Market observers are questioning whether Hong Kong property prices have finally hit bottom. In a June article, I developed a property price model using lagged house prices, GDP with lag terms, inflation and interest rates. The residuals appeared to follow random noise patterns, suggesting the model's accuracy was reasonably sound, if not perfect.
That analysis assumed $X'e = 0$ and estimated coefficients using the standard formula $\hat{\beta} = (X'X)^{-1}X'y$ to measure how various factors influence property price movements. This approach imposes no distributional assumptions on the error terms $e$, simply ensuring $X$ and $e$ remain uncorrelated while maximising the relationship between $X$ and $y$. This constitutes the Generalised Method of Moments (GMM), which demands little from residuals but delivers relatively inefficient estimates.
