Announcement

Collapse
No announcement yet.
X
  • Filter
  • Time
  • Show
Clear All
new posts

  • Rate of change for non-consecutive data

    Hello, this is my first time posting on statalist.

    I have a time-series data for poverty headcount ratio. I'm trying to use the change rate (something similar to gdp growth rate) of this data as the dependent variable for my cross-country regression analysis.

    I have data with many non-consecutive observations which led to stata creating missing values.

    For example,
    country year poverty
    Argentina 1985
    Argentina 1986 0
    Argentina 1987 0
    Argentina 1988
    Argentina 1989
    Argentina 1990
    Argentina 1991 1.1
    Argentina 1992 2.05
    Argentina 1993 2.42
    Argentina 1994 2.28
    Argentina 1995 4.11
    Argentina 1996 4.63
    Argentina 1997 4.14
    Argentina 1998 4.59
    Argentina 1999 4.78
    Argentina 2000 5.7
    Argentina 2001 9.42
    Argentina 2002 13.99
    Argentina 2003 9.79
    Argentina 2004 7.09
    Argentina 2005 5.37
    Argentina 2006 4.12
    Argentina 2007 3.49
    Argentina 2008 2.97
    Argentina 2009 3.05
    Argentina 2010 2.05
    Argentina 2011 1.53
    Argentina 2012 1.63
    Argentina 2013 1.75
    Argentina 2014 1.71
    Armenia 1985
    Armenia 1986
    Armenia 1987
    Armenia 1988
    Armenia 1989
    Armenia 1990
    Armenia 1991
    Armenia 1992
    Armenia 1993
    Armenia 1994
    Armenia 1995
    Armenia 1996 17.9
    Armenia 1997
    Armenia 1998
    Armenia 1999 16.88
    Armenia 2000
    Armenia 2001 19.33
    Armenia 2002 15.06
    Armenia 2003 11.36
    Armenia 2004 7.93
    Armenia 2005 4.45
    Armenia 2006 3.3
    Armenia 2007 3.57
    Armenia 2008 1.49
    Armenia 2009 1.59
    Armenia 2010 2.54
    Armenia 2011 2.7
    Armenia 2012 1.74
    Armenia 2013 2.44
    Armenia 2014 2.31
    Azerbaijan 1985
    Azerbaijan 1986
    Azerbaijan 1987
    Azerbaijan 1988
    Azerbaijan 1989
    Azerbaijan 1990
    Azerbaijan 1991
    Azerbaijan 1992
    Azerbaijan 1993
    Azerbaijan 1994
    Azerbaijan 1995 7.32
    Azerbaijan 1996
    Azerbaijan 1997
    Azerbaijan 1998
    Azerbaijan 1999
    Azerbaijan 2000
    Azerbaijan 2001 2.71
    Azerbaijan 2002 0
    Azerbaijan 2003 0
    Azerbaijan 2004 0
    Azerbaijan 2005 0
    Azerbaijan 2006
    Azerbaijan 2007
    Azerbaijan 2008 .49
    Azerbaijan 2009
    Azerbaijan 2010
    Azerbaijan 2011
    Azerbaijan 2012
    Azerbaijan 2013
    Azerbaijan 2014


    At first, I just applied the same coding for calculating gdp growth rate. My plan from here was to generate average rate of change based on this calculation.
    So I did that and collapsed the data to make it applicable to cross-country regression. However, due to the nature of my data, only 39 observations survived out of 98 samples.
    But for the sake of my research, I have to save as many observations as possible.
    I think I may not need to go through calculating annual growth rate to average growth rate. Rather, I'm wondering if there is any way to calculate overall rate of change when there is non-consecutive observations.

    I hope my question makes sense...

    Thanks in advance.

    Taylor

  • #2
    I imagine many people would fit a regression to the logarithms of the outcome and work from there. Some of your data e.g. Azerbaijan don't look good for any method, however, and there is no simple fudge for zeros in the data. (I guess that negative values really are impossible.)

    Comment


    • #3
      I wonder whether a time-series analysis (you may use operators such as D., for example) wouldn’t fit fine in this case.
      Best regards,

      Marcos

      Comment


      • #4
        You might be able to calculate a change rate over more than one year and then divide it up into the intervening years. However, when I see data saying 0 for poverty in Argentina and Azerbaijan, I question whether the data are even close to correct.

        Comment


        • #5
          Thank you very much for your responses. What I tried to do here was to generate a panel data to replicate a work from the article I read about poverty and governance. My dependent variable would be poverty headcount ratio and independent & control variables include Rule of Law, Voice & Accountability indicators, etc (all from World Governance Indicators) as well as trade openness, etc. I was just having some problems with the poverty data so I described the data as time-series. As suggested by some of the respondents here, I also found the data quite problematic. However, it is the data directly downloaded from World Development Indicators operated by the World Bank..

          I will look into the data more and will post with more understanding of the data.

          Thank you!

          Comment

          Working...
          X