Central banks can move policy rates, but transmission depends on households acting on them. This column shows that central banks can mitigate monetary policy’s ‘last-mile problem’ with direct communication. In Ireland, a mortgage refinancing reminder increased refinancing with borrowers’ existing lender by up to 76% relative to the control group, from 8.9% to 15.7% of borrowers. The results suggest that inattention is a key friction and that central bank communication holds promise.
The ECB has moved policy rates in both directions in quick succession, cutting through 2024 and 2025 before reversing course with a June 2026 hike as energy prices surged. Each turn revives the question of how much of any change in policy rates reaches household budgets. The concern is warranted. Pass-through from policy rates to the retail interest rates that households and firms pay is slow and uneven (Altavilla et al. 2020), and rates on outstanding mortgages move far less than rates on new loans. In logistics, the ‘last mile’ is the disproportionately costly final step of delivery to the end user. Monetary policy has a last-mile problem of its own.
Figure 1 illustrates the last-mile problem in the US mortgage context, showing that while interest rates on the flow of new mortgages are reasonably responsive to monetary policy rates, interest rates on the stock of outstanding mortgages respond much less and with a significant lag.
Figure 1 US policy rates and average interest rates on outstanding and new mortgages
Notes: Figure plots US Federal Reserve policy rates and the average interest rates on outstanding and newly originated mortgages.
Source: Authors’ calculations using ICE McDash Data; FRED.
On the investment side, Gormsen and Huber (2023) show that firms adjust their hurdle rates only sluggishly when their cost of capital falls, muting the response of corporate investment to rate changes. On the household side, a key reason is a well-documented ‘failure to refinance’. Across the US, Denmark, the UK, Italy, Australia, and Ireland, large numbers of mortgage holders leave substantial savings unclaimed (Campbell 2006, Keys et al. 2016, Andersen et al. 2020). This matters at both ends. From a macroeconomic perspective, sluggish refinancing weakens the refinancing channel of monetary policy transmission (Eichenbaum et al. 2018, Beraja et al. 2019, Cloyne et al. 2020), a channel through which even quantitative easing does much of its work (Kermani et al. 2016, Di Maggio et al. 2020). From a microeconomic perspective, it means many households overpay mortgage interest and forgo consumption. Ireland illustrates the point. Byrne et al. (2020) estimate that three in every five Irish mortgages could save over €1,000 within a year of refinancing, yet just 2.9% of mortgages switched lenders during the second half of 2019.
What the literature has largely lacked is an actionable prescription. Candidate frictions abound, including inattention, present bias, distrust, low financial literacy, and hassle costs, but the two prior field experiments on mortgage refinancing found small or insignificant effects (Keys et al. 2016, Johnson et al. 2019), and neither tested reminders. Meanwhile, an emerging literature has considered the possibility that central banks can improve policy effectiveness through more clever communication strategies (Blinder et al. 2024).
The experiment
In a recent paper (Byrne et al. 2025), we analyse a field experiment conducted by a large retail bank in Ireland, built on the mandatory annual disclosures that Irish lenders must send variable-rate mortgage borrowers summarising cheaper products available from their existing provider. In early 2020, 12,050 variable-rate borrowers were randomly assigned to a control group receiving the standard disclosure or to one of six treatment arms receiving behaviourally enhanced versions. 1 Within each treatment arm, a randomly selected half also received a short follow-up reminder letter four to six weeks later.2
The outcome we focus on is internal refinancing, meaning refinancing with the borrower’s current lender, which in Ireland involves no closing costs and far lower time and hassle costs than switching to an external provider.3 The stakes were meaningful, with an average outstanding rate of 4.2% against an available fixed rate from refinancing of 2.9% and average first-year savings of around €1,000.
The reminder mattered more than the redesign
Figure 2 summarises the core results. The left-hand bars are experimental estimates, while the right-hand bars are model-implied and time-series benchmarks discussed below.
Figure 2 Refinancing treatment effects
Note: Blue bars report experimental treatment effects on internal refinancing. Red bars report model-implied effects of a 100-basis-point decrease in mortgage rates holding attention fixed. The green bar reports the US time-series correlation between rate gaps and prepayment. Error bars denote 95% confidence intervals.
Source: Byrne et al. (2025).
Redesigned disclosures on their own moved behaviour modestly. The average treatment arm without a reminder increased internal refinancing by 1.8 percentage points from a control-group baseline of 8.9%. This replicates the small effects found in the earlier refinancing experiments and is consistent with the broader finding that consumers are often inattentive to disclosure (e.g. Adams et al. 2021).
Adding the follow-up reminder raised refinancing by an additional 3.6 percentage points, for a total average communication effect of 61% over the control group (a 5.4 percentage point increase). In the descriptively best-performing arm, the combination of redesigned disclosure and reminder lifted internal refinancing from 8.9% to 15.7%, an increase of 76%. We cannot statistically reject equality of the combined effects across arms, but the total effect is significant in every arm, and reminder effects show little heterogeneity across borrower characteristics.
The contrast between the modest effects of disclosure design and the larger marginal effect of the follow-up points to inattention, in the form of absent-mindedness or procrastination, as a significant impediment to refinancing. The average 12-month savings realised by refinancing borrowers in our data is €1,209. A conservative back-of-the-envelope calculation, applying a marginal propensity to consume of 0.5 out of interest savings, implies €605 of extra consumption per refinancing household and roughly €42 of borrower consumption per €1 spent on the communication.4
A model-based comparison with rate changes
To interpret these effects, we extend the Andersen et al. (2020) model of inattentive refinancing to allow experimental treatments to shift attention. The estimates imply that the enhanced disclosure and reminder combination increased the share of attentive households from 25% to 41%. Allowing for inattention also improves the model’s fit of the data.5
The model then allows a comparison of policy relevance, captured in the right-hand bars of Figure 2. This is not a direct experiment on policy rates but a model-implied, partial-equilibrium comparison at the same four-month horizon, holding attention fixed at its baseline level. On that basis, a 100-basis-point reduction in mortgage rates would raise four-month refinancing by 1.2, 1.4, and 1.5 percentage points in Ireland, the US, and Denmark, respectively. The time-series relationship between rate gaps and US prepayment, a likely upper bound because it bundles in attention responses, implies a 2.8 percentage point increase. Each is smaller than the 6.8 percentage point effect of the best-performing treatment and reminder combination. Two qualifications matter. First, a large rate cut would likely raise aggregate attention through media, marketing, or peer effects, so these figures capture the partial effect of a rate change rather than its total effect. Second, the gap between the model-implied partial effects and the time-series upper bound itself suggests that shifts in aggregate attention are an important component of monetary policy transmission.
Policy implications
Our results indicate that the refinancing channel of monetary policy can be strengthened, at low cost, by addressing attention directly. The package we test consists of a redesigned disclosure followed by a short reminder directing the borrower to the channels for acting. The lever is available to a menu of institutions. Fiscal authorities seeking to stimulate consumption, competition authorities concerned with mortgage market competitiveness, and consumer protection authorities focused on household debt service burdens could each employ it. Central banks easing policy could use such communication to improve pass-through, which may be especially valuable at the effective lower bound or for authorities in a monetary union whose desired stimulus differs from the union-wide setting. Given the importance of household trust in the disclosing entity, communication sent directly by a government agency or central bank may be more effective than communication from a for-profit lender, although emphasising that a letter is mandated could help. More broadly, the effectiveness of reminders in our setting adds momentum to efforts to improve monetary policy effectiveness through central bank communication (Blinder et al. 2024).
Three caveats apply. First, we study a one-shot reminder, and repeated reminders may lose salience or crowd out proactive search (Ericson 2017). Second, the setting involved a trusted sender operating under a regulatory mandate, and mailed reminders may be less effective where trust is lower or junk mail more prevalent. Third, the intervention operated in a falling-rate environment with widespread in-the-money refinancing opportunities, and reminders are unlikely to be as impactful when rates are rising. Some benefits could also be eroded in general equilibrium if lenders reprice against more responsive refinancers, although the countervailing effects estimated in that literature are generally modest (Fisher et al. 2022, Zhang 2022, Berger et al. 2024).
Finally, the results warrant humility as much as optimism. Even in the best-performing arm, internal refinancing reached 15.7%, meaning more than four in five eligible borrowers remained inactive despite personalised information and a follow-up prompt. The failure to refinance persists, and its remaining drivers deserve ongoing examination. Targeted communication shortens the last mile of monetary policy but does not complete it.
Source : VOXeu
