon request.

For robustness, we re-estimate all alternative specifications of the model excluding industry median leverage. Results are in line with those reported in all tables that have industry median leverage as an explanatory variable. In addition to be a proxy for target leverage, industry median leverage is also argued to be a proxy for some omitted common industry factors (Flannery and Rangan, 2006; Frank and Goyal, 2008, 2009); hence, we report results of the model including industry median leverage in line with the capital structure literature. Alternatively, we also include industry x year fixed effects in the model in order to control any possible omitted industry factors (time-variant unobservable industry factors). However, time-variant variables (all macroeconomic and economic environment factors) are dropped from the model because of collinearity. Rest of the variables, namely all firm-specific variables, remain robust. To converse space, they are not reported in the paper but available upon request from authors.

For robustness, another classification scheme based on firm sales is also used. In this approach, firms are divided into quartiles by the value of their net sales, and a firm is classified as “large” if it is in the highest net sales quartile and as an “SME” otherwise. Since the results based on this classification scheme are qualitatively the same as those based on number of employees, they are not reported in the paper but available upon request from authors.

All the models for short-term and long-term leverages and different firm sizes based on net sales, and based on number of employees are re-estimated for the subperiod 2002–2015 as well. Results are in line with those for the subperiod 2003–2015. To conserve space, they are not reported in the study; however, they are available from authors upon request.

There is not a significant variation in financial development index during the first subperiod; thus, coefficient of this variable cannot be estimated.

To conserve space, these results are not reported in the paper. However, they are available from authors upon request.