چکیده:
The Rational Expectations Permanent Income Hypothesis
implies that consumption follows a martingale. However, most
empirical tests have rejected the hypothesis. Those empirical
tests are based on linear models. If the data generating process is
non-linear, conventional tests may not assess some of the
randomness properly. As a result, inference based on
conventional tests of linear models can be misleading.
This paper tests for the presence of stochastic non- linearity
in aggregate consumption of non- durable goods and services,
using US and Canadian data. The two major tests applied are a
test devised by Brock, Dechert, and Scheinkman, and a test
based on an Artificial Nemral Network model. The results
support the hypothesis that there is no non- linearity in the data.
The forecast results, however, suggest that even though linearity
is not rejected, the non-linear ANN model tends to outperform
the linear ARIMA model over three different horizons.
خلاصه ماشینی:
Non-linear dynamic, Rational ExpectationsPermanent Income Hypothesis, BDS test, ANN test •- Saeed Moshiri, Assistant Professor, Faculty of Economics, Allam eh Tabatabie University.
To provide more direct evidence on the presence or absence of a non-linear component, we apply the BDS and ANN tests to the residuals of the Canadian and US models above.
The BDS test statistics are not large enough to reject the hypothesis of white noise residuals implying that the aggregate consumption series is linear.
Table 1: BDS statistics (Wm,T(c:) in equation 5) for filtered residuals of linear model, US and Canadian consumption of non-durable goods and services, 1946:1-1996:4 e a m CAN 2 0.
به تصویر صفحه رجوع شود Embedding Dimension 3 Embedding Dimension Figure 3: Correlation Dimension vs embedding dimension To run an ANN test, the intermediate totals G(xyJ are calculated using random coefficients for y and the lagged consumption series as the input vector x.
Tabel 3: The results of the ANN test for the US and Canadian aggregate consumption به تصویر صفحه رجوع شود " The null hypothesis is that the underlying data generation model is linear.
The ANN test result in case of the US consumption series, as presented in Table 3, clearly supports the result of the BDS test: the hypothesis of linearity cannot be rejected at 5 percent significance level.
Although the ANN test result in case of the Canadian consumption series is not consistent with the BDS test result at 5 percent significance level, the hypothesis of the linearity cannot be rejected at 2.
Most empirical tests, based on a linear model of aggregate consumption behaviour, have rejected the hypothesis.