人工智能技术对就业的影响及趋势预测:基于北京市工业机器人安装量的数据

本文基于德国国际机器人联盟(IFR)统计的中国工业机器人安装量数据,利用回归模型分析检验了人工智能(AI)技术对北京市就业总量、从业人员工资水平、从业人员素质结构的影响。结果显示,北京市机器人安装量对于北京市的就业总量具有替代效应,对于从业人员的工资收入具有补偿效应,对于从业人员的素质提升则具有推动效应,而北京市三次产业增加值的增长会调节机器人安装量对就业总量的影响。运用灰色预测模型法,文章有效预测了2019-2030年北京市机器人安装量变化所引起的就业变化,并提出了相应的政策建议。     

碳达峰目标下国家生态文明试验区碳强度预测及减排潜力研究

基于ARIMA-BP神经网络模型并在碳达峰目标下预测国家生态文明试验区的碳强度；同时，运用非径向方向距离函数、共同前沿理论及共同前沿非径向曼奎斯特二氧化碳排放效率指数(MNMCPI)对减排潜力进行深入探讨。研究发现：(1)由ARIMA-BP神经网络模型所得碳强度预测序列均方误差(MSE)为0.005，平均绝对百分比误差(MAPE)为1.95%，显示了较高的精度；国家生态文明试验区碳强度2028年相较于2005年下降67.5%，提前实现国家2030年碳强度比2005年下降65%以上的目标，其中：贵州碳强度下降幅度最大，2030年相较2005年下降77.5%，福建、江西、海南下降幅度均在65%以上。(2)国家生态文明试验区减排潜力呈波动下滑趋势，2016—2030年均减排空间达26.59%，约21649.33万吨CO₂。到2030减排潜力达21.9%，为2016—2030年最低值，由于减排潜力与碳排放效率负相关，碳排放效率越高则减排潜力越小，这意味着国家生态文明试验区碳排放效率得到优化。(3)通过探究CO₂排放效率变化，间接分析区域内减排潜力变化...     

Why is the forecast error of quarterly GDP in Japan so large? – From an international comparison of quarterly GDP forecast situation

We examined the accuracy of prediction of Canada, Japan, United Kingdom, and United States from the viewpoint of forecast errors. Compared with the forecast error of each country at the around same time, the forecast error of Japan is about 2 times larger. In case of Japan, even immediately before release of quarterly GDP, the forecast error is over 1 %, which is the same level of forecast error as 94 days before in the United States and 135 days before in Canada.Evaluating the characteristics of forecast errors, it can be pointed out that Japan's forecasts are as efficient as those of other countries, and the addition of major economic statistics is unlikely to improve forecast errors. The reason for Japan's large forecast errors is the fluctuations in the GDP growth rate. These results provide evidence that volatile GDP may make the outlook worse.Large fluctuations in Japan's quarterly GDP have already been pointed out. It is necessary to examine the factors behind the large fluctuations in the rate of change in Japan's quarterly GDP.