Interhousehold Transfers in Urban Papua New Guinea 1982 - Papua New Guinea

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

The 1982–83 study reported here is one of only a handful of studies primarily designed to quantify interhousehold transfers in urban Papua New Guinea. The main alternative sources of quantitative information on transfers are the four large-scale household income and expenditure surveys conducted in 1975–76, 1987–88, 1996 and 2009–10 (see Bureau of Statistics, 1977; Gibson, 1998; World Bank, 2000; and National Statistical Office, n.d.).

This study can be set alongside the large-scale household income and expenditure surveys to provide more fine-grained information on how and why transfers flow and their impact on consumption and poverty. The relevance of the study today is not the kina value of transfers, but the description of transfers and the relationships between transfers and other household and community characteristics.

The study adds to what is known from the large-scale household income and expenditure surveys by focussing on four low-income census units (three settlements and one traditional village in two urban areas) and by including some of the poorest urban households. The field methods were designed to capture transfers in more detail than larger surveys could. Unlike other surveys, the study included meals given and received and overnight hospitality in the definition of transfers. The study also recorded for the donor or recipient of every transfer the relationship to the study household, the birthplace, and place of residence.

The main data collection methods were demographic and economic surveys of all 415 households (2,548 residents) in the four low-income study areas, and two-week income and consumption surveys of a sample of 48 households (295 residents) within those areas.

Although initial findings from the study were issued at the time (Morauta, 1983a and 1984a), the full data and analysis were not published. The purpose of this report is to place a fuller set of data, including data by household for all consumption survey sample households, and a more complete analysis in the public domain.

The analysis of the data in this report mainly follows the original design. However, in two areas, the definition of adequate calorie and protein consumption and the development of poverty lines, the analysis draws on studies since the 1980s, particularly the World Bank poverty assessments (World Bank 2000 and 2004) and the work of Gibson (1998, 2000, 2012, and Gibson et al., 2010).

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

Five urban locations:
-In Port Moresby: Nine Mile, Gordons Ridge, Gerehu
-In Madang: Biliau and Wagol.

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

Household.

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

1) 415 citizen households citizen households from 4 census units with high proportions of households without wage-earners.
2) 26 urban citizen households from one high-income census unit.

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

Sample survey data [ssd]

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

* SELECTION OF STUDY CENSUS UNITS
Analysis of the 1980 census: The locations of urban citizen households without wage-earners in all urban areas were identified through special tabulations prepared by the National Census Office (NCO). These tabulations identified the location of 56,912 households according to the number of wage-earners in each household and the census unit (the smallest locational grouping in the census).
Based on the 1980 census information on the location of households without wage-earners, four census units were selected for the study, where there was a high incidence of households without wage-earners. This was to ensure that the study captured some of the poorest urban households. In this report these four census units are referred to as the low-income areas, or low-income census units. A single high-income census unit was also selected for comparison.

* HOUSEHOLD SURVEYS
The household surveys were designed to provide an up-to-date social and economic description of the selected census units through single interviews with every household. The survey was also designed to provide the frame for sample selection for the consumption surveys.
All 415 households were surveyed in the low-income census units, 100 in Nine Mile, 207 in Gordons Ridge, 65 in Biliau and 43 in Wagol. There was no nonresponse in these census units. In addition, 26 out of the 29 citizen households in the Gerehu census unit were surveyed (with three citizen Gerehu households declining to participate).

* CONSUMPTION SURVEYS
Selection of consumption survey sample: Using information from the household surveys, a sample of 48 households was selected across the four low-income census units for the consumption surveys. Two strata of equal size were selected in each census unit, households with and without wage-earners. A further sample of 11 households was selected from the high-income census unit. The 59 households are those presented in this documentation.
One aim in selecting the consumption survey sample in the four low-income census units was to ensure that poor households within each census unit were included. For this reason, the sample was stratified by whether households had wage-earners. There were six sample households with wage-earners and six households without wage-earners in each census unit, giving a total of 48 households in the four low-income census units.
The second aim was to select each sample of six in each low-income census unit to be as representative as possible of the characteristics of the group of households either with or without wage-earners from which it came in the census unit as determined from the household surveys with respect to:
• province of birth;
• age of household head;
• sex of household head;
• whether the household contained female residents;
• the main source of cash income; and
• for households with wage-earners only, the number and education level of wage-earners.

For each stratum in each low-income census unit, a desired profile was constructed for the six sample households and the households which were the best fit to the profile were identified. Where there was a choice of households fitting the profile, preference was given on the basis of location within the census unit (to make it easier to walk around the sample households in one session) and the language spoken (households where the team did not require an interpreter). Broadly speaking, the profiles were achieved.

In the high covenant census unit, there was no stratification by wage-earner status of households. Otherwise, the variables above were used in the same way as for low-income census units to select the sample. One sample household declined to continue to participate in the consumption survey after commencement, and the
sample was reduced to 11 at that point.
The distinction in the sample selection process between households with wageearners and households without wage-earners did not work out quite as planned. Three households selected as having a wage-earner did not have any income from employment in the two weeks of the consumption surveys and one household had
only low income from casual work. This was a consequence of the instability of low-paid work in these areas. Four people lost their jobs between the household surveys and the commencement of the consumption surveys. To maintain the sampling structure, no adjustment was made for this loss of jobs. There was no comparable problem in the sample of households without wage-earners. They all remained without formal employment during the consumption survey periods.

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

Face-to-face [f2f]

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

For the consumption surveys, the design drew heavily on the Household Expenditure Survey (HES). The main differences were in the detailed questions on transfers and the attention to subsistence produce.
The questionnaires are available for download from this website and contain the following thematic modules:

-Individual characteristics;
-Visits to and from other households;
-Meals given and received;
-Income;
-Types of income;
-Transfers;
-Food consumption;
-Calorie consumption;
-Protein consumption;
-Consumption;
-Poverty.

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

Initial transaction records were analysed in 1982 and 1983 using SPSS to produce data by household. Data by household was edited in 2022 using Excel.

There are two gaps in the data set.
1. There are no data for four households on variables relating to calorie and protein consumption. These households were excluded because they consumed more meals away from home than they ate at home in the 14-day study period.
2. Data on two consumption components for the 11 Gerehu households had not been retained but the consumption totals calculated in 1982 and 1983 in the same way as for low-income households were available for the 1983 report and are comparable.

Further information on the construction of the variables in the data set can be found in Appendix II of the full study report which is available for download on this site.

摘要
---------------------------
此处报告的1982-83年研究是极少数旨在量化巴布亚新几内亚城市地区家庭间转移的唯一研究之一。关于转移的定量信息的主要替代来源是1975-76年、1987-88年、1996年和2009-10年进行的四次大规模家庭收入和支出调查（参见统计局，1977年；Gibson，1998年；世界银行，2000年；和国家统计局，不详）。

本研究可以与大规模家庭收入和支出调查并列，以提供更精细的信息，关于转移的流动方式及其对消费和贫困的影响。本研究的现实意义不在于转移的基那价值，而在于对转移的描述以及转移与其他家庭和社区特征之间的关系。

本研究通过聚焦于四个低收入人口普查单位（两个城市地区的三个定居点和一个传统村庄）以及包括一些最贫困的城市家庭，为大规模家庭收入和支出调查已知的信息增添了新的内容。研究设计旨在比大型调查更详细地捕捉转移。与其它调查不同，本研究在转移的定义中包括了提供的和接受的餐食以及过夜款待。对于每一次转移的捐赠者或接受者，研究还记录了其与调查家庭的关系、出生地以及居住地。

主要的数据收集方法是对四个低收入研究区域内所有415个家庭（2,548位居民）进行的人口和经济调查，以及对那些区域内的48个家庭（295位居民）进行的两周收入和消费调查。

尽管本研究的部分初步发现已在当时发布（Morauta，1983a和1984a），但完整的数据和分析并未公开出版。本报告的目的是将更全面的数据集置于公共领域，包括所有消费调查样本家庭的家庭数据，以及更深入的分析。

本报告中的数据分析主要遵循原始设计。然而，在两个领域，即充足卡路里和蛋白质消费的定义以及贫困线的制定，分析借鉴了1980年代以来的研究，特别是世界银行的贫困评估（世界银行，2000年和2004年）以及Gibson（1998年、2000年、2012年）的工作。

地理覆盖范围
---------------------------
五个城市地点：
- 在莫尔斯比港：九英里、戈登岭、杰雷胡
- 在马当：比利亚乌和瓦戈尔。

分析单元
---------------------------
家庭。

总体
---------------------------
1) 415个公民家庭，这些家庭来自4个人口普查单位，其中没有工资收入者的家庭比例很高。
2) 26个来自一个高收入人口普查单位的城镇公民家庭。

数据类型
---------------------------
样本调查数据 [ssd]

抽样程序
---------------------------
* 研究人口普查单位的选定
1980年人口普查的分析：通过国家人口普查办公室（NCO）准备的特别分类，确定了所有城市地区没有工资收入者的城镇公民家庭的地理位置。这些分类根据每个家庭中工资收入者的数量和人口普查单位（人口普查中最小的地理位置分组）确定了56,912个家庭的地理位置。
基于1980年人口普查中关于没有工资收入者的家庭地理位置的信息，选定了四个具有高比例没有工资收入者的家庭的人口普查单位进行研究，以确保研究能够捕捉到一些最贫困的城市家庭。在本文档中，这四个人口普查单位被称为低收入地区或低收入人口普查单位。还选择了一个高收入人口普查单位进行比较。

* 家庭调查
家庭调查旨在通过每个家庭的单独访谈，提供所选人口普查单位最新的社会经济描述，并为消费调查的样本选择提供框架。在低收入人口普查单位中，对所有415个家庭进行了调查，其中九英里100个，戈登岭207个，比利亚乌65个，瓦戈尔43个。在这些人口普查单位中没有出现非响应情况。此外，在杰雷胡人口普查单位的29个公民家庭中有26个接受了调查（有3个杰雷胡公民家庭拒绝参与）。

* 消费调查
消费调查样本的选择：使用家庭调查的信息，从四个低收入人口普查单位中选择了48个家庭进行消费调查。每个人口普查单位中选择了两个大小相等的层，即有和没有工资收入者的家庭。从高收入人口普查单位中又选择了11个家庭。这59个家庭就是本文档中呈现的。
在四个低收入人口普查单位中选择消费调查样本的一个目的是确保每个人口普查单位中的贫困家庭都被包括在内。因此，样本按是否有工资收入者进行了分层。每个人口普查单位中有6个有工资收入者和6个没有工资收入者的样本家庭，总共在四个低收入人口普查单位中有48个家庭。
选择的第二个目的是在每个低收入人口普查单位中选择每个样本的6个家庭，尽可能地代表从人口普查单位中选出的有或没有工资收入者的家庭群体的特征，这些特征是根据家庭调查确定的，包括以下方面：
• 出生省份；
• 家庭户主年龄；
• 家庭户主性别；
• 家庭中是否包含女性居民；
• 现金收入的主要来源；
• 对于仅有的工资收入者的家庭，工资收入者的数量和教育水平。

对于每个低收入人口普查单位的每个层，为六个样本家庭构建了一个期望的轮廓，并确定了最符合轮廓的家庭。在有多个符合轮廓的家庭可供选择的情况下，优先考虑的是人口普查单位内的地理位置（以便在一次会话中更容易走访样本家庭）以及使用的语言（团队不需要翻译的家庭）。总的来说，轮廓已经实现。

在高收入人口普查单位中，没有按家庭是否有工资收入者进行分层。否则，上述变量与低收入人口普查单位相同，用于选择样本。有一户样本家庭在开始消费调查后拒绝继续参与，样本因此减少到11户。
在样本选择过程中，在拥有工资收入者和没有工资收入者的家庭之间的区别并没有完全按计划实现。有三个被选为有工资收入者的家庭在消费调查的两周内没有任何来自就业的收入，有一户家庭只有来自临时工作的低收入。这是这些地区低收入工作不稳定性的结果。在家庭调查和消费调查开始之间，有四个人失去了工作。为了保持抽样结构，没有对这种失业情况做出调整。在无工资收入者的样本中没有出现类似的问题。他们在消费调查期间都没有正式就业。

数据收集方式
---------------------------
面对面 [f2f]

研究工具
---------------------------
对于消费调查，设计主要借鉴了家庭支出调查（HES）。主要区别在于关于转移的详细问题和对生存生产的关注。
问卷可以从本网站下载，并包含以下主题模块：
- 个人特征；
- 访问其他家庭；
- 提供和接受的餐食；
- 收入；
- 收入类型；
- 转移；
- 食物消费；
- 卡路里消费；
- 蛋白质消费；
- 消费；
- 贫困。

数据清洗操作
---------------------------
最初的事务记录于1982年和1983年使用SPSS进行分析，以生成家庭数据。家庭数据于2022年使用Excel进行了编辑。
数据集中有两个缺口。
1. 没有关于四个家庭与卡路里和蛋白质消费相关的变量数据。这些家庭被排除在外，因为在14天的研究期间，他们在家的餐食比在外就餐的餐食少。
2. 11户杰雷胡家庭的两个消费组成部分的数据没有保留，但与低收入家庭相同的方式在1982年和1983年计算的消费总额可用于1983年的报告，并且是可比的。
关于数据集中变量构建的更多信息，可以在完整研究报告中附录II中找到，该报告可在本网站上下载。