We investigate the relationship between money and happiness across the waves of the British Household Panel Study by using a latent class approach which accounts for slope heterogeneity, omitted variable bias and departures from normality assumptions. Our findings reveal the presence of a vast majority of "Easterlin-type" individuals with positive but very weak relationship between changes in income and changes in happiness and a small minority (2 percent) of "frustrated achievers" with negative relationship. Such share is much below descriptive evidence on frustrated achievement (17.5 percent). The probability of belonging to such group is shown to be positively related with divorced status and negatively related to education and relative (personal to reference group) income. Our interpretation of these results is that the standard concave money-happiness relationship provides a partial and incomplete picture of the complex nexus between happiness and income as it does not take into account two important phenomena: the role of peers and of reference group income and that of the dynamics between realisations and expectations.