Data-Driven Bill Gates Wears Rose-Colored Glasses in Mexico

Bill Gates is the world’s most generous humanitarian. He is also very ambitious and self-assured. Last month in the Wall Street Journal, Gates outlined his plan to “fix the world’s greatest problems,” as the headline of his piece stated. (It is reproduced from his annual letter.) This includes reducing child mortality and improving educational performance by better measuring the effectiveness of aid programs.

Nobody can question Gates’ motives, but his methods are not immune to criticism. Some smart people have sharply criticized (even ridiculed) his overall approach to international development. Others are not so blunt, and argue instead that it is the culture of development that needs to be reformed.

It’s no surprise that Gates, the king of all geeks, wants aid programs to be data-driven. But he should lead by example and not be so cavalier with statistics, as he is here, in this recent note lauding the creation of a new agricultural facility in Mexico (my emphasis):

One reason why I always enjoy going to Mexico is because of the country’s incredible progress, which has been really encouraging for me to get to see as I’ve spent time there over the years first for Microsoft and lately for our foundation. Despite the many challenges Mexico still faces, since 1995, the country has been able to cut its poverty rate from about 15% to less than 2% today. A major driver in reducing poverty has been agricultural development, especially innovations that have helped improve crop yields for smallholder farmers, whose lives have improved a lot as a result.

Huh? Since when has Mexico reduced it’s poverty rate so drastically? From the Pew Research Hispanic Center (April, 2012):

Despite a moderate long-term rise in per capita GDP, the share of Mexicans who live below the poverty line has not changed significantly in recent decades. It was 51% in 2010, down slightly from 53% in 1984.

And from the World Bank:

According to CONEVAL (sp) (National Council on Evaluation of Social Development Policy) the number of Mexicans living in poverty to 2010 is estimated in 52 million people. This implies that around 46.2% of Mexico’s total population lives in poverty, mainly in urban areas.

Meanwhile, extreme poverty (those living with less than $978 pesos, US $76) a month in urban areas and less than $684 pesos (US $53) in rural areas reduced slightly from 10.6% to 10.4% (11.7 million people).

This is not to say that Mexico hasn’t made great economic strides, which owes in part to the emergence of the country’s high tech sector. That is something Gates can surely appreciate.

4 Responses to “Data-Driven Bill Gates Wears Rose-Colored Glasses in Mexico”

  1. I’ll bet we find out that figure comes from some other NGO report. I like Gates’ work, but I agree the data needs to be correct. Going after the wrong issues, or thinking something is resolved when it’s not, is a waste of donors funds and gets nowhere.

    But over and over we see that activist and NGO reports are pretty loose with the data and sources. People are getting publicly burned by this (see Attenborough’s recent drama, and the IPCC). I am glad to see some journalists calling this out.

    Too often some activist report uncritically becomes a whole story.

  2. Joshua says:

    I’ll bet we find out that figure comes from some other NGO report.

    Lol! Yeah – Gates makes some odd statistical error – but some “NGO” must be to blame. Nothing like a little confirmation bias with my
    mid-morning coffee.

    I guarantee that no matter how hard you search, you will find no NGO
    report that lists the poverty rate in Mexico to be 2%, let alone 15%.

    My guess is that Gates was mistakenly referring to something like growth in poverty rate?

    And Keith – I’ve seen some good critiques of Gates’ views on
    international development –

    but that Easterly editorial was a joke – not up to your usual standards of quality links.

  3. Tom Scharf says:

    It’s a known fact that the poverty line goalposts have been moving for a long time. Maybe that’s appropriate, but pretending that things haven’t changed much since the 1930’s is a fantasy. So the statistics become an artifact of the measuring process.

    The same is true for measuring hunger. The new euphemism is “food insecurity”. Activist groups (Bill Gates included) cannot be trusted to put out their own statistics.

    Crime rates, unemployment, education progress have all been victims of political manipulations of the measurement process. They changed how they measured high school graduation rates here in St. Petersburg FL and then announced they had improved graduation rates 20%. Totally bogus. With a federally mandated measurement process, it turns out they had some of the worst rates in the entire nation. It’s disgusting to see this.

    Trends in statistics are only useful if things are measured the same way consistently.

  4. Nullius in Verba says:

    There are dozens of different measures of poverty, to measure different aspects of it. Some are based on a percentage of the median salary, some on absolute poverty levels like the “dollar-a-day” extreme poverty level (which has been updated to $1.25-a-day) which may or may not be adjusted for local prices, inflation, etc.

    You can’t just look at a number labelled “poverty rate”, compare it to another number labelled “poverty rate”, and expect the two to come out the same.

    According to the World Bank’s PovcalNet, moderate poverty ($2/day) dropped 24% to 12% and extreme poverty ($1/day) dropped from 6% to 2% between 1990 and 2004. The United Nations gives 5.4% to 1.7% for extreme ($1.25/day) poverty from 1990-2005. Since Bill’s dates are different, he’s presumably talking about a different source. I don’t know where he got his numbers from – 2% looks feasible as an ‘extreme poverty’ measure but the 15% doesn’t fit.

    He may have got his data wrong, but I wouldn’t bet on it. I think there’s a good chance he’s just using a different source/definition.

    That’s the trouble with ‘data’ – you have to know how it was generated to be able to interpret it appropriately.

Leave a Reply

Your email address will not be published. Required fields are marked *