How to explain to non-statisticians that if we have e.g. obesity rate equal to 40% in men and 12% in women, the target values in men and women in 2030 should NOT be equal to 38% in men and 10% in women? Do you know any paper/guide that would explain it to policy-makers that one should take into account the baseline level, when assuming the target values? Actually, I would like to show that time series of social indicators are more likely to show exponential trends instead of linear trends. Any help will be appreciated.
First, there is a difference between a forecast (what we expect to happen given specific circumstances) and a target (what we want to happen, and are prepared to change circumstances to achieve).
Second, to your question: I believe that this point is so trivial that nobody would publish anything on it. And it should be easily explainable. I doubt that someone who will not or cannot follow the explanation would be convinced by a reference.
You may have more luck in framing the development differently. "Obesity incidence will be reduced by 10% in both men and women" is a relative reduction and translates into a reduction from 40% to 36% in men and from 12% to 10.8% in women.