英语论文网

important, before going into further analyses. If the error term is autocorrelated and heteroscadastic, the regressions will be spurious and the results will be unacceptable. To estimate the long run relationships among the variables, the following three step procedure is inherited.

In the first step, the staionarity of the variables is checked. As Brooks (2008) state it “ the stationarity or otherewise of a series can strongly influence its behaviour and properties”. To check for stationarity, the existence of unit root is traced in null and alternative hypothesis. In presence of a unit root, on the level data or its first difference, than it can be stated that the series is stationary in its first difference or second difference. In order to do that, tests for order of integration will be run on the level data, and in their 1st difference. The tests, to check for unit root will be the Augmented Dickey-Fuler test (ADF), which is highly used in recent empirical literatures. The other two tests will be Phillips-Perron tests (PP) and the suggested test by Kwaitkowski et al. (1992), known as KPSS test. The criticisms of these tests are is on their power properties, and thus we will consider all the tests.

In the second step, which is the major part of the analysis, is that we check for co-integration among the variables in the system. The objective of the analysis is to look for the long run dynamic relationship among the varibles, and the direction of movement, as well as significance of the cointegrating relationships are checked. There exists two important estimation criteria suggested by Engle, R.F. and Granger,C.(1987) and by Johansen, S. and Juselius,K.(1990). These two estimation criteria are known Engle-Granger two step procedure based on the estimation of residuals and the Johansen's full information maximum likelihood method. The problem with the former one is that, the Engle-Granger approach is based completely on the residuals of the data, and therefore, may mislead the result. In addition, the Engle-Granger method fails to provide the rank of cointegrating vectors, and the cointegrating vectors. The Johansen's approach overcomes all of these and thus is considered a better estimation method for the analysis. I, therefore, will base my results on Johansen's full information maximum likelihood estimation procedure. This will give us, the rank of cointegration as well as the adjustment coefficients.

Finally, It is not only desirable to look for the long run realtionships among the vrible series, but also desirable to look for shocks in the series and the length of time, the shocks will persist, in the relationship. If suppose, the series of variables are in an equilibrium steady state level, than the deviations should be estimated. It is desirable to look for, when the shocks would converge on the equilibrium condition once again. For this purpose, the Vector error correction model or Equilibrium correction model will be estiamted. The testing of significance of the error correction term is very important for acceptability of the error correction model. The error correction term will give the speed of adjustment coefficient, which will further strengthen the long run relationships of the equation, with considering any further deviations form the ser本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。