Module code and title: Dissertation-6econ012C-n

Discussion of results
Using the corr function in SATA program, the dataset had first been tested to see if there was any relationship between variables. At appendix 1.2.1 shows that the matrix of correlation between the variables, which could be a useful tool for demonstrating the presence of multi - collinearity with in results. That according econometric concepts, correlation variables control the dangerous dependency of factors among 0.1 and 0.9. The significance of the independent variable varies between 0.3 and 0.7, which corresponds to a standard correlation coefficient. To avoid illustrating a multi - collinearity problem in the dataset, these rules state that the magnitude of association components should never be greater over 0.5. There has been valid evidence in the research that there were no multi-collinearity concerns in the full sheet as from covariance value. Previously, the total operation displayed basic analyses of values, such as their average, maximum, and standard deviation in appendix (1.2.2). On conclusion, the mean rate of poverty in the eight nations studied is 26.99 percent, with minimum and maximum rates of 4.07 until 64.3 percent in Sri Lanka and Pakistan, correspondingly. Furthermore, in Nepal and Afghanistan, household debt by financial industries averages 35 percent and varies between 1.8 and 88.07. Moreover, the median rate of urban growing population is 2 percent more than the rate of regional population expansion. The minimum and maximum values vary from -0.7 in Nepal to 4.5 in Afghanistan and from 0.04 in Sri-Lanka to 6.79 in the Maldives accordingly. Furthermore, while the mean rate of interest is 5 percent, it fluctuates greatly depending on the highest and lowest rates (17.58 and -27.41 respectively). And per the Gini coeffitsient, the median wage distribution among eight nations is 35 percent, compared to a standard of 4.84 percent, with a lowest at 26.43 percent in Afghanistan and a highest in 51.89 percent in Sri Lanka. Another crucial measure is the mean GDP growth rate, which is 5.82 percent. In Maldives, it varies by 4.05 percent between -13.12 to 26.11. Nevertheless, the estimated net household savings level is 20 percent, it swings more than another measures Standard Deviation 14.38. As a result, minimum max values were compared within a wide range [ in Bhutan 44.64 and in Afghanistan -29.51]. Lastly, the vif function was introduced to the data to evaluate for auto-correlation between factors in appendix (1.2.3). This outcome indicates that the average VIF is greater than 1.0, while the majority of the 1/VIF parameters have smaller values. As a result, it could be stated that parameters really has not auto-correlated. In the next step, a few of the models mentioned in above literature study, such as Random, Fixed and OLS effect models, where analyzed in selecting the most appropriate theory for the period intervals and collection of data in countries of Asia.
Initially, although no missing data were found, the OLS model was performed in STATA program, and large levels of robust and regular standard errors even among the data turned unimportant in appendix (1.4). Furthermore, the data that was chosen had not been normally distributed, also respect to OLS model, information ought to be regularly divided up in order to make accurate inferences. To evaluate these hypotheses, use rvflot, hist, sktest and imtest for distributions, homoscedasticity, and normal testing in appendix (1.3). The 2-way diagram and 2-way distributing competences of the graph are available over all analysis plotting command. A line zero (0) is disclosed while drawing a line all along graph for y = 0, but it could not be made in comparison to the data shown inside the remnants in a depth model - it does not relate to the theory. When some of the extremes at the bottom and top of the graph are removed, the remaining shaped curves, indicating that perhaps the hypothesis that poor is constant for independent factors has been broken, as well as the volatility or reduction in the residual may be seen due to homoscedasticity. A failure of the deepest quadratic parameters can take any shape. As a result, it is reasonable to conclude that the model of OLS ought to be discarded. 
The fixed and random function hypotheses were then examined, and the Housman theory was used to pick either about them as the major hypothesis in appendix (1.5 ,1.6). Torres O. emphasized that the Hausman theory resulted in Probability>chi2 (0.0000), indicating that Hausman test suggests using a panel data model instead of a random effect model, however this would not rule out the random effect (2011). The Hausman test defines the relationship between instructional factors and error correlations. While collecting and analyzing from a panel, the Hausman theory is used to distinguish between an explanatory variables and a random effects model. Study of Gujarati shows further discusses how this analysis is exclusive in regard of recognizing endogenous variables in the generated model (2009). The model was ineffective during the learning process due to the lack of endogenous variables in standard regression method of OLS. The explanatory parameters that provide the model of random and residuals effects have no discernible connection, as per Hausman theory for Н_о hypothesis. For Н_а hypothesis from the another side emphasizes that standard errors and variables are strongly connected in support of a coefficient of determination. Although some negative results were reported in the Fixed effect model, its level of significance, standard errors and R-square were still satisfactory. However, according to theory of Hausman's, the random model effect produces far more intended results than that of the fixed model effect. As a result, the Random model effect may be chosen as the best fitting model for this short study, and the fixed model effect can be evaluated using Hausman theory to illustrate a clear contrast in between two.
In STATA program, the xtreg, y_ х and хtrеg, y_ х, rе, fе functions were used to regress dependent and independent and variables, with Random model effect (RME) and Fixed model effect (FME) regression findings obtained in appendix (1.5.1, 1.5.2). For both methods, panel data analysis fallouts show there are 152 observed samples by identifying 8 nation groups over the previous 20 years. Although there was fluctuation "within" and "among" factors in RME R-square, there was still a 50 percent confidence that the system was operating in terms of the selected variables, nation group, and duration. In contrast, FME's total R-square was 17 percent, but "among" variable fluctuation is nearly 50 percent, which is appropriate because FME varies "inside" variables. Corr (i u, x) shows whether or not changes between units were connected with the regression coefficients, which is "zero" in RME and 0.26 in FME. Basic on econometric concept, if Probability > chi2 and this value is less than 0.05, our different models are acceptable, and the F value is used to determine that all of the variables in the equation are different from zero. Because both theories are satisfactory of several identical characteristics, yet predicted research outcomes were acquired as from Random model effect, this analysis refers to the RFE model. While they incorporate order and between-entity impacts, correlations are challenging to comprehend.
Continuing on to the regression findings, the association among urban citizen's increase and poverty remained considerable of P-value in RME for the first time that lower than 0.05. The coefficient already has a positive result, indicating that even if the population of an urban grows by 1 percent, the rate of poverty would rise by 2.61 percent, but the P-value and significant coefficients in FME indicate inconsequential results. Interestingly, yearly regional citizen's increase has negative P values and insignificant coefficient values, implying that populations have a greater effect on poverty than people of village, implying that urban populations benefit more from educated people than rural citizens. Nevertheless, coefficients of slope were insignificant values, and the analysis was applied, real interest rate and Gross Domestic Product, both indicate P values that are negative in both cases. This suggests that, according on data obtained over last twenty years, there was no correlation between economic growth real interest rate and GDP in the education progress. From another point of view, the fact that residents in village regions were restricted with funds owing to increasing interest rates was highlighted. Nevertheless, all above factors had become unimportant, and it is thought that there was a combined influence in the entire model. An index of Gini was a represent of income distribution between population that ranges from zero (0) to one (1), or zero (0) percent to hundred (100) percent, absolute equity and complete inequity, respectively in economic principles. The index of Gini also exhibits statistically significant results in RFE it his approach. Furthermore, the value of P is 0.023, which would be smaller than the 0.05 confidence level that was set. from another side the coefficient value indicates an opposite value, implying that if the index of Gini rises by 1 % point (raising the index of Gini indicates increasing inequality, which normally moves in negative directions), level of poverty might fall by 0.49, which is irrational. However, the Gini coefficient is created in conjunction with some other connected causal variables, it is reasonable to presume that they will have a combined effect inside the model. The total contribution is significantly higher than that of the other factors. Consequently, in value of P and coefficients of slope, the last explaining variables, gross domestic savings became statically important and satisfactory. Furthermore, a 1%-point increase in gross domestic savings results in a 0.31-percentage-point reduction in poverty level. Lowering the rate of interest on expenditure or tuition fee for education and providing a better saving rates for the population will improve the middle classes status.
Conclusion and Final chapter
This research examined the empirical evidence on the effect of education on reducing poverty in middle Asian nations in order to determine if there was a link between distributing education system to the poor and improving overall poverty situation. The results of the analysis for regression show that a well education program could aid in the reduction of poverty level, but also that the progress of expansion volume does not occur throughout all orders of magnitude simultaneously. In another terms, the education program has an independent influence on each model estimation, with some becoming statically important and others becoming unimportant, however the total integrated impact may be recognized as a highly efficient way all across the model. Domestic borrowing to the poverty, a growth in assets, rising populations, and a decrease in the index of Gini all contributed to the reduction of poverty level according to the findings. Nevertheless, in aspects of chosen regression models, period intermission, and nation selection, increase of GDP, rises of real interest rate, and citizens of village growth produce negative effects; even so, we assume that while these factors are making an appearance of any economy's total decided to merge impact, they must have an impact on the poor society. In conclusion, from the viewpoint of central Asian nations, this research emphasis on the function of education program in modernizing towns and rural areas. In almost the same way, it demonstrates that education has the ability to significantly aid in the introduction of new economic plans. This also played a major role in creating new work prospects for the poor society, especially some with a low level of education. As a result, education creates opportunities for low income borrowers to play a vital factor in the economic progress of local families and the development of countries. The success of well education program is proposed by MFI in the establishment of substance courses on productivity of education to meet the goal for reduction of poor society, that have blown out their position as an element of development through low income country to developed country, according to the final inquiry. During the development of this study, various constraints were discovered, causing the research project to reach some weaker inferences. First of all, most of the other collected data from the World Bank and IMF data bases were incomplete information for several of the years, which is due to the fact that some of the statistics are released semiannually or once every three to five year periods. We were forced to continue with secondary sets of data due to the global COVID-19 epidemic, which limited our ability to create questioners or find other ways to obtain new primary information. As a result, we must rely on questionable sources for several secondary sets of data.
Furthermore, among some Asian countries many have been disagreements and wars by conflicts for many years (Syria, Afghanistan, Iraq), and portion of the people not formally documented (Afghanistan Bangladesh, Afghanistan, Pakistan and India). Nevertheless, while these datasets are technically available across the globe, they nonetheless have significant limitations due to these governments' internal conflicts.
It is suggested that future research use more nation organizations with much more explanation variables linked to education and poor society indices. The aim of test and create good, stronger and obvious inferences, far more complex tools are also strongly advised.