_{1}

^{*}

The pecking order theory is one of the most crucial theories in corporate finance field. Empirical test on this theory is very common in western countries while domestic empirical study is still relatively limited in China. The majority of previous studies mainly concentrate in the whole capital market. Over the last decade, manufacturing sector experienced incrementally rapid development in China. The manufacturing sector not only makes critical contributions to China’s economy and society development, but also becomes key force in supporting the world economy. Thus, this article tests pecking order theory on Chinese publicly traded firms in manufacturing sector. Using a panel of 1212 observations during the period between 2013 and 2015, we draw the conclusion that there is no evidence to back up the pecking order theory. Further subsample analysis indicates that pecking order does not apply to Chinese manufacturing listed firms with both high float shares ratio and low float shares ratio.

For enterprises, the establishment, business operations and sustainable development all need support of a lot of capital. No matter what a huge difference exists in organization regardless of the asset size, operating conditions, and market share, enterprises desire to raise short-term funds to stabilize the business situation, and to raise long-term funds to seek future development. With the development of modern financial markets, the choices for financing channels are increasingly diverse, yet in general financing method still can be summarized in three main ways: the internal financing, debt financing and equity financing.

Western scholars have been studying enterprises’ financing preference for a long time. The pecking order hypothesis proposed by Myers and Majluf [

But in China, the application of pecking order theory has been challenged. During the period between 2000 and 2005, empirical studies conducted by Chinese scholars confirm that serious equity financing preference does exist in China’s listed companies (Huang and Zhang) [

Two events, “Split-share structure reform” and “the rise of the corporate bond market”, are not the object of our article. This article concerns that, after the split-share reform, whether financing preference of China’s listed companies varies or not?

In the past decade, manufacturing sector experienced incrementally rapid development in China. Both the overall scale and the comprehensive strength have been significantly increasing. Thus, it is of interest to study the financing behavior in Chinese manufacturing sector. Does pecking order theory apply to Chinese manufacturing listed firms?

Our study examines the financing preference of 404 Chinese listed firms with financing deficit in manufacturing industry. And then we test pecking order theory on subsamples classified by the percentage of floatable shares.

Myers and Majluf [

To test the applicability of the pecking order theory, a new model is introduced by Shyam-Sunder and Myers [

According to Shyam-Sunder and Myers [

Δ D i t = α + β D E F i t + ε i t (1)

where Δ D i t t denotes change in debt issues by firm i in year t. Since financial statements of Chinese firms do not have long-term debt, we define Δ D i t t as the change of short-term loan, long-term loan plus payable bonds. ε i t is the errr term. D E F i t is the flow of fund deficit defined as follows:

D E F t = D I V t + X t + Δ W t − C t

where

D I V t : cash payments for dividends, profits and interests in year t.

X t : capital expenditure is defined as the cash for purchasing fixed assets minus net cash received from disposal of fixed assets.

Δ W t : change in working capital, and working capital is the difference between current assets and current liabilities.

C t : cash flows generated by operating business after taxes in year t.

The current portion of long-term debt is also a component in financing deficit in original Shyam-Sunder and Myers [

In the case of ideal conditions, changes in corporate debt issues should usually track the financing deficit perfectly, thus, the pecking order estimate β should equals 1 and constant α approaches 0, statistically significantly. Shyam-Sunder and Myers [

The pecking order theory argues that financing behavior is affected by adverse selection costs to a large extent. Therefore, this theory should best apply to companies facing particularly serious adverse selection problems. Small-sized companies at growth stage are often regarded as companies with large information asymmetries.

Shyam-Sunder and Myers [

Δ D i t = α + β D E F i t + γ D E F i t 2 + ε i t (2)

In this regression, if pecking order theory holds, the estimate of the pecking order parameter (β) and R^{2} of the regression should see an obvious increase. Besides, estimating Equation (2) should lead to a statistically significant negative estimate of squared deficit coefficient.

It should be noticed that not every firm has annual financing needs. A part of DEF (financing deficit) in our sample is less than 0, suggesting that the firm doesn’t need extra financing in corresponding year. Instead, the firm generates cash surplus through operating activities. In order to improve the explanatory power, maintaining empirical conclusions unchanged, three treatment methods are used in data processing:

1) Reserve observations with negative DEF value. We refer to this as sample 1.

2) Follow Liu et al. [

3) Exclude observations with negative DEF value. Since our goal is to examine how the firms with financing needs choose funds, it’s better for enhancing the accuracy of the empirical results by excluding observations which have DEF less than 0. This is referred to as sample 3.

A balanced panel of cross-section observations from manufacturing listed companies between 2013 and 2015 is used in our study on pecking order. First of all, we do the unit root test on all variables to see whether they are stationary or not. If they are non-stationary, we cannot run the regression through OLS method directly. Otherwise, it will result in spurious regression. Unit root test on our variables indicate that all of them have no unit root. They are stationary so that we can use OLS directly. Then, for a panel data, in order to investigate the relation in dimensions from space and time aspect, the pooled OLS (ordinary least squares) regression is the simplest method. If the intercept is a constant in the cross-section observations, then OLS model gives us efficient and unbiased estimates for intercept α and the slope coefficient of β . By doing the redundant fixed effects test, we can know that whether significant individual effects exist in our sample. If our observations show no remarkable individual effects, we could pool them together and run the regression using the ordinary least squares method; if there is individual effects or time effects, Hausman test is needed to decide which model is most appropriate for estimating our regression, namely the fixed effects model or random effects model. In the process of running the regression, panel estimated generalized least squares (EGLS) approach is adopted. It gives the residuals of each observation different weights so as to overcome the heteroscedasticity problem. This estimation gives the error term of regression model homoscedasticity. And it also helps resolve the contemporaneous cross-equation error (residual) correlation problem.

A balanced panel of 1212 observations in manufacturing sector between 2013 and 2015 is collected from Shanghai Stock Exchange and Shenzhen Stock Exchange^{1}.

The pecking order theory hypothesizes that companies do not issue much shares after going public. In other words, equity issues should approach zero. If this is true, in empirical test, the number of financing deficit and of net debt issued must be the same in Equation (1). However, _{t}) accounts for the largest amount among deficit. And the amount of cash flow (C_{t}) driven from operating activities reaches 8.00E+08 on average, illustrating that the firms in our sample have strong profitability and thus have sufficient internal funds to cover the funding gap.

In terms of panel data, the pooled OLS is used when the explanatory variable is independent of the error term. The redundant fixed effects test illustrates that there is no individual effect in our sample so that pooled OLS model is adopted here. ^{2} being 0.09 approximately. After adjustment, making negative financing deficit zero, we get sample 2. As presented in column (2) of ^{2} is 0.13. This pecking order figure indicates that on average only 15% of debt is issued when firms need financing. In sample 3, which we exclude year-firms with negative financing deficit, we observe that pecking order coefficient approaches 0.19 and R^{2} of the re

Year | Average funds flow and financing as a fraction of total assets | ||
---|---|---|---|

2013 | 2014 | 2015 | |

number of firms | 1212 | 1212 | 1212 |

Dividend | 0.0251 | 0.0248 | 0.0240 |

Capital expenditure | 0.0425 | 0.0406 | 0.0344 |

∆Working capital | 0.0204 | 0.0116 | 0.0360 |

Internal cash flow | 0.0373 | 0.0439 | 0.0431 |

Financing deficit | 0.0506 | 0.0331 | 0.0501 |

Net debt | 0.0039 | −0.0006 | −0.0057 |

Net equity | 0.0490 | 0.0538 | 0.0446 |

Net external financing | 0.0529 | 0.0532 | 0.0389 |

Note: recourses from the author.

variables | mean | median | Std. Dev. | minimum | maximum | observations |
---|---|---|---|---|---|---|

X_{t} | 6.91E+08 | 1.71E+08 | 2.12E+09 | −1.52E+09 | 2.48E+10 | 1212 |

Div_{t} | 4.02E+08 | 1.31E+08 | 1.17E+09 | 48491.67 | 2.32E+08 | 1212 |

∆W_{t} | 17042153 | 28425444 | 2.34E+09 | −2.48E+10 | 3.43E+10 | 1212 |

C_{t} | 8.00E+08 | 1.88E+08 | 2.86E+09 | −9.72E+09 | 4.44E+10 | 1212 |

DEF_{t} | 3.10E+08 | 91106959 | 3.59E+09 | −3.54E+10 | 4.90E+10 | 1212 |

DEF^{2}_{t} | 7.88E+18 | 1.28E+17 | 8.20E+19 | 1.06E+10 | 2.40E+21 | 1212 |

∆D_{t} | 1.14E+08 | 3101925 | 1.75E+09 | −1.75E+10 | 1.50E+10 | 1212 |

Note: aE+n is the scientific notation, representing that a × 10n. Recourses from the author.

variable correlation coefficient | |||||||
---|---|---|---|---|---|---|---|

DEF_{t} | D E F t 2 | ∆D_{t} | Div_{t} | ∆W_{t} | C_{t} | X_{t} | |

DEF_{t} | 1 ----- | ||||||

D E F t 2 | 0.347028*** (0.0000) | 1----- | |||||

∆D_{i} | 0.258943*** (0.0000) | 0.073180** (0.0108) | 1----- | ||||

Div_{t} | −0.082908*** (0.0039) | 0.264070*** (0.0000) | 0.101266*** (0.0004) | 1----- | |||

∆W_{t} | 0.626276*** (0.0000) | 0.382163*** (0.0000) | 0.001528*** (0.9576) | −0.239616*** (0.0000) | 1----- | ||

C_{t} | −0.344287*** (0.0000) | 0.346707*** (0.0000) | −0.050527* (0.0787) | 0.719605*** (0.0000) | −0.069993** (0.0418) | 1 ----- | |

X_{t} | 0.207763*** (0.0000) | 0.358020*** (0.0000) | 0.215305*** (0.0000) | 0.57338*** (0.0000) | −0.23992** (0.0000) | 0.576197*** (0.0000) | 1---- |

Note: * demonstrates significance at the 0.1 level. ** demonstrates significance at the 0.05 level. *** demonstrates significance at the 0.01 level. P-values are shown in parentheses. Recourses from the author.

Δ D i t = α + β D E F i t + є i t | |||
---|---|---|---|

sample 1 (1) | sample 2 (2) | sample 3 (3) | |

β | 0.109376*** (0.0000) | 0.145617*** (0.0000) | 0.189042*** (0.0000) |

N | 1212 | 1212 | 1212 |

R^{2} | 0.087667 | 0.132549 | 0.321912 |

F-test | 58.03887*** | 92.29329*** | 159.0365*** |

Note: * demonstrates significance at the 0.1 level. ** demonstrates significance at the 0.05 level. *** demonstrates significance at the 0.01 level. P-values are shown in parentheses. Recourses from the author.

gression increases to 0.32, indicating that 19% of financing deficit is funded by debt financing. It is obvious that the regression has stronger explanatory power after excluding observations with negative financing deficit since our aim is to examine firms financing choices when extra funds are needed. All pecking order coefficients estimated from Equation (1) in three different samples are much smaller than 1, not consistent with the expectation behind Equation (1). Pecking order theory accepts some stocks issues, to a broad extent, so an estimate of pecking order coefficient being a little lower than 1 is acceptable. But this is still not supported in our sample due to the extremely low β estimate and R^{2} (

In order to test the pecking order theory on our manufacturing listed companies more particularly, we break down the data into three subgroups according to the free float methodology. Due to the fact that firms in our sample have relatively high free float ratio, we break down the overall sample into group 1 (float of shares no larger than 60%), group 2 (float of shares between 60% - 80%), and group 3 (float of shares between 80% - 100%). We do both the redundant fixed effects test and Hausman test to choose the most appropriate regression model. And as mentioned above, panel EGLS is used in our regression. In terms of processing data, we follow the previous methods to treat the observations with negative financing deficit.

Analyzing our data, a panel of 99 companies that have float of shares below 60% are obtained. We do not exclude observations that have negative financing deficit when Estimating Equation (1) and Equation (2) on group 1, because the

Δ D i t = α + β D E F i t + γ D E F i t 2 + ε i t | |||
---|---|---|---|

sample 1 (1) | sample 2 (2) | sample 3 (3) | |

β | 0.106247*** (0.0000) | 0.245484*** (0.0000) | 0.311585*** (0.0000) |

γ | 1.55E−12 (0.2817) | −1.19E−12* (0.7089) | −4.22E−12* (0.0938) |

N | 1212 | 1212 | 1212 |

R^{2} | 0.085075 | 0.234589 | 0.695547 |

F-test | 28.03525*** | 92.40618*** | 434.0704**** |

Note: * demonstrates significance at the 0.1 level. ** demonstrates significance at the 0.05 level. *** demonstrates significance at the 0.01 level. aE−n is the scientific notation, representing that a × 10^{−n}. Recourses from the author.

number of observations available in individual group is not very large. The redundant fixed effects test indicates that pooled OLS should be used on group 1. As seen in ^{2} of both samples is very low, which might is resulted from the small sample sizes. So we need to estimate Equation (2) to see more precise results and then make the conclusion.

Then we follow the Lemmon and Zender (2010) model to do more precise test on group 1. First of all, redundant fixed effects test suggests pooled OLS model. As column (1) of ^{2} increases from 0.10 to 0.54. These features are still unable to prove the pecking order since the coefficient of financing deficit is still much smaller than 1. As for sample 2, column (2) of ^{2} variable stays at −7.94E−11,

Δ D i t = α + β D E F i t + ε i t | ||
---|---|---|

sample 1 (1) | sample 2 (2) | |

β | 0.114128**** (0.0070) | 0.149526** (0.0123) |

N | 99 | 99 |

R^{2} | 0.102778 | 0.131678 |

F-test | 7.331285*** | 9.093747** |

Note: * demonstrates significance at the 0.1 level. ** demonstrates significance at the 0.05 level. *** demonstrates significance at the 0.01 level. Recourses from the author.

Δ D i t = α + β D E F i t + γ D E F i t 2 + ε i t | ||
---|---|---|

sample 1 (1) | sample 2 (2) | |

β | 0.101091*** (0.0000) | 0.226921*** (0.0003) |

γ | −9.75E−12*** (0.0001) | −7.94E−11*** (0.0007) |

N | 99 | 99 |

R^{2} | 0.543225 | 0.37053 |

F-test | 34.46177** | 21.68204*** |

Note: * demonstrates significance at the 0.1 level. ** demonstrates significance at the 0.05 level. *** demonstrates significance at the 0.01 level. aE−n is the scientific notation, representing that a × 10^{−n}. Recourses from the author.

suggesting that firms turn to equity issues when the deficits exceeds debt capacity constraint point. And the R^{2} of this regression is 0.37. After running regression on Equation (1) and Equation (2) and analyzing the estimates, it can be concluded that, in manufacturing sector, Chinese listed firms with circulation stocks lower than 60% do not follow pecking order when internal funds are exhausted. And this finding is consistent with the condition of firms that have relatively less tradable shares. At the beginning, the establishment of china’s stock market aims to help state-owned enterprises raise funds. Thus when firms go public, the corporate stocks are artificially divided into floatable stocks and non-floatable stocks, in which non-floatable stocks occupy the absolutely dominant position. Non-tradable stocks are often issued at par, while the outstanding shares are issued at premium. Thus, the holding costs of non-tradable shareholders are much lower than the cost of shareholders who hold outstanding shares. In this case, refinancing will lead non-tradable shares to obtain equity appreciation undoubtedly. Therefore, the board controlled by non-tradable shareholders will not easily decide to give up equity financing when the firm needs extra financing. The absolute control by major shareholders causes the minority of shareholders unable to contend with the resolution of the board. These facts might give an explanation to our empirical results on group 1.

Findings for firms that have 60% - 80% tradable shares show that pecking order theory does not hold in past three years. Redundant fixed effects test and Hausman test demonstrate that fixed effects model should be used here. It can be seen that the estimate of β coefficient is −0.15 (Column (1) in ^{2}. Obviously, these results show that manufacturing firms with 60% - 80% outstanding stocks prefer equity financing instead of debt financing when there is funding deficit. Therefore, in this case, pecking order theory is not applicable to firms in group 2.

And the results obtained from estimating Equation (2) is consistent with it. The significantly negative slope coefficients in column (1) and (2) of ^{2}. These firms have floatable shares between 60% - 80%, yet there is close interest nepotism between the big shareholders and controlling shareholders in most Chinese listed companies, including the manufacturing sector. These minority big shareholders might be the co-founder of the company when it was founded or equity investors introduced by controlling shareholders in order to meet listing requirements. Apart from this, big shareholders might be the family group of controlling shareholders. The presence of these minority big shareholders cannot effectively supervise the controlling shareholders, but will intensify the control by the controlling shareholders.

Results for firms with floatable shares between 80% - 100% indicate that pecking order theory is not applicable for these firms. Based on redundant fixed effects

Δ D i t = α + β D E F i t + ε i t | ||
---|---|---|

sample 1 (1) | sample 2 (2) | |

β | −0.150282*** (0.0000) | −0.174906*** (0.0000) |

N | 153 | 153 |

R^{2} | 0.69766 | 0.719182 |

F-test | 4.514739*** | 5.010690*** |

Note: * demonstrates significance at the 0.1 level. ** demonstrates significance at the 0.05 level. *** demonstrates significance at the 0.01 level. Recourses from the author.

Δ D i t = α + β D E F i t + γ D E F i t 2 + ε i t | ||
---|---|---|

sample 1 (1) | sample 2 (2) | |

β | −0.125220**** (0.0014) | −0.182271*** (0.0000) |

γ | −1.82E−12 (0.4748) | −4.19E−13 (0.8751) |

N | 153 | 153 |

R^{2} | 0.675767 | 0.743556 |

F-test | 3.821031*** | 5.316015*** |

Note: * demonstrates significance at the 0.1 level. ** demonstrates significance at the 0.05 level. *** demonstrates significance at the 0.01 level. aE−n is the scientific notation, representing that a × 10^{−n}. Recourses from the author.

test, pooled OLS model is appropriate when estimating both Equation (1) and Equation (2). According to the column (1) of

Δ D i t = α + β D E F i t + ε i t | |||
---|---|---|---|

sample 1 (1) | sample 2 (2) | sample 3 (3) | |

β | 0.219667*** (0.0000) | 0.313093*** (0.0000) | 0.353369*** (0.0000) |

N | 960 | 960 | 584 |

R^{2} | 0.204101 | 0.31342 | 0.720028 |

F-test | 245.6711*** | 437.3210*** | 1496.777*** |

Note: * demonstrates significance at the 0.1 level. ** demonstrates significance at the 0.05 level. *** demonstrates significance at the 0.01 level. Recourses from the author.

Δ D i t = α + β D E F i t + γ D E F i t 2 + ε i t | |||
---|---|---|---|

sample 1 (1) | sample 2 (2) | sample 3 (3) | |

β | 0.243513**** (0.0000) | 0.214494*** (0.0000) | 0.286690*** (0.0000) |

γ | 9.54E−12*** (0.0000) | 1.28E−11*** (0.0000) | 8.32E−11*** (0.0000) |

N | 960 | 960 | 584 |

R^{2} | 0.372035 | 0.28792 | 0.84293 |

F-test | 283.4850*** | 193.4752*** | 1558.995*** |

^{−n}. Recourses from the author.

constraint, firms will turn to equity issuing and reduce debt issues. The significantly positive DEF^{2} coefficient illustrates that firms with debt capacity constraints still choose issue debt when the funding deficit exceeds constraints point. For sample 2, R^{2} decreases from 32% to 29%. And a decrease of slope deficit coefficient is also observed, which does not match the logic behind Lemmon and Z ender regression. Financing shortage should be more possible to be filled by debt issues so an increase in pecking order coefficient should be observed after adding squared deficit as extra regressor. The γ estimate is significantly positive, indicating that these publicly traded companies still use debt financing when deficit exceeds debt constraint point. Excluding the time series observations with financing deficit lower than zero, we obtain 584 observations left in sample 3. Column (3) of

Modern corporate finance theory originated from MM theorem [

Our article aims to examine the pecking order hypothesis on Chinese publicly traded companies in manufacturing sector during the period from 2013 to 2015. By using Shyam-Sunder and Myers [

The author declares no conflicts of interest regarding the publication of this paper.

Yuan, Y. (2018) Does the Pecking Order Theory Apply to Chinese Publicly Traded Companies? Evidence from Manufacturing Sector. Modern Economy, 9, 2233-2247. https://doi.org/10.4236/me.2018.912138

^{1}http://www.sse.com.cn/, http://www.szse.cn/.