The interest premium for left government: Regression-discontinuity estimates

This paper employs a regression-discontinuity design (RDD) to ascertain the effects of left government on the interest-rate premium that markets build into government-bond prices. One advantage of this approach is that RDD does not require, as have some previously employed strategies, strong assumptions about how market actors form political expectations, about the quality and dissemination of political information, or about functional forms or explanatory-variable selection. We expand from previous RDD studies in exploring effect heterogeneity, namely, whether particular political-economic conditions produce larger or smaller interest costs of left government. Our findings suggest no or very small and insignificant partisan-government effects except under specific circumstances: sharp governing alternatives (low fragmentation and high polarization), in certain eras (around the 1950s-1970s), and for a short term (about 1 year). Under these conditions of stark differences between alternative left/right governments and relatively great domestic policy autonomy, however, there is a statistically discernible and substantively notable government-bond yield increase after left parties enter government following close elections.