Recent Posts

Monday, April 12, 2021

Notes On Shelter Costs

Shelter is a necessity, and so it is certainly an important component of the cost of living. However, the issue that we are concerned with is inflation, as well as the narrower issue of how to shelter should appear in a consumer price index (like the CPI).

Different countries have different norms with respect to how shelter is provided, and those norms have changed over time. My comments here reflect the norms in Canada and the United States, but they should translate to other developed countries once we account for local idiosyncrasies.

NOTE: This article is a draft section from a chapter on asset prices and inflation. It is preliminary, and needs beefing up with data. It is in a far more tentative state than usual, as I will be iterating on it once the rest if the  chapter is finished. It needs more statistical data, and I am largely working from memory. However, chasing the data will take time, and I will only do so once the text is closer to a finished state.


Rented shelter is the simplest one to deal with, at least from a high-level perspective. We just need to survey rents, and then look at how much they change over time, taking into account quality differences. Although most rental units in North America are apartments or other multi-unit dwellings, houses can also be rented.

From a theoretical perspective, rents should fit within our notion of inflation. It is a service that is typically paid monthly and re-negotiated on an annual basis, which is a fairly common situation for services. It seems entirely plausible to argue that if wages are rising as a steady pace, sooner or later, rents will eventually follow. However, I want to underline that the relationship should not expected to be perfect – the rental market is a market, which also faces regulation that depends upon the jurisdiction. (For example, Quebec has rent controls.)

The theoretical simplicity of rents has meant that they are a part of consumer price indices, and effectively stand in for all of shelter.

(Note:The comments in this paragraph need to be validated.) One of the stranger issues associated with rents is the question of utilities. In many multi-unit rentals, heating and central air conditioning expense are embedded in the rent. If the statistical agency wants to split out the energy expense from the rent, we can end up with a strange effect: higher energy prices increase the energy component of the consumer price index, but since rents are fixed for the year, the non-energy component of rent must fall. As such, higher prices in energy causes a fall in “core” consumer prices (ex-food and energy). This effect is one of the weaknesses of looking at core inflation. Nevertheless, this is purely an issue for the interpretation of the consumer price index: the headline level reflects the cost of the overall basket.

House prices are the greater concern and will be the subject of the rest of this section.

What is the Cost of Owning a House?

For an individual who buys their residence, it is easy to measure the shelter component of their cost of living: they can just add up how much they have spent. However, when we are discussing inflation, we need an economy-wide measure, not how much someone spent on something.

Since writing “residence” is somewhat stilted, I will replace “buying a residence” with “buying a house.” The rising importance of multi-unit condominiums might make that somewhat dated. However, it also fits better with how the subject is normally framed – we refer to “house price indices,” and not “residence price indices.”

One of the joys of homeownership is the need for maintenance, as well as paying for insurance and utilities. Those are already incorporated into consumer price indices. Within this area, property taxes are one component of the cost of living that is not directly part of the CPI. As discussed in {another section of the book}, the CPI drops things like taxes, since they reflect the policy structure of the economy.

(Property taxes presumably indirectly appear in the CPI via rents. From the BLA FAQ: "In addition, property taxes are indirectly reflected in the BLS method of measuring the cost of the flow of services provided by shelter, called owners' equivalent rent, to the extent that these taxes influence rental values." Thanks to Jerry Brown for highlighting this point.)

This leaves us with the controversial cost: house prices. The problem with judging the financial effects of house price changes on people is that people are different. Not everyone follows the exact same path towards home ownership.
  • The most common path is for younger people to buy a relatively small house (“starter home”) with a down payment percentage that varies across jurisdictions and over time. In the good old days in North America, a typical down payment was at least 20% of the house value – but lending standards shifted towards lower down payment percentages. The house buyer thus faced an up front cost of 20% of home value, and a mortgage for the remaining 80%. There are also many other costs associated with buying a home, but they would show up in other components of the CPI.
  • Someone who already owns a home may buy another. They are only exposed to the difference in home prices (plus the costs of the move, including broker fees). If they trade up, they need to inject money to make up the difference (either cash, or a larger mortgage), but if they trade down, they receive cash. If we are concerned about “average” behaviour, there would be a lot of cancellation between people trading up and down.
  • Some people could inherit their house. They get it for free.
  • There could be relatively rich people who buy their first house without a mortgage.
  • A relatively small proportion of the population will own second (third…) properties. In Canada, a good portion of this demographic are cottages (“chalets” in Quebec), quite often passed down the family. 

One-Time Cost Plus a Mortgage

One could imagine being concerned with the travails of a multi-millionaire who keeps buying houses and giving them away once they are bored with them. For them, a house price index would reflect the cost of living their chosen lifestyle. However, we are interested in a measure that reflects society at large, and for society at large, house purchases are debt-financed. It would be entirely typical to only make a single down payment once in their working life, and then finance any other houses bought via the equity built up in the home that they sell. 

This has the following effect: for such people, there is only a single payment in their life where their spending is directly proportional to house prices, every other payment they make is intermediated via mortgage payments. This means that that their cost of shelter is intimately tied to mortgage interest rates. The one-time nature of the payment also means that it would end up with a relatively low weighting in any notion of an aggregated cost of living.

Mortgage Payments

The structure of residential mortgages varies across countries. As an example, I will use a 30-year fixed rate mortgage, which is a common structure in the United States. This allows a much longer period of fixed interest rates than is the case of other “Anglo countries” (the United Kingdom, Canada, Australia, and New Zealand).

(Note that this numerical analysis is tentative.)

Using the weekly average fixed rate mortgage series from Freddie Mac, the 30-year conventional mortgage rate dropped from a peak of 18.44% in 1981 to 2.66% in 2020. Unsurprisingly, this makes mortgages more affordable. At the peak interest rate, a monthly payment of $1 per week allows for a mortgage of $64.81 (ignoring other fees buried into the mortgage). At the low level, the same payment allows a mortgage of $247.84, which is a multiple of about 3.8 times.

Meanwhile, $1 in 1981 has a greater value than $1 in 2020, although there was disinflation, inflation rates were still positive. This means that if the “cost of a mortgage payment” kept up with inflation, the monthly payment would be $2.85 in 2020, which means that the equivalent mortgage principal is $706, or about 10.9 times the 1981 level. (Note: I used an inflation calculator, while I should have grabbed the specific monthly index values.)

As such, the secular bull market in housing – which is normally discussed in nominal terms – is not too surprising.

Down Payments and Financial Engineering

Falling interest rates from the Volcker era peaks cushioned the impact of rising house prices on household monthly budgets. Nevertheless, households still were expected to make a down payment on the mortgage. If the terms of borrowing were unchanged, then rising house prices would directly translate into a higher down payment, making houses less affordable.

However, the financial sector is filled with highly paid people who take pride in their ability to conjure financial wizardry. The ingenious financial engineering solution to the greater difficulty of saving for a down payment was to lower lending standards. Traditional mortgages with 20% down payments are an extremely safe lending structure, and default rates on such mortgages were low. “We are being too conservative” was the assessment in the financial sector, and so they created new products to allow for a reduction in down payments. (For those readers with an interest in economic theory, this is exactly the script that Hyman Minsky described.) Meanwhile, governments love rising house prices, since a housing bull market creates jobs in construction – which is one area were college degrees generally superfluous. They also made it easier to make down payments through various housing “affordability” programmes. (These programmes have the effect of increasing the price of houses, which makes housing less affordable to people not using them.) For example, Canada lets workers dip into their registered retirement savings plans (RRSP) to help make their first down payment.

If we cut the down payment from 20% to 10%, we can double the price of the house we can buy, if we assume that the down payment and not the monthly mortgage payment is the constraint on spending. If we drop the down payment to 0% (which some more inventive bankers pioneered), down payments drop out of the picture entirely.

Cost of Living Complexity

The above considerations explain why raw house price changes do not translate one-to-one into a cost of living (or inflation) increase. We need to consider the effect of interest rates, and realistically, the historical reduction in the percentage size of the down payment. 

Such nuances are typically ignored by commentators with an ideological axe to grind, who focus solely on the changes in house prices. Their attitudes might change if and when mortgage interest rates rise, at which point they will probably incorrectly calculate the effect of higher interest rates.

Why Did House Prices Rise?

Although I make barbed editorial comments, my desire is that this book deals mainly with the facts and keep commentary to the absolute minimum. However, the subject of house price increases creates a fair amount of controversy, and so I will offer some of my views.

From my perspective, the aggregate housing market is relatively straightforward to analyse: we can count on households to spend as much on a house that their banks will lend them. (Admittedly, this has not helped my real estate forecasting ability; I have been bearish on house prices in my hometown for at least two decades, when they have course skyrocketed. Luckily, my wife had bought a house before we were a couple.) There can be regional bull and bear markets, owing to rising and falling fortunes of industries in different areas. 

What we see is that lending standards are based on payments as a percentage of household income. This percentage will vary relative to average rents, but these two costs generally do not get too far out of line with each other.

There are three factors that caused the generational bull market in housing from 1980 on.
  1. Average wages rose, generally slightly faster than the inflation rate. (This implies a rising standard of living.) If we keep the payments as a percentage of wages fixed, that means that house payments would rise faster than the inflation rate. Although quality improvements (larger houses) can absorb some of the higher payments, existing home prices are also likely to rise roughly in line with wages.
  2. However, rising standards of living and the falling relative prices of goods has meant that shelter payments are in increased share of incomes. This means that spending on shelter rises even faster than wages, which in turn rise faster than inflation.
  3. Finally, since interest rates fell, this means that house prices rose even faster than the spending on shelter.
Taken together, it should surprise nobody that house prices have risen faster than inflation measures like the CPI. More generally, market forces result in tying shelter costs to wages, and not the price of other consumer goods. In the same way that we do not expect wage rates to rise as the same rate as final consumer goods, we should not expect shelter spending costs to match the price rises of non-shelter components of the CPI, even without taking interest rate changes into account.

Although the inflation calculation is contentious, it is safe to say that many people are very mad about the rise in house prices. Although many economists cheerlead for rising house prices (particularly bank economists, who know where their bread is buttered), my view is that rising house prices damage the standard of living of citizens. Canada used to have cheap housing costs (outside of Toronto and Vancouver), and I felt the difference when I moved to southern England in the early 1990s. Those days are no more, and this has had negative effects.

The policy to deal with this problem is straightforward: tighten mortgage lending requirements. It is difficult to regulate business borrowing without instituting capital controls, but financiers are not going to “arbitrage” regulations in broad daylight for penny ante household borrowers. In principle, rolling back all the idiotic loosening of lending standards that allegedly made home ownership more “affordable” would in fact make them truly affordable. The problem – which is not easily ignored – is that drastic tightening of lending standards would torpedo the housing market, and that would cause hefty job losses in the construction sector. In the absence of some policy to absorb that blow, this would cause a repeat of the Financial Crisis downturn. (My earlier book, Modern Monetary Theory and the Recovery, discusses one policy to help stabilise the economy, the Job Guarantee.)

Raising interest rates would be another way to reduce home prices. However, the theory of interest rate policy is in disarray, and central bankers are casting about in a wide variety of directions to find a new framework for setting rates. Adding targeting house prices to the growing list of things that central banks are attempting to do just adds to the incoherence.

(c) Brian Romanchuk 2021


  1. "One of the joys of homeownership is the need for maintenance, as well as paying for insurance and utilities."

    Yes, I have always found this to be one of the more joyful experiences in life...
    And it is even more joyful when you are renting the place to others and are responsible for repairs. But then I am a carpenter who has made a living repairing houses, so maybe it evens out for me.

    So property taxes are not included in the typical CPI? That seems strange to me. As a landlord I would attest that they are typically passed along to the tenants as part of the rent they pay. If rent prices increase due to property tax increases and that counts as inflation, then why wouldn't the same tax increase for the individual homeowner count as inflation? Not that you devised the CPI methods.

    1. Thanks for spotting the rent issue. Property taxes appear indirectly in the U.S. CPI as being embedded in rents. Since it is not clear how much is being absorbed by the renter versus the landlord, they do not attempt to remove them.

      It might be that other consumer price indices are different, so I might need to revisit my wording. Some taxes do make it into some consumer price indices, with sales tax increases in Japan causing spikes in their CPI being the most well known.

    2. I personally know of at least 3 cases of people owning rental properties and increases in taxes are passed to the renters, absolutely.

    3. The Georgist Land Value Tax types disagree...

  2. Brian I totally agree that the correct way is to intervene on the banking side maybe even for CBs to set up some sophisticated, granular lending guidance in some of the most overheated markets...but, like you hinted, too many constituencies are very happy with the gravy train.
    A friend of mine was able to buy a particular property within his budget in Vancouver two years ago only because the bidding war of offers above his all failed (3 of them) the banks simply did not approve the mortgage for that property at those price.

  3. This comment has been removed by the author.

  4. You might devote a paragraph of thought to the economic choice made when the house buyer decides between new construction and old.

    It seems to me that the cost of new construction drives the price of old. Of course; location, condition, and general suitability also play a role. However, ultimately, the affordability of new construction drives the value of existing construction and the biggest driver of cost is the current value of labor measured in units of money.

    Now, it seems to me that the availability of units of money has increased, until recently, at a rate of about 4% per year. Housing prices have also increased at the rate of about 4% per year. To the extent that data can support this association, housing prices seem to reflect the relative value of money.

    1. My interest in this book is inflation and the cost of living. Discussing the relationship between new and existing homes is a tangent that would just add confusion.


Note: Posts are manually moderated, with a varying delay. Some disappear.

The comment section here is largely dead. My Substack or Twitter are better places to have a conversation.

Given that this is largely a backup way to reach me, I am going to reject posts that annoy me. Please post lengthy essays elsewhere.