Page 62 - Invited Paper Session (IPS) - Volume 2

IPS179 Wian B.
appears to resolve the problem of chain drift and does not differ much from
the fixed base indices in Figure 2(a), although the methodology is statistically
more robust.

Figure 3: PLS Adjusted Chain-Linked Salary EFT Price Indices (July 2017 = 100)

6. Conclusions and Recommendations
Based on the limited results in Section 5, it is recommended that an
annually chained index on a monthly frequency based on the weights of the
respective EFT initiators be tested for purposes of reducing the inherent
volatility in salary EFT payments. An appropriate seasonal adjustment
procedure which properly takes into consideration seasonality over multiple
frequencies, the impact of public holidays and weekends, as well as other
relevant regressors, should be applied to the data. The most appropriate
outlier removal method should be based on an empirical study comparing
various methods using the EFT database. A sufficiently long time series would
be required to properly implement the above discussed recommendations.
There are some outstanding research questions. First, the issue of what
constitutes “price” in the context of salary EFT indexing along with the most
appropriate weighting scheme should be given careful consideration, as some
economists may find the current use of average transaction value problematic.
Second, the view that EFT initiators can be represented as a basket of items
exhibiting different price and volume characteristics could be debated on a
conceptual level. Third, it should be considered whether more than one index
should be produced, given that the users of this index will have different needs
and requirements. For example, certain users may be interested in a
remuneration index consisting of daily, weekly as well as monthly salary
payments, while others may be interested in a subset of payments.

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