Announcement

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

  • Inflation Adjustment and Industry dummies

    Hi!
    In my panel data analysis I am using research and development expenditures (R&D) of firms as an independent variable and I have to calculate inflation adjusted R&D. I'm not sure about whether to use starting year as reference or years on rolling basis. My dataset is unbalanced panel data (1975-2016) and I'm using fixed effect model. If anyone can help me with a code to calculate inflation adjusted R&D expenditure, that would be helpful.

    Secondly, do I need to add industry dummies in the model as someone suggested me. As far as I know, in fixed effect model we don't need to add any cross section related factor which is not variable over time and industry of a company remains fixed over time.

    Code:
    * Example generated by -dataex-. To install: ssc install dataex
    clear
    input int(permno fiscalyear) float(rd inflation)
    10006 1975 3.235  9.13193
    10006 1976 3.786 5.737027
    10006 1977 3.858  6.48646
    10006 1978 4.702 7.647464
    10006 1979  5.04 11.26604
    10006 1980 6.565 13.50937
    10006 1981 8.234 10.31553
    10006 1982 7.661 6.160616
    10006 1983 6.792 3.212435
    10007 1989  .505 4.827003
    10010 1986  .806 1.858736
    10010 1987  .833 3.740876
    10010 1988  .823 4.009088
    10010 1989  .863 4.827003
    10010 1990  .967 5.397956
    10010 1991 1.942 4.234964
    10010 1992 2.469  3.02882
    10010 1993 4.791 2.951657
    10010 1994 4.783 2.607442
    10012 1987  .349 3.740876
    10012 1988  .384 4.009088
    10012 1989  .264 4.827003
    10012 1990  .206 5.397956
    10012 1991   .23 4.234964
    10012 1992  .166  3.02882
    end
    Note: Inflation is in percentage

  • #2
    Shafaq:
    most nationalt statistics for inflation calculate the yearly inflation rate using an index year as a reference (ie, year 2010=100). This detail is not reported in your post.
    I would take a look at the inflation statistics for the country (countries) you're interested in.
    As far as your second question is concerned, the fixed effects machinery gets a rid of any time-invariant predictor. Hence, if industry does not change across years, its coefficient will not be calculated under fixed effects specification.
    Kind regards,
    Carlo
    (Stata 18.0 SE)

    Comment


    • #3
      Dear Carlo, this is US data and I have extracted inflation rates from world bank database

      Comment


      • #4
        Shafaq:
        the (web)source should contain notes/files explaining the way inflation rates were calculated.
        Kind regards,
        Carlo
        (Stata 18.0 SE)

        Comment


        • #5
          Dear Carlo, there are two different inflation rates given in world bank data base. One that I have taken is mentioned as [Inflation, Consumer Prices], while there is no reference given here but in Consumer Price Index it's mentioned 2010 = 100. There is another inflation rate given mentioned as [Inflation, GDP deflator] can I use this deflator directly? If yes, do I just have to divide it by R&D expense?

          Comment


          • #6
            Shafaq:
            I would go with the Inflation, Consumer Prices (which is simpler to apply).
            Kind regards,
            Carlo
            (Stata 18.0 SE)

            Comment


            • #7
              By the way, I don't think you specified what your panel variable is. Carlo assumed it is firm which is almost certainly the best option. And, since firms don't change industry much if at all, firm fixed effects include all stable industry effects.

              Comment


              • #8
                Yes Phil, panels are based on firms. So, If I apply year fixed effect, then I will need to add industry effect?
                One thing more, Do I need to make inflation adjustment if I am taking log value of R&D?

                Comment


                • #9
                  Shafaq:
                  you may want to try:
                  Code:
                  xtset firms year
                  xtreg ln_R&D_exp i.firms i.year <other predictors>
                  Then, you can test -i.firms- and -i.year- via -testparm-.

                  Please note that some variables may be omitted due to collinearity with fixed effects.
                  Kind regards,
                  Carlo
                  (Stata 18.0 SE)

                  Comment


                  • #10
                    Shafaq:
                    yes, you should inflate -R&D- before logging it.
                    Kind regards,
                    Carlo
                    (Stata 18.0 SE)

                    Comment


                    • #11
                      Thanks Carlo.

                      Comment

                      Working...
                      X