Asset-Specific Forecasting: Methods and Approaches

Asset-Specific Forecasting: Methods and Approaches

Understanding the Asset-Specific Context

Asset-specific forecasting delves into the unique characteristics❓ and circumstances of an individual property to predict its future performance. This is a more granular approach than market-level forecasting, requiring a deeper understanding of both the asset and its immediate environment. It’s important to note that “asset-specific” may refer to the physical characteristics of the property, its location, or the specifics of its management and leasing strategy.
The merit of asset-specific forecasting lies in explicitly presenting the strategic forecast and the asset-level forecast. Adjustments can be made based on the benchmark market changes. The most important is the comments, as it permits a clear explanation of why an individual asset may be expected to outperform or underperform the wider market.

Informal Forecasting Approaches

Informal approaches rely heavily on market experience, intuition, and a deep understanding of market fundamentals. These approaches often incorporate market sentiment alongside an appreciation of underlying market drivers.
* Market Sentiment Analysis: Gauging the prevailing optimism or pessimism among investors, developers, and tenants.
* Experience-Based Judgments: Leveraging past successes and failures to anticipate future trends.
* Qualitative Data: Anecdotal evidence, industry gossip, and expert opinions.

Formal Forecasting Approaches

Formal approaches to forecasting can be broadly categorized into quantitative, qualitative, or a combination of both.

Quantitative Methods

Quantitative methods use numerical data and statistical techniques to develop forecasts. They can be further subdivided into time-series/trend-based and causal/structural models.

Time-Series/Trend-Based Models:

These models analyze historical data to identify patterns and trends, projecting them into the future. They generally don’t seek to explain the why behind the trends, focusing instead on identifying repeatable patterns.

• Assumptions: The past is a good indicator of the future.
• Strengths: Limited data needs, quick and easy to develop, useful baseline.
• Weaknesses: Lack of explanatory power, less accurate for longer forecast horizons.

Examples:

• Moving Averages: Calculates the average of a data series over a specific period.
• Exponential Smoothing: Assigns exponentially decreasing weights to older observations.
• Regression Models: Using regression analysis to predict future values based on historical trends.

Examples of time-series/trend-based models for total returns:

Smoothing Models

TRt = α + ρt + ß2ρt-1 + ß2ρt-2 + … + ßqρt-q

Regression Models

TRt = α + ßTRt-1 + ßTRt-2 + … + ßTRt-q + εt

Critical properties:

Constant mean, Constant variance, Autocovariance is non-zero to lag q, Stationarity

Terms:

TRt is the total return in period t

α is a constant

ρ is a disturbance term

ß is a coefficient

ε is an error term

q is the period over which the model is calculated

Causal/Structural Models:

These models aim to identify the underlying drivers of real estate performance and build relationships between independent variables (e.g., economic growth, demographics) and the dependent variable (e.g., property value, rental income).

• Assumptions: A strong theoretical underpinning linking independent variables to real estate performance. The past relationship of the dependent variable to one or more independent variables
• Strengths: Can provide insights into the why behind market movements, potentially more accurate for longer-term forecasts.
• Weaknesses: Requires more data, more complex to develop, relies on accurate forecasts of independent variables.

Examples:

• Multiple Regression: Using multiple independent variables to predict the dependent variable.
  • Equation: TRt = α + ß1X1t + ß2X2t + ß3X3t + ß4X4t + εt
    • TRt is the total return in period t.
    • α is a constant.
    • ßi is the coefficient for variable Xi.
    • Xi is an independent variable.
    • ε is an error term.
  • Critical properties: Identify independent variables from theory. Model diagnostics should be statistically significant and consistent with theory. Error term ε should be minimized.

• Systems of Equations: A nested set of equations where the output of one equation becomes the input of another. Each equation must have strong theoretical underpinnings and rigorously tested model specification.

  • Trade-off between simplicity and complexity. The single-equation approach can be a little more “grey box”, as the interaction between variables is not made explicit. The system of equations has a practical advantage as it allows the forecaster more explicitly to understand the intermediate steps.
  • Greater complexity triggers questions about the potential for positive and negative feedback loops within the model and a prerequisite of the approach is that the nested system of equations should be stable.

Qualitative Methods

Qualitative methods rely on expert opinion, surveys, and other non-numerical data to develop forecasts.

• Expert Opinion/Surveys: gathering insights❓ from industry professionals, appraisers, and other knowledgeable individuals.
  • Advantages: Consensus, grounded in a ‘market view’, can be well suited to the prediction of turning points and non-linear changes
  • Disadvantages: Temtation for expert opinion to read too much into data than is warranted, suffer from ‘anchoring’
• Delphi Method: A structured process for gathering and refining expert opinions.
• Historical or Geographical Analogy: Drawing comparisons to similar assets or markets to anticipate future performance.

Qualitative techniques can be used in a number of ways:
* Whether to use a quantitative model.
* Which quantitative model(s) to use.
* Which variable(s) to use in the model.
* Whether to adjust the model inputs or results manually.

Forecasting Income Return

In forecasting the income return the emphasis is on understanding trends and forecasting real estate occupational markets in order to forecast NOI. This lends itself to formal quantitative approaches – perhaps time-series/trend-based for supply and causal/structural for demand, occupancy and rent.

Forecasting Capital Return

In forecasting the capital return, an understanding of the likely change in the NOI needs to be combined with an understanding of trends and forecasting of capital markets. This requires both a quantitative and qualitative approach, the latter in the forecasting of capital market trends – perhaps with reference to a forecast risk premium over a forecast risk-free rate.

Yields and Cap Rates

Qualitative modelling of yields/capitalisation rates:

K = RFR + RP –G + D (The equation adopts the form of a Gordon’s growth model)

Critical properties:

Real estate value is determined in relation to the performance of other asset types. Investors will determine the risk premium required in relation to wider appetite for risk.

Terms:

K is the yield or capitalisation rate of an asset

RFR is the risk-free rate (usually taken to be the long-term government bond rate of the country in which the asset is located)

RP is the risk premium that an investor would demand for an investment in real estate compared to the RFR

G is the long-term average rental growth rate

D is the long-term average depreciation rate of the property (or the annual average amount of investment required to maintain the quality of the asset)

Practical Applications and Examples

• Retail Property: Consider a shopping center revitalization project. A time-series model could be used to project foot traffic based on historical data, while a causal model could incorporate demographic changes, consumer spending patterns, and the impact of anchor tenants on overall performance. Expert opinions could be surveyed regarding the potential impact of competing retail developments nearby.
• Office Building: Forecasting occupancy rates for an office building might involve a causal model that considers local employment growth, industry trends, and the building’s amenities. A qualitative assessment of tenant creditworthiness and lease renewal probabilities would also be crucial.
• Multifamily Apartment: Analyzing historical occupancy rates, rent growth, and operating expenses using time series models can provide a baseline forecast. Incorporating factors like local job growth, population trends, and the availability of competing units using causal models can refine the prediction. Qualitative assessments of neighborhood quality and property management effectiveness are important considerations.

Demand and Supply Forecasting

Demand Forecasting:

• Demand for real estate (commercial or residential) is a product of the number of people in a defined trade area and their average wealth, and inversely related to the price of real estate in the previous period.
• Equation (Example): Demandt = α + β1*Populationt + β2*Wealtht - β3*Pricet-1 + εt
  • Where Demandt is the demand at time t, Populationt is the population, Wealtht is average wealth, Pricet-1 is the price in the previous period, and α, β1, β2, β3 are coefficients.

Supply Forecasting:

• Short Term (2-3 years): Either monitoring real estate projects under construction and building a granular database or simple causal/structural techniques.
• Long Term (3+ years): Projecting completions using time-series/trend-based techniques.

Model Diagnostics and Back-Testing

It is essential that any quantitative model be subjected to the range of relevant statistical diagnostic tests and wherever possible back-tested against historical data.