We investigate whether nonlocal buyers of real estate pay different prices for similar assets as compared to local buyers. An efficient real estate market would operate against this but empirical studies leave the question unresolved. While some studies fail to detect or confirm price differences (Turnbull and Sirmans, 1993; Watkins, 1998; Clauretie and Thistle, 2007), other studies do detect them (Miller et al., 1988; Lambson et al., 2004; Ihlanfeldt and Mayock, 2012). Most studies of the question have utilized residential real estate and have focused on owner-occupiers. Two studies have used investors, but these have each considered only a single market (Phoenix; Las Vegas). We extend the literature in this study by examining investors in office properties across 138 US markets and, uniquely, by considering the question of price differences when investors sell as well as buy. We investigate this using a CoStar database of 10,971 purchases and 11,444 sales between 1996 and 2012. Aware of limitations in some previous studies, we use a propensity score matching approach to produce matched samples that reduce the potential effects of conflating factors such as selection bias, investor clienteles (Wiley, 2012) and marketing duration. We find nonlocal investors significantly overpay on purchase by an estimated 13.8% relative to similar assets purchased by local investors. We attribute this to a combination of information asymmetry and anchoring (by those investors from markets whose prices are higher than where they buy). We find that, upon exit, nonlocal investors sell a discount of 7% relative to similar assets sold by local investors. This is again attributable to information asymmetry but is not explicable by anchoring. Overall, nonlocal purchases and sales result in significant relative capital value underperformance.