協整:醉漢牽著一條狗

讀研究生的時候,就聽周雨田老師說過這句描述協整精髓的名言:

協整就是:醉漢牽著一條狗 (A Drunk with his Dog)。

今天又被學生問及協整,想起這句話,查了一下,居然源自一篇正式發表的論文:

Murray, Michael P. (1994). "A Drunk and her Dog: An Illustration of Cointegration and Error Correction" (PDF). The American Statistician. 48 (1): 37–39. doi:10.1080/00031305.1994.10476017.

直觀解釋

  • 兩個不認識的醉漢 ? 假設你看到兩個醉漢在路上晃悠 (計量上稱為兩個隨機遊走序列,two random walks),二人彼此不識 (稱為二者彼此獨立,they"re independent)。在這種情況下,二人的行進路徑沒有任何關係。
  • 醉漢牽著他的狗 ? 如果單獨看醉漢或狗的行動軌跡,都是沒有規律的(隨機遊走,random walk),但如果一起看,就會發現兩個重要特徵:(1) 二者之間的距離是有規律的——介於 0 和繩長之間,這是長期關係;(2) 短期來看,要關注的問題在於:到底是醉漢拉著狗,還是狗拉著醉漢。
  • 引申:廣州和北京的房價 ? 單獨看廣州或北京的房價,似乎沒有明確的規律。若放在一起看,會發現二者 纏纏綿綿到天邊。二者的長期關係決定於中國的土地政策和財稅政策。短期關係則需重點分析兩個視角:
    • 其一,是廣州的房價牽制了北京的房價嗎? 有些北漂實在無法忍受帝都的高房價了,選擇跑到廣州發展,導致廣州房價小漲,間接拖住了北京房價上漲的腳步。
    • 其二,還是廣州房價跟著北京房價跑? 投資客們看到帝都房價猛漲,但由於各種限制,無法去摻乎,隨即選擇在廣州投資房產,推高了廣州的房價。

附:協整的發展歷程

The first to introduce and analyse the concept of spurious—or nonsense—correlations was Udne Yule in 1926.[2]. Before the 1980s many economists used linear regressions on (de-trended) non-stationary time series data, which Nobel laureate Clive Granger and Paul Newbold showed to be a dangerous approach that could producespurious correlation,[3][4] since standard detrending techniques can result in data that are still non-stationary.[5] Granger"s 1987 paper with Robert Engle formalized the cointegrating vector approach, and coined the term.[6]

參考資料:

  1. "https://www.quora.com/What-is-cointegration-of-time-series-data-in-statistics-How-do-you-identify-pairs-of-stocks-for-trading-Is-it-based-on-simple-correlation-analysis-or-is-there-a-more-rigorous-method">What is cointegration of time series data in statistics?
  2. Wikipedia-Cointegration
  3. ref="https://r.search.yahoo.com/_ylt=A0LEVwwynPZZ">The Stata Blog ? Cointegration or spurious regression?

延伸閱讀:

  1. Panel-data cointegration tests | New in Stata 15
  2. ref="blog.stata.com/tag/time">The Stata Blog ? time series
  3. ref="blog.stata.com/tag/nons">The Stata Blog ? nonstationary
  4. ref="blog.stata.com/2016/06/">The Stata Blog ? Unit-root tests in Stata
  5. Stata 範例:Unit Roots and Cointegration - Econometrics at Illinois
  6. Cointegration – Johansen Test with Stata

weixin.qq.com/r/7Ujm-tf (二維碼自動識別)

最後,發張偉大的照片!


推薦閱讀:

TAG:Stata | stata连享会 | 计量经济学 |