Causes of Poverty in Mauritius

Keywords: poverty in mauritius, poverty in mauritius essay

Over the past two decades, Mauritius has continuously experienced considerable improvements in both social and economic levels. Mauritius is ranked as an upper middle income country since 2003, with a GDP per capita of $4000. Substantial improvement in life expectancy and literacy, Mauritius has shifted from medium to a high human development country during same period. This is evidenced by the latest UN Human Development Indices: Mauritius is ranked 81 (182 countries) in 2007, with an HDI value of 0.804.

Challenge

Despite these considerable improvements, poverty does exist in Mauritius. Rapid modernization and industrialization has lead to income inequality in the population, leading to an increase in number of pockets of poverty. This is a common phenomenon experienced by most developing countries. Governments or organizations have to reconsider their policies to decrease the level of income inequality in order to eradicate the problem of poverty.

Meeting the challenge

In Mauritius, government has implemented several social welfare programmes to bridge the gap between poor and non-poor. This include the distribution of social aid to needy people, subsidies on basic food item, ZEP programme in schools to enhance level of education, micro-financing to small and medium enterprises, female empowerment in labor market .

In the 2008/09 National Budget, the Government provided Rs 395 million for the setting up of the of the Eradication of Absolute Poverty (EAP) Programme, an integrated development project which targets the households in the 229 pockets of poverty. In 2009, the Government has set up the National Empowerment Fund as an institutional framework to strengthen the role of various policy programmes such as the Empowerment Programme, EAP, Decentralised Cooperation Programme (DCP), etc. Despite the government policies, reforms and actions, poverty is still persistent in Mauritius.

Poverty perceptions

Poverty is a complex issue and multifaceted. This has always been of concern for everybody. Many studies and policy programmes have been put forward to assess poverty situation in the country and also to target the poor. It is worth noting that the proportion of people living below $1.25 per day, international poverty line, is almost negligible in the country. In contrast, past studies have shown that there are people living in severe poverty. The Relative Development Index for administrative regions, Municipal Wards and Village Council Areas (Central Statistics Office, 2000) identified the least developed regions in the country. The Trust Fund of Social Integration for Vulnerable Groups (set up in 2001) identified 229 pockets of poverty. The qualitative study conducted by DCP pointed out that there were people struggling for basic food. As a matter concern poverty has even been linked to a particular ethnic group that is, poverty perceived as ‘malaise creole’.

These perceptions of poverty are evidence that people show their concern and this concern has accentuated the need for good measurement.

Objective of study

The complexity and sensitivity of poverty has accentuated the interest of people in targeting the poor and assessing poverty. Debates, focus group discussions and studies regarding poverty alleviation are still going on. Researchers are looking for new measurement and approaches to assess poverty in the country. Policy makers are looking for high quality inputs to formulate targeted strategies and programmes.

In light of this, this study aims at identifying the determinants of poverty which are very crucial for policy analysis and the design of effective poverty reduction strategies.

Given poverty is a multi-dimensional and cut across various factors, it is of paramount importance to know the factors increasing the likelihood of being poor. So far, there has been no poverty study on the determinants of poverty; most of the studies have mostly dwelled on the profiles of the poor and non-poor based on descriptive statistics. So, this study provides the opportunity to identify the determinants of poverty of the poorest poor using rigorous econometric models.

The study also provides the opportunity to study poverty not as a dichotomy but as a spectrum. There have been several debates on poverty as a dichotomy that is, poor and non-poor (R. Kanbur). Poverty is a spectrum which comprises several grades of being poor. A multinomial logistic regression model is used to analyse the different groups of the poor. This will help policy makers to target the priority areas and formulate proper budgetary measures.

The study also offers the opportunity to assess poverty using the qualitative and quantitative data. The first and last poverty assessment on such data was done in 1996. Today, poverty is assessed on quantitative data only (CSO. Poverty Report). Quantitative and qualitative data allows better understanding of poverty. Today, maximum emphasis is laid on ‘poverty participatory’ that is to hear the voices of the poor. This approach is widely used. However, this study attempts to show that perception survey allows spurious responses and thus, poverty should be assessed on both quantitative and qualitative data.

During the recent years, the study of the poorest poor has been a topic of growing interest for policy makers and researchers. So, an attempt is also made to study the poorest poor people. A threshold on food poverty is derived on the basis of available survey data. The determinants of poverty will be identified using this threshold. It would be desirable to study the ‘malaise creole’. However, given data on ethnicity is not available it is not possible to assess poverty in this particular population group.

The study also analyses the consequences of poverty using the qualitative data of the LCS that is, how poorest poor people had to borrow money from relative/friends, living in poor dwellings etc.

Last but not the least, the study also elaborates the various aspects that need to be addressed to enhance poverty assessment in the country for better formulation of policies and strategies at the conclusion. The welfare of the population and success of Government policies largely depend on the quality of statistics; good quality statistics allows government to monitor and adjust policies to ensure sustainable social and economic development.

This study will, thus, be based on the data collected at the 2006/07 Household Budget Survey (HBS) data and the 2008 Living Conditions Survey (LCS). The HBS is the major source for poverty analysis. Together with income and expenditure data, it contains detailed information on the demographic, educational and economic status of the household members. The LCS is sub-sample survey from the HBS; this survey differs from the usual household survey conducted by the CSO; it is an opinion based survey based on Participatory Assessment approach; together with socio demographic details of the household members, it contains the assessment details of the households with regards to their life style. The information of the LCS survey allows a more in-depth analysis of the living condition of the people.

CHAPTER 2 – COUNTRY PROFILE AND ASSESSMENT

2.1 Background information

The Republic of Mauritius is a group of islands located in the south-west of the Indian Ocean, consisting of two main islands, the island of Mauritius and island of Rodrigues. The island of Mauritius and Rodrigues has a total area of 1865 sq. km. and 140 Sq. Km. respectively. The Republic of Mauritius is a multi-racial country comprising the general population that is, mixed European and African origin, Indo-Mauritians and Sino-Mauritians. The official language being English, but French is widely spoken. Mauritius has been successively a Dutch, French and British colony. It became independent of Britain on 12 March 1968.

Since independence in 1968, Mauritius has achieved considerable progress in both its economic growth and standard of living. In the economic sector, Mauritius has developed from a low-income, agricultural based economy to an upper middle-income diversified economy with growing industrial, financial and tourist sectors. With the continuous expanding economy, Mauritius has continuously moved ahead from the primary to tertiary sector. The share of GDP in agricultural sector has continuously decreased from 7.1% in 1999/2000 to 4.4% in 2008/09; the share of GDP in the tertiary sector has continuously increased from 67.8% to 72.6% during same period. Mauritius is ranked among upper middle-income countries (e.g. example of countries) with a Gross Domestic Product (GDP) per capita of around $4,000 in 2004. Recent figures as at 2007 show that the GDP per capita worked out to above Rs 149,049 (Figure 1); the annual growth rates worked out around 5% for the past three years; the growth of investment (Gross Domestic Fixed Capita Formation) increased significantly from +19.2% in 2006 against -8.3% in 2000. The Foreign Direct Investment (FDI) as a percentage of GDP worked out to 3.5% in 2006 against 1.5% in 2005. Modernisation and expansion of the economy is apparent from the increasing use and availability of telephones (28.6 per 100 people in 2008), mobiles (81.2 per 100 people in 2008), computers (24.2 % of households in 2006) and internet subscribers (15.8 per 100 people in 2008).

Together with economic development, Mauritius has achieved remarkable progress in the social development; the standard of living has changed over the years in terms of increased life-expectancy, lowered infant mortality, high literacy, high participation rate of children in schools, improved infrastructure, leisure and sports etc; Mauritius has a life expectancy at birth of 72 years in 2006 compared to 69 years in 1990; The adult literacy rate rose from 79.9% in 1990 to 84.3% in 2000; the primary school enrolment is almost 100%; the infant mortality rate (per 1,000 live births) dropped from 20.4 in 1990 to 14.4 in 2008; the unemployment rate, being a major concern for the country, has decreased from 9% in 2000 to 7% in 2008; the extreme poverty is almost negligible (less than 1% of the population is found below the poverty line of $1 a day). Due to sustained development in the social sector, Mauritius, second Sub-African country, now stands among high Human Development countries. In 2003, Mauritius shifted from medium development to high with a Human Development Index (HDI) value of 0.800; based on the latest UN Human Development Report 2009, Mauritius has an HDI value of 0.804 and ranked 81 among 182 countries. (Refer figure 2 – HDI trend) According to the progress to tract the 8 Millennium Development Goals , Mauritius, through sustained policies and actions, have already achieved almost 6 goals in the eradication of extreme poor, achievement of primary school enrolment, reduce child mortality, improve maternal health, sustainable environment, and economic development.

Despite these remarkable economic performances and sustained social developments, Mauritius still has to respond to many challenges; there are a number of short and medium term challenges; these relate to productivity, erosion of trade preferences, exchange rate fluctuations, budget deficits and unemployment. Consequently, these are impacting on social development thus leading to environmental degradation, poverty issues such as problem of social exclusions etc. An overall assessment of the poverty situation in the country is highlighted in the following paragraphs.

Poverty

Poverty is not highly prevalent in Mauritius as compared to the Sub-Saharan African countries where millions of people are struggling to live below a $1 a day, millions people dying due to diseases and hunger, millions of children in labour instead of being a school etc. However, poverty does exist in Mauritius; there exist pockets of poverty across the country.

According to the CSO publications on poverty analysis report 2001/02 and 2006/07, it is noted that extreme poverty is almost negligible in Mauritius; the proportion of population living below the $1.25 (PPP) a day, so called US $ 1 a day, is estimated to be less than 1%. As compared to other Sub-Saharan African countries like Zambia, Nigeria, etc., poverty is relatively very low in Mauritius. According to the Millennium Development Goal 1 – Eradicate extreme poverty hunger and the target being to halve the proportion of poor by 2015, Mauritius has already achieved this target. However, an analysis on qualitative assessment on poverty conducted by Decentralised Cooperation Programmes relates that there are Mauritian people who are struggling for basic foods (DCP, 2009).

Mauritius does not have a national poverty line. However, on the basis of relative poverty measurement and data collected at Household Budget Surveys, the poverty situation is assessed by using a poverty line defined as ‘half median monthly household income per adult equivalent’. In 2006/07, the poverty line is estimated at Rs 3,821, around 8.5% of the population is deemed to be poor. The reports relate that poverty is highly prevalent among single member households (10.3%), female headed households (11.9%), one parent households with unmarried children only (13.5%), households with large number of dependent children; heads of households with educational attainment below Standard VI (13.2%) and being inactive (11.0%) were found most vulnerable. The report also highlights that the income disparity between poor and households that is, the household income for the poor (Rs 7,055) was three times lower than that of all households (Rs 22,242); poor households were found highly reliant on government social security benefits that is, basic pensions and social aid. It was noted that if government social security benefits are discontinued the poverty incidence would double; poverty rate would increase from 7.9% to 15.9%).

The report also sheds light on the household tenure of poor persons. It was found that 82% of the poor households owned a dwelling against 92% for all households. In terms of household goods and durables, poor households were more likely to possess television (85%), refrigerator (63%) , fixed telephone (41%); in particular, mobile phone (48.5%); it is worth noting that the proportion of poor households with mobile phones in 2001/02 was almost negligible. As regards principal use of cooking fuels, poor households have already switched off to cooking gas. In 2006/07, nearly 90% of the poor households used cooking gas and thus, only 10% of them had recourse to cheaper fuels like wood and kerosene.

The share of expenditure on food and non-alcoholic beverages, also a measure of economic wellbeing, decreased from 42% in 1986/87 to 32% 2006/07. This implies that people are better off.

The modernization, industrialization and increasing economic growth has lead to the growing income inequality in the population and increasing number of pockets of poverty (NMDG report, 2002). Indeed, this phenomenon is quite common in most developing countries. The Gini Coefficient,a measure of income inequality, dropped from 0.445 in 1980/81 to 0.388 in 2006/07. A Gini nearing to 1 means perfect inequality and 0 no inequality. However, according to the past three Household Budget Surveys, the 2006/07 Gini has deteriorated to some extent (0.387 in 1996/97, 0.371 in 2001/02 and 0.388 in 2006/07) Refer figure 3. The ratio of share of income going to richest decile and share of income going to poorest decile worked at 7.4 in 2006/07 against 7.9% in 1986/87. The unequal distribution of income in the population gives rise to growing number of pockets of poverty. The CSO publication on Relative Development Index based on 2000 Housing and Population Census data shows the administrative regions with least developments. These least developed areas are more concentrated in the island of Rodrigues and the east, west and south part of the island of Mauritius. In 2006, the Trust Fund of Integration of Vulnerable Group has come up with a list of 229 pockets of poverty across the island of Mauritius.

In 2009, the Decentralised Cooperation Programme has come up with a report on qualitative study on poverty assessment. In the overall assessment, the author has highlighted the profiles of the poor and some assessment of policies in the country. The report also presents that poverty is highly correlated with gender, employment, level of income, level of education, geographical areas etc. It is also noted that poverty is also related to ethnicity. The author raised the issue where poverty was characterized as ‘Malaise Creole’. And also that poverty is prevalent among fisherman living in coastal areas. The main assessments of the pilot study are as follows:-

people are finding it difficult to enjoy even a basic diet;

high degree of indebtedness in poor households; and

difficulty in paying utility bills and purchase of basic food items

In 1997, the Appavoo & Associate, together with Data Research Africa has come up with a report on poverty analysis in Mauritius. The report highlighted the poverty incidence in terms of monetary approach in the country, together with an assessment of people perceptions on policies in education, transport, health etc. The report also highlighted the prevalence of poverty in connection with regions, households with large number of dependents, female headed households etc.

2.3 Poverty policies and actions

The eradication of poverty is on the agenda of the government. Government, together with private organisations and assistance of international agencies like the UNDP, IMF, World Bank etc. is making concerted effort to eradicate poverty in the country. Various social welfare programmes and polices have been implemented. Some examples of the social welfare programmes are as follows:-

Distribution of social security benefits – old age pension to ensure proper standard living for elderly people aged 60 years and over, widows pension, invalid pensions, social aid for poor households etc.

Subsidy on flour, ration rice and cooking gas;

Free education at primary and secondary education; distribution of books in primary education; distribution of daily bread ‘pain maison’ in primary schools; distribution of food in selected schools under Zone Education Prioritaire programmes; Industrial and vocational training for children having not passed the final stage of primary education; distribution of computers in schools.

Free health services in government hospital and area health centres; health services through ‘Caravane de Sante’ in different regions of the country; school and domiciliary visit of health personel; Sensitization campaign on HIV in schools and workplaces;

Free transport facility to elderly, invalids and school going children

Low interest housing loan for building of houses;

Empowerment Programme set up in 2006 in view of empowering unemployed people and also women having lost their jobs;

Financial services such as Micro-credit scheme to empower women entrepreneurs

According to figures published by the CSO, the government expenditure on ‘Community and Social Welfare’ worked out around Rs 30 Billion every financial year over a total government expenditure of Rs 50 Billion, thus indicating that Government disburse more than 50% of the government expenditure to social and welfare development.

Together with these social welfare programmes, several poverty alleviation programmes were set up which are as follows:-

Trust Fund for the Social Integration of Vulnerable Group (2001) set up in view of addressing the need of the poor people who are excluded from the main stream of socio-economic development

A Nou Dboute Ensam (1999) aims at promoting subsidies and micro credit schemes to the vulnerable groups.

IFAD, Community Development Programme (2000) aims at bringing disadvantaged people within an organizational framework

Levé Deboute (1999) focuses at income generating activities and community developments in Rodrigues

Decentralised Cooperation Programmes (2006) funded by European Union to fight against poverty Alleviation of poverty in Mauritius and Rodrigues by improving the delivery of social services and complementing the resources of vulnerable groups

National Empowerment Fund (2008) aims to fights against poverty.

Eradication of Absolute Poverty

2.4 Poverty measurement

The measurement of poverty depends on how poverty is perceived. According to the description of poverty assessment in Mauritius, it is clear that there is no single measurement of poverty. Poverty is assessed in terms of ‘Absolute’, ‘Relative’ and ‘Subjective’. The approaches are succinctly described below:-

Mauritius does not have a national poverty for example the ‘minimum vital’ which is frequently updated with price inflation as in the context of absolute poverty. The World Bank $1 purchasing power parity a day international absolute poverty line is found not relevant to the context of Mauritius. The advantage of using an absolute poverty line is that it allows comparison over time thus enhancing continuous poverty assessment and monitoring.

In the absence of the absolute poverty, the CSO uses the relative poverty measurement based on half median household income where adjustment for household size and composition and economies of scale are considered. The relative poverty measurement reveals the prevailing poverty situation for a given time period. This approach is the most commonly used measure particularly in developing countries. The relative poverty measurement still varies because some of the assessments are based on income/expenditure, mean/median income, 40%, 50% or 60% median income etc.

Subjective poverty is the assessment of the poverty situation based on the participatory of the poor persons for example the poverty assessment in 1996 (Appavoo & Associates).

Poverty has also been done on the basis of non-monetary approach that is, other than using income/expenditure data. The Relative Development Index which attempts to identify the least developed administrative regions is based on housing and socio-economic variables at the Housing and Population Census.

CHAPTER 5 – METHODOLOGY

5.1 Introduction

This chapter presents the methodological part of the study. It gives a broad description of the statistical models used and also the determination of the different thresholds used. Given that this study aims to present the determinants of poverty for extreme poverty and also an in-depth analysis of the various sub groups of poor population, two econometric models are being used namely the logistic regression model and the multinomial logistic regression model.

5.2 Regression model

Regression analysis plays an importance role in statistics; it is a very powerful and commonly used technique. This technique provides more meaningful results and conclusions as compared to descriptive statistics. In the context of analysing the determinants of poverty among various explanatory variables, the relevance of using regression model is elaborated on the World Bank website.

5.2.1 Selection of model

Starting with the simplest linear of the General Linear Model (GLM).

The simplest linear regression model that can be used for the analysis is the multiple regression (MR) model where the outcome variable, Y is regressed on a set of predictors X. The MR is in the form of,

Y: outcome continuous variable

X : set of p predictors/ explanatory variables

E : Error term, normally distributed with Mean 0 and variance σ 2

α: intercept term

β: coefficients of explanatory variables

However, given that our dependent/response variable in the analysis is a dichotomous/ categorical variable, the MR is not appropriate. The MR requires the response variable to be continuous and to be normally distributed. In fact, the MR has also been used to identify determinants of poverty, where the response variable was log expenditure of households and Ordinary Least Square was used to estimate parameters.

The regression analysis of categorical response can be made possible by using the models of the Generalised Linear Models (GzLM) family where it relaxes the assumptions of normality. This property of the GzLM has widened the scope of data analysis.

The GzLM is, indeed, an extension of the class of linear model. It provides the opportunity to analyse response variables which follow distributions other than normal distribution; and also the distribution should belong to an exponential family.

Both logistic and Multinomial regression model forms part of GzLM. The application of such regression model is well known in statistics. They are implemented in various fields (educational, health, poverty etc.) to analyse complex data with categorical response variable.

It is worth noting that logistic regression model has largely been used in social sciences since early 1980’s. Its application in social sciences has known no bounds. This technique has been used in education research (success or failure), health(death /survival, with/without of disease) etc. Similarly, in the analysis of poverty, researchers have largely made use of this model and came with useful and effective solutions for decisions makers. The Multinomial model has also been used to show the poverty as a spectrum.

5.2.2 Description of model

LOGISTIC REGRESSION MODEL

Logistic Regression (LR) is a member of the GzLM family where the response variable is dichotomous (1, 0) representing success/failure and presence /absence (Princetone, Chapter 3). It is also called Binary Logistic regression.

So, in this study of the determinants of poverty based on an extreme poverty line, the application of a logistic regression model is relevant where response variable (Y) is binary (poor/non-poor). The explanatory variable /predictor (X) can be either categorical or continuous.

Starting with the simplest logistic model is as follows:

The logistic model predicts the logit of Y from X. The logit is the natural logarithm of odds of Y and odds are the ratio.

The log of the odd ratio, log (π/ 1-π), is the link function called the logit which map the probabilities (0, 1) to (-ω, +ω) that is linearising the distribution and making it unbounded.

The logit, thus, links the response variable (poor/non-poor) to the set of predictors (socio-economic, demographic and housing variables)

Using the logit (π), the intercept (α) and (β) is calculated. And, using antilog, the probability (π) is expressed in the form of

Multiple Logistic Regression

In the case of several predictors, the Multiple Logistics Regression is used. The model is expressed as follows:-

the probability is derived by taking antilog and expressed as follows:-

In this case, for each predictor we have a β; the coefficients in (LR) are estimated using maximum likelihood.

The interpretation of results can be done using the odds ratio or even the probabilities.

π = conditional probability of being poor, P(Y=1 / X1, X2, ….,Xp); it is assumed that the probability of being poor depends on the set of combinations of predictors X.

Y= 1, being poor and Y= 0 being non-poor

the odd ratio is the ratio of the probability to its complement that is, ratio of being poor to non-poor. An odd ratio greater than 1 implies the increase in the likelihood of being poor; if it is less than 1, it decreases the likelihood of being poor.

Evaluation of predictors in model

The contribution of a predictor is assessed by examining the reduction in deviance G statistics, brought by the inclusion of the predictor in the model relative to the null model. The null model, logit (π) = α, is the simplest model with maximum deviance; it indicates that the probability of being poor is constant for all categories. The reduction is deviance is then tested to a chi-sq distribution.

Goodness of Fit Statistics

Goodness of fit of the model is assessed using the Hosmer Lemeshow test. This test is considered more robust than the traditional chi-square test particularly if covariate is in the model or sample size is small. A finding of non significance corresponds to the researcher concluding the model adequately fits the data.

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

MULTINOMIAL LOGISTIC REGRESSION MODEL

The Multinomial Logistic Regression (MLR) model is an extension of the Logistic Regression (LR) model, where the response variable has more than 2 categories. For example, in this study four thresholds of poor are defined which are as follows:-

Poorest – Households with total income below 40% median income

Poorer – Households with total income > 40% median income but less 50% median

Poor – Households with total income > 50% median income but less 60% median

Non-poor- Households with total income >= 60% median

If the Non-poor is chosen as the reference category, the logits for other categories are defined as

Logit (πj) = log (πj/ π4) = XjTβj

j= 1, 2, 3 categories (poorest, poorer and poor)

XT= transpose of set of predictors

Î’ = set of coefficients of explanatory variables

The estimated probabilities are presented as

Estimated πj = Estimated π1exp (XjTβj)

&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&&

5.2.3 Model analysis

In this study, the forward regression is used where the analysis begins with null model and the explanatory variable is added one by one, till the preferred model is generated. After the inclusion of each explanatory variable, the contribution of the variable is measured based on the G statistics (reduction of deviance).

The significance, contribution and interpretation of variables is analysed by considering

positive and negative signs of coefficients of explanatory variables

the z statistics [β/SE(β)]

P values

G statistics – Reduction in deviance

Odd ratios

The Maximum Likelihood is used to estimate the parameters α and β in both regression models.

5.3 Choice of poverty line

Mauritius has no official poverty line. In order to assess poverty in the country, the CSO uses the relative poverty line defined as ‘half median monthly household income per adult equivalent’.

Thresholds for the poorest poor

The poorest poor are those households in the poorest quintile group of household income per adult equivalent which satisfy the following three conditions.

(i) household is having difficulty to obtain daily basic food

(ii) household consume government rice

(iii) household buy food on credit

Food is, in fact, the basic requirement for a person to survive; Government rice is the cheapest rice in the country; if a person is borrowing money to spend on food, it means the person is in severe poverty. at international, food poverty line is used to measure extreme poverty for example UN Millennium Development Goals calls to eradicate hunger worldwide.

Multiple groups of poverty

The 40%, 50% and 60% median of household income per adult equivalent is used. In 2006/07, the 40%, 50% and 60% median income are estimated at Rs3,057, Rs 3,821 and Rs4,585 respectively.

5.4 Statistical package

The logistic regression can be run using many statistical software of which there are STATA, SPSS and MINITAB. For this study, STATA and MINITAB are used; the preparation of data was performed using STATA and the regression model was run on MINITAB.

5.5 Constraints

The constraints obtained when analysing the data is succinctly presented as follows:-

Problems when disaggregating data at lower level; given that the analysis of determinants of poverty has been carried out on a sample of 1,683 households, where 187 were poorest poor households, the analysis of disaggregated data was hindered due to few observations in cells and which subjected to low reliability of estimates. In some cases, some important variables like type of households, number of elderly in households, which are very much linked to poverty were discarded. This constraint has hampered the elaboration of data analysis. This problem of disaggregation is more acute in the analysis of the different subgroups of poverty. When further disaggregating the category of poor by selected variables, very few observations were found in some cells. In order to tackle with this problem, the variables have been aggregated into broader categories.

Problems when combining qualitative and quantitative data, many spurious observations were obtained. Households which were found in high decile group of income at the HBS responded being very poor at the LCS and vice/versa. In this case, the invalid responses were dropped. This also decreases the sample size and thus, we encountered problem as mentioned in (i).

Of course, these constraints hampered the data analysis, since profound analysis could not be done. The lesson from these constraints indicates that a specific poverty survey is required where all the variables related to poverty are collected for example possession of fixed assets (acres of land) and over sampling of households in deprived regions is considered.