You are on page 1of 16

KENYA: THE COVERAGE OF OFFICIAL

LABOUR STATISTICS

by

John Thinguri Mukui

Background report prepared for Development and Employment in


Kenya: A Strategy for the Transformation of the Economy, Report of
the Presidential Committee on Employment, Nairobi, Kenya (January
1991)
13 December 1990

KENYA: THE COVERAGE OF OFFICIAL LABOUR STATISTICS1


INTRODUCTION

1.
The Central Bureau of Statistics (CBS) is the Government department responsible for collection,
processing, analyzing, and disseminating statistics in Kenya. Among data collecting activities, it collects
employment and earnings data on annual basis through establishment-based postal surveys, and on ad
hoc basis through household-based personal interviews.
2.
There are other Government and nongovernmental institutions involved with collection or
culling of employment and earnings data in the country. Among them are the line ministries such the
ministries of Labour and Education. The Registrar of Companies, a unit within Attorney Generals
Chambers, keeps files for all registered companies. Information contained in the files is used to update
the frame CBS uses for collecting employment data through establishment-based surveys. Other
institutions compiling data on employment and earnings include the Federation of Kenya Employers
(FKE), Central Organization of Trade Unions (COTU), educational institutions, and researchers.
International institutions such as the International Labour Organization (ILO) and the World Bank also
cull employment and earnings data from primary sources in compiling country profiles.
2.
The Central Bureau of Statistics collects data on employment and earnings through the Annual
Enumeration of Employees and Self-employed Persons, Survey of Domestic Servants, and annual
sectoral surveys e.g. Survey of Industrial Production, Business Expectation Enquiry, Transport Statistics,
Hotel Statistics, etc. In addition there are ad hoc surveys that generate employment and earnings data,
among other variables the surveys are designed to monitor. Such surveys are carried out solely by CBS
or by CBS in conjunction with other ministries. The surveys include population censuses, household
budget surveys, labour force surveys and manpower surveys. The most regular and important source of
employment and earnings data in the modern sector is the Annual Enumeration of Employees and Selfemployed Persons.
3.
The purpose of the data annex is to evaluate the coverage of labour statistics generated by
Government ministries and agencies. The analysis covers the modern sector, small farms, rural nonfarm
and the urban informal sector. While employment and earnings data for the modern sector is collected
through an annual census, data for the other sectors are derived using very restrictive assumptions; or
residually from the knowledge of composition and magnitude of the labour force. This creates special
problems in that errors of measurement and labour force summary statistics (e.g. unemployment rates)
affect the magnitude of estimates for the sectors whose employment is derived residually.
4.
The Annex is divided into four parts. Section I is a rundown of the regular and occasional
censuses and surveys that gather labour data. Section II focuses on official institutions engaged in labour
data collection and analysis, while Section III briefly discusses delays in official release of labour
statistics. Section IV deals with various uses of labour data, narrowing down to issues of size distribution
of wage earnings, the compilation of labour costs component of Gross Domestic Product (GDP), and the
creation of an employment profile for the entire population.
1 The contents of the paper were liberally used in my paper titled Employment Profiles for Kenya, 1994-96,
Background report prepared for the Seventh National Development Plan, 1994-96, Office of the Vice President
and Ministry of Planning and National Development, Nairobi, Kenya, November 1993.

SECTION I: SOURCES OF LABOUR DATA


A:

Annual Enumeration of Employees and Self Employed Persons

5.
This is an establishment-based annual census which gathers data on employment and earnings
in the modern sector. The information, which is disaggregated by geography, gender, occupation,
citizenship and status of employment (permanent or casual), is collected using two different mail
questionnaires each designated as large or small form. The small form is completed by small
establishments defined as those employing twenty (20) or less persons while large form is completed by
large establishments i.e. those engaging more than 20 persons. For the first category, a detailed
questionnaire (FORM LE/9-/L) is administered to the respondents, while the latter uses a brief schedule
(FORM LE/9-/S). In addition, Government departments receive a schedule (LE/9-/C) to capture
information on casual employees.
6.
The questionnaires are sent to the establishments on the basis of the Register of Establishments,
also known as the Master File (MF), which is maintained and periodically updated by the Central
Bureau of Statistics on the basis of records from several sources, mainly the Registrar of Companies in
the Attorney Generals Chambers, the Ministry of Industry, National Social Security Fund,
Commissioner of Insurance, Central Bank of Kenya, physical listing of new establishments by districtbased CBS officers when such establishments are not contained in the district-specific checklists
supplied to the districts, physical listing of new establishments by headquarters staff during listing
exercise (rarely carried out due to financial constraints), professional/business organizations such as the
Kenya Association of Manufacturers, and the media e.g. the Business Directory, Telephone Directory,
newspapers, magazines and periodicals. The lists are counterchecked with the active MF to avoid any
duplication or omission of a record.
7.
The MF takes the establishment rather than the firm as the unit of the register. The Master File
contains private, parastatal and Government establishments and gives postal address, geographical code,
type of business activity, and employment, which is necessary for stratification. Information obtained
from the survey is also used to update the Master File. For example, changes in address, name, and
activity are noted on the return and the information ultimately used to effect necessary changes in the
Master File. Geographical, activity, and occupational codes are assigned to the Master File to facilitate
analysis of employment data by area, industry and occupation. For instance, urban versus rural analysis
is undertaken since data are available by sub-location, which can be aggregated to location, division and
district; while all urban towns, defined as urban centers with a population of 2,000 or more persons, are
given codes in the Master File. Employment data are presented based on the 1968 edition of the
International Standard Industrial Classification of economic activities (ISIC) at four-digit level.
8.
The response rate for this survey is generally quite low. It rarely exceeds 60 percent for large
establishments and 30 percent for small. In 1985, which was considered a good year in terms of
response, 58 percent was achieved for large establishments and about 27 percent for small
establishments at the close of the survey. In an endeavour to improve response rate, thereby reducing
underestimation, the Central Bureau of Statistics supplements mail questionnaire with personal
interviews carried out by enumerators. Follow-up of non-responding firms is done two months from the
date of posting the questionnaire to the establishment. However, due to budgetary constraints, personal
interviews concentrate only on the large firms and in most years no follow-up is done.

9.
Due to the volume of data, analysis for this survey is done on mainframe at the Government
Computer Services. The receipt of survey forms is monitored through a specifically designed checklist,
where each responding establishment is ticked off against its name, and the date the questionnaire is
received entered into the appropriate column, after which the returns are edited and taken to the
keying room. After closing the survey in the month of March of each year, hard copies from the tapes
containing the data are availed to the Labour Statistics Section to proofread or validate. After validation
(cleaning tapes), data collected by the three survey forms (FORM LE/9-/L, FORM LE/9-/S and FORM
LE/9-/C) are merged to form the Primary Survey File. The survey file is enriched with the Estimation
File (a file that contains computer-generated employment and earnings estimates for establishments that
would not have responded by the close of the survey) to form the Enriched Survey File. The enriched
survey file is merged with two magnetic tapes containing information on civil servants and teachers
employed by the TSC to form the Final Employment File. Thirteen standard tables are then produced
by the mainframe computer and given to the Labour Statistics Section. The Labour Statistics Section
uses these tables to update employment and earnings series appearing in various publications of the CBS
and in meeting data requests from different users.
10.
Data manipulation includes estimation for non-responding establishments. Although an
estimation method for non-responding establishments is activated by the computer at the analysis stage,
the method collapses where an establishment fails to respond for three or more consecutive years. If an
establishment fails to respond in the first year, estimates are made on the assumptions that its
employment has not changed and that earnings followed the general trend of the responding firms. For
those establishments which did not respond in the two previous years, estimates are derived on the
assumption that employment followed a similar pattern as that of establishments of the same size in the
same economic activity which had responded for three consecutive years. The automatic adjustments
done by the computer are then manually scrutinized on a firm-by-firm basis to determine whether the
estimates are reasonable. If an establishment does not respond for three years in a row, it is assumed to
have gone out of business and is therefore removed from the Master File.
11.
The main constraint to the survey has been defects in the Master File. The Master File is known
to contain establishments that have closed down, and some that are dormant. It is also known to omit
establishments that are known to exist. It is against this background that there is a feeling that the
Master File should have at least 70,000 establishments as opposed to the 45,000 establishments it
currently lists. This figure can be derived by projecting the number of establishments from way back in
1980 by use of annual registration of companies as given in Table 70 of the Statistical Abstract and
growth rates of the economy by sectors since then; and using the records held with the NSSF, which
give the number of employers registered with the Fund at 39,000. The number of establishments is
likely to be higher since employers are legally required to register with the Fund if they employ 5 or
more persons and many evade registration for varied reasons; while employers are registered at
firm/enterprise level and some firms may be made up of more than one establishment.
12.
With respect to the establishments of the Central Government and Teachers Service
Commission, data are obtained from their respective nominal rolls pertaining to 30th June of every year.
These are therefore the most reliable datasets on public employment and earnings. In addition, the
Government hires works-paid casual employees for specific short-term projects undertaken mostly by
the following Ministries: Water Development, Public Works, Transport and Communications,

Environment and Natural Resources, and Agriculture. Data pertaining to such employees are gathered
through the Survey of Government Casuals as they are not included in the payroll.
B:

Survey of Domestic Servants

13.
Domestic servants are defined as people who perform household chores and include ayahs,
cooks and maids. The contribution of domestic services to Gross Domestic Product is higher than that of
either monetary forestry, or fishing, or mining and quarrying, or electricity and water. It is for this
reason that accurate data on domestic servants must be gathered.
14.
Prior to 1987, employment and earnings data on domestic servants were collected through the
Survey of Domestic Servants which used the Kenya Post Office Telephone Directory as the frame. The
frame was used on the unjustifiable assumption that everybody with a telephone employs a domestic
servant and vice versa. The frame therefore excluded numerous people who employ domestic servants
and included a lot of people who were not employing domestic servants. In addition, the frame offered
no possibilities for physical follow-ups. It is possible that more employers of domestic servants were left
out than the non-employers who were included. This created serious response problem and the quality
of the data was also questionable.
15.
Consequently, and mainly due to low response rate, possibilities of using other frames started
being explored. In 1982, a question on domestic servants was incorporated in the Rent Survey. The
analysis showed that although the figures on employment of domestic servants were low, response rates
were higher compared to response in the Survey of Domestic Servants. However it was not until 1987
when the old frame of the Telephone Directory was abandoned, and the National Sample Survey and
Evaluation Programme (NASSEP) frame was adopted as the most appropriate frame for collecting data
on employment of domestic servants. The questionnaire was modified to incorporate rent and domestic
servants and became the Survey of Rent and Domestic Servants.
16.
Data from the new survey started being used in 1988. However the results have been giving low
figures since the observation unit of the frame for rent survey is the residential structure, while there is
no discernible relationship between a structure and domestic servant. For instance, a structure may be
occupied by people who do not require hired domestic help, such as single or unmarried persons. In
addition, the rent survey frame is biased towards low income residential areas since the NASSEP frame
is determined by population size. The frame, therefore, does not adequately represent high income
groups who are the main consumers of domestic services. However, reliable data on domestic servants
can be obtained if a specific sample of the frame for the Survey of Rent and Domestic servants is
prepared by stratification so as to give more weight to middle and upper income groups.
17.
Although there is no hard evidence to prove so, it would appear that the number of domestic
servants has been underestimated. At present CBS is trying to capture domestic servants through heads
of households by using the National Sample Survey and Evaluation Programme.
C:

Survey of Small-scale (Informal) Enterprises Sector

18.
Information on small-scale enterprises (informal) sector is collected through an annual survey.
As per CBS definition, small-scale enterprise sector consists of semi-organized and unregulated
activities largely undertaken by self-employed persons in the open markets, in market stalls, in

undeveloped plots or on street pavements within urban centres. They may or may not have licenses
from local authorities for carrying out such activities as tailoring, carpentry, black-smithing, grocery,
kiosks, meat and maize roasting, sale of apparel and shoes, open air restaurants, repair of footwear, car
repair, shoe shining, hair cutting, etc (Central Bureau of Statistics, Employment and Earnings in the
Modern Sector 1981, 1984). There is therefore no frame for these enterprises. Sample units are
systematically selected and interviewed at the same time. Data on people engaged and their earnings are
collected and, for classification purposes, type of business and activity. From 1973 to 1975, the census
was undertaken in Nairobi, Mombasa, Kisumu and Nakuru but since 1976 all urban centers, that is
those having populations of 2,000 or more, excluding those in the North-Eastern Province, have been
covered.
19.

There are some major drawbacks in the analyzed data:


a) Gross earnings are reported without deducting costs of inputs e.g. raw materials, wages and
rent, thereby making it difficult to derive net earnings. This is due to the difficulty of soliciting
reliable response on sources and costs of inputs and wares.
b) Sample units lack permanence and are scattered throughout the urban centers. In addition, ease
of entry and exit makes it extremely difficult to collect reliable data, and some operators who
are there during the interview may leave immediately they dispose their wares.
c) Total employment figures relate to total persons engaged without classifying them by
employer/employee status.
d) The lack of a predetermined frame gives enumerators the discretion to decide the
establishments to interview. This does not guarantee completeness of coverage and reliability of
data collected. It is also not possible to give the trend in total employment in the sector as the
coverage in each town might change from year to year and the number of towns included
changes over time.
e) The survey does not normally cover certain economic activities e.g. the matatu sub-sector.

20.
To arrive at the total wage employment in the modern sector, data from all the above sources
are aggregated and published under the title, Employment and Earnings in the Modern Sector.
However, it should be noted that the data have shortcomings:
a) Data on wage employment covers both rural and urban sectors but those of small scale
enterprises pertain to urban areas only. A lot of informal sector activities in the rural areas are
therefore left out.
b) Due to lack of a suitable sample frame and definitional problems, data on domestic servants are
not reliable and could be underestimated.
A big proportion of labour data that go into the calculation of labour costs components of GDP are
provided by this source. However, it seems that the statistics are underestimated. Caution therefore
needs to be exercised in their interpretation.
21.
Apart from the regular surveys, CBS also conducts ad hoc surveys to meet specific data needs.
These include:
(a)
(b)

Labour Force Survey, 1977/78


Urban Labour Force Survey, 1986

(c)
(d)
(e)
(f)
(g)
(h)

Rural Labour Force Survey, 1988/89


Household budget Survey, 1974/75
Household budget Surveys: Rural (1981/82) and Urban (1982/83)
Rural Non-farm Activity Survey, 1977
Survey of Rural Non-Agricultural Enterprises, 1985
National Manpower Survey, 1986-88.

With exception of the last two surveys, all the others used the household as the unit of investigation.
This has special significance in that households are the suppliers of labour.
D:

Labour Force Surveys

22.
In 1977/78, a labour force survey was undertaken as a module within the Integrated Rural
Surveys. The objective of this Survey was to examine labour supply and its utilization. As detailed in the
Integrated Rural Surveys (1976-79), all the rural households covered by the Integrated Rural Survey and
3,000 urban households were interviewed each month for a period of 12 months. All members of
households were interviewed, including those attending school fulltime. Information on pattern of
work was gathered, in particular, hours worked on and off the holding, occupation and industry of
employment, training received, and whether or not respondents were looking for work and their
reasons. Results of the Survey were analyzed and limited copies of the report published in 1986, while a
summary was included in the 1981 Economic Survey.
23.
In order to update the 1977/78 survey, an Urban Labour Force Survey was undertaken in 1986
and the rural component in 1988/89, using the NASSEP II frame. The two surveys were a joint project
of the Central Bureau of Statistics and the Long Range Planning Unit of the Ministry of Planning and
National Development. The Rural Labour Force Survey was undertaken in two phases (July to
November 1988 and February to June 1989) in order to capture seasonality of rural employment, while
the Urban Labour Force Survey was a single round. The two surveys were based on slightly different
definitions from those used in 1977/78 survey which limited the comparisons which could be made.
The basic difference is in the length of the reference period. In the 1977/78 survey, a single day
reference period or yesterday approach was used while the 1986 Urban Labour Force Survey and 198889 Rural Labour Force Survey used yesterday approach and one week approach. All these surveys
gathered basically the same information but the 1988-89 Rural Labour Force Survey collected additional
information on urban-to-rural migration for purposes of job search. Labour force surveys yield data on
which important microeconomic parameters are calculated e.g. unemployment and participation rates.
However, definitions of the concepts used in the Labour Force Surveys mainly refer to allocation of
time, but not the economic returns of the various activities the labour spend their time on.
24.
One major problem in the interpretation of the data arises from the definition of employment.
Only few rural respondents report as unemployed in the technical sense; but this may be obscuring a
high level of underemployment and hidden unemployment. This means that the unemployment rate of
0.3 percent calculated on the basis of the 1988-89 Rural Labour Force Survey could be hiding a
substantial level of unemployment and underemployment. However, the 16.2 percent urban
unemployment rate may be more representative of the picture in urban areas due to the nature of urban
economies in comparison with rural areas.

E:

Survey of Rural Non-Agricultural Enterprises, 1985

25.
The Survey of Rural Non-Agricultural Enterprises was undertaken from December 1985 to
April 1986 and had 1985 as the reference period. The survey had three main objectives:
a) To obtain information that would help estimate the contribution of rural-based activities to
wage employment and earnings.
b) To secure information on rural sectoral contribution to Gross Domestic Product.
c) To shed light on the sectoral dispersion of rural non-agricultural enterprises.
26.
Three sectors were covered: manufacturing, distribution, and services. The survey covered a
sample of 67 rural and market centers selected from 525 centers listed in the 1979-1983 Development
Plan. Information was collected from all unregistered establishments in each selected centre with
identifiable activities and engaging not more than 10 persons (working for pay or profit). Care was taken
not to include any centre with a population of more than 2,000 persons as this would fall outside the
definition of a rural centre. All establishments included in the Master File were also excluded since they
are covered in the Annual Enumeration of Employees and Self-Employed Persons. Information
regarding persons engaged, gross earnings, and inputs (e.g. wages) were collected. This made it feasible
to compute gross product for this sector. The data was grossed up to obtain district, province and
national totals taking into account the rural and market centers not included in the sample survey.
27.
Information was obtained through personal interviews by permanent field enumerators who
were trained prior to commencement of the survey. Information on small rural non-agricultural
enterprises for the years 1973, 1974, 1975 and 1976 is reported in the Report on Surveys of Industrial
Production. However, only three sectors were covered: mining and quarrying, manufacturing, and
building and construction. As it appears, this information was collected as a subset of the Survey of
Industrial Production. This means that distribution and services were not included. In effect, therefore,
comparison with the results of the 1985 survey is rather difficult.
F:

Survey of Industrial Production

28.
Since the first Survey of Industrial Production (SIP) was conducted for reference year 1954,
others have been conducted for reference years 1956, 1957, 1961 and annually since 1963. Enquiries for
reference years 1961, 1963, 1967 and 1972 took the form of a census. Up to and including 1963, all
establishments engaging 5 or more persons were covered, and from 1964 to 1969 the coverage was
restricted to firms engaging 50 or more persons. Surveys for 1970 and 1971 covered firms engaging fifty
or more persons plus a 25 percent sample of those engaging between 20 and 49 persons. In 1972, the
coverage was extended to all establishments (see 1979 Statistical Abstract).
29.
This survey covers three sectors: mining and quarrying, manufacturing, and building and
construction. Firms covered are stratified by employment. In the surveys of 1973-1976, firms engaging
fifty or more persons were covered in full and a sample (25 percent) was selected from the group
employing between twenty and forty nine persons. Firms employing less than twenty persons were not
covered because they presented particular response problems. For convenience, the latter group of firms
is divided into three groups: those employing 5 to 19 persons; those employing 1 to 4 persons; and rural
non-agricultural enterprises employing 1 to 4 persons.

30.
Information for the category of firms employing less than 20 persons was estimated. Data on
persons engaged, gross earnings, and inputs (e.g. wages, water, transport) was collected through a mail
questionnaire supplemented by secondary statistics from the ex-community Government Corporations.
To obtain aggregate statistics for the annual SIP, data for firms employing 20-49 persons are grossed up
(blown-up) to obtain totals for this category of firms. However, every five years, a Census of Industrial
Production (CIP) is undertaken. The CIP differs from the SIP because it covers all firms employing 20
or more employees, and only takes a 50 percent sample of firms employing 5 to 19 persons and 25
percent of those employing 1 to 4 persons (see 1980 Statistical Abstract). In addition, CBS undertakes a
quarterly Business Expectation Enquiry for all industrial sectors (excluding electricity and water), which
includes employment and earnings statistics. The main limitations is that employment figures from the
Business Expectation Enquiry and Survey of Industrial Production is only for the industrial sector; cover
a sample, usually firms employing 50 or more persons; and normally take the firm rather than the
establishment as the statistical unit of enquiry (respondent) mainly because employment data is not the
domain of their study.
G:

National Manpower Survey, 1986-1988

31.
The first manpower survey was undertaken in 1964 and aimed at collecting data on high level
manpower. Two more surveys were undertaken in 1967 and 1972 and extended their coverage to
include both high-level and middle-level manpower but still low-level manpower was excluded. A
more comprehensive survey was attempted in 1982/83 but due to some constraints the survey was not
completed. This necessitated another survey as a matter of priority. Consequently the 1986/88
Manpower Survey was undertaken. The objectives of the 1986-88 National Manpower Survey were to:
a) Determine stock and distribution of various categories of manpower in the modern sector;
b) Collect and determine characteristics of employed persons in the modern sector;
c) Collect information which would assist in determining the number of vacant posts within
different sectors of the economy;
d) Assess the relationship between demand and supply of manpower in the economy in order to
determine both current and future manpower balances;
e) Determine ways of improving the existing system of occupational classification in Kenya;
f) Assist in the formulation of long-term manpower development and utilization policies and
determine the profile of wages and earnings by occupation and economic activity;
g) Assist the Government in planning and programming education and vocational training systems
more appropriate to the economys manpower requirements; and
h) Contribute towards the development of a sustained manpower databank which would improve
Kenyas capability for overall manpower planning.
32.
The 1986-88 National Manpower Survey was conducted on sample basis selected randomly
from three frames: (a) register of establishments for private sector, state corporations and municipalities,
(b) nominal civil service roll, and (c) a list of educational institutions maintained by the Teachers
Service Commission (TSC). Data were gathered through interviewers who had been thoroughly trained.
Survey work was done in two phases: Phase 1 covered employees and information was collected on a
form designated as Form A, and Phase 2 covered employers and used a form designated as Form B.
The response rates achieved were 88.6 percent and 96.6 percent for phase I and phase II, respectively.
Non-response is attributed to the inadequacy of the frame, in particular the Register of Establishments.
It was in some cases found to include some establishments which had ceased to exist and some

agricultural establishments that had been subdivided into smaller units. Important characteristics of
employees covered in the survey included age, sex, citizenship, and highest level of education and
training attained.
33.
It is noted in An Overview Report of National Manpower Survey 1986-1988 that, while most of
the objectives were addressed, four areas were not fully addressed due to various inadequacies in the
data:
a) Information on manpower supply was incomplete as only output from local education and
training institutions was considered, leaving out the Kenyans training abroad;
b) Data on new occupations which were not in the Kenya National Occupational Classification
System (KNOCS) were not analyzed;
c) The questionnaire used could not determine whether or not employees not physically housed
by the employer were given house allowance; and
d) Complete analysis of employment dynamics could not be ascertained because the data from
Phase II of the survey had not been weighted.
H:

Employment in Small Farm Sector

34.
The Government recognizes that the bulk of employment opportunities will come from small
farms, urban small-scale and rural nonfarm enterprises. In view of this, it is important that reliable
employment data should be available for all those sectors. As pointed out elsewhere, surveys on rural
nonfarm sector are undertaken on ad hoc basis which means that data pertaining to the non-survey
years are estimated. The methodology of collecting data for the survey of small scale enterprises leads to
inadequate and unreliable database. Although the small farm sector is the dominant employer, no
survey has been conducted to gather comprehensive employment data on both crop and livestock
production sector directly.
35.
Prior to 1974, the CBS conducted a wide variety of sample surveys in the rural areas to meet
specific individual objectives. The data were often found to be internally inconsistent and of limited
application, and were therefore not published. This means that data on employment in the small farm
sector do not exist for the period prior to 1974.
36.
The integrated sample surveys programme was established in 1974 and two surveys which
gathered data on employment in the subsistence sector were conducted within the framework of
national sample surveys. These are the 1977/78 Labour Force Survey and 1988-89 Rural Labour Force
Survey. The two surveys, particularly the Rural Labour Force Survey of 1988-89, provided a basis on
which employment in small farms could be estimated.
37.
The analysis of employment in the crop and livestock sectors encounters severe conceptual
problems because the database is extremely weak. However, employment in the crops sector can be
estimated by using estimates of crop planted area while employment in livestock production can be
estimated by applying estimates of labour employed per livestock head to estimates of the number of the
national herd. The 1989 estimates derived in this manner closely compare with the results of the 198889 Rural Labour Force Survey. For example, crop area-based estimate of total employment in 1989/90 is
equivalent to 2.17 million fulltime jobs (comprising 270 eight-hour days per year) compared with 2.62
million jobs estimated through the Rural Labour Force Survey. For livestock production, the Rural

Labour Force Survey estimated 1.44 million jobs while estimates based on herd size are 1.76 million
jobs. The Rural Labour Force Survey did not cover the main pastoralist districts. Employment in these
districts is estimated at 1.03 million jobs. Thus, total employment in the small farm sector is estimated at
about 5 million comprising crop-related activities (2.17 million), livestock in non-pastoralist areas (1.76
million) and pastoralist areas (1.03 million).
I:

Rural Non-Farm Enterprises

38.
According to Rural Household Budget Survey of 1981/82, nonfarm enterprises account for about
17 percent of household incomes. In spite of its importance to rural incomes, data pertaining to
employment in the sector is scanty. CBS conducts sample surveys on rural nonfarm enterprises on ad
hoc basis, the latest of which was undertaken in 1985/86. The surveys cover establishments in rural
trading centers with a population of less than 2,000 persons. Unfortunately, the surveys do not provide
proper estimates of total size of the rural nonfarm sector, though useful for assessing its sub-sectoral
composition.
39.
Most of rural nonfarm activities are household-based, usually undertaken on part-time basis
using unpaid family labour. According to a survey of rural Kenyan households undertaken in 1977 to
determine the nature and extent of nonfarm activities, at least 50 percent of the households were
engaged in at least one of such activity and nearly one out of four engaged in two or more. Estimation of
this component is intrinsically difficult. Due to the inadequacy of data, estimates of employment in
rural nonfarm enterprises are often derived as a residual by subtracting modern wage employment,
small farm and pastoral employment, from estimates of the rural labour force.

SECTION II: INSTITUTIONS ENGAGED IN LABOUR DATA COLLECTION AND ANALYSIS


40.
The Central Bureau of Statistics is the principal collector of labour data in the country, but other
Government departments also collect labour data to satisfy their specific needs which cannot be
adequately met by the CBS. For instance, the Manpower Planning Department of the Ministry of
Manpower Development and Employment conducted the National Manpower Survey in 1986-89 to
update the Manpower Survey of 1972 as the results of the 1981/82 Manpower Survey were not
published. To update the Labour Force Survey of 1977/78, the Long Range Planning Unit of the
Ministry of Planning and National Development conducted the Urban Labour Force Survey in 1986 and
later in 1988-89 undertook the Rural Labour Force Survey.
41.
Participation of CBS in surveys undertaken by other Government departments has always been
very significant due to the following reasons:
a) CBS maintains the Register of Establishments which serves as the frame for establishmentbased surveys. Without the register, the National Manpower Survey would have been difficult
if not impossible.
b) CBS has a well developed infrastructure for household-based surveys including a master sample
frame and trained permanent field enumerators.
c) All national surveys are undertaken under the Statistics Act and any institution intending to
conduct such a survey must obtain prior clearance from the Director of Statistics in accordance
with the Act.

10

42.
The input of the CBS in these surveys has been mainly technical covering questionnaire design,
sampling and data collection. Data processing and analysis is carried out by the department sponsoring
the survey. Analysis of the 1986-89 National Manpower survey was done by the Ministry of Manpower
Development and Employment, while analysis of the 1986 Urban Labour Force Survey and 1988/89
Rural Labour Force Survey was by the Long Range Planning Unit. The arrangement of pooling
resources and skills has worked well and the results are produced and published timely. This applies to
labour-related surveys and all the surveys undertaken jointly by the Bureau and other Government
departments.
43.
The Directorate of Personnel Management and Teachers Service Commission produce labour
data on numbers and characteristics of civil service and teaching establishments, respectively, as an
administrative by product using the respective nominal rolls. Data from the nominal rolls are directly
used to determine employees earnings and statutory deductions which make the quality of civil service
and teachers statistics to be unquestionably high. This is particularly important in that Central
Government and Teachers Service Commission employees comprise about 35 percent of total wage
employment.
44.
Some labour data requirements go beyond the competence of the CBS e.g. those relating to
industrial relations which fall under the Ministry of Labour. Consequently, statistics on collective
bargaining, strikes, lockouts and time lost, and occupational accidents are produced and compiled by the
Ministry of Labour.
45.
In spite of the existence of many departments collecting labour data, information of vital
importance in manpower planning is either deficient or nonexistent altogether. These include
comprehensive data on skills needs, employment needs, and information on available occupational
skills.

SECTION III: DELAYS IN RELEASING LABOUR STATISTICS


46.
Considerable delays occur between completion of surveys and publication of results. Sometimes
results of surveys are never published. The impact of the results of a survey is seriously eroded when
such delays occur. Such delays are usually caused by the following factors: budgetary constraints, staff
shortage and turnover, and inadequate computer time.
47.
CBS, like other Government departments, faces financial constraints. The department is
therefore unable to contract printing jobs commercially and is often unable to purchase items such as
paper and ink for its own press. Printing of publications outside CBS and the Government Press would
have considerably reduced the delay if funds were available to contract private printers. It is appropriate
to mention that cost overruns affect a survey at the data processing stage, leading to delays in
publication of the results.
48.
The Bureau experiences a high staff turnover. It is essential that a subject matter specialist
assigned a particular job takes it through from the design of the questionnaire to analysis and
preparation of final documents for publication. However, turnover of this category of workers is

11

unfortunately very high. This has inevitably resulted in delays in publication of results. The 1981/82
Rural Household Budget Survey, whose results have not been published to date, is a case in point.
49.
The Government Computer Services (GCS) provides computer time to Government
departments and institutions. CBS is therefore forced to share and compete for computer time with
other Government departments. CBS work is given less priority to, say, the payroll. In such cases, CBS
work suffers, resulting to delays in publication of survey results. These delays could be reduced if funds
were available to contract for private computer time or to expand GCS so as to give CBS uninterrupted
computer time.
50.
The table below gives an idea of delays involved in printing results of various labour surveys and
censuses. It should be emphasized that a period of about one year from completion of a survey to
publication of the results is considered normal. Any period in excess of one and half years is
considered excessive.
Table 1: Delays in Official Release of Recent Employment and Labour Force Surveys
NAME OF SURVEY
ISSUE DATE
OF DATE OF PUBLICATION
COMPLETION
Employment and Earnings in the 1985
February 1986
July 1989
Modern Sector
1986
March 1987
Not published
1987
March 1988
Not published
1988
March 1989
Not published
1989
March 1990
Not published
National Manpower Survey
March 1988
October 1988 (normal
period)
Urban Labour Force Survey
January 1987
July 1988 (normal period)
Rural Labour Force Survey
April 1989
June 1990 (normal period)
Survey of Rural Non-Agricultural
April 1986
1989
Enterprises
Labour Force Survey
1977/78
Draft report published in
1986
51.
It is evident that considerable delays occur between completion of surveys and their
publication. However, users are occasionally allowed access to the unpublished information at the
discretion of the CBS. To reduce the delays, it is recommended that:
a) Provision of microcomputers and training of professional officers in their use should be stepped
up.
b) Finance, especially at the data processing stage of the survey process, should contribute towards
reduction of delays.
c) To accord survey continuity when an officer leaves, there should be a requirement that survey
knowledge be shared within a section or persons in the CBS.
d) To enhance survey and research capability of CBS, training should be expanded and officers
given wide exposure on survey methodology.

12

SECTION IV: USES OF LABOUR DATA


52.
Labour data have three basic uses: (a) to give an insight to the dispersion of wages in the
country; (b) to provide labour costs data needed for compilation of national accounts; and (c) to create
employment profile for the entire population. Data which allow analysis of wage dispersion and wage
trends are regularly gathered through the Annual Enumeration of Employees and Self-employed
Persons. Similar information was also gathered by the National Manpower Survey of 1986-89. However,
the data collected through the two surveys cover wages in the modern sector. Wages in informal sector
and in small farms are excluded. In addition, non-salary benefits e.g. cost of rations, leave passages,
bonus and gratuity payments, contribution to private provident funds, cost of uniforms and medical
benefits, are excluded.
53.
Labour costs are a major component in the systems of national accounts. Information on nonsalary benefits relating to self-employed persons and unpaid family workers, is practically difficult to
gather. The publication titled Sources and Methods used for the National Accounts of Kenya (Central
Bureau of Statistics, 1977) acknowledges problems involved in compiling labour data needed for
national accounts. For instance, no imputations are made for wages and salaries of self-employed
persons and unpaid family workers. Similarly, non-salary benefits are not imputed for this category of
workers which means that some labour costs are concealed in the gross operating surplus component of
the national accounts.
54.
To create an employment profile, the entire population is divided between rural and urban,
using population census data and population projections for inter-census years. The rural and urban
populations are further subdivided into those of working age and those not of working age. In Kenya,
the rural working age population is defined to be eight years of age and above, and 15-64 years for the
urban population. The working age population is divided into persons who are part of the labour force
(or economically active) and those who are inactive. Those not in the labour force includes housewives,
students at school fulltime, retirees and those unable to work through, say, illness. The participation
rate is the proportion of working age population who are in the labour force.
55.
Results of the 1986 Urban Labour Force Survey show that 70.4 percent of the total working age
population are in the labour force, while the 1988-89 Rural Labour Force Survey shows that almost
everyone in the rural areas work, i.e. almost no one in the rural areas is openly unemployed. Urban
unemployment was 16 percent of the labour force or 59 percent of the working age population.
56.
Data on total modern sector employment for private and public sectors is computed from the
Annual Enumeration of Employees and Self-Employed Persons and returns from the Teachers Service
Commission and the Directorate of Personnel Management. However, the basic data on Civil Service
and teaching establishments does not categorize employment data between rural and urban. The
distribution of civil service employment will be therefore split between rural and urban in a rather
arbitrary way. Urban public sector will be assumed to consist of central Government employment net of
those engaged in agriculture and forestry, teaching establishment weighted by the share of urban in
total population, local Government and parastatal sector employment. The rural public sector consists of
those working in agriculture and forestry and the teaching establishment weighted by the share of rural
in total population. Private rural and urban employment are then determined residually from the
published modern sector employment. Indeed, it appears that this was the approach adopted in the
1989-93 Development Plan.

13

57.
The urban self-employed are determined as part of the questionnaire of Annual Enumeration of
Employees and Self-Employed Persons (Excluding Domestic Servants). Total urban unemployment is
determined by applying unemployment rate (16 percent of the labour force). The urban informal sector
is derived residually by netting out urban modern sector employment and the unemployed from the
labour force.
58.
The rural nonfarm employment can be determined from the results of the 1981/82 Rural
Household Budget Survey (RHBS) or the 1988-89 Rural Labour Force Survey (RLFS). The RHBS shows
that nonfarm enterprise was the main source of income for an estimated 9.9 percent of the rural
household population. Since all members of the households included would not be engaged in nonfarm
enterprise and the returns from each activity are not proportional to time spent, this methodology is
likely to overestimate the contribution of nonfarm enterprise to total employment. The use of RLFS
data is also laden with definitional problems. However, if we restrict ourselves to nonfarm profitmaking, then the hours spent on nonfarm enterprise as proportion of time spent on all rural economic
activities would 7.7 percent for the 8-64 years age range and 6.6 percent for all rural population of eight
years of age and above. The latter methodology is more appropriate, although it is not a proxy for rural
nonfarm employment but time spent by the rural labour force on nonfarm profit-making activities.

CONCLUSION
59.
The data on the modern sector is collected through an annual census. Despite the low response
rate, the sectoral composition of employment is fairly reliable. Employment and earnings data on civil
service and teachers are probably the most reliable as they are collected directly from their respective
nominal rolls. Data on domestic servants is underestimated due to an inappropriate frame. Information
on small-scale enterprises is collected through an annual census. Although the aggregate figures are
unreliable, the census is likely to give a fair representation of the sectoral breakdown of informal
activities. A survey of rural non-agricultural enterprises was conducted in 1985, but its usefulness is
more in characteristics of these activities, rather than in their aggregate contribution to employment. In
addition, the survey did not adequately capture household-based nonfarm activities. In general, the
labour data generated by Government gives a fairly good base, especially of the industrial occupations of
wage employment. It also gives a fairly accurate picture of growth rates in modern sector employment,
both by sectors and in the aggregate, as estimation problems appears consistent over time. However, the
maintenance of an accurate record of firms is rather difficult, especially for the small firms where the
rate of entry and exit is expected to be high.
60.
Labour force surveys were conducted for urban (1986) and rural (1988/89) areas. The surveys
give useful parameters (e.g. participation rates and unemployment rates) which are useful in mapping
out an employment profile for the entire population. Since labour force and household budget surveys
collect data from the households, they offer a good supply-side complement to labour statistics gathered
from employers.
61.
There are various Government agencies involved in collection of labour data. However, the
proliferation of institutions concomitantly diverts CBS staff time to help the other Government
departments, and is also not conducive to raising the skills levels in CBS. CBS does not have adequate

14

financial allocations to undertake major surveys, and it would be useful to concentrate resources rather
than spread them thinly over many institutions.
62.
The creation of an employment profile confronts at least four major problems. First, there is no
single year when all the surveys were conducted at the same time. Second, there is some undercoverage of the modern sector due to non-response and the comprehensiveness of the Master File.
Third, some sectors e.g. rural nonfarm and small farms are not covered, and have to be derived either
indirectly through labour force surveys (e.g. nonfarm) or residually (e.g. small farms). Fourth, as
illustrated by the distinction between small farms and rural nonfarm, there are definitional problems. In
case of the rural sector, it not clear whether marketing and fetching of water and firewood are nonfarm.
In this annex, these activities are included in small farms while rural nonfarm is restricted to non-farm
profit-making activities.

REFERENCES
Brown, Laura, The 1988-89 Rural Labour Force Survey: Description and Structure of Database,
Technical Paper 90-02, Long Range Planning Unit, Ministry of Planning and National Development,
Nairobi, Kenya, 1990
Kenya, Central Bureau of Statistics, Ministry of Planning and National Development, Labour Force
Survey, 1977/78: Basic Report, Nairobi, 1986
Kenya, Central Bureau of Statistics, Ministry of Planning and National Development, Employment and
Earnings in the Modern Sector, 1984 and 1985
Kenya, Central Bureau of Statistics and Long Range Planning Unit, Ministry of Planning and National
Development, Urban Labour Force Survey, 1986, 1988
Kenya, Central Bureau of Statistics, Ministry of Planning and National Development, Survey of Rural
Non-Agricultural Enterprises 1985, 1989
Kenya, Ministry of Manpower Development and Employment, An Overview Report of Manpower
Survey 1986-88, Nairobi, Kenya, 1988
Kenya, Central Bureau of Statistics, Ministry of Planning and National Development, Economic Survey
1990 (Chapter 3: Urban Labour Survey, 1986), 1990
World Bank, Employment and Growth in Kenya: A World Bank Economic Report, Eastern Africa
Department, Washington D.C., 1988

15