One way of viewing microservices is that each microservice is potentially a separate three-tier system. We don't normally implement each tier for each microservice, though. With this in mind, we see that each microservice can implement its own data layer. The benefit would be a potential increase of separation between services.
There are pros and cons to both scenarios. It is easier to deploy changes when the systems are clearly separate from each other. On the other hand, data modeling is easier when everything is stored ...