The point estimate is typically not correct for the entire population, is typically not very accurate when looking at larger sample sizes, and is also unable, when used by itself, to show how far the estimate is from the population parameter. A single value is not usually a good way to describe an entire population, without some sort of alpha to show a confidence interval.
