本文的研究在回顾已有研究方法和成果的前提下，以对我国CPI统计过程的分析为基础，在延续城乡住户调查消费支出方法的主要框架的同时，在CPI权数估算思路、内容和方法上均取得了一定的发展。本文的主要结论是：第一，CPI权数实际计算过程难以完全复刻。我们的研究发现，从学术上完全复刻其统计过程几乎是不可能的，通常能做的只是尽可能把握其生产过程，从而有助于对CPI数据有更为深刻的理解。第二，我们的CPI权数估算方法具有较好的可靠性。经过检验，我们发现，使用我们的权数估算方法对CPI大类全国指数和总指数的计算偏差较小，表明我们对CPI大类全国权数的估算有较好的可靠性。未来，为了得到更加切近的结果，还应该在现有计算中考虑自有住房调整等因素。第三，在权数估算中城乡常住人口统计优于市县常住人口估计。通过检验，我们发现，与市县常住人口统计相比，以城乡常住人口资料为基础计算得到的城乡权数能够反映城市权数总体增加的趋势，而市县常住人口则不能反映，因而对于CPI权数估算而言，城乡常住人口统计表现更优。第四，公开信息反映的居住类权数时间趋势仍然有待解释。我们的估算结果无法反映上述时间趋势，自有住房调整的估计方法差异、人口数据差异和全国城乡人均消费支出数据差异可能是主要原因，但具体原因仍然有待进一步确定。第五，人口数据和自有住房调整计算方法是相关研究的最大难点。不仅全国居住类大类权数时间趋势问题的解释可能取决于人口数据和自有住房调整计算方法，而且人口数据和自有住房调整计算方法也是影响其他权数的基本因素。

Based on the survey of past research method and finished papers, and also detailed analysis of the statistical process of CPI in China, the thesis inherited the main framework of the method of the surveyed consumption expenditure data from urban and rural household survey and, further on, developed an estimation method of CPI weights, from all classified levels. In the meantime, the thesis also made progresses regarding the estimation design, content, and method.To sum up, the thesis reached to three major conclusions: First of all, it was found that the real process of CPI weights calculation could hardly be reengineered. Our research found that, academic reengineering of the real process could hardly be possible for the complexity involved. However, more sophisticated mater of the production process of CPI weights could lead to better understanding of CPI statistics. Secondly, the estimation method of CPI weights introduced by the thesis was both consistent and reliable. With the test of the bias between public CPI and estimated CPI, we found that the comparatively small bias possibly was the result of sound and reliable estimation method constructed by the thesis. Further on, the adjustment of self-owned residences could be added to our calculation to further reduce the bias.Thirdly, the demographic statistics of urban and rural resident was possibly superior to the one of city and county in the regard of CPI weights estimating. With the test of the bias between public CPI and estimated CPI, with different datasets from the demographic statistics of urban and rural resident and the demographic statistics of city and county residents, we found that the former showed a growing trend of national urban weight that was consistent with the publicized CPI data, which meant it was a choice with better chance for the calculation of CPI weights. Fourthly, the cause of the growing time trend of the weight of residence class still remained to be explained. While our estimates could not reflect the above trend, self-owned residence rejustment method, difference in demographic statistics, and difference in national urban and rural consumption per capita statistics are possible reasons. However, the detailed reason should be positioned further.Lastly, demographic statistics and self-owned residence rejustment method perhaps weretopics of the most difficult nature in related field of research. The significance of the two issues was not only because they were possible candidates for the explain of the time trend of the weight of residence class, but also because they were basic factors that influence the estimation of other weights.