Macroprudential policy is designed to make the financial system safer, but it can also reshape the finance available to innovative firms. Using patent data matched to 2,844 firms in 21 European countries over 1990-2021, this column finds that macroprudential tightening is followed by lower patenting and lower patent quality. The effect is strongest when credit growth is weak and among financially constrained firms, and it is concentrated in instruments that directly restrict credit. By contrast, financial-resilience tools carry no measurable short-run innovation cost and may support innovation over longer horizons. The findings suggest that policymakers can protect financial stability without systematically penalising innovation, provided they are selective about which instruments they deploy.
Macroprudential policy has become a central pillar of the post-crisis regulatory framework, designed to limit systemic risk by restraining excessive credit growth, strengthening bank balance sheets, and tightening borrowing conditions (Akinci and Olmstead-Rumsey 2018). These objectives are indispensable when financial vulnerabilities build up. Yet the same measures inevitably affect the supply and cost of external finance available to non-financial firms – and through that channel, the capacity of economies to innovate.
The spillover to innovation is especially consequential in Europe, where financial systems remain predominantly bank-based. Innovative investment is by nature intangible, risky, slow to generate returns, and difficult to pledge as collateral – precisely the characteristics that make it vulnerable when lenders grow more selective. Faced with tighter credit conditions, firms may postpone research programmes, discontinue early-stage projects, dissolve development units, or redirect scarce resources toward shorter-horizon activities (Kerr and Nanda 2015). The policy question is therefore not whether financial stability or innovation matters more, but whether the design and calibration of macroprudential instruments can serve both objectives simultaneously.
In a recent paper (Ho et al. 2026), we examine this question across 21 European countries from 1990 to 2021 – a setting that provides rich and meaningful variation in policy. While the European Union operates a common macroprudential framework under the single rulebook of the Capital Requirements Regulation and Capital Requirements Directive IV (CRR/CRD IV), national authorities retain discretion over the timing and intensity with which individual instruments are activated. Figure 1 illustrates how the average frequency of annual tightening episodes intensified following the framework’s introduction.
Figure 1 Macroprudential tightening increased across European countries after CRR/CRD IV
Our study links country-level macroprudential policy data from the IMF iMaPP database (Alam et al. 2025) with firm-level financials from LSEG Worldscope and patent data from Orbis Intellectual Property, yielding a sample of 401,755 granted patents and 588,266 forward citations across 2,844 innovative non-financial firms. Patent counts measure innovation quantity; adjusted forward citations measure quality, benchmarking each firm’s citations against peers in the same industry, country, and year. Macroprudential instruments are classified along two dimensions: credit-oriented versus non-credit-oriented (Gonzalez 2022), and credit cycle-smoothing versus financial resilience-building (Claessens et al. 2013).
Identification is non-trivial, since policymakers typically tighten when economic and financial conditions are themselves unfavourable to innovation. We address this through three complementary strategies: multiway industry, country, and year fixed effects; an instrumental-variable design exploiting variation in the policymaking-autonomy component of central bank independence; and a difference-in-differences estimator leveraging the staggered national activation of CRR/CRD IV. Estimates are negative and consistent across all three approaches, under both OLS and Poisson pseudo-maximum likelihood estimation, and survive a broad range of robustness checks including alternative specifications, functional forms, additional controls, exclusion of crisis years, and placebo exercises.
The magnitude of effect is noteworthy. A one standard deviation increase in the overall macroprudential policy index is associated with 0.263 fewer patents and 0.518 fewer adjusted forward citations per firm-year – declines of approximately 10.1% and 8.4% relative to sample means. Tighter policy thus reduces not only the quantity of innovation but also its quality.
Figure 2 The innovation response to macroprudential tightening
The effects are persistent. Patent counts remain depressed for up to five years after a tightening episode, with the largest impact in the first three years; innovation quality is significantly affected for roughly the same window. These lags reflect the funding structure of research: a credit contraction disrupts the innovation pipeline immediately, while lost patents and citations materialise only later. Crucially, easing after a tightening does not reverse the damage – consistent with the difficulty of recreating a discontinued research programme or rebuilding a dissolved team.
The evidence points to a credit channel, consistent with prior research showing that credit supply affects innovation (Amore et al. 2013). The negative effect is stronger when credit growth to non-financial firms is weak and for firms facing greater financial constraints. With less internal liquidity, weaker collateral, and fewer alternatives to bank finance, these firms are more likely to cut risky, long-term innovation projects when credit conditions deteriorate.
W also identify a partial buffer: in regions with greater high-tech manufacturing human capital, the decline in innovation is smaller. Skilled labour, local knowledge networks, and collaboration may help firms partly offset restricted finance. Human capital does not eliminate financing constraints, but it can reduce firms’ dependence on bank credit.
The main policy result is that the innovation cost is not uniform across the macroprudential toolkit. It is concentrated in credit-oriented and credit-cycle-smoothing instruments – tools whose primary effect is to restrict credit directly to firms or borrowers. By contrast, financial-resilience instruments, including liquidity requirements, capital buffers, and provisioning measures, carry no measurable short-run innovation cost (Muñoz et al. 2025).
Over longer horizons of nine to fifteen years, resilience instruments are associated with positive innovation effects – plausibly because a better-capitalised financial system sustains lending through future stress. We flag this result with appropriate caution, given the smaller samples and greater exposure to confounding shocks at long horizons, but the direction is consistent with the underlying rationale for resilience-oriented policy.
The lesson is not to abandon macroprudential policy – instability itself damages investment and innovation. It is rather that instrument choice carries real-economy consequences that deserve explicit recognition. When direct credit restraint is necessary to contain systemic risk, policymakers face a genuine short-run trade-off requiring transparent calibration and communication. Where vulnerabilities permit, resilience-building measures offer a lower-cost alternative (Wedow et al. 2022).
Complementary policies can cushion the impact: funding facilities, public guarantees, and well-designed R&D support can help financially constrained firms preserve research programmes during tightening episodes, while longer-term investment in high-tech skills and alternative financing can reduce firms’ dependence on bank credit. The broader implication is that macroprudential policy functions as a portfolio – its consequences for the wider economy depend not only on whether it tightens, but on which instruments are used, when, and in combination with what else.
Source : VOXeu
The voluntary carbon market was built on a single promise: that one credit cancels one…
Anti-immigration policies often rest on the premise that immigrants worsen native-born employment and wages. This…
Discussions on Japan’s ‘lost decades’ have long focused on sluggish demand and the survival of…
Lithium, cobalt, nickel, copper, and rare earth elements are reshaping global trade and great-power politics…
If frontier AI models can be used to develop their next models, this may lead…
Lithium, cobalt, nickel, copper, and rare earth elements are reshaping global trade and great-power politics…