SPAR or stratified model house SPAR or stratified model house price - - PowerPoint PPT Presentation

SPAR or stratified‐model house SPAR or stratified‐model house price indices p

G ð ú R Jó dó i Guðrún R. Jónsdóttir

Statistics Iceland

Ott ti Ottawa group meeting Wellington , New Zealand, 4‐6 May 2011 y

The aim of the study The aim of the study

• To make a contribution to the research in the

field by comparing a stratified model and SPAR

• Can the comparison tell us something about

the effectiveness of a simple metod like a few the effectiveness of a simple metod like a few variable stratification model house price i d ? index?

• Can a stratification model be recommended?

2 Guðrún R. Jónsdóttir Statistics Iceland, www.statice.

Background Background

• Stratified model has been used since 1992 in

the simple user cost model for housing in the Icelandic CPI

• Since 2010 Iceland is participating in the
• Since 2010 Iceland is participating in the

fourth pilot project on OOH led by Eurostat

• SPAR one of the methods under scrutiny as a

candidate for RPPI to be used for the OOH‐ component

3 Guðrún R. Jónsdóttir Statistics Iceland, www.statice.

Present method stratified model Present method‐ stratified model

P i h l l t d f 26 t t

• Price changes calculated for 26 strata

– Three geographic areas

• Inner capital area
• Outer capital area
• Outside capital area

– Two types of properties

• Detached and semi‐detached houses
• Flats

– Size categories

• 4 – 5 categories for each area and type of property

Guðrún R. Jónsdóttir Statistics Iceland, www.statice. 4

Present method calculation Present method‐ calculation

• Measures changes in price per m2 Weighted with
• Measures changes in price per m2. Weighted with

total value of sales in the stratum the last 36 Months Months.

• Calculation:

– Three month’s data used – moving average Three month s data used moving average – Geometric mean within stratum – Fisher index

• Quality adjustment with few variables
• Problematic when the market is in downturn with

Problematic when the market is in downturn with low sales volumes → missing data in some strata

5 Guðrún R. Jónsdóttir Statistics Iceland, www.statice.

SPAR SPAR

• SPAR: Sale Price Appraisal Ratio
• Assumptions:

Assumptions:

– Information on appraisals and prices available A i l t bi d/ t diff tl bi d – Appraisals not biased/not differently biased

• First possible to use in Iceland in 2010 when

valuation methods greatly improved

• Has been used in New Zealand Denmark
• Has been used in New‐Zealand, Denmark,

Sweden and the Netherlands

6 Guðrún R. Jónsdóttir Statistics Iceland, www.statice.

The appraisals The appraisals

• Valuation made by the Property registry
• Based on market prices

Based on market prices

• Regression model where between 11 and 22

i bl h t i ti f th h i variables on characteristic of the housing are used

• The appraisal for year 2011 uses data from the

period 2004 to April 2010 period 2004 to April 2010

– Nearly 47 thousand sales contracts

7 Guðrún R. Jónsdóttir Statistics Iceland, www.statice.

SPAR SPAR

i 1 h i b

• i=1,...,n, where n is number
• f contracts in Month t
• j= 1

m where m is j= 1,...,m, where m is number of contracts in Month 0

• ai and aj are the properties

appraisals (2011, at February 2010 prices) February 2010 prices)

• The index for each region is

calculated and the results weighted with sales volumes

8 Guðrún R. Jónsdóttir Statistics Iceland, www.statice.

Further on the SPAR calculation Further on the SPAR calculation

D t f th P t i t

• Data from the Property registry

– Not fully comparable with CPI data → For the comparison exactly the same data are used – 25.980 sales from January 2006 to March 2010

• Conditions

SPAR between 0 5 and 2 0 Same size categories as in the – SPAR between 0,5 and 2,0 Same size categories as in the stratified model

• SPAR calculation

– Two regions: The Capital area and outside the Capital area – Present value of the sales contracts compared with the appraisals pp – Geometric mean of the ratios for each category using three Month’s data – The categories are then weighted using the value share of sales – The categories are then weighted using the value share of sales in the last twelve months

9 Guðrún R. Jónsdóttir Statistics Iceland, www.statice.

Monthly changes Monthly changes

Guðrún R. Jónsdóttir Statistics Iceland, www.statice. 10

Comparison of the indices Comparison of the indices

Guðrún R. Jónsdóttir Statistics Iceland, www.statice. 11

The advantages of SPAR The advantages of SPAR

• Simple to produce
• Less (no?) stratification needed

Less (no?) stratification needed

• If the appraisals are quality adjusted by using

h d i th i d h i b ilt lit hedonics the index has an inbuilt quality adjustment

• Easier to find outliers
• Effective use of resources Important for small
• Effective use of resources. Important for small

countries.

12 Guðrún R. Jónsdóttir Statistics Iceland, www.statice.

Conclusions Conclusions

Alth h th t tifi d d l l th

• Although the stratified model only uses three

determining variables the comparison shows h h l i f i l ll h i that they explain fairly well the price development in the housing market.

• The deviation in some points can be explained

by quality issues that the SPAR better address. y q y

• Stratified model can be recommended where

hedonics are not used in valuation process for hedonics are not used in valuation process for housing.

Guðrún R. Jónsdóttir Statistics Iceland, www.statice. 13

More conclusions More conclusions

• In small markets many methods must be

excluded, e.g. repeated sales method and rental equivalence.

• With small resources in the Statistical office
• With small resources in the Statistical office

and lack of data own research on hedonics is li i unrealistic

• Take advance of the work already undertaken

y by the Property registry

Guðrún R. Jónsdóttir Statistics Iceland, www.statice. 14