|by Abbe Will|
The weighted average of the LIRA inputs produces the LIRA estimates and projections as seen in Figure 2A (for modeling improvements spending trends) and Figure 2B (for modeling maintenance and repair spending trends) compared to the now updated AHS-based benchmark data series for 1994-2015. The improvements LIRA continued to track the reference series very closely in 2014 and 2015. The estimates produced by the improvements LIRA model and the AHS-based benchmark now have a correlation coefficient of 0.84 (p-value of 0.00). And a simple regression of the LIRA output on the benchmark spending series results in an R-squared value of 0.6759, which suggests that upwards of 70% of the variation, or movement, in the improvements spending benchmark can be explained by the LIRA model.
Similarly, Figure 2B compares the weighted average output of the maintenance and repair LIRA model to its AHS-based reference series. The maintenance LIRA has also tracked its benchmark fairly well since 2013. The maintenance and repair LIRA and its reference series have a correlation coefficient of 0.73 (p-value of 0.00) and a simple regression of the LIRA output on the benchmark results in an R-squared value of 0.5362, which suggests that about 54% of the movement in the home maintenance and repair spending benchmark can be explained by this LIRA model.
Sources: JCHS calculations using HUD, American Housing Surveys; Department of Commerce, Retail Sales of Building Materials; and US Census Bureau, Survey of Residential Alterations and Repairs (C-50); Leading Indicator of Remodeling Activity.
Last spring, the LIRA was re-benchmarked to a measure of home improvement and repair spending based on estimates from HUD’s biennial American Housing Survey, and at that time, historical remodeling and repair data from the AHS was available for 1994–2013. Until the 2015 AHS data became available, the LIRA model was used to estimate historical improvement and repair spending levels since 2013. Once every two years, with new historical AHS data, the LIRA benchmark series will be updated. With the January 2017 release, the LIRA model will be used to estimate historical spending levels since 2015 until the next biennial release of the American Housing Survey allows for actual 2016 and 2017 spending data to replace modeled estimates.