Two researchers crunching population statistics have confirmed an unsettling reality. Siwan Anderson and Debraj Ray noticed the ratio of women to men in developing regions and in some cultures is suspiciously below the norm.

    In India, China and sub-Saharan Africa, millions upon millions of women are missing. They are not lost, but dead: victims of violence, discrimination and neglect.

    A University of British Columbia economist is amongst those trying to find them - not the women themselves, who are long gone, but their numbers and ages, which paint a sad and startling picture of gender discrimination in the developing world.

    The term "missing women" was coined in 1990, when Indian economist Amartya Sen calculated a shocking figure. In parts of Asia and Africa, he wrote in The New York Review of Books, 100 million women who should be alive are not, because of unequal access to medical care, food and social services. These are excess deaths: women "missing" above and beyond natural mortality rates, compared to their male counterparts.

    Women who are dead because their lives were undervalued.

    Around the world boys outnumber girls at birth, but in countries where women and men receive equal care, women have proved hardier and more resistant to disease, and thus live longer. In most of Asia and North Africa, however, Sen found that women die with startlingly higher frequency.

    His research began a flutter of activity in academic circles and by 2005, the United Nations produced a much higher estimate for how many women could be "missing": 200 million.

    From her office at the University of British Columbia, economics professor Siwan Anderson has been crunching numbers to try and understand why so many women are dying. "If you're interested in gender discrimination, it's really one of the starkest measures of discrimination, because it's women who should be alive, but aren't," she says.

    The 40-year-old researcher recently co-authored a paper with New York University's Debraj Ray, focusing on figures from China, India and sub-Saharan Africa for the year 2000. What they discovered flew in the face of existing literature and commonly held beliefs about the missing women phenomenon.

    "Previously, people had thought that they (the missing women) were all at the very early stages of life, prenatal or just after, so before four years old," Anderson says. "But what we found is that the majority are actually later." Female infanticide has been endemic in India and China for some time, which she says led researchers to assume that it was the source of all the missing women. But the truth is much more complicated.

    Once she and Ray broke down the numbers by age group, they found that the majority of excess female deaths came later in life: 66 per cent in India, 55 per cent in China and 83 per cent in sub-Saharan Africa.

    One of their colleagues in the economics department at the University of British Columbia says this finding is striking, and points the way for future research and advocacy.

    "Why would there be excess mortality of, let's say, 45-year-old women versus 45-year-old men?" asks economics professor Kevin Milligan. "And what they find is ... they have the same set of diseases, they just seem to die more frequently. The explanation that seems most consistent with that is differential access to health care. And so that's a really striking finding."

    Anderson says that lack of health care is likely a big part of the problem, but that there are numerous cultural and social factors at play that can be difficult to pinpoint.

    In their "elementary accounting exercise" published this February, Anderson and Ray began to plot the causes of excess death in 2000 by age group, and produced some interesting figures.

    In sub-Saharan Africa, the dominant source of missing women was HIV and AIDS, the cause of more than 600,000 excess female deaths each year.

    In China, Anderson says, most of the 141,000 excess female deaths by injury were suicides, making China the only place in the world where women are more likely than men to kill themselves, often by eating pesticides used for crops.

    And in India, a category called "injuries" yielded ominously high figures: 86,000 excess deaths in the age group 15-29 in 2000 alone. Anderson has done extensive research in India, and says the numbers beg the question of exactly how many deaths were so-called "kitchen fires" - often used to mask dowry-related killings, the result of a new bride being tortured by her new family until her parents pay their debts.

    Contrary to what you might expect, Anderson says, dowry prices have not dropped off with improvements in education in India. Instead, they have gotten worse, with educated brides and their families willing to pay even more for high-quality grooms.

    Anderson says dowry payments can be six times a family's annual wealth - an excruciating price, especially for poor villagers. The implications of this hefty sum trickle down to the first moments of a child's life. While conducting recent field work in India, Anderson asked villagers about selective abortions and found them open about the fact that they use ultrasound to determine the baby's gender and help them decide whether or not to keep it.

    "They see no other options," she says. "They really cannot afford to have a daughter."

    Future research will delve deeper, seeking answers to questions such as: How often are men given mosquito nets to protect themselves from malaria, but not women? How many women die because they are not taken to the hospital when they are sick?

    Anderson is using data gathered primarily from the World Bank, the United Nations and the World Health Organization, but admits that getting the figures can be a huge challenge. In sub-Saharan Africa, for example, many deaths go undocumented, and in India, it is virtually impossible to know how many "unintentional" deaths are actually dowry killings, because they are not accurately reported to the authorities.

    It is also difficult to separate direct gender discrimination from biological, social, environmental, behavioural and economic factors. That will be part of the task as Anderson works on calculating missing women by region in India, and isolating gender discrimination from other factors that might contribute to uneven male-to-female ratios.

    When asked what can be done to combat such deep-seated inequality, Anderson pauses. Even when governments outlaw root causes, such as the Indian dowry system, violence persists, she says. "It's too embedded in the system in their world."