Conclusion and policy implication

TABLE OF CONTENT Content  Page   Introduction Methodology Econometric model Data description Results and test Results and analysis Results Analyze some basic content of results Detect and cure default model Normality Multicollinearity Heteroscedasticity Autocorrelation Detect and cure default new model Normality Multicollinearity Heteroscedasticity Autocorrelation 6.Conclusion and policy implication a. Conclusion b. Recommendation c. Policy implication APPENDIX REFERENCES  2 2 5 6 7 7 8 10 10 12 13 16 19 19 21 22 23 24 24 24 25 28 31   1. Introduction Issue: Try to establish an econometrics model to analyse the impacts and influences of Foreign Direct Investment (FDI) and urban unemployment ratio U on Gross Domestic Products (GDP). Reason for researching: Firstly, this is an issue relating to economics. All the knowledge we can gain from this researching will be helpful for other economics subjects such as Macroeconomics, International Economics….and our future jobs as well. Secondly, our country started to innovate in 1986; foreign investment law in Viet Nam was promulgated on 29th December, 1987 to make a legal basis for the investment in Viet Nam from foreign investors. The fact is that since Viet Nam opened to integrate, foreign investment has become a very important source of capital for Viet Nam economy in industrialization and modernization. Being a member of World Trade Organization (WTO), Viet Nam has many chances to gain more FDI. However, now the issue is that how to use FDI effectively, make FDI be an important factor to develop the economy. The study of the effects of foreign direct investment and unemployment on economic growth helps us to know the extent of the impact of FDI to GDP as well as U to GDP. According to learning the theories and features, understanding characteristics of this and trends to develop, we can make the directions and solutions to attract FDI and use FDI in the most effective way; besides, try to bring back unemployment ratio to nature unemployment standard in order to help GDP grow up. That is all the reasons why we choose to research this topic! 2. Methodology *Economic theories: Gross domestic product (GDP) is the market value of all final goods and services produced within a country in a given period of time. In the real world, the market values of many goods and services must be calculated to determine GDP. While the total output of GDP is important, the breakdown of this output into the large structures of the economy can often be just as important. In general, macroeconomists use a standard set of categories to breakdown an economy into its major constituent parts; in these instances, GDP is the sum of consumer spending, investment, government purchases, and net exports, as represented by the equation: Y = C + I + G + NX Because in this equation Y captures every segment of the national economy, Y represents both GDP and the national income. This because when money changes hands, it is expenditure for one party and income for the other, and Y, capturing all these values, thus represents the net of the entire economy. Four components of GDP: Consumer spending, C, is the sum of expenditures by households on durable goods, nondurable goods, and services. Examples include clothing, food, and health care. Investment, I, is the sum of expenditures on capital equipment, inventories, and structures. Examples include machinery, unsold products, and housing. Government spending, G, is the sum of expenditures by all government bodies on goods and services. Examples include naval ships and salaries to government employees. Net export, NX, equals the difference between spending on domestic goods by foreigners and spending on foreign goods by domestic residents. In other words, net export describes the difference between exports and imports. FDI is a form of international investment, in which the investors bring the means to invest abroad to directly organize the production process management and business profits. FDI plays a huge role in economic development: Add to domestic capital. Acquisition of technology and management know-how. Join the global production network. Increase the number of jobs and trained workers. Bring a large budget inflow. Unemployment is always a concern of society; long-term macroeconomic policies of the government are aiming to achieve the natural rate of unemployment in the economy. It reflects the prosperity of the country in each period of time. The some following simple analysis shows us that unemployment occupies an important position, is one of the objectives of government activities: High unemployment rate means that GDP is lower – human resource is not use effectively, we are wasting opportunities to produce more products and services. Unemployment also means less production, reducing the efficiency of production scale. Unemployment leads to social demand reduction. Moreover, goods and services are less consumed, business opportunities are smaller, quality and quantity of product reduces. Besides, high unemployment ratio can lead to the less consumers’ demand compared with when they are employed, as the result, the investment opportunities reduces. Relationship between gross domestic product GDP and foreign direct investment FDI: The relationship between the GDP and the level of FDI has always been a matter of discussion between economists. There is a widespread belief among policymakers that foreign direct investment (FDI) generates positive productivity effects for the host countries. The neoclassical growth model states that FDI cause an increase in investments and their efficiency leading to increases in growth. In the long-run, according to the endogenous growth model, FDI promote growth, which is considered a function of technological progress, originating from diffusion and spillover effects. The main mechanism for these externalities is the adoption of foreign technology, which can happen via licensing agreements, imitation, competition for resources, employee training, knowledge and export spillovers. These benefits, together with the direct capital financing it provides, suggest that FDI can play an important role in modernizing a national economy and promoting economic development. Relationship between gross domestic product GDP and utility U: GDP only measures production and consumption, not the level of utility people gain from producing and consuming. There is much economic activity (for example, replacing a low quality product, or repairing damage from war or natural disaster) that does not improve quality of life (compared to having a high quality product to begin with, or no war). The result can be a very high GDP combined with low customer satisfaction. *We collect the data and statistics of GDP, FDI and U to prove relations between GDP, FDI and U and by using regression model in econometrics. 3. Econometric model Model includes three variables: dependent variable: GDP (billion dong), independent variables: FDI( million USD) and U (%) GDPi= β1 + β2 FDIi +β3Ui + Vi This is multi regression model. Many economic models express the negative relation between inflation and unemployment (Phillip curve). Generally, high GDP leads to high inflation because of growth objectives of government. As the result, relation between GDP and unemployment is negative. 4. Data description - Data collected from website: www.gso.gov.vn, GDP, FDI and U in Vietnam from 1995-2009. - Correlated analysis between variables: During one year, if the total capital of foreign direct investment in Vietnam increases, there will be more capital for other projects. This will encourage produce more; therefore GDP increases accordingly. Unemployment rate increasing means GDP decreasing. - Table of data: see table in the appendix - Relation between variables: see graph in the appendix - Description:  Mean  Standard deviation  Minimum  Maximum   Median      GDP(billion dong)  697572.1      441975.8  228292  1658389   535760      FDI(million USD)  4198.913    3028.292  2334.9  11500   2714.0      U(%)  5.71  0.76  4.60  6.85   5.8800      5. Results and test A. Results and analysis 1. Results Model’s result from the gretl software ( Model-> Ordinary Least Squares ) Model 1: OLS, using observations 1995-2009 (T = 15) Dependent variable: GDP coefficient std. error t-ratio p-value --------------------------------------------------------------------------------------------- const 1.68744e+06 624740 2.701 0.0193 ** FDI 85.6018 23.6463 3.620 0.0035 *** U -236250 94698.9 -2.495 0.0282 ** Mean dependent var 697572.1 S.D. dependent var 441975.8 Sum squared resid 3.06e+11 S.E. of regression 159783.1 R-squared 0.887974 Adjusted R-squared 0.869303 F(2, 12) 47.55913 P-value(F) 1.98e-06 Log-likelihood -199.3341 Akaike criterion 404.6682 Schwarz criterion 406.7923 Hannan-Quinn 404.6455 rho 0.525136 Durbin-Watson 0.766908 2. Analyze the basic content of results. a. Population regression model: (PRM) GDPi =
1+
2 FDIi+
3 Ui+ Vi Sample regression model: (SRM)
=
+
FDI i+
3Ui +ei ( ei is estimator of Vi) (SRM) GDPi = 1.68744e+06 + 85.6018.FDIi – 236250.Ui + ei
= 1.68744e+06 means that if FDI=0 and U=0 then GDP = 1.68744e+06 billion dong (holding inflation rate, CPI equal to 0, population is constant)
= 85.6018 means that when FDI increases 1 million USD then GDP increases 85.6018 billion dong (holding other factors constant)
3 = – 236250 means that when U increases 1% then GDP decreases 236250 billion dong (holding other factors constant) b. Measure of fit + Intercept:
Test the hypothesis:

= 2.701 With
= 5% :
= 2.179 Reject Ho if:
>

Reject H0 ->

0 -> intercept is statistical significance + Slope: *
Test the hypothesis:

= 3.620 With
= 5% :
= 2.179 Reject Ho if:
>
3.620 > 2.179 => Reject H0 (
≠ 0 (
is statistical significance *
3 Test the hypothesis:

= -2.495 With
= 5% :
= 2.179 Reject Ho if:
>
2.495 > 2.179 => Reject H0 (
≠ 0 (
is statistical significance + Model R2= 0.887974 indicates that FDI and U explain about 88.7974 % for the variation of dependent variable GDP. Test the hypothesis:
(H0: the model is significant H1: the model is not significant)
= 47.5590 F0.05(2;12)= 3.89 Reject H0 if F > F0.05(2;12) 47.5590 > 3.89 => reject H0 ( R2> 0 ( model is significant B. Detect and cure default of model 1. Normality H0: error is normal distribution H1: error is non-normal distribution Use gretl software: Test ( Normality of residual Frequency distribution for uhat1, obs 1-15 number of bins = 5, mean = -3.88051e-010, sd = 159783 interval midpt frequency rel. cum. < -2.010e+005 -2.586e+005 3 20.00% 20.00% ******* -2.010e+005 - -8.598e+004 -1.435e+005 2 13.33% 33.33% **** -8.598e+004 - 2.908e+004 -2.845e+004 0 0.00% 33.33% 2.908e+004 - 1.441e+005 8.662e+004 9 60.00% 93.33% ********************* >= 1.441e+005 2.017e+005 1 6.67% 100.00% ** Test for null hypothesis of normal distribution: Chi-square(2) = 5.815 with p-value 0.05461
p-value = 0.05461 > 0.05 ( accept H0 => Error is normal distribution. 2. Multicollinearity H0: No multicollinearity in the model H1: Multicollinearity in the model Use gretl software Test (collinearity Variance Inflation Factors Minimum possible value = 1.0 Values > 10.0 may indicate a collinearity problem FDI 2.812 U 2.812 VIF(j) = 1/(1 - R(j)^2), where R(j) is the multiple correlation coefficient between variable j and the other independent variables Properties of matrix X'X: 1-norm = 3.9324782e+008 Determinant = 5.4825961e+009 Reciprocal condition number = 1.4456799e-010 VIF (FDI) = VIF (U) = 2.812 < 10 Accept H0 No multicollinearity in the model. 3. Heteroscedasticity H0: Var(ui)= σ2 for all i H1: Var(ui) = σ2i Use gretl software: + Tests ( heterokesdasticity ( white test White's test for heteroskedasticity OLS, using observations 1995-2009 (T = 15) Dependent variable: uhat^2 coefficient std. error t-ratio p-value ---------------------------------------------------------------------------- const -1.31163e+012 1.19540e+012 -1.097 0.3010 FDI 5.75038e+06 1.47086e+08 0.03910 0.9697 U 4.02859e+011 3.58620e+011 1.123 0.2904 sq_FDI -2697.74 2068.51 -1.304 0.2245 X2_X3 8.86852e+06 2.97815e+07 0.2978 0.7726 sq_U -3.37756e+010 2.61055e+010 -1.294 0.2279 Warning: data matrix close to singularity! Unadjusted R-squared = 0.278199 Test statistic: TR^2 = 4.172989, with p-value = P(Chi-square(5) > 4.172989) = 0.524788 n.R2= 15x0.278199 = 4.172989 χ2α (k-1) = χ20.05(5) = 11.07 reject H0 if n.R2 > χ20.05(5) 4.172989 < 11.07 => accept H0 + Test ( heteroskedasticity ( white test ( squares only) White's test for heteroskedasticity (squares only) OLS, using observations 1995-2009 (T = 15) Dependent variable: uhat^2 coefficient std. error t-ratio p-value ----------------------------------------------------------------------------- const -1.55410e+012 8.34385e+011 -1.863 0.0921 * FDI 4.84548e+07 3.11676e+07 1.555 0.1511 U 4.74663e+011 2.53071e+011 1.876 0.0902 * sq_FDI -2807.88 1940.23 -1.447 0.1785 sq_U -3.81973e+010 2.04696e+010 -1.866 0.0916 * Warning: data matrix close to singularity! Unadjusted R-squared = 0.271087 Test statistic: TR^2 = 4.066311, with p-value = P(Chi-square(4) > 4.066311) = 0.397106 n.R2 = 15x0.271087 = 4.066311 χ2α (k-1) = χ20.05(4)= 9.49 reject H0 if n.R2 > χ20.05(4) 4.066311 < 9.49 => accept H0 ( Var(ui) = σ2 for all i No hereroscedasticity in the model. 4. Autocorrelation *Hypothesis: H0: cov (ui;uj) = 0 H1: cov (ui;uj) ≠ 0 d =
= 0.766908 with n=15,
k=3
k'=3-1=2 Use Durbin-Watson statistics
= 0.946





0

d
4-d
4-d

Ta có d= 0.766908
Autocorrelation of order 1 - Use gretl: Test( autocorrelation (Lag of oder: 1 Breusch-Godfrey test for first-order autocorrelation OLS, using observations 1995-2009 (T = 15) Dependent variable: uhat coefficient std. error t-ratio p-value ------------------------------------------------------------- const -529735 599418 -0.8837 0.3957 FDI 20.9536 22.8623 0.9165 0.3791 U 78884.8 90594.5 0.8707 0.4025 uhat_1 0.653404 0.302686 2.159 0.0538 * Unadjusted R-squared = 0.297571 Test statistic: LMF = 4.659936, with p-value = P(F(1,11) > 4.65994) = 0.0538 Alternative statistic: TR^2 = 4.463558, with p-value = P(Chi-square(1) > 4.46356) = 0.0346 Ljung-Box Q' = 3.7777, with p-value = P(Chi-square(1) > 3.7777) = 0.0519 p-value=0.0346 <0.05 ( autocorrelation of order 1 *Cure: (AR1): Ut =
.Ut-1 + v
= 0.653404 Set: GDP* = GDP – 0.653404*GDP(-1) FDI* = FDI – 0.653404*FDI(-1) U* = U – 0.653404*U(-1) Then run the regression model: GDP* =
1 +
2GDP* +
3U* + v Use gretl: + Add ( lags of selected variables: 1 + Add ( Define new variables ( newGDP = GDP - 0.653404*GDP_1 ( newGDP = GDP* ) newFDI = FDI - 0.653404*FDI_1 (newFDI = FDI*) newU = U – 0.653404*U_1 (newU=U*) + Model( Ordinary Least Squares Model 5: OLS, using observations 1996-2009 (T = 14) Dependent variable: newGDP coefficient std. error t-ratio p-value ---------------------------------------------------------------------------- const 494695 211153 2.343 0.0390 ** newFDI 72.9059 21.6312 3.370 0.0062 *** newU -162241 96558.4 -1.680 0.1211 Mean dependent var 320349.9 S.D. dependent var 202377.8 Sum squared resid 1.42e+11 S.E. of regression 113681.5 R-squared 0.733005 Adjusted R-squared 0.684460 F(2, 11) 15.09963 P-value(F) 0.000701 Log-likelihood -181.1532 Akaike criterion 368.3064 Schwarz criterion 370.2235 Hannan-Quinn 368.1289 rho 0.153244 Durbin-Watson 1.213297 New regression model: (SRM): GDP* = 494695 + 72.9059 FDI*i - 162241U*i + ui + F = 15.09963 > F(2,11) = 3.98 ( model is statistical significance + β1: |t|= 2.343 > t0.0511=2.201 ( intercept is statistical significance + β2: |t|= 3.370 > t0.0511=2.201 ( slope β2 is statistical significance + β3: |t| = 1.680 < t0.0511=2.201 ( slope β3 is not statistical significance C. Detect and cure default of new model (SRM): GDP* = 494695 + 72.9059 FDI*i - 162241U*i + ui 1. Normality H0: error is normal distribution H1: error is non-normal distribution Frequency distribution for uhat5, obs 2-15 number of bins = 5, mean = 4.15769e-012, sd = 113681 interval midpt frequency rel. cum. < -8.790e+004 -1.385e+005 3 21.43% 21.43% ******* -8.790e+004 - 1.323e+004 -3.733e+004 6 42.86% 64.29% *************** 1.323e+004 - 1.144e+005 6.380e+004 4 28.57% 92.86% ********** 1.144e+005 - 2.155e+005 1.649e+005 0 0.00% 92.86% >= 2.155e+005 2.661e+005 1 7.14% 100.00% ** Test for null hypothesis of normal distribution: Chi-square(2) = 4.788 with p-value 0.09128
p-value = 0.09128 Accept H0 Error is normal distribution 2. Multicollinearity H0: No multicollinearity in the model H1: Multicollinearity in the model Use gretl software Test (collinearity Variance Inflation Factors Minimum possible value = 1.0 Values > 10.0 may indicate a collinearity problem newFDI 1.468 newU 1.468 VIF(j) = 1/(1 - R(j)^2), where R(j) is the multiple correlation coefficient between variable j and the other independent variables Properties of matrix X'X: 1-norm = 88260386 Determinant = 7.8663739e+008 Reciprocal condition number = 2.2696973e-009 VIF(FDI*) = VIF(U*) = 1.468 <10 Accept H0 No multicollinearity in the model 3. Heteroscedasticity H0: Var(ui)= σ2 for all i H1: Var(ui) = σ2i Use gretl software: + Tests ( heterokesdasticity ( white test White's test for heteroskedasticity OLS, using observations 1996-2009 (T = 14) Dependent variable: uhat^2 coefficient std. error t-ratio p-value -------------------------------------------------------------------------------- const -1.93698e+011 2.30951e+011 -0.8387 0.4260 newFDI 8.09341e+07 6.29831e+07 1.285 0.2347 newU 8.12109e+010 1.97847e+011 0.4105 0.6922 sq_newFDI -10103.1 3338.49 -3.026 0.0164 ** X2_X3 -3.34721e+06 3.51410e+07 -0.09525 0.9265 sq_newU -6.86930e+09 4.04854e+010 -0.1697 0.8695 Warning: data matrix close to singularity! Unadjusted R-squared = 0.569928