I am evaluating impacts of major health financing policy changes in Georgia (country). The database is household level and it is not panel data. Continues outcome variable is out-of-pocket health spending (OOPs) and it exhibits skewed distribution as well as seasonality. The residuals are positively autocorrelated. The regression also takes independent variables connected with each household characteristics into account. My goal is to evaluate impact of health policies on the financial wellbeing of population connected with health care utilization determinants. Should I aggregate the dataset or keep it as it is?