Page 165 - Special Topic Session (STS) - Volume 3

STS538 Ibrahim S Y.
ii. Data processing
The processing of cost-of-living data is conducted through an integrated
data management system. The main issue here is that of the treatment of
outliers, and missing or erroneous data. The small samples of observations
that characterize the datasets from many locations do not comply with the
usual assumptions underlying the application of the standard statistical criteria
for the identification and exclusion of outliers. To address this problem, the
ICSC’s standard operating procedures provide for several layers of analysis and
quality control checks, in which the analyst is granted discretion to apply
experience accumulated from processing price data, as well as to consider
various factors, both quantitative and qualitative, to determine whether, or
not, a given price should be considered an outlier. In any case, the results of
data processing are subjected to a multi-stage peer-reviewed quality control
process before they are published.

3. Methodological challenges
The methodological challenges are related to the compilation of a cost-of-
living index that reflects a trade-off between pure statistical measurement and
compensation policy considerations. They include the determination of
expenditure weights needed for compilation of the index, as well as the
specification of the index formula and aggregation. Essentially, the PAI is
based on a system of bilateral comparisons between each duty station and
New York. Yansaneh and Pagan (2011) provides a brief overview of the
methodology for compilation of the PAI. More details on the PAI
methodology, context, governance, and institutional operational environment,
can be found in ICSC (2018). Essentially, the PAI consists of the following five
major components:
(i) In-area, excluding housing (IA-H)- relates to living costs incurred
locally. It has an internal hierarchical structure, in line with the Classification of
Individual Consumption According to Purpose (COICOP). Its index is a
weighted geometric average of the cost-of-living relativities of its basic
headings, which are estimated as an unweighted geometric average of item
price ratios (Jevons), whereas each item ratio is the ratio of unweighted
average prices at the duty station relative to New York (Dutot). Its weight is
obtained as a residual, after subtracting from the reference net remuneration,
the sum of the weights of the other four components; and is pro-rated to all
lower level components, down to the basic heading level, in proportion to a
set of “common expenditure weights” (average expenditure shares for each
component across a subset of locations where about half of eligible staff are
assigned, which are considered as representative of the expenditure patterns
of the average UN staff).

154 | I S I W S C 2 0 1 9