|Unit||Index Jun 2005=100, NSA|
|Adjustments||Not Seasonally Adjusted|
|Release||Teranet - National Bank National Composite House Price Index|
|Housing Starts||Oct 2019||201.97||221.14||Ths. #, SAAR||Monthly|
|Building Completions||2019 Q3||51,865||48,526||#, NSA||Quarterly|
|Building Permits||Sep 2019||8,339,580||8,920,465||Ths. CAD, SA||Monthly|
|House Price Index||Sep 2019||103.08||102.86||Index Dec2016=100, SA||Monthly|
|House Price Index for New Homes||Sep 2019||103.2||103||Index Dec2016=100, NSA||Monthly|
|House Price Value for Existing Homes||Sep 2019||227.72||227.51||Index Jun 2005=100, NSA||Monthly|
|Residential Building Permits||Sep 2019||19,087||20,389||#, SA||Monthly|
The national bank house price index is an estimated index that measeures the increase/decrease in the value of properties over time by tracking home sale prices for properties that have been sold twice, allowing the source to calculate the increase or decrease in the value of the home over a given time period.
The source trackes the registered home sale prices over time, looking for properties that have been sold at least twice, making it possible to see the increase or decrease in the value or the property over a period of time.
A constant level quality of each property is assumed, however some properties do not meet requirments for this assumption and are therefore removed from the survey. Factors that could cause a property to be removed from the sample include a non-arms-length-sale, renovations that occured between salles that changed the value of the property, data error, and high turnover frequency.
The source then uses a linear regression algorithm to estimate the index using the qualifying properties.
The source also uses a process to weight poperties differently in the index. This weighting process uses geographical area of interest, frequency of the percent change of the property in the set, and time interval between sales.
It is noted that the above weighting schemes may not be exactly representative of the reality because there are many unknown factors.
The repeat sales index construction is based on a simple linear regression model, whose regression coefficient is the reciprocal of the desired index.
The source uses three different estimators to calculate all pre base period indices simulataneously. These estimators include least squares (LS), instrumental variables (IV), and generalized method of moments (GGM).
The source then refines the initial weights and downweights data flagged as an outlier.
Data is subject to revision.