The Housing Crisis Hits Wisconsin

The National Housing Crisis

The national housing crisis has grown to epic proportions in the last five years. Nationwide, housing prices have continued to climb year after year, with the average U.S. home today selling for 25% more than in January 2020. But this dramatic rise in housing costs is merely the continuation of a longstanding trend: home prices have increased 150% since the year 2000 and 547% since 1980.

In 1980 the median home in the U.S. sold for roughly $64,000. Then, by the year 2000, the median home cost $165,000. Today, the median home sells for more than $412,000.

The principal reason for the unprecedented rise in home prices is the same reason all assets prices have increased: decades of inflationary policy have continually debased the Dollar and reduced its purchasing power. But importantly, when compared to other goods, housing prices are particularly high, indicating that rising housing costs are not only the result of monetary policy but of housing policy generally.

A detailed discussion of the individual policies which have contributed to this situation will be discussed in the ensuing pages. For now, the policies most responsible for the current housing crisis are expansionary monetary policy, burdensome building codes, and zoning and land use regulations. Together, these three things have cartelized the housing market––restricting supply and preventing first-time homeowners from entering the market.

As a result, homeownership rates are artificially lower than they otherwise would be; the cost of living is the worst its been in 50 years; median household savings is only $6,100 and makes the traditional 20% downpayment nearly unattainable; and the percentage of young adults living at home with their parents is the highest its been since 1940.

The housing situation can best be understood the same way all economic things are: It's supply and demand. There simply are not enough houses to go around.

Home Prices: Through The Roof!

The increase in housing costs have long-outstripped that of all other consumer goods when compared to the Consumer Price Index (energy and food included). The rate at which housing price increases began to diverge from all other goods started in 1995. Incidentally, the '90s also mark the beginning of the Federal government's intense housing market activism (Peter Lewin, p. 54-56) which peaked in the early 2000s and culminated in the Subprime Mortgage Crisis of 2008.

According to the chart above, since 1995 CPI increased by roughly 208 points, whereas housing increased by 443 points––more than double that of the CPI. This disproportionate increase in housing costs reflects a real increase in the cost of housing.

This fact is also borne out by the national income data. If we look at the ratio of median home prices to median household incomes, we see that the median home was 3.5 times the median income in 1984, 4 times the median income in 2000, 4.5 times the median income in 2010 and 5.8 times the median income in 2022.

In just 40 years, then, the nominal cost of housing has not only increased five-fold, the relative cost of housing as a factor of household income has increased by 65%. Housing is more expensive today than ever before (see also: S&P National Home Price Index).

Housing in the U.S. has become increasingly difficult to come by and exceedingly costly for those who happen to be able to buy a home at all. A recent report by the Joint Center for Housing Studies of Harvard University included a graph that indexes current housing costs to the year 1990––including a breakdown of home prices, monthly housing costs and mortgage rates.

Their research shows that since 1990 mortgage rates are 30% lower but rising quickly relative to the all-time low in 2021, while housing prices are 80% higher and monthly payments are 45% higher. 

The same report also found that monthly payments for the median-priced home now exceed $3,200 (p. 6) per month.

The National Association of Home Builder's (NAHB) July 2024 issue of their Cost of Housing Index (CHI) reported that not only are homes priced at the median largely unaffordable, but that households who do purchase homes spend 38% of their income on mortgage payments. This rate of housing expense far exceeds the long-time rule that housing expenses ought to be "no more than 30% of your income."

With respect to the number of households who can afford to buy a home at the median price, in March 2024 the NAHB found that "77% of households cannot afford a median-priced new home." They also went on to say that for every "$1,000 increase in the median price of a new home... 106,031 households [are priced] out of the market," such that the minimum income required to afford the median-priced home is more than $150,000!

Are Home Prices Falling?

Yes but total housing costs are not. The relative decline in home prices since Q3:2023 has been understood by some to be a sign that the housing market is responding to increased home prices by increasing the number of new constructions. But this recent decline in prices is actually only the effect of a high-interest-rate environment.

Higher mortgage rates have forced sellers to lower their prices in order to attract buyers who cannot afford high-interest mortgage payments at high home prices. In effect, slightly lower housing prices and high mortgage rates have kept the total cost of owning a home the same.

In the graph above you can see that the rise in the average 30-year fixed mortgage rate (orange line) precedes the decrease in median U.S. home prices (dark blue line). This demonstrates that the recent fall in home prices is interest-rate induced, not supply-induced.

Build, Baby, Build?

The number of new housing starts does not reflect a significant increase in housing supply that would decrease home prices in any considerable way. In fact, the ratio of new housing units to households is at an all-time low excepting for the Housing Crisis of '08.

The number of new housing units as a proportion of total U.S. households is astonishingly low––lower, even, than what it was in 1990 when the Federal government first embarked on its "affordable housing" crusade.

The graph above is probably the clearest depiction of the housing crisis. It shows that the number of new housing units completed as a percentage of total households is a meager 1.1%––a third of what it was in 1972. The number of new housing starts as a percentage of total households is equally disappointing at 1.08%––also a third of what it was in 1972.

Keep in mind that even though the Subprime Mortgage Crisis occurred more than 15 years ago, its effects on housing development have lingered. Many developers feel "burned" by the last housing bubble and have still yet to recover. Add in pessimistic inflation expectations and rising material costs and it's little wonder developers aren't chomping at the bit to build more houses.

In the NAHB's 2023 survey of home builders, 80% reported concerns over interest rates and inflation; 73% were concerned with the cost of labor; 63% were concerned with the cost of building materials; and 49% were concerned with the difficulty of getting zoning and permitting approvals (p. 15). Consequently, housing units are not being built. 

Estimates of the current housing shortage run the gamut from 1.5 million to more than 7 million units. But whatever the true number is, the claim that there is a shortage is corroborated by the fact that total housing unit vacancy rates are among the lowest rates observed in the last 24 years.

In a housing market with a ready supply of dwellings we would expect moderate (12-14%) vacancy rates, not near-all-time-low vacancy rates. Yet that is precisely the situation the U.S. finds itself in: total housing vacancy rates are roughly 10%––lower than the entire period spanning from 2000-2020.

Gen-X Still Nesting: Millennials & Gen-Z Can't Fly the Coop

In what is probably the most startling statistic pointing to the presence of a housing crisis, 17% of 25 to 35 year-olds are still living with their parents––the highest rate since 1940.

Chris Salviati of Apartment List looked at U.S. Census data and produced the following graph showing the share of 25-35 year-olds still living at home...

The percentage of 18-24 year-olds living at home has increased since 1960, too. The Census Bureau tracked this demographic and found that the share of young adults living at home with their parents has increased by only 5% for men but 20% for women! In 1960 approximately 52% of 18-24 year-old men lived with their parents––today, it's roughly 58%. For women in this same age group, however, 35% lived at home with their parents in 1960 whereas 55% of women report living with their parents in 2023.

Furthermore, Salviati also looked at the income data available for the 25-35 year-old age group and assessed their ability to afford rent in their counties of residence.

Since 1960 the share of 25 to 35 year-olds who can afford median rent in their county has decreased, on average, by 7% per decade. Note, also, that the decline in this demographic's ability to afford their local rent in the '90s and 2000s remained constant. As was explained earlier, aggressive Federal housing policies were able to artificially increase the supply of housing by making it generally more affordable––in the short run. When the housing bubble burst in 2008, however, the artificial nature of the housing market was revealed and the decline in rent affordability resumed.

That this age group is least likely to be able to afford housing is not at all surprising, however. Income is strongly correlated with age and experience, and 18-35 year-olds have relatively little of either. As a result, half of this age group is near the bottom of the income distribution, has little accumulated savings, and would be disproportionately affected by housing costs anyway. 

But the extent to which they have been disproportionately affected is exacerbated by a precarious labor market and housing cost increases that exceed wage increases. Their frustrations are primarily the result of a hampered housing market's inability to supply additional housing to the largest home-buying generation to ever exist in the United States. On top of all that, their ability to buy homes is doubly complicated by the fact that longer life expectancies have reduced the turnover rate of existing homes. 

While many might chalk this up to poor work ethic, bad spending decisions or younger generations' attainment of college degrees with low financial returns, the data set forth thus far hopefully temper such opinions. Readers are encouraged to read MacIver's other reports on the state of the economy––The Real Wisconsin Economy and Recession Outlook, in particular––to better understand the labor market.

As for poor college degree selection, degrees with low returns are relatively rare and still manage to improve graduates' job prospects––even if they end up working in an industry outside of their subject area. But most importantly, all degrees, regardless of major, yield higher lifetime returns, on average, than just a high school degree. For example, degrees in Agriculture, Culture & Gender Studies, Foreign Languages & Linguistics, Liberal Arts & Humanities, Philosophy & Religious Studies, Theology, and Visual & Performing Arts were only 10.27% of Bachelor's degrees conferred in 2020-21. As for lifetime earnings, a recent paper by the Center for Research on the Wisconsin Economy (CROWE) at UW-Madison shows that the lifetime earnings of all UW-Madison graduates, regardless of major, are greater than those of their peers who hold only high school degrees (see graphs below).

As was said in the opening section, monetary policy, building codes, and zoning and land use regulations are principally responsible for the housing crisis. Their influence will be expanded on in the next section: Wisconsin Housing.

Worst in the Nation

Wisconsin home prices are some of the worst in the nation. While states like California, Hawaii and New York have higher average and median home prices, housing affordability in Wisconsin is worse than nearly any other state, with prices rising much faster than incomes.

While median home prices in Wisconsin are on par with the rest of the U.S., compared to its four neighboring states Wisconsin ranks 2nd for the highest median home prices.

The graph below shows that the median home costs roughly $400,000 at a cost of $212 per square foot. These are the costs of the typical home in Wisconsin––not the statistically average home. In 2016 Wisconsinites could purchase a typical home for roughly $190,000––a price which increased steadily by $20,000 per year over the next five years.

But after 2021, Wisconsin home prices began to rise even more rapidly––by more than $36,000 per year up to the current date. Why? The short answer is that home prices in Wisconsin––and the Midwest, generally––had more room to grow. States like California may have millions of homes selling for millions of dollars, but even California salaries aren't enough to afford a modest 5% increase in housing costs. But a 10% increase on a $200K Midwestern home? Now that's affordable.

Wisconsin has had below average housing prices for a long time; prices which attracted homeowners from other states with disproportionately high housing costs. The result of this new demand for Wisconsin housing was a rapid increase in housing costs, inflated by immigrants from other states with higher incomes (p. 16).

These lower initial housing prices, in conjunction with the net increase in migration, is a big reason housing prices in Midwestern metropolitan areas have risen faster than metros in other parts of the U.S. as measured by the AEI (p. 13).

 The Best & Worst Counties to Buy a Home

For those looking to buy a home in Wisconsin, the table below ought to be especially helpful––it utilizes median home price data as of June '24 and ranks Wisconsin's 72 counties from worst to best in terms of median home prices. And in a result that will surprise few, Waukesha, Door, Ozaukee, Vilas, Dane, Brown and Walworth county all have median home prices of half a million dollars or more.

If we look at the heat map below, the reason for higher home prices becomes rather intuitive. When looking at the counties with major metros––Madison, Milwaukee and Kenosha, for example––the counties surrounding those metros tend to have higher home prices. This indicates that the metros are "good" places to work because they offer better-paying jobs, but are not good places to live. A cursory glance at the crime rate statistics of these metros makes this fact clear.

The overall crime rate in these metros deters homeownership and property ownership in general because property ownership becomes more costly. Consequently, Wisconsinites commute into these metros from surrounding counties where crime rates are relatively lower. This increases the demand for housing in those areas, further exacerbated by the higher salaries of metro-area-employed people, and results in higher housing costs.

On the other hand, housing price in counties like Walworth and Door county are heavily influenced by tourism. For example, Lake Geneva in Walworth county is so popular a tourist destination that housing prices in Fontana, Lake Geneva, and Lyons are pushed ever higher by retirees or tourists looking to vacation or purchase second homes near the lake, even if they can't afford to purchase homes in the immediate vicinity of the lake. The effects thus emanate outward into the surrounding cities and towns.

To truly understand which counties are most affordable, however, we need to look at the housing expense ratio. The median home prices are a good starting point for those looking to buy a home provided they already have an income, but the housing expense ratio by county tells us which people in which counties are the most cost-burdened. It tells us which counties have the greatest "bang" for the "buck."

To that end, the table below ranks counties from best to worst with respect to their Housing Expense (HE) Ratio.

The graph above ranks Wisconsin counties in the year 2022 when average home prices were $328,000 and median household income was $73,000. That year the average Housing Expense (HE) Ratio was 4.57 and the median was 4.3. A total of 19 counties had HE ratios over 5, and 49 counties had HEs of 4 or more. This is a far cry from from the housing expenses that were typical of the 1980s. 

2024 is even worse in terms of housing affordability. The graph below uses median listing price data from realtor.com and median household income projections based off the 2012-2022 Census Bureau SAIPE data. The graph below estimates current housing expenses in Wisconsin by county and shows that the housing crisis has not only hit Wisconsin, but has hit it hard.

Based off our estimates, the average HE ratio for Wisconsin counties in 2024 is 6.09 and the median HE is 5.87. Most Wisconsinites, therefore, can expect to buy a home at a cost that is roughly six times their household income.

Four counties in the state currently have HEs of 9 or more––Richland, Washburn, Vilas and Menominee––and 55 counties have HEs greater than 5. In relative terms, then, Wisconsin housing is 67% more expensive today than it was in 1980. And in the span of just two years, the typical Wisconsin household buying the typical Wisconsin home can now expect to pay 36% more to purchase a home.

How Monetary Policy Contributes to the Crisis

The Federal Reserve's role in contributing to the housing crisis is perhaps the easiest to understand. As MacIver has shown in The Real Wisconsin Economy and Recession Outlook, the FED's policy of continual inflation at a targeted rate of 2% per year dilutes the purchasing power of the Dollar and causes wages to lag behind general increases in the price level.

Since the FED's creation of money necessarily increases the supply of loanable funds in the banking sector, interest rates are artificially lowered and asset prices increase. These effects are the result of both the debasement of the currency and Cantillon or Injection Effects. A Cantillon Effect is when the first recipients of newly created money benefit from an increase in their real purchasing power by purchasing goods today at yesterday's prices. Since the new money has not yet percolated throughout the economy, those who receive the new money first are enriched by a policy of inflation; those who receive the new money later, after the general price level has gone up, are hurt.

When the FED creates money, some of this new money is used to purchase real estate and housing which disproportionately drives up their prices. These two things tend to be affected by a policy of inflation because both overwhelmingly require financing via a mortgage and are therefore sensitive to interest rates (hardly anyone buys either of these with cash), and because State and local regulations have created shortages of each (more on that in the next section) that restricts their supply and drives up prices.

As was said in the first part of the report, intervention in the housing market on the part of the Federal government has been especially activist since the 1990s. One of the effects of this interventionism has been that banks, investors, entrepreneurs and consumers have changed their perception of housing. No longer is housing considered a consumption good, but rather, an investment to be held to maturity.

Normal buyers, too, begin to "buy and hold" housing units in an attempt to capitalize on the artificially high housing prices. This means that the turnover rate of existing housing units is artificially lowered, further exacerbating the shortage.

This perception is entirely warranted given the existence of Government-Sponsored Enterprises (GSEs) like Fannie Mae and Freddie Mac whose modi operandi is to subsidize housing and bailout housing speculators if need be. The FED's role in all this has been to purchase these mortgage-backed securities––to the tune of $2.3 Trillion according to their latest Balance Sheet report––from GSEs, creating a moral hazard which further incentivizes speculation.

This creation of money and injection of credit is what creates the "Boom" phase of the business cycle, marked by overconsumption and malinvestment. The actions of the FED, the GSEs and looser lending policies on the part of the FHA are precisely the cause of the housing boom and bust of the 2000s. With respect to the FED, their creation of money and injection of credit finances the boom by devaluing the currency––an effective tax on savings.

Thus, real distortions in the market are materialized by real changes in the value of savings and money.

How Zoning and Lot Regulations Contribute to the Crisis

The housing crisis would not be possible without the existence of Zoning and Lot regulations. At the exact same time that the Federal government is "boosting" demand for housing via subsidized housing policies and low interest rate environments that encourage speculation, local governments restrict supply by regulating how land can be used, and how much can be used. Further restrictions on the supply of housing are imposed by the Federal and State governments by imposing building codes––things like which materials must be used, the width of doors and stairwells, and how energy efficient a water heater must be.

When all these things are added up, the final summation is a housing market that lacks variability in housing types, restricts the number of dwellings, and sends housing prices through the roof.

Of all these factors, however, none is more important than the regulations imposed at the local level by municipal and county-level zoning authorities. As I discussed in a previous article (Zoning in Wisconsin), even very modest zoning reform can have a huge impact on improving housing costs and raising incomes. 

As Emily Hamilton found in her study of Houston, Texas's minimum lot size reform, reducing minimum lot size requirements from 5,000 sq/ft to 3,500 and 1,400 sq/ft resulted in:

during the same period.

The result of this reform was that Houston's housing expense ratio (HE ratio) fell to 3.3 times the median income in 2020 at the same time that Wisconsin's HE ratio was 3.9 and Milwaukee County's was 3.5.

Mike Mei's study of Houston's lot size reform found that "Over a lifetime, the deregulation amounts to $18,000 which is about one third of the median income" (pg. 28) and that "the overall base price of housing decreased by about $12,000" (pg. 23). These sorts of improvements are unsurprising to those who understand economic theory, but the magnitude of these effects are likely shocking to everyone. Pursuing similar policies here in Wisconsin would have similarly beneficial effects, and would likely eliminate the housing shortage and end the crisis if the policy changes are liberal enough in their scope.

A Study of Zoning and Lot Regulations' Effect on Housing Prices in Walworth County

In the section below I discuss some of the results of a study I conducted on the effect of Zoning and Lot regulations on housing prices in Walworth County. Before discussing those results, however, first I will introduce the kinds of Zoning and Lot regulations that are typically enforced by zoning authorities, and then review the expected effects of these regulations according to economic theory.

Zoning and Lot regulations entail a number of restrictions placed on lots and building development. On the one hand, Zoning regulations dictate how a piece of land may be used; and on the other hand, Lot regulations dictate how much of a piece of land may be used.

In general, Zoning classifications consist of six different categories: 

• Agricultural zoning;
• Commercial zoning;
• Industrial zoning;
• Mixed-use zoning;
• Multifamily zoning; and
• Single Family zoning

Typically, each category has several distinct zoning designations. Single Family zoning, for example, will usually be subdivided into SF-1, SF-2 and SF-3––each with their own lot regulations. 

As for lot regulations, each category typically implements the following regulations: 

• Minimum Lot Sizes (MLS);
• Minimum Lot Widths (MLW);
• Front, Side and Rear Setbacks (FS, SS, RS);
• Maximum Building Heights (MBH);
• Parking Space Requirements (PR);
• Maximum Dwelling Density (MGD);
• Minimum Landscaping Requirements (MLSR);
• Maximum Coverage (MC); and
• Floor Area Ratio (FAR)

In some instances, lot regulations are even more particular and restrictive. One such example is the implementation of a maximum Floor Area Ratio (FAR) which limits the square footage of a building to be no more than a certain percentage of the square footage of the lot. If a home on a 10,000 square foot lot is only 1-story then it can only be 1,000 square feet in size. But even if it's 2 stories, the square footage of each floor can't add up to more than 1,000 square feet––even though the zoning board would have approved a 1-story home covering 1,000 square feet.

Other such regulations can be even more granular in their requirements. In one "Central Business" district the zoning authority required that "First floor facades facing a public street shall have a minimum of 60% window coverage." 

Now, some people might like this requirement and enjoy being able to peruse a shop from the comfort of the sidewalk before deciding to enter the store. Two things are of not here: first, just because something isn't mandated doesn't mean it won't occur. There is no mandate to exercise yet people do it anyway because they find it valuable. Second, give some thought to the zoning authority's specification that it be sixty percent rather than forty or eighty percent. That's a very specific number. Surely its purpose is well-founded and based on the principles of engineering or thermodynamics. But as a matter of fact, speculation here isn't at all necessary because the zoning authority states the purpose of their regulations at the beginning of the zoning section:

These kinds of rationales are ubiquitous in every zoning code. The first two purposes––appearance and character––are entirely aesthetic. Perhaps the municipal architect fancies themself an aficionado of style and taste. But the last purpose––the preservation and enhancement of property values––is truly revealing.

Given the zoning authority's stated purpose, zoning's basis for existence is overwhelmingly to benefit property owners. To use an analogy, a zoning authority is quite like a feudal lord who aims at improving the wealth of the gentry at the expense of the unlanded peasantry.

But an even more apt description is this: according to their own admission, zoning authorities are literally cartels, and the regulations they impose have the desired effect of preserving and enhancing property values by limiting market entry of competitors and would-be buyers. The purpose of a cartel, of course, is to collude with other members of the cartel to limit supply so that prices, and ultimately profits, rise. Artificially restricting supply to increase prices and profits necessarily entails excluding marginal buyers––increasing producer surplus, reducing consumer surplus, and decreasing social welfare. To understand this better think of the effect of a minimum wage: in one sense a minimum wage says that no one can be paid less than $X/hr; but another way of stating the same thing is to say that no one may have a job who is worth less than $X/hr. The result of a minimum wage, then, is to cause people worth less than $X/hr to become unemployed. Zoning and Lot regulations––often imposing minimums of their own––have the same effect.

In short, theory says that imposing minimum or maximum prices produce either surpluses or shortages, respectively. In the case of the lot regulations listed above, every minimum and maximum produces only shortages because all of them attempt to restrict the supply of land and houses by imposing barriers to entry (minimums) and limits on output (maximums). Listed below are the theoretical effects of Zoning and each Lot regulation.

Zones: Designating how pieces of land may be used places a "cap" on the supply of a particular use which prevents its supply from increasing beyond the cap. Changes in price associated with a particular land use, therefore, will be determined entirely by changes in demand if the entirety of the supply has been consumed (in this case, used). If zoning regulations state that only 30% of the land in a jurisdiction may be used for residential purposes, and there is excess demand for residences, prices for housing will be bid up by buyers because the zoning authority has "fixed" the supply of housing. Rezoning must occur before the demand can be met––a tedious and time-intensive process, to say the least.

Minimum Lot Sizes (MLS): A minimum lot size says that no one may purchase a lot unless they can afford a lot that is at least X acres in size. This regulation directly parallels that of a minimum wage. If someone is attempting to purchase a developed lot, like a single-family home, a minimum lot size of 12,000 square feet rather than 9,000 square feet can be the difference between being homeless and owning a home.

Minimum Lot Widths (MLW): Minimum lot widths have the same effect as minimum lot sizes: they ensure that lots are of a minimum size by imposing a "floor" on one of the dimensions of the lot. They also have the added effect of preventing irregularly-shaped lots which obstruct the vision of city planners who tend to desire a Euclidean layout––a.k.a. a grid.

Front, Side and Rear Setbacks (FS, SS, RS): Setbacks place a limit on the fraction of a lot which can be developed and ensure building separation. The first is effect is, again, to reduce the size and quantity of living space and the second effect is to achieve certain aesthetic qualities. Zoning authorities tend to like grass and insist that every zone (except one) require them. As an example, the SR3 zone in Lake Geneva requires minimum lots of 15,000 square feet and front, side and rear setbacks of 25, 10 and 30 feet, respectively. This ensures that no more than 50.6% of the lot are can be developed meaning the rest of the lot will be landscaped.

Maximum Building Heights (MBH): Maximum building heights place a "cap" on building height which also restricts the supply of housing. All else equal, a building twice as tall can house twice as many people. Since doubling the height of every residential building would double the supply of houses without consuming anymore land, zoning authorities put in place height restrictions to prevent property values from falling. To look at this in a different way, if every residential building were twice as tall, we could destroy half of the residential buildings and double the amount of landscaping. Yet this is never considered by zoning authorities who overwhelmingly desire traditional single-family homes as opposed to other kinds of dwellings.

Parking Space Requirements (PR): Parking space requirements, also called "minimum off-street parking requirements" cap the amount of living space as well. Parking spaces––whether in the form of a driveway or an attached or detached garage––count as developments. If we consider the Lake Geneva example from earlier, remember that only 50.6% of an SR3 lot can be developed. In SR3 each lot must have three parking spaces 157 square feet in size for a total of 470 square feet, reducing the maximum living space to just 47.5% of the lot. Parking requirements also have the effect of encouraging driving and what is know as "sprawl" by increasing the distance between locations and making pedestrianism more difficult. For commercial zones, parking requirements are even more intensive. A place like Walmart is required to have one parking space for every "300 square feet of gross floor area," effectively guaranteeing that a third of Walmart's lot is a parking lot. 

Maximum Dwelling Density (MGD): This regulation caps the number of dwellings units per acre within a zone. Again, this prevents residential construction by prohibiting higher-density development and encouraging sprawl. With respect to Lake Geneva's SR3 zone, the MGD is three dwellings per acre. Since SR3 only allows single-family homes there can be no more than three homes in a one acre SR3 zone. But remember: the minimum lot size in SR3 is 15,000 square feet and one acre is 43,560 square feet. That means that one acre of SR3 couldn't accommodate three single-family homes anyway––it can only accommodate two! So while the de jure limit is three homes per acre, the de facto limit is two. This is an incredibly common practice I have observed in my review of the zoning ordinances in Walworth County.

Minimum Landscaping Requirements (MLSR): Minimum landscaping requirements are another cap on development. Not only do they prevent development on lots (and typically fall within the range of 30-50%) but they also require that what is not developed is covered with "approved" foliage. Yes, approved foliage. As described for SR3, that means "The area of a site which is planted and continually maintained in vegetation, including grasses, flowers, herbs, garden plants, native or introduced ground covers, shrubs, bushes, and trees."

Maximum Coverage (MC): This regulation places a cap on the area of a lot which may be developed. It includes both the "primary" developments (like a house or business) and the "accessory" developments (like sheds, garages or parking spaces). Zoning codes often break this down into "Maximum Building Coverage" and "Maximum Accessory Building Coverage". Accessory coverage is usually limited to 10% and primary coverage is typically limited to 40% in residential zones. The effect of this in the SR3 zone, again, is to further limit the amount of living space. Originally the maximum lot development was 50.6% but was revised down to 47.5% after accounting for parking requirements. But now that the maximum building coverage (for primary buildings) has been specified, only 40% of the lot can be used for living space.

Floor Area Ratio (FAR): FAR requirements are the most stringent of lot regulations because they tend to be much lower than other caps placed on lots. As explained earlier, FAR requirements place a cap on the total floor area (square footage) of a primary building. It is "the gross horizontal areas of the several floors of a building, including interior balconies, mezzanines, basements and attached accessory buildings, fitting rooms, stairs, escalators, unenclosed porches, detached accessory buildings utilized as dead storage, heating and utility rooms, and inside off-street parking or loading space." In SR3 there's no FAR requirement for residential uses but there is for nonresidential uses. In the event that someone doesn't succeed in getting a lot rezoned they could purchase a lot in the SR3 zone and use it for certain nonresidential purposes. In that event, the lot now needs to be a minimum of 40,000 square feet (20,000 with a permit) and the FAR requirement is 10%. That means that a single-story development on this 40,000 square foot SR3 lot can be at most 4,000 square feet compared to the maximum 7,125 square feet if it is used as a residence. FAR requirements of this sort can severely limit living space in residential zones far below the maximums imposed by other lot regulations. And additionally, imposing FAR requirements on residentially zoned lots being used for nonresidential purposes discourages lot conversion by reducing the productive value of the alternative use. This discourages or prevents entirely market actors from effectively re-zoning lots and cartelizes land use.

A Regression of Single Family Home Prices on Lot Regulations in Lake Geneva and Williams Bay

*The following charts can be enlarged by clicking on the images. Alternatively, a .pdf of the charts can be viewed by clicking here.

* The details of the study will not be discussed at length and will instead be made available in a later publication.

While the theoretical explanations of the effects of Zoning and Lot regulations on housing prices are unassailable, I sought to demonstrate empirically the effect of Zoning and Lot regulations on housing prices in an attempt to quantify their impact on housing affordability.

To keep the scope of the project manageable and to control for differences that could bias the results, two municipalities in the same region surrounding Geneva Lake were investigated: the City of Lake Geneva and the Village of Williams Bay. These two municipalities were selected for several important reasons: the two are both positioned along the periphery of Geneva Lake which helps to control for price differences due to location and beauty factor premiums; the distribution and level of household incomes are roughly the same in each area; lot regulations were sufficiently varied to yield statistically significant effects; their zoning ordinances were detailed, timely and freely available; and my familiarity with each area made data cleaning both possible and manageable.

Parcel data was sourced from the Statewide Parcel Initiative and merged with the Zoning data provided by the municipal GIS analysts. Joining these data was conducted using QGIS. In some instances it was necessary to georeference the official zoning maps to assign zoning codes to parcels where that data was missing or incorrect. The merged parcel and zoning data were then exported to Excel, cleaned, and analyzed in R to produce a linear regression model.

Studies of this sort are incredibly difficult to conduct. This is especially true of economic studies employing econometric methods such as this one. Empirical studies which attempt to explain economic phenomena are notoriously difficult to perform and often produce results that contradict economic theory––even in situations where the results are known a priori. Empirical findings which contradict theory are primarily the result of two things: bad data and immeasurable economic phenomena.

In econometrics, immeasurable phenomena are referred to as "unexplained variables" or the "error term" of a model. Unlike the natural sciences and their study of individual particles, atoms and molecules––carefully observed and studied in a controlled laboratory environment––the human sciences and economics in particular cannot control for what could be an infinite number of variables affecting economic decision-making. Consequently, empirical economic studies cannot produce the same level of exactitude that is expected of the natural sciences. To paraphrase Friedrich Hayek, we must be content with more general knowledge that is certain, rather than aspire for specific knowledge which is likely to be incorrect.

That said, the results of this study are reliable to the extent that the direction and relative magnitude of the explanatory variables agrees with what we expect from economic theory. Furthermore, other studies conducted by much more competent and experienced econometricians corroborate these findings. No torture was necessary (or even possible) to produce these results.

Lake Geneva and Williams Bay

The chart below shows the regression statistics from a linear model of Lake Geneva and Williams Bay single-family housing prices regressed on a number of lot regulations. The model is a "log – semi-log" model.

The values with asterisks represent the coefficients of the variables; the asterisks (*) indicate levels of statistical significance; and values inside of parantheses (#) represent robust standard errors. Here is what these variables and their coefficients mean:

The first variable, "log(Acres)" uses lot size to predict home prices. As expected, its coefficient (.308) is both positive and highly significant(***). This makes economic sense: large lots, all else equal, ought to cost more money. Since "Acres" is "logged", a coefficient of .308 means that for every 1% increase in lot size, the price of a home increases by .308%. Another way of expressing this is to say that if the size of a lot is doubled, all else equal, the price of a home will increase by 30.8%.

Second, the variable "log(Minimum Lot Size)" uses minimum lot size requirements to predict home prices. As expected, the coefficient (.666) is both positive and highly significant. This makes economic sense: higher minimum lot sizes mean you must pay more to acquire a home. Since "Minimum Lot Size" is "logged", a coefficient of .666 means that for every 1% increase in minimum lot size requirements, the price of a home increases by .666%. Another way of expressing this is to say that if the minimum lot size is doubled, all else equal, the price of a home will increase by 66.6%. This is the second most powerful effect measured by the model and also falls in line with economic expectations.

Third, "log(Minimum Landscape Surface Ratio)" uses MLSR requirements to predict home prices. As expected, the coefficient (1..876) is both positive and significant. This makes economic sense: higher landscaping requirements would reduce the proportion of a lot that can be developed, decreasing the amount of livable space, thus decreasing the supply of housing and raising home prices. Since "MLSR" is "logged" a 1% increase in MLSR requirements would increase home prices by 1.876%. However, note that this variable is significant only at the .05 level and is likely explained by the many hundreds of "Estate Residential" single-family homes in each municipality. These tend to be lakefront properties with large lot sizes and large MLSR requirements to preserve the "estate" character of those respective lots. Additionally, remember that these results apply only to Lake Geneva and Williams Bay––not where you live.

Fourth are the "Maximum Density" variables. "Maximum Density" uses the number of dwellings allowed per acre to predict home prices. As expected, the coefficients are all negative and four of the "levels" are statistically significant. This makes economic sense: if more homes are allowed per acre then more homes can be built, increasing the supply and lowering prices. The maximum density variable is being treated as a "factor" to account for the difference in expected home prices given the maximum number of dwelling units allowed per acre. Each of these factor levels are being compared to the "base" level of density which is 0.7 dwellings per acre. So, for example, since the coefficient on "Maximum Density of 2du/acre" is -0.684, this means that if a home is in a residential zone with a maximum density of two dwellings per acre, its price will be 68.4% lower that it would be in the base level where maximum dwelling density is only 0.7. As can be seen below, homes in zoning districts with dwelling densities between two and four are all less expensive than homes in zoning districts where the maximum dwelling density is only 0.7. Home prices in zones with a maximum density of one isn't statistically different from those in zones where density is capped at 0.7, likely because the difference in density requirements isn't large enough to produce any variation in housing prices, but also because there were only seven observations in the base level. Maximum densities of six aren't statistically different from the base layer either, surprisingly. There were more than 500 observations in that factor category so the fact that prices differences do not register statistically will be reviewed before the final iteration of this study is published.

Lastly, at the bottom of the the chart there are two fields of note: "Observations" and "Adjusted R^2". This study examined 2,920 single-family lots in Lake Geneva and Williams Bay and the R^2 (r squared) value reports the "goodness" of the model. With a value of 0.530 this model explains 53% of the variance of housing prices in these two municipalities without looking at anything other than Zoning and Lot regulations.

The significance of this number is that roughly 53% of the price of homes in these two places is explained by lot size and only three lot regulations. Further studies would do better to control for the square footage of homes, the number of bedrooms and bathrooms, the year the home was built, and other measures of housing quality. Additionally, since these two areas are popular tourist destinations due to their proximity to Geneva Lake, a better study would also control for homes' proximity to the lake. 

As for the charts below, these are the regression results for the two municipalities individually. I have reported the best fitting models for each of these separately so the reader may see which factors most affect prices for single family homes in each locale. The reader will remember that logged variables are interpreted as a percent while non-logged variables are multiplied by 100 and then interpreted as a percent.