Imagine you're a vegetarian avoiding meat, an environmentalist reducing power consumption, or a fair-trade supporter who buys fair trade coffee.

You might naively expect that not eating a chicken saves a chicken. This is not true. Likewise, for the other cases. Likewise, if you go out of your way to eat chicken, this doesn't cause an entire chicken to die.

The reason? Market power. Just as your decision to avoid eating chicken slightly reduces the demand for chickens (in expectation), it also slightly reduces the price of chickens (in expectation). This fall in price, causes people to eat slightly more chickens. While these second-order effects don't negate the expected impact of avoiding chicken, they do reduce it.

It can help by considering edge cases. Imagine the demand for chicken is perfectly elastic. For instance, people might only buy chicken if it costs less than say $2 per pound. In this case, producers will keep farming chickens until they reach that marginal cost - regardless of whether you buy chicken. Hence, your decision to avoid chicken would have literally 0 impact on the number of chickens slaughtered.

On the other hand, if the demand for chicken is perfectly inelastic, then your decisions will not affect the price at all, meaning that every chicken you avoid is one saved.

It turns out there's an elegant formula for determining your true ethical impact. If you don't eat one chicken, the number of chickens saved (in expected value) is given by $$\frac{\epsilon_S}{\epsilon_D + \epsilon_S}$$ where $\epsilon_S$ is the elasticity of supply and $\epsilon_D$ is the elasticity of demand.

Of course, this doesn't just apply to avoiding meat. It applies to all purchasing decisions. Similarly, if you make decisions with ethical costs, not accounting for market power will overstate how much harm you do.