What to watch for in the 2018 Census data on earnings, incomes, and poverty

Next Tuesday is the Census Bureau’s release of annual data on earnings, income, poverty, and health insurance coverage for 2018, which will give us a picture of the economic status of working families 11 years into what is now the longest economic expansion in United States history. This data is particularly important because it gives us insight into how evenly (or unevenly) economic growth has been distributed across U.S. households. Other data sources that are released more than once a year too often provide only averages or aggregates— but next week’s Census release gives a much more textured picture of how the U.S. economy is working for typical households. In particular, next week’s release will help us chart the progress made by the typical American household in clawing back nearly two decades of lost income growth—the result of a failure of incomes to return to the business cycle peaks of 2000 during the slow early-2000s recovery and expansion, and the Great Recession. We’ll be paying particular attention to differences in the recovery across racial and ethnic groups.

What happened with incomes in recent years?

After adjusting the series to account for changes to the survey made in 2013, in 2017 real (inflation-adjusted) median incomes for American households rose just 1.8 percent and only managed to return to their pre-Great Recession peaks, even coming off of two years (2015 and 2016) of impressive across-the-board improvements. It is important to note, however, that some of the improvements in inflation-adjusted income we saw in 2015 and 2016 were driven by atypically low inflation—0.1% in 2015, and 1.3% in 2016. We didn’t get a similar boost from low inflation in 2017 (inflation increased 2.2% in 2017), and don’t expect one in 2018 (inflation increased 2.4% in 2018). We anticipate that an additional year of even modest growth will likely bring the broad middle class back to 2000 incomes. But, for non-elderly households, the latest data will be likely still below the peak reached 18 years prior.


Real median household income, all and non-elderly, 1995–2017

All households All households- imputed series All households- new series Non-elderly households Non-elderly households- imputed series Non-elderly households- new series
1995 $54,600 $56,330 $62,727 $64,677
1996 $55,394 $57,150 $63,898 $65,885
1997 $56,533 $58,325 $64,722 $66,734
1998 $58,612 $60,470 $67,372 $69,467
1999 $60,062 $61,966 $69,079 $71,226
2000 $59,938 $61,838 $69,419 $71,577
2001 $58,609 $60,466 $68,324 $70,448
2002 $57,947 $59,784 $67,650 $69,753
2003 $57,875 $59,709 $67,031 $69,115
2004 $57,674 $59,502 $66,246 $68,305
2005 $58,291 $60,138 $65,792 $67,837
2006 $58,746 $60,608 $66,698 $68,772
2007 $59,534 $61,421 $67,015 $69,098
2008 $57,412 $59,232 $64,817 $66,832
2009 $57,010 $58,817 $63,932 $65,920
2010 $55,520 $57,280 $62,280 $64,217
2011 $54,673 $56,406 $60,775 $62,664
2012 $54,569 $56,298 $61,346 $63,254
2013 $54,744 $56,479 $56,479 $61,605 $63,520 $63,520
2014 $55,613 $62,667
2015 $58,476 $65,541
2016 $60,309 $67,917
2017 $61,372  $69,628
ChartData Download data

The data below can be saved or copied directly into Excel.

Note: Because of a redesign in the CPS ASEC income questions in 2013, we imputed the historical series using the ratio of the old and new method in 2013. Solid lines are actual CPS ASEC data; dashed lines denote historical values imputed by applying the new methodology to past income trends. Non-elderly households are those in which the head of household is younger than age 65. Shaded areas denote recessions.

Source: EPI analysis of Current Population Survey Annual Social and Economic Supplement Historical Income Tables (Tables H-5 and HINC-02)

Copy the code below to embed this chart on your website.

What do we expect in this year’s release?

Given the data we’ve seen for 2018 from other sources, it is likely that earnings, income, and poverty in the 2018 Census data will show some improvement over the past year. But it is also likely that this pace of improvement will be significantly slower than the average of the previous three years. As the economy steadily strengthens, we’ve seen progress in key labor market indicators, including participation in the labor market and payroll employment, which should boost household labor earnings. The unemployment rate ticked down another 0.5 percentage points in 2018, similar to the drop between 2016 and 2017. The overall labor force participation rate was unchanged between 2017 and 2018, but the employment-to-population ratio continued to increase, 0.3 percentage points overall and 0.8 percentage points for the prime-age population (25-54 years old). These are similar to the increases found between 2016 and 2017.

However, our earlier analysis of hourly wage from the Current Population Survey (CPS) data—of which Tuesday’s release is a supplement to—suggests that real wage growth in 2018 continues to be unequal and slower than expected at this point in the business cycle. In 2018, strong growth in hourly wages continued at the top (2.7% at the 95th percentile), while the 20th percentile saw the strongest growth at 4.8% due in part to a tightening labor market as well as state-level minimum wage increases. However, median wages grew only 1.6%.

On Tuesday, we will compare changes earnings, income, and poverty against several benchmarks: over the last year, as well as changes since before the Great Recession and since 2000—the last business cycle peak that can be confidently associated with something close to genuine full employment. Women and men working full-time, full-year, both experienced earnings losses between 2016 and 2017. As of 2017, full-time men had yet to return to 2007 or 2000 levels of earnings. We’ll also analyze these changes by race and ethnicity to understand how the economy has treated different demographic groups. Again, the hourly wage data are likely the best predictor of what we can expect for these groups. We will also analyze gender and racial wage gaps to see whether we’ve made any progress in closing these among full-time, full-year workers. If the hourly wage data are any indication, we expect little change, but perhaps a mild narrowing of the gender wage gap as well as a mild widening of the black-white wage gap.

Second on our agenda will be an examination of trends in median household incomes. Again, we will be looking at these data across a range of households: all households, non-elderly households, and by race and ethnicity. As people continue to return to the labor force and get jobs, we should see improvements in incomes, since labor income is the primary form of income for non-elderly households in the middle of the income distribution. Even if individual earnings do not improve significantly, more working members of a household will increase household incomes.

(Also, just a side note on the income data, because of the redesign in 2013, we make an imputation to the historical series, 2000 to 2012 to make them directly comparable to the latest income data. This entails creating a ratio of the original and redesigned 2013 income data within each demographic subgroup, and imputing that backwards to create a consistent series. This is the same adjustment we made last year measuring elderly and nonelderly household incomes as well as incomes by race and ethnicity. The Census Bureau also announced a processing change that will affect data years 2017 and 2018, making them directly incomparable to prior years. Though this is most likely to have the largest effect on health insurance data, incomes and earnings trends are also affected.)

In addition to looking at how median household income growth differed by race and ethnicity, we will examine changes in incomes across the income distribution. Specifically, we will be presenting the growth in income by income fifth and the top 5 percent to assess how much inequality has grown or shrunk over the last few years. Unfortunately, again, the hourly wage data through 2018 indicate that top earners have continued to pull away from the median, and therefore growth in income inequality is likely. Last year, while there was broad-based income growth, the bottom 40% of households still had incomes below their 2000 income levels.

Third, we will provide an analysis of recent trends in poverty. Similar to the previous discussion, we will analyze poverty in 2018 and then make comparisons to 2007 and 2000. We’ll also look at poverty by race and ethnicity, and separately for children—who tend to have particularly elevated levels of poverty. Assuming incomes continue to rise, particularly if there’s broad-based growth, poverty will hopefully continue to fall.

In addition to the official poverty rate, we will highlight recent trends in the Supplemental Poverty Measure (SPM) being released on Tuesday. The SPM is an alternative to the long-running official poverty measure that attempts to correct some of the substantial weaknesses of the official poverty measure (e.g., the fact that it counts only cash income, often sets possibly too-low thresholds for poverty, and doesn’t allow for geographic variability). Because it includes non-cash measures of family income (like food stamps) and the effect of refundable tax credits, the SPM lets us assess how effective public assistance and safety net programs are at lifting people out of poverty. On Tuesday, we will take a deeper look at these data to make an assessment of how poverty, as measured by the SPM, has changed in the recovery and the importance of shoring up not tearing down these programs in future policymaking.

It is important to remember that the data reported next week is the latest installment in just one of the Census Bureau’s many long-running series that have been essential to assessing the needs of the American people and the effectiveness of national policies. It’s a reminder that fully funding the Census Bureau and all of its useful surveys is critical for accurate measurement of the population and providing necessary insights into the role of policy in improving incomes and reducing poverty.