PCA-Biplots as a Visual Aid to Regression Analysis: With an Application to Household Finance

Title: PCA-Biplots as a Visual Aid to Regression Analysis: With an Application to Household Finance

Abstract: PCA-biplots can be used to visualize variation in high-dimensional data. I apply these tools to household finance: studying heterogeneity in consumption propensities. PCA-biplots offer a quick and intuitive view of the dataset. This helps in designing bespoke regression specifi- cations incorporating interaction terms and non-linearities. Specifically, I test the liquidity constraint theory and unveil a finding at odds with the theory for poor elderly households: their MPCs are much lower than by liquidity constraint theory and traditional regression analysis suggested.