Insights on Canadian Society
The association between skills and low income

Warning View the most recent version.

Archived Content

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 Andrew Heisz, Geranda Notten and Jerry Situ

Release date: February 24, 2016

Skip to text

Text begins

Overview of the study

This article explores how skill proficiencies are related to household income for Canadians aged 16 to 65 using data from the first wave of the Longitudinal and International Study of Adults (LISA), conducted in 2012. The article also demonstrates how the relationship between skill level and low income changes after controlling for other characteristics known to increase the risk of low income.

  • In 2012, 17% of Canadian adults aged 16 to 65 had a literacy score corresponding to level 1 and below, meaning that they could only find single pieces of information in shorts texts or only had a basic vocabulary. About 13% were in the two highest categories of literacy skills (level 4 and level 5).
  • The median household income for individuals who were in the lowest category of literacy proficiency (level 1 and below) was $49,700, compared with $84,600 among those who were in the two highest categories (level 4 and level 5).
  • Among individuals in the lowest category of literacy proficiency in 2012, 29% were in a low-income household (households whose income is below the after-tax Low Income Measure), compared with approximately 8% for those in the two highest categories.
  • After controlling for other characteristics known to increase the risk of low income, individuals who were in the lowest level of literacy proficiency continued to have a low-income rate higher than individuals who were in the highest levels. The magnitude of the difference, however, was smaller (12 percentage points instead of 21).
  • Differences in skill level also help explain part of the higher incidence of low income among certain groups, including groups with low educational attainment and recent immigrants.

End of text box

Introduction

Certain population characteristics such as immigrant status, being an Aboriginal person, having an activity limitation, having low education and being in a lone-parent family have been found to be concentrated among the low-income population.Note 1 This information could be useful to policymakers since understanding the characteristics of individuals in low income can inform the design of effective programs.

One population characteristic that has received less attention is skill level, which refers specifically to fundamental literacy, numeracy and problem-solving skills. Intuitively, skills should also be negatively correlated with low income. Presumably, persons with lower skills have lower wages and thus a higher likelihood that their income falls below the low-income threshold. However, there is little evidence on the scope of these relationships.

Other research has shown that skills have a positive effect on individual earnings even after controlling for other characteristics such as educational attainment.Note 2 Building on this evidence, this article examines the relationship between skills and household income. Specifically, this study examines whether having low skills increases the likelihood of living in a low-income family, and whether the effects persist after controlling for other factors known to increase the risk of low income.

The study uses data from the Longitudinal and International Study of Adults (LISA). LISA is unique in that it combines skill data from the Programme for the International Assessment of Adult Competencies (PIAAC) with household income information (see Data sources, methods and definitions). The skills data consist of an assessment in each of three domains: literacy, numeracy and problem solving in technology-rich environments. As done in other studies, the focus is on literacy scores because of a high correlation between the three domains.Note 3 The three domains assessed in PIAAC rely on the same core cognitive processes,Note 4 therefore using only numeracy or problem-solving scores would produce similar results.

LISA interviews were conducted between November 2011 and June 2012, whereas income information refers to the 2011 calendar year. This article refers to this collection period as “2012” for brevity.

Literacy skills and income

In 2012, 17% of Canadians aged 16 to 65 had a literacy score that placed them in level 1 and below (Table 1). Individuals in this category, for example, can only locate single pieces of information in shorts texts in the absence of other distracting information, or demonstrate only basic vocabulary. Remaining Canadians scored in higher categories with 13% achieving level 4 and level 5. Scoring in this category indicates a higher literacy proficiency, with individuals in it demonstrating, for example, being able to integrate information from multiple dense texts and being able to reason by inference.

Table 1
Distribution, median income and low-income rate of Canadians aged 16 to 65 by PIAAC literacy level, 2012
Table summary
This table displays the results of Distribution All, PIAAC literacy level, Level 1 and below (low), Level 2, Level 3 and Level 4 and level 5 (high), calculated using percent and dollar units of measure (appearing as column headers).
  All PIAAC literacy level
Level 1 and below (low) Level 2 Level 3 Level 4 and level 5 (high)
percent
Percent with this literacy level 100 17 32 37 13
  dollar
Median household income (in 2011) 67,958 49,696 63,520 74,568 84,583
  percent
Low-income rate (in 2011) 16 29 17 11 8

Literacy skill level and household income are positively related. At $84,600, the median household income for individuals with the highest literacy proficiency (level 4 and level 5) was 70% higher than it was for those with lowest literacy proficiency (level 1 and below), and 33% higher than for those in the level 2 proficiency category.

The bottom row of Table 1 shows the low-income rates, using the after-tax Low Income Measure (LIM–AT). With the LIM–AT, individuals are considered to be in low income if they are living in a household whose income (adjusted for the size of the family) is lower than 50% of the overall median.

The results show a clear association between skill level and the incidence of low income. The low-income rate for the highest literacy proficiency group was, at 8%, one-half the level observed in skill-proficiency group 2 and nearly one-quarter the level seen in the lowest proficiency group. In contrast, the low-income rate was about 29% among those who were in skill-proficiency group 1 or below.

The subgroup with both lower skills and low income can be considered at a particular disadvantage.Note 5 Estimates suggest that, in 2012, this group represented about 5% of Canadians aged 16 to 65.Note 6

Skills and other factors associated with low income

Research on low income has emphasised certain at-risk groups. These are demographic groups that have higher low-income rates, such as recent immigrants, Aboriginal persons, unattached non-elderly persons and people with activity limitations.

In each case, at-risk groups have a skill profile that tends to be closer to having a lower literacy level (Chart 1). For example, while 17% of all persons had a literacy score of in the lowest category, 30% of recent immigrants, 26% of Aboriginal persons, 27% of unattached non-elderly persons and 23% of people with an activity limitation had a literacy score level in the lowest category in 2012.

Chart 1 of article 14322

Description for Chart 1
Data table for chart 1
Table summary
This table displays the results of PIAAC literacy level for groups at high risk of low income PIAAC literacy level, Level 1 and below (low), Level 2, Level 3 and Level 4 and level 5 (high), calculated using percent distribution units of measure (appearing as column headers).
  PIAAC literacy level
Level 1 and below (low) Level 2 Level 3 Level 4 and level 5 (high)
percent distribution
All individuals 17.4 32.4 37.0 13.2
Recent immigrantsNote 1 for chart 1 table1 29.5 33.5 28.3 8.7
Lone parents 22.7 35.9 32.4 9.1
Aboriginal people 25.7 34.5 32.7 7.0
Unattached non-elderly personsNote 2 for chart 1 table2 27.5 34.7 28.8 9.1
People with activity limitations 23.3 32.9 32.3 11.6

Education and skills are also highly correlated—the generation of skills is an important output of the educational system, and the presence of basic skills such as literacy are important prerequisites to entering higher education. For example, among individuals with a high school diploma as their highest education level, 8% scored a level 4 or higher on the literacy assessment compared with 28% among those with a university degree or higher (Chart 2).

Chart 2 of article 14322

Description for chart 2
Data table for Chart 2
Table summary
This table displays the results of PIAAC literacy level by highest level of education PIAAC literacy level, Level 1 and below (low), Level 2, Level 3 and Level 4 and level 5 (high), calculated using percent distribution units of measure (appearing as column headers).
  PIAAC literacy level
Level 1 and below (low) Level 2 Level 3 Level 4 and level 5 (high)
percent distribution
Highest education level  
All individuals 17.4 32.4 37.0 13.2
No certificate, diploma or degree, or non-response 44.4 35.7 18.0 1.9
High school diploma or equivalent 17.3 38.5 36.6 7.6
College, CEGEP or other non-university certificate or diploma 13.2 35.2 41.0 10.6
University certificate or diplomaNote 1 for chart 2 table1 6.5 21.2 44.0 28.3

Educational attainment, however, is not perfectly correlated with literacy levels—some people with a university degree had lower literacy scores,Note 7 while others had higher literacy scores despite having only a high school education.

The relationship between low income and skills after controlling for other factors

The previous section showed the relationships between skills and low income. It showed that skills were related to demographic characteristics, such as educational attainment, as well as whether the individual was a member of an at-risk group. In this section, a regression model is used to test whether the relationship between skills and low income persists once controls for other characteristics common to low-income populations are applied.

Several low-income models are tested. Each model controls for a different number of factors that are expected to affect low income such as educational attainment, immigrant status, the presence of activity limitations, family type and other demographic characteristics. The strategy is to compare the regression coefficient estimates across multiple models to see if the relationship between skills and low income persists as various control factors are added.

The descriptive statistics for the variables included in the regression models are shown in Table 2. Low income was higher for persons with less education (19% among persons with at most a high school diploma versus 9% among university degree holders). Immigrants who arrived in Canada in the 10 years preceding the interview had a low income rate of 25%, which was higher than the 15% rate for the Canadian-born. Persons with activity limitations had a low-income rate twice that of those without a limitation (27% versus 13%). Unattached persons (defined as persons not in a census family) and persons in a lone-parent family each had a low-income rate three times larger than that of couple families (about 30% versus 9%).

Table 2
Percentage of individuals whose household income was below the after-tax Low Income Measure (LIM–AT), 2012
Table summary
This table displays the results of Percentage of individuals whose household income was below the after-tax Low Income Measure (LIM–AT) Percent of population and Percent below LIM (in 2011) (appearing as column headers).
  Percent of population Percent below LIM (in 2011)
PIAAC literacy level  
Level 1 and below 17.4 29.3
Level 2 32.4 16.7
Level 3 37.0 11.1
Level 4 and level 5 13.2 8.3
Highest education level  
No certificate, diploma or degree 16.0 26.5
High school diploma or equivalent 25.9 18.7
College, CEGEP or other non-university certificate or diploma 31.3 13.3
University certificate or diploma 26.8 9.2
Immigrant  
Yes – immigrated after 2002 6.4 24.8
Yes – immigrated in 2002 or earlier 13.8 16.1
No 79.8 14.9
Activity limitation  
Activity limitation 19.3 26.5
No activity limitation 80.7 13.1
Family type  
Couple family 67.6 8.7
Lone parent 10.2 30.7
Person not in a census family 22.2 30.2
Aboriginal people  
Yes 3.2 22.7
No 96.8 15.5
Unattached non-elderly personsNote 1 of table 21  
Yes 9.4 33.5
No 90.6 13.9
Sex  
Male 50.1 15.2
Female 49.9 16.3
Age  
16 to 19 7.1 14.9
20 to 24 10.3 25.6
25 to 34 20.1 14.3
35 to 44 19.6 14.0
45 to 54 22.7 14.5
55 to 65 20.3 15.6
Province  
Atlantic 6.8 19.0
Quebec 23.2 18.0
Ontario 39.2 16.6
Prairies 17.9 10.2
British Columbia 12.9 14.8

The results of the regression exercise are shown in Table 3. The coefficients represent the increased or decreased probability of being in low income relative to the omitted category. Five models are shown, each with a different set of independent variables.

Table 3
Regression models with low income as a dependant variable
Table summary
This table displays the results of Regression models with low income as a dependant variable Regression model, calculated using Model 1, Model 2, Model 3, Model 4 and Model 5 units of measure (appearing as column headers).
  Regression model
Model 1 Model 2 Model 3 Model 4 Model 5
PIAAC literacy level  
Level 1 and below 0.213Note with 3 asterisk for table 3*** 0.162Note with 3 asterisk for table 3*** 0.149Note with 3 asterisk for table 3*** 0.121Note with 3 asterisk for table 3*** Note ...: not applicable
Level 2 0.085Note with 3 asterisk for table 3*** 0.058Note with 3 asterisk for table 3*** 0.050Note with 3 asterisk for table 3*** 0.042Note with 2 asterisk for table 3** Note ...: not applicable
Level 3 0.028Note with 1 asterisk for table 3* 0.015 0.011 0.010 Note ...: not applicable
Level 4 and level 5 (ref.) Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable
Highest education level  
No certificate, diploma or degree Note ...: not applicable 0.124Note with 3 asterisk for table 3*** 0.134Note with 3 asterisk for table 3*** 0.122Note with 3 asterisk for table 3*** 0.177Note with 3 asterisk for table 3***
High school diploma or equivalent Note ...: not applicable 0.064Note with 3 asterisk for table 3*** 0.072Note with 3 asterisk for table 3*** 0.062Note with 3 asterisk for table 3*** 0.086Note with 3 asterisk for table 3***
College, CEGEP or other non-university certificate or diploma Note ...: not applicable 0.020 0.029Note with 2 asterisk for table 3** 0.023Note with 1 asterisk for table 3* 0.039Note with 3 asterisk for table 3***
University certificate or diploma (ref.) Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable
Immigrant  
Yes – immigrated after 2002 Note ...: not applicable Note ...: not applicable 0.090Note with 3 asterisk for table 3*** 0.122Note with 3 asterisk for table 3*** 0.150Note with 3 asterisk for table 3***
Yes – immigrated in 2002 or earlier Note ...: not applicable Note ...: not applicable 0.005 0.025 0.044Note with 3 asterisk for table 3***
No (ref.) Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable
Activity limitation  
Activity limitation Note ...: not applicable Note ...: not applicable Note ...: not applicable 0.106Note with 3 asterisk for table 3*** 0.108Note with 3 asterisk for table 3***
No activity limitation (ref.) Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable
Family type  
Lone parent Note ...: not applicable Note ...: not applicable Note ...: not applicable 0.183Note with 3 asterisk for table 3*** 0.190Note with 3 asterisk for table 3***
Person not in a census family Note ...: not applicable Note ...: not applicable Note ...: not applicable 0.202Note with 3 asterisk for table 3*** 0.208Note with 3 asterisk for table 3***
Couple family (ref.) Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable Note ...: not applicable
Intercept 0.096Note with 3 asterisk for table 3*** 0.074Note with 3 asterisk for table 3*** 0.076Note with 3 asterisk for table 3*** 0.000 0.024

Model 1 shows the results when including only basic demographic controls (age, sex and province) and literacy variables. It shows that the low-income rate for persons scoring level 1 and below is 21 percentage points higher than the rate for those scoring level 4 or level 5 (the omitted category), when holding basic demographic variables constant.

Model 2 adds controls for educational attainment. The addition of educational attainment to the model reduces the size of the coefficients on the skill variables. For example, the coefficient for proficiency level 1 and below drops from 0.213 to 0.162, or by nearly one-quarter.

Model 3 adds controls for immigrant status and Model 4 adds having an activity limitation and family controls. Each successive entry of control variables into the model reduces the size of the coefficient on skills with little change to the education variables.

After all controls are entered into the model, the disadvantage associated with having lower literacy (level 1 and below) drops to 12 points, which continues to be a large and significant effect, but much smaller than the 21 points obtained without controls.Note 8

Model 5 includes all variables except the literacy variables. Changes in the coefficients between Model 5 and Model 4 show the effect of literacy on the other variables. Coefficients for education and immigration status are larger than in Model 4, while coefficients for activity limitation and family types are unchanged. Thus, differences in skill level help explain part of the higher incidence of low income among low educational attainment groups and recent immigrants. The disadvantage associated with having at most a high school diploma falls from 9 percentage points (Model 5) to 6 points (Model 4). For recent immigrants, it reduces the disadvantage from 15 points to 12 points.

Conclusion

This study examines the relationship between skills and low income using data from the Longitudinal and International Study of Adults (LISA). The results show that being in low income is related to having lower literacy skills. The results indicate that the low income rate among persons with literacy skills in the lowest category was, at 29%, more than three times that of those who were in the highest categories.

This study also examined whether the relationship between skills and low income persists after taking other demographic factors such as education, immigrant status and having an activity limitation because previous research has shown that these are important “risk-factors” of being in low income. It finds that skills continue to matter even after these other controls are added in. The effect of skills on low-income, however, is muted due to correlations between skills and other risk factors. Moreover, the inclusion of skills reduces the importance of educational attainment and immigrant status, indicating that part of the higher incidence of low income among these groups is associated with lower skills.

Andrew Heisz is Assistant Director with the Income Statistics Division at Statistics Canada; Jerry Situ is an analyst also with the Income Statistics Division; and Geranda Notten is Associate Professor with the Graduate School of Public and International Affairs at the University of Ottawa.

This article is an abridged version of the paper “The role of skills in understanding low income in Canada,” originally published in 2015 in Measurement of Poverty, Deprivation, and Economic Mobility.Note 9

Start of text box

Data sources, methods and definitions

Data sources

The dataset used in this research is from the first wave of the Longitudinal and International Study of Adults (LISA). LISA is a longitudinal household survey that collects social and economic data about the Canadian population every two years. The first wave collected data between November 2011 and June 2012.

LISA data contains both high-quality information on household income and direct measures of respondents’ literacy, numeracy and problem-solving skills.

This study uses income data from LISA, which are obtained from annual income tax files provided by Canada Revenue Agency and other income sources for the year 2011. These data are collected for each family member aged 15 and over and summed to yield household income. In total, including children aged 0 to 14 from whom no income data were captured, LISA contains the income information of 32,133 respondents living in 11,458 households.

For households containing at least one individual between the ages of 16 and 65 (inclusive), one person per household was selected to complete the Programme for the International Assessment of Adult Competencies (PIAAC) assessment. Initiated by the Organisation for Co-operation and Development (OECD), PIAAC is designed to assess the skills and competencies of working-age adults in 26 countries. In total, 8,598 LISA respondents completed the PIAAC assessment some time between November 2011 and June 2012. This study focuses on this subset of LISA respondents who completed a PIAAC assessment.Note 10

Methods and definitions

PIAAC skills

The PIAAC skills measures consist of a direct assessment in three domains: literacy, numeracy and problem solving in technology-rich environments (PS–TRE). The assessed score in each of the domains is divided into different levels to facilitate interpretation.

The three domains assessed in PIAAC rely on the same core cognitive processes.Note 11 Thus, there is a high degree of correlation between the assessment scores. The correlation coefficient between literacy and numeracy scores is 0.87. Between literacy and PS–TRE, the correlation is 0.82 when the sample is restricted to respondents with assessments in both domains. Lastly, the correlation is 0.75 between numeracy and PS–TRE.

This study’s models were tested using each skill score separately and combined. When included separately, each skill score had a statistically significant model coefficient; however, when all skill scores were included, only literacy remained statistically significant. Thus, this study focuses mainly on literacy skills.Note 12

In PIAAC, a skill assessment corresponding to level 1 and below represents a proficiency such as only being able to find single pieces of information in short texts in the absence of distracting information or having only a basic vocabulary. Scoring in the top categories (level 4 or level 5) indicates a higher proficiency. Scoring in these categories represents a proficiency such as being able to integrate information from multiple dense texts and reasoning by inference.Note 13

Low Income Measure

For this study, the after-tax Low Income Measure (LIM–AT) is used. That is, individuals are considered to have low income if their adult equivalent adjusted (AEA) after-tax household income is below a threshold defined as 50% of the median AEA after-tax household income. The AEA adjustment accounts for potential economies of scale enjoyed by larger families.

End of text box

Date modified: