States and cities have fallen over themselves to offer huge tax breaks to Amazon in the hope of securing one of the tech giant’s order fulfillment centers. But an analysis of the impact of the centers released on Thursday found the facilities do not boost overall employment in the counties where they open.

After analyzing data for counties in 25 states containing Amazon fulfillment centers, the Economic Policy Institute (EPI) found that within two years the centers lead to a 30% increase in warehouse and storage employment in the surrounding county. But the analysis also found no increase in overall employment in the county and, in some cases, the data suggested a reduction in overall employment.

Amazon has received over $1bn in state and local subsidies to open fulfillment centers across the US. The report questions the efficacy of such tax breaks if they do not contribute to a net growth in jobs.

“If policymakers instead invested in public services – particularly in early-childhood education and infrastructure – that would be a much stronger recipe for long-term economic development, rather than giving tax breaks to national employers like Amazon,” said EPI economist Ben Zipperer.

Zipperer and economic analyst Janelle Jones speculate that the jobs created by the centers are either being offset by job losses in other industries or that the growth is too small to affect the jobs figures.

The report comes as Amazon has shortlisted 20 US cities for its $5bn second headquarters. Many of the cities on the list, such as Detroit and Newark, have offered huge tax breaks in order to lure Amazon.

As cities and counties compete to host new Amazon facilities and its additional headquarters, policymakers should be cautious about “giving away the store”, said Jones. “Instead of pre-committing to giving away public funds to attract employers, communities should demand a concrete demonstration that an employer’s arrival will make their region a more prosperous place for working people.”

Amazon disputed the report saying it was “misleading as it includes the time in which communities struggled during the recession and when Amazon was not building out its network of fulfillment centers as it is today.”

“In addition to the 200,000 Amazon employees in the US, we know from 2016 data, which is more current than the EPI data, Amazon’s investments led to the creation of 200,000 additional non-Amazon jobs, ranging from construction jobs to healthcare industry positions. In fact, over the last five years, counties that have received Amazon investment have seen the unemployment rate drop by 4.8 percentage points on average, and in some areas, the rate has been lower than the state average,” the company said.

Zipperer said he would expect unemployment to fall - in areas with or without Amazon fulfillment centers - as the recovery in the jobs market continues.

He said EPI’s calculations had factored in differences in local and state jobs’ numbers and that EPI would continue to assess Amazon’s employment record as numbers became available.