Economic activity is unevenly distributed across space, or spiky. Measuring this spikiness is not trivial. A good measure is comparable across space, comparable across sectors, unbiased regarding spatial and sector classification, and should provide a measure of significance. No measure fulfils all these criteria, but some are better than others. Once spikiness is identified the next question is ‘so what?’ Does spatial agglomeration stimulate productivity? Econometric methods to deal with this question and tackle the problem of reverse causality are: difference-in-differences, natural experiments, and regression discontinuity design. These techniques are introduced in this chapter.