Pingdingshan University, Pingdingshan 467000, China

Abstract After entering into WTO, export volume of agricultural products in China quickly increases, but its proportion in total trade volume becomes lower and lower, and there are more and more trade barriers of agricultural products. In this paper, based on the data during 1994-2016, error correction model is established to test and analyze the relationship between economic growth and export of agricultural products in China. The results show that change of agricultural products export in China has positive impact on GDP in short time, and they also have long-term stable relationship. When they lag for different periods, economic growth and export of agricultural products have unidirectional causality.

Key words Economic growth, Export of agricultural products, Error correction model

1 Introduction

After entering into WTO, with economic growth and gradual opening of domestic market, trade scale of agricultural products in China gradually enlarges. It’s worth noting that although trade volume of agricultural products export is gradually rising, its proportion in total export volume of China becomes lower and lower. Moreover, primary agricultural product has larger proportion, and its encountered trade barrier is more and more. In post financial crisis era, it has important significance to study how to promote export of agricultural products and further promote China’s economic growth. Based on the existing theoretic model, empirical analysis on the influence of agricultural products export on economic growth in China is conducted, and corresponding countermeasures and suggestions are proposed according to test result.

2 Empirical analysis

In this paper, error correction model among variables is established to test the relationship between export of agricultural products and economic growth, to avoid "pseudo regression phenomenon" of time series data. Eviews7.0 econometric software is used for test[1-4].

2.1DataselectionTime range of the sample is during 1994-2016, and illustration of these data is as below. (i)Economic growth is measured by GDP. (ii)Export volume of agricultural products is based on "classification of import and export goods" from theChinaStatisticalYearbook, and the data from the first class (live animals and animal products), the second class (plant products), the third class (animal and vegetable oils and fat, edible oil refined by its decomposing products, animal and plant wax), the fourth class (food; drink, wine and vinegar; tobacco, product of tobacco and tobacco substitute) are summed. (iii)To maintain consistency with dollar unit of agricultural products export data, dollar value of GDP during 1994-2016 is calculated according to average price of RMB and US dollar from theChinaStatisticalYearbookduring 1994-2016.(iv)Because that statistics of import and export trade is not perfect in China, and it lacks statistics of import and export trade indexes, the nominal value of agricultural products export could not be converted to the actual value. To maintain the consistency of calculation caliber, both export of agricultural products and GDP take nominal value. (v)In test and analysis, to eliminate heteroscedasticity of time series data, GDP and export volume of agricultural products respectively take natural logarithm.lnGDPandlnAErespectively show GDP and export volume of agricultural products, whileΔlnGDPandΔlnAErespectively show first order difference sequence of GDP and export volume of agricultural products.

2.2UnitroottestUnit root tests oflnGDPtandlnAEtare respectively conducted by ADF test method (Augmented Dickey-Fuller, proposed by Dickey and Fuller in 1979), namely stationarity test. After repeated test by adjusting lagging period number, original serieslnGDPtandlnAEtdo not pass through the test, illustrating that original serieslnGDPtandlnAEthave unit roots. Stationarity test of their first order difference sequencesΔlnGDPtandΔlnAEtis conducted, and results are shown as Table 1. According to test result, ADF test value ofΔlnGDPtis -2.938 470, which is smaller than critical value -3.297 799 at 10% of significance level. It accepts original hypothesis, and first order difference sequenceΔlnGDPtis stable, and original serieslnGDPtis first order single integer sequence, namelylnGDPt∽I(1). Similarly,lnAEt∽I(1). Therefore,lnGDPtandlnAEthave the condition of cointegration relationship. To establish relationship model betweenlnGDPtandlnAEt, it also needs testing their cointegration relationship.

2.3CointegrationtestSeen from above stationarity test, serieslnGDPtandlnAEtare both first order single integer sequences. Cointegration regression of serieslnGDPtandlnAEtis conducted, and below regression equation is obtained.

lnGDPt=1.445 380+1.547 390lnAEt

(5.102 344)(30.719 83)

R2=0.979 247s.e.= 0.144 793DW=0.587 965

Residual sequence of regression is marked aset. By testing stationarity of serieset, it could judge if serieslnGDPtandlnAEthave long-term stable relationship. Similarly, ADF is used to conduct stationarity test of the serieset. If seriesetis stable, stationary sequenceslnGDPtandlnAEtare CI(1，1), and test result is shown as Table 1. Test results show that ADF statistical value -4.481 272 of residual seriesetis smaller than critical value -2.771 129 at 10% of significant level. It is rejected the nihilism hypothesis that residual sequenceethas a unit root, that is to say,etis stationary sequence, marked as I(0). It indicates that serieslnGDPtandlnAEthave cointegration relationship, and there exists long-term stable relationship.

Table1StationaritytestoftimeseriesdataΔlnGDPtandΔlnAEtandresidualet

Variable Test type ADF valueCritical valueIf stableΔlnGDPtc, t, 1-2.938 470-2.650 413*StableΔlnAEtc, t, 1-3.182 075-2.650 413*Stableetc, t, 1-4.481 272-2.771 129*Stable

Note: In test form,cis constant item, andtis tendency item, and 1 shows 1-year lagging order number. Selection standard is that AIC and SC are the minimum, while*shows 10% of significant level.

2.4EstablishmentoferrorcorrectionmodelStationary sequenceslnGDPtandlnAEtare (1，1) order of cointegration, and they must have expression form of error correction model. Among them, error correction item isEMCt=ΔlnGDPt-1.547 390ΔlnAEt. The established error correction model is as below:

ΔlnGDPt=1.77E-13+1.547 390ΔlnAEt-EMCt

(4.226 792)(3.11E+14)(-4.77E+13)

R2=1.000 000DW=1.000 000s.e.=1.35E-14

The above results show that short-term change of agricultural products exportAEhas positive impact on GDP, and the coefficient of error correction item -1 is smaller than 0, which corresponds with reverse correction mechanism.

2.5GrangercausalitytestSeen from above analysis, economic growth GDP and export of agricultural products have long-term equilibrium relationship, but it needs testing if the two variables have causal relationship. Granger causality test method is used to test variableslnGDPtandlnAEt, and the results are shown as Table 2. Seen from test results, in 1-year lagging period and at 10% of significance level, economic growth of China and export of agricultural products have Granger unidirectional causality, that is to say, export of agricultural products is Granger cause of economic growth. In 4-year lagging period and at 5% of significance level, economic growth is Granger cause of agricultural products export.

Table2TestresultofcausalrelationshipbetweenlnGDPandlnAE

Original hypothesisNumber of lagging periodF valueP valueResultGranger cause for lnGDP not being lnAE12.589 920.124 9AcceptGranger cause for lnAE not being lnGDP13.773 380.067 9RejectGranger cause for llnGDP not being lnAE22.625 510.105 3AcceptGranger cause for lnAE not being lnGDP20.771 790.479 6AcceptGranger cause for llnGDP not being lnAE31.432 690.281 8AcceptGranger cause for lnAE not being lnGDP31.269 090.329 1AcceptGranger cause for lnGDP not being lnAE44.589 170.027 0RejectGranger cause for lnAE not being lnGDP41.468 520.289 6AcceptGranger cause for lnGDP not being lnAE51.499 020.315 8AcceptGranger cause for lnAE not being lnGDP50.681 340.654 7Accept

3 Conclusions and enlightenment

3.1BasicconclusionsSeen from cointegration test, economic growth and export of agricultural products have long-term cointegration relationship. When export volume of agricultural products changes by 1%, it will cause that economic growth changes by 1.547 390% in the same direction. Via error correction model, it displays that export volume of agricultural products significantly affects economic growth under long-term condition. Granger causality test shows that export of agricultural products and economic growth have unidirectional Granger cause.

3.2CountermeasuresandsuggestionsFirstly, it should concern the tendency that trade deficit of agricultural products in China continuously enlarges. When trade of agricultural products is normally implemented, it should establish and perfect corresponding trade rule, increase agricultural protection and enlarge export of agricultural products. Secondly, it should sufficiently play comparative advantage of agricultural products, optimize trade structure of agricultural products, accelerate to construct safety monitoring system of agricultural products, and enhance safety quality management of agricultural products, to enhance export competitiveness of agricultural products. Meanwhile, it should establish and perfect agricultural standardization system, improve technology content of agricultural products, increase added value of agricultural products, and break through trade barrier of agricultural products. Relevant system and policy of agricultural products trade should be perfected, and the role of trade association should be played, to actively respond to various trade frictions of agricultural products.