Observables can be directly counted or measured, once a definition  
has been selected. 
    For example, inflation requires an agreement (not commonly  
accepted, viz. the "Griliches et al.  report" on the proper use of  
index numbers in U.S. social policy) on the selection of  basket of  
goods and services and  the way in which the corresponding prices are  
measured in (at least) two different periods. 
   GDP measurement  is the outcome of a process whereby some money  
flows in the economy are estimated; nonmoney flows (such as  
ecological damage and social degradation following from the  
generation of GDP) are commonly disregarded -- but, otherwise, GDP  
can  be counted "objectively". 
    Unemployment, once the definition of an actual labour force and a  
potential labour force has been made, is a matter of counting.  
Needless to say that measurement and counting errors are always  
present. Intelligent economists and policy makers take these into  
account. 
 
NAIRU is statistically derived from a complex estimation process of  
the operations of an economy, in which process some assumptions about  
human behaviour are made. Indeed, one should say that forms of human  
behaviour is imputed (assumed to be present) in the way the model is  
set up. Take away the imputations and NAIRU stands stark naked. The  
reason why it is so popular in some circles is that its  use  
justifies attacks on the welfare state and labour rights -- nothing  
objective about that, and not at all in the same league as  the  
measurement  (even when subject to error) of inflation, GDP and  
unemployment 
 
Jesse Vorst