Evidence for policymakers in real time: a blueprint

cube-no-animation-2

Evidence for policymakers in real time: a blueprint

By Thiemo Fetzer, Christina Palmou and Jakob Schneebacher

The last two decades have seen a Global Financial Crisis, the aftermath of the EU referendum vote, the COVID-19 pandemic and most recently an energy crisis triggered by Russia’s invasion of Ukraine. In the face of unexpected events, policymakers must design policies quickly.

Accurate economic data allows policymakers to make informed decisions in response to economic shocks. But relevant economic statistics and analysis are often published with a significant lag. Therefore, policymakers may be forced to rely on economic narratives provided by the media, interest groups or financial markets to guide them. These narratives may be oversimplified or biased. Our new ESCoE paper “How do firms respond to economic shocks in real time”, offers a blueprint for real-time, comprehensive analysis that considers the complexity of firm responses across adjustment margins and different parts of the economy.

Our blueprint has three building blocks:

  • The integration of already existing, high-frequency UK microdata sources to capture firm reactions
  • A pre-registered analysis plan to discipline our analysis
  • The application of advanced analytical techniques to explore heterogeneity in a systematic fashion and help test the evidence base behind prevailing economic narratives

In the process, we also document the challenges involved in measuring firm responses with existing public data sources, including discrepancies between survey and administrative data, source transparency and update frequency of administrative data, issues with industry classification and the differences between the legal and economic definitions of a firm.

Integration of diverse microdata sources

The integration of real-time survey data with administrative data is at the heart of our approach. We link the Longitudinal Business Database (LBD), which provides sparse but near-universal data on the UK business population, with rich, qualitative data for a smaller sample of firms from the Business Insights and Conditions Survey (BICS).

This combination of diverse data sources enables us to track multiple dimensions of firm behaviour—such as pricing strategies, capital investments, changes in cash reserves, and perceptions of insolvency—every few months during periods of economic turmoil. As a result, we can observe adjustments to shocks that occur within relatively short timeframes, such as weeks or months. This is an advantage over traditional methodologies that often depend on static, historical data. Using our linked, high-frequency data, policymakers can identify emergent trends across multiple adjustment margins in near-real time that might otherwise remain hidden.

Creating a pre-registered analysis plan

The second building block of our approach is our commitment to a pre-registered analysis plan. This is a document outlining the technical details of our study. Publicly defining our research framework and hypotheses in advance helps to promote transparency and accountability in our analytical process.

Where competing results vie for visibility, a pre-registration plan adds credibility to empirical research. Clearly describing our hypotheses and analytical strategies before engaging with the data enables other researchers and policymakers to validate our methods and results. This fosters a culture of reproducibility in economic research. The pre-analysis plan also allows us to examine the complexity of potential adjustments across multiple margins without data mining.

Our estimating strategy also stems from the pre-analysis plan: since we committed to a research design before touching the data, we couldn’t use data exploration to decide how to analyse it. Instead, we progressively saturated the specification with more and more fine-grained fixed effects and reported the entire suite of results. In other words, we looked at narrower and narrower comparisons of firms to understand where the energy shock really made a difference. This allows researchers and policymakers to understand the sensitivity of our results to methodological choices.

Applying advanced analytical techniques

Economic shocks will affect businesses in different ways. To understand these different responses, we used complex analysis techniques to look for patterns and similarities across different businesses. We used a method called a k-means algorithm to group similar businesses together. We looked at how much each business’s sales changed when energy prices went up or down. Businesses with similar reactions were grouped together. This helps us see patterns and understand how different industries respond to economic shocks. We then looked at how individual businesses within these groups reacted to the shock. We found that there were different types of reactions, or “archetypes,” among these businesses. Some businesses might have increased their prices, while others might have cut costs or borrowed more money.

This approach allows for a clearer understanding of the diversity of adjustments firms undertake, which often reflect underlying structural differences. For instance, Figure 1 shows that in response to the energy price shock, manufacturing firms where more likely to increase cash reserves and debt repayments than hospitality firms, and less likely to pass through price increases.

Figure 1 shows how different types of businesses, like manufacturing and accommodation and food services, reacted to a sudden increase in energy prices. The figure shows the impact of the price increase on different aspects of these businesses, such as their prices, costs, and financial health.

For example, we can differentiate between firms that act as “shock absorbers” and those that pass increased costs directly to consumers. In the paper, we show that response patterns are correlated with firm size, industry type, and operational efficiencies. This aspect is crucial; it acknowledges the complexity inherent in economic shocks, where simplistic one-dimensional cause-and-effect narratives often fail to capture the underlying dynamics at play.

By mapping these archetypes onto broader narratives, our research connects empirical evidence with storytelling—a critical component of effective policymaking. Policymakers by necessity often rely on simplified narratives to justify their interventions. Our framework helps illuminate the myriad ways that firms adjust to shocks, enabling the construction of more nuanced policy responses.

Why does this matter?

As firms navigate the complexities of economic shocks, understanding their responses can bridge the gap between empirical research and practical policy implementation, ultimately supporting more resilient policymaking. This paper provides policymakers with a blueprint to do so and highlights key challenges in economic measurement for researchers to address going forward.

 

 

ESCoE blogs are published to further debate. Any opinions expressed are those of the authors only and do not necessarily reflect those of the Competition and Markets Authority (CMA) or the Office for National Statistics (ONS). This paper uses ONS statistical research datasets via the Secure Research Service (SRS). Outputs may not exactly reproduce National Statistics aggregates.