Determinants of capital structure: evidence from a major developing economy

Fischer et al. (1989) and Hennessy and Whited (2005) present different formulations of what is known as “dynamic trade-off theory”, where the firm has a target level of leverage and deviations from that target are gradually removed over time. In this paper, trade-off theory refers to static trade-off theory, unless stated otherwise.

See, for example, Wiwattanakantang (1999) for Thailand, Pandey (2004) for Malaysia, Huang and Song (2006) for China, Correa et al. (2007) for Brazil, Qureshi (2009) for Pakistan, and Espinosa et al. (2012) for Chile.

Previous studies on the Turkish economy include, among others, Durukan (1997, 1998), Gönenç (2003), Acaravcı and Doğukanlı (2004), Aydın et al. (2006), Sayilgan et al. (2006), Korkmaz et al. (2007), Yıldız et al. (2009), Demirhan (2009), and Okuyan and Taşçı (2010).

For developed economies, support for the trade-off theory can be found, among others, in Antoniou et al. (2008) for for UK, US, Germany, France, and Japan, and Frank and Goyal (2009) for the US, whereas support for the pecking order theory can be found in Zoppa and McMahon (2002) for Australia. For developing economies, support for the trade-off theory can be found in Wiwattanakantang (1999) for Thailand, Huang and Song (2006) for China, and Espinosa et al. (2012) for Chile, whereas support for the pecking order theory can be found in Pandey (2004) for Malaysia, Korkmaz et al. (2007) for Turkey, Correa et al. (2007) for Brazil, and Qureshi (2009) for Pakistan.

There is a related but distinct strand of literature devoted to analyzing the determinants of corporate debt maturity. Barclay and Smith (1995) and Stohs and Mauer (1996) are notable early papers on the subject.

Another important risk is the exchange rate mismatch, which may occur due to a rise in foreign currency debt, while the firm’s income is in domestic currency.

We use total sales rather than total assets to alleviate the problem of multicollinearity since many of our variables are scaled by total assets, including those for debt ratios. These two measures are highly correlated, indicating that each of them should be a sound proxy for size.

Studies such as Miller (1977) and Rajan and Zingales (1995) recommend including both personal and corporate taxes in studies of capital structure. However, in Turkey, tax rates on equity and debt income at the personal level are extremely complicated and have gone through several reforms during the past two decades. This makes it almost impossible to come up with good indicators of personal tax rates on different sources of income that would also be consistent over time. As a result, we are forced to do away with personal taxes in our analyses.

This measure is also desirable because it takes into account the actual statutory tax rates and the fact that firms focus on the amount of income that can be shielded from tax using interest payments.

For confidentiality purposes, the SBS data are not available to the general public in its raw form. However, an aggregated version of the data is available on the Bank’s website at http://​www.​tcmb.​gov.​tr/​sektor/​mainmenu.​htm along with a report based on the data annually.

SBS firms report unconsolidated balance sheets. As noted by Rajan and Zingales (1995), this may cause firms to incorrectly appear to have lower leverage than otherwise identical firms who report consolidated balance sheets. This should not be a serious problem for our analyses since most of the sample firms are stand-alone enterprises. Still, some care is warranted in interpreting the results.

By micro-sized, small and medium-sized, and large-sized firms we mean firms with 1–9 employees, 10–249 employees, and >250 employees, respectively. According to this classification, there are on average about 1,350 (15 %) micro-sized firms, 6,390 (71 %) SMEs, and 1,260 (14 %) large firms in our sample. Classifications based on sales or assets would likely yield similar size distributions.

The ratio of the total number of workers employed by the SBS firms to non-financial sector employment in Turkey is likely to be considerably greater than 11 %. Unfortunately, the non-financial sector employment number is not published by the Turkish Statistical Institute. Moreover, although the lack of appropriate data makes the calculation impossible, the fact that almost all large firms and a large number of medium-sized firms are included in our dataset implies that our sample is likely considerably more representative of the Turkish non-financial sector in terms of sales; at least about 40–50 % in any given year.

The manufacturing industries are: (1) food products, beverages, and tobacco, (2) textiles and textile products, (3) leather and leather products, (4) wood and wood products, (5) pulp, paper, paper products, publishing, and printing, (6) chemicals, chemical products, and man-made fibers, (7) rubber and plastic products, (8) other non-metalic mineral products, (9) basic metals and fabricated metal products, (10) machinery and equipment N.E.C., (11) electrical and optical equipment, (12) transport equipment, and (13) furniture, manufacturing N.E.C. The non-manufacturing industries are: (1) construction, (2) wholesale and retail trade, (3) hotels and restaurants, and (4) transport, storage, and communications. On the other hand, we exclude some industries from our analyses. These are: (1) agriculture, hunting, and forestry, (2) fishing, (3) mining, (4) electricity, gas, and water supply, (5) real estate, renting, and business activities, (6) education, (7) health and social work, and (8) other community, social, and personal service activities. Unlike the industries included in our analyses, these industries are generally under the influence of various sorts of government intervention that distort the operation of market forces.

Possible bias introduced by firm entry or exit is discussed in Sect. 5.4.

Dell'Ariccia et al. (2008)among others, argue that capital inflows can ease financing constraints for productive investment projects, foster the diversification of investment risk, promote intertemporal trade, and contribute to the development of financial markets.

To economize on space, we do not report detailed results for our robustness analyses, except for the last one. However, these results are available upon request.

Dividing the sample period in this way is reasonable given the fact that the Turkish economy went through a dramatic transformation following the 2001 crisis. See, for example, Turhan (2008).

This is, to some extent, expected as each of the industry- and macro-level factors (unlike the firm-level factors) has only two and one observations per year, respectively, and splitting the sample into two further reduces the sample size, making precise estimation of their coefficients difficult.

A third possibility is linked directly to our measure of capital flows, which is an aggregation of equity and debt flows. If public firms take advantage of both types of flows, this might leave their leverage unchanged.

Our finding that pecking order theory is more powerful in explaining the financing behavior of young and small firms is consistent with notion that the pecking order theory may be more suitable for explaining the financing behavior of smaller firms since information costs are more important for SMEs than for large companies. See, for example, Zoppa and McMahon (2002) and the references therein.

Recall that our analysis of the economic significance of leverage determinants also lends more support to the trade-off theory than to the pecking order theory. There, we found that determinants that are more closely associated with the trade-off theory such as potential debt tax shields and firm size are economically more significant than those determinants that are more closely associated with the pecking order theory such as profitability.