Forecast Audit towards 2016 Gross Domestic Product as Influence of Financial Growth and the ASEAN Economic Community Preparation Mutiara Shifa Economics Department, State University of Medan, Medan 20221, North Sumatra, Indonesia; Email: mumut9793@gmail.com Abstract As a country facing global challenges, it will be said as ready when it experiences domestic growth measured through gross domestic product (GDP). Welcoming the ASEAN economic community 2017, necessarily the role of growth will show the readiness of community as a whole. In this study, the author will predict the calculation of GDP based on the influences of the bank financing growth. In banking, GDP is an indicator of economic growth which is an important measure in explaining direct economic performance and a performance of economic actors providing goods and services including the banking industry. Economic growth can increase the cash flow of banks by increasing demand for financing from companies and households. During periods of strong economic growth, financing demand tends to increase. Because the financing that tends to obtain better profits. Keywords: Gross Domestic Product (GDP), forecasting, growth, financing, AEC. Introduction A readiness of a country to face free market and ASEAN economic community in 2017 especially in Indonesia, the growth of Gross Domestic Product (GDP). GDP is used to measure all goods and services produced by a country in certain period. The components of GDP are income, expenditure/investment, government expenditure and difference of export-import. Stiglitz and Walsh (2006) explain that GDP provides the best measurement to measure the level of production. However, the changes of nature of production from the growth in underground economy become a technological innovation in order to influence GDP ability to provide accurate illustration about economic performance. Furthermore, GDP illustrate the whole economic activity level in a country, that is, the number of goods and services produced by a market. It shows that GDP is an indicator of economic growth which is an important measurement in explaining direct economic performance providing goods and services including banking industry. As a form of indication of growth from community, then financial cycle in banking will be greater, and it shows the ability of community s purchasing power and welfare that become higher, and it can be stated as an indication of community readiness in ASEAN economic 2016. Economic growth can increase banking cash flow by increasing demand for financing from companies and households. During periods of strong economic growth, financing demand tends 109
Mutiara Shifa to increase. Because the financing that tends to obtain better profits than investment of securities, then expected cash flow will be higher. Other reason for the high cash flow is the less default risk level that happened during a period of strong economic growth (Madura, 2006). Literature Review Growth of GDP has some effects on the quality of loan given by bank. Furthermore, it is stated that if an economy has decreased in terms of negative GDP growth, then it will impact on lower quality of banking. This phenomenon can be seen in 2008 when Indonesia had experienced economic crisis and it influenced the decreasing activities of real sectors. GDP can also be predicted, therefore, predicting activity is needed for economic growth. According to Chang (2012:234) a predicting activity is a business function that tries to predict sales and the use of products so the products can be made in the proper quantity. Thus, prediction is a presumption of demands that will come based on some predicted variables; it is often based on historical time series data. Prediction is an art and science to predict future events. Prediction needs historical data collecting and to project it for future with some mathematical models. It may be subjective or intuitive prediction about the future. Or prediction can cover the combination of mathematical model adjusted by good assessment by the manager. Prediction is typically categorized by future time horizon that is underlying it. Three categories are beneficial for operational manager: a. Short-term Prediction, the range of time is reaching over one year but in general it is less than three months. Short-term prediction is used to plan the sale, work schedule, the number of employees, assignment and level of production. b. Mid-term Prediction, mid-term production typically measures from three months to three years. This prediction is beneficial to plan the sale, planning and budgeting product, budgeting cash, and to analyze various operational planning. c. Long-term Prediction, the range of time is typically three years or more; it is used in planning new products, capital expenditure, location of facility, or expansion and research and development. Research Method Gross Domestic Product is the total monetary value of all goods and services produced by an economic in a period. In this research the variable of GDP used was in the form of real GDP growth Quarter on Quarter. This variable is notated by notation GDP obtained by this formula: Analysis data method used in this research was Multiple Linear Regression analysis method. In doing multiple linear regression analysis, this method requires classic assumption test to obtain good regression result (Zharnowitz, 2013) based on the prediction of GDP that will be occur in 2017. In analyzing multiple linear regression model in order to obtain good estimator, that is the linear which is unbiased with minimum variance (best linear unbiased estimator = blue) is the complete basic regression assumption by doing a series of classic assumption. 110
Multicollinearity is caused by correlation between free variables, so in this case it is difficult to be found which variable that completes the bond variables. The effects of multicollinearity can be seen from VIF if < 10 it indicates that in the model there is no multicollinearity, but if VIF > 10, it indicates that in the model there is serious multicollinearity. Hesteroscedasticity shows the effects of interference variance cause uneven probability of independent variances. It is a way to detect whether there is heteroscedasticity according to Crushore (2011:9). To finish the problems as well as to prove whether the hypothesis is accepted or rejected in this research, multiple linear regression analysis tools are used. According to Rangkuti (2007: 23-25) the formula of multiple linear regressions as follows: Y = b o + b 1 X 1 + + b n X n + e Where: Y = Dependent variable X = Independent variable bo = Constanta b 1 -b n = Coefficient of regression e = Standard of error If the formula is used in this research, then the regression equation is: Y = b o + b 1 X 1 + b 2 X 2 + b 3 X 3 + b 4 X 4 + b 5 X 5 + e Y = GDP Prediction X 1, X 2, X 3, X 4, X 5 = Growth of GDP bo = Constanta b 1 b 4 = Coefficient of regression e = The limit of error To test the meaning of coefficient of regression simultaneously, statistical F test is used by formula (Rangkuti, 2007: 27) as follows: Where: F = Obtained from distribution table F R 2 = Multiple determination coefficient k = The number of independent variable n = The number of sample 2 R / k UjiF 2 (1 R ) /( n k 1) By the rules of decision making as follows: a. If F calculate > F table on confident level 95% (α = 0,05 ), then it is proved that GDP prediction can occur based on its growth. Thus, alternative hypothesis (H 1 ) is accepted and the null (Ho) is rejected. b. If F calculate < F table on confident level 95% (α =0,05 ), then it is proved that the five growths significantly does not influence of GDP prediction can occur based on the growth. Thus, alternative hypothesis (H 1 ) is rejected and the null (Ho) is accepted. With the decision-making as follows: a. If t calculate >t table on confident level 95% (α = 0,05 ), then it is proved that GDP growth can occur based on the significant growth, it influences GDP prediction in 2016. 111
Mutiara Shifa b. If t calculate <t table on confident level 95% (α = 0,05),then it is proved that the five GDP growths can occur based on the significant growth, it does not influence GDP prediction in 2016. The valid research instrument is an instrument that can measure the variables. Validity test is done to find out whether measurement tool can actually measure what needs to be measured. A measurement test will have small error variance or in other hand it does the measurement by giving the proper result with the purpose of the test. To find out whether the instrument is valid, then validity test is used by analytical point validation, with correlational technique of product moment. Masrun (Ogneva, 2012: 70) states that if the correlation of coefficient between score of an indicator and score of total indicators is positive and bigger than 0,3 (r 0,3) then the instrument is valid. Reliability test is done to valid statements, in order to find out how the measurement remains consistent if the measurement is done again the same effect. In this research internal consistency will be used to measure the reliability of measurement tools. There is a method of calculating the coefficient of reliability which is used in this research, it is Alpha Cronbach. After the coefficient value is obtained, and then the coefficient value of reliability is need to be determined. It is suggested that the coefficient of reliability is between 0,70 0,80 it is quite good for the purposes of basic research (Kaplan et al, 2013: 126). Whereas according to Malhotra (Ogneva, 2012: 71) that an instrument is reliable if it meets the standard of alpha Cronbach coefficient that is bigger than 0,6 (α > 0.6). Results and Discussion Ogneva (2012: 70) states that if coefficient of correlation between score of an indicator and score of total indicators is positive and bigger than 0,3 (r 0,3) then the instrument is valid. Whereas according to Konchitchki (2011: 1078) states that instrument can be considered valid if the obtained r calculate is bigger than r table. In this research, the reference used to determine the validity of each question refers to Ogneva s suggestion, it is expected that the higher limit value of validity requirement can make the questionnaire accurate in measuring respondents perceptions so that it also give the real conclusion. Scale Mean if Item Deleted Table 1. The result of Validity Test. Scale Corrected Squared Variance if Item-Total Multiple Item Deleted Correlation Correlation Cronbach's Alpha if Item Deleted X1.1 8,1700 2,062,672,455,735 X1.2 8,2700 2,200,674,456,738 X2.1 8,6700 2,910,640,461,612 X2.2 8,8500 2,553,648,464,598 X3.1 11,9100 3,376,602,396,714 X3.2 12,2900 4,006,410,191,805 X4.1 6,1900 1,307,384,149,639 X4.2 6,4700,918,491,255,506 X5.1 12,0200 5,353,843,715,869 X5.2 12,2000 4,970,853,733,865 X5.3 12,1000 6,111,775,631,898 X5.4 11,8900 5,109,756,585,904 Y1 12,0800 2,377,634,482,733 Y2 12,5100 2,818,476,272,807 112
From the data above, to collect data in this research questionnaire was used. Therefore, questionnaire, its question consistency had not been measured, then to measure the level of consistency, reliability test was needed. Malhotra (Ogneva, 2012: 71) states that an instrument is reliable if it meets the standard of alpha Cronbach coefficient that is bigger than 0,6 (α >0.6).Based on the reliability test of all research variables it was found that each variable had alpha Cronbach as follows: Table 2. The result of Reliability Test. Cronbach's Cronbach's Alpha Based on Alpha Standardized Items N of Items X1,813,816 3 X2,736,749 4 X3,778,776 4 X4,646,647 3 X5,911,915 4 Y,796,797 4 Referring to Malhotra s suggestion (Ogneva, 2012: 71) that an instrument is reliable if it meets the standard of alpha Cronbach coefficient that is bigger than 0,6 (α > 0.6) then it can be said that the numbers of obtained coefficient of reliability on table 5.2 had met the requirement of reliability, thus questionnaire used in this research was quite reliable in measuring respondents perceptions toward the observed variables. To find out the effects of multicol can be seen on VIF value if >10 it indicates than in the model there is serious multicollinearity. In this research the obtained VIF value was<10 so it can be said that there was no multicollinearty, thus the analytical regression could be continued. Table 3. Collinearity Statistics. No Variable Collinearity Statistics Tolerance VIF 1 (X 1 ),702 3,316 2 (X 2 ),951 1,051 3 (X 3 ),888 2,306 4 (X 4 ),974 2,671 5 (X 5 ),955 3,918 Heteroscedasticity can show the effects of interference of variance error which cause the inequity of variance probability for each observation of all independent variables. The effects of heteroscedasticity simply can be seen on scatterplot table where point distribution on that Table does not form systematic patter. Besides that, Rank Spearman correlation can also be used by correlating each independent variable with variance interference (residual). If sig. values (2-tailed test) < 0,05 then the correlation between independent variable and residual value is said to be significant or in other word there is an effect of heteroscedasticity. Conversely, if sig. value (2-tailed test) > 0,05then the correlation between independent variable and residual value is said to be not significant or the data is homocedasticity From the result of heteroscedasticity test by using Rank Spearman correlation, it shows that there was no independent variable which correlated significantly to residual value. Thus, he data used was homocedasticity. 113
Mutiara Shifa Table 4. The result of Heteroscedasticity Test. No Independent Correlation Variable 1. (X 1 ) -,037,714 2. (X 2 ) -,104,302 3. (X 3 ) -,031,760 4 (X 4 ) -,086,396 5 (X 5 ),003,976 To analyze the problems, multiple linear regressions were used. The summary of the result of analysis used SPSS and it can be seen on Table 5. Table 5. The result of Multiple Linear Regression. No Independent Variable Coefficient of Regression Sig. t 1 2 3 4 5 (X1) (X2) (X3) (X4) (X5),497,348,159,502,207,000,001,047,000,004 Constanta (Beta) =,849 Sig. F =0.000 Multiple-R =,940 α =0.05 Adjusted R Square =,878 Durbin-Watson = 1,875 From the result of multiple linear regression as seen on Table 5 above, it was then transformed into regression equation as follows: Y = 0,849 + 0,497X 1 + 0,348X 2 + 0,159X 3 + 0,502X 4 + 0,207X 5 + e. The equation shows the independent variables that were analyzed (X 1, X 2, X 3, X 4 and X 5 ) positively influenced the value of GD growth. Where +e shows the standard error. The large influence of independent variable as a whole was shown by the coefficient value of Adjusted R Square for 0,878. That value can be interpreted that the change of five independent growth had big influence for 87,8% toward the variation of the change of growth GDP loyalty value. Whereas the rest was 12,2%, it was influenced by other variables that were not included in the model (e). The obtained coefficient value (Multiple-R) was 0,940, it can be interpreted that the independent variable correlation (X 1, X 2, X 3, X 4 and X 5 ) toward dependent variable (Y) was 94%. The first hypothesis in this research is, growth GDP variables (X1-X5) simultaneously influence significantly to growth GDP prediction. To test if the first hypothesis is accepted or rejected, F test was used. From the result of the test Sig F 0,000 < α = 0.05, it means that five variables X1-X5 simultaneously influenced significantly to growth GDP prediction. Thus, the first hypothesis was accepted or the validity was proved, in other word it accepted hypothesis (H1) and rejected hypothesis (Ho). Therefore, the GDP prediction as follows: 114
Table 6. GDP Prediction. It can be seen that GDP prediction in naïve methods model would show the value of Bias or mean error was,33. While the MSE was 48,67 which was the Mean Square Error, MSE was the second way to measure the prediction error as a whole. MSE was the mean square of difference between predicted and observed values. The weakness of MSE is it tends to show large deviation due to squaring. MAPE value was,25 that was the mean Absolute Percentage Error, MAPE was calculated as mean of absolute differentiation between predicted and actual values, and it was stated as actual percentage value. Then, the next analysis was by looking at the analysis detail and error, it obtained the GDP forecast data for the next year by the data presented as follows: Table 7. GDP Prediction for the next year. GGDP (Y) Forecast Error Error Error^2 Pct Error X1.1 28 X1.2 28 28-10 10 100.56 X2.1 20 18 2 2 4.1 X2.2 29 20 9 9 81.31 X3.1 35 29 6 6 36.17 X3.2 23 35-12 12 144.52 X4.1 31 23 8 8 64.26 X4.2 28 31-3 3 9.11 X5.1 20 28-8 8 64.4 X5.2 27 20 7 7 49.26 X5.3 29 27 2 2 4.07 X5.4 34 29 5 5 25.15 X6 32 34-2 2 4.06 TOTALS 354 4 74 584 2.96 AVERAGE 27.23,33 6.17 48.67.25 Next Period 32 (Bias) (MAD) (MSE) (MAPE) Forecast Std Err 7.64 The result of forecast data for the next year in doing prediction of GDP growth was obtained, so the view between demand and prediction is presented in the graphic below: 115
Mutiara Shifa Figure 1. The view between demand and prediction. Conclusions The result of the research shows that GDP growth could be predicted based on the calculation of the growth occurred in previous year. From the result of the test it obtained that Sig F0,000< α = 0.05, it means that five growth aspects through financial which was observed simultaneously influenced significantly to GDP growth prediction. Some elements of GDP growth show the positive significance to growth in GDP prediction. Then, from the research, the author also got research limit that in measuring GDP as one of the indications of economic community readiness in ASEAN market 2016 could be done with details of increasing with macro economy. The limit is that this research only measured the financing in community that influenced GDP. It is recommended as a suggestion from the author that to create a measurement of GDP in showing community readiness it can be identified by some aspects, one of them is inflation, the increasing of rate of interest and some other aspects. When all is integrated, then it will result in accurate value from the prediction. References Bureau of Economic Analysis, 2014. Corporate profits in the GDP accounts. BEA paper series, No. 0040.United States Department of Commerce, Economics and Statistics Administration, Bureau of Economic Analysis. Chung, R., Kryzanowski, L., 2012. Accuracy of consensus expectations for top-down earnings per share forecasts for two S&P indexes. Applied Financial Economics 9 (3), 233 238. Croushore, D.D., 2011. Real-time forecasting. In:Higgins, MatthewL. (Ed.), Advances in Economic Forecasting. W.E.Up John Institute for Employment Research. Kalamazoo:MI, pp.7 24. Fischer, S., Merton, R.C., 1984. Macroeconomics and finance: the role of the stock market. NBER working paper series, No. 1291. NBER: Cambridge, MA. Hann, R.N., Ogneva, M., Sapriza, H., 2012.Forecasting the macroeconomy: analysts versus economists. NBER: Cambridge, MA. Kaplan, Robert M. and Saccuzo, Dennis. 2003. Psychological Testing, Principles, Applications, and Issues, Brool/Cole Publishing Company, a division of Wadsworth, Inc. Konchitchki, Y., 2011. Inflation and nominal financial reporting: implications for performance and stock prices. Journal of Accounting Review86 (3), 1045 1085. 116
Nasution, Mulia. 2008. Ekonomi Moneter, Uang dan Bank. Jakarta: Djambatan. Rangkuti, Freddy. 2007. Riset Pemasaran Cetakan Pertama. Jakarta: PT. Gramedia Pustaka Utama. Solimun. 2002. Multivariate Analysis Structural Equation Modelling (SEM), Lisrel dan Amos: Aplikasi di Manajemen, Ekonomi Pembangunan, Psikologi, Sosial, Kedokteran dan Agrokompleks. Malang: Pusat Dokumentasi Ilmiah Universitas Negeri Malang. White, H., 1980. A heteroskedasticity- consistent covariancematrix estimator and a directtest for heteroskedasticity. Econometrica 48(4), 817 838. Zarnowitz, V., Braun P., 2013. Twenty-two years of the NBER-ASA quarterly economic outlook surveys: aspects and comparisons of forecasting performance. Business Cycles, Indicators, and Forecasting, James Stock and Mark Watson, eds. Chicago: The University of Chicago Press. 117