The Cambodian government updated the poverty line in April 2013, based on Cambodia Socio-Economic Survey (CSES) 2009. In a recent post, I have updated poverty indicators, referred to the latest poverty assessment by the World Bank. This post will review a technical side of the poverty assessment.

The government previously used the poverty line set in 1997 based on data from of CSES 1993-1994 but found that it did not capture the reality of recent Cambodian livelihoods. The major change is made in the component of consumption basket and the increase in food consumption threshold by 100 K-Calories. The detailed calculation method of new poverty line is available at Ministry of Planning (pdf).

How much does it differ poverty headcounts?

The graph shows the world bank estimation published in the latest poverty assessment. The new estimation is clearly higher than the old one.

Why is the estimation different between World Bank and the government?

It is not very clear but I guess there are two reasons: data cleaning and imputation. Firstly, it is in the black box how both parties have done data cleaning before starting calculation. From the technical note of poverty assessment, the bank seems to remove some outliers in their standard. Secondly, the government take actual value in aggregating consumption while the bank appears to impute some of the missing values, particularly on housing rent component. I think this is the main reason for the estimation gap. If you look at the datasets over time, you would find more missing values in CSES 2004 and less in CSES 2010 (the imputed value is usually large). Consistently, the gap becomes smaller in 2010 in the above graph.

Reference

• Cambodia Ministry of Planning (2013) New Poverty Line in Cambodia.
• World Bank (2013) Where Have All The Poor Gone? Cambodia Poverty Assessment 2013.