Someone sent a question to my blog a few weeks ago, asking what I thought of the then-recent news stories saying that census information indicated that areas with wealthy Asians were reporting low incomes and/or housing costs exceeding their incomes and therefore must be avoiding taxes in some illegal way. (And maybe even collecting welfare to boot!)

Links to the stories here, here and here, which varied in how cautious they were about these stats, if you can access these.

It’s taken a while to do the research, with my two day jobs and all, but here I am.

Okay, there will be a lot of nerd stuff later on and many references to dissemination areas and census tracts and rate of return.

But, for those of you with lives to live, here’s the short answer.

The two researchers whose numbers and maps have been used as the basis of those stories say, in essence, “It’s wrong to jump to those quick conclusions.”

There may be evidence out there that someone will find someday to confirm the dark suspicions that wealthy mainland Chinese are immigrating here, using Canadian services but avoiding Canadian taxes. This census information isn’t that evidence.

(Neither, by the way, are stories of this or that proven case of unreported foreign income, any more than stories about wealthy white Canadians hiding their money in offshore trusts on the Isle of Man proof that all wealthy white Canadians are crooks. But that’s for another day.)

Why it’s dangerous to draw inferences too quickly

 To put it in the words of one of them, Jens von Bergmann: ”It’s tricky to make substantive claims based on maps showing geographic overlap of immigration from China and poverty and these maps need more explanation than I have been willing to put in the map descriptions.”

He said he wishes now that he had put more cautions alongside his maps when he tweeted them out. Von Bergmann’s delightful and rabbit-hole seductive site that colour-codes census information onto maps is here: censusmapper.ca

Besides von Bergmann, UBC prof Daniel Hiebert also said it’s premature come up with definitive conclusions just by looking at a couple of variables from the census information. (And three other stats specialists I talked to said the same and added more qualifiers.)

“There are cautionary tales around the whole project,” said Hiebert, who listed off several reasons why people living in expensive areas might be reporting low income.

Hiebert didn’t actually do a study on wealthy Asians declaring low incomes, by the way, as some might have inferred from the coverage.

He did a study published in August about ethnic enclaves in Canada. He found that enclaves are becoming more prevalent, but that they are generally not low-income, marginalized ghettos. Instead, he reported on many positives: immigrants are becoming homeowners at surprisingly high rates, those in enclaves are less dependent on government payments, and the enclaves work as springboards for new immigrants into Canadian society.

He did include maps that combined information from census-tract data on enclaves and high levels of poverty (more than 30 per cent of people in the area classified as low income). His study is here.

You can also look at the maps Stats Canada produced showing which census tracts across the country showed high levels of poverty.

As you’ll see if you read on to my most wonkish and detailed section at the end, there are many census tracts in wealthy areas that have a high proportion of Chinese or Asian residents and are NOT reporting unusual levels of poverty.

There are also wealthy areas reporting high levels of poverty and with a lot of new Chinese immigrants, but where there are also a lot of renters or people in subsidized housing.

We sometimes forget how many basement suites, special-needs homes, and social housing are secretly sprinkled around the city, even in the poshest areas. Even in the four census tracts that are, generally, well-off Dunbar, the proportion of renters ranges from 19 to 23 per cent.

I’ll also list all the reasons later people gave me about why these numbers need more scrutiny, ranging from “this isn’t tax-return information so we don’t actually know what taxes were paid” to the usual cautions about all of this data because it comes from, guess what, the long-form census. But in the meantime …

Something is going on that we’d like to look more closely at

That isn’t to say anyone dismisses the possibility that there is something odd happening. The two researchers do say the incidence of high poverty rates in some unexpected areas, which happens in Toronto also, with high immigrant (often Chinese) populations is interesting and worthy of further analysis.

Hiebert calls it non-commonsensical and says he’s undertaking research on the puzzling stats. Von Bergmann says he hopes university researchers, who have powers and budgets greater than his, will request the kind of fine-grained stats searches that might yield some answers. For those not in the stats know, Stats Canada can be requested to do searches that combine various items (called crosstabs), like household income, ethnicity, length of time in Canada, age, etc., to get a true read on what is happening in census tracts showing unusual patterns. That shows true correlations, which amateur scanning of maps for crude correlations does not. You have to pay for those cross-tabs, so it’s generally something only people with funding — municipal governments, university researchers, etc. — do.

The reasons why the numbers could be wrong

Okay, that was my sort of quick answer. Now for more detailed info, in two stages.

First, the reasons researchers gave me in general for why to be wary about the census numbers behind those maps. (Later, a close look at some of the areas that were identified as high-poverty, high-Chinese and in unexpected parts of the city.)

Category 1: The data is problematic

1. This census information is based on the National Household Survey, aka the long-form census. It is not Canada Revenue Agency information. (That’s available elsewhere, in great detail.) We don’t know what anyone actually declared as their income on their tax forms and we don’t know what taxes they paid. All the NHS has is what respondents wrote on the form as their income.
2. All this information comes from, of course, the now-voluntary long-form census, making it less reliable than the old census and more easily skewed if certain groups are over- or under-represented. Census tract areas that I looked at were in the range of 1,500 or so households. That means the long-form census went to 500 households (one in three) approximately in any census tract. Response rates ranged from 50 to 90 per cent in different areas of the Lower Mainland. That could mean as few as 250 non-random households are representing the whole 1,500. That would be fine if the responses were random, but as researchers have noted, certain groups tend to be less likely to have completed their long-form census questionnaires. Dissemination areas, which is what von Bergmann’s maps used, are even smaller and therefore more subject to the pitfalls of limited, non-random data.
3. The questions on income in the very lengthy long-form census come at the end. People filling out the form sometimes give up before the end. That means income information is the least reliable. As well, one stats specialist pointed out to me that the NHS asked people to state their 2010 income and their 2011 housing costs.
4. It seems implausible, a couple of stats analysis suggested, that someone trying to hide income would be motivated to fill out the long-form census at all.
5. This real-estate blogger also raised some other questions about the inferences related to the “people paying more for housing than they earn in income” conclusions. He’s not the only one. No one really understands how anyone, whether rich or poor, manages to spend more on housing than they have in income. It makes marginally more sense among the wealthy, who may be living off their wealth or have jobs that pay a lot in one year but not so much in the next and the census happened to catch them in the low year.

Category 2: Just because someone is reporting low income doesn’t mean they’re breaking the law or deliberately avoiding taxes

1. An accountant I talked to and several others noted that reporting low income doesn’t mean you’re lying about your wealth. Wealth is different from income. You pay tax on income, but not wealth. There have been people who’ve moved here with considerable assets, enough to live on. (Of course, if invested, that would generate some taxable income, but who knows how people are using their assets.)
2. The same accountant I talked to noted that there is nothing illegal or fraudulent about having, say, a father who remains as a non-resident of Canada and doesn’t report his income there, while the wife and children live here, supported by money sent by the dad. We might not like it, but it’s not any more illegal than, say, Canadian retirees who pack up and move to Mexico or Thailand for 20 years, declare themselves non-residents, avoid paying Canadian taxes on their incomes for those two decades, and then move back in their 80s when they need Canadian health care.

Category 3: The perfectly legal tax-avoidance mechanism that no one seems to know about

          One interesting aspect of tax policy I discovered while trying to learn more about this issue is that the Canadian government had, until December 2014, a specific policy that allowed (and I’d say encouraged) wealthy immigrants to put their money into an offshore trust for five years after arriving, precisely so they could avoid paying high Canadian taxes. Again, we may not like it, but this was something that our government promoted.

Several researchers and accountants I talked to had never heard of this policy, which makes me guess it was only well-known within a small community – would-be wealthy immigrants and their lawyers and accountants. (Below a couple million in assets and there wasn’t really a point in setting them up.)

As you can see from these older links here, here, and here, these immigrant offshore trusts were promoted as a distinct advantage to extremely wealthy clients.  “Five-year tax holiday” was the headline on one adviser’s site.

From what I can gather, permitting these trusts put Canada on an equal footing with the UK in being able to attract wealthy immigrants.

And there were many warnings in late 2014 and early 2015 about what the impact might be on their elimination, along with some general hand-wringing by law and accounting firms.

Again, it seems like this policy was only known in limited circles.

Boring down to the census tracts

 I don’t claim to do the kind of detailed research that a university professor paid to do this kind of thing has done. But I did take the trouble to spend a few hours looking closely at some of the census tracts identified as anomalous.

Richmond

A number of census tracts in north-central Richmond indicate more than 30 per cent of households are considered low income. In some stories, these areas were described as neighbourhoods of “mansions.”

The mansion part is true on the far west side of Richmond, the areas that nudge up against the dike. But it’s less true for north-central Richmond. I drove there weekly every year for 15 years to teach at Kwantlen, then a college, between Garden City and No. 3 Road. I’m pretty familiar with the area. It’s a mix of older split-levels, some frankly crummy looking bungalows, a lot of 1970s-era apartments, and, increasingly, new condos, as well as some newer houses.

I looked at one census tract area, 9330148.00, between No. 2 and 3 Road, and between River Road and Granville. That’s the area the furthest west of the low-income-reporting districts, the closest to the true mansion areas. It has 8,125 people living there, with almost a third of them having immigrated in the 10 years between 2001 and 2011. Many immigrants are from China, Honk Kong, and Taiwan, but almost 700 are Filipino. This area reported 38 per cent of households were “low income,” by the Stats Can definition.

The median value of a dwelling in 2011 was $449,876, half of the median value of homes in the areas to the west. Just over 40 per cent of the respondents were renters. The proportion of tenants in subsidized housing: 20 per cent.

Wealthy Asians living in mansions? You decide.

Vancouver-Oakridge

This census tract 9330010.02, just east of Granville in what I’d call upper Marpole. (49^th to 57^th, Granville to Cambie.) It had 34 per cent of households defined as low income, and 79 per cent called said they were Chinese. It showed up as both an ethnic enclave and a low-income tract in Hiebert’s maps. Renters were 43 per cent of the population of 710 of the 1,635 households were renters. Seven per cent of tenants were in subsidized housing.

Vancouver-Arbutus Ridge

There are two census tracts here. The one with the little curve in it, 9330027.02, is the most interesting. It’s in the curvy-streets area around Prince of Wales high school, generally bounded by Trafalgar, Maple, 37th, and 27thish. One of von Bergmann’s maps shows that it had the highest number of immigrants from China arrive 2006-2011 of any census tract in the Lower Mainland. In total, only 1,440 of the 3,865 people living there identify as non-immigrants.

Almost one-third of households are low income, say NHS stats. About 21 per cent of households are renters, and two-thirds of them spend more than 30 per cent of their income on housing. Ten per cent are in subsidized housing. Of the people defined as low-income, 39 per cent of them are under 18, which is a strikingly high number. (The non-response rate for the long-form census here, by the way, was 27.2 per cent.)

There’s something happening in this area but what? Is this where a big group of wealthy immigrants with legal offshore trusts arrived? Are there renter families, classified as low-income, with huge numbers of children? Is it possible there is a bunch of under-18s living on their own, with income being transferred from abroad, while they go to school in Vancouver? We await answers

There’s a second census tract right next door to this one, just south of it, 9330022.00. That is even stranger. But not really when you look at it closely. It’s the little rectangle bounded by 37th, 41st, Trafalgar and Arbutus. This has a 34-per-cent low-income rate for its households, which seems odd for the west side indeed. But a quarter of the people living here are renters and a quarter, not necessarily the same quarter, are seniors. But then, I realize as I look at the map, this is the cluster of apartment buildings just north of Kerrisdale Village. So it makes sense in the end.

Some might say, well, Frances just went cherry-picking looking for contradictions or things that proved her thesis. That’s right, a person could say that. And the same could be said for people who came up with the other interpretations of these maps. That’s why it’s probably best to wait for the experts to do this stuff.