One of the surprising things I noticed when I was poring over data from the NHANES survey (US CDC National Health and Nutrition Examination Survey) from 1975 to 2006 is that the number of inactive people has diminished in that same time period from 50% to 24%. This is shocking to most people. We have this romanticized idea that in the 1970s people were more active, as if everyone chopped wood and walked 15 miles to work in the morning. The reality is, there were office jobs, housewives and cars without the large numbers of runners and gym-goers we have today.
Granted, NHANES data are self-reported and should be taken with a grain of salt. However, Chris at Conditioning Research pointed me to a study looking at changes in energy expenditure from the 1980s to the present in North America and Europe. It doesn't suffer from the same biases because it's based on direct measurement rather than self-reporting. Here's the executive summary: we're expending slightly more energy than we used to, partly because we exercise more and partly because it takes more energy to move our heavier bodies around.
I'm certainly not blaming the obesity problem on an increase in physical activity, but I do think we can safely rule out inactivity as the reason we've gotten fatter. In my mind, this only leaves one major possible cause for the obesity epidemic: changes in diet. Don't get me wrong, I think exercise is good. It has numerous positive effects on physical and mental health. But it's not as powerful of a tool for fat loss and general health as diet.
Anecdotally, I do know several people who lose fat when they exercise regularly. I also know some who don't lose fat when they exercise. Exercise and a healthy diet converge on some of the same metabolic pathways, such as sensitivity to insulin. But diet changes are far more effective than exercise at correcting metabolic problems. The reason is simple: the problems a person corrects with a good diet are caused by a poor diet to begin with.