House size doubles?

by John Q on May 14, 2008

It’s regularly stated that the size of the average (new) American home has doubled since 1950, and implied that this increase has continued fairly steadily over the intervening decades. This seems a bit surprising given that (on standard measures) real wages for large groups of workers have not increased since the 1970s. Some, but not all, of the story can be explained by the fact that a fixed stock like housing takes a long time to adjust and so would continue to improve in response to the big increase in both income and equality that took place from the 40s to the 70s. And of course, the top quintile of the income distribution is doing well, and they have a big influence on the market.

But that still leaves a puzzle I think. Here’s one little piece. As far as I can tell, the statistical basis for the statement comes from the National Association of Home Builders and compares the houses being built today with those built by Levitt and others in the 1950s. But not everyone lives in houses. To get the full story you’d need to take account of apartments (I haven’t looked at this).

Even more important, though, are manufactured homes (aka trailers). There are about 8 million of these up from essentially zero in the 1950s. They constitute about 8 per cent of the housing stock now (housing around 19 million people), which must imply a substantially larger proportion of homes built each year. Although they can be quite large, most are a lot smaller than the average home built on-site. Taking manufactured homes into account would substantially lower the size of the average new home. It would also fit the income data, showing rising inequality, a lot better.



Joshua W. Burton 05.14.08 at 9:05 pm

Also, and fairly obviously, not everybody gets to live in a new house. Even when we were building over two million new houses a year, that was under 2% of 125 million existing houses.

So a decade of new houses all double the size of existing houses would only increase the size of the average house by 15% or so — and would not increase the size of the median house, in which the median family is likely to live, at all.


Tim Worstall 05.14.08 at 9:18 pm

Sounds like one of those interesting questions that an academic economist could usefully explore.

What, exactly, has been the expansion (or not) of the average living space of the average (I guess we’d need to say median, right?) household in the US?

Tap into the house, apartment, mobile home information, come up with a number. Correct by the change in household size and we would see something about:

“This seems a bit surprising given that (on standard measures) real wages for large groups of workers have not increased since the 1970s.”

We would find out something about whether real wages, as measured by consumption (of housing, to be sure), had in fact gone up or down, would we not?

Hmm, anyone know an academic economist interested in this question?


matt 05.14.08 at 9:25 pm

John- are you considering real wages per capita or per house-hold? My understanding is that much of the increase in consumer goods since the 70’s has come from two-worker families. If that’s right I’d expect it to apply to homes, too- so, real wages per worker have stagnated but per family have gone up since more women work and it’s this extra work that’s buying stuff. At least that’s the store I’ve heard. Is it wrong?


abb1 05.14.08 at 9:28 pm

What Matt said, obviously. Families live in houses, incomes of families went up, problem solved.


John Quiggin 05.14.08 at 9:31 pm

I’m picking up the sound of snark here, Tim, so please forgive me if I respond in kind.

You may have heard of this cool new thing the kids are into called “blogging” (honestly, it’s like they speak a foreign language!). Anyway, it’s a new Intertubes thingy where, for example, an academic economist can raise a point they don’t have time to explore in detail, and find out if anyone else has worked on this, or is interested in doing so. You can find an example here.


Chris 05.14.08 at 9:32 pm

How do apartments impact average house size? Is it that people with lower incomes don’t have to purchase small housesa, thereby increasinf the average size?


John Quiggin 05.14.08 at 9:37 pm

Matt & Abb1, increased workforce participation of women is certainly part of the story, but not all of it. Again, no time to spell this out in a blog comment, but I’m confident the increase from this source is not enough by itself to explain various puzzles in the data.


Slocum 05.14.08 at 9:43 pm

Taking manufactured homes into account would substantially lower the size of the average new home. It would also fit the income data, showing rising inequality, a lot better.

Perhaps. But, the manufactured house my sister lives is about the same size as the family home we lived in as kids (about 1200 square feet). Hers has two bathrooms compared to our one and central A/C (which we didn’t have for the first 10 years we lived there). There were six of us in that house, while my sister lives alone. And my father had a higher status job than my sister does.

In general, I think you’ll find that the size of manufactured homes in the U.S. has ballooned, too. And keep in mind that the average number of persons in a household has dropped significantly, so the amount of living space per person has increased much faster than the average house size. Even for people living in unexpanded older homes, the decline in family sizes has translated into much more living space, on average (what used to be a family home is now considered appropriate for a single or a couple).

I really don’t think you’re going to be able to explain away this affluence.


joel hanes 05.14.08 at 9:51 pm

I really really want the median and standard deviation figures to help interpret this “average” (usually mean) figure.

I’d guess that much of the effect is produced by the recent skew toward enormous showplace homes at the high end, and that the median has moved much less.

You know: Bill Gates walks into a room, and the average income of people in that room jumps into the stratosphere ….


Slocum 05.14.08 at 9:53 pm

With respect to manufactured homes, there’s this:

Manufactured Housing: Growing in Size and Popularity

According to the Manufactured Housing Institute’s analysis of U.S. Commerce Department and Census Bureau data, the average square footage of a new manufactured home increased from 1,335 in 1996 to 1,595 in 2002. The average size of a conventional home increased from 2,090 square feet to 2,301 square feet over the same period.

Many customers are now purchasing multisection homes — known in the popular parlance as “doublewides” — or even two-story homes that are erected on site and sell for $85,000 and up, according to Savage.

It appears that manufactured homes have expanded size slightly faster on a percentage basis than conventional homes.


Brett Bellmore 05.14.08 at 9:54 pm

I live in a home approximately 2-3 times the size of the one I grew up in, and I know exactly why: The damned building code here doesn’t permit you to build small houses! Otherwise I’d have built much smaller, and better quality.

You might consider the possibility that revenue hungry local governments are using zoning and building codes to force people to build big, or not at all, in order to boost property taxes.


John Quiggin 05.14.08 at 10:02 pm

Slocum, I don’t need to “explain it away” in the sense of showing there has been no increase in the consumption per person of the median household.

The combination of increased LF participation of women, longer working hours, and increased household debt implies an increase in consumption even with stable real wages.

The problem is to reconcile the observed data with anecdotal evidence (like a doubling of house size) that would appear inconsistent with the fact that real wages haven’t changed, even taking these factors into account. As you’d be aware, a lot of people have pointed to this kind of anecdotal stuff to “prove” that the wage data must be wrong. In this case, once you take account of manufactured homes, along with the other points already mentioned, this counterclaim looks a lot weaker.


F. Blair 05.14.08 at 10:04 pm

I don’t think the mobile-homes explanation holds water. There are actually fewer mobile homes sold nowadays than there were twenty years ago — in 1987, something like 300,000 were sold (, while these days it’s closer to 100,000, which is what sales were back in the 1960s.

There’s no way that has any meaningful impact on average house size in the U.S., without even mentioning what everyone else has pointed out, which is that mobile homes have been getting bigger, too.


John Quiggin 05.14.08 at 10:05 pm

On #9, there’s a fallacy of composition here. If the share of manufactured homes is increasing, as claimed in the headline, the average growth rate can be lower than the rate for either type of home taken separately.


lemuel pitkin 05.14.08 at 10:22 pm

Just for the record, household and family median income, and not just wages, have been fairly stagnant since the 1970s. For both, the total increase between 1979 and 2006 was just over 10 percent. (The entirety of which, by the way, came in the second half of the 1990s.) This was a drastic deceleration compared with the quarter century after World War II, during which family income doubled.

So for those who think the entry of women in the labor force has made up for the decline in wages, the answer is, sadly no. (And abb1, dude, doesn’t it occur to you to spend 30 seconds checking something before making a confident — and in this case completely wrong — declaration of fact?)


lemuel pitkin 05.14.08 at 10:24 pm


lemuel pitkin 05.14.08 at 10:31 pm

Also, the census has numbers on average and median home size here. They seem to back up the NAHB survey.

John, why don’t you find it plausible that suburbanization has reduced the cost of large houses, and/or that people are spending a larger share of their income on housing?


lemuel pitkin 05.14.08 at 10:37 pm

Here’s another table, on the numebr of housing starts. One thing that’s clear is that the proportion of single-family homes has risen sharply — from about 60 percent in 1968 (where the table starts) to 80 percent today. So unless new apartments are shrinking awfully fast, the average size of houses+apartments is growing even faster than the average size of houses.


Jake 05.14.08 at 10:38 pm

I’m having trouble understanding what exactly you are trying to say here. Do you think that the size of the average American house has not in fact doubled? Do you wonder how the size of the average American house could have doubled given flat-ish real wages?

Even a little bit of thought as to how house sizes could double with real wages staying constant would yield a bunch of plausible explanations. The two most obvious questions to ask are “how does the conversion from nominal to real wages affect this?” and “has the price per square foot of the average house changed?”

I think the answer to both questions will be “yes.”


Slocum 05.14.08 at 10:44 pm

In this case, once you take account of manufactured homes, along with the other points already mentioned, this counterclaim looks a lot weaker.

Mobile homes growing in size faster than conventional homes would tend to make the claim stronger. Mobile homes making up an increasing percentage would tend to weaken the claim. Taking the two together, it’s probably a wash or at most a very small effect (after all, we’re talking about 8 percent of the market).

And you’re neglecting the steady reduction in persons per household. Even if house sizes had remained constant (which, of course, they have not), the housing space per person would have increased 30% just based on the decline in household size (from 3.4 to 2.6) between 1950 and 2000.

Lastly, this is one of those areas where consumption inequality (in terms of utility, not dollars) has decreased, regardless of what has happened with income inequality. The average family that earns $500K a year certainly doesn’t live in a house that’s 10 times larger than the family making $50K. The increase in house size is obviously not linear with income.

And in the last decades, the houses of people of the middle class have been equipped with all of the features that were formerly luxury items (washers, dryers, microwaves, central A/C, video systems, cable-TV, game consoles, computers, etc) whereas no significant new types have been invented that remain the exclusive province of the rich (unless, perhaps, you count ‘whirlpool tubs’).

If you’re looking for a place to confirm rising inequality, I suspect you’ll be disappointed looking at housing.


lemuel pitkin 05.14.08 at 10:49 pm

… aaaand one more table.

Between 1984 and 2006, the average share of household expenditures on “shelter” (inclduing mortgages, property taxes, maintenance of own home, and rent) increased from 15.9 percent to 20.0 percent.

So there you go. It’s true that houses are getting bigger. And it’s true that incomes are flat. The explanation is simply that people are spending more of their incomes, on housing.


ScentOfViolets 05.14.08 at 10:50 pm

Keep in mind that doubling over a period of fifty years is really a quite modest annual qrowth – less than 1.4%/year, if I can trust my botec. So it would not be surprising to see people purchasing larger houses as long as their real income was keeping pace.


flubber 05.14.08 at 10:57 pm

What lemuel said: “household and family median income, and not just wages, have been fairly stagnant since the 1970s”

Increasing labor force participation hasn’t done much at all for the median family.

The past 20 years has seen the rise of the exurbs. Houses are built on super cheap land – former farmland. Even if “house size” graphed linearly with cost in established cities, probably not in the burbs.

I’d also guess that, ignoring land, manufacturing a house has a decreasing marginal cost curve, as size increases. (Increasing at a decreasing rate).

And the cost of financing has decreased with financial innovation (securitization of loans) and historically low interest rates (since the early 1980s). People can afford more for a given income.


SG 05.14.08 at 11:06 pm

Western housing patterns haven’t kept up with western living patterns. Explain why the average size of the western home keeps increasing but young people keep paying increasingly large rents for rooms in share houses, while their parents upsize to increasingly large homes with no-one in them. Meanwhile in Japan single young people can live in dirt cheap studios. Western housing markets are, in general, a complete mess, as are western planning rules. In combination with the so-called housing “boom”, the result is an increasingly stupid and complicated housing market, where everyone pays too much for something they don’t need.


Matt 05.14.08 at 11:23 pm

Thanks for the data, Lemuel.


Brad 05.14.08 at 11:29 pm

What about the productivity increases of construction in that period? Even if construction workers they haven’t – perhaps especially if they haven’t – received increased wages as a result, more productivity means more home for the money. Standardized home plans, more pre-fab, mechanization, better tools (pneumatic nail guns anyone?), cheaper, better materials (plastic pipe) etc. all have made bigger homes cheaper.


sara 05.15.08 at 12:37 am

New McMansions are often remarkably flimsy compared with their early twentieth-century counterparts.

This effect is most dramatically viewed in upscale beachfront communities where warm weather makes insulation unnecessary and the possibility of hurricane promotes cheap building techniques.

Do you want a walk-in closet and dressing room or do you want to be able to open and close the bedroom door in its frame and not feel the entire wall shake?


Justin 05.15.08 at 12:42 am

I feel as if it’s conventional wisdom that one reason why ‘mcmansions’ have become so common in the States is that material costs have been decreasing, and this in turn pushes down revenues and profits for anything other than really large houses.

That said, I looked where I thought I remembered seeing that claim, and didn’t find it. So conventional wisdom-eh….


roger 05.15.08 at 12:54 am

There are all kinds of amazing statistics churned out by government economists that are much like the bigger houses one. For instance, this bit about the latest soviet style information from the Bureau of Labor today in the NYT brought a smile to my lips:

Transportation costs dipped by 0.7 percent in “April, and recreation costs dropped by 0.1 percent.

A primary factor holding down the inflation rate was a surprising dip in gasoline prices — at least in the way the government tracks the data. The Labor Department calculated that gasoline prices actually increased 5.6 percent in April, but when seasonal fluctuations for demand were factored in, that increase was penciled in as a 0.2 percent dip.”

As anybody buying an airplane ticket knows, it is an absolute illusion that ticket prices are going up. One simply has to apply the Soviet People’s of America rule, which says that all price rises are solely in order to honor our boy king for making the world such a progressive and People’s choice place to travel. And because, hedonically, those places, like Las Vegas, look so much better than they did in the 60s, travel prices are going down all the time.

But the best is the dip in gasoline prices. That was special. I think such exercises in fiction are deserving of some prize. Maybe the pulitzers could start a special category: best prize in statistical fiction. This year shared by the U.S. military (about the weapons flow into Iraq from Iran) and the Labor and Commerce department.


grackle 05.15.08 at 1:42 am

A little anecdotal info from 30+ years in construction. In the early seventies, IIRC, new houses were around 12-1500 sq. ft. in places I worked, NY and northern CA, with many on the low end of those sizes. Over the next 5-10 years, during periods not in recession, it became difficult for small builders and for homeowners to finance such “small” houses. The sizes grew to 2000 plus, with lenders demanding at least two full baths or two and a half baths. As the pattern was set, every few years the current size-on-average became a defacto minimum with the size tied by lenders to their sense of what would constitute safe loans. A kind of economic red-lining.


Cranky Observer 05.15.08 at 1:49 am

> In general, I think you’ll find that
> the size of manufactured homes in the
> U.S. has ballooned, too.

Just a note on terminology here: while all trailer houses are manufactured, by no means are all manufactured houses trailers. There is a growing trend to pre-assemble large structural components of houses in factories and then do final assembly on the building site. Much the same as the transformation in shipbuilding in the 1950s and for many of the same reasons: lower cost and greater precision/quality of the manufactured components vs. site-built. These houses range up to mansion size in every quality level.



vivian 05.15.08 at 1:57 am

John (or Tim), I happen to have the American Housing Survey 2003 semi-massaged and SAS-ready. It has four housing types (single family detached, attached, multi-unit and manufactured). Not perfect because it compares units of all ages still around. However, for houses of 2-6 bedrooms, for example, the median home_sq_ft used to be much lower than the mean, but is inching up towards it for newer homes. And if you want a one- or two-bedroom detached house, you’re probably getting an old one.

Within the house-type and number-of-rooms (or bedrooms) square footage has risen by a more modest 25-40% or so (by the intraocular comparative test) and the fraction of new houses with more bedrooms is up compared to the fraction of older homes still around.

I haven’t looked at income, rents, rural/urban or pmsa variations, but they’re in the dataset, alogn with mortgage payments, rents and state of repair. It’s off the Census website, or I could send you my 850MB SAS file.


Pete 05.15.08 at 2:45 am

I think we have much longer mortgages than we used to. Several people I know have 30-year mortgages. This, theoretically, makes it possible to buy a much more expensive place than a 0- or 15-year, say.


Tom T. 05.15.08 at 2:48 am

John Q, what about the effect of land cost? Since new home construction is generally farther out in the suburbs, the house itself can be affordable at a larger size because the land (which is a substantial portion of the overall cost in many places) is that much cheaper.


Quo Vadis 05.15.08 at 5:01 am

New McMansions are often remarkably flimsy compared with their early twentieth-century counterparts.

I used to own a house built in 1925 and I had some foundation work done on it. The foundation guy told me that back in the old days they had to build really solid houses because they built them on very poor foundations. These days they build relatively flimsy houses on really solid reinforced concrete foundations. The latter works better in the long run.


nick s 05.15.08 at 5:44 am

a fixed stock like housing takes a long time to adjust

Though in the US, my guess is that it adjusts faster than most places. Americans do like knocking down buildings.

(My parents are still not comfortable with the whole idea of wooden houses. The Three Little Pigs shapes the British homebuilding industry fare more than its American counterpart.)

The flatapartment I’m in is, I’d guess, more or less the same square footage (967 sq ft) as the 3-BR semi I grew up in. But it’s what would be considered temporary housing for most people in the area. Indeed, with apartments, you have to consider that outside urban areas, they fall into two categories: where poor people live, or where more aflfuent people live for a short period while looking for a house.

In general, I’d agree with sg: housing is one of those markets where there’s always a gap between what’s optimal and what’s available, but the disparity between the two in the west (and certainly the US) is pretty big. I’m interested in grackle’s comment #30, though, which ties square footage to the willingness of lenders to finance purchases.


R 05.15.08 at 6:34 am

I’d be pretty sure that above median income families today live in substantially larger houses (on average) than they’re above median income counter parts 50 years ago. Though I would guess that the growth in square feet per capital is nothing compared to the growth in bathrooms per capita.

Where to condos/apartments fit in to the the picture though? They might have the same effect on the calculations as mobile homes, making the average “home” growth smaller than the growth of detached single-family homes.


abb1 05.15.08 at 6:56 am

And abb1, dude, doesn’t it occur to you to spend 30 seconds checking something before making a confident—and in this case completely wrong—declaration of fact?

I didn’t feel it was necessary because I still remember the lecture – video linked by Chris Bertram a few days ago on this site. The very first graph in that lecture showed the average family income going up, seemed like quite a bit higher than 10%. Perhaps those were different kinds of stats? I dunno, she was talking about the average family.

Incidentally, something was said there about the average house size as well. I remember she said that it increased slightly – one more room or one more bathroom, but not both. That’s from 1970 to 2005.



novakant 05.15.08 at 9:05 am

To get the full story you’d need to take account of apartments (I haven’t looked at this).

Considering the fact that somewhere between 75-80% of the US population lives in urban areas, which are dominated by apartments and older houses, I really think this factor has to be taken into account before any conclusions are drawn.


Nick 05.15.08 at 9:37 am

“You will observe, Watson, that every dwelling in this street has a double-garage which the owners use to keep the wind and rain off Mr Daimler’s invention.”

From Sherlock Holmes & the Strange Death Of Off-Street Parking 1896


SamChevre 05.15.08 at 1:14 pm

Another confounding factor, IIRC.

The NAHB statistics track homes built by members of NAHB–who tend to be large-tract suburban builder/developers. I’m pretty certain that the growing house size is the case most strongly in that market; it is very very difficult to get planning approval or financing for large small-home developments. And as Brett points out, in many counties the minimum legal size is around 1000 sq. ft. (In mine, it thankfully is not, as my house is half that size.)

Also second brad, in #26. Improved materials and tools have considerably decreased the cost/sq ft relative to older techniques.

Another important fact is that more sq ft has never increased cost linearly. What it used to increase nearly linearly was maintenance cost; better materials, and most especially better insulation and windows, has made heating/cooling a large house much less more expensive than it used to be.


Nick 05.15.08 at 1:19 pm

New McMansions are often remarkably flimsy compared with their early twentieth-century counterparts.

I’m not sure that it was a steady trend, though. The first house my wife and I bought was built in 1978. It felt significantly more flimsy (with shoddy wiring too) than either the 1950s house I grew up in or the 2006 house I now inhabit. The 2006 house has various features (e.g. hurricane clips now required by code) which probably make it better able to withstand high winds than the early 20th century homes nearby.

As quo vadis suggests above , many of the older homes in my area have nice work inside, but the foundations are crumbling and floor joists are rotting due to the vented crawlspace. Newer homes built on slabs or sealed crawlspaces seem better designed for the climate in the southeast.


Nick 05.15.08 at 1:41 pm

re: #31 and #32

In the context of housing surveys, does “manufactured home” include trailer/mobile homes? As cranky observer writes, a manufactured home built on a permanent foundation on land that the homeowner owns is a whole different kettle of fish than a single- or double-wide on a temperary foundation on rented land. The latter is more like a car whose value depreciates over time than a house that appreciates in value. If only the former is included in the average of home sizes, then the numbers will be skewed.


Sock Puppet of the Great Satan 05.15.08 at 2:32 pm

“My parents are still not comfortable with the whole idea of wooden houses. The Three Little Pigs shapes the British homebuilding industry fare more than its American counterpart.”

Yeah, but the story leaves out digging the Third Little Pig’s corpse out of the rubble after a big earthquake.

At least on the West Coast, wood’s the way to go for structural reasons, unless you’re going to make your house out of reinforced concrete.


Andrew 05.15.08 at 3:36 pm

A few of you may have watched the Elizabeth Warren video posted here.

In the comments, Aaron Swartz linked to a review of the video in the Boston Review which notes:

The median owner-occupied home grew from 5.7 rooms in 1975 to 6.1 rooms in the late 1990s—an increase of less than half of a room in more than two decades. What was this half a room used for? Was it an “exercise room,” a “media room,” or any of the other exotic uses of space that critics have so widely mocked? Nope. The data show that most often that extra room was a second bathroom or a third bedroom.

The wealthy may be living in spacious new digs, but middle-class families are not. The proportion of families living in older homes has increased by nearly 50 percent over the past generation, leaving a growing number of homeowners grappling with deteriorating roofs, peeling paint, and old wiring. Today, nearly six out of ten families own a home that is more than 25 years old, and nearly a quarter own a home that is more than 50 years old.

Even as middle-class living conditions have improved only modestly, the burden of paying for a home has increased dramatically. Over a generation, the average number of rooms in a home increased by seven percent as average mortgage expenses increased by 69 percent—at a time when other family expenses were falling. The impact of rising mortgage costs has been huge. The proportion of families who are “house-poor”—that is, who spend more than 35 percent of their incomes on housing—has quadrupled in a single generation. Today it often takes two working people to support a mortgage. A police officer or elementary-school teacher earning an average salary could not afford to pay the mortgage of a median-priced home in two thirds of the nation’s metropolitan areas.


Tim Worstall 05.15.08 at 4:25 pm

“I’m picking up the sound of snark here, Tim,”…true, written pre-caffeine, apologies.

With no snark: the expansion or not of housing area per capita, if we knew it accurately, and could adjust for the different qualities (mobiles, prefabs, McMansions, apartments) might tell us something about whether incomes have in fact increased or not?


F. Blair 05.15.08 at 7:20 pm

How much does the increase in the average number of rooms tell us? If, as my own experience suggests, rooms today are much bigger (certainly bathrooms and kitchens, for instance, are huge today relative to what they were), the small increase in the number of rooms could still translate to much bigger home sizes. Given Warren’s general argument, we know she’s going to spin the numbers to look as if things have gotten better, so I’m not sure there’s reason to take them at face value.


Andrew 05.15.08 at 7:55 pm

Fair enough, f. blair.

Here is the data on median and average square feet of new homes.

1973 = 1525′
2006 = 2248′
increase = 68%

1973 = 1660′
2006 = 2469′
increase = 67%

In neither case did the size double, but there has clearly been an increase in size for new constructions.

However, as noted by Warren above, “Today, nearly six out of ten families own a home that is more than 25 years old, and nearly a quarter own a home that is more than 50 years old.”

Which means that 60% live in homes with a median size of 1550′, which is was the median in 1981.


Andrew 05.15.08 at 8:10 pm

Addendum: The original post suggests the size of new constructions doubled since 1950. The dataset I provided begins in 1973, so does not address a possible doubling since 1950.


abb1 05.15.08 at 8:18 pm

Which means that 60% live in homes with a median size of 1550’, which is was the median in 1981.

No it doesn’t, necessarily. Additions. My house, built in 1956, was first expanded in the early 1980s. then I bought it and built another addition. In the end it had become at least twice the original size.


Slocum 05.15.08 at 8:25 pm

Which means that 60% live in homes with a median size of 1550’, which is was the median in 1981.

No it doesn’t — eventually most old homes are modified and expanded. For example, almost none of the the originally tiny Levittown houses look like they did when they were built:

“These days, many of the original Levittown homes have been remodeled and enlarged beyond recognition. A decade ago, there were perhaps 200 unaltered Levitts left, but only a handful remain today.”

And I don’t mean to be tiresome, but let me point out one more time — even if houses were not a single square foot larger than they were in 1950, we’d still have experienced a 30% increase in living space per capita because of decline in the average number of persons per household.


F. Blair 05.15.08 at 8:54 pm

Thanks for the data, Andrew. I actually think the data you posted is more relevant to the original question at the heart of John’s post than the 1950 data is, since 1973 was also when the postwar boom ended. This is precisely what John was trying — and failing — to explain with his mobile-home theory: even as wages have stayed flat in the last 35 years, the size of new homes has risen sharply (more sharply, I suspect, than it did between 1950 and 1973).


Andrew 05.15.08 at 10:13 pm

Slocum (and abb1),
It is a good and relevant point that number of average household members has been declining.

However, pointing to ‘additions’ is rather a rather spurious argument unless you believe for some reason that putting additions on homes is a recent innovation.


Slocum 05.15.08 at 10:29 pm

However, pointing to ‘additions’ is rather a rather spurious argument unless you believe for some reason that putting additions on homes is a recent innovation.

No — the point is that the houses that existed in 1981 whose median size was then 1550′ — those exact houses — are larger on average now than they were 27 years ago because of additions during that time.

As the NY Times Levittown article noted — a decade ago there were still almost 200 unmodified Levittown houses whereas now there are almost none. So the average size of even the average 50-year-old Levittown house increased during the last decade.


James Wimberley 05.15.08 at 11:04 pm

I don’t get John’s original problem. The increment to the housing stock has consisted of bigger houses for the few whose incomes have risen sharply.

Is the tale of the Three Little Pigs really so silly when you look at Hurricane Katrina? I built a house in Alsace in the 1970s: concrete pier foundations (floodplain), reinforced concrete basement walls and floor slabs, double brick walls (22 cm + 6cm) with 10cm insulation between, kingpost wooden roof beams. Now retired, I live in a smaller terrace house in Spain built a decade later by a developer: 25 cm reinforced concrete frame set directly into rock, double non-load-bearing brick walls, compound concrete floor slabs. Either property would be thought ridiculously small by US suburban standards; but both will still be there in a hundred years.

I wouldn’t put the European preference for solid construction down to age-old tradition. I was shocked to read the surveyor’s report on a terrace house we bought in Brighton, built around 1870 of “bungarouche”, a cheap rubble concrete invented by local speculative builders for the Regency building boom. Basically the house stays up because of the similar ones on either side.


Andrew 05.15.08 at 11:46 pm


My mistake. Thanks for pointing out the flaw in my logic.

The AHS has the median square ft/person.
In 1993 (the earliest date I could easily find data on), median sq ft/ person was 689′.
In 2005 it was 752′.

If anyone knows where to find data from 1973 or so that would be super.


soru 05.16.08 at 12:06 am

What’s the average/median inter-generational wealth transfer? How much of a house bought in 2008 is paid for with money earned:

1. in 1973
2. expected to be earned in 2033?

How have those proportions changed between 73 and 07?


1. those proportions were fixed.
2. you assume capital available to spend on housing is proportional to, say, the last 30 years of family income
3. you use the median figures from

1973 household has an income of 47,768
2006 dollars and a housing budget proportional to $909,730.
2006 household has an income of 58,407
2006 dollars and a housing budget proportional to $1,595,254.

Picking 1970 as the reference year instead of the boom year of 1973 gets you a budget of $770,488, less than half the 2006 figure. So you don’t actually need anything else to explain a change in house size, as long as you don’t think of houses as being bought like sweets out of weekly wages.


soru 05.16.08 at 12:17 am

Ignore that last figure – the data doesn’t go back far enough to calculate it properly. But presumably, median family income during WWII wasn’t so high as to invalidate the basic argument: the average non-immigrant in 2006 had parents with a 1973 level of income, the 1973 generation didn’t.


Martin James 05.16.08 at 1:15 am

It seems like John Q. is looking at housing mainly as a consumption good. For the bottom four income quintiles home equity is from 73% to 80% of net worth, with the peak in the median incomes. Only in the highest income quintile does anybody really own much of anything but part of their house.

For the third income quintile, total net worth in 2002 was $48,200 of which $38,694 was home equity. The answer to John’s question may be that housing is one of the few ways to consume like a Proud American and build up some equity, too. Gotta luv that American dream.

The depressing part was realizing that since my annual federal taxes exceed the median net worth, I’m pretty much screwed once the Democrats win. The only consolation is that like a good Laffer curve supply-sider, I can quite working and live off that home equity until the republicans take over again.


reason 05.16.08 at 7:33 am

Martin James…
Strange comment (not sure how to take it actually), it clearly isn’t fully serious. And well who cares abut you, to be blunt. You don’t seem to worry about anybody else.

But back to the theme. Surely the telling point is that housing is a stock and income is a flow. You pointed this out and then have forgotten about it. That plus the trade off poeple have made with increased commutes. Whether people really are better off like that, is another point (they spend less and less time in their big houses because of the commutes and working to pay off the mortgage) and a large part of the time they are there, they are asleep. But then housing is a positional good. The utility comes not from owning the house, but from bragging about it.


Martin James 05.16.08 at 1:40 pm


You’re right that it wasn’t fully serious. I also phrased it poorly. The depressing part wasn’t really the future taxes I might pay, but that I realize that the electoral math is such that increasing taxes on those with higher income makes sense, yet those with higher incomes still will have more options.

Stating that fact bluntly does seem uncaring, but as Jesus would say anybody can love their friends. Its loving your enemies that’s earns you the big rewards.


lemuel pitkin 05.16.08 at 3:43 pm

John Q. (and Tim Worstall, from the other side) think the puzzle is that a major increase in housing consumption is incompatible with flat household income. John Q. suggests the former must therefore be false, Worstall the latter.

But the evidence is pretty clear that both facts are true, partly because housing costs have declined, but mostly because people are choosing to spend more on housing.

Why people choose to devote to more of their incomes to housing is the puzzle. Presumably, it’s some mix of:

1. Regulations have made it more difficult too build smaller homes.

2. The specialized mortgage-lending system makes it easier to finance housing purchases than alternative forms of consumption. (Yes, this was probably at least as true 30 years ago, but as more and more consumption is debt-driven, ease of financing becomes more important.)

3. Houses becoming more attractive as an investment as well as consumption good. Especially true in the bubble period, obviosuly, but probably a longer-term trend.

Untangling these explanations — and whatever others I’m not thinking of — is an interesting question. But John Q.’s approach, of trying to find reasons why housing consumption has not increased as much as it seems to have, is pretty clearly barking up the wrong tree.


lemuel pitkin 05.16.08 at 3:45 pm

Oh, and

4. The tradeoff between longer commutes and larger houses got more favorable in a period of suburbanization and cheap oil.


SamChevre 05.16.08 at 4:49 pm

And 5:

Changes in school costs and the legal framework in which schools operate made lving in a uniformly high-income area much more valuable.


abb1 05.16.08 at 4:59 pm

Is it also possible that house-size increase is somewhat exaggerated while family-income increase is somewhat underestimated? If it was so, would it make everybody happy?


Martin James 05.16.08 at 5:36 pm


Good try at cutting a deal, but you’re never going to make it as a real estate broker unless you tell the buyer the house is BIGGER than reported and tell the mortgage company that the buyer’s income is BIGGER than reported.

These arguments became a lot less meaningful to me when I started looking carefully at the time use statistics. Across all adults in the USA, the average work hours per day is 3.75 hours, the average leisure is 5.09 and of that 5.09, 2.58 is spent watching TV. Big house, flat income, who cares? What people really seem to want is a good TV show.


abb1 05.16.08 at 6:43 pm

I have hard time accepting that with so many more wives working the average family income has only grown 10%. Need to calculate the average income of families living in houses or something.


Martin James 05.16.08 at 6:54 pm

Average income for married households is up more than 10%, but the percent of households that are married persons has dropped significantly.


spencer 05.16.08 at 7:51 pm

Because the housing stock is a very long lived asset the average age of the housing stock is around 22 years old.

The median home in the US in 2005 was built in 1973.

Because of this using the size of new homes is using data on the marginal home rather then on the average home.

You need to go to the American Housing Survey at Census to get the data you want.

For example, single family detached homes only account for around 62% of the housing stock.


spencer 05.16.08 at 8:03 pm

The national association data you link to gives the average size of a new home in 2005 as 2,434 sq. ft..

But the Census housing survey has the median size of the existing housing stock at 1,768 sq ft.==
this is 72% of the data you linked to.

I did not check to see how far back the Census housing survey goes back, but you might want to check that data.


spencer 05.16.08 at 8:50 pm

I found Census data for 1985 that had the median square footage at 1,583 sq. ft.

That yields an 11% increase to get the 2005 1,758 sp ft.

That is over 20 years or about a 0.5% growth rate.

This implies that the increase over 60 years from 1945 would be about a third as compared to the Builders Association statement that new home sizes doubled.


John Quiggin 05.16.08 at 9:25 pm

Thanks for this, Spencer. I suspect the increase may have been more rapid in the decades immediately following WWII, but my interest is mainly in the period since the 1970s which your data covers.


paul 05.17.08 at 4:59 pm

Isn’t tracking square footage of houses a really serious example of looking under the lamppost? Adjusted floor area doesn’t capture the utility of a house, its construction cost, or its selling price. Number of rooms and utility are somewhat better correlated, but even that is seriously dependent on how you define “room” and which rooms (e.g. bathrooms, as mentioned above) get added.

I live in a house whose utility would probably be improved if it were several hundred square feet smaller, because the rooms that are larger than they need to be cost more to furnish, heat and cool; they also make it harder to clean and lead to unpleasant furniture arrangements. I’ve seen plenty of mcmansions where a normal house worth of furniture is just lost in the big rooms, and people turn out to live in a few reasonably-sized corners. Unfortunately, most such houses are built (structurally and architecturally) in ways that make it impossible to rejigger the space into more livable condition.


lemuel pitkin 05.17.08 at 7:05 pm

Isn’t tracking square footage of houses a really serious example of looking under the lamppost?

At 21, I posted census data showing a significant increase between 1984 and 2006 in the share of household spending going to housing. So the phenomenon does not seem to be just an artifact of looking at square footage.


paul 05.19.08 at 7:55 pm


Perhaps you misunderstood what I meant by “looking under the lamppost”. The increase in square footage is typically invoked as a response to the increases in share of household spending going to housing. Implicitly or explicitly, it’s “but look how much more people are getting for the additional money they’re spending”.

My point was that increased square footage doesn’t necessarily mean increased utility (which tends to correlate better, although still imperfectly, with number of rooms). Increase square footage also doesn’t correlate all that well with increased construction cost. All it correlates with is spending more money, which is something we already knew.


Frank the sales forecaster 05.19.08 at 9:12 pm

dang ya’ll
In 1950 less than 8% of households had a college grad as head of household and less than 33% of households had 2 wage earners. By 2005 more than 20% of households had a college grad as head and almost 66% of all households have 2 wage earners. Lots of these households have 2 college grads. Housing is a function of income with housing costs increasing at rate greater than income. Given the puzzling evidence that a much greater percentage of households own a home and that it is a much nicer home than previous generations, maybe your understanding of income growth/inflation needs adjusting.

Comments on this entry are closed.