Efforts to level the informational playing field across firms — by mandating data-sharing or expanding access to consumer risk information — are increasingly at the centre of Europe’s digital regulatory agenda. Such policies aim to enhance competition by reducing informational advantages held by dominant players, but they also raise important questions about how such interventions may affect firms’ incentives to innovate, consumers’ welfare and privacy, and overall market efficiency. Initiatives like the Data Act, the Digital Markets Act, and the Second Payment Services Directive (PSD2) reflect a broader shift towards opening up data-driven markets to new entrants (Caffarra and Scott Morton 2021). These discussions have gained new urgency in light of Mario Draghi’s report on European competitiveness (Draghi 2024).

There has been a growing body of evidence showing how data access and privacy restrictions may shape innovation and competition. Early studies, such as Goldfarb and Tucker (2012), document the negative impact of privacy regulation on online innovation, while more recent work (Doerr et al. 2023) shows that privacy protection laws can help balance privacy protection with market competition. Other studies have examined the effects of data-sharing policies ex post: for instance, Babina et al. (2024) find that open banking initiatives stimulate fintech entry and innovation but also increase prices for consumers who are costlier to serve.

To date, however, there is little direct evidence on the consequences of information equalisation, the very outcome that data-sharing policies are designed to achieve. Focusing on imperfectly competitive selection markets, in this column, based on Cosconati et al. (2025), we examine how levelling the playing field in firms’ predictive capabilities affects competition and consumer outcomes.

The Italian auto insurance market: A case study

The Italian auto insurance market has several characteristics that simplify contract structures and highlight the role of pricing. Insurance is mandatory, and insurers cannot reject consumers. Contracts offer full liability coverage with virtually no deductibles, and the mandatory minimum coverage limits (over €6 million) are rarely binding in practice. Premiums are calculated using insurers’ proprietary algorithms based on risk factors such as age, location, vehicle type, and claims history. While Italy employs a public bonus-malus (BM) system with 18 risk classes, its granularity is limited: around 80% of drivers are in the best class, pushing insurers to develop more precise private risk scores.

We estimate a structural model of imperfect competition in the Italian auto insurance market. The model allows insurers to differ along three key dimensions: risk-rating precision, cost structures, and product differentiation.

The estimation reveals substantial heterogeneity on the supply side. These differences translate into comparative advantages that drive distinct pricing and targeting strategies. Firms with superior risk-rating precision tend to focus on low-risk customers (Figure 1), while less precise firms end up serving riskier pools. This pattern is consistent with cream-skimming: better-informed firms use their informational advantage to selectively attract safer drivers.

Figure 1 Who goes where: Rank correlation between information precision and average consumer risks across firms

Notes: Firms are ranked by information precision (highest to lowest on the x-axis) and average consumer risks (lowest to highest on the y-axis). The red solid line represents a linear fit between the two rankings. Firm IDs are displayed next to each dot.

Counterfactual results

We evaluate three counterfactual scenarios: full transparency (firms observe consumers’ true risk types), a centralised risk bureau, and a privacy-oriented regime limiting firms to basic public information.

When all firms have full knowledge of consumers’ true risk types, consumer surplus rises by 16.9%. A centralised risk bureau delivers nearly the same improvement, raising consumer surplus by 15.7%. In contrast, a privacy-oriented regime generates a more modest gain of 3.6%.

These gains are driven by a reduction in premiums, which fall by between 21.6% and 25.7% under the full-transparency and centralised-bureau scenarios. With more accurate and commonly shared information, firms engage in stronger head-to-head price competition.

Important distributional effects also emerge. Low-risk consumers gain the most when firms have access to detailed or shared risk data: under a centralised risk bureau, their surplus increases by over 78%, while high-risk consumers experience a reduction. With full information, this gap widens further due to more precise price discrimination. By contrast, high-risk consumers benefit from privacy regulation, which limits firms’ ability to differentiate across risk types and leads to more uniform pricing.

Firm profits decline on average, but the impact is highly uneven across firms (Figure 2). Firms with weaker predictive capabilities gain from data-sharing, as they can now compete using information previously unavailable to them. By contrast, firms with more precise risk models experience sizeable profit losses, reflecting the erosion of their informational advantage.

The figure makes this redistribution clear: gains are concentrated among less informed firms, while losses are largest for those at the top of the information hierarchy.

Figure 2 Differential impact on firms: Percentage change in profit relative to the baseline

Notes: Firms are ranked from the highest to the lowest risk-rating precision on the x-axis. The orange bars represent the percentage changes in profit under the centralised risk bureau scenario, while the blue bars indicate changes under the privacy regulation. Firm IDs are displayed next to each bar.

Matching and efficiency

Information policies influence not only prices but also how consumers are matched with insurers. When firms have access to full or centralised risk information, they can better identify the consumers they are most efficient at serving. This generates market segmentation based on comparative advantage.

As illustrated in Figure 3, sorting becomes markedly stronger when firms have access to richer information. Under full information or a centralised bureau, firms specialise more clearly: some concentrate on low-risk drivers, while others attract higher-risk customers. Under privacy regulation, instead, these patterns flatten out, and consumers are more evenly — and less efficiently — distributed across firms.

This difference in sorting is the key driver of the efficiency results. Perfect information reduces average costs by 3.7% (about €32 per contract per year), while a centralised risk bureau achieves roughly 40% of these gains (around €12 per contract). With over 31 million contracts in the Italian market, even partial improvements in matching produce substantial aggregate benefits.

Together, Figures 2 and 3 highlight two sides of the same mechanism: information equalisation redistributes profits across firms while improving how consumers are allocated across them.

Figure 3 Sorting patterns: Average risk levels among consumers (measured in euros) within each firm under the baseline and three counterfactual scenarios

Conclusions

As regulators grapple with how to govern data in competitive markets, the case of auto insurance provides a sharp lens into the trade-offs at stake. Policy interventions such as the creation of a centralised risk bureau can generate substantial welfare gains for consumers by strengthening competition and improving market efficiency through better matching, even without fully eliminating information asymmetries.

At the same time, these benefits come with important caveats. By narrowing the informational advantage of the most sophisticated firms, data-sharing policies may weaken incentives to invest in better risk-prediction technologies. The findings thus highlight a key trade-off between promoting competition through information sharing and preserving incentives for innovation and privacy safeguards.

References

Babina, T, S Bahaj, G Buchak, F De Marco, A Foulis, W Gornall, F Mazzola and T Yu (2024), “Customer data access and fintech entry: Early evidence from open banking”, VoxEU.org, 17 July.

Caffarra, C and F Scott Morton (2021), “The European Commission Digital Markets Act: A translation”, VoxEU.org, 5 January.

Cosconati, M, Y Xin, F Wu and Y Jin (2025), “Competing under Information Heterogeneity: Evidence from Auto Insurance”, forthcoming, Review of Economic Studies.

Crawford, G, N Pavanini and F Schivardi (2015), “Asymmetric information and imperfect competition in lending markets”, VoxEU.org, 30 April.

Doerr, S, L Gambacorta, L Guiso and M Sanchez del Villar (2023), “Privacy regulation, fintech lending, and financial inclusion”, VoxEU.org, 16 August.

Draghi, M (2024), Report on the Future of European Competitiveness, European Commission.

European Insurance and Occupational Pensions Authority (EIOPA) (2021), “Open Insurance: Accessing and Sharing Insurance-Related Data – Discussion Paper”, 28 January.

Goldfarb, A and C Tucker (2012), “Privacy and innovation”, Innovation Policy and the Economy, 12(1): 65–90.

Saal, M, H Natarajan, J Frost, L Gambacorta and E Feyen (2021), “A policy triangle for Big Techs in finance”, VoxEU.org, 23 October.