• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文学位 >Bayesian variable and model selection in economics: An application in urban economics.
【24h】

Bayesian variable and model selection in economics: An application in urban economics.

机译：经济中的贝叶斯变量和模型选择：在城市经济学中的应用。

获取原文
获取原文并翻译 | 示例
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Scope and method of study. The purpose of this study is to investigate the use of Bayesian variable selection procedures in the field of urban economics. Three Bayesian procedures are used with two data sets. One set of data consists of observations on the population growth of the central city (referred as CITY) and the other of observations on Metropolitan Statistical Areas (MSA). The sample period for each is 1970 to 1990. The determinants of growth consist of 23 variables for CITY, and 21 variables for the MSA. Bayesian procedures are used to determine the appropriate subset of variables required to model the economic growth of cities and MSAs and the outcomes are compared to more traditional procedures from classical statistics.; Findings and conclusion. The Bayesian variable selection procedures are more flexible than the traditional approach. This occurs because prior information can be used systematically based on the expert knowledge of the researcher. For the two of the three procedures (MBVS and Geweke), the posterior marginal probabilities that individual variables appear in the model are highly correlated with one another. This is taken as evidence that the two procedures produce results that are consistent; the results from MBVS and the Geweke procedure are not highly correlated with those from the other technique employed, Bayesian model averaging (BMA), despite the fact that similar prior information is used for each estimator. The likely reason is slower convergence of the Markov Chain in the BMA algorithm. When smaller subsets of the variables are searched, correlations among the marginal probabilities improve, suggesting that a very large number of iterations is required to ensure convergence of the Markov Chains.
机译：研究范围和方法。这项研究的目的是调查贝叶斯变量选择程序在城市经济学领域的使用。三个贝叶斯过程用于两个数据集。一组数据包括对中心城市（称为CITY）人口增长的观察，另一组对大都市统计区（MSA）的观察。每个阶段的采样期为1970年至1990年。增长的决定因素包括CITY的23个变量和MSA的21个变量。贝叶斯方法用于确定对城市和MSA的经济增长进行建模所需的变量的适当子集，并将结果与​​经典统计中的更多传统方法进行比较。 发现和结论。贝叶斯变量选择过程比传统方法更灵活。这是因为可以根据研究人员的专业知识来系统地使用先验信息。对于这三个过程中的两个（MBVS和Geweke），各个变量在模型中出现的后缘概率彼此高度相关。这是两个程序产生一致结果的证据。尽管每个估计量都使用了类似的先验信息，但是MBVS和Geweke过程的结果与采用的其他技术贝叶斯模型平均（BMA）的结果并没有高度相关。可能的原因是BMA算法中马尔可夫链的收敛速度较慢。当搜索变量的较小子集时，边际概率之间的相关性会提高，这表明需要大量的迭代才能确保马尔可夫链的收敛。

著录项

  • 作者

    Tien, Jui-Chu.;

  • 作者单位

    Oklahoma State University.;

  • 授予单位 Oklahoma State University.;
  • 学科 Economics Theory.
  • 学位 Ph.D.
  • 年度 2001
  • 页码 147 p.
  • 总页数 147
  • 原文格式 PDF
  • 正文语种 eng
  • 中图分类 经济学;
  • 关键词

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号