Nonlinear dynamic models : applications of chaos theory and Threshold Autoregressive model to agricultural prices /
This dissertation investigates nonlinear dynamic time series models. In chapter 11, we investigated the presence of the nonlinear deterministic dynamic behavior, which is called chaos. Barrel and block behavior, which is called chaos. Barrel and block a recently introduced Nycva-Ellner-Gallant-McCaf...
| Main Author: | |
|---|---|
| Format: | Thesis Book |
| Language: | English |
| Published: |
[Place of publication not identified] :
[publisher not identified] ;
1998.
|
| Subjects: | |
| Online Access: | http://proxy.library.tamu.edu/login?url=http://proquest.umi.com/pqdweb?did=732514561&sid=1&Fmt=2&clientId=2945&RQT=309&VName=PQD |
| Summary: | This dissertation investigates nonlinear dynamic time series models. In chapter 11, we investigated the presence of the nonlinear deterministic dynamic behavior, which is called chaos. Barrel and block behavior, which is called chaos. Barrel and block a recently introduced Nycva-Ellner-Gallant-McCaffrey (NEGM) method for the calculation of positive Lyapunov exponent (LE), we found that these data show nonlinearity. We argue that seasonality shown in these price data, combined with the government's price support policy, creates a limit cycle. These results are consistent with nonlinear dynamic behavior. Certain nonlinear time series tools and techniques have been inspired by the chaos theory. The Threshold Autoregressive (TAR) model is one of these. In chapter III, the TAR model is used for modeling the limit III, the TAR model is used for modeling the limit explaining this phenomenon. We found that the TAR model is superior to the AR model in 2- and 4-step model is superior to the AR model in 2- and 4-step not for barrel cheese prices. This may be due to the not for barrel cheese prices. This may be due to the products. In chapter IV, we applied the generalized TAR model to the speculative storage theory. According to Deaton and Laroque (1992, 1996), speculators expect price increases in the future if the current price falls below the expected price level of the future. Hence, below the expected price level of the future. Hence, On the contrary, if the current price rises above that level, they expect no future profits by storage, and they do not store the commodity. We investigated whether there exists such a critical level (threshold) and a nonlinear relationship between the world-ending stock (WES) and A-lndex (AI) of cotton. Our empirical results show the nonlinear relationship between these data. That is, WES is more significantly influenced by M if M is below 80 than above 80. Also, we found that the multivariate TAR model is superior to the linear model in out-of-sample forecasting. In addition, we investigated the autocorrelation characteristics of AI. Our results are as follows: first, if .AI is below 80, it is strongly autocorrelated; second, if AI is above 80, its autocorrelation is weakened. The storage model provides a good explanation for the dynamic behavior of AI. Our empirical results confirm predictions made by Deaton and Laroque: first, the speculative storage model can explain the high autocorrelations of prices that look like random walk; second, speculation can substantially increase price autocorrelation that is substantially weakened in the absence of storage. In conclusion, nonlinear dynamic models, especially threshold models, have rich potential for fruitful applications in the study of agricultural markets. Our efforts have provided strong evidence that storage can induce nonlinearity. |
|---|---|
| Item Description: | Vita. |
| Physical Description: | x, 102 leaves : illustrations ; 28 cm. Issued also on microfiche from University Microfilm Inc. |
| Bibliography: | Includes bibliographical references: pages 84-89. |