Most of the proposed macro models and Phillips curves for policy design and analysis in Colombia depend on estimates of the potential output. However, it is widely known that this estimates are highly unreliable because of its level of estimation uncertainty. Following Staiger et al.(1996), we explore some common and not very common "fully structural" estimates of the NAIRU, the Non Accelerating Rate of Unemployment, provide confidence bands, and formally test the constancy hypothesis on the NAIRU. We also study the robustness of this results to the specification of the Phillips Curve. We find more reliable estimates of the NAIRU than previous estimates of the output gap, and find evidence in favor of a non-constant NAIRU. Our results indicate that it has increased about 4 percentage points along the sample span. However, uncertainty results are not robust to specification. There a single policy implication: An increasing NAIRU along with a policy of reduction and stabilization of inflation may imply increasing unemployment levels. Since the NAIRU is the component of unemployment that do not respond to monetary policy, it is up to the government to design policies for its reduction.