Foundations of Real Estate Forecasting: Approaches and Techniques

Foundations of Real Estate Forecastin\g\\❓\\gle="modal" data-bs-target="#questionModal-72649" role="button" aria-label="Open Question" class="keyword-wrapper question-trigger">G❓: Approaches and Techniques

1. Informal Approaches

• Rely heavily on market experience and intuition.
• Reflect the professional and entrepreneurial traditions of real estate.
• Emphasize market sentiment alongside market fundamentals.
• Useful for understanding current trends and quick assessments.

2. Formal Approaches

Formal approaches can be broadly categorized into:
* Quantitative
* Qualitative
* Mixed (Quantitative/Qualitative)

2.1 Quantitative Approaches

Quantitative approaches are further subdivided into two primary methods:

• Time-Series/Trend-Based Analysis:

  • Focuses on identifying patterns in historical data to project future trends.
  • Generally avoids developing explanatory theories or causal relationships.

• Causal/Structural Analysis:

  • Involves building, testing, and using models with strong theoretical underpinnings.
  • Seeks to identify and quantify the causal relationships between various factors and real estate market performance.

2.1.1 Time-Series/Trend-Based Models

• Employs historical data patterns for prediction.
• Assumes past patterns will continue into the future.
• Examples include smoothing models and regression models.
• Limited data needs and easy development.
• Useful as a baseline for comparison against other models.
• Best for basic forecasting of non-volatile data series.

• Examples: Smoothing (Moving Averages, Exponential Smoothing), Regression (Autoregressive, Partial Autocorrelation).

  Examples of Time-Series/Trend-Based Models:

  • Smoothing Models:

    Equation: TR_t = <a data-bs-toggle="modal" data-bs-target="#questionModal-72639" role="button" aria-label="Open Question" class="keyword-wrapper question-trigger"><span class="keyword-container"><a data-bs-toggle="modal" data-bs-target="#questionModal-289636" role="button" aria-label="Open Question" class="keyword-wrapper question-trigger"><span class="keyword-container">α</span><span class="flag-trigger">❓</span></a></span><span class="flag-trigger">❓</span></a> + ρ_t + β_1 * ρ_{t-1} + β_2 * ρ_{t-2} + … + β_q * ρ_{t-q}

    Critical Properties: Constant mean, Constant variance, Autocovariance is non-zero to lag q.

  • Regression Models:

    Equation: TR_t = α + β_1 * TR_{t-1} + β_2 * TR_{t-2} + … + β_q * TR_{t-q} + ε_t

    Critical Properties: Stationarity.

    Where:

    • TR_t is the total return in period t.
    • α is a constant.
    • ρ is a disturbance term.
    • β is a coefficient.
    • ε is an error term.
    • q is the period over which the model is calculated.

  Types of Time Series Techniques:

  1. Moving Averages: Calculates the average of a set of data points over a specific period.

     • Simple Moving Average (SMA):

       SMA = (P_1 + P_2 + ... + P_n) / n

       Where:

       • P_i is the data point at time i.
       • n is the number of periods.
         * Weighted Moving Average (WMA): Assigns different weights to different data points.

       WMA = (w_1*P_1 + w_2*P_2 + ... + w_n*P_n) / (w_1 + w_2 + ... + w_n)

       Where:

       • w_i is the weight assigned to data point P_i.

  2. Exponential Smoothing: Assigns exponentially decreasing weights to data points over time.

     • Simple Exponential Smoothing (SES):

       S_t = α * X_t + (1 - α) * S_{t-1}

       Where:

       • S_t is the smoothed value at time t.
       • X_t is the actual value at time t.
       • α is the smoothing constant (0 < α < 1).
         3. Autoregressive (AR) Models: Models the current value as a function of past values.

     • AR(p) Model:

       X_t = c + φ_1*X_{t-1} + φ_2*X_{t-2} + ... + φ_p*X_{t-p} + ε_t

       Where:

       • X_t is the value at time t.
       • c is a constant.
       • φ_i are the parameters of the model.
       • p is the order of the model.
       • ε_t is white noise.
         4. Autoregressive Integrated Moving Average (ARIMA) Models: Combines autoregressive and moving average components with differencing to achieve stationarity.

     • ARIMA(p, d, q) Model:

       AR: X'_t = c + φ_1*X'_{t-1} + φ_2*X'_{t-2} + ... + φ_p*X'_{t-p} + θ_1*ε_{t-1} + θ_2*ε_{t-2} + ... + θ_q*ε_{t-q} + ε_t

       Where:

       • p is the order of autoregression.
       • d is the degree of differencing.
       • q is the order of the moving average.

2.1.2 Causal/Structural Models

• Relies on theoretical relationships between real estate returns and independent variables.
• Links real estate returns with fundamental drivers such as demographics, economics.
• Independent variable changes often lead the dependent variable.
• Simplicity is favored; complex models can be difficult to calibrate.
• Examples: Single Equation (Multiple Regression), Systems of Equations.

• Require more data and are potentially more complex.

  Examples of Causal/Structural Models:

  • Single Equation (Multiple Regression):

    Equation: TR_t = α + β_1 * X_{1t} + β_2 * X_{2t} + β_3 * X_{3t} + β_4 * X_{4t} + ε_t

    Critical Properties: Independent variables identified from theory, Model diagnostics statistically significant and consistent with theory, Error term minimized.

  Where:

  • TR_t is the total return in period t.
  • α is a constant.
  • β_i is the coefficient for variable X_i.
  • X_i is an independent variable.

  • ε is an error term.

  • Systems of Equations: A nested set of individual equations. Results from one equation feed into another as an input. Each equation has strong theoretical underpinnings and rigorously tested model specification.

  Experiment Example: Impact of Population Growth on Housing Prices
  * Hypothesis: An increase in population will lead to an increase in housing prices.
  * Data: Collect annual data on population and average housing prices in a specific city over a period of 20 years.
  * Model: Develop a simple linear regression model:
  * Housing Price = α + β * Population + ε
  * Where α is the intercept, β is the coefficient that quantifies the impact of population on housing prices, and ε is the error term.
  * Analysis: Evaluate the statistical significance of the coefficient β. A statistically significant positive coefficient would support the hypothesis.

2.2 Qualitative Approaches

• Relies on expert opinions, surveys, and Delphi methods.
• Historical or geographical analogy.
• Used as alternative, check, or combination with quantitative.
• Useful for predicting turning points and non-linear changes.
• Can be used to inform judgement in the forecasting process.

• Judgements required in real estate return forecasting need to be grounded in robust qualitative techniques.

  Example: Qualitative Modeling of Yields/Capitalization Rates

  Equation: K = RFR + RP – G + D

  Critical Properties: Real estate value determined in relation to other asset types, Investors determine the risk premium required in relation to wider appetite for risk.

  Where:

  • K is the yield or capitalization rate of an asset.
  • RFR is the risk-free rate (long-term government bond rate).
  • RP is the risk premium demanded for real estate investment compared to RFR.
  • G is the long-term average rental growth rate.
  • D is the long-term average depreciation rate of the property.

  Qualitative Techniques Examples:

  1. Expert Opinion:
     • Consulting with experienced real estate professionals and economists to gather insights into market trends and future conditions.
  2. Surveys:
     • Distributing questionnaires to a large group of market participants (investors, developers, brokers) to assess their expectations and opinions about future market conditions.
  3. Delphi Method:
     • A structured communication technique that relies on a panel of experts who provide their opinions anonymously. The experts revise their opinions based on feedback from other experts until a consensus is reached.
  4. Historical Analogy:
     • Examining past market cycles and events to understand how similar conditions might influence future market performance.

3. Hybrid Approaches

• Combines quantitative methods with qualitative judgement.
• Useful for balancing statistical rigor with real-world insights.
• Involves adjusting model inputs or results based on expert opinions.
• Often more accurate than either approach used alone.

4. Forecasting Process

1. Understanding Trends and Forecasting Real Estate Occupational Markets: Primarily utilizes formal quantitative approaches, such as time-series/trend-based models for supply and causal/structural models for demand, occupancy, and rent.

2. Combining Understanding of NOI Changes with Trends and Forecasting of Capital Markets: Requires both quantitative and qualitative approaches. Quantitative methods are used to analyze NOI, while qualitative assessments are necessary for forecasting capital market trends, considering factors such as risk premiums over risk-free rates.

5. Practical Considerations

• No single “right” approach; depends on the forecaster’s goals and data availability.
• The best forecasting system is the one that works, having a significant correlation with actual results.
• Statistical diagnostic tests and back-testing are essential for quantitative models.
• Scenarios and sensitivity tests are vital due to the limited reliability of long-term forecasts.
• Model selection and specification should be carefully considered.

6. Conceptual Framework for Forecasting Real Estate Returns

1. System of Equations for Occupational Markets:

   • Model occupational markets and forecast NOI determinants using causal/structural quantitative techniques to derive the income return.

2. Qualitative Technique for Yield or Cap Rate:

   • Forecast yield or cap rate as an input to calculate capital value and hence the capital return.

3. Total Return Calculation:

   • Summation of the income and capital return to calculate the total return.

7. Forecasting Demand, Supply, Occupancy, Rent and NOI

• Demand: Causal/structural econometric techniques. Dependent on demographic, economic, and price variables. The relationship between demographic and economic factors is through employment.
• Supply: Short term supply (2-3 years) can be forecast by monitoring real estate projects or using causal/structural techniques. Long term (more than 3 years) supply forecast is projected using time-series/trend-based techniques.
• Occupancy, Rent and NOI: A system of equations.
• Yield or cap rate: Qualitative techniques may provide a reasonable basis to forecast yields or cap rates and hence to estimate the capital return.

8. Explicit Strategic and Asset-Specific Interconnection

The principal merit of the four-column approach lies in the explicit presentation of the interconnection between the strategic forecast and the asset-level forecast.

Explicitly presenting the interconnection between strategic and asset-level forecasts is useful. Making reasoned adjustments to the asset-level forecast should the outlook for the benchmark market change, is useful. The fourth column – comment – permits a clear explanation of why an individual asset may be expected to outperform or underperform the wider market.

9. Forecasting and Investment Decisions

In practice many organizations take a pragmatic view. The best forecasting system is the one that works or appears to work; that is, it appears to have some significant correlation to actual results over time. Therefore, where quantitative techniques are involved it is essential that the model is subjected to the range of relevant statistical diagnostic tests and wherever possible back-tested against historical data.

10. Limitations of Forecasting

Underlying all this has been an undercurrent that the available quantitative, statistical techniques are to some extent let down by the lack of good quality, long-term data sets both of real estate investment performance and of many of the independent variables that theory suggests would help to forecast real estate returns.

11. The future of Forecasting

This is beginning to change. As time goes by and as the industry becomes more sophisticated and data become more widely available and cost effectiv