Skip to main content

Advertisement

Table 2 Capital accumulation equations (a) IIa and (b) IIb.

From: Asymmetric real-exchange-rate effects on capital accumulation: evidence from non-linear ARDL models for Mexico

  (2a.1) (2a.2) (2a.3) (2a.4)
Manufacturing Tradables Non-tradables Whole economy (tradables + non-tradables)
(a) Capital accumulation equations, IIa
Estimated coefficients
Depreciation component, RERD 0.203 (0.02) 0.182 (0.02) − 0.054 (0.05) 0.080 (0.01)
Appreciation component, RERA 0.247 (0.00) 0.244 (0.00) − 0.055 (0.02) 0.096 (0.00)
Production growth rate 0.756 (0.06) 1.052 (0.00) 0.638 (0.00) 0.948 (0.00)
Error-correction coefficient, ρ − 0.278 (0.00) − 0.215 (0.00) − 1.016 (0.00) − 0.329 (0.00)
Diagnostics
Jarque–Bera 0.765 0.750 0.534 0.639
Breusch‒Godfrey, n = 2 0.552 0.316 0.489 0.763
ARCH 0.799 0.421 0.987 0.178
Ramsey’s RESET 0.910 0.999 0.729 0.795
Adjusted R-squared 0.938 0.947 0.866 0.954
Wald and bounds tests
Wald RERD = RERA 0.035 0.005 0.965 0.146
Bounds t statistic − 3.87** − 3.27# − 4.70*** − 3.88**
Bounds F statistic 24.13*** 33.42*** 11.93*** 36.06***
  1. Dependent variable: capital accumulation rate
  2. Long-run coefficients from non-linear EC ARDL models, 1992–2015, 24 annual observations
  3. For illustrative purposes, the p values for the di coefficients from Eq. (2a) in the text are shown in parenthesis, next to the estimated coefficients
  4. For an explanation of the diagnostic and bounds tests, see Table 1
  5. All the equations include intercept and an outlier year dummy for 1995 (not shown), except column (2b.3). In addition, columns (2a.1) and (2a.2) include outlier dummies for 2009 and 2013, column (2a.3) includes an outlier dummy for 2009 and no intercept, and columns (2b.1) and (2b.2) include an outlier year dummy for 2013
  6. Due to increased number of lags in the final equation, the sample in columns (2b.2) and (2b.3) was reduced to 1993–2015
  7. All the variables are expressed in %, except RER, which is in natural logs times 100
  8. Wald tests: p values from tests applied to the di coefficients of Eq. (2a) in the main text. For the bounds F test, RERD and RERA were counted as a single regressor, corresponding to the stricter test (i.e., the test with higher critical values). For the same reason, for the t test they were counted as two separate regressors. See the discussion in Shin et al. (2014)
  9. Panel a—#Rejects the null at 10% under the condition that RERD and RERA are considered a single regressor
  10. Panel b—#Rejects the null at 10% under the condition that RERD and RERA are considered a single regressor, and the same for QH and QL
  (2b.1) (2b.2) (2b.3) (2b.4)
Manufacturing Tradables Non-tradables Whole economy (tradables + non-tradables)
(b) Capital accumulation equations, IIb
Estimated coefficients
Depreciation component, RERD 0.175 (0.04) 0.230 (0.00) − 0.034 (0.02) 0.084 (0.00)
Appreciation component, RERA 0.228 (0.00) 0.271 (0.00) − 0.040 (0.00) 0.093 (0.00)
Production growth rate, high, QH 0.954 (0.01) 0.650 (0.00) 0.662 (0.00) 0.779 (0.00)
Production growth rate, low, QL 1.735 (0.00) 1.837 (0.00) 0.406 (0.00) 1.115 (0.00)
Error-correction coefficient, ρ − 0.232 (0.00) − 0.272 (0.00) − 1.366 (0.00) − 0.360 (0.00)
Diagnostics
Jarque–Bera 0.992 0.969 0.855 0.626
Breusch‒Godfrey, n = 2 0.878 0.719 0.299 0.593
ARCH 0.815 0.647 0.754 0.953
Ramsey’s RESET 0.750 0.599 0.912 0.484
Adjusted R-squared 0.933 0.969 0.940 0.966
Wald and bounds tests
Wald QH = QL 0.164 0.004 0.035 0.135
Wald RERD = RERA 0.045 0.019 0.192 0.310
Bounds t statistic − 3.33# − 4.19** − 7.33*** − 4.82***
Bounds F statistic 23.37*** 35.02*** 13.96*** 39.08***
  1. Dependent variable: capital accumulation rate
  2. Long-run coefficients from non-linear EC ARDL models, 1992–2015, 24 annual observations
  3. For illustrative purposes, the p values for the di coefficients from Eq. (2a) in the text are shown in parenthesis, next to the estimated coefficients
  4. For an explanation of the diagnostic and bounds tests, see Table 1
  5. All the equations include intercept and an outlier year dummy for 1995 (not shown), except column (2b.3). In addition, columns (2a.1) and (2a.2) include outlier dummies for 2009 and 2013, column (2a.3) includes an outlier dummy for 2009 and no intercept, and columns (2b.1) and (2b.2) include an outlier year dummy for 2013
  6. Due to increased number of lags in the final equation, the sample in columns (2b.2) and (2b.3) was reduced to 1993–2015
  7. All the variables are expressed in %, except RER, which is in natural logs times 100
  8. Wald tests: p values from tests applied to the di coefficients of Eq. (2a) in the main text. For the bounds F test, RERD and RERA were counted as a single regressor, corresponding to the stricter test (i.e., the test with higher critical values). For the same reason, for the t test they were counted as two separate regressors. See the discussion in Shin et al. (2014)
  9. Panel a—#Rejects the null at 10% under the condition that RERD and RERA are considered a single regressor
  10. Panel b—#Rejects the null at 10% under the condition that RERD and RERA are considered a single regressor, and the same for QH and QL