Geranda Notten, University of Ottawa Julie Charest, Statistics Canada An evidence-informed rule to setting a poverty threshold - Illustrated with Canadian data - Geranda Notten, University of Ottawa Julie Charest, Statistics Canada Canadian Economics Association - Annual Conference 2018
Where to set the poverty threshold? Canadian Economics Association - Annual Conference 2018
Commonly used decision rules for setting the poverty threshold Norms Composite poverty indicators: union, intermediate & intersection identification criteria for second threshold Single poverty indicators: poverty concept Problem: either very extreme or leaving lots of space for arbitrary decisions Budgetary or political Threshold depends on government budget for poverty reduction or on the political message of a particular government at a particular time Problem: inappropriate for measuring & tracking a societal phenomenon Practical Threshold depends on past practice, convention or other poverty indicators Problem: the nature of poverty can change over time, may be different for different groups or societies, and conceptually distinct indicators may have different strengths and weaknesses in measuring poverty Canadian Economics Association - Annual Conference 2018
Canadian Economics Association - Annual Conference 2018 Argument Setting the threshold is usually the most influential choice when measuring poverty Commonly used selection rules either leave much room for arbitrary choices or are problematic We develop an evidence-informed rule to setting the poverty threshold, which: Reduces the scope for arbitrary decisions Relies on theory and data on the drivers & consequences of poverty Works for unidimensional poverty indicators Works when your data hold extra information closely associated with poverty Canadian Economics Association - Annual Conference 2018
An evidence-informed decision rule Review your data for additional information that, according to theory and empirical evidence, is closely associated with the poverty concept that you are measuring (i.e. a direct driver or consequence) Set up a cross-validation frame that lists the attributes available in the data and their association with the likelihood of poverty (i.e. high / low) Apply this to your observations so that each gets assigned a value for his/her likelihood of poverty Look for the poverty threshold that optimizes the trade-off between (likely) false negatives and (likely) false positives Canadian Economics Association - Annual Conference 2018
Illustration: a marginal increase of the poverty threshold…. Likely false negatives Likely false positives Canadian Economics Association - Annual Conference 2018
Illustration with Canadian data (I) Canadian Survey of Economic Well-being (2013) One time survey, supplement to Labour Force Survey The only representative material deprivation data for Canada Material deprivation is an objective, unidimensional, outcome-based poverty indicator In Notten, Charest & Heisz (2017) we provide the first material deprivation estimates for Canada See paper for more details on data, methodology & results Canadian Economics Association - Annual Conference 2018
Material deprivation – indicator operationalization Step 1: Derive deprivation items from 17 survey questions “Are you and your household able to ….” If yes → Not deprived If no → Follow up question “Is this because you cannot afford it, or for some other reason?" ” If cannot afford → Deprived If other reasons → Not deprived Includes deprivation items such: eating fresh fruits & vegetables once a day, get regular dental care, cover an unexpected expense of $500 from own resources Step 2: Aggregate deprivation items into a composite poverty indicator Applying a counting approach, dual cut-off & a headcount poverty measure Our big issue: Where to set the poverty threshold? Canadian Economics Association - Annual Conference 2018
Where to set the poverty threshold? Canadian Economics Association - Annual Conference 2018
Materially deprived persons in Canada and their likelihood of poverty Variable Status Material deprivation Deprived Income Low income Low-middle income High(er) income Economic hardship Yes No Group A B C D E F Likelihood of being poor Highest High Low Lowest Canadian Economics Association - Annual Conference 2018
Material deprivation and (changes in) the likelihood of poverty Canadian Economics Association - Annual Conference 2018
Material deprivation and (changes in) the likelihood of poverty Canadian Economics Association - Annual Conference 2018
Material deprivation and (changes in) the likelihood of poverty Canadian Economics Association - Annual Conference 2018
Material deprivation and (changes in) likelihood of poverty Optimal threshold Canadian Economics Association - Annual Conference 2018
Canadian Economics Association - Annual Conference 2018 Conclusion Poverty is a normative concept so normative judgements are unavoidable Our decision rule makes some those judgements more evidence-based, more transparent, while also reducing the scope for arbitrary choices Be they politically, budget, convenience or belief driven The rule works For unidimensional poverty indicators (low-income & material deprivation) When your data hold extra information closely associated with poverty When the task involves introducing new or updating old poverty indicators Limitations Optimal threshold depends on information & poverty risk assessment Sample sizes typically too small for statistically significance of differences between false negative & false positive groups (esp. between neighbouring thresholds) Canadian Economics Association - Annual Conference 2018
Canadian Economics Association - Annual Conference 2018 References Notten, G. & Charest, J., An evidence-informed rule to setting a poverty threshold, (soon to be finished) draft manuscript, please contact gnotten@uottawa.ca if interested in a copy. Notten, G., Charest, J., & Heisz, A. (2017), Material deprivation in Canada. Faculty of Social Sciences, University of Ottawa, 1-59. Canadian Economics Association - Annual Conference 2018
Geranda Notten gnotten@uottawa.ca Thank you! Geranda Notten gnotten@uottawa.ca Canadian Economics Association - Annual Conference 2018