Is economics reductionism or emergentism ?Reductionism is the idea that knowledge at a higher level can be deduced from the entities and their interaction at a lower level.Emergentism rejects reductionism, and olds that even though some of the properties of the whole cannot be deduced from its properties of the parts and their interaction, that some of the organisational principles at the higher level organisation cannot be reduced organisational principles at a lower level.
There is not a correct or a wrong answer to this question. The approach to economics continously changes (economists rarely arrive to precise and widely accepted answers to their research questions, every new model will sooner or later be replaced by a (maybe) "better" one).
Classical economists would be probably similar to what you call "emergentism": hypothesis were made directly to the aggregated economy (think about Say's law or quantity theory of money: there was no attempt to build these theories from the bottom, i.e. from a lower level).
For a long time, economists would not question this approach (Keynes' general theory follows the same approach: building a model starting directly from aggregate hypothesis), same for Neoclassical (ex. Hicks' very famous IS-LM model or Phillips curve).
It was not until the seventies that the debate between "recuctionism" and "emergentism" became crucial in economics debat. In 1976 a certain gentleman called Robert Lucas proposed the so called "Lucas critique". The critique was a critique to large econometric models in fashion at the time, because estimating parameters is close to useless if you have no reason to believe that they will remain constant over time. The solution proposed by Lucas was to build models from the bottom using representative rational agents with rational expectations (you would call it a sort of reductionism, even though interactions are not yet crucial because of representative agent hypothesis). Even though it was not completely accepted at the time (another gentlman called Solow heavily criticez Lucas critique), it ended up having an enormous impact on the development of economic literature. Since then microfounded models became increasingly popular: think about New Keynesian school with their "DSGE" models or about "ABM" models.
In particular ABMs are probably the model that are the most similar to "reductionism": they are essentially built by making some hypothesis on the behavior of single agents (possibly heterogeneous), that will start to interact among them generating models' dynamics. On the other hand nowadays it seems that pure reductionism is outdated, and researchers/professors that work with these models often use parameters independent of interactions among agents in order to reproduce empirical macro trends that don't emerge from these interactions, so it seems they are closer to emergentism.
Anyway, this is not a topic on which I expect uniformity (it rarely happens in economics!), so I wouldn't be surprise if other answers will show different point of views or if this will change in the future.