By using a unique 17-source database that combines individual survey data, air pollution data at the city and country level, climate information and macroeconomic indicators, this paper explores the relationship between happiness and air pollution in Latin America. Three different analyses are carried out: a macro analysis, in which a country level panel data is used; a hybrid analysis, using country level air pollution data and individual level happiness and controls; and a micro analysis with city level pollution data and individual level happiness data and controls. The comparison of the three analyses shows the distortionary effect that aggregation bias has on the estimation results. From the preferred model, i.e. ordered probit estimation of the micro analysis, it follows that the probability of a person answering “not at all satisfied” with his/her life grows from 0 to 0.65 when the annual mean concentration of PM10 increases within the range observed in the sample. The predicted probability of answering “very satisfied” with his/her life decreases from 0.94 to 0.13 when the annual mean concentration of PM10 increases within the same interval. These results are consistent with those found in the previous literature.By using a unique 17-source database that combines individual survey data, air pollution data at the city and country level, climate information and macroeconomic indicators, this paper explores the relationship between happiness and air pollution in Latin America. Three different analyses are carried out: a macro analysis, in which a country level panel data is used; a hybrid analysis, using country level air pollution data and individual level happiness and controls; and a micro analysis with city level pollution data and individual level happiness data and controls. The comparison of the three analyses shows the distortionary effect that aggregation bias has on the estimation results. From the preferred model, i.e. ordered probit estimation of the micro analysis, it follows that the probability of a person answering “not at all satisfied” with his/her life grows from 0 to 0.65 when the annual mean concentration of PM10 increases within the range observed in the sample. The predicted probability of answering “very satisfied” with his/her life decreases from 0.94 to 0.13 when the annual mean concentration of PM10 increases within the same interval. These results are consistent with those found in the previous literature.PFCHA-BecasPFCHA-Becas