Page 200 - Contributed Paper Session (CPS) - Volume 1

CPS1278 Desiree R. et al.
together with the updating of the list of establishments, commenced on the
16th of April 2018.
The scope of the CPBI is confined only to all existing establishments and
enterprises in the formal sector which are engaged in economic activities, as
classified in the 2009 Philippine Standard Industrial Classification (PSIC). The
rationale for only including the formal sector was based on the share of the
group to the value added on the 18 major sections of the Philippine Economy
covered by the CPBI.
The results using the current methodology produced high values of CVs at
the establishment level. In addition, the distribution of the target variables
such as the total employment from the last CPBI, is highly skewed. Hence it is
important to develop a sampling methodology which would be able to
produce more precise and accurate estimates as well to address the skewness
of the data.

2. Methodology
In order to efficiently sample highly skewed population, there is a need to
stratify the population into a take-all stratum and a number of take-some
strata1. This method was adopted in a business survey where a generalized
stratification using LH was proposed2. The proposed methodology accounts
for the difference between the discrepancy of the stratification and survey
variables. The results of his study showed that using the generalized LH
algorithm, the CV of the estimators were close to the target CVs. In relation to
this a study was conducted which discussed two approaches in estimating
sample size for survey of enterprises in the Philippines3. One of the methods
was used is the LH method. In this study, it was mentioned that the CV’s of the
samples obtained using the LH at the domain level is expected to be small as
these are based on the aggregation of the primary strata for which the CV’s
were controlled. It also added that the main advantage of using LH is that it
determines the boundaries as well as the sample size for each stratum.
Aside from stratification of the population, it is important to address the
establishments which contribute very small amount to the total of the target
variables. An example was the study which used a “cut-off” sampling in order
to create a framework for business survey4. This method eliminates the
sampling units whose contribution is very small to the total of the target
variables.
The previous sampling design of the CPBI involves two method of
sampling, one for the establishment and another for the enterprise level,
respectively. For the proposed methodology, the sampling design would
adapt the cut-off sampling and generalized LH algorithm at the establishment
level only.

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