Based on a sample of 296 stocks from the S&P 500, the time-varying network structure within three distinct two-year periods since the beginning of the 21st century was analyzed. Logged first-differences of daily stock prices serve as input for a correlation-based distance measure between any two of the 296 stocks. The computation of a Minimal Spanning Tree then abstracts from a complete network and allows for a topological analysis of the resulting community structure. Both the Great Recession (2007–2008) and the Global Commodity Crisis (2010–2011) reveal tendencies of enhanced community formation compared to a formerly rather randomized network structure. Nevertheless, the drivers of the resulting clustering are found not to be related to industry sector affiliation.