Labour Force Survey 2008 - Nepal|劳动力调查数据集|就业统计数据集

Abstract
---------------------------

The major aim of the 2008 Nepal Labour Force Survey was to update a set of comprehensive statistics on employment, unemployment, under-employment. As essential tools, such information play key role in assessing the impact of various government policies and programs planned for employment generation. Likewise, the results from the survey do provide information required for skill development, for managing the flow of migrant workers, for improving the status of women and children, and for assessing the role and importance of the informal sector. However, this time, the scope was further extended to generate information required to improve infrastructure of households and for appropriate management of remittances flowing into the country.

Geographic coverage
---------------------------

National

Analysis unit
---------------------------

- Households
- Individuals

Universe
---------------------------

The survey covered all households excluding households of diplomatic missions and institutional households such as school hostels, prisons, army camps and hospitals. The homeless and those people living for six months or more away from the households were not considered eligible and hence were excluded from the survey.

Kind of data
---------------------------

Sample survey data [ssd]

Sampling procedure
---------------------------

Sample design
The sampling design adopted in NLSS-II is slightly modified from NLFS-I. The design is based on two stages stratified sampling technique with equal PSUs or households distributed between urban and rural areas as done in NLFS-I considering the heterogeneous labour force activities to provide a detailed picture of employment situation in the urban areas. So the prescribed 800 PSUs are divided equally in two parts, i.e., 400 PSUs each for urban and rural. Urban areas are stratified into three strata as Urban Kathmandu Valley, Other Urban Hills and Urban Tarai, and rural areas are stratified into Rural Mountains, Rural Hills and Rural Tarai. The sample size of 400 PSUs in each urban and rural area will be proportionately distributed within their respective strata. In the first stage, the prescribed PSUs (wards or sub-wards or combination of wards) are selected by Probability Proportional to Size (PPS) from each stratum, i.e. the number of households in the ward as the measure of the size. Using PPS at the first stage is followed by selecting a fixed number of households at the second stage that provides a convenient workload for each field team. This design has the benefit that the sample, in principle, is self-weighted. In the second stage, 20 households are selected from each PSU (ward) of urban and rural strata with systematic random sampling procedure giving a total of 16000 households from 800 PSUs.

The minimum sample size required is estimated as 244 households for urban and 361 households for rural domain based on the currently economically active population taken from Nepal Living Standards Survey 2003-04. Based on employment status of the population of this fresh survey information, the minimum households to be selected turns out to be 265 for urban and 277 for rural. However, 400 households each in urban and rural area have been allocated in this survey to provide reliable estimates for lower level of disaggregation up to 5 years age group (10 groups) by sex (2 groups). There will be altogether 8000 households in each urban and rural area.

Note: See Annex A (in the final survey report) for the details of the sample design.

Mode of data collection
---------------------------

Face-to-face [f2f]

Research instrument
---------------------------

The questionnaire for the survey was prepared in Nepali language. The questionnaire was based on the survey questionnaire used in the first Nepal Labour Force Survey 1998/99. There were some additional subjects such as household characteristics, migration and remittances and absentees information included in the questionnaire for this second round.

The NLFS-I questionnaire (which was developed on the basis of an ILO manual and comprised of 77 questions) provided a concrete base for developing the NLFS-II questionnaire. The NLFS-II questionnaire was substantially expanded to 130 questions in eight sections. The questionnaire structure ensured that respondents (other than heads of households who also had a few additional questions) generally answered a maximum of about 80 questions and most were asked significantly fewer questions.

For the sake of maintaining comparability many questions that were asked in the NLFS-I were retained. However, some important additions were made to address some contemporary issues such as household amenities, absentee population and remittances flowing into the country. Some questions were modified in order to have better response than in the previous survey. The eight sections in the NLFS-II questionnaire were designed to capture data on general information, household information, current activities, unemployment, activity in last 12 months, past employment record, absentee information and remittances received by households. The sections on household information, absentee information and remittances were asked only to head of household.

As already indicated, some modifications were made on questions that aimed to collect accurate and reliable information on informal activity and usual activity. Questions relating to usual activity were asked to collect the number of months he/she had spent in three different categories of activity (working, not working but available for work, and not working and not available for work) over the last 12 months. A month was considered to be "worked" if the person spent most days in that month at work. A month was considered to be "not working but available for work" if the person spent most days in that month not working but available to work. This method of determining usual activity status is simpler than that used in the NLFS-I (see footnote 1) but remains comparable with the international standards on labour statistics. In the household information, some of the questions are intended to provide statistics for monitoring selected Millennium Development Goals. Furthermore, some questions that have been added relate to ownership of house, type of energy used for cooking, lighting source and size of agriculture land. A copy of the questionnaire is mentioned in Annex B, while Annex C contains a detailed flow chart which can be used to identify the paths followed through the questionnaire by people with different characteristics.

As in the NLFS-I, the lower age cut-off point for the questions on economic activity was maintained at 5 years. This enables in the collection of data on the economic activities of children. This time as well, the objective to make the questionnaire more gender sensitive has been retained. Information on those activities such as cooking, cleaning and childminding which are performed without pay for the household, mainly by women, was collected.

Initially, the draft questionnaire was presented to the Technical Committee and underwent several rounds of discussions. Improvements and modifications were made according to the suggestions and comments received from members of the Technical Committee representing various interest groups. The questionnaire thereafter was pre-tested several times and revised intensively in accordance to the feedback received from each pre-test. The pre-tests were carried out to cover a wide range of areas and included different ecological zones, development regions and urban/rural areas. The questionnaire in the final Nepali version was translated into English to see if there were any misunderstandings likely to emerge. Probable confusions were removed in the final Nepali version of the questionnaire that was to be administered to the field for data collection so that there would be less chance of misunderstanding about the intended meaning of each question. At the same time an interviewers manual was prepared to elaborate the concept and objective of each question that led to collect reliable and accurate information.

Cleaning operations
---------------------------

A centralised data processing system was used for this survey. There could have been some advantages in using decentralised processing, since it would have permitted field editing immediately after the fieldwork. However, as in the past, the need to maintain a careful control over data entry and verification of all coding, especially of difficult topics such as occupation, industry, and subject of training was difficult to resolve in the field.

The data entry exercise in whole was carried out by at most three data entry operators for 12 months simultaneously with the field work. The data entry activity was supervised by an experienced computer officer in the section. Labour force surveys generally use very technical definitions of key terms (such as the ‘currently active’) and as in the past, special programs were therefore written in CSPro to produce these derived variables, based on how each person responded to various different questions.

Response rate
---------------------------

There was very little non-response on the survey, with data not collected for only 24 households out of 16,000. Twenty of these households are accounted for by one PSU in the Far-western region. This PSU could not be covered in the third season because the selected households had been evicted to other wards of next VDC in order to declare it a wildlife conservation area. The eviction had taken place after the 2001 Population Census. The weights for the two other PSUs selected in that area were therefore adjusted at the analysis stage to compensate for the 20 missing households.

Sampling error estimates
---------------------------

As with all surveys, the results from the NLFS II are subject to two major sources of error. These can be broadly termed as sampling error and non-sampling error. In a survey of this size, the robustness of the sample design means that the sampling errors for statistics at the national level are likely to be fairly small. Non-sampling errors are likely to be the major source of concern, and every effort has been made at all stages of the survey to try to minimise these non-sampling errors.

Sampling errors have been calculated in STATA application that was used for processing this survey. In order to derive these estimates of sampling error, account was taken of the structural design of the survey, with PSUs being assigned to six strata consisting three urban and three rural stratum, and with different sampling fractions being used in each stratum. However the resulting sampling errors probably substantially overstate the width of the true confidence intervals, since they take no account of the very strong implicit stratification by region and ecological zone incorporated into the design.

More details of the estimation of sampling errors are given in Annex A of the final survey report.

摘要
---------------------------
2008年尼泊尔劳动力调查的主要目标是为就业、失业和低就业等全面统计数据更新一套综合统计。此类信息作为基本工具，在评估旨在促进就业的政府政策和计划的影响方面发挥着关键作用。同样，调查结果也提供了所需的信息，以促进技能发展、管理移民工人的流动、提高妇女和儿童的地位以及评估非正式部门的作用和重要性。然而，此次调查范围进一步扩展，以生成改善家庭基础设施和适当管理流入国家的外汇汇款所需的信息。
地理覆盖范围
---------------------------
全国
分析单位
---------------------------
- 家庭
- 个人
调查范围
---------------------------
本调查涵盖了所有家庭，不包括外交使团家庭以及如学校宿舍、监狱、军队营地和医院等机构家庭。无家可归者和那些离开家庭六个月或更长时间的人被认为不符合资格，因此被排除在调查之外。
数据类型
---------------------------
样本调查数据 [ssd]
抽样程序
---------------------------
抽样设计
NLSS-II采用的抽样设计在NLFS-I的基础上略有修改。该设计基于两阶段分层抽样技术，与NLFS-I一样，将等量的抽样单元或家庭分配到城市和农村地区，以考虑到异质劳动力活动，为城市地区的就业情况提供详细图景。因此，规定的800个抽样单元被平分为两部分，即城市和农村各400个抽样单元。城市地区分为三个层次：城市加德满都谷、其他城市山区和城市塔拉伊，农村地区分为农村山区、农村山区和农村塔拉伊。每个城市和农村地区的400个抽样单元将在各自的层次内按比例分配。在第一阶段，从每个层次中按照规模成比例的概率（PPS）选择了规定的抽样单元（街区或次街区或街区组合），即街区内家庭数量作为规模的衡量标准。在第一阶段使用PPS之后，在第二阶段选择固定数量的家庭，以便为每个现场团队提供方便的工作量。这种设计的好处是，样本原则上具有自加权性。在第二阶段，从城市和农村各层的每个抽样单元（街区）中按照系统随机抽样程序选择20个家庭，总共从800个抽样单元中选择16000个家庭。
抽样误差估计
---------------------------
根据从2003-04年尼泊尔生活标准调查中获取的当前经济活动人口，估计所需的最低样本量为城市244户，农村361户。然而，根据这次新调查中人口就业状况的信息，需要选择的最低家庭数量为城市265户，农村277户。然而，在城市和农村地区各分配了400户，以提供可靠的下层分解估计，直至5岁年龄组（10个组）按性别（2个组）分解。每个城市和农村地区将有总共8000户。

注意：有关抽样设计的详细信息，请参阅最终调查报告的附件A。
数据收集方式
---------------------------
面对面 [f2f]
研究工具
---------------------------
调查的问卷是用尼泊尔语准备的。问卷基于第一次尼泊尔劳动力调查1998/99年使用的调查问卷。问卷中包含了一些额外的主题，例如家庭特征、移民和汇款以及缺席人员信息。
NLFS-I问卷（基于国际劳工组织手册制定，包含77个问题）为NLFS-II问卷的制定提供了坚实的基础。NLFS-II问卷大幅扩展到130个问题，分为八个部分。问卷结构确保受访者（除了家庭户主，他们也有一些额外的问题）通常最多回答大约80个问题，其中大多数人回答的问题数量显著较少。
为了保持可比性，保留了NLFS-I中提出的一些问题。然而，增加了一些重要内容，以解决一些当代问题，例如家庭设施、缺席人口和流入国家的汇款。为了获得比以前更好的回应，对一些问题进行了修改。NLFS-II问卷的八个部分旨在收集关于一般信息、家庭信息、当前活动、失业、过去12个月的活动、过去就业记录、缺席信息和家庭收到的汇款的数据。关于家庭信息、缺席信息和汇款的问题只问家庭户主。
如前所述，对旨在收集关于非正式活动和通常活动的准确可靠信息的题目进行了一些修改。询问通常活动的问题是为了收集过去12个月内在三个不同活动类别（工作、未工作但可供工作、未工作且不可供工作）中他/她花了多少个月的时间。如果一个人在那个月的大部分时间都在工作，那么这个月被认为是“工作”的。如果一个人在那个月的大部分时间没有工作但可供工作，那么这个月被认为是“未工作但可供工作”。这种方法确定通常活动状态比NLFS-I中使用的方法简单，但与国际劳工统计标准保持一致。在家庭信息中，一些问题旨在提供用于监测选定千年发展目标的统计数据。此外，增加了一些相关问题，涉及房屋所有权、烹饪用的能源类型、照明来源和农业用地的大小。问卷副本见附件B，附件C包含一个详细的流程图，可用于识别具有不同特征的人通过问卷的路径。
与NLFS-I一样，关于经济活动的问题的下限年龄为5岁。这使能够收集关于儿童经济活动的数据。这次也保留了使问卷更具性别敏感性的目标。收集了如烹饪、清洁和照看孩子等活动信息，这些活动主要由妇女免费为家庭提供。
最初，草案问卷提交给技术委员会，并经过了几轮讨论。根据来自代表不同利益集团的技术委员会成员的建议和评论进行了改进和修改。此后，问卷进行了多次预测试，并根据每次预测试的反馈进行了彻底的修订。预测试的范围很广，包括不同的生态区、发展区域和城市/农村地区。最终尼泊尔语版本的问卷被翻译成英语，以查看是否可能产生任何误解。在最终用于现场数据收集的尼泊尔语问卷的最终版本中消除了可能的混淆，以减少对每个问题意图含义的误解的可能性。同时，准备了一份访谈者手册，详细说明了每个问题的概念和目标，以便收集可靠和准确的信息。
数据清理操作
---------------------------
本调查使用了集中的数据处理系统。虽然使用分散式处理可能会有一些优势，因为它允许在实地工作后立即进行现场编辑，但正如过去一样，需要仔细控制数据录入和所有编码的验证，尤其是在诸如职业、行业和培训主题等难题上，这在现场很难解决。
整个数据录入活动最多由三个数据录入操作员在12个月内同时进行。数据录入活动由该部门的经验丰富的计算机官员监督。劳动力调查通常使用非常技术性的关键术语定义（例如“当前活跃”），因此，像过去一样，因此编写了CSPro的特殊程序，根据每个人对各种不同问题的回答产生这些派生变量。
响应率
---------------------------
调查的响应率很低，只有24户家庭（16000户中的24户）未收集数据。其中20户是由于一个PSU在远西部地区，该PSU在第三季节无法覆盖，因为选定的家庭已被驱逐到下一个VDC的其他街区，以宣布其为野生动物保护区。驱逐发生在2001年人口普查之后。因此，该地区另外两个选定的PSU的权重在分析阶段进行了调整，以补偿20个缺失的家庭。
抽样误差估计
---------------------------
与所有调查一样，NLFS II的结果受两种主要误差来源的影响。这些可以大致称为抽样误差和非抽样误差。在这样规模的调查中，样本设计的稳健性意味着国家层面的统计数据抽样误差可能相对较小。非抽样误差可能是关注的重点，并且在调查的所有阶段都尽力尽量减少这些非抽样误差。
在STATA应用程序中计算了抽样误差，该应用程序用于处理此调查。为了推导这些抽样误差估计，考虑了调查的结构设计，将抽样单元分配到由三个城市和三个农村层组成的六个层，并在每个层中使用不同的抽样比例。然而，由此产生的抽样误差可能大大高估了真实置信区间的宽度，因为它们没有考虑到设计中包含的非常强烈的区域和生态区隐含分层。
关于抽样误差估计的更多详细信息，请参阅最终调查报告的附件A。