Ben Harris is a Fellow with the Brookings Institution and Policy Director of The Hamilton Project. Harris focuses on federal tax and budget policy, state and local public finance, and retirement security. He was previously a senior economist with the President’s Council of Economic Advisers, a research economist with the Brookings Institution, and a senior economist with the House Budget Committee.
Most federal subsidies aimed at helping households accumulate financial and tangible assets are delivered through tax expenditures—spending through the tax code. In 2013, these deductions, deferrals, and credits aimed at encouraging such asset building totaled over $350 billion, most devoted to homeownership and retirement saving incentives. Yet, these tax incentives do a remarkably poor job.
In contrast to government appropriations, which at least sometimes are scrutinized though the congressional budget process and are currently limited by statutory caps, many tax expenditures grow automatically and are subject to little oversight.
The mortgage interest deduction, for example, has been part of the income tax since its beginning in 1913. Even though there is much evidence that it is an inefficient tool to encourage home ownership, American taxpayers annually spend roughly $70 billion on the mortgage interest deduction—about the amount expected to be spent on food stamps. But while policymakers have just concluded an intense debate over whether to cut food stamps, few ask whether the mortgage interest deduction and other tax expenditures that support asset development achieve this goal.
In a new paper, my colleagues Gene Steuerle, Signe-Mary McKernan, Caleb Quakenbush, Caroline Ratcliffe, and I reviewed the academic evidence and found that many of these tax expenditures fail to promote asset development. The mortgage interest deduction is not alone. For instance, retirement saving subsidies do little to increase total overall wealth of lower-income households.
Tax subsidies for activities such as home ownership and retirement savings benefit those at the top of the income distribution more than those at the bottom. Most come in the form of current deductions from income or as deferral of taxable income, which are typically worth more to taxpayers with higher marginal tax rates and can be worth little or nothing to taxpayers in the lowest tax brackets. For instance, about 70 percent of the tax savings from the mortgage interest and property tax deductions accrue to the households in the top 20 percent of income. Just 8 percent goes to middle-income households, and almost nothing to the bottom 40 percent.
Similarly, roughly 70 percent of the tax subsidies for employer-based retirement savings and 65 percent of IRAs benefit the highest-income 20 percent of households. Although education subsidies are more progressive, the size of these tax expenditures is dwarfed by those for housing and retirement.
The evidence from years of research is pretty clear: Current tax subsidies for asset building are poorly designed. They fail to increase savings much and overwhelmingly benefit the highest-income taxpayers who are least likely to need these incentives. When it comes to tax reform, they are low-hanging fruit.
Taxpayer use of itemized deductions varies widely by location, according to a new analysis of 2007 IRS data. In about one in ten counties, 11 percent or fewer taxpayers itemize while in another 10 percent at least 38 percent of taxpayers claim deductions. In a handful of counties, more than half of taxpayers itemize. In general, taxpayers in high-cost, high-tax counties located along the coasts are far more likely to itemize than those in living in in the middle of the country (see map).
About one in three tax filers itemizes. The largest itemized deductions are for mortgage interest, state and local taxes, and gifts to charity. Others include certain medical expenses, job-related expenses, and casualty and theft losses. The Pease provision limits itemized deductions for upper-income taxpayers.
The variation in itemized deductions depends on many factors, but is driven by differences in income. For example, taxpayers with higher incomes own more expensive homes and carry larger mortgages and thus pay more in mortgage interest. Some live in states and counties with higher taxes. Taxpayers with higher incomes tend to give larger gifts to charity, and they tend to cluster along the east and west coasts.
Not only does the share of taxpayers who itemize vary greatly from county-to-county, so does the amount they claim. Among those taxpayers itemizing, the median county-level deduction was $18,590 in 2007. However, in 10 percent of counties, taxpayers itemized an average of $16,640 or less, while in another 10 percent they deducted an average of $22,990 or more. In about 1 percent, taxpayers deducted an average in excess of $35,020.
Where do the biggest itemizers live? Not surprisingly in high-cost, high-income urban counties located in southern California and the corridor between Washington, DC and Boston, and around major inland cities such as Chicago and Denver. Inland and rural counties tend to benefit less from itemization.
The Earned Income Tax Credit (EITC) is one of our nation’s most effective anti-poverty policies, bringing 10.1 million families out of poverty in 2012. The EITC is designed to reward work by increasing wages for low-income workers; workers with low incomes can receive up to an additional 45 cents for every dollar they earn. Recognizing the ability of the EITC to ease poverty and incent work, the President recently called for an expansion of the EITC for childless workers in his State of the Union address.
The EITC is only available to low-income workers, and phases out as taxpayer income increases. The value of the credit is not only based on income, but family structure as well: taxpayers with children receive substantially larger benefits relative to those without. Since these income and demographic characteristics vary across states and counties, there is wide variation in EITC take-up across localities. The remainder of this brief describes this variation in greater detail, using IRS data from 2007.
There is notable variation across counties in EITC take-up. Ranked by share of taxpayers claiming the EITC, the bottom 10 percent of counties have claiming rates of 10.7 percent or lower, while the top 10 percent of counties have claiming rates of 27.1 percent or higher. The median county sees 16.2 percent of taxpayers claiming the EITC.
The size of the average EITC, among those claiming the credit, varies less than take-up rates. Most counties have average credits between $1,500 and $2,500: 10 percent of counties have average credit values of $1,657 or less, while 10 percent of counties have average credit values of $2,236 or more.
The regional variation in the EITC is stark. The counties with the highest share of taxpayers taking up the EITC are overwhelming located in the Southeast. As can be seen in the accompanying map, a large share of counties in Alabama, Georgia, and Mississippi have over half of their taxpayers claiming the EITC. With few exceptions, almost all counties with high rates of EITC take-up are located in the South. Relative to the South, the Northeast and the Midwest have much lower rates of EITC take-up. Moreover, average EITC benefit closely follows the pattern for share of taxpayers taking up the credit.
Tax expenditures for homeownership, such as deductions for mortgage interest and property taxes and the partial exclusion for capital gains on the sale of a primary residence, have long been recognized as ineffective, regressive, and extraordinarily expensive—costing $121 billion in 2013 alone. Until now, most reforms—including the Bowles-Simpson deficit-reduction plan—have focused on restructuring the mortgage deduction into a flat-rate credit. But what if we largely replaced the deduction with incentives to buy a house, rather than to run up a lot of mortgage debt?
In a new Tax Policy Center paper, my TPC colleagues Gene Steuerle, Amanda Eng, and I examine three very different tax subsidies for housing. Instead of encouraging people to borrow, they would create incentives for homeownership without tying tax breaks to mortgage debt. Each would be financed by eliminating the deduction for property taxes paid and capping the mortgage interest deduction at 15 percent. Thus, none would add to the deficit over 10 years. Here, briefly, are the three options:
- A permanent First-Time Homebuyers Tax Credit, similar to the provision temporarily in effect during the Great Recession. The credit—$12,000 for singles and $18,000 for married couples—would give new homeowners a one-time lump sum subsidy in the year they purchased a home. The credit would be refundable, allowing households with low income tax burdens to claim the full value of the credit.
- A flat annual subsidy for homeowners. Taxpayers could claim the credit—$870 for singles and $1,300 for couples—in any year in which they owned their home. The credit would be refundable and would phase out for upper-income households.
- An annual 36 percent flat-rate credit for property taxes paid, up to a maximum of $1,400 for single filers and $2,100 for couples. Like the other alternatives, the credit would be refundable.
Each would offer homeownership subsidies to a wide range of taxpayers, rather than concentrating the benefit on those with high-incomes who itemize deductions. All three would raise after-tax incomes for the bottom 80 percent of taxpayers while trimming after-tax income for the top 20 percent. And each would raise housing values by about 1 percent by lowering the after-tax cost of housing investment.
They also are more in line with economic arguments for subsidizing homeownership. Economists justify these tax subsidies on two grounds: 1) homeowners have a financial investment in their neighborhoods and therefore work to maintain them and 2) homes contribute to asset accumulation, and therefore to higher saving that provides greater security when incomes fall, such as when people lose jobs or retire. In contrast to current preferences, which are worth more to those with higher mortgage debt and in higher tax brackets, these alternatives tie the value of the subsidy more directly to homeownership.
None are perfect. For example, the First-Time Homeownership Tax Credit does not reward long-tenured homeowners, the annual tax credit does not reward upper-income homeowners, and the property tax credit would probably cause local property tax rates to rise.
The social value to homeownership may or may not be worth more than $100 billion-a-year in government subsidies. But regardless, if the goal really is to encourage people to buy homes, any of these alternatives would be superior to the current deductions for mortgage interest and property taxes.
Differences in income and other characteristics mean that federal income tax burdens vary substantially across counties. While the median federal income tax burden across counties is about $3,400, approximately 10 percent of counties have average tax burdens less than $2,100 and around 10 percent of counties have average tax burdens over $6,700. Counties with high federal income tax burdens are concentrated around large cities, the California coast, southern Florida, and the corridor between Washington DC and Boston.
Income taxes comprise a substantial fraction of the country’s revenue stream. In recent years, around 45 percent of federal revenue was derived from income taxes, with federal income tax revenue amounting to about 7 percent of GDP. Income tax burdens vary for a host of reasons, including family demographic characteristics, level and composition of income, and deductions and tax benefits claimed. Since these characteristics vary by geographic location, it is natural to see variation in income tax burdens across localities.
The map below presents variation in income tax burdens for each county in the United States. (Data are from 2007, the most recent data available for this type of analysis.) The variation—due in large part due to differences in incomes—is stark. In absolute dollars, the average tax burden across counties is $3,419, but some counties’ average tax burden is substantially more and some counties’ burden is substantially less. Ten percent of counties have average tax bills of $2,102 or less, while ten percent of counties had average tax bills of $6,745 or more. As a share of adjusted gross income, the lowest 10 percent of counties had average tax bills under 7.4 percent of income while the top 10 percent of counties had average tax bills in excess of 11.7 percent of income. (See here for an interactive map that also presents county-level tax burdens as a share of income.)
In general, counties with higher income tax burdens—due in large part to higher incomes—tend to be clustered around large cities. For example, most counties in Illinois have low average federal income tax burdens, with the exception of those counties surrounding Chicago. In addition, Texas counties close to Dallas, Houston, and San Antonio generally have higher tax federal income tax burdens relative to those counties farther from a major city. Counties located on the California coast, in southern Florida, and within the corridor between Washington DC and Boston also tend to have higher federal income tax burdens. In addition, some counties that host popular ski resorts—such as Pitkin County, CO (which contains the city of Aspen) and Teton County, WY (which contains Jackson Hole)—also have notably high income tax burdens.
Several caveats to the data are worth noting. One, the data only include those families who filed income tax returns. Thus, families who did not file returns are not counted in the average tax burdens. Two, in 2007 many low- and middle-income families could receive a stimulus-based rebate, but only if they filed an income tax form. This induced some families who would not have filled income taxes to submit a tax return in order to receive the rebate. Lastly, to protect the confidentiality of taxpayers, the Statistics of Income division at the Internal Revenue Service has omitted information from some counties; data on these counties is not publicly available.
Congress’s latest flirtation with debt-limit default caused barely a ripple in the financial markets. Rates on short-term Treasuries spiked in early October, before quickly subsiding to more normal levels. The spread between one- and three-year Treasuries temporarily widened, but quickly fell back to a more normal trend. All told, financial markets barely blinked. Unfortunately, the next time may be worse. Here’s why.
If Congress is going to threaten the country with defaulting on bond payments, late October is a relatively good time to do it. In the year spanning October 1, 2013 to September 30, 2014, just 0.8 percent of interest payments were due on October 15st and only 6.3 percent were due on October 31st. (These calculations include Treasury bonds and notes, but not T-Bills.)
The small share of interest payments due in October meant that financial markets could absorb the threat of default—however small—without causing too much disorder. Money market funds could request that collateral come from the vast majority of bonds that would not be in default if no agreement were reached by the end of the month. Repo markets could charge higher rates on the small share of bonds scheduled to receive a payment on these days without disrupting the market for Treasuries too much.
The key point is that the threat of default does not affect all Treasuries equally, especially if the default is expected to be short-lived. If the market expects the fiscal disturbance to last only, say, a few weeks, payments would be missed (or more likely delayed) on only those bonds due to receive payment in the given timeframe. While a missed or delayed payment would undoubtedly have dire consequences for all Treasury securities, the expectation of a missed or delayed payment appears to have more severe effects only for those Treasuries with payments due near the expiration of Treasury’s borrowing authority.
Under the last-minute agreement reached on October 17th, the debt limit is suspended through February 8th. Unlike last October, large shares of Treasury interest payments are in mid- and late-February. As shown below, about one-third—32.8 percent—of Treasuries are owed payments on February 15th and another 5.5 percent are owed payments on February 28th (see chart). This time, money market funds and the repo markets might have more trouble handling the threat of default.
With such a large share of Treasuries potentially becoming toxic from the perspective of repo market and money market funds, the worry is that yields might spike. Demand for Treasuries not owing interest on February 15th or February 28th could rapidly increase, and market players might seek to replace these Treasuries with other highly rated securities. If this happens, we could see the differentiation between those Treasuries with payments due in February 2014 and those without.
The folks at the Bipartisan Policy Center, who have been closely tracking government cash flows, estimate that a binding borrowing limit will be reached sometime towards the end of February or early March, but this date could be moved up if expected tax revenues fall short. The Congressional Budget Office expects the hard borrowing limit to be hit in March, but notes that this date could fall to April, May, or early June if tax receipts exceed expectations. Judging by the chart below, a mid-May borrowing limit would be much worse than April or June.
The country’s last go-round with the debt limit was resolved before catastrophe, perhaps indicating that members of Congress do indeed understand the dire consequences of failing to pay bills owed. But the real point here is that even kicking the tires on default might be costly, and that the costs will probably rise if the budget mess resurfaces in February or May. Let’s not find out
When it comes to property taxes, location matters. In a new TPC report, my colleague Brian David Moore and I look at just how much property taxes vary across states and counties.
Using self-reported American Community Survey data, we find that residential property taxes tend to be close to $1,000 per year, with a small share of households paying substantially more, especially in Connecticut, New Jersey, New York and New Hampshire. In recent years, 48 percent of homeowners paid between $750 and $1,750 in property taxes. About one-third—31 percent—paid less than $750 and 21 percent paid more than $1,750. Just 3 percent paid more than $4,000, with a miniscule share of homeowners (0.2 percent) paying more than $8,000.
These tax burdens vary by state and county. As shown below, seven states—Connecticut, Illinois, Massachusetts, New Hampshire, New Jersey, New York, and Rhode Island—had per household property tax collections in excess of $3,500. Five states—Alabama, Arkansas, Louisiana, Mississippi, and West Virginia—had per household property taxes of $750 or less. In dollar terms, property taxes tend to be highest in counties in the northeast and the west.
The counties with the highest property taxes paid per homeowner are those surrounding New York City. Westchester, Nassau, and Bergen counties had the three highest average tax burdens, all in excess of $8,500; this in part reflects higher house prices and higher reliance on property taxes to provide state and local services. In general, the counties with the highest property tax burdens tend to be in New York and New Jersey, with all but three of the top 25 counties being from these two states.
Some of these differences are related to house prices, as rising housing values can increase property tax burdens. As a share of housing price, property tax burdens tend to be less than 1 percent—60 percent of homeowners fall into this group. An additional 37 percent of homeowners had property tax bills between 1 percent and 2 percent of their home’s value, and just 3 percent had property tax bills in excess of 2 percent.
Residents in all but fifteen states pay between 0.5 percent and 1.5 percent in taxes. Residents in four states pay less, while residents in 11 states pay more. The states with the highest average property tax burden as a share of housing price are New Jersey (2.0 percent), Texas (1.9 percent), and New Hampshire (1.9 percent).
Moving forward, the Tax Policy Center will regularly update these calculations as additional data become available.
Tax preferences for housing are under fire, with mounting evidence that these preferences are inefficient, unequal, and too expensive to warrant a place in the tax code. Critics of proposed changes in the tax treatment of home ownership argue that these reforms would slash home prices at the very time they are showing signs of recovery. But in a new paper, I find that changes to the deductions for mortgage interest and property tax payments might not reduce prices much at all, and some reforms might even boost prices.
I estimate the impact on metropolitan housing prices of the higher tax rates imposed this year on high income households as well as three proposed tax changes: President Obama’s 28 percent limit on selected tax expenditures; eliminating deductions for mortgage interest and property taxes; and limiting the tax savings from the mortgage interest deduction to 20 percent while providing a flat credit for closing costs.
The president’s 28 percent limit on itemized deductions would barely move housing prices at all, causing them to fall just 0.3 percent. The higher top tax rates put in place in 2013, which drive up the value of deductions for housing for high-income taxpayers, are estimated to increase home prices by an even smaller margin.
By contrast, completely eliminating the mortgage interest and property tax deduction—a drastic change that probably would only happen if accompanied by a new tax preference for housing—would cause housing prices to fall by an average of 11.8 percent in the 23 cities studied. Estimated price declines would range from 10.3 percent in Seattle to 13.8 percent in Milwaukee.
Some proposed tax changes might even boost home prices. I estimate that limiting the mortgage interest deduction to 20 percent while offering a flat credit equal to two percent of a home’s value would raise prices an average of 3.0 percent. Price gains would vary from 1.5 percent in Pittsburgh to 4.6 percent in Portland.
Price effects could differ from my estimates, depending on how individual markets actually work. The effects could be lower, especially in the long-run, if the stock of owner-occupied housing adjusts to price changes—this can happen through either new construction or conversions between rentals and owner-occupied units. Or the price changes could be larger if metropolitan housing markets are driven by a marginal homebuyer in a high tax bracket, rather than being segmented across taxpayers in various tax brackets.
These findings diverge from prior studies, including one of my own, because they derive from a different way of thinking about housing investment. Earlier studies analyzed only the ongoing costs of homeownership—including mortgage interest, property taxes, and foregone return to housing equity. My approach adds transaction costs such as transfer taxes, title insurance, and realtor and broker fees to the costs of homeownership.
Transaction costs can dominate the cost of homeownership, especially for homes owned for short periods. For a home owned for less than four years, transaction costs are the largest expense, exceeding interest, property taxes, or foregone return to housing equity. For a home owned for 11 years—the median ownership across cities in the study—transactions costs still make up about one-fifth the total cost of homeownership (see chart).
The findings offer two important lessons for the tax expenditure debate. First, although completely eliminating the tax preferences for owner-occupied housing would reduce home prices substantially, curbing those preferences probably would not. Second, reform that limits deductions for mortgage interest while subsidizing the cost of housing transactions could actually raise home prices.
There are lots of barriers to a more efficient tax code. This research suggests that concern over lower home prices should not be one of them.
Until the Great Recession, state and local governments played a remarkably constant role through down business cycles. For four decades, when the economy turned sour, state and local governments boosted their spending—mitigating the depths of recessions and adding to growth when the economy revived. (Of course, this growth was partially offset by the negative effect of taxes collected to pay for that extra government spending.)
It was different this time. State and local governments addressed looming budget gaps by cutting expenditures. Instead of contributing to growth, their budget cuts dragged down the recession even further. And when the economy turned a corner in mid-2009, state and local government consumption (i.e., current spending on government programs like police and fire departments, education, and health) and investment remained depressed—a big reason why the recovery has been so weak.
In a new paper, my Tax Policy Center colleague Yuri Shadunsky and I show what happened. Never before had the state and local consumption and investment been negative three years into a recovery, but in 2009, it was down by about 4 percent (see chart immediately below).
What was different? First, the Great Recession was longer and more destructive than any other post-war recession. The severity meant that many state rainy-day funds, which had been replenished after the 2001 recession and are designed to help states weather downturns, were insufficient to protect against the slump. Second, state and local governments were reluctant to sufficiently raise taxes to cover the decline in revenue (many states did raise taxes through higher rates or base broadening, but these measures replaced only a fraction of the lost revenue due to the recession). And third, the steep decline in housing prices meant that property tax revenues collapsed, though this took a while because assessments often lag changes in market value.
Like the decline in state and local consumption and investment, this fall in property tax revenue was unprecedented. In prior recoveries, inflation-adjusted property tax revenue typically grew by 10 percent three years after the recession had ended. But in 2009, property tax revenue was down by 1 percent (see chart below).
These trends reinforce what millions of homeowners and the long-term unemployed already know: This recession and subsequent recovery are very, very different.
The president’s FY 2014 Budget would limit tax benefits for workers with high-balance retirement saving accounts. Although critics call the plan a blow to workers’ retirement saving, I consider the plan a smart way to roll back the billions in tax breaks that go to investors who don’t need tax incentives to save for retirement. (As a recent senior economist with the President’s Council of Economic Advisers, my support for this provision might not come as a surprise, but note that I didn’t work on this proposal during my tenure at the White House.)
Under current law, annual defined-benefit distributions are limited to $205,000 per plan. The president’s proposal extends the limitation to defined-contribution accounts like 401(k)s and IRAs and recognizes that, unlike in the past, individuals may have multiple pensions. If the combined value of a worker’s retirement accounts exceeds the amount necessary to provide a $205,000 annuity, they can no longer receive tax benefits for retirement saving. As under current law, the maximum benefit level would be indexed to the cost-of-living and would be sensitive to interest rates, which determine the price of an annuity. This year, the cap would affect individuals with defined-contribution account balances exceeding about $3.4 million.
The absence of a cap on defined-contribution accounts allows some high-income workers to shield large amounts of saving from tax. A worker and his employer can contribute up to $51,000 each year to a workplace retirement account (a worker can contribute up to $17,500 on their own) and a worker without a retirement plan can generally contribute $5,500 annually to an IRA. Limits are higher for workers over age 50, and contributions can be made regardless of an account’s balance. The president’s plan would disallow new contributions if account balances exceed the limit, although balances could still grow tax-free.
One analysis estimated that the cap would apply to only one in a thousand current account holders aged 60 and older and would eventually affect just one in a hundred current workers later in their careers. While there are caveats with the analysis—the data are for 2011 and do not include defined-benefit pensions—the point remains that the proposal would affect few workers now or in the future.
Wouldn’t the president’s limit discourage saving? Probably not. Research has found that tax incentives for retirement saving have only miniscule impacts on overall net saving—contributions to retirement accounts mostly represent saving that would have happened anyway.
Besides, whether the president’s limit would reduce incentives to save is the wrong question. We should be asking whether it’s worth nearly $10 billion in tax breaks over the next decade—the amount of revenue this provision would save—to encourage wealthy, mostly elderly households to save perhaps a little bit more. The answer, especially for those who think the tax code is too riddled with tax expenditures, is that it’s not.