Cross and Taylor: Lies, Damned Lies, And Race-Based Statistics

Reading this commentary reminded me of an anecdote that I can’t unfortunately locate: former PM Harper’s decision to replace the mandatory census with the voluntary, and less accurate, National Household Survey in 20ll was driven in part by the data being used by academics, advocates and activists as a basis for more progressive policies.

The alternative, as Cross and Taylor appear to advocate, is not to have visible minority breakdowns in the labour force survey to avoid this use of data. To my mind, it is a head in the sand approach as such data is needed to understand how well Canadian society is working in terms of economic integration.

COVID-19 has demonstrated the various inequalities between different groups. The regular censuses have also captured these inequalities as well so expanding this to the labour force survey (and public service employment equity reports) is consistent with long-standing practice.

To my mind, issues lie more with respect to how the disparities are interpreted, whether narrowly or looking at the range of factors that influence these disparities.

For example, when I look at public service employment equity data, groups that have lower levels of educational attainment (e.g., Blacks, Latin Americans) are less represented among occupations requiring university degrees. This disparity, of course, likely reflects in part earlier barriers and discrimination encountered by those groups (e.g., streaming of Blacks into non-academic streams, recently addressed by the Ford government).

Disaggregated date is need to be aware of disparities and point towards questions regarding the reasons for these disparities, and assess the degree to which policy interventions, and which kinds, may be warranted.

To their credit, Cross and Taylor do some analysis, looking at occupations and visible minorities, highlighting that Koreans, Filipinos and Southeast Asians are more concentrated in the accommodation and food service industry than not visible minorities as an explanation of why these groups were more affected by COVID-19 lockdowns.

But it is ingenuous, at best, to present socioeconomic circumstances as completely unrelated to barriers faced by some groups.

And of course, the data will be used and sometimes misused by advocates and activists, and one could argue that Cross and Taylor are equally and legitimately using data to support their position.

But curious for a former statistician to be arguing for less data and thus less needed information for evidence-based policy. And using France as a model?:

Since July, Statistics Canada has been publishing labour-market data divided into 12 ethnocultural categories including Chinese, South Asian, Southeast Asian, West Asian, Korean, Japanese, Arab, Black, Filipino, Latin American, White, and Others. Sliced this way, Statcan’s figures reveal the unsurprising fact that unemployment is unevenly distributed across Canada’s racial populations, just as it varies by region, gender and age. The adult Canadian unemployment rate in January was 9.4 percent, but 20.1 percent for Southeast Asians, 16.4 percent for Blacks and 16.6 percent for Latin American Canadians. “Others” had a slightly-better-than-average unemployment rate of 8.9 percent.

This move to produce racially-specific labour-market data may well have been inevitable, given the intersectional enthusiasms of Prime Minister Justin Trudeau, who recently declared his next budget will be an explicitly “feminist”document. It also follows logically from his government’s creation of Statcan’s Centre for Gender, Diversity and Inclusion Statistics in 2018. Equally predictable is the effect this new information has had on public discourse.

The release of race-based labour-market data has provided further fuel to the ascendent view that Canada is an inherently unfair and racist country. Lobby groups and organizations representing the various racial groupsidentified by Statcan have latched onto the new data to back up claims regarding the “negative labour market impact of racism on Black youth” and other collective sins aimed at Canadian society. The figures are also frequently held as proof that employment equity programs and other government market interventions must be scaled to industrial proportions to eliminate the discrimination baked into Canada’s labour market.

But when it comes to fomenting outrage, Statcan is just getting started. In a recent commentary in The Globe and Mail, Anil Arora, Chief Statistician at Statistics Canada, explained his organization’s intention to double down on the collection and dissemination of race-based data. Because the initial effort last July revealed such glaring “racial disparities”, he wrote, Statcan will now be using “data from varying lenses…to measure those inequalities and track the progress being made to address them.”

French law specifically forbids INSEE from processing or analyzing data regarding “ethno-racial classifications” because it could violate constitutional requirements that all citizens must be treated equally.Tweet

As exciting and progressive as all this may seem, however, Statcan should tread carefully. Collecting race data is inherently contentious and divisive, something all national statistical agencies must recognize. While the United States has a long history of collecting very detailed race-based data, others such as France’s Institut national de la statistique et des études économiques (INSEE) do not disseminate any race statistics. In fact, French law specifically forbids INSEE from processing or analyzing data regarding “ethno-racial classifications” because it could violate constitutional requirements that all citizens must be treated equally “without distinction of origin, race or religion.”

As we shall see, unequal racial outcomes revealed by national statistics do not necessarily prove racism, and often lead to intractable debates. This is especially so in a country like Canada, where there’s a large overlap between visible minorities and immigrants who historically take years to match the outcomes for Canadians born in the country. Feeding a culture of grievance that denies any role for cultural differences in generating observed inequality can, paradoxically, perpetuate unequal racial outcomes. And as the state of affairs in the U.S. suggests, a surfeit of race-based statistics is no guarantee of racial harmony.

Neither is Statcan exempt from the principle of opportunity cost. Collecting one set of data inevitably means foregoing others – some of which may be of greater value. For years, researchers from social policy groups such as Cardus have asked for better data on how marital status affects employment and income. This would provide more detail on the important role played by family in the labour market. Yet these requests have long been ignored for cost reasons. Statcan presumably has better things to do with its limited resources. Now, however, in the middle of a pandemic, the agency has suddenly discovered the means to produce divisive race-based unemployment data.

Pandemic and Race

There are many pitfalls and risks associated with attributing different outcomes experienced by different racial groups exclusively to race, especially when these accusations are based on superficial statistics. In its July 2020 labour market report that, for the first time, segmented unemployment by race, Statcan itself noted that the top line figures showing poorer outcomes for most visible minorities categories reflected, in large part, the tendency of certain racial groups to work in industries deeply affected by the pandemic.

For example, 19.1 percent of Koreans, 14.2 percent of Filipinos and 14.0 percent of Southeast Asians were employed in the accommodation and food industry, according to the 2016 Census, compared with only 5.9 percent of Whites. Given the dramatic effect the pandemic-related lockdowns and other measures have had on the hospitality sector, it seems reasonable to conclude that race played little or no role in these unequal outcomes. Rather, it was the circumstances of the industries they were working in.

It has also been widely reported that different racial groups contract Covid-19 at different rates. Some concluded that this was because these groups are particularly disadvantaged by a racist society, while others wondered whether particular racial groups might have a different genetic susceptibility to the virus. As a recent U.S. National Bureau of Economic Research study warns, merely noting differences among racial groups without knowing their source means “the political discourse can gravitate toward ‘biologic explanations’ or explanations based on racial stereotypes which are harmful in themselves and get in the way of policy solutions.” The same study made plain that it was the socio-economic circumstances of particular groups that affected their exposure to the virus. This was due to working in particular industries and using public transit, which increased their contact with other people and, in turn, led to a higher rate of infection. Once the data was corrected for these variables, visible minorities in the U.S. were found to be no more susceptible to the virus than whites.

Given how easily some data can be misinterpreted or misrepresented, it would seem that Statcan has a clear responsibility to caution users about its proper use. Figures regarding the distribution of federal government revenues and spending by province, for example, are regularly twisted by politically-motivated analysts and governments. As a result, Statcan published an article in 2007 offering a detailed explanation for why these statistics should not be considered a scorecard for which provinces are gaining or losing from their dealings with the federal government.

Much of the current debate over racism in Canada arises from the presumption that all aspects of life should be perfectly evenly distributed, and that any deviation from pure equality must be considered prima facie evidence of systemic racism. Tweet

It is, accordingly, curious that these new race-based labour-market figures do not come with a similar warning; race data is far more emotionally and politically incendiary than provincial fiscal data. It is also surprising that Statcan did not directly address the issues raised by France’s refusal to collect race-based data.

Racism of the Gaps

Much of the current debate over racism in Canada arises from the relatively recent presumption that all aspects of life should be perfectly evenly distributed, and that any deviation from pure equality (a term also recently redefined from equality of opportunity to sameness of outcomes) must be considered prima facie evidence of systemic racism. With dizzying speed, this eminently contestable claim has been elevated nearly to conventional wisdom.

In an insightful commentary published earlier this month by the Macdonald-Laurier Institute, Vancouver-based writer Sonia Orlu tackled head-on the notion that “any disparity in outcomes between blacks and whites is the direct result of racism, as opposed to class differences, culture, personal ‘(ir)responsibility,’ or any other myriad of situational factors.” As Orlu, who is black, points out, this “racism of the gaps” generally relies on surface-level observations lacking in context or detail.

Nowhere is this assumption more explosive than regarding claims that members of visible minorities are disproportionately targeted, arrested or killed by police. As Orlu points out, a case for systemic racism in policing can only be proven with detailed race-based data showing police interactions as a share of the overall criminal population, rather than the population at large. While racism may create the conditions in which visible minorities commit more crime, simply arresting more visible minorities is not, in and of itself, proof police are acting with racist intent.

Orlu notes, however, that Canada does not collect the sophisticated race-based data necessary to come to an informed observation on this heated topic. With only the most basic statistics available regarding race, arrests and incarcerations, it is easy to conclude that police actions are driven by racism rather than other factors. And even when detailed race and crime evidence is available, as it is in the U.S., Orlu points out it is generally ignored by the media and public because it does not align with popular “anti-racism ideology” narratives. More information, in this case, does not produce a better debate or better decisions.

This problem is further illuminated by economist Tim Harford in his fascinating new book The Data Detective. Harford offers the example of an algorithm called COMPAS (Correctional Offender Management Profiling for Alternative Sanctions) used in the U.S. to predict the probability of a criminal being re-arrested. Because the algorithm produced racially disparate results – giving higher probabilities for blacks to be re-arrested than whites under similar circumstances – it was accused of perpetuating systemic racism. And yet the algorithm itself was colour-blind; race was not an input factor.

A detailed investigation by a team of statisticians revealed that the differing results were the product of members of different races behaving differently and/or living in different neighborhoods. As Harford concluded, “The only way in which an algorithm could be constructed to produce equal results for different groups…would be if the groups otherwise behaved and were treated identically.” Such an outcome adds evidence to the proposition that unequal results between races do not prove racism if behaviour and circumstances differ.

Examples of the racism gap fallacy are in ample supply elsewhere. Last month, for example, Akim Aliu, a former NHL player and founding member of the lobby group Hockey Diversity Alliance, claimed that an observed lack of racial diversity in the National Hockey League could only have one possible source. “There are still owners in the league who don’t even believe [racism] is a problem,” Aliu complained to Reuters in a Black History Month article. “To me that is just unfathomable, 95 per cent of your league is white and you don’t see there is an issue of race.”

Yet visible minorities make up a vast majority of the lineups in many other sports. The National Basketball Association is 74 percent black, and the National Football League 68 percent. While the Canadian Football League does not provide readily-accessible race-based statistics, the number of black players in this league also appears to far exceed representation in the general population. Should all this be taken as self-evident proof that football and basketball are equally prejudiced, but in favour of visible minorities? Of course not.

In another fixation on gaps, Statcan’s Arora in his Globe and Mail commentary emphasizes the importance of moving “toward levelling the uneven economic playing field”, citing the unequal unemployment and poverty rates among immigrant women as a key example. It must be noted, however, that there is a large overlap in Canada between visible minorities and immigrants. The lagging labour market outcomes for visible minorities and other Canadians reflect the long-standing challenges of immigrants establishing themselves in Canada.

In 2016, for example, Canada admitted thousands of Syrian refugees, many with limited education and little or no knowledge of either of Canada’s official languages. Do inferior incomes and more joblessness among the women of this group in the short time since they arrived prove the “playing field” in Canada is uneven? Inferior outcomes for some players don’t necessarily indicate a tilted field, it may merely demonstrate that they were sent out onto the field without the skills and training needed to compete. It is also worth remembering that the prevalence of poverty and inequality of income is much greater in the countries most immigrants come from, than is their inequality compared with native-born Canadians.

 The Inconvenient Truth that some Minority Groups Outperform the Majority

Racism – defined as the presence of deep-seated prejudices that affect individual and collective behaviour – certainly exists in Canada, as it does in all countries. And wherever present, it should be challenged and overcome. That said, collecting race-based data may not contribute to that worthy goal at all. It could instead cultivate a mentality of grievance and entitlement that undermines the impetus for individuals to strive to achieve more for themselves and their children. Look how easy it was for Aliu, for example, to take a simple statistic regarding the race of NHL players and turn it into a bitter accusation.

Arora’s recent Globe commentary, meanwhile, laments the “many economic challenges facing racialized populations, Indigenous people, persons with disabilities and other marginalized groups” as proof of the need for Statcan’s big move into race data. But might it not be more useful to study how certain minority groups have overcome even-greater challenges in the past? Few groups have suffered more persecution and discrimination than Jews, yet their internal culture enabled Jewish people to achieve superior results in multiple fields of endeavour in country after country. Japanese Canadians are another example, overcoming their forcible removal from their homes to be quarantined in remote camps during the Second World War, and going on to achieve one of the highest income levels of any racial group.

It is too easy to dismiss the achievements of certain races or ethnic groups as the result of advantages and privileges. While the lagging performance of some visible minorities is automatically assumed to be evidence of Canada’s innate racism, the opposite conclusion is never drawn from the superior results displayed by other minority groups (such as Chinese, to use Statcan’s terminology) in terms of employment, scholastic achievement or avoidance of crime. Looking south of the border, pre-Covid U.S. Census Bureau data revealed that the real median household income of Asian-Americans is nearly 30 percent higher than that of whites.

Thomas Sowell, the renowned black economist at the prestigious Hoover Institution at Stanford University, has written extensively on the use and misuse of race-based data. His insights on the importance of the culture internalized within racial groups provides a good lesson on the pitfalls of superficial interpretations of race data. As Sowell observed in his 2013 book Intellectuals and Race: “Different races, after all, developed in different parts of the world, in very different geographical settings, which presented very different opportunities and restrictions on their economic and cultural evolution over a period of centuries.” Further, people tend to blame racial differences on bias, which ignores “internal explanations of intergroup differences in favor of external explanations.”

As Sowell noted wryly in his 1996 book Black Rednecks and White Liberals, “all things are the same except for the differences, and different except for the similarities.” Given current demands for diversity in all things, he was observing, why should anyone expect identical outcomes as a result? Perhaps that comment should be attached to every Statcan press release on racial differences in the labour force survey.

The Politics of Distribution Versus the Economics of Growth

There is a growing sense of malaise in Canada, including worry that we are falling well short of our economic potential. Our political and economic leaders ought to be focusing on creating the macroeconomic and cultural conditions wherein all groups can thrive. Instead, our country’s growing fixation on racial issues – including the collection of race data – invites policymakers to think in terms of improving Canada one micro-group at a time.

We have already seen its nefarious impact. The most salient fact of the Covid-19 pandemic has been its devastating impact on all of Canada, with 5.5 million people losing their jobs or having their work severely curtailed in the spring of 2020. Rather than proposing general solutions to support growth and allow the reopening of the economy, numerous special interest groups have used the pandemic as an excuse to advance their particular pet policy projects, re-packaging old proposals that have circulated for years or decades to “solve” a once-in-a-lifetime crisis. The ideas include greater provision of day care, universal basic income, universal Pharmacare, extended employer-paid sick leave and so on, almost ad infinitum, as if budget constraints no longer existed.

The greater influence of broad economic conditions than specific social policies is revealed by Arora’s own reference to the “significant progress” visible minorities were making towards equality before the pandemic set them back. He cited a sharp drop in poverty and “rapidly rising employment rates among working-age immigrant women” as evidence of this happy situation. Such pre-pandemic levels of achievement – closing numerous gaps with the rest of society – was not the product of programs targeting specific aggrieved minority groups, but the result of an improving and robust economy-at-large. As Canada as a whole grows, its gaps shrink.

The best way to resuscitate the fortunes of visible minorities is the same as for all other Canadians: reopen the economy as quickly as possible and adopt policies and attitudes aimed at supporting long-term economic growth. Tweet

The same phenomenon was in even greater evidence in the U.S., where wage gains in 2019 were led by the lowest wage-earners, especially visible minorities. In recognition of this, more Latino and black voters cast ballots for Donald Trump last November than in 2016, despite his obvious negative attributes. The clear lesson is that better macroeconomic policy and economic growth always outweigh the impact of targeted government programs.

It is important to remember that the reversal of fortunes for minorities during the pandemic was because our economy was struck by the economic equivalent of a thermonuclear device, not because Canada overnight became more racist. The best way to resuscitate the fortunes of visible minorities, therefore, is the same way as for all other Canadians: reopen the economy as quickly as possible and adopt policies and attitudes aimed at supporting long-term economic growth.

Statcan’s new race-based data invites the facile conclusion that one group’s success explains another group’s relative failure and justifies its grievance. And our faltering economic growth reinforces the sterile view that the size of the economic pie is fixed and any gain by one group comes at the expense of others. The result is a focus on the politics of distribution instead of the economics of growth.

To be fair, Statcan did a lot of good work in response to the pandemic. This includes flash estimates of GDP, adjustments to how it measures labour under-utilization, more timely data on firm turnover, and innovative ways to track population mobility during a lockdown. The agency’s recent move into race-based data does not, however, rank among these useful innovations. And its effects may outlast all the others due to the appeal it holds for groups dedicated to fanning the flames of internal complaint.

With race-based data now being widely disseminated, this process may be unstoppable. Any move to cut off funding for this project will be widely condemned by the many vocal advocates of the “race industry”. Canadians should thus prepare themselves for a steady stream of studies in the coming years declaring the presence of gaps that allegedly prove the existence of systemic racism, but which tell us nothing about their origin or the best way to reduce them. All this is an unfortunate but costly distraction from the bigger and more important issues of innovation, investment and entrepreneurship that will be necessary to restore an economy that will benefit all Canadians – of every race and colour.

Philip Cross is a senior fellow of the Macdonald-Laurier Institute and the former chief economic analyst at Statistics Canada. Peter Shawn Taylor is senior features editor of C2C Journal.

Source: https://c2cjournal.us19.list-manage.com/track/click?u=e8efce716429c34122979e2de&id=ca577642d5&e=4174a59277

About Andrew
Andrew blogs and tweets public policy issues, particularly the relationship between the political and bureaucratic levels, citizenship and multiculturalism. His latest book, Policy Arrogance or Innocent Bias, recounts his experience as a senior public servant in this area.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: