Debt-fueled consumption is a real concern at the median (visible in the post-crisis decline as people repay debts and worry about the economy) but is not a big driver of increased consumption at the bottom near the poverty line, where little borrowing occurs.

The popular claims that the middle class is being hollowed out, incomes are stagnating for all but the top, and the rising tide did not lift all boats are basically wrong, as shown by ten years of research with James Sullivan; properly accounting for inflation, median incomes have risen about 50% since 1980, and consumption has risen by a similar amount.

The picture of well-being looks much better if you account for what people can actually spend (after-tax income); taxes matter substantially at the median, particularly in the 2000s, because of the 2001 and 2003 income tax cuts which were spread fairly evenly across the income distribution in percentage terms rather than going entirely to the rich.

The CPI overstates inflation because it fails to adequately keep track of new goods and to adjust for quality improvements; the Boskin Commission concluded in 1996 it overstated inflation by about 1.1 percentage points per year, and members polled more recently still estimate a bias of about 0.7 to 0.9 percentage points per year.

It took the BLS about 15 years to include cell phones in the CPI basket after their introduction, and historically most goods (electric refrigerators, used cars) were included only long after the majority of people owned them; because new goods often fall dramatically in price, leaving them out misses a large component of real price decline.

An independent way to confirm CPI overstates inflation is to find the income adjustment that keeps spending patterns (shares spent on food, clothing, leisure) constant over time at a given real income level; researchers like Costa and Bruce Hamilton have done this and arrive at CPI biases not very different from the Boskin Commission's.

Physical ownership data show the bottom 20% are materially much better off than in 1980: in 1980, 27% of those in the bottom income quintile had central air, versus 67% in 2009 (nearly triple), with similar gains in dishwasher, washer, and dryer ownership; structural housing problems like roof leaks and plumbing problems have nearly disappeared.

Housing is about 40% of consumption across income levels, and the size of housing units has gone up dramatically while problems like roof leaks and plumbing defects have nearly disappeared, undercutting the claim that gains are merely cheap electronic 'toys' while big-ticket items got more expensive.

Healthcare is deliberately excluded from Meyer and Sullivan's research because there is less professional agreement on how to handle the tremendous quality changes; children's access (e.g., share with annual checkups) has improved, but the share of median-income people with insurance has declined over time, which would offset some gains if access to better technology fell.

Education has become more expensive primarily because demand rose sharply (more people attending high school, college, community college, and graduate school) against a relatively inelastic supply, so an increase in demand pushes price up a lot relative to quantity; this reflects greater access, not unaffordability.

Most charts circulated by the Census Bureau and media report pre-tax income and ignore transfers (welfare, etc.), and poverty is measured by people's own earnings rather than living standard after transfers; correcting for taxes and transfers substantially changes the picture, especially at the bottom.

Most of the measured 50% improvement at the median comes from the inflation adjustment, but a substantial part (particularly in the 2000s) comes from tax cuts; for those in the bottom 10-20%, transfers matter much more than taxes, and the combination of raising exemptions and expanding the Earned Income Tax Credit dramatically increased after-tax incomes of the poor relative to pre-tax income.

Comparisons of the median or quintiles across decades are misleading because they are not the same people over time; in particular, the rising divorce rate since the 1970s splits households so that household-level analysis can show median income falling even when the original individuals are better off—a statistical paradox often ignored by those with an agenda.

Over the last 10 years, poverty measured by consumption has fallen much faster than poverty based on income (the official income measure shows poverty rising), and longer-term indicators like housing types and cars driven suggest the bottom has seen real improvement in living standards despite the awful last three years.

Meyer and Sullivan disaggregate by household type (married-parent, single-parent, single families without kids, married couples without kids, 65+ headed) and by race, employment, and education, then aggregate; doing so yields different answers than lumping everyone together, but they find changes in family composition do not play a big part in changes in median income or poverty over their period.

The Hoynes, Page, and Stevens study found that between 1967 and 2003 every single demographic group's poverty rate fell by double digits (most by 20% or more, except single men), yet the overall poverty rate fell only about 4% (and official measures show rises) because the highest-poverty group—single women with children—grew as a share of the population (Simpson's paradox).

Poverty has declined much faster for households headed by someone 65 and older and for single-mother-headed households than for the overall population, while married couples with children have seen their poverty rate decline more slowly than other groups—even more so when measured by spending rather than income.

Consumption is a better measure of well-being than one year's income because families save for rainy days and can consume out of past savings; this is especially true for retired households (over 80% of those headed by someone 65+ own their home and at least one car), whose low current incomes badly understate their well-being.

For people at the bottom, income is hard to measure because it comes from many irregular sources (formal/informal work, money from family, friends, boyfriends, fathers of children, and government transfers), as documented by Edin and Lein; these same people report rent and weekly food spending fairly accurately, so consumption gives a better read of their living standard.

Surveys like the Current Population Survey poorly capture government transfers: matching CPS data to government records shows the survey captures only about half of food stamps actually paid out, not necessarily from shame but because respondents skip the laborious reporting of benefits, so official income/inequality data understate the well-being of the poor.

There was a large increase in inequality in the late 1970s and early 1980s by every measure, but since the late 1980s—when measured with after-tax income or better yet consumption—inequality has grown only modestly for the bulk of the population, and the 10th percentile has actually risen relative to the median, especially in the last 10 years.

The after-tax 90/10 income inequality ratio in 2009 was about the same as in 1993 (dipping in the 1990s, rising in the 2001-02 recession), largely because the 10th percentile did quite well after accounting for taxes, including payroll taxes; meanwhile the 90/50 ratio rose from about 2.5 around 1990 to roughly 2.75 today—a relatively small increase compared to public perception.

Much of the increase in inequality has been attributed to rising demand for high-skilled labor combined with the failure of the US to supply enough high-skilled labor; minimum wage changes and high early-1980s inflation are other suggested but uncertain factors.

Globalization and technological change both reduced demand for low-skilled US workers (their products are made overseas or by machines), and because the US K-12 school system is mediocre and has not responded vigorously to prepare strugglers for college, this is part of the inequality problem—yet because most of the period shows little inequality increase, globalization and immigration are not the threats to living standards that some claim, especially since their pace intensified precisely when inequality was not rising.

Inequality is not an inexorable fact but emerges from a complex process with good and bad sources: good inequality comes from value-creating entrepreneurs like Brin and Page who made others better off while capturing only a fraction of the value; bad inequality comes from bailouts of Wall Street and subsidies to rich farmers.

Many in the top 1% are entertainers, movie stars, sports figures, and entrepreneurs who have leveraged their skills via improved technology so they are more valuable than before, alongside some who earn through subsidies (e.g., subsidized borrowing/lending by large financial institutions), so the group is heterogeneous rather than simply an exclusive club extracting money.

Policy should keep its eye on what sustains high long-term economic growth (which has been pretty widely distributed), distinguish recession-appropriate policies from long-term growth policies, and avoid using false claims of stagnation to justify measures like closing borders to goods and immigrants.

The pace of improvement in daily life is itself what makes living standards hard to measure and creates the false impression of stagnation; in stagnant eras like the 1950s and 1960s cars and refrigerators looked similar decade-to-decade, whereas now everything improves within ten years, leading people to wrongly conclude they aren't catching up because the improvements are mismeasured.

The 1950s and 1960s were relatively dull and stagnant with little innovation, so the popular nostalgia that those decades were 'more fair' confuses fairness with a lack of dynamic improvement.

A homeless person having a heart attack will receive emergency-room care in both 1980 and 2011 (required by law), but the drugs, treatment, and technology available are much better today, illustrating that healthcare quality has improved dramatically even at the bottom.

Looking around at your immediate environment is a dangerous basis for inference because it produces misperceptions—e.g., concluding a recession isn't bad because a thriving Washington DC mall is crowded.

Low-income workers pay a healthy amount of payroll tax as a percentage of income (because it is effectively a flat tax), and those rates have risen over time, so analyses must account for federal and state income taxes and payroll taxes when measuring after-tax income.

Mark Bils at Rochester has done careful work estimating quality-change bias on manufactured goods, finding the bias is quite significant for that important subset of goods, demonstrating that systematic correction is possible for narrow categories even if hard to generalize.