A STATISTICAL ANALYSIS OF GDP AND FINAL CONSUMPTION USING SIMPLE LINEAR REGRESSION. THE CASE OF ROMANIA 990 200 Bălăcescu Aniela Lecturer PhD, Constantin Brancusi University of Targu Jiu, Faculty of Economics and Business Administration, Romania, anielabalacescu@yahoo.com Zaharia Marian Professor, Petroleum-Gas University of Ploiesti, Romania, marianzaharia53@gmail.com Abstract: This paper aims to examine the causal relationship between GDP and final consumption. The authors used linear regression model in which GDP is considered variable results, and final consumption variable factor. In drafting article we used Excel software application that is a modern computing and statistical data analysis. Keywords: GDP, final consumption, model, linear regression Cod JEL: C0, C20, C22. Cod REL: 0A 0C.. The simple linear regression model For expressing how a dependent variable is modified under the action of independent variables we use the simple linear regression: = β + () Yx i 0 βx i Due to the randomness of economic and social processes in the theoretical model we introduce a disturbance variableε, which summarizes the influence of variables unspecified in the model. ε has the following feature: M ( ε ) = 0, σ ( ε ) = ct. We have: Yx i = β 0 + βxi + ε (2) β The parameter 0 expresses the dependent variable Y when x = 0. From geometric point of view β0 where the regression right intersects the axis Oy. In terms of statistical and economic interpretation 0 will be closely examined with practical example. The parameter β is called "regression coefficient", given the key role in the regression analysis, with its help determining the influence of independent variables X on the variability of dependent variables Y. The regression coefficient shows the dependence of torque correlative variables, respectively as increases or decreases in average variable Y positive results from a unit change of the variable factor X. From geometric point of view, β is the regression slope, illustrating how much the change to alter the outcome Y with a unit X variable. If the two variables analyzed there is correlation, slope is equal to zero. β The sign of regression coefficient indicates the direction of the link between the two correlated variables, as follows: if β =0, then the variability Y variable does not depend on the influence of X variable, the two variables X and Y are independent; if β >0, then the connection between the two variables analyzed is direct, positive; if β <0, then the connection between the two variables analyzed is negative feedback. β is 26
In the regression equation is added the error term (ε ) because not all coordinate points are right on the line average variation. Random variable collects the influences of the variables that are not included in the model with on the variable Y. ε is estimated by the. (figure no.) y i Y (x i y i ) Regression line Error x i X Figure no. Regression line and error 2. Evolution of gross domestic product and final consumption of Romania in the period 990-200 Gross Domestic Product is one of the main macroeconomic aggregates of National Accounts and reflects specific gross final output of goods and services produced by factors of production that their activity within the country during a period. By cost method (the method for final use), GDP is obtained by adding the value of final goods and services produced in the interior during the period of analysis and consumed by the population, invested or exported. Thus: where: CP CPL CF FBC EXN PIB pp = CP + CPL + FBC + EXN, (3) CF = private consumption, which includes all goods and services purchased by private households and households consumed from own production (self-consumption); = public consumption (consumption state), which includes production state (the non-market services produced by government and private benefit of the community) that are eliminated from services sold and capital investments; = final consumption, viewed as a sum of private consumption and public consumption; = gross capital formation, which includes net investment and depreciation (i.e. gross investments), with the change in inventories from manufacturers; = net export, determined as the difference between the value of exported goods and services and the value of those imported. To identify the dependence of final consumption by a single factor (GDP) we will use unifactorial model with autocorrelation errors, so we will consider final consumption is independent variable and dependent variable is GDP, and the relationship between two variables is given by the relation: PIB = β + β CF + ε 0 (4) To calculate the unifactorial regression parameters propose using data published by NIS for the period 990-200 on the development of the two macroeconomic indicators of results. This information can be summarized in a table of the form: Evolution of gross domestic product and final consumption of Romania in the period 990-200 27
Table Year GDP Final consumption (million lei) (million lei) 990 85.8 68.0 99 220.4 67.3 992 602.9 464.3 993 2003.9 523.6 994 4977.3 3845.2 995 7648.9 6257.7 996 384.2 973.8 997 25529.8 2972.2 998 37055. 333.2 999 559.4 493.9 2000 80984.6 69587.4 200 7945.8 0073.7 2002 5207.0 278.8 2003 97427.6 6888.7 2004 247368.0 2054.6 2005 288954.6 25038. 2006 344650.6 294867.6 2007 46006.8 344937.0 2008 54700.0 42097.5 2009 498007.5 402246.0 200 53640.8 405422.4 Analyzing statistical data on the evolution of the gross domestic product (GDP) in the period 990-200, we can point out the following conclusions: a) G.D.P.: GDP throughout the period under review, has seen a steady growth from one year to another, making an exception to this rule in 2009, when it decreased by 3.20% over the previous year (figure no. 2); Figure no. 2 28
average value of this indicator for the period 990-200 is 67.447,8 lei, with a variation between a minimum 85.8 million (recorded the end of 990) and a maximum of 54.700 million lei (recorded the end of 2008); distribution of gross domestic product for the range of values considered is not a perfectly symmetrical (the value of the skewness test is different from zero, i.e. 0.86386) distribution is platykurtic (kurtosis < 3, -0.7476). It can be seen that in the data series considered, the values between minimum and average the series are much more numerous than those contained in the second half of the variation of the indicator subject of this research. b) Final consumption (public and private) throughout the period under review the final consumption increased steadily from year to year, making an exception to this rule throughout the year 2009, when it decreased by 4.44% over the previous year (figure no. 3); Figure no. 3 the average value of this indicator for the time period 990-200 is 39208.33 million lei, with a variation between a minim by 68.0 million lei (recorded at the end of the year 990) and a maximum by 42097.5 million lei (measured at the end of the year 2008) distribution of final consumption values for the range considered is not a perfectly symmetrical (the value of the skewness test is different from zero, i.e. 0.79004835), distribution is platykurtic (kurtosis < 3, i.e. -0.922073704). It can be seen that the evolution of the two macroeconomic indicators is similar, with sharp increases for the period 990-2008, and a decrease of 4-5% in recent years included the subject of research time. 3. Analysis of correlation between GDP and final consumption by means of simple linear regression In order to identify the existence and form the link between values of GDP and final consumption we construct the scatter plot diagram using Excel. 29
Figure no. 4. Scatter plot diagram The scatter plot diagram approximates a linear form, as the pairs of points gross domestic product final consumption almost perfectly describe the trajectory of a line, which allows us to affirm that the linear regression model can describe unifactorial very successful relationship between the two indicators analyzed. In determining the parameters of this regression model we used the Regression module of Excel software application, where we defined the equation which has dependent variable of GDP and final consumption value as an independent variable. Regression analysis tool performs linear regression analysis method using the method of least squares to find a line corresponds to a set of observations. By using the Excel application module data Regression, we obtain the following results: Regression Statistics Regression Statistics* Multiple R 0.9992293 R Square 0.9984598 Adjusted R Square 0.99837809 Standard Error 7600.37774 Observations 2 *Dependent variable: PIB ANOVA Table No. 2 Table no. 3 df SS MS F Significance F Regression 7.29E+ 7.29E+ 232.2877 3.47366E-28 Residual 9 097549093 5776574.75 Total 20 7.237E+ Correlation coefficients Table no. 4 Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Intercept -327.5 226.40253 -.382993533 0.8270525-7860.674956 605.665 X Variable.225324 0.0042928 0.9600323 3.47366E-28.20220563.248437 In the table no. 2 can be seen that all indicators of reliability of the model are close to which allows us to say that simple linear regression model is good, this conclusion can be made based on the values determined using Excel for an R-squared (0,9984) and Adjusted R-squared (0,9983). Table no. 3 shows the results of the variance of the dependent variable (GDP) under the influence of regression and residual factor. The values in columns F and Significance F provides important information underlying the regression model validation. The rule of decision on acceptance of the model is: high values for the F-test statistics and small values for Significance F. In our study, the value of the test F = 232.2877 and 30
Significance F = 3.47366E-28, so we accept that the linear relationship between two variables is considered significant. Table no. 4 presents the standardized coefficients of regression model estimates, standard errors of these statistics and t-test values for each factor (columns t State and P-value.). For a coefficient to be significantly different from zero, P-value column must have small values, for example 5% or under 5% (then obviously we have high values in t Stat column, in mode). In our study, for the theoretically term of model we have P-value = 0.82, i.e. we can say that if we reject the hypothesis that the intercept is equal to zero, we make an error of only 2%. So we reject this assertion and accept as true assumption that intercept is different from zero. The last two columns of Table no. 4 give us information about the 95% confidence intervals for each coefficient of the model: the constant term (theoretically) get the range model (-7860.674956, 605.665) and the slopes of the regression equation have confidence interval (.20220563,.248437). Significant amount of free time means that the factors were not included in the model shows a very high influence on the value of gross domestic product, and free term negative value indicates that the variables that were not included in the model have a negative effect on the evolution gross domestic product. However, we can appreciate that linear regression model describes the correlation between the value of gross domestic product and the value of final consumption and may be transcribed following form: PIB = -327.5+,22 CF (5) Analysis of correlation between GDP and final consumption (private consumption and public consumption) will result in an increase of.22 units of monetary value of gross domestic product. We can conclude that the Gross Domestic Product of our country is strongly influenced by the private and public consumption. 4. Bibliography. Anghelache, C., Economie teoretică şi aplicată, Volumul XVIII (20), No. 9(562), pp. 84-93 2. Oprea C., Zaharia M., Elemente de analiza datelor şi modelare utilizând Excel, Editura Universitară, Bucureşti, 20 3. Zaharia M., Bălăcescu A., Modelarea deciziei monetar financiare, Editura Universitaria, Craiova, 20 4. ***Anuarul Statistic al României, Institutul Naţional de Statistică 3