Page 60 - Invited Paper Session (IPS) - Volume 2

IPS179 Wian B.
The range of average transaction values coupled with high heterogeneity in
terms of the transaction volume distribution among EFT initiators lead to the
conclusion that value or volume weighting schemes would not be appropriate
for purposes of monthly aggregation. To therefore obtain the average
transaction value per EFT initiator per month, the harmonic mean of the daily
average transaction value per EFT initiator falling within a calendar month are
calculated (not shown).
The focus of the methodological framework will be on the following price
indices, both in chain-linked and unchained format: (1) Laspeyres, (2) Paasche,
(3) Fisher, and (4) Törnqvist. The Laspeyres based salary EFT price index
consisting of EFT initiators can be written as follows (SNA 2008):

0 0
∑ ( 0 ) ∑

=1
0

0

= ∑ ( 0 ) = ∑ 0 ≡ ∑ =1

0

0 0
=1 =1 =1
where represents average transaction value per EFT initiator per month in
the base period while represents the total number of transactions. To
0
remain consistent with the literature, the price variable is displayed as
instead of . The variable represents the base period proportions. The
0
corresponding Paasche price index can be expressed as:
−1 −1 ∑
0

= [∑ ( 0 ) ] ≡ ∑ =1

0 0
=1 =1

The Fisher index is calculated as the geometric mean of and , or

= √ . The Törnqvist index is calculated as the geometric average of

the price changes, with a weighting scheme equal to the average value shares
0

( + )/2

in the two periods, or = ∏ ( ) .

=1

0

A case can be made that the chain-linked versions of the above discussed
index formulas would under certain circumstances be more appropriate for
purposes of developing a salary EFT index. Goodridge (2007) states, for
instance, that chain-linking allows for more frequent changes to the basket
composition, while current period information can be compared with the
previous period instead of some arbitrary base period. Chained versions of the
Laspeyres, Paasche, Fisher and Törnqvist indices will therefore be tested on
the current salary EFT sample.

Outlier detection will be done on a quantile basis. The levels associated with
minimum wages for domestic workers and an approximate threshold for high
income earners based on the 2017 Tax Statistics will serve to guide the
47 | I S I W S C 2 0 1 9