Table of Contents

Log Periodic Power Law Singularity (LPPLS) model

The asset price p(t) follows a standard diffusive dynamics with varying drift (or conditional expected return) μ(t) in the presence of discontinuous jumps (where σ (t) is the volatility and dW is the increment of a Wiener process, dj represents a discontinuous jump $E_{t}[dj] = h(t)dt$):

\[\frac{dp}{p} = \mu(t)dt + \sigma(t)dW - \kappa dj\]

LPPLS:

\[E[\ln p(t)] = A + B | t_{c} - t |^{m} + C|t_{c} - t |^{m}cos(\omega \ln |t_{c} - t| + \phi)\]

Rent-Price Ratio

Rent-Price Ratio (price / rent-annual)

  • High Ratio (>20): potential overvaluation or lower rental yields.
  • Moderate Ratio (15-20): a balanced market .
  • Low Ratio (<15): buying may be more advantageous than renting.
  • Interest Rates: Lower interest rates make mortgages cheaper, potentially increasing property prices.

Papers

Real estate price forecasting methods (from The Oxford Handbook of Economic Forecasting):

  1. Uses lagged return. housing price changes exhibit positive serial correlation.
    • But difficult to disentangle spurious correlations from actual market inefficiencies.
  2. Uses valuation ratios : the rent-price ratio, price-income ratio.
    • But not be able to capture fully the time variations in the conditioning set.
  3. Evidence of the relevance of property, and/or region-specific, economic variables : such as demography, income, construction costs, and zoning restrictions.

Understanding the impact of city government relocation on local residential property prices in Hangzhou, China 2024. The impact of government relocation on the property market.

  • The impact is spatially concentrated and temporally transient.
  • The average treatment effect over the seven-year post-relocation period was positive (4.76%), but it did not reach statistical significance.
  • There was a significantly positive effect (12.7%) on transaction prices within a 5 km radius of the new city hall during the first year following relocation; however, this effect diminished rapidly in subsequent years.
  • For transactions within 1 km of the new government site, the significant positive effect persisted for five years.

Forecasting Methods for the Real Estate Market: A Review 2023

  • Time Series Analysis:
    • ARIMA Models (AutoRegressive Integrated Moving Average)
    • Exponential Smoothing. work for seasonality.
    • Regression Analysis. no non-linearity captured.
  • Hedonic Pricing Model, based on the fundamental characteristics of the property.
  • Artificial Intelligent Modeling.
  • Econometric Forecasting Method : (1) VAR Models (Vector Autoregression); (2) Panel Data Models.
  • Spatial Analysis Forecasting Methods: (1) Geospatial Forecasting; (2) Spatial Autoregressive Models (SAR).

Anticipating critical transitions of the housing market: new evidence from China 2019 detect housing bubbles by finding the evidence of market inefficiencies. use Log Periodic Power Law Singularity (LPPLS) model to predict critical time (~ bubbles break time).

House Price Index Construction in the Nascent Housing Market: The Case of China 2014. Chinese housing market is facing a greater risk of mispricing than reported by the existing official metrics.

  • Main methods:
    1. the simple average method without quality adjustment.
      • without the substantial complex-level quality
    2. the matching approach with the repeat sales modeling framework.
      • without effect of developers’ pricing behaviors.
    3. the hedonic modeling approach.
  • Average Price Index, 70 Cities Index. mistrusted and widely criticized: conflict, underestimated,
  • Newly-built sector prices could reflect real pricing.
  • Hedonic modeling: $P_{ijt} = \alpha OU_{it} + \lambda UU_{it} + \beta OC_{jt} + \phi UC_{jt} + \theta PB_{ijt} + \delta_{t} D_{ijt} + \mu_{ijt}$
    • U : unit-level housing characteristics; C: complex-level housing characteristics. O: observed; U: unobserved. PB : developers’ pricing behaviors; D: time dummies; μ : error term.