4.6 Details of Methodology
In the study, the methodology is basing on Tanzi currency-demand method to link with the Chinese situation to use the appropriate currency-demand model, and the model is established by Aiting Xu in 2005. Basing on the model, cointegration method is used to get the equilibrium expression and confirm the long term equilibrium relationship of demand for currency and other economic variables. Before that, the unit root test should be made to analyse the stability of time series; in the other words, the test can check whether the data is random walk. Finally, the size of black economy of China of 1980-2005 is measured by the equilibrium expression.     

4.6.1 Unit root test
The unit root test that is proposed by David Dickey and Wayne Fuller (1979) is used to test whether the data is random walk so that it can evaluate the order of integration of each variable. Following the operating steps of Eview 5.0, the test is to determine whether the variables are null hypothesis of I(1) or an alternative hypothesis of I(0). According to Table 4.2, it shows that all of the variables have intercept, and C, TT and J have trend except intercept. Additionally, the appropriate lag length was chose, and then the result was obtained and exhibited in Table 4.2. It presents that the value of t-statistic of all of variables is less than the value of 10% level, and the probability of all of variables is more than 5%. Thus, the variables are all I(1), and they have the unit roots. It means that the data of the variable are random walk. The unit root test provides the good base to the cointegration tests.
     
Table 4.2: Unit Root Test (ADF Test)
　 Trend & Intercept Lag length t-statistc 10% Level Prob.
C T & I 4 1.92587 -3.261452 1.0000
P I 2 -1.22179 -http://www.51lunwen.com/Academicwriting/2009/0829/lw200908291708414787.html2.638752 0.6468
TT T & I 4 2.206132 -3.261452 1.0000
J T & I 4 1.511848 -3.261452 0.9999
R I 4 -1.274171 -2.646119 0.6214
π I 4 -1.83602 -2.646119 0.3539

4.6.2 Cointegration method
Due to the variables are all non-stationary i.e. I(1), the cointegration method can be used to obtain the equilibrium expression that measures the size of black economy. The multivariate cointegration test that supplies maximum likelihood estimates of the cointegrating vectors was developed by Johansen and Juselius (1990). There are two steps to complete the cointegration tests. Firstly, the form of each