The sharp rise in interest rates since 2021 has not been matched by equal increases in household savings rates in advanced economies. This column uses transaction-level data from Iceland to study how changes in interest rates affect household deposit holdings. It shows that for the average household, low-yield deposits barely respond to interest rate differentials, even when there are equally safe alternatives available. This inaction means households forgo significant income by not moving balances into higher-yielding accounts. The findings imply that monetary policy transmission through deposits is weaker and more uneven than standard models imply.
Across advanced economies, the sharp rise in policy rates since 2021 has not been matched by an equally sharp increase in the interest households receive on their savings. Banks in the US and Europe have kept deposit rates persistently low, even during the steepest monetary tightening in decades (Drechsler et al. 2017). This raises a central question: when policy rates rise, do households actively shift their money out of deposits into higher-yielding alternatives, or do they leave their balances untouched? And if many remain inactive, who are the households that actually move funds — and how much income is forgone by those who do not?
In a new study (Cirelli and Olafsson 2025), we use transaction-level data covering roughly one-third of Iceland’s adult population to provide answers. Iceland provides a perfect environment to study these questions: all households have free, instant access to a range of deposit products that are identical in risk, maturity, and liquidity but differ substantially in the interest rates they pay. Transfers between accounts can be executed instantly through a mobile app or an internet bank. Despite this frictionless setting, we find large differences in portfolio holdings, widespread inaction, and we show that only wealthy and financially literate households adjust meaningfully when interest rates rise.
Wealth matters, interest rates not so much
What drives households to reallocate their portfolios? The cost of mismanaging savings depends both on how much households have and on the opportunity cost of leaving money in low-yield accounts — that is, the difference between what they actually earn and what they could earn elsewhere (the spread). In Cirelli and Olafsson (2025), we show that even within the universe of liquid, short-term, safe assets, households adjust their portfolios when their wealth changes, not when spreads change.
- As households become wealthier, they keep far less in low-yield accounts. The top decile allocates more than 40 percentage points less to low-yield deposits than the bottom decile.
- By contrast, the average household barely responds to spreads: a one-percentage-point rise in the rate differential reduces the share in low-yield deposits by only 0.1 percentage points.
This inaction is especially striking given that the assets under study are equally liquid, safe, and insured, with transfers that are instant and costless. In other words, households leave money on the table even when switching requires no more than a tap on a mobile app.
Figure 1 Portfolio shares across the asset distribution and over time
Notes: The left panel shows average portfolio shares across the three liquid safe assets available to Icelandic households — checking, savings, and liquid funds (a money-market-type account) — across the wealth distribution. Interest rates on savings and liquid funds consistently dominate those on checking deposits over the sample period. The right panel shows the average portfolio composition across households, plotted against the interest rate on liquid funds (black line), which serves as a proxy for the spread between checking and higher-yielding alternatives.
How costly is the inaction?
Inaction is not free, especially for wealthy households, who — even though they invest more efficiently — still forgo substantial income simply by not moving balances into higher-yielding accounts:
- On average, wealthy households earn about two percentage points more on their liquid safe portfolios than poorer households.
- Yet they leave large sums on the table: for the top decile, forgone interest income amounts to roughly 2.5% of annual consumption.
These magnitudes are large. They are comparable to the return per dollar households forgo by not participating in stock markets or to the welfare losses from under-diversification documented in Sweden (Calvet et al. 2007). The key difference is that in this setting, the losses arise without any trade-off in risk, maturity, or liquidity, and within nearly identical products offered by the same bank on the same platform.
Figure 2 Average return and forgone interest income across the asset distribution
Notes: The left panel shows the average return on liquid, short-term safe assets by wealth percentile. The right panel displays forgone interest income as a share of annual consumption expenditure.
Who moves aggregate deposits?
It is well known that aggregate deposits decline when central banks tighten policy rates (Drechsler et al. 2017). But who actually moves these deposits? The answer lies in heterogeneity.
- Wealthy households are far more elastic. They reduce their low-rate deposit share by about one percentage point for every one-point increase in spreads — almost ten times the population average.
- By contrast, the bottom 60% of households show little or no response.
- Beyond wealth, financial literacy and inflation knowledge matter: households scoring higher on survey-based measures of financial understanding reallocate much more actively.
Because deposits are highly concentrated — the top decile alone holds over 60% of balances — these wealthy and financially informed households drive the aggregate flows that matter for banks and policymakers. In other words, understanding aggregate deposit fluctuations ultimately means understanding what makes wealthy depositors tick.
Figure 3 Predicted 12-month change in the share of checking within core deposits
Notes: Predicted 12-month change in the share of checking within core deposits — aggregate (left), top wealth decile (centre), and fifth decile (right). We decompose each change into (i) the component due to interest-rate terms and (ii) the component due to other covariates (wealth levels, group-specific trends, demographics). Interest-rate effects are computed by varying only the interest-rate terms while holding other covariates fixed. For the top decile, which drives the aggregate, changes are driven almost entirely by rates; for the fifth decile, most variation reflects other covariates rather than rates.
Theories under stress
How do these facts square with standard macroeconomic theories widely used to study monetary policy transmission? To find out, we build a two-asset, incomplete-markets model with portfolio adjustment frictions, in the tradition of Auclert et al. (2020, 2025).
These models incorporate frictions to trading between assets with high or low returns. When calibrated to match the frequency of monthly adjustments between accounts, they do a decent job of replicating the cross-sectional distribution of holdings. But they fail where it matters most: they vastly overstate households’ sensitivity to interest rates. In the models, rate hikes trigger large reallocations; in the data, households hardly budge.
This mismatch suggests that something beyond standard adjustment frictions is at work — perhaps wealth-dependent inattention. For policy, the message is clear: if models exaggerate deposit elasticity, they will also exaggerate how forcefully monetary policy transmits through household portfolios to bank funding and lending.
Conclusion
Our evidence from bank transaction data highlights that deposit stickiness is not a technological phenomenon. Even in one of the most advanced digital banking systems in the world, with frictionless, instantaneous transfers, households fail to respond to interest rate incentives. The key drivers of aggregate deposit dynamics are wealthy, financially literate households — who both invest better on average and account for most balances yet still leave significant sums unclaimed.
For policymakers, the lesson is sobering: monetary policy transmission through deposits is weaker and more uneven than standard models imply. For researchers, the challenge is to incorporate wealth- and knowledge-dependent frictions, or behavioural inertia, into models of household portfolio choice. Deposit stickiness is not just a curiosity of household finance — it fundamentally shapes the effectiveness of monetary policy and the resilience of the banking system.
Source: cepr.org
