MONETARY AND REGULATORY POLICIES

Impact of monetary and regulatory policies on loans



Impact of monetary and regulatory policies on loans

The regression model

Dependent Variable: LNLOANS

Method: Least Squares

Date: 09/04/12 Time: 20:18

Sample: 2000Q1 2010Q4

Included observations: 44

Variable

Coefficient

Std. Error

t-Statistic

Prob.  

C

6.170011

0.321512

19.19064

0.0000

LNRATES

0.502397

0.047715

10.52921

0.0000

LNRESV

-0.296644

0.131872

-2.249488

0.0302

LNCAP

0.544757

0.023281

23.39969

0.0000

LNNPL

0.066204

0.023883

2.771955

0.0085

R-squared

0.987731

    Mean dependent var

14.43216

Adjusted R-squared

0.986472

    S.D. dependent var

0.625468

S.E. of regression

0.072747

    Akaike info criterion

-2.297003

Sum squared resid

0.206395

    Schwarz criterion

-2.094254

Log likelihood

55.53406

    Hannan-Quinn criter.

-2.221814

F-statistic

784.9154

    Durbin-Watson stat

1.305039

Prob(F-statistic)

0.000000

The results of the regression model are very much significant with a high and significant f - statistic of the model because its significance value is less than 0.05. The t - values of the coefficients are also highly significant which shows that our model is reliable for future estimates. The value of R - square is certainly very high showing a very high linear association between the variables and on the other hand simultaneously proving no existence of multicollinearity between the variables and if it could arise so for managing it we have taken the logarithmic model to make our model highly significant.

The model though looks significant and reliable for future estimates but it should be analyzed whether the data in the model is stationary or not i.e. having long run relationship with each other that could result in biased future estimates. For this purpose cointegration test will be applied to find the non - stationary points in the data to make the model non - spurious. One local way of finding the spurious relationship is that the Durbin - Watson value of the model should be less than the R - square value. Now, from the results we can see that the model looks spurious because and can give non - sense results. Therefore, cointegration testing is compulsory in order to reach reliable results.

Johansson Cointegration Test

The basic ideas and calculations of cointegration analysis require knowledge and application of a least-squares method and are based on the concepts of stationary and non-stationary processes. Non-stationary time series has always been a problem in the econometric analysis. As it has been shown in a number of theoretical papers (Phillips, 1986), the statistical properties of the regression analysis used for non-stationary time series, is doubtful.

If the variables included in the model as covariates, non-stationary, then the estimates will be very bad and the model will be spurious. They will not have the property of consistency, i.e. it does not converge in probability to the true value with increasing sample size. Indicators such as the coefficient of determination, t-statistics, F-statistics point to a link where it is actually not. It also affects the Durban - Watson value which is considered to be the most reliable for the change and differences in the spurious model.

Date: 09/04/12 Time: 20:23

Sample (adjusted): 2001Q1 2010Q4

Included observations: 40 after adjustments

Trend assumption: Linear deterministic trend

Series: LNLOANS LNRATES LNRESV LNCAP LNNPL 

Lags interval (in first differences): 1 to 3

Unrestricted Cointegration Rank Test (Trace)

Hypothesized

Trace

0.05

No. of CE(s)

Eigenvalue

Statistic

Critical Value

Prob.**

None *

 0.723348

 99.09513

 69.81889

 0.0000

At most 1

 0.418591

 47.69530

 47.85613

 0.0518

At most 2

 0.284995

 26.00328

 29.79707

 0.1286

At most 3

 0.236706

 12.58467

 15.49471

 0.1310

At most 4

 0.043529

 1.780213

 3.841466

 0.1821

 Trace test indicates 1 cointegrating eqn(s) at the 0.05 level

 * denotes rejection of the hypothesis at the 0.05 level

 **MacKinnon-Haug-Michelis (1999) p-values

Unrestricted Cointegration Rank Test (Maximum ...