Predicting future (home) prices

[Stumped]

Office 2007 / Windows 7
------------------------

I am curious to know if future home prices can be predicted with reasonable accuracy by Excel. Consider these prices of a hypothetical house:
330000
280000
248000
196000
190000

I know I can input them in a column and then just select them all and drag down one more row & Excel will put in a value in that cell, but:
1) How is Excel calculating this value;
2) Is it accurate?
3) Are there well known & established methods I can use to get a reliable result?

[sdckapr]

1) it is doing a linear regression of the 5 (Y)values with X presumed to be 1-5, to get a slope and intercept. It then predicts the 6th (and future values from this).

2) It will depend on if the trend matches that equation. What is the value of the next point in your hypothetical? Is it close enough if not, then no it is not accurate.

As an exercise, you can use the first 4 to predict the 5th and compare it the actual 5th. The predicted value is 155,000, 18% lower than the actual. Is that close enough accuracy for you? Your data suggest non-linear at the end with a change in formula or that the equation is not the simple line.

3) There are many different types of strategies. [Google can be your friend here, it is not an excel question, though excel can be used once you have a strategy chosen, it can do more than simple linear modeling]. The problem in exptrapolating is that you move outside the area where a "model" was based. Experimenter George Box used to say that no models are accurate though some are useful...

Steve

[Stumped]

>>2) It will depend on if the trend matches that equation. What is the value of the next point in your hypothetical? <<
That is what I am trying to calculate.

>>As an exercise, you can use the first 4 to predict the 5th and compare it the actual 5th. The predicted value is 155,000, 18% lower than the actual. Is that close enough accuracy for you? <<
No. I would like the results to be within 5% of actual.

>>3) There are many different types of strategies. [Google can be your friend here, it is not an excel question, though excel can be used once you have a strategy chosen, it can do more than simple linear modeling]. <<
What phrase do you suggest I search for?

[tfspry]

What phrase do you suggest I search for?

Home price volatility can be affected by so many factors like location, location, location!!

If you want to play in Excel, within Excel Help try a search for "trend" which will lead to many analysis functions. Be sure to also install the "Analysis ToolPak add-in".

For a Google search try something like "Case–Shiller". There are so many factors that go into home pricing.

HTH

PS: Omit the quotes "" in all the searches.

[sdckapr]

>>2) It will depend on if the trend matches that equation. What is the value of the next point in your hypothetical? <<
That is what I am trying to calculate.

It was your hypothetical, if you don't know the value then who does? You need to actual to test the accuracy of the prediction...

>>As an exercise, you can use the first 4 to predict the 5th and compare it the actual 5th. The predicted value is 155,000, 18% lower than the actual. Is that close enough accuracy for you? <<
No. I would like the results to be within 5% of actual.

Then you need a different model than a linear one based on just the year. That means you probably need more variables than just the past prices and the year.

>>3) There are many different types of strategies. [Google can be your friend here, it is not an excel question, though excel can be used once you have a strategy chosen, it can do more than simple linear modeling]. <<
What phrase do you suggest I search for?

In addition to "Case-shiller" Try "forecasting home prices".

Steve

[BATcher]

I am curious to know if future home prices can be predicted with reasonable accuracy by...

...any method?

Answer: No.

If the answer was yes, then someone would make a fortune!

[Stumped]

I didn't think forecasting home prices would be easy, and from the responses clearly it isn't. BATcher sums it up nicely: If the answer was yes, then someone would make a fortune!

Thanks for all the replies.

[mediamavin]

Using simple regression to explain a complex economic question is iike "drawing a mathmaticly precise line from a unwarranted assumption to a forgone conclusion!".
you would need some/several leading economic indices to use in a multiple regression analysis to construct a model that MIGHT predict the future.
My suggestion is to forget it!

[edmcguirk]

The near future prices of houses are not really a mathematical progression. It's really more of an emotional progression, probably for the next 5 to 10 years.

The last 10 or so years of house prices were part of a speculative bubble so the prices are higher than they should be. What's happening now is not a crash in house prices, it is a return to normal. Unfortunately the prices will probably overshoot low until people calm down.

Remember, houses have no intrinsic value. They are only worth what people are willing and able to pay.

Let's look at where house prices will eventually settle. A normal safe loan really needs to start with 20% down payment and a total that is fairly close to triple your yearly salary. If you get a loan that is higher than that, the bank is really betting that it can make more money in fees or foreclosure than a normal payback of the loan. (yes, the banks got greedy and lost that bet, but that is where normal banking will eventually return to)

It's not an exact equation but it generally means that if the average neighborhood family salary is around 50,000 than the average house will be somewhere around 200,000. And if the average family income is around 100,000 the average home should settle around 375,000.

Some neighborhoods are already below that while some are still above that.

Throw into that, during the bubble contractors decided that they would make more money if they built more expensive houses. Now there is a glut of luxury homes that are really more valuable than people can afford to pay. That pushes prices down.

Interest rates also sway the numbers. If the rate goes down, people can afford a few more dollars so the value of the houses will get pushed up. Be warned that when those interest rates go back up to normal, the value of the houses will fall to what people will be able to pay at the new rates.

Add into that the confidence level of the potential buyers. Do they really feel that they will have a job one or two years from now?

I'm sure there are a few more factors I haven't even thought of but you can see how complicated forecasting house values can be.

Personally I would keep that number 3 to 4 times average salary in mind and compare that number to the local emotional background to decide if I wanted to buy anything.

[macropod]

I guess it all depends on what you mean by 'reasonable accuracy'. It's a bit like forecasting global warming. Come to think of it, no-one knows whther any of the forecasting models is any good, but there's plenty of people making money out of it. Maybe you should turn your attention to that instead of house prices

[Stumped]

Thanks for those good comments, edmcguirk. I was actually looking to buy a house and the prices I posted in my first post are close to the appraised value of a house I was looking at. I was just trying to figure out how much my own offer should be.
-----------------------
Out of curiosity, let me ask another question: If I were to have cash and don't need mortgage for the house, is it better to:
- buy a house
OR
- invest the money (and where?) and rent in the meantime!

[tfspry]

....... the prices I posted in my first post are close to the appraised value of a house I was looking at. ......
-----------------------
...... is it better to:
- buy a house
OR
- invest the money (and where?) and rent in the meantime!

What is meant by "the prices I posted in my first post are close to the appraised value of a house I was looking at"? The 5 prices in Post #1 range from a low of $190,000 to a high of $330,000. Are those 5 different recent appraisals for the same property?

The Rent vs. Buy decision also has many factors to be considered. You might want to setup an Excel analysis for such a comparison. Some "What Ifs" tables may be helpful,


HTH