Transcript Estimation methods of the informal sector

Estimation methods of the informal sector:
strengths and weaknesses
Aloke Kar and Giovanni Savio, UN-ESCWA
Expert Group Meeting on National Accounts
Cairo, 12-14 May 2009
Presentation Plan
1.
2.
3.
4.
Brief background
Methods for measuring the informal sector
Data needs and strategies for data collection
Pros and cons
Brief background




ILO’s Report of a Comprehensive Employment
Mission to Kenya (1972) used the term “informal
sector”
Since then the term has gained acceptance in
international official documents
Initially, labour statisticians considered “informal
sector” as composed of urban “working poor”
migrated from rural areas in search of work
Later, recognised as an important employment-,
production- and income- generating sector present
as much as in rural areas as in urban areas
Brief background




Though the term “informal sector” is of relatively
recent origin, the idea of such a segment of
economy has existed since long
Particularly, for ensuring complete coverage of the
national accounts statistics
The official statisticians have referred to this
segment – or a closely comparable one – by
various names like “unregistered”, “unorganised”
and “unrecorded” segment of the economy
The coverage of the segment of the economy
referred to by these terms were indeed varied
Brief background



The 15th ICLS (1993) adopted a resolution
relating to statistics of employment in the
informal sector
This provided an international statistical
standard definition of informal sector
The SNA 1993 identified “informal sector” as
one of the segments of economy that are not
covered or not adequately covered in the
national accounts
Brief background

In measuring the activities within 1993 SNA
production boundary exhaustively, the 5 problem
areas that are most likely to be non-observed are
(OECD (2002)):






Underground
Illegal
Informal sector
Household production for self consumption
Deficiencies of basic data collection programme
Clearly, the NOE problem areas are not mutually
exclusive but, the informal sector is the most
important component in many countries, especially
in developing countries
Methods of measuring

Two broad groups of methods:




Macro-model methods
National accounts methods to achieve
exhaustiveness
Both these groups of methods attempt to measure
NOE in general, of which informal sector is but a
component – albeit the major component
Macro-model methods produce an estimate of the
entire NOE by means of a single model
Methods of measuring

OECD (2002) – “… not considered useful in
obtaining exhaustive estimates of GDP or NOE
…”, “… and tend to produce spectacularly high
measures …”

Three broad types of macro-model methods:



Monetary methods
Global Indicator methods
Latent Variable methods
Monetary methods





These rely on the assumption
unexplained (by the model) = (proportional to) NOE
Start with Fisher’s quantity theory equation (MV=PT)
Obtain an estimate of model-based GDP
Difference between the model-based estimate and
official estimate is considered a measure or indicator
of NOE with respect to the benchmark
Three variants:

Transaction method (the one above, discussed in
Feige, 1979)

Cash/deposit ratio method (Guttman, 1977)

Cash demand method (Tanzi, 1982)
Global Indicator methods


This method uses a single variable, like
electricity consumption (Kaufmann and
Kaliberda (1996), as the indicator of entire
economic activity
The difference between the estimate of GDP
obtained under the model and the official GDP
is claimed to represent a measure of NOE
Latent Variable methods

These are multiple regression models with a nonobserved dependent variable (the NOE) and a number
of observed explanatory variables for countries and
years, such as:
Actual tax burden and its perception, unemployment
rate, the regulation burden, the attitude towards paying
taxes (tax morality), the per capita available income,
the labour force participation rate of the male
population, the number of weekly working hours and
the growth of the GNP (Frey and Weck, 1983)
Indirect methods

Indirect methods of covering NOE in the GDP
estimate:
–
–
–
–

Supply based approaches, including labour input
method
Demand based approaches
Income based approaches
Commodity flow approaches
Labour Input Method (LIM) is perhaps the most
important procedure to measure contribution of
informal sector to GDP
LIM method

The LIM of estimating value added / output for
the informal segment (for an economic activity
or a group of economic activities) consists of:
1. obtaining an exhaustive estimate of labour input
from various sources (LFS, Population Census and
other, i.e. demographic and administrative, sources
including Establishment Surveys: demand/supply
comparison);
2. obtaining estimates – after corrections, i.e. using the
Franz approach - of output or value added per unit
of labour input from Establishment Survey; and
3. multiplying the estimate of labour input by the
estimate of per unit value added / output to arrive at
an aggregate estimate of value added / output
LIM method

Measuring the value of production of goods
and services by LIM, therefore, demand a fair
degree of precision in the activity group
estimates of:
–
–
Labour input (i.e. number of workers adjusted for
multiple employment) based on data from
households (LFS) and other sources
Gross value added per worker (GVAPW) from
enterprises obtained from enterprise survey
Data Needs


The requirements of estimates on informal
economy can be summarised as follows:
For Informal Employment:
–
–
–
–

parameters defining informal employment (LFS)
terms & conditions of employment (LFS)
structural information (ES)
productivity (ES)
For Informal Sector:
–
–
parameters defining informal sector enterprise (ES)
production related parameters (ES), including
labour input, output, intermediate consumption and
GVA
Data Collection Strategy


For measurement of informal employment and
informal sector employment the LFS is the
preferred instrument
The main options of colleting data on
production of informal sector are:
1. List-frame based establishment survey
2. Area-frame based establishment survey
3. Area-frame based mixed household and enterprise
surveys
4. Area-frame based “1-2” surveys
Data Collection Strategy
Universe of Enterprises
List-frame segment
Large
units
Public
sector
Small
units
Area-based frame
segment
With fixed
premises
Without
fixed
premises
Private
sector
In business register
Gray boxes: contain HUEM
units
Not in business register
Data Collection Strategy

List-frame based establishment survey
–
–
–
–
business registers does not usually cover the
informal sector units
list-frame has to be developed from a general
economic/establishment census
but, these censuses do not ensure full coverage
thus, list-frame based establishment surveys can
not ensure complete coverage of informal sector
Data Collection Strategy




Thus, area-frame based surveys are essential
for complete coverage of informal sector
However, the conventional area-frame based
establishment surveys suffer from the same
limitation as the Establishment Census
In these surveys, a list frame of establishments
is developed for each selected area unit by
door-to-door enumeration
This procedure is prone to omission of
activities carried out inside the owner's home
Data Collection Strategy

The choice of method is, therefore, practically
restricted to the last two area-frame surveys,
viz.
–
–


Mixed household enterprise surveys and
Integrated “1-2” like surveys (1-2-3 survey)
Both these methods use a multi-stage (usually
two-stage) sampling scheme
A sample of area units are selected as the first
stage unit (fsu) in both the methods
Mixed Household Enterprise Surveys



In the first phase (the household survey
component), a sampling frame for informal sector
enterprises or, more generally, small enterprises is
obtained through a household listing or survey in
the selected sample areas (primary sampling
units)
Data often have to be obtained from household
members other than the enterprise owners (proxy
respondents)
In the second phase (the enterprise survey
component), a sample (or all) of the enterprise
owners are interviewed to obtain detailed
information about them, their enterprises, and their
employees (if any)
Mixed Household Enterprise Surveys

Variously described in literature. As per the (Draft)
Manual on Surveys of Informal Employment and
Informal Sector, its sampling frame at the second
stage consists of the following:
i.
ii.
all identifiable establishments outside the owners’ home
located in the selected area unit;
household-based enterprises located within home; and
iii. the units without any fixed premises of operation
are listed by a structure-to-structure visit

The units of last two categories are listed against
and interviewed in the owners’ households
Mixed Household Enterprise Surveys

Units excluded from the survey coverage are not
listed, e.g.
–
–



agricultural or formal sector establishments and
units covered in a list frame based survey
Within-scope units without fixed premises of within
owner’s home are identified through additional
questions to households during listing; and
are listed against the household where the proprietor
(or a partner of a partnership concern) resides
This way, all establishments within the scope of the
survey in the fsu are included in the list
Integrated “1-2 Survey”

This approach consists of two phases:
–
–



First phase: a household survey (LFS)
Second phase: an enterprise survey
The first phase used also for constructing the
sampling frame for the enterprise survey
From the sample households in the first phase,
the within-scope enterprises owned by the
households are identified
In the 2nd phase, a sample of within-scope
enterprises that are owned by the households
is drawn for the enterprise survey
Integrated “1-2 Survey”

The within-scope enterprises selected for
survey may either be:
–
–
–

Within the fsu without fixed premises, or
Within the fsu with fixed premises
Outside the fsu
In all these cases, the enterprise is surveyed
Integrated “1-2 Survey”
Households owning/
operating enterprises
Within the sample
area
Outside the sample area
and with fixed premises
Capture at premise
With fixed
premises
Capture at premise
Without fixed
premises
Capture at household
Pros and Cons
Mixed Hhd-Entp.Survey
“1-2 Survey”
• May be conducted
independent of other
surveys
• Involves extra costs for
enterprise listing
• Does not cover owner
households within the s.a.
with premises outside s.a.
• Is likely to have reduced
sampling errors in both
stages
• Provides data on informal
employment & informal
sector
• Estimates of informal
sector employment from
2 sources likely to be
consistent
• Involves extra costs for
travel – for surveying
enterprises outside the
fsu
Pros and Cons
Type of
premises
Location of owner
household
Location of
production unit
With fixed
premises
With non-fixed
premises
Inside the sample area
Outside the sample
area
NA
Owner
households
within the
sample area
Owner
households
outside the
sample area
MHES, 1-2
MHES
1-2
MHES, 1-2
Pros and Cons



In a stage sampling, efficiency of an estimate
depends on how well the inclusion probability
of the fsu’s (pa) are correlated to the fsu-level
value of the parameter, e.g. sum of GVAs of all
within-scope enterprises of the fsu
In a mixed household enterprise survey, this
can be attempted by using EC data
In an “1-2” survey, the choice of size variable
for fsu selection is based on distribution of
population and not on that of within-scope
enterprises
Pros and Cons


The establishments owned by non-residents
can not be captured by the “1-2 Survey”
approach – whether with fixed premises or not
The results of a survey conducted with “1-2
Survey” might be subject to higher sampling
error than MHES – at both the stages
–
–
The first-stage variance is higher, since the size
variable for fsu selection is based on distribution of
population and not on that of within-scope
enterprises, and …
… the second-stage variance is higher, since
establishments are not selected from a complete list
of establishments