Philippine Statistics Authority
Other Opendata Websites

Metadata for National Agricultural Statistics of the Philippines

# Chapter 2. Major Domains and Selected Indicators of Agricultural Statistics

### 2.2.1.3a Data Processing, Estimation and Revision Methodology

#### Volume of Palay and Corn Production

Data from the Palay and Corn Production Survey (PCPS) are processed using two (2)-customized Windows-based systems developed by the Information and Communications Technology Division (ICTD). The two (2) customized systems, Palay Production System (PPS) for palay and Corn Production Survey (CPS) for corn, were developed using the software Census and Survey Processing System version 4.1 (CSPro 4.1).

Decentralized processing is applied for both the PPS and CPS. At the Provincial Operations Center (POC), processing activities include encoding of data from survey questionnaires; computerized editing, completeness check, generation of expansion factor and generation of output tables. These procedures are the same for both PPS and CPS.

Prior to data encoding, the accomplished survey returns are manually edited and coded. Manual editing involves the checking of data items based on pre-set criteria, data ranges, completeness and consistency with other data items. Coding is the assignment of alpha-numeric codes to questionnaire items to facilitate data entry.

To validate, encoded data are subjected to computerized editing using a customized editing program. The editing program takes into consideration the validation criteria such as validity, completeness and consistency with other data items. This activity is done to capture invalid entries that are overlooked during manual editing.

An error listing is produced as output of the process. The errors reflected in said lists will be verified vis-à-vis the questionnaires. The data file will be updated based on the corrections made. Editing and updating are performed iteratively until a clean, error-free data file is generated.

Completeness check is done to compare the data file against a master file of barangays to check if the sample barangays have been completely surveyed or not. This activity is done after a clean, error-free data file is generated.

A program generating the appropriate household weights or correction factor is run using the clean data file. The generated household weights will then be used in the estimation.

Output table generation is performed only after the activities of completeness check and generation of correction factor have been done. The PPS and CPS systems generate 12 provincial output tables. Soft copies of provincial data, specifically the clean data and the barangay master file, are submitted to the ICTD for national consolidation while hard copies of the provincial reports are submitted to the Crops Statistics Division (CSD).

#### Estimation and/or Compilation Procedure

Each replicate (represented by the sample psu) in a stratum will yield an independent estimate for the stratum. Hence, there will be four (4) independent estimates and the mean of these four (4) estimates will be the unbiased estimate for the stratum.

${x}_{khi}^{{}^{\prime }}=\left(\frac{{P}_{kh}}{{P}_{khi}}\right)\cdot \left(\frac{{N}_{khi}}{{n}_{khi}}\right)\cdot {W}_{khi}\cdot \sum _{j-1}^{{n}_{khi}}{x}_{khij}$
$={b}_{h}\cdot {R}_{k}\cdot {x}_{hi}$
$=4\cdot {R}_{k}\cdot {x}_{hi}$

where:

xkhij= value obtained from th jth sample farm household of the jth baranngay in the hth stratum of the kth province;

xhi= weighted total for the jth barangay in the hth stratum;

$={W}_{khi}\cdot \sum _{j-1}^{{n}_{khi}}{x}_{khij}$

and

wkhi,nkhi,Pkh and bh are the ones defined in the sampling design

The unbiased estimate of total for the hth stratum is simply the mean of the four (4) independent estimates, that is,

${x}_{kh}^{,}=\left(\frac{1}{{b}_{h}}\right)\cdot \sum _{i-1}^{{b}_{k}}{x}_{khi}^{,}$
$={R}_{k}\cdot {x}_{h}$

where xh is the weighted total for the hth stratum.

#### a. Provincial Estimates

Estimates of total for the province are obtained simply by aggregating all the stratum estimates in the province. Hence, the estimate of total for the kth province is given by

${x}_{k}^{,}=\sum _{h-1}^{{H}_{k}}{x}_{kh}^{,}$

where Hk is the total number of strata in the kth province (domain) and its variance is estimated by the sum of stratum variances, that is

$v\left({x}_{k}^{,}\right)=\sum _{h-1}^{{H}_{k}}v\left({x}_{kh}^{,}\right)$

#### b. Regional and National Estimates

Estimates of total for the region and for the whole country, together with their respective variances, are obtained in the same manner as those for the province, that is, by aggregating relevant stratum estimates. These may also be obtained by aggregating relevant provincial estimates (for the region) and aggregating relevant provincial estimates (for the whole country).

#### Revision of Estimates

The BAS has adopted a policy on revision of estimates approved under the NSCB Resolution No.7 dated May 18, 2005. It basically informs producers and users of agricultural statistics generated by the BAS that revision of quarterly estimates on the agricultural production, prices and related statistics be limited to the immediately preceding quarter and for the past three (3) years with quarterly breakdown to be done only during May of the current year. This happens when additional statistics and/or indicators are made available to support the change in the original data.

#### Estimation and/or Compilation Procedure

Percent change for each type of farm for any given commodity (crops other than palay and corn) is computed as follows:

For small farms,

$\mathrm{%}chang{e}_{s}=\left(\frac{\sum _{i-1}^{n}{P}_{sci}-\sum _{i-1}^{n}{P}_{spi}}{\sum _{i-1}^{n}{P}_{spi}}\right)\ast 100\mathrm{%}$
where:

%changes - percent change for small farms

Psci - actual level of production /area/number of bearing trees for the current period reported by the jth sample farmer

Pspi - actual level of production/area/number of bearing trees for the same period last year reported by the jth sample farmer

n - number of sample farmers

For large farms / plantations,

$\mathrm{%}chang{e}_{i}=\left(\frac{\sum _{i-1}^{m}{P}_{lci}-\sum _{i-1}^{m}{P}_{lpi}}{\sum _{i-1}^{m}{P}_{lpi}}\right)\ast 100\mathrm{%}$
where:

%changei - percent change for large farms/plantations

Plci - actual level of production/area/number of bearing trees of the jth sample large farm/plantation for the current period

Plpi - actual level of production/area/number of bearing trees of the jth sample large farm/plantation for the same period last year

m - number of sample large farms/plantations

Each type of farm has a corresponding weight which is determined as follows:

For large farms / plantations,

${W}_{l}=\frac{{A}_{i}}{{A}_{p}}$
where:

Wl - weight for large farms/plantations

Ai - total area of large farms in the province

Ap - total area of small farms in the province

For small farms,

${W}_{s}=\frac{{A}_{s}}{{A}_{p}}$
where:

Ws - weight for small farms

As - total area of small farms in the province

Ap - total area of small and large farms in the province

The overall percent change for the province is computed as the sum of the weighted percent change for each type of farm, that is,

$\mathrm{%}change=\left({w}_{s}\ast {\mathrm{%}change}_{s}\right)+\left({w}_{l}\ast {\mathrm{%}change}_{l}\right)$

The estimated total production / area / number of bearing trees for the current period for the province, denoted  by Ec, is computed as

${E}_{c}={E}_{p}\left(1+\frac{\mathrm{%}change}{100\mathrm{%}}\right)$
where,

Ep - estimated total production/area/number of bearing trees for the same period last year, i.e. the base data

%change - overall percent change for the province

Estimates of total production / area / number of bearing trees for the region are obtained by aggregating the estimated total production / area / number of bearing trees of the provinces within the region.  Estimates at the national level are the sum of the estimates of the regions.

#### Data Processing

The existing Crops Compilation System (CCS) adopted by BAS is in MS Excelbased template that utilizes the links and protection commands. The system electronically consolidates the different data sets from the provinces to the region up to the national level. An identical and independent system is provided for each of the two sub-commodity groups which are classified further into major and other crops so that three data files were created to accommodate these crops. The major crops are contain in two data files, one for production and the other file for area and bearing trees. A third data file which is identical for both sub-commodity groups was provided and is for other crops. Production, area and number of bearing trees were lumped in this data file. All the worksheets provided are protected except in cells for the reference periods covered in each reporting period. Updating the links and formula are limited in these reference periods. Thus, the need to use the updated file intended for the specific reporting period. Otherwise, the data in the un-updated files could not be captured which could result to underestimation of the regional and national totals.

The POC files comprise several worksheets by reference period and another worksheet consolidating the data in all reference periods by crop. The data series in the system is from 1999 to the most current period. For the current year, a column is provided for the preliminary estimates and another column for the final estimates.

Preliminary estimates are reported in the current reference period and final data for the previous reference period. POCs encode the provincial estimates in the Crops Compilation System, which electronically computes the totals and the year-on-year percent changes of the reference period.

The compiling system also electronically computes the planting density and yield by crop. This shall facilitate the checking of the computed levels against the parameters by crop. Hard copies of the corresponding worksheets are simultaneously submitted to the ROC and the Central Office as advanced copies and ready reference. Aside from the hard copy, the POCs provide the ROCs with the soft copy to electronically generate the regional total.

The ROC compiling system also comes in two files. The files have worksheets corresponding to the POCs’ files. Additional worksheets are provided as summary worksheets by reference period showing the data by province to facilitate comparison and summarization of reasons. Another summary worksheet presents the electronically computed planting density and yield by crop showing the data by province. No encoding is done at the ROC except for the summary of the reasons for changes. Other crops include production and area in one file as a result of the validation. The regional total and the corresponding percent change for all crops and data items are electronically computed using the links command and formulae. The ROC takes note of the problematic entries and observations on the submitted reports. These are referred back to the concerned POC for appropriate action prior to submission to the Central Office.

The central compiling system is maintained at the Central Office. The data submitted in soft copy are copied on the corresponding worksheets of the central compiling system. This ensures that any unauthorized data adjustment in the submitted files would not corrupt the main file. Otherwise, the data in the main file would become inconsistent to released data. Pasting the data on the main file is done in several workstations and accessed through the Local Area Network (LAN) facility.

The system electronically generates the national level data. As an added measure and for quick reference, upon the release of data, summary tables are generated deleting the links. As done at the ROC, the problematic entries and observations on the provincial and regional reports are referred back to the concerned reporting unit for appropriate action.

Apart from the provincial and regional files, the Central Office also maintains commodity files. Separate files are maintained for the data series on production, area and bearing trees by crop. From these files, data on planting density and yield by commodity are generated for use in the data review.

For crops covered by specialized agencies of the government, the scheme varies. For sugarcane, the data for centrifugal sugar in ton canes are obtained from the Sugar Regulatory Administration (SRA). These are from the reports of sugar mills operating in the country. The BAS Operations Centers collect data on production of canes for chewing, basi and muscovado through the quarterly Crops Production Survey. These two data sets are incorporated to account for the production of sugarcane.

In the case of fiber crops, the national total is the summary comprising the data from both the Fiber Industry Development Authority (FIDA) and BAS. Data of FIDA are from the baling stations while the data of provinces with no baling stations are derived from the Crops Production Survey of the BAS.

For cotton, the Cotton Development Administration (CODA) provides the Bureau with data from their monitoring system. This includes the 10 CODA monitored provinces which are also covered in the BAS Quarterly Crop Production Survey. Meanwhile, for coconut, the data is a product of the reconciled data of the Quarterly Coconut Production Survey (QCPS), a joint undertaking of the Philippine Coconut Authority (PCA) and BAS, and the Crops Production Survey of BAS.

For tobacco, data is obtained from the Quarterly Crop Production Survey of the BAS. In the review and analysis of data, the National Tobacco Administration (NTA) is consulted and it also provides auxiliary information, which serves as inputs for data checking.

#### Data Processing

Similarly with the PCPS, the processing of returns of the Palay and Corn Stock Survey 1 (PCSS1) is decentralized. In the Operations Centers, this processing is a combination of manual and Excel-based processing systems developed at the Cereals Statistics Section. The provincial estimates are summarized using the prescribed format and forwarded to the Central Office for review and consolidation.

#### Estimation and/or Compilation Procedure

The provincial estimate is given as follows:

where:
${\stackrel{^}{Y}}_{p}$
- estimated total stock in the province for jth type of cereal during the specified period

Yijk - observation from the kth sample in the jth catergory (farming or non-farming) for the jth type of cereal

nj - number of responding households samples for in the jth category

Nj - total number of households in the jth category

The estimate of the total stock in the province is simply obtained by:

1. for each type of cereal, multiply the average stock held by the reporting households by the total number of households under the farming and  non-farming category; and then
2. for each type of cereal, add the estimated stock held by the farming and non-farming households.

The expansion factor of the PCSS 1 is based on the 1991 CAF number of farming and non-farming households which is updated in July every year using the projected mid-year population estimate.

The rice and corn grain equivalents of the estimated palay and corn stock are computed using the formula below:

$riceequivalent=estimatedpalaystock\ast 0.65$