Riksbank's and NIER's forecasts of Swedish inflation

The data on Swedish Riksbank's density forecasts of CPI inflation, actual inflation and NIER's forecasts of Swedish inflation are used in an illustration of testing forecast efficiency conditions.<br>FParam.tsv<br>Sweden's central bank (Riksbank) started to publish its density forecasts of inflation in June 1998. The forecasts are in the form of two-piece normal distribution. The target variable is the yearly CPI inflation. We evaluate only one-year-ahead forecasts. There are 64 forecasts available for evaluation for the period 1999–2015. They were issued 4 times a year with approximately quarterly frequency. Forecasts before 2007 are conditional, assuming a predetermined trajectory of the repo rate. The file contains 4 variables: the 3 parameters of two-piece normal distribution mu, s1, s2 and the date (month).<br>Actual.tsv<br>The file contains the actual values of the CPI innflation (the so called shadow version).<br>Actual_Last.tsv<br>The file contains the actual lagged values of the CPI innflation.<br>NIER.tsv<br>The file contains the point forecasts of inflation by the National Institute of Economic Research.<br>

本数据集用于演示预测效率条件检验，所用数据包含瑞典中央银行（Riksbank）发布的消费者物价指数（CPI, Consumer Price Index）通胀密度预测、实际通胀数据，以及瑞典国家经济研究所（NIER, National Institute of Economic Research）的瑞典通胀预测数据。<br>FParam.tsv<br>瑞典中央银行（Riksbank）于1998年6月首次发布通胀密度预测，该预测采用双分段正态分布（two-piece normal distribution）形式，目标变量为年度消费者物价指数（CPI）通胀率。本研究仅针对提前一年的预测展开评估，1999年至2015年间共有64组可用于评估的预测数据，发布频率近似季度频率，每年发布4次。2007年之前的预测为条件预测，假设回购利率（repo rate）路径已预先确定。该文件包含4个变量：双分段正态分布的3个参数μ、s₁、s₂，以及预测发布的月份日期。<br>Actual.tsv<br>该文件包含消费者物价指数通胀率的实际观测值（即所谓的影子版本数据）。<br>Actual_Last.tsv<br>该文件包含消费者物价指数通胀率的滞后实际观测值。<br>NIER.tsv<br>该文件包含瑞典国家经济研究所（NIER）发布的通胀点预测数据。