Enders-Siklos JBES 2001

[TomDoan]

This is the replication file for Enders and Siklos(2001), "Cointegration and Threshold Adjustment," JBES, vol. 19, no. 2, 166-76 which estimates and tests for cointegration in a bivariate system allowing for asymmetrical error corrections.

Zip with program and data

Most of the work is done with the @EndersSiklos procedure.

[dominrod]

Dear Tom Doan,

How do I estimate the error-correction model presented in the equation (20 ') by Enders and Siklos (2001)?

Thanks in advance

[TomDoan]

This estimates the error correction models (20) and (21) that are derived from the MTAR estimates of the threshold values. (%%BREAKVALUE is defined by the @EndersSiklos procedure). They differ very slightly from (20) and (21) in the paper which may be due to rounding on the break value (%%BREAKVALUE is the exact estimated break value). (20') would be estimated the same way, but it uses a different data set, which I don't have. (rates.xls starts in 1964:1).

*
* Estimate the momentum error correction models: (20) and (21) in the paper.
*
set plus  = (dseries>=%%breakvalue)*u{1}
set minus = (dseries<%%breakvalue)*u{1}
*
set df  = logff-logff{1}
set d10 = log10-log10{1}
*
linreg df
# constant plus minus df{1 2} d10{1 2}
linreg d10
# constant plus minus df{1 2} d10{1 2}