Research Of Marine Economy Growth In China Based On Mixed Frequency Data Model

In the 21st century, China government has begun to focus on the development of Ocean Economy and the Report to the Eighteenth National Congress of the Communist Party of China has already upgraded "Powerful Marine Nation Construction" to the level of national strategy. State Oceanic Administration has established the Gross Ocean Product (GOP) accounting regulation to evaluate the growth of ocean economy, referring to the Gross Domestic Product (GDP) accounting methodology in the field of national economic accounting. However, due to the imperfection of national ocean economy related index data, the government can only publish the annual GOP in the beginning of the year rather than publish the seasonal result regularly. In the meantime, because the sample data length of Chinese ocean economy is short, so the analyzed results based on annual low frequency data are inaccurate. To deal with this condition, some scholars apply interpolation method to transfer low frequency data to high frequency data in order. Nevertheless, the results may be suspected to be artificial and many interpolation methods tend to have no economics theoretical support as they are pure mathematical method. Mixed-frequency data model is able to use different frequency data, to solve the short sample length and to make full use of high frequency data as well.Mixed-frequency data model has been applied and extended gradually. But this model has not been applied into national ocean economy growth analysis because of the shortage of ocean economy related index data and unpublished seasonal GOP, which has also restrained its application to some extent.This dissertation intends to use mixed-frequency data model to analyze national GOP with the combination of a small number of high frequency ocean economy index data and low frequency GOP data, thereby exploring the feasibility of mixed-frequency data model in national ocean economy growth analysis, and compare and analyze them with benchmark data which are in same frequency aiming to understand whether the mixed-frequency data model is effective in Chinese ocean economy growth analysis. The innovative contributions of this dissertation are as follows:Firstly, this paper establishes a Mixed-Frequency Data Model that is suitable in China to analyze ocean economy growth. According to the accessibility of ocean economy related index data and referring to the selection of Chinese ocean economy business index, this paper uses high frequency monthly index and quarterly data of GOP got by data strip method to study the applicability and efficiency of Mixed-Frequency Data Model in the fields of China’s ocean economy growth. Secondly, this paper uses two models (MIDAS Model & MF-VAR Model) to study ocean economy growth in China, and tests the advantage of Mixed-Frequency model by comparing with baseline model.This dissertation builds the mixed-frequency data model of Chinese ocean economy growth, based on mixed-frequency data model related theoretical methods and referring to the methods of building a macro-economy mixed-frequency data model, and tests model fitting and makes a short-term prediction of Chinese ocean economy gross index, thus effectively analyzing and investing the feasibility of mixed-frequency data model in the field of Chinese ocean economy. The conclusions of the paper are as follows. Firstly, the mixed frequency data model can fully snatched up the sample information of the high frequency variable. And the Cargo throughput of the coastal port over the scale which is closely connected with the growth of China’s ocean economy shows a more accurate estimation and prediction results. The mixed-frequency data model has significant advantages compared with the same frequency benchmark model. Secondly, the two kinds of mixed frequency data model all have merit, with the increase of lag, the precision of estimation and prediction increases, especially for the first-order autoregressive MIDAS model. MF-VAR model has obvious advantage in short-term prediction. If the smoothing forecasting method is adopted in the MF-VAR model, the result of prediction is significantly superior to the MIDAS model.