Hong Kong, China (2)

Title of the survey

Survey of Wages, Salaries and Employee Benefits

Organization responsible

Census and Statistics Department

Periodicity of the survey

Half-yearly, in March and September.

Objectives of the survey

To compile nominal wage indices for individual industries analysed by broad occupational group as well as for all the industries covered as a whole, in order to measure the changes in wage rate in money terms;
To derive real wage indices by deflating the nominal wage indices by the Consumer Price Index, for measuring the changes in purchasing power of wages; and
As a by-product, to compile statistics on the absolute level of prevailing wage rates and normal hours of work of principal occupations in the major industries, for reference in career counselling and salary determination.

Main labour topics covered by the survey

number of employees
wage rates
normal hours of work per day
standard working days per month

for each selected occupation in each industry. In addition, about once every two years, information is collected on establishment pay practices in terms of employee entitlements to various types of benefits other than wage rates.

Reference period

Employment: the last working day of the reference month. Wage rates and hours of work: the whole month of March and September for salaried employees; a fixed day for wage earners, (i.e. the last normal working day of the month).

Coverage of the survey

Geographical

The whole territory.

Industrial

Since the March 1993 survey, 46 selected industries are included, covering: manufacturing, electricity and gas, wholesale, retail and import-export trades, restaurants and hotels, transport and communication services, business services and personal services.

Establishments

Medium and large establishments above a pre-set size of employment which is determined separately for each industry. In general, establishments with less than 10 persons employed are excluded.

Persons

Employees up to the supervisory level, aged 15 years and over. Proprietors, family workers and out-workers who work in their own premises are excluded.

Occupations

In each selected industry, a number of occupations which have clearly understood and definable duties are surveyed if they are either numerically important or identified as being of special interest. Individual occupations are those classified under the following two occupational groups:

Supervisory, technical, clerical and miscellaneous non-production workers, and
Craftsmen and operatives.

Managerial and professional level occupations are excluded from the scope of the survey.

Concepts and definitions

Employment

Data refer to full-time employees directly paid by the establishment, whose duties and tasks correspond to the selected occupations. The main categories of employees covered in the survey are time-rated and piece-rated workers. Within these categories, workers of both sexes are included. Time-rated employees cover all employees paid on a time basis, whether this be hourly, daily, weekly, fortnightly or monthly. For monthly-rated employees, only those receiving a full month's salary in the reference month are included. For daily-rated and piece-rated employees, only those working on the reference date are included. Employees who are paid less than the established wage rates, such as handicapped workers whose wage rates are reduced because of their handicap, part-time workers, apprentices, learners, trainees and workers on probation are not covered. Data are also collected on the total number of persons engaged by the establishment on the last working day of the reference month. This includes working proprietors and business partners, and all full-time employees, both permanent and temporary, who are either at work or temporarily absent from work on the reference date, because of sick leave, maternity leave, annual vacation or casual leave, and on strike. Part-time employees and unpaid family workers who work more than three hours on the reference date are also included. Excluded from persons engaged are persons on indefinite leave and on pension, and outworkers who work in their own premises on materials supplied by the establishment.

Earnings

Not relevant.

Wage/salary rates

They comprise the following elements:

basic wages and salaries, including paid holidays and leave
commission and tips (excluding tips received directly from customers)
shift allowance
cost-of-living allowance
meal allowance and meal benefits
good attendance bonus
guaranteed year-end bonus, and
other regular and guaranteed bonuses and allowances

About once every two years, information is collected on establishment pay practices in terms of employees' entitlements to various types of benefits other than wage rates. These include:

paid holidays per annum
housing benefits
provident fund
pension scheme
medical benefits
profit-sharing bonuses
rates of overtime premium, and
other benefits

Hours of work

Data are collected on normal hours of work per day. They refer to the hours of work, excluding meal breaks and rest time, stipulated by the establishment for each occupation, and beyond which any time worked is remunerated at overtime rate or forms an exception to the rules and customs of the establishment. Data on wage rates and normal hours of work cover the same categories of employees as employment data and are collected separately by sex and mode of payment for each selected occupation in each industry. Data are also collected on the number of standard working days per month, by sex and mode of payment, for each selected occupation in each industry. If there is no rest days granted at all in a month, the number of standard working days is 30. Where rest days are granted on a monthly basis, the number of standard working days is the difference between 30 and the number of rest days granted in a month. If rest days are granted on a weekly basis, the assumption of four weeks in a month is made. The standard working days per week are calculated as follows:

26 working days per month correspond to 6 days a week,
24 working days per month correspond to 5 1/2 days a week,
23 working days per month correspond to working half days on alternate Saturdays,
22 working days per month correspond to five days a week.

International recommendations

The components of wage and salary rates mentioned above conform to the international recommendations on wage rates. They refer to the rates actually paid by the establishment for the specified occupations during the reference period, and comprise basic wage rates as well as guaranteed and regularly paid allowances. The concept of normal hours of work used in this survey conforms to the international guidelines.

Classifications

Industrial

Data on wage and salary rates and normal hours of work are classified according to the Hong Kong Standard Industrial Classification (HSIC) and selected industries are coded mainly at the four-digit level of this classification. The HSIC can be linked to the International Standard Industrial Classification of all economic activities (ISIC), Rev. 2, 1968 at the four-digit level, with some slight modifications.

Occupational

Data on wage and salary rates and normal hours of work are classified according to the International Standard Classification of Occupations (ISCO-68). Individual occupations are selected from major groups 3, 4, 5, 7, 8 and 9 and classified according to the following occupational groups and subgroups:

Supervisory, technical, clerical and miscellaneous non-production workers:
- supervisory and technical workers
- clerical and secretarial workers
- service workers, and
- miscellaneous non-production workers
Craftsmen and operatives:
- craftsmen
- operatives

Occupational classifications and job specifications are defined in advance of the survey. The job specifications may differ significantly from those in use in individual establishments. In order to improve the inter-establishment comparability of occupational content, the primary objective of the job specifications is to identify the essential elements of skill, difficulty and responsibility that establish the basic concept of the job.

Others

Data on wage and salary rates and hours of work are also classified by sex and mode of payment.

Sample size and design

Statistical unit

The sampling and reporting unit is the establishment, defined as an economic unit which engages, under a single ownership or control, in one, or predominantly one kind of economic activity at a single physical location. The unit of inquiry is an occupation in an establishment (and not the individual worker).

Survey universe / sample frame

This consists of the Central Register of Establishments kept by the Census and Statistics Department. It covers all non-agricultural establishments engaged in economic activities classified under HSIC major divisions one to nine. The Register is updated on a continuous basis to take account of birth and death cases, on the basis of data from the Business Registration Office, administrative records and results from various economic surveys of the Census and Statistics Department. For records which are new to the Register, establishment particulars are collected by postal, then field enumeration method. New insertions and amendments are updated on a quarterly basis.

Sample design

The survey is based on a rotational stratified random sampling method. A main sample is drawn from the frame once every two years. It is stratified by industry and employment size and divided into eight replicates. Four replicates are selected for each round of the survey; each round, one replicate is replaced in order to spread the burden of selected establishments, so that 75 per cent form an overlapping sample for two consecutive rounds. A sample unit is thus surveyed a maximum of four times and the whole sample is completely renewed after two years. Each round, the sample consists of some 2,000 establishments, with an average of 50 establishments per industry, out of a total of some 7,000 establishments falling within the scope of the survey. This represents an employment size of approximately 270,000 persons, out of some 680,000 employees. The sampling fraction and sample size in terms of number of establishments varies with each stratum. The sampling fraction ranges between 7.55 and 50 per cent, and public utility companies (i.e. electricity and gas and transport services such as motor buses, tramways and ferries) are included with certainty. Each round of the survey, supplementary samples are chosen and incorporated into the main sample in order to take account of death cases (i.e. closed-down establishments, establishments that have fallen below the minimum size limit and those that have changed the nature of their business) and to incorporate birth cases (i.e. newly opened establishments that fall within the scope of the survey and establishments not previously within the scope of the survey that have shifted in during the course of the survey).

Field work

Data collection

This takes place during the periods 1 April - 20 May (for the March round) and 1 October-20 November (for the September round). At the start of the survey round, notification letters are first sent out to the sampled establishments. Field officers, who are members of a permanent survey organization, then visit the respondents and assist them in job matching and filling in the questionnaire.

Survey questionnaire

This consists of two parts and is complemented by explanatory notes corresponding to most questions in the questionnaire. Part A provides general establishment information such as name, main product or activity, total number of persons engaged, total number of employees covered in Part B, and particulars of the establishment covered in the questionnaire (name, address, business registration number, etc.). Part B is designed to provide occupational data on wages and salaries. Respondents are required to fill in one form per occupation. For each occupation, detailed information is required on the number of employees covered, by sex and mode of payment, and for each sub-category, on the following items: number of normal working hours per day; number of standard working days per month; minimum and maximum basic wages and salaries actually paid to employees and weighted average basic wages/salaries; amount of each benefit listed under "Wage/salary rates"; and item of work and normal output per day for piece-rated employees.

Substitution of sampling units

There is no substitution of sampling units in case of total non-response (due to refusal, establishments which have closed down or moved with an unknown address, etc.).

Data processing and editing

Data are processed both manually and by computer. Coding of occupation is done manually; coded data are transcribed onto data preparation forms according to coding instructions. Data are checked first manually, then through computer programmes. Three types of checks are made: (i) level checks for overlapping replicates, to check whether the change in wage rate is acceptable, (ii) range check for all replicates, to check whether wage rates are within an acceptable range, and (iii) logical check on individual data items for internal consistency. Individual occupations with missing fields are discarded. Starting from the March 1993 survey round, data from a certain proportion of establishments are input directly into laptop computers during field enumeration and validated on the spot. For the other establishments, data are collected using the conventional pen and paper method and input into the computer (with on-line validation) by indoor staff. The input data are then further checked by computer programmes to detect outliers.

Types of estimates

The survey is designed to estimate both the changes in and the levels of wage rates. Three types of estimates are derived from the survey:

average daily wage rates and monthly salary rates,
nominal and real wage indices, and
median, and lower and upper quartiles of wage/salary rates.

Estimates are also made of mean normal hours of work per day and mean number of standard working days per month. Median and quartiles: since wage rates of individual workers are not available through this survey, the computation of median and quartile wages is based on the assumption that the average wage rate for a particular occupation in a particular establishment is representative of the wage rates of all workers in that occupation in that establishment. Establishment pay practices: statistics on various benefits other than wage rates are expressed in terms of the percentage of workers included in the survey and entitled to the benefits.

Construction of indices

The Nominal Wage Index is calculated using a Laspeyres base-weighted formula, so that occupational and industrial structure is held constant. Starting from the March 1993 survey round, the base period is September 1992 and the weights were determined in the September 1992 survey (previously, March 1982). To compare wage indices of the old series (March 1982=100) and those of the new series (September 1992=100), a splicing method is used, whereby (i) a conversion factor is derived from the indices for the overlapping period of September 1992 in both index series, and (ii) the index for a particular month with the old base period is converted to the new base period by multiplying the index by the conversion factor. The Real Wage Index is obtained by deflating the Nominal Wage Index by the Consumer Price Index (A). Although the same deflator is used for each industry to derive the real wage index from its nominal counterpart, the difference between the movements of the nominal wage index and the real wage index between any two time periods may vary slightly from industry to industry, due to rounding.

Weighting of sample results

Average daily wage rate or monthly salary rate of a particular occupation in a particular industry is given by the ratio of the estimated total wage bill for that category of workers to the estimated total number of workers in that category, according to the following simplified formula:

where,

R=: average wage rate for employees of a certain sex and mode of payment of a particular occupation in a particular industry;
F_i=: grossing-up factor for the i^th establishment;
e_i=: reported number of employees in that category in the i^th establishment;
x_i=: reported average wage rate of employees in that category in the i^th establishment.

For calculating average wage rates for higher levels of aggregation (e.g. an occupation, an occupational group, a particular industry and all industries combined), the same method is used, extending the summation sign in both the numerator and the denominator to cover all workers in the required level of aggregation.

Adjustments

Non-response

Since the March 1993 survey, non-response is taken into account by adjusting the sampling fractions so that in effect, the values of stratum averages are imputed for the non-response cases.

Other bias

Changes in industrial classification are accounted for by adjusting the sampling fraction.

Use of benchmark data

Not relevant.

Seasonal variations

Not relevant.

Indicators of reliability of the estimates

Coverage of the sampling frame

Total coverage of all non-agricultural establishments engaged in economic activities is aimed at. The Central Register of Establishments covered some 480,000 establishments as at December 1990.

Sampling error / sampling variance

The coefficient of variation of daily and monthly wage/salary rates at broad occupational group level mostly ranges between five and ten per cent. The sampling variance is computed as follows:

where, for four independent replicates,

y_i=: estimate for the i^th replicate,
=: overal estimate = .

Non-response rate

It averages 9 per cent, in terms of refusals, closed-down establishments or establishments which moved to an unknown address, etc.

Non-sampling errors

The following remarks should be taken into account when interpreting the estimates:

Sample size: a sample of 2000 establishments cannot bear detailed analysis. The detailed breakdowns of wage rates for individual occupations are therefore given for rough guidance only.
Changes in wage rates: as a result of changes in sample composition and staff movement, the movement of the absolute level of wage rates may differ slightly from the movement of the wage indices. Where this happens, the wage indices should be used as indicator of changes;
Wage indices: the wage indices are designed to measure the "pure" changes in wage rates by holding constant the structure of the labour force with respect to industry, occupation, sex and mode of payment between two successive rounds of the survey. However, it is not possible to standardize the experience and quality of workers within the basic unit which is defined by an occupation, sex and mode of payment. Hence, changes in composition of workers in terms of experience and quality within a basic unit may still distort, though not significantly so at the higher levels of aggregation, the movements of the wage indices;
Wage rates of piece-rated workers: a time basis is used to measure piece-rated wages. When asking for piece-rated wages, the normal output of an average worker within a certain time period is obtained together with the rates per unit of output. An assumption is made that there is sufficient work for the worker to work unceasingly within the time period. In view of this, the wage rate of piece-rated workers should not be used to calculate actual earnings.
Linking wage index series of different base periods: the method used to compare wage indices of old and new series should be applied with particular care in comparisons at the economic sector level and the "all industries" level, in view of the fact that the coverage of some industries may be different for the two index series, and the industry coverage of each economic sector has been revised in 1992.

Conformity with other sources

A reconciliation of the survey results is made with the findings of the Index of Payroll Per Person Engaged by industry group, obtained from the Survey of Employment, Vacancies and Payroll (SEVP).

Available series

Two sets of data are tabulated:

summary statistics in the form of short-term economic indicators, relating to nominal and real wage indices analysed by industry and broad occupational group, and by economic sector and occupational group, and medians and quartiles of daily wages and monthly salaries analysed by economic sector and broad occupational group;
detailed results on average daily wages and average monthly salaries, normal hours of work and standard working days, separately for craftsmen and operatives, and for supervisory, technical, clerical and other non-production workers, analysed by industry, occupation and sex; as well as percentage of each category of workers entitled to various benefits other than wage rates, analysed by industry.

History of the survey

The Survey of Wages, Salaries and Employee Benefits was introduced in March 1982 on a quarterly basis to replace the old Wage Survey which had been conducted since 1964 at half-yearly intervals. At that time, a number of improvements were made, which included:

the use of a larger sample (from 350 to some 2,000 establishments) to achieve a higher degree of precision in the estimates,
an extension in coverage to include the non-manufacturing sectors (a total of 41 industries were covered, representing most economic sectors) and non-manual workers (i.e. supervisory, technical, clerical and miscellaneous non-production workers),
adoption of the rotational sampling method, in order to reduce the burden on respondents, and
collection of more detailed information.

After the September 1989 survey round, the frequency of the survey was changed to half-yearly. In 1992, a review of the survey was undertaken with a view to improving the survey design. As a result, the industrial coverage has been updated, mainly by dropping some declining manufacturing industries and adding industries in the services sector which have grown in importance. The occupational coverage has also been updated. Moreover, improvements to the sample design, estimation methods and questionnaire design have been introduced. A parallel run of the new and old survey design was conducted in September 1992. Starting from the March 1993 round, wage statistics are published based on the new survey design and the base period of the wage indices changed from March 1982 to September 1992. Since March 1994, the collection of payroll statistics has been transferred from the Survey of Employment, Vacancies and Payroll to the Survey of Wages, Salaries and Employee Benefits. The latter is subsequently renamed Labour Earnings Survey.

Documentation

Census and Statistics Department: Report on Half-yearly Survey of Wages, Salaries and Employee Benefits, volumes I and II (half-yearly, Hong Kong). The first volume includes summary statistics in the nature of short-term economic indicators and the second volume contains detailed occupational wage data. They are published some 3.5 months after the survey reference period. Both volumes contain methodological information on the survey. Starting from the March 1993 survey round, the content of the report and the table layouts have been slightly modified. idem: Hong Kong Monthly Digest of Statistics (monthly. ibid.); the survey results appear in this publication some 5.5 months after the survey reference period. idem: Hong Kong Annual Digest of Statistics (annual, ibid.); published some 14 months after the survey reference period. Ad hoc requests for data which do not appear in national publications are handled on a case-by-case basis by the Census and Statistics Department.

Confidentiality / Reliability criteria

Statistics compiled from data collected from less than ten establishments are suppressed from publication because either (a) data are insufficient to provide meaningful statistical results, or (b) there is a possibility of disclosure of individual establishment data.

Other information

Data supplied to the ILO for publication

Statistics of average daily wage rates in non-agricultural activities, manufacturing, construction and transport, storage and communication are provided each year for publication in the Yearbook of Labour Statistics. Statistics of average monthly wage rates in trades, restaurants and hotels, financing, insurance, real estate and business services, as well as personal services, are contained in the ILO data base and available upon request. Half-yearly series of average daily wage rates (wage earners) and monthly wage rates (salaried employees) in non-agricultural activities and manufacturing are published in Tables 7 and 8 of the Bulletin of Labour Statistics. Statistics of average basic daily wage rates, average basic monthly salary rates and normal hours of work, as well as average wage and salary rates including fringe benefits, by occupation and industry, are published in Statistics on occupational wages and hours of work and on food prices - October Inquiry results, a special supplement to the Bulletin of Labour Statistics.