The poor often spend more on all kinds of things. Households that have less money to spare in any given week, for example, are forced to buy toilet paper and similar goods in small packages, increasing the prices they pay. In addition, poor families must rely on a whole range of alternative financial services, which might charge exorbitant fees and expose customers to serious risks.

New research suggests that these disparities might only be getting worse.

Xavier Jaravel, a graduate student at Harvard University, has been studying the prices people pay in the retail sector -- their everyday purchases at grocery and drug stores. He has found that prices are increasing by more than 2 percent a year on average for goods purchased by consumers with household incomes under $30,000, but by just 1.4 percent annually for those with incomes above $100,000.

While apparently small, that divergence -- if it continues -- would become hugely important in a relatively short period of time. After 20 years, for example, every dollar in the pocket of a poor consumer would be worth just 88 cents compared to what a wealthier consumer would be able to buy with it at the grocery store, given the differences in inflation and in both consumers' preferences.

Jaravel's research is preliminary, and his data includes only information on prices in one sector of the economy, excluding major household expenses outside of retail trade. All the same, the numbers are striking.

"If you’re one of the consumers who is stuck at the bottom of the distribution, then you suffer more than you would think," Jaravel said.

He used data from Nielsen on retail prices and household incomes from 2004 to 2013 to determine how quickly prices increased for the rich and the poor. First, he measured the change in price for identical products -- identical models from the same manufacturer -- from one year to the next. The products purchased mainly by the poor, he found, increased in price much more quickly than those purchased by the wealthy.

This difference wasn't a result of the fact that the poor spend proportionally more on certain broad categories of goods, such as produce or paper towels. The differences in inflation between these categories weren't all that large for different groups of consumers.

Rather, Jaravel found that the difference resulted from the fact that within specific kinds of products -- beef jerky, say, or tortilla chips -- more affluent consumers were buying premium brands, and those brands' prices were more stable over time.

"It’s not that some buy wine and others buy beer," Jaravel said. "Some are going to buy craft beer, and others will buy Budweiser or Bud Lite."

Jaravel next examined the products that manufacturers introduced in each year, and he found that relatively few new products made it into stores that weren't premium goods. In other words, poorer consumers were more likely to be buying the same products from year to the next.

When a new product arrives on shelves, retailers typically have to discount the price of older products somewhat. Using data on markups from a large national grocer with a few hundred stores, Jaravel found that the absence of new products for poorer consumers allowed grocers to increase the prices on the old products more from year to year.

The fact that prices in the retail sector seem to be increasing more quickly for one group of consumers than another is worrisome, but Jaravel's explanation for the trend is even more disconcerting.

He argues that as society becomes wealthier in general, and as the rich in particular become wealthier because of increasing inequality, the market for premium retail goods expands. Manufacturers compete with one another to sway consumers' preferences in this segment, busily inventing new ways for the affluent to dispose of their income.

By contrast, there is less new territory up for grabs in the market for bargain goods. Innovation is not a priority for manufacturers, resulting in increased prices for the poorer consumers who shop in this segment.

As a result, Jaravel said, inequality engenders more inequality. Earnings for the poor and the middle class aren't increasing as rapidly as they are for the rich, but the prices they pay are going up more quickly.

Jaravel's research contradicts a couple of assumptions that economists typically make about consumer goods. One is that when new kinds of products are introduced, they'll be purchased mainly by the wealthiest consumers, who will pay higher prices for them. Over time, prices will fall, and less affluent consumers will purchase the goods at lower prices.

Jaravel found little evidence of that pattern in the data. Whole categories of new, premium products never made it into the shopping cart for poorer families.

He noted that trends in the rest of the economy could be different. The data from Nielsen is representative of only about 16 percent of household expenditures. Families spend heavily on health care, rent, energy and education, too.

Yet if further research corroborates Jaravel's findings so far, economists may have to change the way they think, a whole range of government policies.

As an example, Jaravel said his work supports the case for the national food-stamp program, regardless of trends in the rest of the economy. By expanding the market for inexpensive goods at groceries, food stamps likely encourage manufacturers to devote more resources toward innovation in this segment in this market, reducing inflation and creating additional benefits beyond the food stamps themselves that will accrue over time for poor families.

Measuring inflation is a messy business. Economists have spent decades debating how best to compare prices for one year to another as technologies and consumers' preferences evolve. Jaravel's research suggests solving the problem will mean accounting for inequality, too.