Household Sorting Economics 312 Martin Farnham
Segregation Different household types are not uniformly distributed across space; instead tend to cluster by type Segregation by race and income are common Why do we care? Segregation may compound problems of poverty, by leaving poor children with few positive role models Segregation may lead certain segments of the population at higher risk for crime May compound inequity On the other hand, segregation (say by demand for school quality) can lead to a better (efficient) match of local public services to individuals 1
Household Sorting--Local Publicly- Provided Goods as Possible Cause Households may self-segregate if different people want different local services and amenities; for example People who like nature may choose to live near parklands People who like city life may live downtown People who want good schools may sort into good school districts People who want low crime may sort to areas that spend a lot on policing (or that have low populations of poor youth) 2
Sorting by Demand for Local Publicly- Provided Goods (fiscal sorting) Can think of each community as local service provider (of parks, schools, streets, etc.) Each community offers a different local bundle (and tax bill to pay for the bundle) People shop for the bundle and tax bill that suits them best People vote with their feet by choosing the community with the best local bundle This allows different people to consume different public services, rather than forcing same services on everyone Under fairly extreme assumptions this fiscal sorting leads to an efficient allocation of local publicly-provided goods. 3
Why is Fiscal Sorting Efficient? When people in a community want different things, voting tends to be inefficient Policy tends to be one-size-fits-all Elderly pay for education they don t benefit from; young pay for bowling greens they don t benefit from Voting majority may choose to spend public money on things people in the minority don t want. If people sort by the local goods and services they want, then (in theory) everyone can get exactly what they want no more, no less. Young communities with good schools Old communities with good bowling greens This is distributive efficiency like you learn about in 203/313 4
Dark Side of Local Fiscal Sorting Suppose all people have same underlying preferences, but some are rich and some are poor. Rich will demand large parks Poor will demand small parks (they like large parks, but can t afford them) If everyone pays the same tax per person (a head tax), fiscal sorting will lead to income segregation due to sorting over local publicly-provided goods (LPGs). Note: We don t need rich to dislike poor to get this outcome just need them to want different LPGs or different quantities of LPGs This is (partly) why local funding of schools in US leads to such high degrees of income segregation 5
Evidence on Fiscal Sorting (sometimes called Tiebout sorting) There are different ways to test whether people really choose where to live according to differences in local services Survey people on tastes Do people with similar tastes live together? Do people with different tastes live apart? Gramlich and Rubinfeld (1982) do this, find evidence supporting Tiebout 6
Evidence on Fiscal Sorting Test for whether communities become more homogeneous over time Tiebout hypothesis assumes costless mobilityècostly mobility should lead to incomplete sorting. Hypothesis: As mobility costs have declined over time, people should have engaged in more complete Tiebout sorting Are communities more homogeneous now than 100 years ago? Rhode and Strumpf (2003) test this; find little increase in sorting over past 100 years Suggests Tiebout may not explain residential choice However, rely on proxies for fiscal sorting like whether jurisdictions are homogeneous with respect to race or with respect to income (don t directly measure fiscal demands) 7
Evidence on Fiscal Sorting Another test: look at how people change their fiscal bundle when they move. Do changes match what we d expect given where households are in lifecycle? Example: Young couples with kids should want good schools; old couples should not care as much about good schools Do old couples reduce their exposure to local school taxes when they make their empty-nest move? (first move after youngest kid turns 18) If Tiebout explains sorting behavior, should expect empty-nest movers reduce exposure to school spending Farnham and Sevak (2006) find mixed evidence of this in US Residential sorting is inhibited if higher level of government (such as state or province) forces communities to spend equal amounts on education F&S 2006 find evidence of strong fiscal sorting in states that don t equalize school spending across districts. 8
Sorting over Characteristics of Neighbours People also seem to have preferences for certain types of neighbours. People generally like quiet, non-disruptive neighbors People like their kids to have positive role models Hardworking neighbors who convey (through their success) the virtues of education and work Hardworking friends who do homework on time and don t get into trouble. If people believe rich neighbors will tend to be relatively desirable along these dimensions, people will try to choose rich neighbors. 9
Why Might Rich be Better Neighbors? Rich tend to commit less crime (or at least less crime that affects their neighbours) Opportunity cost of getting caught is higher than for poor (because a year of foregone income due to going to jail is higher) Rich tend to have prestigious jobs A doctor living next door may (on average) better model what you want your kids to turn into than an unskilled worker or someone who is unemployed. Rich tend to value education highly They teach their kids to value education, so their kids are likely to be a positive influence in school. 10
Why Might Rich be Better Neighbors? Obviously this is more than a bit elitist, but it s not meant to be judgmental Rich may get ahead by good luck (being born to rich parents, or winning the lottery in life) So this argument doesn t have to imply that there is virtue associated with being rich. Given that they re rich, they may have the luxury of pampering their kids, not resorting to crime, drinking fancy wine instead of doing drugs, etc. Also, we re just talking averages and perceptions here 11
Sorting by Income Over Neighbor Quality Suppose that everyone prefers (ceteris paribus) rich neighbors to poor. Will this lead to mixing or segregration? Answer depends on a number of things If rich and poor are equally willing to pay for rich neighbors, then we ll observe mixing If rich are willing to pay more for rich neighbors, rich will outbid poor in competition for homes in rich neighborhoods. Segregation will occur. Recall that willingness to pay is partly a function of preferences, but also partly a function of wealth. People with higher ability to pay tend to have higher willingness to pay; hence we ll generally expect the rich to be more willing to pay for rich neighbors than the poor. 12
Sorting by Income Over Neighbor Quality Assume everyone wants rich neighbors 2 neighborhoods with 100 identical housing units each 2 types of people (rich and poor); 100 of each type Degree of segregation depends on differences in WTP between rich and poor; in this case we assume rich are WTP more than poor for rich neighbors r A -r B A Segregating Equilibrium Rent Premium (high income) Rent Premium (low income) s u 50 High Income Households in 100 Neighborhood A 13
Sorting by Income Over Neighbor Quality Note that the rent premium is just the extra amount that a person is willing to pay to live in N hood A rather than N hood B Why is the point where there are 50 rich people in N hood A not a stable equilibrium? Suppose one rich person moves in from N hood B Now there are 51 rich people, which makes N hood A more desirable than N hood B; homes in N hood A will go to people willing to pay the most for them. The 52nd will be rich, the 53rd will be rich, etc. until all 100 in N hood A are rich. Why is the point where there are 100 rich people in N hood A a stable equilibrium? 14
Sorting by Income Over Neighbor Same scenario, except poor willing to pay more than rich for rich neighbors Note that the stable equilibrium now involves complete integration Each neighborhood has 50 rich and 50 poor Neighborhoods are identical to each other (by income composition) Rent is the same in each Quality r A -r B s An Integrating Equilibrium Rent Premium (low income) Rent Premium (high income) 50 High Income Households in 100 Neighborhood A 15
Sorting by Income Over Neighbor Quality Why is s an equilibrium in this case? If one rich person moves from B to A, then then everyone will prefer A (because it will have 51 rich and 49 poor, while B will have 51 poor and 49 rich) The poor will outbid the rich to obtain a spot in A, hence bringing the number of rich and poor back to 50 of each. 16
Sorting by Income Over Neighbor Here the rent premium curves cross. If they cross such that the low-income premium curve is steeper than the high-income premium curve, then there is a stable equilibrium at the crossing In this case, A is richer than B, but there is not complete segregation. Equilibrium rents will be higher in A. Quality r A -r B Equil. w/ Incomplete Integration u Rent Premium (low income) s Rent Premium (high income) 50 High Income Households in 100 Neighborhood A 17
The Role of Lot Size in Income Sorting In above setup, we assumed rich and poor lived in identical houses What if rich buy twice as much house as poor? Then the premium they are willing to pay to live in a richer neighborhood is half that of the premium the poor are willing to pay, on a per unit of house basis If the rent premium for the rich is less than twice the rent premium for the poor, then the poor will outbid the rich for housing. Say the rich rent or buy 1/2 acre lots, and the poor rent or buy 1/4 acre lots; 2 poor people will combine to outbid 1 rich person for a 1/2 acre lot which they can subdivide or 18 share.
Consider Rich Person who wants to combine 2 small units into 1 big one Assume poor rent 1 unit in a duplex; rich rent whole duplex and remove wall Assume 1 neighborhood richer, 1 poorer; poor willing to pay $1100 per unit in rich N hood, $900 per unit in poor N hoodèpoor willing to pay $200 premium to live in rich N hood Rich would have to pay at least $2200 for a whole duplex in rich N hood (to not be outbid by poor), $1800 for whole duplex in poor N hoodèrich has to pay a $400 premium to live in rich N hood (to not be outbid by poor) If poor are willing to live in smaller units/on smaller lots, they may be able to outbid richèintegration (Table 8-3 O Sullivan illustrates) 19
Property Tax and Segregation Most local communities in Canada use the property tax as their main local tax Tax liability=t*(assessed Value) Those with more land pay higher taxes This provides an incentive for people who can get by on little land to try to mix into the same communities with big landowners Why? In that situation, rich will subsidize poor Rich don t avoid poor just because they don t like them it s expensive to live near them. 20
21
This Provides Incentive for Exclusionary Zoning Everyone wants to live in a rich community Both rich and poor have lowest tax burden there Poor will try to move in Rich will try to keep them out Will resort to legal means to zone out poor Lot-size minimums, housing code requirements, etc. èsystem of local property-tax financing likely contributes to income segregation If provinces simply raised income taxes, and redistributed the money to communities, there would probably be more mixing by income Only way to avoid subsidizing the poor would be to leave province 22
Sorting By Other Local Characteristics Many parents choose a neighbourhood on the basis of local schools Schools or districts that are perceived to be better may attract wealthier families (if education is a normal good) Others (especially elderly) choose neighbourhoods on the basis of crime rates Wealthier may move to areas with lower crime rates (if safety is a normal good) Environmental amenities such as trees or views may attract richer families 23
Sorting By Other Local Characteristics Some people may sort by language (it s easier to gain some benefits of clustering when you can communicate with those you re clustered with) Chinese neighbourhoods in Richmond English and French neighbourhoods in Montreal. Punjabi neighbourhoods in Surrey. Spanish speaking neighbourhoods in NY City In this case, what may look like segregation driven by race or ethnicity might be driven more by convenience of communication. 24
Sorting by Race Cities are often highly racially segregated. Why? 1 (somewhat) innocuous reason and 1 pernicious reason. (somewhat) Innocuous reason: Race and income are correlated; racial minorities tend to earn less than whites So even if people are race-blind, income segregation will cause some racial segregation Kain (1985) suggests non-racial household characteristics predict about twice the number of blacks should live in US suburbs as actually did at the time, which suggests... A more recent study of the San Francisco Bay Area suggests that 30% of African American segregation is attributable to non-race household characteristics, like income. Pernicious reason: People are directly motivated by racial characteristics of neighborhoods 25
Differing Perceptions of Ideal Racial Integration In US, most whites prefer to live in lightly integrated neighborhoods; most blacks prefer to live in heavily integrated neighborhoods. A 1980s study found whites and blacks who claimed to prefer integrated neighborhoods had very different visions of the ideal level of integration Whites viewed 20% black as ideally integrated Blacks viewed 50% black as ideally integrated This creates the perfect environment for a coordination failure, whereby neither group s ideal (or even a compromise) is achieved if the real estate market is left to its own devices. 26
Neighbourhood transition Consider an all-white neighbourhood With the above-noted preferences, blacks could start to move into the neighborhood and existing residents would be happy until the percent black in the neighborhood was 20% Beyond that level, whites might begin to move out, further raising the percent black If the neighborhood reaches an integration tipping point all whites might move out and it might become completely segregated again (all black now) Everyone s unhappy with that outcome. 27
Neighborhood Transition With strongly divergent views of what constitutes ideal integration equilibrium segregation may be very strong (neighbourhoods almost all black or all white) Even when everyone wants some degree of integration, we can end up with heavy segregation. Thomas Schelling (economist) developed a model that produces such tipping points, with white neighbourhoods transitioning to black neighborhoods This white flight is thought to be a major cause of the isolation of black families in American inner cities Deprives them not just of a sense of inclusion, but of economic opportunity, as jobs tend to move away too. 28
Neighborhood Transition in Canada Fong and Gulia (2000) find that white flight is not a prominent feature of Canadian neighborhood change Instead neighborhoods are becoming more diverse, though this is due more to minorities trying to mix with whites than whites trying to mix with minorities According to their findings, in cases where neighborhoods become more uniform, it s due to minorities moving out 29
How Real Estate Agents and Banks Facilitate Racial Discrimination If whites prefer low-integration neighborhoods, then integration may lower white willingness to pay for housing--so integration could lower property values Real estate agents have vested interest in property values in their market Also don t want to annoy potential white clients It can quickly become in one s economic interest to be racist Steer black clients away from white neighborhoods Studies in US show evidence of discriminatory practices by real estate agents US Fair Housing Act (1968) meant to counter this. Similar laws do not exist in Canada 30
Real Estate and Lending Discrimination Banks can also prevent integration by refusing to lend to racial minorities (or by lending to them at higher rates or by only lending to them if they steer clear of white neighborhoods) This may prevent minorities/immigrants from moving from renting to owning May make move from segregated inner city neighborhoods to suburbs more difficult 1975 Home Mortgage Disclosure Act (US) 1977 Community Reinvestment Act (US) No similar laws in Canada 31
Residential Discrimination in Canada Paired audits show evidence of rentalhousing discrimination against blacks and south Asians Diminishing over time, but still present Studies of rental market discrimination (e.g. Dion) in Toronto suggest that black immigrants to Toronto (Somali, Jamaican) experience more discrimination than white (Polish) immigrants 32
Consequences of Segregation We ve identified a large set of reasons why people may segregate. Why is racial segregation problematic? It suggests discrimination (an equity concern) and likely exacerbates problems of poverty among minorities which is often caused by labour-market discrimination. Income segregation in general can be problematic Kids isolated in poor neighborhoods may have limited opportunity for upward mobility School quality may suffer, due to negative peer effects, low levels of resources, bad teachers, etc. Hopelessness may cause spillovers for rest of society in form of crime, drug addiction, welfare dependency 33
Consequences of Segregation Segregation interacts with some major trends in the spatial arrangement of jobs in cities Recall discussion of suburbanization of manufacturing. Blacks in US used to work heavily in manufacturing in the industrial north. Relatively prosperous due to high-paying jobs When those jobs left, blacks found it difficult to follow jobs to suburbs--this is called the spatial mismatch problem (John Kain). Income segregation and housing discrimination made it hard for blacks to move to suburbs; they were trapped in inner cities with jobs moving away. Joblessness and poverty rose (starting in 1950s and reaching crisis by late 1960s) Race riots and other forms of unrest contributed to middle-class flight, hollowing out of tax base, city schools, etc.èvicious cycle 34
Why do anti-discrimination policy? Obvious justifications based on equity. For economists, the more interesting question is whether there are justifications based on efficiency. Arguably there is a large negative fiscal externality from restricting economic opportunity to large groups of the population If these groups weren t discriminated against, government would collect more revenues; spend less on welfare, criminal justice system etc. 35
Why do anti-discrimination policy? Also, remember many of the gains from clustering come from exchange of ideas and information. If we live in a segregated society, each group operates in a silo, with limited opportunities for information exchange Policies to promote integration may increase the positive externalities associated with agglomeration. 36
Other anti-segregation Policies Busing Kids were bused from white schools to minority schools and vice versa, to improve racial mixing in schools Created angry backlash, especially among white parents; some among black parents (kids had to travel long distances) Thought to have contributed to further suburbanization, which would have undermined the effectiveness of the program 37
Housing Voucher Programs Gatreaux (Chicago): Courts ruled public housing in Chicago to be discriminatory (segregationist). Remedy was to give 7000 African American families housing vouchers to move into less segregated neighbourhoods Kids from families with vouchers had better educational outcomes than those from families without. 38
Housing Voucher Programs Moving to Opportunity (US): Lowincome families given vouchers to move to richer neighborhoods Kids of families offered vouchers have better educational and work outcomes than those not offered vouchers (lottery determined who got vouchers) Kids who moved as teens suffered (difficult adjustment to new school, etc.); but kids who moved young did substantially better than those left behind. 39
Centralizing School Finance School funding in Canada is fairly centralized (i.e., provincially rather than locally controlled) Many US states have moved in this direction over last 30 years (under court order) Cutting link between local income and local service levels removes one incentive for the rich to isolate themselves from the poor; may help to lower income (and therefore racial) segregation 40
Attitudes toward racial residential mixing are improving In US in 1958, 44% of whites said they would move if a black family moved in next door; in 1997 same poll showed just 1% said they would move In 1958, 80% of whites said they would move if great numbers of blacks moved into neighbourhood; 18% said they would move under such circumstances in 1997. Measures of segregation are falling for American cities (suggesting more racial mixing) 41
But it will take a lot to undo the history of segregation Segregation in US remains high See Vi Hart s segregation simulator on course website (illustrates Schelling tipping model). Shows low levels of individual bias can lead to high levels of segregation. Also demonstrates that segregation is hard to undo. If there was segregation historically, then it will only be undone if people seek out mixing 42