You can easily find studies that cut both ways. Ultimately it depends on how and what you're measuring and how you adjust your data. The economic and social costs of smoking/bad diet are undeniably greater than healthcare savings from early death. If someone had worked an extra 20 years instead of dying from lung cancer at 50, their extra productivity likely exceeds the extra healthcare costs. Let's say they incur $50,000 in costs from cancer at 50 and avoid $100,000 in costs from death; if the worker earned $30,000 a year, $600,000 of economic value is lost. And I'm connecting disease with bad diet, not obesity. You can eat a terrible, terrible diet that skyrockets health risks and not be obese.