Income Research Paper Series – Research Paper
Low income Measurement in Canada: What do different Lines and Indexes tell us?
- Main page
- A comparison of LICO, LIMs and MBM
- Low income indexes under alternative lines
- Who fall between the lines?
- Who contributes more to overall low income? A decomposition analysis
- Summary and conclusions
- Tables and figures
- Appendix 1 Methodology
- More information
- PDF version
Low income Measurement in Canada: What do different Lines and Indexes tell us?
Information identified as archived is provided for reference, research or recordkeeping purposes. It is not subject to the Government of Canada Web Standards and has not been altered or updated since it was archived. Please "contact us" to request a format other than those available.
by Xuelin Zhang
While Canada has never had an official poverty line, there are a number of low income lines widely employed to inform public debates and program initiatives. These lines are designed to identify low income individuals from different angles. Together with several aggregate indexes, they may be used to obtain a sensible picture of low income in Canada. But in practice, researches often rely on a single line and a single index. Would we observe the same trend by using multiple lines and indexes? What would be the best practices when different indexes contradict to each other?
This study assesses the existing Low Income Cut-Offs (LICO), Low Income Measures (LIM), and Market Basket Measure (MBM) lines, together with a fixed LIM, by using several distribution sensitive indexes. We found that the low income lines tracked each other well in the long-run. But, in the short-run, they often behaved differently. The same was observed when examining different indexes under the same line. In the long-run, the low income rate, gap, and severity indexes all moved in the same direction. However in the short-run, they sometimes varied in opposite directions, or in the same direction with different magnitudes, suggesting that a single line or index can be misleading in some circumstances.
Examining different low income lines across several disadvantaged groups of individuals, we found that fixed LIM was not as inclusive as MBM, or as capable as MBM in capturing individuals from families headed by recent immigrants, although the groups as a whole contributed more to low income incidence under fixed LIM than under MBM. The result suggests that future development of the LIM lines needs to take regional variations in costs of living into consideration, and that the fixed LIM needs to be re-based periodically.
In Canada, although there has never had an official poverty line, there are several regularly published low income lines in use, including LICO and LIM of Statistics Canada and MBM of Human Resources and Skill Development Canada, to support public debates and policy developments. These lines were designed to identify low income individuals from different angles. But in practice, researchers tend to rely on a single line. This leads to some critical questions: what happens if we apply different lines to the same population? Would we observe a different low income trend? Who falls into low income under one line but not under the others? What happens to groups of disadvantaged individuals if a different line is employed?
On the other hand, although leading poverty researchers have proposed numerous aggregate indexes by which we may answer questions such as how many individuals are in low income, what their income shortfalls are, and how the income shortfalls are distributed, in practice, poverty debates often focus on a single index, namely the "headcount" and tend to compare aggregate indexes over time or across individuals without noticing whether a change of, say, half a percentage point in low income rate is statistically significant. Can these practices be misleading?
To answer these questions on a broad base of low income lines, we also included a fixed or anchored LIM. It is a modification of the existing LIM – which we shall refer to as the variable or floating LIM line hereafter. To classify individuals as low income the MBM uses the cost of a predetermined basket of goods and services, the LICO uses a fixed spending pattern, while the variable LIM uses the median of the contemporary income distribution, and the fixed LIM uses the median of an income distribution in a pre-determined year. By design, fixed LIM and variable LIM complement each other, with variable LIM being a "relative" low income line and the fixed LIM the "real" version of the variable LIM line.
We first assessed the sensitivities of these low income lines by examining a number of distribution sensitive indexes from the Foster-Greer-Thorbecke family. We also compared how different indexes moved under the same line and how different lines interacted with each other. Particular attention was paid to decomposing the aggregate low income incidence into contributions by several disadvantaged groups of individuals who are often referred to as groups at high-risk of social exclusion.
We found that, in the long-run, low income movements under different lines were similar and they were sensitive to business cycle indicators such as unemployment rate. In particular, the fixed LIM tracked LICO and MBM well, suggesting that it can be used as a credible measure to monitor low income trend in Canada. However, in the short-run, the variable LIM line behaved differently. For example, low income rate under variable LIM was flat in certain periods while low income rate under other lines changed significantly. This does not mean that the variable LIM line is misleading. On the contrary, it provides useful information that is not available under other lines. In particular, the flat low income rate under variable LIM in certain periods in Canada suggested that individuals from the lower tail of the distribution did not equally share the benefits of economic growth.
We also examined various low income indexes using one line at a time. Our results indicate that different indexes moved in the same direction in the long-run, and even in the short-run, they changed in the same direction most of the time. However, sometimes they did move in opposite directions or in the same direction with different magnitudes. Thus, although it may not be harmful to focus on a simple index such as the headcount for public debate, it is necessary to look at various indexes together for policy development, and given that various indexes can be produced with little cost, the best practice would be to use multiple indexes.
Examining the capability of different lines in capturing low income individuals, we found that for individuals being captured by MBM, there was a fair chance that other low income lines failed to identify them as being in low income; while for those who were above MBM, there was little chance they would be counted as low income persons by other lines. However, among individuals who were above the fixed LIM line, there was still a non-trivial chance for them to be captured by other lines. Thus, it appeared that the MBM line would capture more individuals than the fixed LIM line, suggesting that a fixed LIM line needs to be re-based periodically to maintain its relevance.
Our decomposition result suggests that disadvantaged individuals as a whole contributed more under the fixed LIM line than under MBM, and fixed LIM was stronger than other lines in terms of its capability to capture individuals exposed to multiple risks of social exclusion. However, the fixed and variable LIM lines were relatively weak in capturing individuals from families headed by recent immigrants, mostly likely due to the fact that recent immigrants overwhelmingly chose to reside in large cities where costs of living were high, indicating that future development of the LIM line should take regional differences in costs of living into consideration.Finally, while comparisons of low income statistics under different lines with different indexes are necessary in order to obtain a more complete picture of Canadians' economic well-being, our exercises in this study suggest that a simple comparison of low income indexes over time or across regions without rigorous statistical test may not be the best practice in low income study.