# How to handle change in base year of index?

I have monthly IIP data from 1996-2016. But the issue is the change of base year. For the year 1996- 2011 the base year is 1993-94 and the index is disconnected from Feb 2011. For the year 2005-2016 the base year is 2004-2005.

I intend to use the IIP monthly index from April 1996 - March 2016 but as you can see their exists two set of different index. Can you please help suggest some technique/method so that the entire index is on the same base year of 2004-2005?

For your reference I have attached the data.

Edit: see the data and the necessary calculation here.

• 1) I don't see any data 2) This question would be more useful to others if you spelled out what IIP was in the first use 3) Is it really the case that the 96-11 data uses as a base year values that are not in the data (93-94)?
– BKay
Mar 27 '17 at 12:17
• The IIP numbers are for India and i have attached the excel sheet for your reference. Mar 28 '17 at 17:31

You need to rebase the oldest series, using the base from the new series. This is an example of how to do this, with annual data (same for monthly data):

So, basically, you need to find a period in which both series have an entry, compute the ratio between them, and then transform the old series using the ratio just calculated. This is how you link two series in different base years.

Remarks: you will have several months in which the series have values. If you have reasons to believe that the newest series is better, then I would make sense to take the earliest month in which the two have data (probably January of 2005). Nonetheless, the two series might be in fact different (e.g. sector composition or definitions vary). Compare the rate of monthly growth of both indexes for the months in which both have data (probably January 2005 to Feb 2011). Are they very different? If so, you might try to adjust the old series. Perhaps, there is a persistent difference, which you can account for when expanding the newest series backwards. See more in this document.

• Thanks a lot for your reply. I have computed the rebase. If possible can you please have a look at my calculations and methodology. Please find the link below. LINK :- spaces.hightail.com/space/WRlEc Mar 28 '17 at 17:32
• @SOUMYAJITRAY It looks fine. An alternative is to use the most of the newest series. So, just make the connection in April 2005 (row 116). Use the value in D116 to adjust earlier periods. Comparing the two series might be interesting. Mar 29 '17 at 10:29

An alternative procedure would be to extrapolate the data for the most recent time backwards at the growth rate of the first set (earlier periods). By doing this, you do not have to rebase, adjust or splice the two sets. You will have the base year of the data for the most recent years (2004-5) and also will preserve the trend growth of the data for the earlier periods.