The Shelter (housing) category makes up for 45.2% of the whole truCPI index (as of Sept 2021) and contains the biggest subcategory vital to the index changes: Rent.
The government CPI breaks up rent into tenant rent and the owner rent. Truflation also separates those categories to further expand the index later. However, we use the same number for both of the subcategories right now, derived from the Pennsylvania State University data on rents for both tenants and owners based on the owners' rental incomes.
Apart from rents, the shelter category includes energy utilities, furniture and household items. This category within the govCPI reported by the BLS does not directly include property taxes (because those are said to be included indirectly through their market forces' influence on the final rent expressed as the owner's equivalent rent). This category also does not include mortgages and house prices because houses are considered assets by the BLS.
In future updates, Truflation CPI will add property taxes and actual mortgages data on interest rates and the principal to our index to better reflect the homeowner's cost.

Rent (31,2%)

The Truflation Oracle exchanges the CPI rent prices obtained from BLS surveys in two subcategories:
  • Rent of primary residence (tenant survey)
  • Owners' equivalent rent of residence) (landlord surveys)
with a new dataset introduced by and in collaboration with Penn State University.
The traditional rent prices used by CPI are collected rotationally, with tenants from various regions surveyed once every 6 months.
Some downsides of BLS data collection include:
  • Significant data lags (6 months to 2 years before data in the system)
  • Self-reported surveys rather than objective market information
  • A limited number of cities surveyed or included in the CPI model
  • No correction for residential unit quality
  • Omitting rental updates upon tenant turnover when most rent prices increase (BLS keeps track of the same tenants over time which gives priority to existing tenants)
  • Statistical smoothing of the data
    • CPI uses the growth of the average rent instead of the average of growth rates
    • CPI uses a 6-month average growth rate corresponding to their 6-month data collection intervals
In effect, the BLS’s rent metric doesn’t correspond to the market prices, especially in times of global crises. For example, it didn’t go down with the market rents in 2009-2010.
To create an alternative, the researchers at Penn State use a database of the Commercial Property Price Index (CPP) and adjust it for the average capitalization rates of the apartments.
This way, they obtain a Net Rent Index (NRI), which describes the net operating incomes (NOI) for apartment investors (gross income minus management expenses, maintenance, and taxes). From there, they derive a Marginal Rent Index, an average monthly rent paid by the tenants.
Net Rent Income = Property Price x Capitalization rate.
The MRI has been shown to reflect the actual rental costs when compared to historical data previously obtained by the University, which contained 1.4 million lease contracts of over half-million family residences, including lease characteristics and property locations, for the period between 1998 and 2010.

Energy Services (3.2%)

Energy services are part of the housing category in the Truflation basket and contain house electricity and utility gas services. Our index uses open-source data on the monthly bills paid by consumers, gathered by a credible source at
EIA is a US Energy Information Administration providing monthly statistics related to energy, including utility costs. EIA's reports are required by law to be objective and independent of potential policy considerations.
Surprisingly, the BLS energy data seems to use subjective surveys rather than comprehensive statistics of market prices offered by EIA. Additionally, EIA results tend to be higher than CPI reported prices of electricity and gas bills.
Last modified 8mo ago