MPCA Vulnerability Model Review

August 2021

August 2021

Overview

​

• Background – why the need for a review
• Current model – development, approach (why used), model
• New model:
• Data used, collection process
• Approach / methods
• Findings Inclusion/exclusion analysis
• Cut-offs
• Next steps
• Q&A

​

Background and Current Model

​

• Why?

​

• Context has changed – less new displacement, more returnees
• Current model coefficients of indicators are based on experts opinion only and not based on evidences.
• Current model not adaptive to different geographical areas ie not considering contextual factors
• Less urgent reason is to try to prepare the ground for linkages with Social Protection schemes.

​

August 2021

The New Model I: Data Collection

​

• Building a PMT requires good data

​

• Opted to use REACH Multi-sector Needs Assessment (MSNA) data, conducted in July/August 2021

​

• Why? Good data – randomised sampling, representative at the Mantika and baladya level, full sectors input, contained all necessary indicators

​

• Contains information on expenditure, collected as cumulative number and as a breakdown

​

August 2021

The New Model II: Approach

​

• Approach used: regression analysis on MCNA data, using per capita monthly consumption (expenditure) as the proxy for vulnerability

​

• Why consumption? Better predictor of household welfare (than e.g. income) – can calculate spending per person

​

• This approach estimates relationships between multiple household characteristics and behaviours and their ability to consume goods and services

​

• E.g., how does consumption vary with household size?

​

• To account for differences between regions, regional datasets can be created, clustered as per the MEB???

August 2021

August 2021

August 2021

The New Model III: Approach

Why expenditure?

The New Model III:

Conducting the Analysis

​

• First step: create variables, both continuous (e.g., household size) and categorical (e.g., has regular income)

​

• Second step: run linear regression

​

• What does this mean? We take an outcome or ‘dependent variable’ (expenditute) and see to what degree other factors (‘independent variables’) impact on the dependent variable

​

• Remove variables that have no significance (p = > 0.05)

​

• Results in a list of significant variables, with ‘coefficients’

​

• Coefficients tell us to what degree the variables effect per capita daily consumption, whether increase or decrease

August 2021

The New Model III:

Selecting the variables

​

• Looking at the available indicators in MSNA, create dummies, continuous, ordinal or categorical variables.
• Dummies 0 / 1 🡪 example: if we have 5 option in “type of shelter” we can decide that 2 of them are not acceptable for dignified living conditions, those will be a 1
• Continuous 0 – infinite 🡪 a classic continuous variable is the family size
• Ordinal 1,2,3,4… 🡪 we assign an order to the available options, for example from best to worst on the employment question: 3 for regular job, 3 for self-employment, 2 for casual worker, 1 for no job
• Categorical A, B, C, D 🡪 we are not aware of the order but we have options which we consider equal: for example living in an house or in an apartment.

​

• Simpler way to use dummies as much as possible!!!!!!!

August 2021

The New Model III:

Selecting the variables

​

• Look at the questionnaire and group questions (eliminate repetitive questions or select among a number or questions the one that provides the best answer)

​

• Consult with the sectors

​

• Group the options available

​

• Decide what kind of variables they are

August 2021

The New Model IV:

Defining Vulnerability

​

• Further refines how we define vulnerability for an MPCA programme:

​

• Always at the household level. All household characteristics presented determine consumption across the whole population in a given region.
• As an inability to consume goods and services OR the particular household attributes which depress consumption

​

• New models are much more complex than the first iteration:

​

• Reflects expanded scope of MPCA programme and more in-depth analysis
• Result of our desire to be more inclusive and reflective of the diversity of the communities we work with

August 2021

Example from Iraq : Northern Iraq

​

Includes: Anbar, Diyala, Kirkuk, Ninewa, Salah al-Din

​

• MCNA contained 14 30-day negative coping strategies, each added as a binary variable
• NCS = Negative Coping Strategy
• CSI = Coping Strategy Index score (7 day)
• HH size = (members) X (score)
• Employment rate is positive, i.e., the higher the rate, the more people in the household working, therefore the score is negative
• HH employment rate =

(working members/size) X (score)

​

August 2021