台灣景氣對策信號趨勢之研究與預測 - 應用類神經網路

Abstract

景氣循環與未來經濟趨勢向來是社會大眾所關注的問題，國家經濟之盛衰亦隨著景氣的波動而受影響，因此世界各國為了定位目前國家景氣的好壞選擇適合該國情之總體經濟指標以計算景氣指標。景氣指標包括領先指標、同時指標與落後指標，分別代表未來經濟走勢與目前經濟現況，這些景氣指標使得政府與國民得以了解目前的景氣狀況，因此景氣指標一值以來是企業與投資者在投資決策時所關注的焦點。 台灣行政院經濟建設委員會亦定期公告目前景氣對策信號、領先指標與同時指標以代表台灣經濟循環之現況。以不同燈號提供各企業與投資者決策的參考依據，於景氣過熱時減少產能，景氣低迷時調整產出。政府在發展經濟時亦可將景氣的好壞納入考量，以調整目前經濟發展的速度。 本文旨在研究相關總體經濟因素的變動與景氣對策信號的趨勢之關聯。以實質景氣循環理論為基礎，並從文獻中選取影響景氣變動之因子做為類神經網路之輸入項，類神經網路藉由學習總體經濟歷史資料以預測未來的景氣對策信號，類神經網路良好的預測效果證實總體經濟指標係以非線性之方式影響經濟景氣。

The Business Cycle and future economic situation are always be interested by people. Rise and fall of the nation’s economy also moves along with the fluctuations of Business Cycle. In order to orientate the economic situation, the feasible economic indicators are chose to make up business indicators in many countries. Each of the business indicators, including Leading Index, Coincident Index and Monitoring Index, represent the economic condition and future economy trend. These business indicators can help the government officials and investors to measure the situation of economic development, and provide valued information for decision making. This research focus on the connection of the macro economy factors between the Monitoring Index in the future. The experiment chooses many macro economy factors from previous literatures about the Business Cycle and macro economy, which may influence economy. In the consequent, the Artificial Neural Networks are trained using these economy factors. After training, the future Monitoring Index is predicted as the Artificial Neural Networks’ output. The results of this experiment indicate that some of the macro economic indicators can affect future Monitoring Index non-linearly and the Real Business Cycle Theory is available.