DOI: doi.org/10.71734/DP‑2026‑18

How do German households adjust their savings in relation to rising inflation expectations? Using survey data from 4,824 households collected during historically high inflation in March 2023, this empirical study examines the relationship between inflation expectations and saving decisions across seven categories. We document that higher inflation expectations are associated with a greater likelihood of maintaining or expanding overall savings – consistent with precautionary motives – but that the relationship is non-linear: only households with strongly elevated expectations exhibit economically meaningful behavioral responses. Decomposing portfolio shifts by inflation expectation level further reveals substantial heterogeneity in how households reallocate funds across saving categories, with implications for bank funding stability that depend on which segment of households is responding.

The relationship between inflation expectations and household saving behavior has generated conflicting findings in the literature, with some studies documenting negative relationships suggesting households increase current consumption to avoid future price increases (e.g., Dräger and Nghiem 2021, Vellekoop and Wiederholt 2017) while others find positive relationships, indicating that precautionary saving motives dominate (Rondinelli and Zizza 2020, Coibion et al. 2023). This disagreement partly stems from focusing on aggregate saving measures that obscure underlying portfolio reallocation patterns. Our study based on German household data addresses this gap through comprehensive disaggregated analysis of household saving decisions across multiple categories during 2023, a period of unprecedentedly high inflation. Germany provides an ideal setting for this analysis given the crucial role of retail deposits in bank funding.

A non-linear threshold and flight-to-liquidity at the upper tail

Using generalized ordered logit models, we analyze saving intentions across seven categories: cash, demand deposits, short-term deposits, long-term deposits, bonds, shares, and digital formats. We find that the continuous inflation expectations measure is associated with a greater probability of maintaining or expanding overall savings, but the marginal effects are economically modest (approximately 0.1 percentage points per one-percentage-point increase). Replacing the continuous measure with quartile dummies reveals that this aggregate result masks a sharp threshold pattern. Only households in the fourth quartile of the expectations distribution – with a mean expectation of 12.3 % – exhibit economically meaningful responses, including a 6.2 percentage point increase in the probability of saving more in cash. This pattern is robust to survey weighting (where the cash effect strengthens to 10.3 percentage points) and to a qualitative reformulation of the expectations measure.

This pattern aligns with liquidity preference theory (Keynes 1936, Tobin 1958), whereby households respond to inflation-related uncertainty by building liquid buffers rather than simply accelerating consumption. Our findings contrast with recent working papers suggesting households shift from safe to risky assets when inflation expectations rise (Yang 2022) or toward bank deposits when expectations decline (Agarwal et al. 2022), highlighting the importance of disaggregated analysis and institutional context in understanding household portfolio responses.

Furthermore, we find that static inflation expectations do not correlate with saving behavior, suggesting that dynamic inflation perceptions play a more critical role. The mechanism matters because static expectations – where households maintain unchanged beliefs – may not capture persistent inflation dynamics, while dynamic expectations reflect active updating in response to evolving conditions. This methodological insight aligns with research on expectation formation mechanisms (Mankiw et al. 2003, Carroll et al. 2020).

Heterogeneity in portfolio reallocation by expectation level

A core contribution of our analysis lies in decomposing portfolio shifts by inflation expectation level. In the full sample, households reducing bonds show a 15.0 percentage point higher probability of increasing cash, and those reducing cash show a 12.9 percentage point higher probability of increasing bonds – a bidirectional cash-bond substitution. We additionally document that households reducing demand deposits show significantly higher probabilities of increasing long-term deposits (18.6 percentage points) and bonds (12.8 percentage points), suggesting maturity extension and a partial shift away from bank liabilities.

Splitting the sample at the median expectation (7 %) reveals that these aggregate patterns are not uniform across households. The cash-bond substitution is driven exclusively by low-expectation households. Among high-expectation households, the relationship reverses: saving more in bonds is associated with simultaneously saving more in cash, and reducing bonds is associated with a 25.1 percentage point higher probability of increasing cash. High-expectation households thus appear to treat cash and bonds as complements within a broader portfolio expansion rather than as substitutes within a portfolio reallocation. The maturity extension from demand deposits toward long-term deposits is similarly concentrated among low-expectation households (a 23.0 percentage point higher probability), suggesting yield-seeking behavior based on moderate inflation concerns rather than inflation-driven precautionary motives. To our knowledge, this distinction between reallocation and portfolio expansion as responses to inflation concerns is novel in the household finance literature.

Implications for bank funding stability

The implications for bank funding stability depend on which household segment is responding and are therefore ambiguous. The maturity extension observed among low-expectation households may reduce rollover risk and extend liability duration at the aggregate level, but it also imposes higher funding costs as banks pay term-deposit premiums, compressing net interest margins. At the bank level, depositors migrating to competitors offering higher rates can transform sector-level stability into bank-level liquidity pressure. Substitution from deposits toward bonds – observed primarily among low-expectation households – directly reduces banks' deposit base.

These implications should be interpreted with care. Our cross-sectional design identifies associations rather than causal effects, and the dependent variable captures saving intentions rather than realized flows. The setting itself – German households at a single high-inflation moment in March 2023 – limits external validity: the threshold effects we identify at the upper tail of the expectations distribution may not arise in low or moderate-inflation regimes, where the distribution of expectations is more compressed and precautionary motives less salient. The Bundesbank Online Panel Households (BOP‑HH) survey also lacks asset-specific return expectations for bonds and shares, so we cannot fully separate inflation-driven portfolio shifts from return-seeking responses to monetary policy tightening, although our control for interest rate expectations partially addresses this. 

References

Agarwal, S., Y. H. Chua, P. Ghosh, and C. Song (2022). Inflation expectations of households and portfolio rebalancing: Evidence from inflation targeting in India. SSRN Electronic Journal, Available at SSRN 4069564.

Carroll, C. D., E. Crawley, J. Slacalek, K. Tokuoka, and M. N. White (2020). Sticky expectations and consumption dynamics. American Economic Journal: Macroeconomics, 12(3), 40‑76.

Coibion, O., D. Georgarakos, Y. Gorodnichenko, and M. van Rooij (2023). How does consumption respond to news about inflation? Field evidence from a randomized control trial. American Economic Journal: Macroeconomics, 15(3), 109‑52.

Dräger, L. and G. Nghiem (2021). Are consumers' spending decisions in line with an Euler equation? The Review of Economics and Statistics, 103(3), 580‑596.

Keynes, J. M. (1936). The general theory of employment, interest and money. London: Macmillan.

Mankiw, N. G., R. Reis, and J. Wolfers (2003). Disagreement about inflation expectations. National Bureau of Economic Research Macroeconomics Annual, 18, 209‑248.

Rondinelli, C. and R. Zizza (2020). Spend today or spend tomorrow? The role of inflation expectations in consumer behaviour. Banca d'Italia Working Paper, No. 1276.

Tobin, J. (1958). Liquidity preference as behavior towards risk. The Review of Economic Studies, 25(2), 65‑86.

Vellekoop, N. and M. Wiederholt (2019). Inflation expectations and choices of households. SAFE Working Paper, No. 250.

Yang, Q. (2022). Inflation expectations and households portfolio choice. SSRN Electronic Journal, Available at SSRN 4137431.

Krüger, J., V. Dinger (2026), Expected inflation and saving in categories, Bundesbank Discussion Paper, No 18/2026.