WASHINGTON, July 15 (Reuters) - The U.S. Federal Reserve could more accurately estimate underlying inflation by including food and utility prices in its "core" Personal Consumption Expenditures Price Index, capturing a greater share of household spending and better reflecting people's experience with the cost of living, according to St. Louis Federal Reserve research released on Wednesday.

The overhaul would exclude from underlying inflation only energy goods, including things like gasoline, that are tightly tied to global oil prices and subject to the most volatile short-term swings, making it appropriate for the central bank to ignore them in making interest rate decisions aimed at keeping inflation at 2% over the medium to long term, the research concluded.

Measures of core inflation currently exclude food purchased to prepare and eat at home, energy services that largely reflect utility prices, and energy goods like fuel, on the grounds that those are tied to volatile commodity markets and move in ways that don't necessarily reflect overall supply and demand conditions in the economy - the thing the Fed is trying to keep in balance.

But, in the case of the PCE Price Index, which the Fed uses to set its inflation target, that means excluding about 13% of consumer spending and also divorcing the measure from some of the most sensitive household costs like grocery bills - a fact that many Fed policymakers agree is hard to explain or even justify when food costs are rising.

Since it is energy goods that show by far the most volatility, with food and energy services behaving more like other items, Fernando Martin, a senior economic policy advisor at the St. Louis Fed, argued that central bankers could get a better measure of underlying inflation by excluding only energy goods, which make up less than 3% of the consumer basket.

'SMOOTHER AND MORE REPRESENTATIVE INFLATION MEASURE'

Current measures of core PCE remove "prices that are too volatile and may mask underlying inflation, at the cost of ignoring a substantial share of consumer spending," Martin wrote in a blog post about his research.

Excluding only energy goods "creates a smoother and more representative inflation measure, as these goods are highly volatile and closely linked to oil prices," he wrote. He said the proposed new index "preserves a larger share of consumer spending ... without overreacting to short-term shocks."

Based on the most recent data, for May, Martin's revised PCE data would show a headline reading of 3.36% versus the current 4.1% and a core number of 3.42%.

Martin told Reuters the approach was similar to using "a scalpel" compared to the current method for calculating core inflation.

"What is the minimum amount that you can take out that remains representative?" he said. "Month to month what is the simplest way to filter out the noise?"

His proposal is relevant to an ongoing debate within the U.S. central bank about how best to measure inflation for the purpose of setting monetary policy, an issue Fed Chairman Kevin Warsh has assigned to one of five task forces.

Before becoming Fed chief in late May, Warsh said he favored measures like the "trimmed mean" that exclude on a month-to-month basis the items that show the most price volatility, whether higher or lower.

But in its common formulation, that measure excludes an even greater 55% of the spending basket and, Martin noted, shows key turning points "with a significant lag. ... In the last 12 months, trimmed mean inflation has been trending down rather than up. Altogether, it is a series that appears to struggle."