What is the future of official economic statistics?

What would the world look like without statistics? In recent times, issues such as the cost of living, health care, the labour market, and wellbeing are increasingly playing out not only in official briefings but also across social media platforms. There seems to be heightened public scrutiny of how statistical agencies capture demographic change, low-income pressures, and trade-offs facing both the private sector and governments in the US economy and beyond.

Economic statistics have long played a foundational role in public policy; yet the institutions responsible for producing them have been under growing pressure to deliver faster, more insightful data with limited resources, particularly in the US.

In a conversation following his keynote address at the recent UNSW-ESCoE Conference on Economic Measurement, Vipin Arora, Director of the Bureau of Economic Analysis (BEA), reflected on the evolution of the agency’s role, its core products, and how official statistics fit into an increasingly complex data ecosystem.

Interviewed by Dr Paul Schreyer, Research and Enterprise Director at the UK Economic Statistics Centre of Excellence, the discussion examined how BEA balances tradition with modernisation, public trust with private data innovation, and human expertise with emerging artificial intelligence (AI) tools.

Flagship statistics and who uses them

For economists, policymakers, and professional statisticians, official economic statistics, such as gross domestic product (GDP) and productivity growth, remain central to understanding economic activity, economic development, and the trajectory of the global economy.

Measures including the consumer price index (CPI), the unemployment rate, and other core metrics shaped macroeconomic analysis, monetary policy, and policy development, influencing decisions on interest rates, tariffs, and responses to geopolitical shocks and disrupted supply chains.

Discussing the US bureau’s core “flagship products”, Dr Arora identified three long-standing pillars of BEA’s work. “GDP is the flagship,” he said. “It's part of a broader set of statistics called the national income and product accounts.”

Photo gallery: The UNSW-ESCoE Conference on Economic Measurement 2025

He also pointed to international transactions accounts, including the balance of payments, which he noted were the agency’s longest-running continuous statistical product, as well as personal income statistics, which had emerged directly in response to policy needs.

Beyond these headline indicators, Dr Arora highlighted a lesser-known but increasingly valuable dataset. “We have these things called regional price parities,” he said. “They actually help allow you to look at price differences between states, and we do them down to the metro area.”

He described them as “the most underappreciated BEA product,” noting that once users understood their value, “they absolutely love them.”

So who are BEA’s key users? Policymakers at federal, state, and local levels are all users of BEA data, as are institutions such as Congress, the Federal Reserve, and the Congressional Budget Office. Financial markets and the press also rely heavily on BEA data.

But Dr Arora emphasised that broadening public access remained a personal priority. “I really want eighth-grade teachers at some middle school in Nebraska using BEA statistics to teach their classes.”

Doing more with less: leadership, modernisation, and process

As demand for more granular data intensifies worldwide, Dr Schreyer said resources for statistical agencies rarely increase. In agreement, Dr Arora said the challenge was global and longstanding. “Everyone around the world is struggling with this,” he said.

To address this challenge, he argued that leadership played a central, sometimes underappreciated, role. “Have we as leaders of different organisations been very clear about the vision for what we want to do? We can't be everything to everybody,” he said.

He emphasised the importance of setting clear priorities, assigning the right people to the right roles, and empowering staff to execute effectively. “You tell them your goals, and then you let them do what they do, and you stay out of their way,” he said.

Subscribe to BusinessThink for the latest research, analysis and insights from UNSW Business School

Modernisation, especially when it comes to updating information software, is a critical productivity enhancer, he said. While a significant investment in technology is required, it achieves long-term efficiency gains. “We've had a lot of success, particularly over the last 20 years, in being able to do more with less,” Dr Arora said.

He also emphasised the cumulative power of incremental improvements. “Just imagine if you made a 1% improvement in just a few of those processes, and everybody in the organisation did it,” he said. “You're going to see huge productivity improvements.”

Official statistics, private data, and the impact of artificial intelligence

Dr Schreyer raised the question of whether private-sector data sources – such as credit card transaction data or data from online platforms – could replace official statistics altogether. But Dr Arora disagreed with framing private-sector data as a form of competition, arguing that both public and private data providers had distinct strengths and limitations.

“There’s almost like a comparative advantage that each of us has in certain things,” Dr Arora said. “We want to leverage that as much as possible.”

Looking ahead, the conversation turned to (unsurprisingly) AI and its implications for the statistical profession. Dr Arora said he was not concerned about AI replacing statisticians. “I actually view it as a compliment,” he said.

He said AI could likely improve speed and efficiency in areas such as survey processing, imputation, and aggregation, freeing skilled staff to focus on higher-value analytical work. “I think it's going to help us do the things that we do better, faster,” he said.

He also highlighted the importance of ensuring official statistics were usable by AI systems themselves. “We need to make it as easy as possible,” he said, referring to data standards and structure.

Despite rapid technological change, he said he remained confident that the need for human judgment and expertise would remain central. “I don't think the next BEA director is going to be an algorithm,” he said. “I think it's going to help us become more productive.”