ciTarLag {apt}

Lag Selection for Threshold Cointegration Regression

Description

Select the best lag for threshold cointegration regression by AIC and BIC

Usage

  ciTarLag(y, x, model = c("tar", "mtar"), maxlag = 4, 
  thresh = 0, adjust = TRUE) 

Arguments

y

dependent or left-side variable for the long-run regression.

x

independent or right-side variable for the long-run regression.

model

a choice of two models, either tar or mtar.

maxlag

maximum number of lags allowed in the search process.

thresh

a threshold value.

adjust

logical value (default of TRUE) of whether to adjust the window widths so all regressions by lag have the same number of observations

Details

Estimate the threshold cointegration regressions by lag and then select the best regression by AIC or BIC value. The longer the lag, the smaller the number of observations availabe for estimation. If the windows of regressions by lag are not ajusted, the maximum lag is usually the best lag by AIC or BIC. Theorectially, AIC and BIC from different models should be compared on the basis of the same observation numbers (Ender 2004). adjust shows the effect of this adjustment on the estimation window. By default, the value of adjust should be TRUE.

Value

Return a list object of class "ciTarLag" with the following components:

path

a data frame of model criterion values by lag, including lag for the current lag, totObs for total observations in the raw data, coinObs for observations used in the cointegration regression, sse for the sum of squared errors, aic for AIC value, bic for BIC value, LB4 for the p-value of Ljung_Box Q statistic with 4 autocorrelation coefficients, LB8 with 8 coefficients, LB12 for Q statistic with 12 coefficients

out

a data frame of the final model selection, including the values of model, maximum lag, threshold value, best lag by AIC, best lag by BIC

Methods

Two methods are defined as follows:

print:

This shows the out component in the returned list.

plot:

This demonstrates the trend of AIC and BIC changes of threshold cointegration regressions by lag. It facilitates the selection of the best lag for a threshold cointegration model.

Author(s)

Changyou Sun ([e-mail])

References

Enders, W. 2004. Applied Econometric Time Series. John Wiley & Sons, Inc., New York. 480 P.

Enders, W., and C.W.J. Granger. 1998. Unit-root tests and asymmetric adjustment with an example using the term structure of interest rates. Journal of Business & Economic Statistics 16(3):304-311.

See Also

ciTarFit; and ciTarThd;

Examples

# see example at daVich