Data
1.5.1.1 Household Samples
This thesis uses primary data from 422 households and 17 recruitment agencies. For the household survey, we selected provinces that are involved in large-scale international migration activity, drawing on data compiled by the Provincial Department of Planning (2015) and the Ministry of Planning (MoP, 2015). Three provinces, located in the northern region of Cambodia, and known to be the most common provinces of origin of migrant workers, have been selected. These three, namely Banteay Menchey, Battambang, and Siem Reap (See Figure 1.1), represent the highest proportion of international migration from Cambodia (MoP, 2015; IOM, 2022). The three provinces account for 53% of all international migrants from Cambodia (MoP, 2015; Dickson & Koenig, 2016).
Figure 1.1 Administrative Map of Cambodia showing Survey Location within the Square
Note: Banteay Menchey, Siem Reap, and Battambang Province. Source: NIS (2020)
Multi-stage stratified random sampling is utilised to determine household location and selection.
According to the MoP (2013), 81% of migrants come from rural villages while 19% come from urban areas. Consequently, rural and urban classification is required for district (Srok) and commune (Khum) selection. The designation is based primarily on the classification provided by the MoP (2011).Based on the above MoP information (2013), probability proportional to size (PPS) sampling is used to provide the household sample distribution in each commune. A total of 12 villages were randomly drawn from the selected communes (Table 6). Based on MoP (2017),it is suggested that approximately 21% of households reside in urban areas and 79% in rural areas. Therefore, the sample size is proportionally distributed based on that information.
Thailand Laos
Vietnam
Table 1.1. Sample Distribution by Migration Status
Province Number of
villages
Non-migrant households
Migrant households
Total samples
Banteay Menchey 6 90 52 142
Battambang 6 96 49 145
Siem Reap 5 89 46 135
Total 17 275 147 422
Note. Data collection was conducted from 10th December 2019 to 18th December 2019.
Source: Author’s fieldwork
Table 1.2. Sample Distribution by Village
Provinces Village Non Migrant HH Migrant HH Total Samples
Banteay Menchey
Banoy 16 9 25
Keab 8 15 23
Kork Svay 10 5 15
Russie Kroak 23 5 28
Tuek Thla 20 5 25
Tomnob Chrey 13 13 26
Battambong
Kroper Chueng 14 7 21
Kroper Tboung 9 11 20
Sampov Lech 24 11 35
Ou Many I 14 11 25
Ou Many II 15 4 19
Samnanh 20 5 25
Siem Reap
Chambak Hae 25 8 33
Chanleas Dai 19 30 49
kork Thnot 18 0 18
Rolum Svay 15 6 21
Trapeang Ses 12 2 14
Total 275 147 422
Note. Data collection was conducted from 10th December 2019 to 18th December 2019.
Source: Author’s fieldwork
1.5.1.2 Recruitment Agency Samples
A key reason for undertaking a recruitment agency survey is to obtain migration cost information that cannot be observed or gathered from a household survey. The fee charged was then computed as a cost relative to the household’s wealth. This type of cost is primarily used in Chapter 2 of this thesis.
The Ministry of Labour and Vocational Training (MoLVT) granted licences to private recruitment firms providing recruitment services for labour migration overseas to Thailand, Malaysia, South Korea, Hong Kong, Saudi Arabia, Japan and Qatar (MoLVT, 2019). Since this study focuses on the SSM aspect, the sampling design for recruitment agencies is based primarily on recruitment firms providing services to Thailand, which is considered a South-South labour migration corridor (Bylander, 2017). To identify registered recruitment companies, the study used a list provided by MoLVT (2019a), and the MoLVT Prakas (Cambodian ministerial order) on Private Recruitment Agency MoLVT (2013) that indicates the regular legal status of a recruitment agency sending workers abroad. We used a separate version of the household questionnaire to obtain information concerning recruitment fees and migration costs through official channels. We selected only 30 registered companies based in Phnom Penh that are responsible for managing and sending labour migrants to Thailand. However, only 17 out of 63 active PRAs participated in the survey (see Thesis Appendix for a list of PRAs).
Research Methods
This thesis encompasses four primary research papers; each emphasises a separate research method and technique to attain research objectives. In the first paper, to address research problems, the alternative-specific conditional logit model is used to gauge the effect of the direct monetary costs of migration on decisions about migration channels. However, as the costs of migration appear to be
endogenous, the control function method is employed to estimate and predict the generalised residual from the costs’ function. The predicted residual is then utilised by substituting in the migration channel decision function. After estimation, the average and conditional marginal effects are employed to measure the changes in migration costs that could affect migration channel decisions.
Finally, we use cost simulation to see how changes in cost affect migration decisions.
The second research paper investigates the effect of the uptake of microcredit on migration decisions.
Since borrowing choices are subject to selection bias, we first estimate the migration decision function using the instrumental variable probit model, which differentiates between borrowers and non-borrowers based on formal and informal credit. To enhance our estimation, we employ the endogenous switching probit (ESP) model, allowing us to account for the differences between borrowers and non-borrowers embedded in the migration decisions function. We provide a robustness test using the seemingly unrelated bivariate probit model and a simple probit model. These approaches permit us to account for the coexistence of official and unofficial loan sources from which households can obtain finance. After instrumentation and using a simple probit model, we are able to examine the effect of borrowing on migration decisions.
To gain insight into the impact of remittances on household indebtedness, the third paper employs several econometric approaches for estimating this impact and for dealing with any bias that might yield inconsistent results. First, the Heckman selection model and the two-stage least square (2SLS) regression model are employed to assess the determinants of remittances. Then, the generalised residual is predicted and substituted into the structural equations of household indebtedness. Finally, we check the robustness of our result, using the instrumental variable Tobit model (IV-Tobit) that estimates directly the selected instrumental variables and the outcome variables.
In light of the present dire situation caused by the COVID-19 global pandemic, the last chapter of the thesis focuses on labour mobility challenges and prospects in the Cambodia–Thailand migration corridor. This paper uses the latest current research, policy papers, and government and official reports from the Asian Development Bank, IOM, ILO, the United States Agency for International Development, the World Bank, United Nations Development Programmes, other international non- governmental organisations (INGOs) and local non-governmental organisations (NGOs), together with grass-roots reports focusing on contemporary labour migration. All these are employed to analyse how pre- existing labour mobility challenges have delayed responses aimed at curbing the adverse effects of COVID- 19. Furthermore, journal articles related to the impact of COVID-19 are carefully evaluated in order to pinpoint the determinants of labour migration as well as the magnitude of the influence of COVID-19 on the management of this migration.
Ethical Approval
Ethical approval (WMS 19/53) was granted by the Waikato Management School Human Research Ethics Committee on 7 August 2019.