I don't insure my car. Well, I have liability insurance, but I don't insure the car itself. So if I drive it into a ditch, or it gets stolen, I have to pay to repair or replace it, out of my own pocket.
Why did I take this decision? I could have estimated the probability of it getting stolen, or my driving into a ditch, estimated the cost of repairs or replacement, and done a very complicated calculation of my expected utility with and without insurance, and chosen whichever option gives me the highest expected utility.
But I didn't. Instead I assumed that the market for collision insurance was reasonably efficient, made a couple of judgment calls, and relied on economic theory.
I don't have a clue what the probability is of my driving into a ditch or my car getting stolen. I couldn't even hazard a guess and expect to get it within the right order of magnitude. Or rather, it would take me hours of searching to get a reasonable estimate. I figure my insurance company knows the probability much better than I do.
"Actuarily fair" insurance means that the premium equals the expected payout. So if there is a 1% probability per year of my car getting stolen, and the payout would be $3,000 if it does get stolen, the premium would be $30 per year.
I figure that even if I shopped around, and got the best deal, any insurance I could get would have a higher than actuarily fair premium. The insurance company has to cover its administrative costs. Plus there's the moral hazard problem of people who have insurance against theft taking less care to prevent theft. Plus the adverse selection problem of people who have the higher risk of theft tending to get insurance against theft.
With collision insurance, there's an additional argument against insurance. If I drive into a ditch, the insurance company would have to pay for repairs that would make the car as good as it was before the accident. I would rather have the cash, and either fix the car myself, or just leave a few dents in it. In fact, taking half the cash would still probably leave me better off. The premiums must be more than double what they would be for insurance that would be actuarily fair from my perspective.
My car is probably worth around $3,000. That's much less than 1% of my total wealth (pension plan, house, human capital, etc.). It's peanuts. My utility function is very close to linear over that small a range. Anyone with a linear utility function would be indifferent to taking an actuarily fair bet. My utility function is a lot closer to linear over a $3,000 range than insurance would be close to actuarily fair.
So I don't insure against small losses. I only insure against big losses, where my utility function would be a long way from linear.
I don't do any calculations in deciding whether or not to purchase insurance. I rely on economic theory, plus the assumption that the insurance market is reasonably efficient, except for the inefficiencies I can identify using economic theory.
There's just a chance, of course, that I am totally wrong. Maybe, just maybe, there is some insurance company out there that is willing to offer me premiums that are so low that I would be better off getting theft and collision insurance. But I doubt it. Insurance companies are unlikely to be that stupid, and they have much better data than I do. I trust their judgment of the probabilities more than I do my own.
I don't trust the housing market to be as efficient as the insurance market. Insurance premiums are set by companies that specialise in providing insurance, in competition with other companies, and they have a big incentive to get their estimates right (at least for things like car insurance). They know a lot more than I do. So I don't even try to second-guess the insurance companies on whether insurance premiums are too high or too low. But house prices are set by people like me, or even more ignorant than me. I flatter myself that I can do a better job of estimating whether house prices are too low or too high than most people. So I would compare house prices to the Net Present Value of rents before deciding on whether to buy or rent a house.
But even if I am smarter than the average person buying or selling a house, there's a lot more of them than there are of me. I would be unwise to ignore the "wisdom of crowds" altogether. And while I recognise that other people can be collectively wrong, getting caught up in manias or panics, I have to recognise that i can get caught in the same manias and panics too. I ought to hedge my bets, and rely partly on my own estimates, and partly on what economic theory tells us should happen if the housing market is efficient.
If everybody else were rational, and were doing their own estimates, and if house prices reflected those rational decisions, would it make sense for me to own a house? What sort of people should own a house in equilibrium? Who should be a tenant? Who should own their own home? Who should be a landlord?
First off, it doesn't make sense to say that nobody should own a house, and everybody should rent. Nor does it make sense to say that everybody should own two houses and rent one out. Since every house must be owned by someone, and every family needs a home, in equilibrium the average family should own the average number of homes. One.
The average family should own one home in equilibrium. But there's a strong argument that each family should own one home too. And it should be the home they live in.
As I argued in a previous post, we are born with a short position in housing. We need somewhere to live, and so are born with a liability of the net present value of the rents that we will need to pay. Having a short position is risky. Buying a house is like covering that innate short position. We are insuring ourselves against the risk of rents rising or falling, and making our future utility less uncertain. Buying two houses is like taking a long position in housing. And that's risky too.
And if you own a house, you need a tenant to live in it. The best tenant would be one you know well, who would be reliable in paying you the rent, and who would treat the house as if it were his own. Hmmm. What better tenant than you yourself? Being your own landlord, and own tenant, solves the principal-agent problem between landlord and tenant.
Plus, you won't get into any arguments about what colour paint and wallpaper, or whether to upgrade the insulation or fix the roof, if the tenant and landlord are the same person. All the externalities between landlord and tenant are internalised; all the transactions costs of negotiations between landlord and tenant are eliminated.
And, there's the tax angle. You don't pay income tax on the rents you pay yourself. Nor capital gains on your own home.
It is easy to imagine an equilibrium where every family owns just one home, and lives in it. So how could we explain an equilibrium with tenants and landlords?
Don't say "But some people can't afford to buy a house!". That sort of statement just drives economists crazy. In equilibrium, the price of a house should equal the Net Present Value of the rents minus taxes, repairs, etc. If you can afford the rents, you can afford to buy the house and pay the taxes and repairs etc. Sure, you can't afford to buy too expensive a house; but then you can't afford to rent too expensive a house either. That's got nothing to do with the rent vs. buy decision.
There is however some truth in the "people who can't afford to buy should rent" argument. If you don't have the liquid assets to use as a down-payment, and so have no equity in the house, you don't really own it. You are renting it off the mortgage company. Only it's a tenancy you are locked into. If your income falls, and you want to rent somewhere cheaper, you have to sell the house and repay the mortgage to get out of the tenancy. And if the price of your house falls below the value of the mortgage, so you have negative equity, you have to pay the difference to the mortgage company to get out of the lease.
House prices are uncertain, and your future desired rents are uncertain too. If there's a correlation between falling house prices and your need to downsize, then buying a house is risky. The greater that risk, in proportion to your wealth, the better off you would be as a tenant.
The other people who should rent rather than buy are people who plan to move soon. There are significant transactions costs in selling one house and buying another. Agent's commissions, lawyer's fees, home inspection, taxes, plus the sheer hassle of showing a house and waiting for the right price, may well rise to 10% of the value of the house. (Costs of moving furniture don't count, since renters face those too.) If you move every two years, that reduces the rate of return on owning a house by 5% per year. If interest rates are 5%, that's equivalent to doubling your cost of borrowed funds, or earning 0% on your own funds. People who plan to move soon should rent, not buy.
What other individual characteristics have I missed?
Worthwhile Canadian Initiative 
Perhaps people who know they will be reckless with their home should rent. If you know you’re way more reckless a driver than the average joe and are super likely to drive your car into a tree, you should probably buy the car insurance. Similarly, if you love to throw parties and hate home maintenance, you may be better off renting.
Perhaps people in tenuous relationships should rent too. I guess that kind of ties into your “people who plan to move soon” should rent argument. But houses are large assets that are not easily divisible. So if you’re searching for property with a life partner, it’s a lot easier to move out of a rented place than split up a house if the marriage goes sour. Maybe Tiger Woods should have been a renter.
David: Yes. I agree on both. The “home maintenance” angle deserves a category to itself. People who have a comparative advantage in home repairs should own. People who don’t should rent, because the landlord probably does have a comparative advantage in home repairs, or can exploit economies of scale in hiring outside contractors.
I read on some blogs (Garth Turner?) that it’s usually the woman in a relationship who wants to buy rather than rent. If it’s true that women are also the ones who want men to “commit” (Roissy’s blog, ahem), then it all fits in. Home ownership is the signal of strategic investment in a relationship?
Maybe people should rent instead of buy if they can’t borrow at a decent interest rate. Welfare recipients, for instance: suppose they’re renting a $300 apartment on their $600 welfare cheque. Even if they could find a similar apartment for sale, if they couldn’t get a loan from anyone other than a loan shark, it isn’t worth it. The interest payments alone would be more than $300.
In the National Post every week, there’s a list of selected real estate transactions. You see buildings in Toronto selling for less than $100K a unit. But are there any actual condo units in Toronto for less than $100,000? Probably the people who can only afford to live in a $100K apartment aren’t good enough credit risks to buy one, so there’s no market for such things.
Phil: agreed. Borrowing rates are usually higher than lending rates, so the opportunity cost of funds for people who can buy a house outright is lower than for those who must borrow. And the lower the downpayment the higher the interest rate. and as your credit risk worsens, the interest rate gets higher still.
“Don’t say “But some people can’t afford to buy a house!”. That sort of statement just drives economists crazy. In equilibrium, the price of a house should equal the Net Present Value of the rents minus taxes, repairs, etc. If you can afford the rents, you can afford to buy the house and pay the taxes and repairs etc. Sure, you can’t afford to buy too expensive a house; but then you can’t afford to rent too expensive a house either. ”
Houses are idiosyncratic assets — they are not substitutable and they are not shortable. What this means is that the price of a house is determined by the highest bid in each auction. Therefore “equilibrium” prices of houses are not equal to the net present value of rents, as estimated by the neighborhood as a whole, but by the net present value estimate of future rents — that is future incomes — as estimated by the most optimistic bidders within the neighborhood.
You can only ignore this effect if you believe that everyone has exactly the same opinion about future incomes and yields.
This is a fundamental difference that biases houses to be on the expensive side vis-a-vis rents, even in the scenario of perfect future foresight with a log-normally distributed error term, as only the positive biases affect house prices, whereas the negative biases do not. This is unlike financial assets.
So the only reason to buy a house — if you believe in this rational expectations with error model — is for personal non-financial reasons.
There is also the information gap between the buyer and seller of the house, as in the market for lemons. One would presume that homes that require major repairs are overrepresented among homes for sale, and despite home inspections, these problems may not be discovered until it is too late.
What about the fed having low interest rates and lax regulation to allow financial traders to speculate in housing and also hope that people will borrow against their rising home values to spend on a lot of chinese imports?
“What about the fed having low interest rates and lax regulation ..”
Now you are talking about the real world, Too Much Fed — we were talking about equilibrium valuation models, that were independent of financing and budget constraints.
In the real world, a household can lever 10:1 (and now, with 3.5% FHA loans, they can lever 28:1) and not be subject to real-time mark-to-market margin calls. Moreover, they can walk away from the debt if the trade goes against them, as mortgage debt is typically non-recourse.
So little Joe Schmoe real estate investor gets huge leverage with an asymmetric payoff scheme and avoidance of mark-to-market requirements. Think of him as a mini-banker, doing his part to destroy the financial system, one “American Dream” at a time. (Btw — is there a “Canadian Dream?” — does it involve leverage?)
That is enough reason to drive up house prices way beyond net present value of rents, and still make buying a house a good trade, due to the embedded call option.
Q: Btw — is there a “Canadian Dream?”
A: Yes, it invlolves the Maple Leafs winning the Stanley Cup.
Q: — does it involve leverage?
A: It doesn’t involve leverage but would probably require it.
With car insurance aren’t you ignoring the no-claim bonus angle? Having a history of having paid insurance and not claimed makes it cheaper for you (and in some countries) for other family members (like children). For them a car (and maybe you will someday buy a new car) is not peanuts.
RSJ: I think your argument is an example of the “winner’s curse” argument. The most optimistic bidder wins the auction, and pays a price close to his own valuation (depends on the form of the auction). If the bidders on average are unbiased, the winner on average will be biased upwards, and (depending on the form of auction), the price paid may be biased upwards.
It’s an interesting argument, and not obviously wrong. But I’m not sure it’s right either.
1. If people knew about the winner’s curse, they would take it into account under rational expectations, and bid lower than their personal expectation of the value of the asset.
2. I’m not sure it applies to housing anyway. The sellers’ expectations ought to influence price as much as buyers’, and the sellers would be the most pessimistic. And if everyone buys one house in equilibrium, are there really multiple bidders?
Dunno. I haven’t been following the winner’s curse literature to its conclusion (if it ever reached one).
Andrew F: agreed. The market for lemons adds an additional transactions cost on buying and selling a house. Relevant for people who move often.
RSJ: nearly all Canadian mortgages are recourse. You can’t just walk away from a negative equity house. The bank will come after you. But yes, this makes a big Canada vs US difference.
Too much Fed: you are describing the mania phase, where houses were overvalued. Maybe now the US is in the panic phase, where houses are undervalued. And Canada is different again.
Morning laugh from Adam!
reason: yes, I was ignoring the no claims reputation angle. But, if I (or the kids) wanted to buy that reputation, wouldn’t it be cheaper just to wait till we needed it, and buy it then? Plus, if you ever do make a claim, your future premiums often rise. So it isn’t really full insurance. The increased premiums should be added to the deductible. When you get to the point that it’s not worthwhile making a claim, you wonder what it is you are actually buying?
“mortgage debt is typically non-recourse”
Not in Canada it isn’t. In Canada the lender can sue to be made whole.
I beat you to the draw, Patrick!
Great discussion so far. I’m going to call the insurance company right now and cancel my collision.
As an aside: most rickshaw drivers don’t own their own vehicles, but instead rent them at extremely high rates. Microcredit agencies are promoting rickshaw ownership as a development strategy, but some are running into precisely the problems mentioned above – it’s very risky for a poor person to own a rickshaw.
In Canada and the US we’ve seen a few trends in the last few years:
– a big increase in (especially male) individual income inequality
– smaller family sizes and more “assortative mating” increasing family income inequality
– more instability in the labour market/shorter job tenures
– stagnant or declining wages at the bottom of the earnings distribution (trade in a good factory job for work as a greeter at Walmart)
So wouldn’t we expect a booming rental market as it’s too risky for people to buy? Is the housing boom driven entirely by the escalating incomes at the top end of the distribution?
I think lenders typically require you to carry collision insurance while you are paying off the loan.
Given that there’s no walking away in Canada, it’d be interesting to know what the relationship between foreclosures and personal bankruptcies is in Canada. I suspect most people would be forced into bankruptcy if they were sued for default. I also wonder how often lenders actually sue. Legal proceedings are expensive and time consuming, so I doubt lenders sue if they can just take the house and sell it for close to what they’re owed.
Adam P: by ‘leverage’, do you mean leverage with the Almighty? 40+ years of prayer doesn’t seem to have done any good.
“Q: Btw — is there a “Canadian Dream?”
A: Yes, it invlolves the Maple Leafs winning the Stanley Cup. ”
Woah, emphasis on the word “dream”. BTW, I know Leafs fans all turn off Hockey Night in Canada after the Leafs lose the early game every Saturday, but there are actually other Canadian teams out there. In fact, the Flames were “Canada’s Team” in 2004 and the Oilers were in 2006, but I don’t remember the last time the Leafs did much worth getting excited about.
As for the post, don’t insurance premiums usually increase after an accident? It seems like that would strengthen your argument about not getting insurance by making it cost even more than the expected value of the payouts.
miko, point well taken. How about Olympic Hockey Gold in Vancouver for the Canadian Dream?
We are born short only if population is increasing and if it is not we will inherit one or have to acquire one from someone without heirs. While population is increasing here, it is with retirees and stabilizing overall, though this will vary with country and location. Generally the young should rent and the old should own and the poor have to rent and the rich have to own, if not directly, at least through their investments.
Back to insurance bit for a minute — this is the same argument that says no reasonably affluent person should ever buy an extended warranty. Good rule of thumb.
It’s pretty easy to calculate what portion of an insurance premium is transaction cost, especially in a mutual. They just add up the premiums that come in, subtract what gets paid out in claims and premium refund (in a mutual), the difference is the TC. I remember reading that this typically runs about 30% in the standard categories (home, auto, life), but I can’t find the ref. Anyhow, I use this number whenever I’m contemplating buying insurance. Your example picks up two poles: a $3000 car (no brainer, don’t insure), and (probably) $1 million in liability (no brainer, insure). But there are a lot of intermediate cases, where you need more than just economic theory, but some information about both your own utility function and the product you’re buying. I have a two-step decision procedure:
1. You are pooling your money with a bunch of other people. Who are these people? Are they more likely to get a payout than you? (By the way, this consideration applies to things like eating at an all-you-can eat buffet as well, which is why I never do. Adverse selection and all… I know when I’m outgunned.) Getting an actuarially fair premium is the best-case scenario here (experience-rated car insurance is probably the only time you’ll get anything close to it, usually there’s huge cross-subsidization).
(As an aside, in Ontario you get something closer to actuarial fairness than in provinces like Alberta that prohibit gender-based risk-classification in car insurance. As a man you’d be benefiting from cross-subsidization in those jurisdictions, so you’d have a more complex decision.)
2. Two-thirds of what everyone pays goes into the pool, one-third covers the costs of organizing the pool. So think of the two-thirds as the money that may or may not come back to you, but the one-third as the price tag. Is it still worthwhile?
For anyone earning a professor’s salary, facing no unemployment risk, declining insurance is a pretty easy decision in most cases. Employment benefits are a bit trickier (i.e. when there is an employer co-payment). Then the 30% number is handy. The only one I decline is the “vision insurance” that my university offers, on the basis that it can’t be a real insurance pool — people know if they need glasses, and will obviously sign up for it only if they do, so one would expect the rate of claims to be 100%, making the transaction cost issue decisive.
The only big-ticket one I’ve opted out of is that I won’t let my wife buy life insurance, even though she’s the big earner in the family. The rationale is that self-insurance is still more attractive — push comes to shove, the family could live on just my salary without a huge lifestyle adjustment.
Mortgage insurance is an interesting variant of life insurance. My suspicion is that it’s a bad deal, one that exploits people’s tendency to think in overly concrete terms.
With your car: the liability insurance is obviously the one that’s worth buying. But I should point out, for those of us not driving old beaters, you can achieve something fairly similar just by taking standard insurance with the maximum deductible. In my case it’s $1000, so that’s what I have. In my experience you get a huge reduction in premium for that, I suspect partly because of what it signals — that you don’t intend to make nuisance claims. I didn’t know you could buy car insurance with literally no collision protection. I wonder if one can disaggregate home insurance policies as well? I’ve heard it said that one should always max out available deductibles (again, assuming reasonable affluence).
Oh yeah, and step 3 is to remember that I suffer from optimism bias and hyperbolic discounting, and repeat steps 1 and 2 in the light of that.
Nick’s post said: “RSJ: nearly all Canadian mortgages are recourse. You can’t just walk away from a negative equity house. The bank will come after you. But yes, this makes a big Canada vs US difference.
Too much Fed: you are describing the mania phase, where houses were overvalued. Maybe now the US is in the panic phase, where houses are undervalued. And Canada is different again.”
Didn’t know that about most Canadian mortgages. Learn something new every day. What if someone sells a negative equity house (short sale)? Does recourse still apply?
With the housing tax credits and other means, I’d say the “panic” phase has been drawn out over time (some people call it extend and pretend). Calculated Risk has good posts about the lower/lowest prices homes being near their price bottom. Higher/highest price homes near their price bottom? Maybe not?
Lord: the fact that we need somewhere to live is what gives us our innate short position in housing. If we inherit a house, our parents cover that short position for us. It doesn’t have anything to do with population growth, though that will indeed increase the demand for housing.
Joseph:
Yep, same argument for not buying extended warranties.
For some risks, yes, you could get a good measure of how far off insurance is from actuarily fair by comparing premiums to payouts, relying on the Law of Large Numbers. I hadn’t thought of that.
Dental and vision (eyeglasses) insurance also looks very undesirable, as you say. When dental is covering $1,000 per year, and eyeglasses $200, it is indeed ridiculous for university profs to get those in the benefits package. Which makes me wonder why we do get them? Some sort of union bias? Some tax angle? Bias towards kids??
Can anyone explain it?
“Mortgage insurance is an interesting variant of life insurance.” You mean insurance that pays off your mortgage if you die?
Yes, choosing a big deductible is another way of getting what I am arguing for. But I can never find a big enough deductible.
I was arguing the other day that Carleton should have a $1 million deductible on all its insurance, at least. About 0.2% of its annual revenues.
I’m in Quebec, so that might make a difference to what insurance is available. I remember when I lived in California, in the 1970’s. You didn’t need auto insurance at all, including liability insurance, if you could just post a large enough bond.
Too much Fed: I don’t know the rules on short sales. I don’t know if the bank can stop you selling for less than the mortgage. But you still owe the bank the full amount.
Great post Nick! Anybody with sufficient cash reserves should think about self-insuring for small, manageable events.
Landlords tend to underinvest in energy efficiency enhancing improvements because tenants usually pay all energy costs. These kinds of improvements can impact comfort. Owned homes and rented homes are not identical.
BTW, is everybody aware that it is not possible to disentangle property insurance from liability insurance for tenants? They come bundled. So guess what happens? Apparently, very few tenants buy home insurance. Good? Bad? Indifferent? I dunno.
I would guess that there is some adverse selection motive for bundling the two.
Nick @8:24:
Yes, you can call this a “winner’s curse” argument, or The Law of Demand — they are equivalent. The point is that in a heterogeneous population in which people have different utility functions for a good, the good will go to the one who values it most. If there are N identical goods, it will go to the Nth most eager buyers, and the price will be set by the marginal purchaser. The statement that demand curves slope downward is then equivalent to the statement of the winner’s curse.
In the same way, although for everyone to buy a car, someone else needs to sell a car, these are different groups of people with different cost structures (and utility functions), so it is possible for demand to increase but not for supply to increase, etc.
Now what happens when the utility is an expectation of future returns (in which case it is a number)? At what point will the clearing price be equal to the “mean” expectation? You need 1) a large number of substitutable goods, 2) require that a shift in the demand curve occurs simultaneously with an equivalent shift in the supply curve, so that the two always intersect in the exact middle, and 3) assume that both curves are linear so that the intersection point corresponds to the mean (and not just the median utility).
So there is no equilibrium in which expectations for houses are such that the clearing price is equal to the mean of the expectations for the market as a whole. And as you point out, we are all in the housing market.
With financial assets these conditions are (generally) met, modulo frictions. People tend to have similar cost of capital structures (the opportunity cost, as discovered ex-post, or as measured by the risk free rate ex-ante). The fact that you 1) only care about cash returns and 2) can short one asset and use the proceeds to buy another means that cash-flows are substitutable for each other, so you can pretend (ideally) that there is just a single market of cash-flows directly comparable by their NPV. In this case, as every buyer of one financial asset must first obtain the funds to do so by selling some other financial asset of equal value, and as everyone has similar risk-adjusted cost of capital, the supply and demand curves always move together, and therefore the situation is indistinguishable from a single horizontal curve, whose height is the mean of the expected return for the market as a whole.
What, from all of this, is missing from housing? The incomparability of housing, leading to an inability to short. If houses were perfectly substitutable with each other, anyone could “borrow” a house and then sell it to someone else, with a promise to replace the borrowed house by an identical one later on. But different houses are not identical, and therefore you cannot short. As you point out, the seller can short by selling their own house, but this is insufficient to guarantee convergence to mean expectation of the market as a whole — you need the entire market to be able to short the asset, not just the current owner, or a subset of participants. As it is, the owners will themselves be the winners of the previous auction, and so they will have similar biases to the current stock of winners in equilibrium. In order to prevent these types of equilibria, everyone needs to be able to short a house, whether they own one or not. And renting (or withdrawing from the purchase market) does not have the same effect on auctions as both refusing to purchase and adding to supply by shorting.
Therefore houses can (and do) remain overvalued for long periods of time, whereas they will remain undervalued for brief periods of time, as there are plenty of arbitrage opportunities to go long housing. Therefore in aggregate, you would expect housing to be overvalued in general, at least if your definition of “overvalued” is an average price greater than the mean expectation of equivalent rental returns.
But this does not mean that houses are overvalued using the valuation approaches used in the goods markets — i.e. prices are set by the marginal purchaser to reflect their marginal utility, which is what they believe the house is worth. But with goods, utility functions are not comparable, and there is no notion of having your prediction turn out to be false (i.e. discovering that you did not have a certain utility ex-post). So housing is at the intersection of requiring expectations to set prices, yet the prices are set by the marginal expectation of a sloping demand curve rather than the mean expectation of the market as a whole.
P.S.
Glad to hear loans are recourse in Canada.
I think you might be mistaken that one cannot short a house. You can short a house you do not own by renting it from a landlord with an option to buy at any point in the future at the then current market price (using say, a second-bid auction, appraisal or whatever), and ‘sell’ it to a counterparty. As part of the sale agreement, you are required to hold a certain margin in trust equal to at least the market value of the house less the sale price to the counterparty. Using the option, at any time you can purchase the house at market value and deliver it to the counterparty.
You take the sale price of the house up-front, pay rent to the landlord, and give the counterparty the benefit of the implied rent/service of the house. Unless I am mistaken, this is directly analogous to shorting an equity position, where the exchange will allow you to sell a share you do not possess using a cash margin, provided you provide to the counterparty all the benefits of owning a ‘real’ share, such as dividend payments. At any time, you have the right to cover the short position by purchasing a share at market value and delivering it to the counterparty.
So, there is nothing theoretically special about housing that does not allow one to short it, except regulation where it applies. Obviously, in practice the transaction cost is higher, because houses are not fungible like shares in a firm.
Andrew,
That is not shorting, because you are not increasing the number of units offered for sale, but merely transferring an offer of ownership extended to you to someone else. But in aggregate, you are not bringing more inventory onto the market, since in this case the total number of houses available for sale is constrained by the total number of people who own houses but do not want to live in them.
When you short, you can increase the supply of instruments available for sale to equal the total number of instruments that were ever issued, or close to this amount (actually, in excess of this amount as market makers can write naked positions, but this is offset by the fact that instruments in cash accounts cannot be shorted. In reality, there are very instruments in cash accounts).
In any case, if there are X issues of IBM stock outstanding, then (close to) X issues can be shorted at any time. This is not the case with housing. Only those houses that are not owner occupied — a small minority — can be shorted. Moreover, the change in price of the landlord-owned housing stock does not translate into a change in price for owner occupied stock, since each house is unique and houses are not substitutable.
Thanks westslope!
“Landlords tend to underinvest in energy efficiency enhancing improvements because tenants usually pay all energy costs.”
This looks to me like another principal-agent problem of renting, only in reverse. The landlord is now agent, and the tenant is the principal. The landlord has insufficient incentive to take proper care of the tenant’s heating bill.
I didn’t know that tenants’ liability insurance and property insurance were bundled. If I were a tenant, I would want liability, but not property insurance.
RSJ: Interesting and unorthodox as always!
But I would disagree with you on one point. People are different in their preferences, and their beliefs. I think it is important to distinguish the two. Your beliefs can be wrong in a way that your preferences cannot.
(And differences betwen people are only one reason why demand curves slope down. Even if everyone were identical, demand curves would normally slope down. I demand more if the price falls.)
By the way, is what you are arguing about with Andrew the same as the difference between covered and naked shorting? I’m not sure.
Nick asked: Why do profs get insurance in their benefits packages? Taxes – employer contributions aren’t taxed (the employer pays over half the cost of my plans).
From a policy point of view, subsidizing employer-provided health insurance isn’t a totally dumb idea, since adverse selection plus the eyeglasses problem (you can’t insure against certain events) means private health insurance markets don’t work very well. Pooling through an employer solves adverse selection to some extent.
Perhaps the university figures the three-times-a-year cleanings make us more productive people?
Unions? Insurance is a fixed per person benefit, so tends to equalize the salary distribution – so, sure, unions would favour coverage. But I think coverage predates university unions.
Thanks Frances. So if university profs pay about 50% marginal tax rate, and the employer pays half, tax-free, that means a 25% tax savings. Maybe just covers the transactions cost. Half the time I lose the accursed receipts anyway, and can’t figure out how to do the claim, or forget. There’s also our transactions costs.
Just an example of the deadweight costs of auto collision insurance: a couple of months ago someone reversed into my daughter’s car. He didn’t want his insurance rates to go up, so we negotiated a deal. The repair shop quoted $3,000. (Which was legitimate, but also ridiculous on a car worth about $4,000). Plus, I didn’t want the hassle of her car being out of action in the shop, plus taking it back and forth. I asked him for $1,500, which he gave me. Bought parts from a wreckers for a couple of hundred. Spent a fun day fixing it. It looks very nearly as good as before. He saved (at least) $1,500 (because he was prepared to pay me the $3,000). I was better off too. Total deadweight cost of insurance, around $2,000, or maybe 2/3 of the expected premium.
Frances: In the case of dental coverage I think insurance companies would rather ‘cover’ predictable events like cleanings and charge a fee for managing your cash flow for you than to really provide insurance against major events that are both unpredictable and potentially expensive.
I once tried to find an insurance company that offered real insurance for dental (wife’s family has unfortunate genetics where teeth are concerned), as opposed to cash flow management for predictable expenses, and I couldn’t. It simply wasn’t offered.
I live in Montreal, and I suspect the Montreal certain aspects of the Montreal real estate and rental markets make it very difficult to compare housing prices with rental values, and make rental values a poor guide to housing prices.
For one thing, Montreals real estate market is differnt from most cities in one aspect: we dont seem to have a large market for detached rental houses. I may be wrong, but my impression is that house rentals are rare, and usually occur when an executive or professional needs to live out of town for a year or two (for example a university professor on sabatical) and for sentimental reasons don’t want to sell their house, so they try to rent it for a year (maybe to another professor on sabatical). In such a case, the house is not rented to the highest bidder, but rather to someone the home-owner trusts will not damage their property, so the rents would tend to be lower than expected.
Furthermore, in Montreal, most single detached homes are owner-occupied, most older (pre 1970) apartments are rented, and the newer ones (post 1980) tend to be condos. We also have a large market of “duplexes”, which in Montreal is a two-story building with one apartment on the ground floor and another on the second floor. Often the owner lives in one apartment and rents the other one. As in house rentals, choosing a quiet reliable responsible tenant is more important than maximising rental income.
For these reasons, I think it’s much more difficult to use the cost of renting versus the cost of buying in Montreal as a tool to analyse real estate prices than it would be in other cities where you can more easily compare the cost of buying versus renting a 30-year old suburban bungalow.
Nick,
I think you are spot on here, especially about the efficiency of the housing market versus the efficiency of insurance. I’m actually baffled to read people arguing as if the EMH holds in the housing market. There’s a long post someone on the blogosphere needs to right about how the heterogeneity of housing, transaction costs, and local knowledge, and the thinness of the market, all prevent the housing market from being efficient.
And I’m glad to read that somebody agrees with my “be your own landlord” metaphor! There’s another metaphor to be had that buying a home is really just vertical integration, and as you’ve pointed out, there are many efficiencies to be had in that vertical integration. I’ve only got so much energy left to write about this issue though, so I’m not going to bother with another clever metaphor.
Thanks Adam. Reading your post inspired me to write this. Metaphors can be so useful. “Be your own landlord”. Or, “Be your own tenant”?
It’s almost as though Felix Salmon were arguing that it never makes sense for anyone ever to own a house. Which can’t make sense. We can’t all be tenants, and own zero houses. We can’t all be landlords and own two houses either. There might be times when house prices are everywhere far too high, so it makes sense for almost nobody to buy. But there are presumably also times when house prices are far too low, so it makes sense for almost everyone to own 2 houses.
I’m not sure about EMH applied to the housing market. I’m not even sure how one would define EMH, given idiosyncratic houses and idiosyncratic demands for places to live. But I don’t think I would ignore the judgment of others either. Maybe they know something I don’t, and what they know is reflected in market prices.
Even in a world of second-best transaction cost constrained alternative equilibria, markets clear, and equilibrium is still obtained.
“is what you are arguing about with Andrew the same as the difference between covered and naked shorting?”
The arrangement that Andrew described was buying a call option (e.g. the lease to own arrangement), exercising it, and then selling to someone else (e.g. a covered short), as a sale occurs only if title is transferred, so the call needed to be exercised prior to the sale.
With financial securities, you can cover your short position by purchasing some other share that is indistinguishable from the one you sold, and can cover at a later time. This delay in covering allows supply (as a flow) to expand via shorting, just as the flow of spending increases when you borrow. This must happen if supply and demand always move together, and this cannot happen with non-fungible assets.
As an aside, businesses could form a “central bank” in which they would agree to issue more stock or repurchase stock in order to maintain a target short margin rate. They could have a discount window or overdraft facility allowing them to loan shares to short-sellers directly. As the monopoly issuers of equity, they can control the own rate of equity. If they were to do this, they would be about as effective at spurring investment and controlling yields (e.g. price/dividend) as the central bank is effective at spurring investment and controlling the cost of capital.
The only effect this would have would be to prevent short squeezes, and there would be a reduction in the relative price of equity with other financial assets (including money) as the risk of a liquidity crisis in equity has been removed. But other than this one-time reduction, market participants would still short one stock and buy another asset only if they thought the cash-flow characteristics of the second was superior to the first. And the long-run return from equity would remain the same as the long run nominal GDP growth rate, as it has been for the last 150 years, regardless of interbank rates or short margin rates.
But back to marginal vs. mean, I thought of a simpler way to see the difference:
When you are talking about shares of IBM, then anyone with an opinion about IBM, and a financial portfolio that they are willing to risk on this opinion, is able to either buy or sell shares of IBM, thus moving the price. In this way, you can argue that the share price of IBM converges to the (asset weighted) mean opinion about IBM.
When you are talking about 1 Market Plaza, only the owner of 1 Market Plaza can sell it. They have much more influence than just anyone with an opinion. So the sale price will converge to the opinion of the most eager buyer and the current owner — the marginal opinion, not the mean opinion.
One effect of this distinction is to look at what happens in market crashes: in the housing market transactions decline sharply, whereas in the financial markets, transactions increase.
In the housing market, you can almost see the utility of house buyers adjusting differently from that of house sellers, leading to an independent move of the demand curve, causing a reduction in quantity transacted. During booms, the opposite occurs. Even though most transactions are for used houses, still there is a strong correlation between gross sales and net sales (e.g. transaction volume and final housing output).
But when there is a financial market crash, new perceptions are formed and there is a rush to make new predictions. Unlike the goods markets; increased volume of financial transactions does not lead to an increase in net new issues, and there is no correlation between the number of transactions and final output or national income.
I’ve skimmed the comments, so maybe I missed this, but the auto market in Ontario is not a profitable line of business. Try this (pdf): http://bit.ly/5XlsO2 http://bit.ly/6UQdxj (it shows the dismal state of Ontario auto on slides 21 and 23 – granted it includes more than physical damage)
Ontario auto is unprofitable because the government regulates the bejesus out of it and forces statutory rates on insurers, especially for Accident Benefits coverage.
Canada’s weird place in the international insurance universe is striking: companies here routinely run small losses on their books(by that I mean actuarially sound rates are rarely achieved). Why so competitive? Because right now we’re sitting in the fat end of the insurance cycle where international insurers (especially Bermudians) are flush with cash and trying to expand. Many have placed bets in Canada and are losing on those bets.
That might make the market more efficient, but it doesn’t mean that you shouldn’t screw your friendly local insurance company over. Besides, Nick, you’re understating the risk to yourself by piling your illiquid assets together and comparing them to the value of your car. Insurance protects against liquidity events. If you total your car, you’re more likely to need a new car faster than you can sell your house or cash in your pension to buy a new one, right? You’re better off comparing the replacement value of your car (more than 3k?) to your liquid savings + line of credit facilities and associated future income to pay it off.
With a stable population, the only way we can be born short housing on average is if we die long housing on average or at least for some period of our lives. For everyone that rents, someone else owns, so for everyone that is short, someone else is long. Society as a whole can only be short if growing and can only be long if declining, neglecting vacancies. If we are born short and only rent, we never cover and never have to cover. If we are born short, buy but never pay it off, can we really say we ever cover it, or perhaps covered at the cost of uncovering a payment stream. If we are born short, buy, and sell in retirement, covering is only temporary, though it may always be considered temporary since since life is temporary. If we are born short, buy, pay it off, and payand revers mortgage in retirement, we have covered and covered the payment stream, only to uncover the stream in retirement. If someone goes long with an income stream covering a possible payment stream, the renter is short at the cost of a payment stream. Housing can also be a consumption good leading to depreciation uncovering a covering.
Maybe its been brought up already, but at least in the US, most benefits packages stem from the wage fixing that occurred around ww2.
Also, what you said reminded me of something my dad told me. “A house isn’t an investment, a house is an expenditure.”
westslope: “Even in a world of second-best transaction cost constrained alternative equilibria, markets clear, and equilibrium is still obtained.”
Yes, there’s an equilibrium, but what sort of equilibrium is it? How well do prices capture the information available? And I’m not sure what “market-clearing” means when each house is unique in some way, so we can’t as easily talk about excess demand or supply at the “prevailing market price”.
RSJ: The housing market is like the stock market, except there are 20 million (or whatever) different companies, and each company issues (normally) only 1 share. That fact has to be important in some way. And one of the ways it may be important is in explaining why, when demand for houses falls, house prices seem to be so sticky, and trading volume falls, and we get what looks like excess supply.
You are very probably onto something. But I really don’t find your explanation of the relation between those things very clear. Might be you, might be me; or it might just be that it’s damn hard getting one’s head around this stuff!
DW: Yes. You are right to bring up the fact that insurance provides liquidity. Something I hadn’t thought about. No problem in my case, since I would just walk straight out and buy another $5,000 used car. But if I were replacing it with a new car, and were liquidity constrained, it might be different. But then the $3,000 insurance payout still would only cover a small fraction of the liquidity needed for a $20,000 new car.
Lord:
Yes. In aggregate, since every house is owned by someone, we are not short housing. We are covered. But if you don’t own a house, you are short housing until you die. We are on the hook for the present value of the rents until death. If we buy a house on a 100% mortgage, and never pay it off, we cover our short position in housing, but replace it with a short position in bonds. If you pay off your mortgage, you cover your short position in bonds, and also cover your short position in housing.
That’s a better way of looking at it, I think.
BTW. Does anyone understand “balance protection” on credit cards? Is this another insurance rip-off, that very few people would benefit from buying? It sounds like it.
Nick: The second-best, transaction-cost constrained market clearing price is a real world equilibrium. Many goods and services exhibit the same attributes as residential homes.
BTW, this reminds me of taking an Industrial Organization theory course with creative genius Curtis Eaton who would turned and said to the graduate seminar with solemn gravity “There are a lot of monopolies out there” before he dove into another one of his insightful, simple monopoly models.
Nick:
“The housing market is like the stock market, except there are 20 million (or whatever) different companies, and each company issues (normally) only 1 share. That fact has to be important in some way.”
No, because I can still short IBM and buy Boeing. In fact, if you are the “holder” of IBM, then you would not even know that I had shorted it, since the short position would be an agreement that I made with your broker — something that you authorized when you opened your margin account.
The reason why this can happen is that financial assets are just contracts that entitle you to an income stream. Therefore if I agree to supply you with the income stream, then you cannot distinguish between a situation in which you have my IOU in your account or one in which you have a share of Boeing in your account. In practice, brokers do not notify their customers when shares in their accounts are borrowed.
It is this indistinguishability that allows anyone to supply shares of IBM by agreeing to match the income stream. As a result, the share price of IBM reflects the mean expectation of the cost of replicating the income stream.
With housing, it would be like you coming home to find another family living in your house, for as long as they wanted, provided that you were re-imbursed for the fair market rental value, as someone shorted your house and sold it to the family.
In that situation, house prices would stay in line with mean expectations of rental values. No one would be able to “sit” on a house and try to wait out the downturn, since their house would be sold as soon as anyone thought it was overvalued vis-a-vis rents, and it would be bought as soon as anyone thought is was undervalued. It would repeatedly bought and sold by anyone with an opinion about the house, whether they owned it or not.
“But I really don’t find your explanation of the relation between those things very clear.”
I’m sure the fault is mine. Here is another attempt:
Suppose there are 3 assets, and 3 investors
X = [x1, x2, x3] = assets
I = [y1, y2, y3] = investors
Each asset is to promise an unknown return next period, between 0 and 1. investor has an opinion about the return of each asset in the next period — a number between 0 and 1. The opinion of investor i about the return of asset j forms a matrix, M.
Now assume that each investor has holdings, H of assets. For assets, y1 holds x1 and y2 holds x2, etc. For cash, each investor starts with 0 cash but is able to borrow as much as they want, but must return this amount in the next period from their cash proceeds.
You can get into the following situation:
Expectations Matrix:
[.3, .6, .7]
[.8, .5, .3]
[.4, .2, .5]
y1 owns an asset he believes is worth .3, but y2 believes it is worth .8 and y3 believes it is worth .4
Similarly y2 holds an asset that he believes is worth .5, but y1 believes it is worth .6 and y3 believes it is worth .2, etc.
For the sake of simplicity, assume when two investors transact, the price paid is the average of the opinions about the assets, and that transactions only occur when the current owner believes an asset that they hold is worth less than what someone else is willing to pay.
In this case, we would have the following transactions and market prices:
investor 1 sells x1 to investor 2 for .55
investor 2 sells x2 to investor 2 for .55
investor 3 sells x3 to investor 1 for .60
And these are all the transactions that occur. Investor 3 holds only cash, and the assets flow to investor 1 and investor 2 only. Note that these are not the average prices. The mean prices are .5, .43, .5, for x1, x2, and x3, which are less than the marginal prices.
Suppose, on the other hand, that now you can sell a security without owning it, simply by agreeing to supply a return at period 2. In this case, you have a complete market and more transactions will occur:
Investor 1 also sells asset 1 to investor 3 for .35
Investor 3 also sells asset 1 to investor 2 for .60
So now the average transaction price of asset 1 is the average of .55, .35, and .60 = 0.5 — the mean price. You have convergence to mean once those investors that do not hold the asset are allowed to both buy and sell it. You converge to a price above the mean when anyone can buy it, but only those who hold it can sell it.
RSJ:
“investor 2 sells x2 to investor 2 for .55”
Typo? Should read: “investor 2 sells x2 to investor 1 for .55”?
Just want to make sure I am following.
Yes, you are right about the typo.
I would also add (if it wasn’t obvious), that this prevents the market from “stalling” and adds to volume, as someone in Florida may have debt commitments and other reasons that prevent them from selling the house for less, whereas I, as the short-seller, have no such barriers preventing me from selling their house. Many of us would have shorted Florida real estate prices in exchange for incurring the obligation of paying market rents 🙂
OK. I am getting the gist of your argument.
Let me think about it.
RSJ:
“For the sake of simplicity, assume when two investors transact, the price paid is the average of the opinions about the assets,..”
I wouldn’t generally accept that assumption, because it might create a price that is lower than the second-highest bidder would pay. (But this is not a problem in your numerical example, and in any case I could replace it with a more reasonable assumption, and it wouldn’t affect the substance of your argument.)
Suppose instead of houses, there are 26 pieces of fruit: one apple, one banana, one cantaloupe, one date….
And 26 people: A has the highest value of an apple, B the highest value of a banana, C the highest value of a carrot, etc.
Each person values their most preferred fruit at $26, their second at $25, etc.
Price of fruit, with no short sales, = $25.50 (or $25 with English auctions of fruit).
Average valuation of fruit, = $13.50
Price of fruit, with unlimited short sales: $2 (?)
Yet, I wouldn’t want to say that fruit is priced “too high” in this example.
So, what is the difference between fruit and assets like houses? (That is not intended rhetorically, because I can see one difference I think is important: the value of fruit is pure taste; the value of a house is part taste (do I want to live there?) and part beliefs (what will it be worth next year?).
I’m going to sleep on this one.
By the way: did you just think this stuff up? Or is it well-understood in some circles?
“I wouldn’t generally accept that assumption, because it might create a price that is lower than the second-highest bidder would pay.”
Yes, this means that prices jump up and down unless you want to change the model for more than two-party sales — which would be more realistic, but wouldn’t really change the conclusion. You could always schedule the trades in such a way as to make them into a sequence of two-party trades, and in practice this is close to what happens with market-makers. In any case, I don’t think this matters.
“Suppose instead of houses, there are 26 pieces of fruit:”
Woah.. you are talking about consumption. With consumption it is the case that the benefit I get from eating an apple is different from your benefit, in which case you should pay more if there is a limited amount of apples. That is a good thing, and promotes trade as well as production and consumption as we make money off these relative gaps in preferences. And there is a cost of gluts to be borne from this, too.
But with a financial asset, the utility is just the expected cash-flow. I agree that houses are bit of a hybrid here, in that expected cash-flow plays a role, but is not the whole story. With housing, people are financially stupid because it’s such an emotional issue. I was arguing “in principle” why house prices cannot be equal to mean expectations of rental returns.
Only financial assets are determined purely by prediction and allow for the type of cash-flow substitutability that can make these markets complete. All other markets are not complete and are set by marginal concerns.
I guess this is part of a long-standing discussion in which I keep arguing that the “nth-market” — the financial asset market, or what you call the money market — is fundamentally different from the other markets, and this changes some of the disequilibrium analysis.
“By the way: did you just think this stuff up? Or is it well-understood in some circles?”
Hmm, this is just how I think about financial markets, from some experience trading them. I thought I was describing orthodoxy as expressed by EMH, or at least the “no arbitrage” version of EMH. It is also orthodoxy to observe that prices for goods are determined by marginal cost and benefit, so pointing out that these two price-setting mechanisms result in different prices is not so radical if you are willing to assume a diversity of expectations. I can’t point you to a paper, though. Maybe this is orthodoxy among financial engineers?
“Only those houses that are not owner occupied — a small minority — can be shorted”
Not at all. Just sell to a landlord, invest the proceeds, rent the house from him, and offer to buy it back at the end of any chosen rental period. Very simple. And can be done with any single house, uniquely.
RSJ: We are (mostly) on the same page here.
A house is like a fruit tree. Every year it produces one piece of fruit. It’s an asset, but its dividend is a consumption good. People have different tastes for the fruit. I have already built that in. But your argument really concerns the idea that people have different beliefs about the future. I haven’t built that in yet. I need to introduce some sort of uncertainty in order to capture what you are saying. It could be uncertainty about future fruit yields. Or uncertainty about future valuations of the fruit. Or uncertainty about future interest rates. I’m going to think about what’s simplest.
I want to stick with my “alphabet” example, because it is symmetric, and so simpler to do the arithmetic and capture the intuition than in your matrix. Make my alphabet “circular”, so that when we get to Z the next letter is A. (This is what is called the “circular beach” model of preferences). That makes it truly symmetric.
And I want to stick to pricing by English auction. So the price is equal to the valuation of the second highest bidder. Simplest roughly realistic assumption. (Wouldn’t make much difference if we made the price the average of the highest and second highest bidder.)
I will return.
As an aside:
“I guess this is part of a long-standing discussion in which I keep arguing that the “nth-market” — the financial asset market, or what you call the money market — is fundamentally different from the other markets, and this changes some of the disequilibrium analysis.”
Other people call that the “money market”. I don’t. With n goods including the medium of exchange, there are n-1 markets. There is no nth market. The term “money market” is a misnomer, because every market is a money market. Money (the medium of exchange) is traded in each of the n-1 markets.
If there are 4 goods, A,B,C,and D, and if A is the medium of exchange, then there are 3 markets: a market where A is traded with B; a market where A is traded with C; and a market where A is traded with D. There’s a B market, a C market, and a D market. All 3 markets are A markets.