Income Research Paper Series – Research Paper

    Low-income Dynamics and Determinants under Different Thresholds: New Findings for Canada in 2000 and Beyond


    Much of the debate on poverty and low income in Canada focuses on cross-sectional evidence.1 While this is important, low-income dynamics from the longitudinal perspective should also be examined. How often do Canadians get in and out of low income within a given period of time? How long do they stay there and how fast do they escape from low income? Who are more likely to be in low income for a long episode? What are the factors affecting the dynamics and persistence of low income? In this paper, we will address these questions to better understand low income from the dynamic perspective.

    Existing literature provides some evidence on low-income dynamics and persistence for the 1990s and earlier. Duncan et al. (1993) is probably among the earliest studies on low-income dynamics for a number of countries, including Canada. Morissette and Zhang (2001) is one of the first fully dedicated studies on Canadian low-income dynamics and persistence. Using data from the 1993-to-1998 panel of the Survey of Labour and Income Dynamics (SLID), Morissette and Zhang showed that about 8% of Canadians experienced low income for four years or more in the six-year period. Only 3% were in low income for all six years. They also highlighted the low-income experiences of various high-risk groups in the population. Finnie and Sweetman (2003) is another major Canadian study on low-income dynamics using data from the Longitudinal Administrative Database (LAD) from 1992 to 1996 (a five-year panel). They found that about 6% of Canadians were in low income for all five years from 1992 to 1996. Almost 21% experienced low income for one to four years. They showed that compared to married couples, lone parents, in particular lone mothers and unattached individuals, had greater rates of low-income entry and exit during the period. However, the above studies are based on data from the 1990s or earlier. In this paper, we wish to study what has happened to Canadian low-income dynamics and persistence since then.

    Another feature of previous studies is their use of a single low-income threshold. Morissette and Zhang (2001) employ the Low Income Cut-Off (LICO); Finnie and Sweetman (2003) create a relative threshold similar to the Low Income Measure (LIM). LICO and LIM are two low-income thresholds established by Statistics Canada. In addition, Human Resources and Skills Development Canada (HRSDC) also introduced the Market Basket Measure (MBM) in the early 2000s. With these three low-income thresholds readily available, we wish to evaluate how robust our updated analysis on low-income dynamics and persistence would be using these low-income thresholds.
    Assessing low-income thresholds touches the core of low-income measurement. In the literature, some authors advocate the relative approach; others advocate the absolute approach emphasizing the use of food, clothing, shelter and other essentials for defining the poverty threshold. Osberg and Xu (1999, 2000a), Myles and Picot (2000), Morissette and Zhang (2001), Finnie and Sweetman (2003) and the World Bank Institute (2005) have adopted the relative approach. Others such as Sarlo (1996) and Pendakur (2001) consider the LICO and LIM too generous: they advocate using the essential costs of living to construct a threshold such as MBM.

    There is a tendency to focus on relative low income or poverty within developed countries. Galbraith (1998) advocates the idea that low-income thresholds must be established with reference to specific communities. He implicitly suggests the relative approach. On the relevance of relative low-income thresholds to other absolute deprivations, Sen (1992, p. 115) points out that "[r]elative deprivation in the space of incomes can yield absolute deprivation in the space of capabilities." More recently, Osberg and Xu (2008) and Ravallion (2010) provide new evidence to support low-income relativism. On the magnitude of low-income thresholds, Osberg and Xu (2008) further argue that if zA is an 'absolute' (however defined) low-income threshold, and zR is a 'relative' (as a fraction of median equivalent income) low-income threshold, the optimal choice of low-income thresholds, z, should be the maximum of the two—that is, z = maxi(zR , zA).

    Nevertheless, when we face several low-income thresholds—all of them could be relative or absolute—how to evaluate the impact of various threshold choices? In this study, we will try to answer the question in the context of analysing low-income dynamics and persistence for Canada in 2000 and beyond.

    Our study is unique in the following ways. First, we rely on more recent data to examine low-income dynamics and persistence in Canada. Second, unlike previous studies we employ several low-income thresholds at the same time to study low-income dynamics and persistence. Third, we pay particular attention to the development of research strategies on low-income dynamics and persistence. Finally, we attempt to shed some light on the choice of different low-income thresholds, both theoretically and empirically.2
    The rest of the paper is organized as follows. Section 2 discusses the conceptual questions and our research strategies. Section 3 presents the empirical results. In Section 4, we conclude.


    1. We have adopted the term 'low income' in our analysis. Although low income is not interpreted as poverty by Statistics Canada, we use this concept to refer to both low income and poverty broadly in our analysis. We believe, from the conceptual point of view, low income defined by Statistics Canada nests poverty, but the inverse is not necessarily true.
    2. Theoretical discussions and empirical results on the choice of different thresholds are contained in the full version of the report, available from the authors on request.
    Date modified: