Creating a Residential Real-Estate Timing Model

In a recent thread discussing macroeconomic data pointing to increasing likelihood of a hard-landing for real-estate in the US, we began brainstorming what type of timing model could be created given current technology and technique.  In this thread let's move the concept forward by more formally documenting some of our ideas, testing those ideas for validity, and perhaps even preparing a simplified simulation.

I've also started a thread on Patrick.net soliciting more general views and opinions on market timing (from a "when to buy" perspective).

The consensus in our past discussion was that psychology is a major factor propping and fueling the RE housing bubble (HB).  I contend that a lot of this psychology is well captured by Richard Thaler's "Mental Accounting Matters".  The real question is whether we can extract useful information about how the masses behave in aggregate so that we can detect signals caused before (or just as) popular sentiment shifts.  RE transactions are slow, prices are sticky, and HB's pop in slow motion compared to highly liquid markets like stocks.  So, it should therefore be theoretically possible to determine and isolate market timing signals which are useful for forward decision making.

After we flush the model out some more I intend to document everything with some use-cases and a couple basic UML models.  But for now, some suggestions from our past discussion that I've organized (thanks to all comments from contributors):

• Maybe we can take a hint from the guys who simulate the response to bird flu pandemics using monte-carlo methods:  From Wired Magazine.
• All conceptual actors participating in a buy-sell model:

1. RE selling agent
2. MLS service, newspaper
3. RE buying agent
4. owner buyer, investor buyer
5. owner seller, builder seller, investor seller
6. mortgage broker
7. mortgage bank

• Assets:

1. cash
2. (real) property
3. mortgages (owned by bank)

• Logical (simplified) Actors:

1. Agents
2. Buyers
3. Sellers
4. Appraisers
5. Brokers
6. Banks

• Asset Class Usage:

1. Asset class for everyone: Cash
2. Asset class for Buyers/Sellers: Property
3. Asset class for Banks: Mortgages
4. Asset classes are swapped using actors

• Factors:

1. Spatial/Geographic
2. Interest Rates
3. Wage Growth

Algorithms and Pseudo-code:

Buyers:

Income:  monthly cash flow
Cash in bank: 
Time Window: 

Bid Price:  probability buyer will buy increases as offered price/intrinsic value decreases

Algorithm:

Maximize house value for dollar paid over all houses(ie, spend all money) within time window else buyer disapears (ie he rents, but were not modeling rent right now)

Sellers:

   Income: monthly cash flow
   Mortgage:  debt/interest rate
   Cash: 
   House:  (with intrinsic value)

• Previous purchase price and date of house

   Ask price:  (only cares about previous house price, not intrinsic value)
   Time Window:

Generalized Algorithm:

Maximize profit (offered price - purchase price) until Time Window expires, then lower offered price by random value or until Bankruptcy

Banks:

FED Interest rate: 
Reserve Rate: 
Algorithm: 

    Analyze Buyers Income to determine
Mortgage size and intrest rates by simple formulas:  Maximum debt/income ratio = 33%, minimum rate = FED Funds rate+1%
maximum rate = FED Funds rate+3%

Have a function that maps debt/income in a linear fashion from min-max from debt/income 15%->33%

Bankruptcy Algorithm:

Minimize loss, ie maximize (offered price - loan amount) within sell Time Window (short for a bank), if expire, then lower offered price -20%. If zero, exclude from future iterations.

General:

Initial distributions will be simplified standard probability distributions and assume no reflectivity. 

• Population of sellers (MLS statistics, income statistics, etc),
• Population of Buyers
• Macro Simulation ideas (Fewlesh)

References:

• perturbation theory (or sensitivity analysis)

 

Comments

[Fewlesh]

Broken link for the Population of buyers link.

I am being pulled by external commitments (recent Vegas trips aren't helping), but I'm defintely thinking of this in the background.

I like UML, but I'm wondering if anyone in the bloggosphere hasn't seen this design pattern before.

Fewlesh.

[Fewlesh]

Cathing up on the patrick.net thread I've come to a thought:

The intrinsic value variable is closely linked to the rent price. It seems the majority of the thread is using the rental price as the determing factor in the rent/own decision.

Therefore, we can use the monthly rent statistics to determine the "intrinsic value" for a housing region.

Much like P/E ratio for stocks, Monthly housing buy cost/rent ratio is a key factor in a bear market.

Ie, you get a massive cascade of investment after the mortgage paymet+taxes < rent: ie guaranteed profits, and the loss of psycology determing house prices.

Fewlesh.

[randolfe]

Links fixed.

I agree on the rent-to-mortgage ratio driving buyer psychology and determination of "intrinsic value". I assert that the basic invariant is:

PITI + infl[e] > Rent + infl[e]

therefore:

PITI > Rent

A theoretical equality for returns should be:

PITI + infl[e] + popg + prem =
Rent + infl[e] + m + e

where:

infl[e] is expected inflation

popg is popluation growth for the region

prem is local comparative premium

m is market index returns based on historicals

e is error;

Finally, the buyer psychology will be an invariant based on the basic invariant such that the buyer _perceives_ greater value from buying:

PITI + infl[e] - hob - r[e] <
Rent + infl[e] - reb

With:

hob: home owner's benefits

r[e]: expected appreciation of property in excess of theoretical returns (that is, the "I've timed it right" return)

reb: renter's benefits