1. Introduction
Regional financial dependence is the degree of contribution of transfer income to total regional revenue. Regional financial dependence shows how dependent a government is on the transfer revenue of the central and provincial governments Suwandi et al (2015), Nyoman et al (2015) examines the Influence of Financial Performance on Capital Expenditure Allocation and Economic Growth of Regency / City Government of Bali Province in 20062013. The results showed that the degree of decentralization and effectiveness of PAD has a positive effect on capital expenditure, while the regional financial dependence has positive effect on capital expenditure and capital expenditure allocation has positive effect on economic growth.
In the financial management should be done by the local government to demand regional independence in exploring local potential and improve its financial performance. The independence of this region is reflected by the regional capability of generating revenue earned from regions derived from regional economic potentials or socalled indigenous revenues [12].
The results of Sularso et al (2011) showed that financial performance in the form of degrees of decentralization, financial dependency, financial independence, effectiveness of PAD and direct contribution of BUMN significantly influence the allocation of capital expenditure while the allocation of direct capital expenditure significantly affects economic growth and financial performance the degree of decentralization, financial dependency, financial independence, the effectiveness of PAD and the contribution of BUMN indirectly have a significant effect on economic growth.
Local governments allocate their financial resources into capital expenditures to meet public needs for public facilities and infrastructure. Capital spending is used to improve product quality to reduce costs or improve production performance and increase customer satisfaction about products to promote revenue growth in the future So capital spending uses enormous resources and investment periods over several long years (Liao et, al., 2016). Local governments to managing their finances are sourced from local revenue and regional expenditure as a tool in produce regional of development rules. (Yuskov, et al 2015).
Ndede et al. (2016) states that simultaneously or collectively local revenues and special allocation funds can have a significant effect on capital expenditure on the government of Manado. According to Liao et al (2016) states that the nonfinancial performance in the product market to capital expenditure on enterprise companies in Taiwan in 20072012. The results of this study showed that there is a negative relationship between nonfinancial performance with capital expenditure.
Capital expenditure allocation is a budget expenditure used to acquire or add to fixed assets and other assets and may benefit more than one accounting period (Yunistin, et al 2016) Based on data sources LKPD that the development of capital expenditures for fiscal year 2014 and fiscal year 2015 increased 944,002.14 or 60.10% from 2014sebesar 1,570,679 to 2,514,682 in 2015. While the development of capital expenditure for fiscal year 2015 and fiscal year 2016 increased by 300,996.63 or 11.97% from 2,514,682 in 2015 to 2,815,678 in 2016. Thus the capital expenditure of 20142016 has decreased by 643,005.52 or about 48.13%. This can be seen from the difference in the percentage decreased by 60.10% in 20142015 to 11.97% in 20152016.
The problem of achieving budget realization in 20142016 did not increase because of the unoptimal governance of budget allocation of capital expenditure of regency / city government in Central Java. Therefore, to optimize it according to Regulation of the Minister of Finance number 101 / PMK02 / 2011 on budget classification influenced by several factors, among them the dependence of regional finance, the degree of decentralization and financing of SILPA towards the allocation of capital expenditure of regency / municipality of Central Java. According to Gerungan et al (2015) shows that fiscal decentralization positively affects the allocation of capital expenditure. However, different from the research conducted by Suwandi et al (2015) shows that the degree of fiscal decentralization negatively affects the allocation of capital expenditure.
Regional financial dependence is a balancing fund from the central government to the local government to be managed by the region in improving the welfare of its people, so the higher the dependency of the regions to the central government then the area is not said to be independent in the management of local finances. The degree of fiscal decentralization is the ability of regions to increase local revenues (PAD) to finance regional development. SILPA is the remaining budget of local government revenue in financing local expenditure [3]. Based on the above explanation, this study aims to determine the effect of regional financial dependence, fiscal decentralization, the level of SILPA financing to the allocation of capital expenditure, to the allocation of capital expenditure
2. Literature Review
Agency theory
Agency theory is the theory of a relationship that exists under a contractual agreement between two or more parties in which the first party is called the principal and the other is called the agent. Theprincipal is the party acting as the giver of command and is tasked with overseeing, providing assessment and input on the task has been run by the agent. The agent is the party who receives and undertakes the task in accordance with the will of the principal [12].
According to Yushkov et al (2015) is a theory of a relationship established under an agreement agreement between two or more parties where the first party is called the principal and the other is called the agent. Agency theory according to Jensen and Meckling et al (1976) states that agency relations is a contract between two parties, principals and agents, in which the principal authorizes the agent to make decisions on behalf of the principal.
Dependence on regional finance
Regional financial dependence is the degree of contribution of transfer income to total regional revenue. The regional financial dependence shows how dependent a local government is on the transfer revenue of the central and provincial governments [3]. The greatest of local financial dependence on central and provincial government revenue transfers, it can be said that the local or regional government has an increasing dependence on central government transfers. In general, the greatest contribution to transfer revenues lies in balancing funds such as general allocation funds, which are funds used for equal distribution of financial capacity [10].
The higher the regional financial dependence, it can be interpreted that the regional government has a greater dependence on the central and provincial governments. In general, the largest contribution to transfer income is in balancing funds such as general allocation funds, which are funds used for the equitable distribution of local financial capacity. The calculation of regional financial dependence is the ratio between total transfer revenue realization and total realized regional revenues [10].
Degree of fiscal decentralization
The degree of fiscal decentralization is the extent of the ability on a region in the independence of the region in accordance with the revenue it generates. If a local revenue is large then it will be great ability of local government in doing regional autonomy, and vice versa. The degree of fiscal decentralization is measured based on the amount of revenue of the original region with the amount of regional income (Simamora et, al 2011).
Research conducted by Nyoman et, al (2015), Sularso et, al (2011) stated that the degree of fiscal decentralization has a significant effect on the allocation of capital expenditure. However, the results of this study contradict the research conducted by Suwandi et, al (2011) which states that the degree of fiscal decentralization has no positive effect on the allocation of capital expenditure.
SILPA financing rate
According to Simamora (2014) SiLPA is the remaining funds obtained from the actualization of revenues and expenditure of local budgets during one period. There are two forms of SiLPA usage: (1) to continue unfinished activities in the previous year (slide) and (2) To finance new activities not budgeted in the pure budget.
Some studies have found that SILPA earned is largely contributed to operational expenditure versus capital expenditures. Dwiranda et, al (2015) finds that SILPA financing negatively affects capital expenditure allocation this is because most of the SILPA earned is donated to local government operational expenditures. This research is supported by previous research such as Hidayat et, al (2013), Novandi et, al. (2016), and Simamora et, al (2014) different from the above research shows that SILPA financing positively affects the allocation of capital expenditure.
Capital expenditure allocation
Capital expenditure is a local government expenditure that benefits more than one budget year and will add assets or wealth to the area which will add routine spending (Uhise, et al 2013). According to Yunistin, et al (2016) capital expenditure is the cost incurred by the local government which is used to acquire and supplement its local assets used within a period of more than one year of accounting period.
3. Research Methods
The population in this study is the financial statements of local government (LKPD) districts/cities throughout Central Java 20142016.sourced from the Supreme Audit Agency (BPK RI). The total population is 35 local government financial statements (LKPD) with details of 29 local government financial statements (LKPD) of District Government and 6 local government financial reports (LKPD) of city government.The sampling technique in this research is saturated sampling technique of sample determination if all members of the population are used as sample (Sugiyono, 2008). the sample in this study is 35 local government financial statements (LKPD) districts / cities in Central Java period 2014 2016.
Data analysis method
Multiple linear regression model
The regression model used in this study are as follows:
Information:
BM: Capital expenditure
KKD: Dependence on regional finance
DDF: The degree of fiscal decentralization
TPSILPA: The remaining financing rate is more budget financing
α,β
ε
4. Result and Discussion
Descriptive analysis result
The depiction of the data to be discussed in this chapter includes the minimum values, the maximum values, the mean values, and the standard deviations on the research variables. The following are presented descriptive statistics for five observation periods in this study:
Table 1
Based on the results of the data in Table 1 above shows the number of observations in the study (N) of 105 district / municipal governments in Central Java, the regional financial dependency variable has an average of 4.37%. The highest regional financial dependency value was 4.59%, while the lowest regional financial dependency value was 3.20%. The standard deviation value of regional financial dependence is 0.15%.
Fiscal decentralization degree variable has an average of 2.72%. the highest degree of fiscal decentralization of 3.59%, while the lowest regional financial dependence of 2.17%. The standard deviation of the fiscal decentralization degree is 0.30%. The variable rate of SILPA financing has an average of 2.66%. The highest SILPA financing rate was 4.58%, while the lowest SILPA financing rate was 0.69%. The standard deviation of SILPA financing rate is 0.58%. The capital expenditure variable has an average of 2.89%. The highest capital expenditure is 3.88%. While the lowest capital expenditure of 0.69. The standard value of capital expenditure deviation is 0.33%.
Hypothesis testing results
Test Data NormalityTo know the normality of data is to use kolmogorovsmirnov result significant value above 0.05 then research data assume normal distribution (Ghozali, 2011). From the test conducted then the data obtained are as follows:
Table 2


Variabel  N  Kolmogov Smirnov  Sig.  Std  Conclusion 
Standardized Residual  105  0,077  0,133  0,05  Normal distributed datarmal 
Based on the results of 2 table test data of KolmogorovSmirnov normality (KS), Asymp. Sig. (2tailed) in Table 2 shows a value of 0.133, which is greater than the 0.05 or 5 percent significance level. So it can be concluded that the normal distributed data and regression model can be used as the next test.
Multicolinearity test
The multicolonierity test was performed using tollerance values and variant inflation factor (VIF). If the VIF value is less than 10 and the tollernce value above 0.1, it is concluded that there are no symptoms of multicollinearity.
Table 3


Variabel Independen  Tolerance  Std  VIF  Std  Ket 
DDF  0,654  0,10  1,529  10  No Multikolinearitas 
KKD  0,633  0,10  1,580  10  No Multikolinearitas 
TPSILPA  0,960  0,10  1,042  10  No Multikolinearitas 
Based on the results of multicollinearity test in Table 3 note that the tolerance value of each independent variable greater than 0.10 and VIF value of each independent variable less than 10. It can be concluded that the test results indicate that there is no multicolonierity in the variables in the study.
Autocorrelation test
Autocorrelation testing is performed using a run test. The run test is part of the nonparametric statistics used to determine whether interresidual values have a high correlation. Test results can be seen in table 4 below:
Table 4


Information  Asymp. Sig. (2tailed)  Sig  Conclusion 
Autocorrelation Test  0,142  0,05  Autofree or worth using 
Based on the results of autocorrelation testing in Table 4 with Run Test method, Asymp. Sig. (2tailed) shows a value of 0.142, which is greater than the 0.05 or 5 percent significance level. This shows that there is no autocorrelation in the research data.
Heteroscedasticity test
One way to detect whether or not heteroskedasitas is by using a plot graph. If the variance from one observation to another observes remains, then this is called heterokedesitas. A good regression model is a regression model that does not occur heterokedesitas. The results of heteroscedasticity testing are as follows.
Based on the results of heteroskedastisitas test in Figure 1 above it can be concluded that there is no clear pattern, and the points spread above and below the number 0 on the Y axis. It can be concluded that there is no heteroskedasitas on the regression model.
Multiple linear regression test
Regression model
Regression analysis used in this study is multiple linear regression. The dependent variable in this study is the capital expenditure allocation while the independent variables are regional financial dependency, the degree of fiscal decentralization, and the financing of SILPA. Results of multiple regression analysis presented in the table.
Based on table 5 results of multiple linear regression test, then obtained the following equation:
BM it = 0,944+0,333DDF it + 0,303 KKD it  0,126TPSILPA it ε it
The equation can be explained as follows:
Constant value for capital expenditure variable equal to 0.944 this indicates that independent variable is considered constant, hence average allocation of capital expenditure equal to 0,944. The regression coefficient for the regional financial dependency variable is 0.333, it means that every increase of Rp 1 regional financial dependence will increase the allocation of capital expenditure by 0,333.The regression coefficient for the fiscal decentralization degree variable is 0.303, meaning that any increase of Rp 1 degree of fiscal decentralization will increase the allocation of capital expenditure by 0.303.The regression coefficient for the variable rate of SILPA financing is 0.126, meaning that each increase of Rp 1 of SILPA financing rate will decrease the capital expenditure allocation by 0.126.
Simultaneous test (F Test)
The result of statistical calculation of F test with result of significance value of F below 0,05 mean simultaneously all independent variable have significant effect to dependent variable.
Table 6


information 
F

F

Sig

Sig  Result 
Model Accuracy Test  2,931  2,69  0,037 

Model Eligible 
F value of table is obtained based on significant 0,05 with df1: (number of variables  1) and df2: (n k  1) so that the value of 2,69. The test results show that Fcount
Test results hypothesis 1
Based on the result of partial test (ttest) shows that the variable of financial dependency of area has titung equal to 1,292. It turns out that tcount is smaller than ttable (1,292
The result of hypothesis testing 2
Based on the result of partial test (ttest) shows that the variable of fiscal decentralization degree has titung 2,306. It turns out that tcount is bigger than ttable (2,306
The results of hypothesis testing 3
Based on the result of partial test (Testt) in table 4:15 show that SILPA variable have t count equal to 2,283. It turns out that tcount is bigger than ttable (2,283
Determination coefficient test (R2)
coefficient of determination is to measure how far the ability of the model in explaining the dependent variable seen from the value of the coefficient determinant determination (adjusted Rsquare).
Table 8


Model  R  R Square  Adjusted R Square  Conclusion 
1 
2,83

0,080  0,053  The independent variable can explain the dependent variable of 5.3% 
Based on the results of the test coefficient of determination in Table 9 shows that the value adjusted Rsquare 0.053 (5.3%). Based on the test it can be said that the research model is able to explain the dependent variable of 5.3% while the rest of 94.7% percent explained by other variables outside the research model that affect the allocation of capital expenditure of regency / city government in Central Java
Discussion
Based on the result of partial test (ttest) shows that the variable of financial dependency of area has titung equal to 1,292. It turns out that tcount is smaller than ttable (1,292
5. Conclusion
The results of this study show that hypothesis 1 rejected regional financial dependence has no significant effect on capital expenditure allocation. While hypothesis 2 is accepted, which means the degree of fiscal decentralization affects the allocation of expenditure. While the hypothesis 3 accepted the financing SILPA effect on capital expenditure.