This paper investigates the existence of speculative bubbles in the US national and in 23 regional housing markets over three decades (1978-2012). A new method for detecting exuberance in housing markets is proposed. By taking changes in the macroeconomic conditions (such as interest rate, per-capita income, employment, and population growth) into consideration, the new method provides a better control for housing market fundamentals and thereby it is expected to significantly reduce the chance of false positive identification. Compared with the method of Phillips, Shi and Yu (2015a,b), the new approach finds a dramatic reduction in the number of speculative housing markets and shorter bubble episodes in the US. It locates only one bubble episode in the early-to-mid 2000s over the whole sample period in the national housing market. At the regional level, it identifies three periods of speculation: late 1980s, early-to-mid 2000s, and the post-crisis period in 2011-2012. The early-to-mid 2000s bubble episode is the most severe one involving nine major metropolitan statistical areas.