Malmquist index is commonly calculated in DEA studies looking at structural changes over time. See for example the attached google search for examples.

For statistical testing I would point you to the work of Simar but this uses FDH not DEA 

http://scholar.google.co.uk/scholar?q=structural+changes+and+malmquist+index+DEA&btnG=&hl=en&as_sdt=0%2C5&as_ylo=2013&as_vis=1

https://www.researchgate.net/profile/Leopold_Simar

If a structural change happens, then a new production possibility set arises. In that case you may have to employ a metafrontier approach where technology heterogeneity is closely associated to time heterogeneity and affects all the examined DMUs. That is, you have a frontier before the structural change and one after the structural change. If the technology gap is significant you may argue that there is a structural change. My publication may be helpful: http://www.sciencedirect.com/science/article/pii/S0377221708009958

In addition to above suggestions, you may chain the computed Malmquist index. That is relating the t-1 year index with t year. If the drop in the chain is very sharp for any given year, then obviously there is a break in continuation at that point, which you can argue as structural change, or more correctly technological progress, which is causing a shift in the frontier. Remember, DEA is a non-parametric approach and the question you are asking requires a parametric procedure for detection. 

I upload DEAwithCrossEfficient.pdf that is one of first chapter of DEA with CCR and Inverted DEA models and crossEficient published from Amazon.

I explain how to build LINGO SEts model and DEA.