From Philip Glandon, Kenneth N. Kuttner, Sandeep Mazumder, and Caleb Stroup:
We document eight facts about published macroeconomics research by hand collecting information about the epistemological approaches, methods, and data appearing in over a thousand published papers. Macroeconomics journals have published an increasing share of theory papers over the past 38 years, with theory-based papers now comprising the majority of published macroeconomics research. The increase in quantitative models (e.g., DSGE methods) masks a decline in publication of pure theory research. Financial intermediation played an important role in about a third of macroeconomic theory papers in the 1980s and 1990s, but became less frequent until the financial crisis, at which point it once again became an important area of focus. Only a quarter of macroeconomics publications conduct falsification exercises. This finding contrasts with the year 1980, when these empirical approaches dominated macroeconomics publishing. Yet the fraction of empirical papers that rely on microdata or proprietary data has increased dramatically over the past decade, with these features now appearing in the majority of published empirical papers. All of these findings vary dramatically across individual macroeconomics field journals.
Here is the SSRN link.