Measuring income is complicated

Household income has long been used as a proxy for the standard of living or the level of material wellbeing of that particular household. But how easy is it to actually measure income?

In earlier decades in Canada, Statistics Canada would do surveys where it asked “household heads” to self-report their income from all sources, and provide other important demographic information. The census has always been our largest survey, but it’s only conducted every five years. In between those periods, other surveys have been conducted and those focusing on the household’s economic situation typically collected income information.

The problem with self-reporting is that not everyone accurately reports income. In some cases, the inaccuracy can be attributed to poor record-keeping, sloppiness, honest mistakes, and in some cases, respondents deliberately provide misleading answers. For example, some people who receive cash payments such as tips in the hospitality sector, contract workers in construction, renovation and some trades, and people working in illegal sectors of the economy such as the drug trade, often fail to accurately report their income. As an example, Statistics Canada researchers in a 1990 study found that unemployment insurance benefits and social assistance income payments were substantially underreported by respondents.

More recently, Statistics Canada has used tax return records for the bulk of census and survey respondents. That helps somewhat with accuracy, but clearly does not address deliberate underreporting. The broadly defined “underground” economy, which includes unreported or underreported income from activities (both legal and illegal) is significant. According to a recent Statistics Canada estimate, underground economic activity totalled about $72.4 billion (or 2.5 per cent of GDP) in 2023. The study found that residential construction and renovation were a significant contributor to that total, as were wages and tips that went unreported, and the sale of liquor, tobacco and cannabis from unlicensed and unregulated sources.

An earlier study published by the Bank of Canada, using a different methodology, found that unreported income was much more pervasive and accounted for between 14 per cent and 19 per cent of GDP, depending on the year. These authors also noted that estimates of poverty are not reliable given the extent of underreported income. They estimated that “Sixty to seventy per cent of households reporting income of less than $20,000 a year appear to be under-reporting income.”

There are additional sources of potential inaccuracy in the measurement of income. As with any survey, the design is important. It must be clear, comprehensive and avoid ambiguity. The selection of respondents must also be random to fairly represent the population. Even if all of that is perfect, a small percentage of those selected will not complete the survey. Various methods are used to “impute” values that are missing. This is another source of potential error.

Finally, it’s important to mention that the purpose of collecting and publishing income information is that it’s often used as a proxy for wellbeing. There are instances, however, where just looking at income is insufficient to accurately gauge economic wellbeing. Consider, for instance, households that are able to afford a much higher living standard than their income would allow by drawing down their savings or selling off assets. Similarly, consider university students whose actual living standards are often supported by income transfers from parents or other relatives and/or borrowing, none of which shows up in the data on their income. And consider how accurate income measures are for families that save a portion of the income, particularly families that save a substantial portion.

These discrepancies between income and what people actually consume have led many economists to prefer consumption as a better representation of living standards than income, which will be addressed in another essay in this series. While income is useful in many cases, especially for middle-income households, it appears to be less adequate when we try to measure the extent of poverty.

The key is understanding the many nuances and potential sources of error involved in measuring income for the purposes of assessing poverty and inequality, and the possibility that other measures such as actual consumption may be better markers of economic wellbeing, particularly for lower-income households.

This is the second essay of a five-part series on income inequality, which will appear on the Fraser Institute blog. The authors would like to recognize the critical contributions of the authors of the various essays in the 2017 collected series on inequality and poverty as well as Christopher Sarlo’s three decade-plus work on poverty and inequality for the Institute. In addition, the authors thank Christopher Sarlo for his work on early drafts of this series.