0% found this document useful (0 votes)
51 views3 pages

Test 3

The document describes using two different methods - general-to-specific and specific-to-general - to build a regression model of the federal funds rate on other economic variables. Both methods result in the same final model. This model is then compared to an alternative Taylor rule model, with the document's model having a slightly higher R-squared and lower AIC and BIC, suggesting it is preferable. Diagnostic tests on the Taylor rule model show evidence against the null hypotheses of the RESET, Chow breakpoint, and forecast tests, indicating issues with the Taylor rule model specification.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
51 views3 pages

Test 3

The document describes using two different methods - general-to-specific and specific-to-general - to build a regression model of the federal funds rate on other economic variables. Both methods result in the same final model. This model is then compared to an alternative Taylor rule model, with the document's model having a slightly higher R-squared and lower AIC and BIC, suggesting it is preferable. Diagnostic tests on the Taylor rule model show evidence against the null hypotheses of the RESET, Chow breakpoint, and forecast tests, indicating issues with the Taylor rule model specification.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

a) Use general-to-specific to come to a model.

Start by regressing the federal funds rate on


the other 7 variables and eliminate 1 variable at a time.

Linear regression
INTRATE Coef. St.Err. t- p- [95% Conf Interval] Sig
value value
INFL 0.696 0.062 11.19 0 0.574 0.818 ***
PROD -0.058 0.04 -1.45 0.148 -0.136 0.021
UNEMPL 0.102 0.097 1.06 0.290 -0.088 0.292
COMMPRI -0.006 0.003 -1.86 0.064 -0.011 0 *
PCE 0.344 0.069 4.96 0 .0208 0.481 ***
PERSINC 0.247 0.061 4.08 0 0.128 0.366 ***
HOUST -0.019 0.005 -4.16 0 -0.029 -0.01 ***
Constant -0.221 0.245 -0.90 0.367 -0.702 0.26

Mean dependent var 5.348 SD dependent var 3.619


R-squared 0.639 Number of obs 660
F-test 164.532 Prob > F 0.000
Akaike crit. (AIC) 2914.301 Bayesian crit. (BIC) 2950.239
*** p<.01, ** p<.05, * p<.1

When the model is run, we have that the PROD, UNEMPL and COMMPRI variables do not meet the
requirement that the p-value be less than 0.05. Complying with the exercise, the least significant
variable, which is UNEMP, is removed from the model.

Linear regression
INTRATE Coef. St.Err. t- p- [95% Conf Interval] Sig
value value
INFL 0.693 0.062 11.15 0 0.571 0.815 ***
PROD -0.025 0.026 -0.99 0.323 -0.076 0.025
COMMPRI -0.007 0.003 -2.31 0.021 -0.012 -0.001 **
PCE 0.369 0.066 5.62 0 0.24 0.497 ***
PERSINC 0.252 0.06 4.16 0 0.133 0.37 ***
HOUST -0.021 0.004 -4.76 0 -0.03 -0.012 ***
Constant -0.291 0.236 -1.23 0.218 -0.754 0.173

Mean dependent var 5.348 SD dependent var 3.619


R-squared 0.638 Number of obs 660
F-test 191.731 Prob > F 0.000
Akaike crit. (AIC) 2913.436 Bayesian crit. (BIC) 2944.882
*** p<.01, ** p<.05, * p<.1

We have that the PROD variable is not significant, so it is removed from the model.
Linear regression
INTRATE Coef. St.Err. t- p- [95% Conf Interval] Sig
value value
INFL 0.718 0.057 12.55 0 0.605 0.83 ***
COMMPRI -0.008 0.003 -2.84 0.005 -0.013 -0.002 ***
PCE 0.341 0.059 5.76 0 0.224 0.457 ***
PERSINC 0.24 0.059 4.05 0 0.124 0.357 ***
HOUST -0.021 0.004 -4.68 0 -0.029 -0.012 ***
Constant -0.24 0.23 -1.04 0.298 -0.692 0.212

Mean dependent var 5.348 SD dependent var 3.619


R-squared 0.637 Number of obs 660
F-test 229.889 Prob > F 0.000
Akaike crit. (AIC) 2912.423 Bayesian crit. (BIC) 2939.377
*** p<.01, ** p<.05, * p<.1

The coefficients indicate the relationship with the dependent variable, which is the interest rate,
has a positive relationship with inflation, personal consumption spending and personal income.
That this agrees with economic theory. It has a negative relationship with the prices of raw
materials and with housing starts. Which conforms to economic theory because a lower level of
interest rate increases the value over time.

b) Use specific-to-general to come to a model. Start by regressing the federal funds rate on
only a constant and add 1 variable at a time. Is the model the same as in (a)?
Linear regression
INTRATE Coef. St.Err. t- p- [95% Conf Interval] Sig
value value
INFL 0.718 0.057 12.55 0 0.605 0.83 ***
COMMPRI -0.008 0.003 -2.84 0.005 -0.013 -0.002 ***
PCE 0.341 0.059 5.76 0 0.224 0.457 ***
PERSINC 0.24 0.059 4.05 0 0.124 0.357 ***
HOUST -0.021 0.004 -4.68 0 -0.029 -0.012 ***
Constant -0.24 0.23 -1.04 0.298 -0.692 0.212

Mean dependent var 5.348 SD dependent var 3.619


R-squared 0.637 Number of obs 660
F-test 229.889 Prob > F 0.000
Akaike crit. (AIC) 2912.423 Bayesian crit. (BIC) 2939.377
*** p<.01, ** p<.05, * p<.1
Doing the one-by-one methodology we have the same model as in point a.

c) Compare your model from (a) and the Taylor rule of equation (1). Consider R 2 , AIC and
BIC. Which of the models do you prefer?
Linear regression
INTRATE Coef. St.Err. t- p- [95% Conf Interval] Sig
value value
INFL .975 .033 29.78 0 .911 1.039 ***
PROD .095 .02 4.80 0 .056 .133 ***
Constant 1.249 .176 7.09 0 .903 1.595 ***

Mean dependent var 5.348 SD dependent var 3.619


R-squared 0.575 Number of obs 660
F-test 443.899 Prob > F 0.000
Akaike crit. (AIC) 3011.616 Bayesian crit. (BIC) 3025.093
*** p<.01, ** p<.05, * p<.1

This is the Taylor model where you have an R2 with 0.575. AIC: 3013.62 - BIC: 3031.59
With the Model that we specify we have that the R2 0.637.

d) Test the Taylor rule of equation (1) using the RESET test, Chow break and forecast test
(with in both tests as break date January 1980) and a Jarque-Bera test. What do you
conclude?

RESET test
data: taylor
RESET = 2.2578, df1 = 2, df2 = 655, p-value = 0.1054

M-fluctuation test
data: taylor
f(efp) = 5.6454, p-value < 2.2e-16

Chow break testing


F value d.f.1 d.f.2 P value
2.873501e+01 3.000000e+00 6.540000e+02 1.836802e-17

So we can conclude that we need to reject the Ho, and say that there is no normal distribution
in the residuals.

You might also like