Modeling And Analysis Of Unemployment Early Warning

With Chinese GDP growth down, the economy has entered a new normal stage. "Economic growth in a reasonable interval" concept putting forward, in policy goals full employment has been regarded as the bottom line. At present, the overall unemployment situation in Jiangsu province is controlled within a reasonable range, but hidden unemployment and serious structural unemployment problem will be a potential threat that the government need monitor. The establishment of early warning system of unemployment for 10 years in Jiangsu Province, which relies on the information system and forth public employment service network that covered urban and rural. Jiangsu province actively carry out the exploration and practice of employment and unemployment monitoring and warning mechanism construction. A number of monitoring and early warning index system has initially formed, which collects enterprise employment survey in spring, unemployment dynamic monitor-ing, enterprise monitoring, human resources survey in urban and rural areas and unemployment early warning application system. But currently there are still many problems in unemployment early warning system in Jiangsu Province.This paper is based on unemployment theory and the development of the early warning system, the status quo of unemployment in Jiangsu province are analyzed in detail. In addition to economic development, population resources, social security and the cost of living, it also selected the unemployed structure index as the main factors that influence the unemployment for empirical analysis, with the time series data from 1990 to 2013. To avoid using a single model making the empirical results are biased, testing data respectively with the linear relationship model and the nonlinear relation-ship model, and adopting comparative analysis method to determine the appropriate model. Then with the best dialectical prediction model of unemployment, making structural unemployment early-warning model, carrying on empirical analysis and improving the early warning system of unemployment in Jiangsu province. This is the contribution of this paper.The main conclusions of this paper are as follows:(1) Dialectically comparing the effect of unemployment prediction between BP neural network and multiple regression model, the result is that linear relationship is more significant than nonlin-ear relationship between the independent and dependent variables in this paper, and this is different from the conclusion that machine learning has an overwhelming advantage in forecasting the unemployment rate which some scholars believe. (2) Economic growth, promoting employment, increasing cost of living can restrain the rise of unemployment rate. And the industrial structure inclined to the third industry, increasing the quantity and cost of labor resources, improving unemploy-ment insurance are the factors to promote the increase in the unemployment rate. (3) The police intelligence forecast by the early warning system of unemployment in Jiangsu province and the development of its economy work together. There was a mild unemployment risk warning between 2001 and 2006. Jiangsu province suffered a unemployment risk moderate warning with a financial crisis between 2007 and 2008.And the unemployment situation is good and no warning between 2009 and 2013. Based on the above conclusion, this paper also puts forward some suggestions as references when Jiangsu provincial government makes decision.