The demand for real estate

General Premises

  • Real‐estate (RE) demand must be analysed by use type because each segment (housing, office, retail, industrial) has its own drivers, constraints and interactions.
    • Land is fixed  aggregate supply is highly inelastic in the short run.
    • Buildable capacity on a parcel is limited by technology & regulation.
    • Location is non-replicable: two properties cannot occupy the exact same spot.
  • Inter-market spill-overs
    • Excess demand in one segment (e.g.
      offices) can induce conversion of stock from another segment (e.g.
      residential).
    • Short- vs long-run elasticity: in the very long run buildings may be demolished, redeveloped or repurposed.
  • Supply lags and conversion frictions create cyclical behaviour that links RE markets to macro-business cycles.

Office Space Demand

  • Distinctive traits versus housing
    • Higher prevalence of vacant space.
    • Demand is tightly linked to business cycles & (office) employment.
  • ALWAYS some vacancy → Natural Vacancy Rate (NVR)
    • Analogous to the Natural Unemployment Rate in macroeconomics.
    • At equilibrium the market clears with NVR > 0 because of:
      • Frictional vacancy (search time to locate/replace tenants).
      • Value of waiting (option value of postponing lease to capture higher future rents).
  • Stylised market cycle (Wheaton-type):
    1. Boom → ↑ demand, ↑ rents, ↓ vacancy (short run).
    2. Sustained boom → ↑ construction → ↑ supply, ↓ rents, ↑ vacancy (medium run).
    3. Slowdown → ↓ demand, ↓ rents, ↑ vacancy.
    4. Prolonged recession → ↓ construction/maintenance → eventual ↓ supply, rents stabilise/↑.
  • Imperfect correlation with macro cycle
    • Diverse shocks; spatial contagion; size of existing NVR acts as buffer (larger NVR smooths transmission).
    • Empirical example: Dublin NVR ≈ 5 % (1978-98) vs 15 % (1999-2009) → slower price reaction in latter period (McCartney, 2010).

Determinants of the Natural Vacancy Rate

  • Shorter lease terms → ↑ tenant turnover → ↑ NVR.
  • Contractual/regulatory restrictions on termination → longer matching search for both parties → ↑ NVR.
  • High construction rate (e.g.
    low interest rates, permissive zoning) → rapid space creation → ↑ NVR.
  • Optimistic rent expectations → landlords exercise option to wait → ↑ NVR.
  • High tenant heterogeneity (specialised fit-out needs) → ↑ search value → ↑ NVR.
  • Inflation + restricted indexation → landlords leave space vacant to renegotiate at higher nominal levels → ↑ NVR.
  • Real interest rate effect:
    • High real r → holding vacant incurs higher opportunity cost → ↓ desired vacancy.
    • Low real r → opposite.

Quantitative Office-Demand Models

  1. Hendershott-Lizieri-MacGregor (HLM, 2008) – long-run demand only ln(D(R,E)) = \lambda0 + \lambdaR \, ln(R) + \lambda_E \, ln(E)
    • Treats employment (E) & rent (R) as log-elastic drivers; ignores vacancy dynamics.
  2. Wheaton-Torto-Evans (1997) – joint demand/supply Potential demand: OSt^* = \alpha0 + Et(\alpha1 + \alpha2 R{t-1}) Construction: Ct = \beta0 + \beta1 Rt + \beta2 Vt + \beta3 It + \beta4 RCt
    • Supply responds to rent–replacement-cost wedge, vacancies and interest rates.
  3. Brunes (2010) – investment focus Pt = \alpha + \beta \, TQ{t-m}
    • TQ is Tobin’s Q (price/replacement-cost ratio) lagged m years → predicts construction starts.
  4. Fuerst (2006) – enriched potential demand OSt^* = \alpha0 + Et\big(\alpha1 + \phi1 (Et - E{t-1})/Et - \phi2 R{t-1}\big) + Z_t
    • Adds employment growth, space-per-worker & exogenous shock dummy Z_t (e.g.
      9/11).

Information set needed for market analysis

  • Stock and current vacancy (survey of buildings, occupancy, space/tenant).
  • Pipeline supply: approved, under-construction, proposed.
  • Demand indicators: realised rents, employment growth; estimate using models above.

Retail Space Demand

  • Choice of location aims at profit maximisation: maximise expected revenues – minimise costs.
  • Complex calculus mixing economic, marketing, geography, strategy and logistics.

Small-scale Trader – Location Methods

  1. Checklist / Scorecard
    • List success factors (nearby customers, income, parking, competition…).
    • Assign weights summing to 100 %.
    • Score each potential site; compute weighted sum → ranking.
    • Pros: intuitive, cheap; Cons: subjective, unsuitable for large capital sums.
  2. Analogue method
    • Benchmark similar shops in candidate areas (survey-based).
    • Pros: empirical, inexpensive; Cons: subjective variable selection, ignores competition/distance explicitly.

Large-scale Retail (Malls) – Gravity & Spatial Models

  • Gravitational intuition: customer attraction ∝ “mass” (population, spending power) / (distance)².

Converse Break-Point Model (1949)

  • Trade ratio: \frac{TA}{TB} = \frac{PA}{PB} \left(\frac{DB}{DA}\right)^2
  • Distance from city A where a shopper is indifferent:
    DA = \frac{D}{1 + (PA/PB)^{1/2}} Example with PA = 300k, PB = 100k, D=20 km → DA \approx 7.3 \text{ km} (≈ one-third distance, closer to larger city).
  • Extensions: pairwise break-points for multiple towns.
  • Limitations: ignores travel cost heterogeneity, existing competition, transport infrastructure.

Modified Break-Point (competition-adjusted)

  • Swap population ratio to push optimal site away from dominant metro: DA = \frac{D}{1 + (PB/P_A)^{1/2}}
    • Suitable where big city already saturated with retail choices.

Huff Probabilistic Model (1964)

  1. Compute total expenditure from origin i: B_i .
  2. Probability consumer i patronises store j: P{ij} = \frac{Sj^a / T{ij}^b}{\sum{j}(Sj^a / T{ij}^b)}
    • Sj = size (m²), T{ij} = travel time, a & b = category-specific elasticities.
  3. Expected revenue from i to j:
    E{ij} = P{ij} B_i
  • Strengths: incorporates size, travel time, competing sites; flexible by category.
  • Weaknesses: data intensive (expenditures, travel matrix, mall sizes) & requires econometric estimation of a,b.

Geometrical / GIS Approaches

  • Voronoi (Thiessen) diagrams: partition space so every point is assigned to nearest city/site.
  • Use as preliminary delimitation of catchment areas.
  • Sophisticated GIS can weight by travel time, account for barriers (mountains, rivers), road speeds.

Industrial Space Demand

  • Categories:
    • Production/factory floors.
    • Warehousing & logistics.
    • R&D labs / testing grounds.
  • Key locational factors
    • Proximity to raw materials or suppliers (agro, heavy manufacturing).
    • Transport access (rail, highway, ports, airports).
    • Parcel size adequacy + expansion options.
    • Building suitability (floor load, ceiling height, utilities, wiring, HVAC, specialised installations).
    • Utility reliability (water/electric, broadband).
    • Energy prices, labour availability, environmental regulations.
  • Cyclicality: least volatile RE class because
    • Space usually owner-occupied; involved capital expenditures are sunk & long-term financed.
    • Logistics networks and complementary public/private investments create high relocation cost.
  • Macro drivers & complications
    • Manufacturing employment is an imperfect proxy due to technological change (automation raises output, may cut labour whilst still needing space or different quality of space).
    • Must consider GDP growth, product-specific demand, existing vacancy, energy prices.

Case Illustration – Werra Papier (Sofidel) near Schmalkalden

  1. Riverside siting → abundant process water for paper making.
  2. Local forest resources → cellulose supply.
  3. Central Germany + highway access → within 300-400 km of dense consumer bases; reduces bulky finished-goods logistics cost.
  4. Ample land enabled successive plant expansions.

Housing Demand

  • Highly heterogeneous good; quality variation even within identical floor area & neighbourhood.
  • Evaluation involves BOTH consumption & investment motives; subject to personal preferences, lifestyle, bequest motives.
  • Social & political relevance: basic necessity → objects of public policy (subsidies, rent control, social housing).

Price–Rent Relationship (simplified finance view)

  • Perpetual NPV formula:
    Pt = \frac{Rt}{c}
    where R_t = net rent (excluding maintenance) & c = discount factor (opportunity cost).
  • Practical challenges
    1. Owner-occupied units lack observable rent ⇒ need comparables; heterogeneity hampers accuracy & local supply shock if unit were rented.
    2. Substitute user-cost for R_t (interest, maintenance, depreciation, taxes, appreciation).
      • But housing markets feature monopolistic competition → market rents include mark-ups > user cost.
    3. Non-monetary utilities (security, emotional value, inflation hedge) break pure NPV logic.
  • Despite imperfections, rents and asset prices co-move because households weigh rent vs buy (often mortgage-financed).
    • Owned housing offers hedge & potential capital gain; rented offers flexibility & different risk.

Dynamic Demand Drivers

  • Income growth → ↑ demand; premium locations (CBD, waterfront) see disproportionate rent escalation.
  • Demographics
    • Population growth exceeding construction pace → tight market, ↑ prices.
    • Local migrations (centre vs periphery) create spatially divergent trends.
  • Mortgage interest rates
    • Low nominal/real rates can spur purchases, yet often coincide with recessions (counter-effects on income/unemployment).
  • Fertility / household size trends
    • Lower fertility → smaller households → demand shifts towards smaller units; influences amenities (schools vs nightlife).
  • Price expectations
    • Anticipated appreciation → higher willingness to buy; anticipated decline → shift to renting unless price drop enables previously constrained buyers.
  • Operating costs (utilities, energy) can crowd out housing budgets.
  • Commuting times/costs & technology
    • New plant opening raises nearby demand; plant closure reverses.
    • Teleworking reduces importance of commute, raises weight of environmental quality, amenities.
    • Digital substitutes (e-commerce, streaming, online courses) may lessen central-city amenity advantage, boost greener suburbs.
    • Short-term rental (Airbnb) in tourist areas raises local prices, induces resident displacement (tourist-driven gentrification).

Empirical Example – Sevilla (Spain)

  • Metro population ≈ 1.1 M.
  • Web-portal listing data (Idealista) indicate stable avg price €/m² over 5 years, concealing heterogeneity:
    • Inner city last-12-month price change ≈ +10 % (Bellavista +35 %).
      • Tourism drives conversion to short-term lets.
      • Dense bar/restaurant network attracts young residents.
    • Satellite towns: Umbrete +19.5 % (West) vs Villamanrique –20.8 % (remote SW).
      • Plausibly linked to commuting convenience to southern industrial plants (e.g.
      aviation manufacturers).

Cross-Cutting Ethical & Practical Considerations

  • Vacancy as social inefficiency (unused urban space) vs rational profit-seeking.
  • Gentrification, displacement, touristification in heritage centres raise equity issues.
  • Zoning & rent control policies influence supply elasticity, investment incentives, spatial segregation.
  • Environmental footprint: industrial siting near water sources may conflict with conservation; retail sprawl vs sustainable mobility.

Key Equations Summary (LaTeX-formatted)

  • Office demand (HLM): ln(D) = \lambda0 + \lambdaR ln(R) + \lambda_E ln(E)
  • Construction (Wheaton): Ct = \beta0 + \beta1 Rt + \beta2 Vt + \beta3 It + \beta4 RCt
  • Investment (Brunes): Pt = \alpha + \beta \, TQ{t-m}
  • Fuerst demand: OSt^* = \alpha0 + Et(\alpha1 + \phi1 \frac{Et - E{t-1}}{Et} - \phi2 R{t-1}) + Z_t
  • Simplified housing asset value: Pt = \frac{Rt}{c}
  • Converse trade gravity: \frac{TA}{TB} = \frac{PA}{PB} \left(\frac{DB}{DA}\right)^2
  • Break-point distance: DA = \frac{D}{1 + (PA/P_B)^{1/2}} (or swapped ratio variant).
  • Huff probability: P{ij} = \frac{Sj^a / T{ij}^b}{\sumj Sj^a / T{ij}^b} ; Expected revenue: E{ij} = P{ij} B_i