Harvard Professor Martin Feldstein: Are US Middle-Class Incomes Really Stagnating?

 The challenge of raising the incomes of middle-class families has emerged as an important focus of the presidential election campaign in the United States. Everyone agrees that incomes at the top have surged ahead in recent decades, helped by soaring rewards for those with a high-tech education and rising share prices. And there is general support for improving programs – such as food stamps and means-tested retiree benefits – that help those who would otherwise be poor. But the public debate is largely about how to help the more numerous (and politically more important) middle class.

Here, much can be done by improving existing government programs: expanding market-relevant training, increasing opportunities for married women to join or rejoin the labor force, reducing the penalties in Social Security rules for continued employment by older workers, and changing tax rules in ways that will increase productivity and wages.

But, while strengthening such programs should be a high priority, we should not lose sight of how well middle-income families have actually done over the past few decades. Unfortunately, the political debate is distorted by misleading statistics that grossly understate these gains.

For example, it is frequently said that the average household income has risen only slightly, or not at all, for the past few decades. Some US Census figures seem to support that conclusion. But more accurate government statistics imply that the real incomes of those at the middle of the income distribution have increased about 50% since 1980. And a more appropriate adjustment for changes in the cost of living implies a substantially greater gain.

The US Census Bureau estimates the money income that households receive from all sources and identifies the income level that divides the top and bottom halves of the distribution. This is the median household income. To compare median household incomes over time, the authorities divide these annual dollar values by the consumer price index to create annual real median household incomes. The resulting numbers imply that the cumulative increase from 1984 through 2013 was less than 10%, equivalent to less than 0.3% per year.

Any adult who was alive in the US during these three decades realizes that this number grossly understates the gains of the typical household. One indication that something is wrong with this figure is that the government also estimates that real hourly compensation of employees in the non-farm business sector rose 39% from 1985 to 2015.

The official Census estimate suffers from three important problems. For starters, it fails to recognize the changing composition of the population; the household of today is quite different from the household of 30 years ago. Moreover, the Census Bureau’s estimate of income is too narrow, given that middle-income families have received increasing government transfers while benefiting from lower income-tax rates. Finally, the price index used by the Census Bureau fails to capture the important contributions of new products and product improvements to Americans’ standard of living.
Consider first the changing nature of households. From 1980 to 2010, the share of “households” that consisted of just a single man or woman rose from 26% to 33%, while the share that contained married couples declined from 60% to 50%.

When the nonpartisan Congressional Budget Office (CBO) conducted a detailed study of changes in household incomes from 1979 to 2011, it expanded the definition of income to include near-cash benefits like food stamps and in-kind benefits like health care. It also subtracted federal taxes, which fell from 19% of pretax income for middle-income households in 1980 to just 11.5% in 2010. To convert annual incomes to real incomes, the CBO used the price deflator for consumer expenditures, which many believe is better for this purpose than the consumer price index. The CBO also presented a separate analysis that adjusted for household size.

With the traditional definition of money income, the CBO found that real median household income rose by just 15% from 1980 to 2010, similar to the Census Bureau’s estimate. But when they expanded the definition of income to include benefits and subtracted taxes, they found that the median household’s real income rose by 45%. Adjusting for household size boosted this gain to 53%.

And, again, even this more substantial rise probably represents a substantial underestimate of the increase in the real standard of living. The authorities arrive at their estimates by converting dollar incomes into a measure of real income by using a price index that reflects the changes in the prices of existing goods and services. But that price index does not reflect new products or improvements to existing goods and services.

Thus, if everyone’s money incomes rose by 2% from one year to the next, while the prices of all goods and services also rose by 2%, the official calculation would show no change in real incomes, even if new products and important quality improvements contributed to our wellbeing. Indeed, the US government does not count the value created by Internet services like Google and Facebook as income at all, because these services are not purchased.

No one knows how much such product innovations and improvements have added to our wellbeing. But if the gains have been worth just 1% a year, over the past 30 years that would cumulate to a gain of 35%. And combining that with the CBO estimate of a gain of about 50% would imply that the real income of the median household is up nearly 2.5% a year over the past 30 years.

So the US middle class has been doing much better than the statistical pessimists assert. And with better policies, these households can do even better in the future.

project-syndicate.org