UNCTADStat produces Commodity Price Index going back decades
Goal: track impact of commodity price changes on developing
countries>>weights are developing countries’ exports
Prices are collected specifically for this, so (most of them) are
also published
The Problem
Involves prices collected from both automatic (easy!) and manual
sources (PDFs, emails: time wasting and mistakes)
Australian Wool Innovation, Cotton Outlook, International Rubber
Study Group
Discussion with DITC@SCC: the raw prices are not widely used, but the
UNCTAD-branded CPI has institutional value so should be maintained
The Approach
Many of these prices are collected by IMF or World Bank anyway, so
let’s collect directly from them
Reduced the number of sources from 18 to 3: World Bank, IMF and
FAO.
Estimation approach: when these sources don’t go far back enough,
the previously-collected data are used.
If new series are missing in latest months, the growth rate of the
rest of the group is used
The Results
Most prices agree strongly. Even when different values (quality,
markets, units), growth rates are often similar
Despite 47 input prices, top five products (crude oil, natural gas,
gold, copper, soybeans) make up 76% of the index; top 15 products make
up 90%. Most of these changes very little
On advice of DITC we left the old index published in parallel
Slide with Plot
The Impact
Index is virtually unchanged
Faster publicatrion in the future (e.g. Jan can be published
end-February instead of ~10th March)
Reduced staff time from ~1 full day a month to ~1 hour (to be
confirmed…)
Reduced subscription fees
The Lessons
Positive engagement between STAT and Divisions at SCC
Strong interns enabled this update
Perfection versus pragmatism
Strategy: focus on what matters to our work. Let others be price
disseminators; we can better add value with (just) the index?