The overall F-statistics of the meta-regression is 5.551. The corresponding p-value is near 0 which proves that the theoretical account is important. The theoretical account allows anticipation of the consequence size. To what extent, nevertheless, is the inquiry. The multiple R2 is the index for the strength of the relationship. A value of 0.3780 implies that all the independent variables have the joint power of 37.80 % of foretelling the consequence size. This stands good with the decisions of Lingwall and Tingwall ( 2010 ) in their meta-analysis of export-led growing in China and Martins and Yang ( 2009 ) in their meta-analysis of the learning-by-exporting hypothesis. Lingwall and Tingwall ( 2010 ) ‘s meta-regression yielded a maximal R2 of 0.18 throughout his 8 theoretical accounts. Similarly, Martins and Yang ( 2009 ) had changing coefficient of finding around the point 0.3. The magnitude of the R, hence, indicates a comparatively strong relationship between the consequence size and the forecaster variables.

After dropping two variables, out of the staying 11, 6 prove to be important with p-values of less than 0.05. Thereby, there is less than 5 % opportunity that they do non impact the consequence size. The t-statistics are above 2 which farther confirm significance.

The consequence size beginning silent person is statistically important to the relationship between exports enlargement and economic growing. The beta coefficient is used “ to compare the comparative influence of each forecaster variable ” ( Darlington, 1968 ) . The beginning silent person has a positive coefficient of 0.433. This denotes that usage of independent t-statistics influences the export-expansion hypothesis positively at a rate of 0.433. Other statistics at the case of p-values and chi-square do non impact the consequence size.

Equally far as the entire sample size is concerned, it is undistinguished in regard of the export-expansion hypothesis. The sample size is non a good forecaster of the consequence size. An account can be the little sum of fluctuation across the reported observations as indicated by the numeration analysis above. St. martins and Yang ( 2009 ) had the same consequences when they included works control in their meta-analysis arrested development. However, it is interesting to observe the negative mark of the beta coefficient of the entire sample size. It implies that as the sample size additions by 1 unit, the consequence size, i.e. the export enlargement consequence on economic growing, lessenings by 0.027 units. Besides, a big sample size involves long run surveies. St. martins and Yang ( 2009 ) , likewise, found that long-term surveies display lower relationship between exports and growing. This can be justified by the fact that reported surveies have inflated the sample size unsuccessfully.

Traveling to the publication type silent person, a p-value of 0.659 indicates low significance. Whether observations are derived from diaries, thesis or paper do non hold an impact on the consequence size. Furthermore, the publication coefficient is positive at 0.048. This is an indicant that there is no publication bias across literature analysing the export-led growing scheme. Observations taken from working documents tend to be higher by 0.048. Diaries and theses do non, merely, publish positive consequences. In fact, research workers are more inclined towards describing controversial results and inquiry the export-driven scheme. The empirical reappraisal of this survey highlights the considerable grounds against the fact that an addition in exports contribute to economic growing.

In the same line, the informations type silent person is undistinguished. Tingwall and Ljungwall ( 2010 ) in their meta-analysis particular to China reported a t-value of -1.170 for industry informations and -1.33 for aggregated information. Use of aggregative or sector informations, hence, is instead irrelevant to the consequence size of the survey. This can be accounted for by the observations which use chiefly aggregated information. Merely 6.2 % of the surveies use industry degree informations. The negative mark, nevertheless, is explained by the fact that at industry degree or sector wise, the export-growth link is observed marginally. Aggregated information, nevertheless, positively act upon the export-growth relationship.

The development degree silent person is based on the HDI degree of states under survey. As expected, the p-value is less than 0.05, denoting that the silent person variable is important and therefore, related to the consequence size. The t-statistic is supra 2 every bit good. The beta coefficient is -0.109. The state type is, therefore, negatively associated with export-led growing. If a state has low HDI, economic growing is decreased by 0.109 for every addition in 1 unit of exports. Critics have argued that the export-expansion hypothesis has non operated as predicted in most underdeveloped states. The good rippling effects in footings of technological invention, acquisition, and economic systems of graduated table amongst others have non materialised. There is no grounds of linkages between export and economic growing. Palley ( 2002 cited by Felipe 2003: p10 ) highlighted that the export-led scheme in fact hinder growing of domestic markets in less developed states. The latter besides suffer from direct competition with developed economic systems. Exports strongly depend on foreign demand. Hence, low developed states face fiscal uncertainness and instability. Recession in international markets further worsen the vulnerable conditions of those states. In the empirical reappraisal subdivision of this survey, accent has been laid on the critics of the export-expansion link. The consequences, hence, is in line with theory which question the part of expanded exports to a better economic system. Harmonizing to Vohra ( 2001 ) a minimal degree of industrialization is a requirement to profit from the export-expansion hypothesis.

Another forecaster variable of the consequence size across surveies of the export-expansion hypothesis is the mean clip for informations. The p-value of 0.008 proves that the clip variable is important. Tingwall and Ljunwall ( 2010 ) meta-analysed clip span as an independent variable. They reported a t-value of 2.35. In this survey, the arrested development coefficient of clip is 0.147. Thereby, as the mean twelvemonth for the sample size additions by one unit, the consequence on the export-expansion hypothesis is positive and of a magnitude of 0.147 unit. It can be deduced that inclusion of more recent informations sets in the sample lead to positive consequences. Another account would be that with the progress of globalization, exports-driven scheme is being employed in many economic systems and the positive results are ever-present.

Whether the value of exports included in surveies is absolute sums or a rate is of import to the finding of the consequence size is tested. A high p-value ascertains, nevertheless, that it is undistinguished in the context of this survey. The coefficient of the step of exports is 0.058 but with a negative way. Therefore, when observations include existent sums to mensurate exports, the silent person variable takes the value of 1. In this instance, the export-growth consequence decreases at a rate of 0.058 per unit.

The independent variable for the control of labor is confirmed to lend to the export-expansion hypothesis by the p-value of 0.006. Tingvall and Ljungwall ( 2010 ) provided grounds of a important labor silent person in two of his theoretical accounts of his meta-regression analysis. If labor is non accounted for in a survey theoretical account, the consequence that exports has on growing is inflated by 0.181 units. It can, therefore, be established that it is of import to include a control labor variable as it is a major determiner of economic growing. The usage of an drawn-out production attack with labor, capital, exports as maps of growing, to prove the export-expansion hypothesis is, hence, the most appropriate theoretical account.

The methodological analysis used in the assorted surveies to prove if exports do take to an addition in economic growing is important with a p-value of 0.036 and t-statistics of 2.075. St. martins and Yang ( 2009 ) included a non- OLS silent person equal to one if the observation is based on other econometrics methods than OLS or fixed effects. The method variable is, likewise, important. The beta coefficient is -0.247. This verifies that an OLS methodological analysis decreases the consequence of exports enlargement on economic growing by 0.247. Non-OLS methods, therefore, take to higher values of the export-growth relationship as per the consequences of the meta-regression.

In add-on to the above, the publication twelvemonth of each observation proves to be undistinguished. The arrested development coefficient, however, bears a positive mark. This implies that more recent publications study in favor of the export- enlargement hypothesis. Tingvall and Ljungwall ( 2010 ) likewise demonstrated an undistinguished period silent person. This can be attributed to the fact that though old ages have gone by since the first survey on the hypothesis, the theories behind this relationship and the theoretical accounts used do non change to a big extent. Arrested developments are still conducted of GDP or a alteration in GDP on exports. No fiscal technology has taken topographic point in order to measure the export-expansion hypothesis otherwise.

The last silent person variable is the relationship mark. This has been reported as the expression to calculate the consequence size automatically zeros down a negative t-statistics or f-statistics as the square of the statistics is used. With a p-value of 0.007 and a t-ratio of -2.691, the relationship mark is important to the consequence size. This is logical as a negative mark will needfully convey down the consequence that exports enlargement is likely to hold on economic growing. The magnitude is of 0.024 for an addition of 1 unit in exports.

The beginning of the consequence size, the mean clip for the information set, the control variable for labor, the methodological analysis used and the degree of development are the variables important to the anticipation of the dependant variable. Hence, they influence the results of surveies in literature in footings of whether an enlargement in exports will take to an addition or lessening in economic growing. The other variables: the sample size, publication and informations type, the step for exports and the publication twelvemonth are non related to the consequence size. These endogenous variables are undistinguished with high p-values and make non impact on the exports-growth link.

## 6.0 CONCLUSION AND IMPLICATIONS

This meta-analysis is a survey of the surveies on the export-expansion hypothesis. Outcomes from 82 surveies with 447 observations have been analysed to generalize the factors impacting the hypothesis and the repeating features of the research work in the field. A meta-analysis is the appropriate technique in this context as many surveies with different features are available. Research workers have concluded that the consequence of exports on economic growing is ill-defined and varied.

First, a qualitative analysis has been conducted. The numeration analysis has ascertained that the mean clip for the informations sets is the twelvemonth 1978 with a sample size of between 20 and 125. The surveies use largely aggregated information. Exports and growing are measured in rates and no control variables for labor and capital are normally included. NON-OLS methodological analysiss are the common manner to arrested development. The more frequent statistic reported is the t-value. More of import is the state under survey: low and medium HDI states are tested often for grounds of the export-expansion hypothesis. The result is positive in most cases and the consequence size tends to 1.064.

For a statistical analysis, a meta-regression is conducted. The method used, the control for labor, the mean clip of the information set, the relationship mark and the degree of development are the independent variables act uponing the export-expansion hypothesis across the 447 observations. Use of NON-OLS methodological analysiss, non-inclusion of a control variable for labor and usage of recent informations sets contribute positively to the export-expansion hypothesis. This implies that when these features are present in a survey, export enlargement is deemed to be lending to economic growing. However, it is noted that surveies of states with low HDI, ever yield consequences against the export-growth link. The latter does non keep valid in these economic systems. Other survey features such as the sample size, the publication type and twelvemonth, the informations type and the step for exports are undistinguished to the export-led growing scheme. These factors do non impact on the consequences of surveies across literature. For illustration, a Journal publication a survey on the export-growth relationship will non needfully be in favor of the latter. Similarly, usage of rates to mensurate exports and aggregated informations in the observations do non foretell the result of the survey.

The present survey has several deductions for export-expansion research. Low and medium degree HDI states have been widely analysed. Research workers should pitch their probes towards really high and high degree HDI economic systems to measure the cogency of the hypothesis. It is of import to foreground that the pick of the methodological analysis, clip span of the information set and the control variables have an of import impact on the result of the exports-growth relationship. The significance of the degree of development in the econometric theoretical account implies that less developed economic systems should upgrade their place state in footings of better substructure, equip themselves with engineering, accomplishments and know-how and better the local criterion of life before implementing export-expansion schemes. Along these lines, the state will accomplish some development which will guarantee the success of an enlargement in exports. The export-expansion hypothesis will, later, keep true.