Decoded, Economics & finance

Rentenangst: Quantifying How Pension Reform Uncertainty Freezes German Consumption

Rentenangst: Quantifying How Pension Reform Uncertainty Freezes German Consumption
The German Economy in 2026 | Analytical Series, Paper 7

Executive Summary

German households entered the autumn of 2026 in a strange frame of mind. Their view of the economy improved for six months running, yet they grew gloomier about their own income and more willing to save. The consumer climate fell to minus 30.8 points for November, eight points below its level a year earlier. Real consumption growth fell from 1.6 to 0.1 percent within two quarters. A popular explanation has formed around one word, Rentenangst, the fear for one’s pension. This paper tests it.

Nominal disposable income kept growing at about 3 percent a year; what changed was the saving rate. It fell from 11.1 to 10.0 percent in 2025, adding more than a point to consumption growth, then stopped falling at 10.1 percent. That freeze, plus the energy-driven inflation spike, explains almost all of the consumption slowdown. The question is why saving froze.

A saving-rate equation estimated on quarterly data from 2012 to 2024, with inflation, the ECB deposit rate, real income growth, unemployment and policy uncertainty as regressors, predicts the 2026 saving rate of 10.1 percent almost exactly, out of sample.

Almost: a pension-debate term reaches 0.10 percentage points from the second quarter of 2026 (p-value 0.08): about EUR 2.9 billion a year of withheld consumption, or 0.04 percent of GDP. In the two months dominated by pension news, the survey propensity to save rose about 3 points more than income expectations would predict.

A calibrated pension-wealth model puts the rational response to what the commission actually proposed at 0.24 points of disposable income, EUR 6.7 billion a year, 90 percent of it from the abolition of the deduction-free pension after 45 contribution years. The measured effect is only 43 percent of this benchmark. Households are not overreacting; most have not yet absorbed what the reform means for them, and that gap is a risk rather than a comfort.

The macroeconomic cost is therefore mostly ahead of us. Under a clear, quickly legislated reform that grandfathers cohorts close to retirement, pension-related saving fades to a planning residual and costs 0.04 percent of GDP in 2027. If legislation drifts, as the minister’s October resignation suggests it may, the cost triples to 0.12 percent. If the reform becomes disorderly and households price in broader cuts, at 0.29 saving-rate points per perceived percentage point of cuts, the cost reaches 0.24 percent of GDP, EUR 14.7 billion of consumption. Speed, grandfathering and clarity, which cost the budget nothing, will decide the bill.

Key findings.

  1. The 2026 consumption freeze is a saving freeze. Real private consumption growth fell from 1.6 percent year on year in the fourth quarter of 2025 to 0.1 percent in the second quarter of 2026. The swing in the contribution of the saving rate, from plus 1.1 to minus 0.1 percentage points, accounts for about 1.2 points of that 1.5 point slowdown.
  2. Fundamentals explain most of the saving rate, but not all of it. A saving-rate equation estimated on 51 quarters to 2024 predicts 10.0 to 10.1 percent for the first half of 2026, against an observed 10.1 percent. The fall of 1.0 point since late 2024 is explained by ECB rate cuts, weak real income and rising unemployment.
  3. Pension anxiety is measurable but still small in the national accounts. A pension-debate term adds 0.10 points to the saving rate from the second quarter of 2026 (significant at the 10 percent level), equal to about EUR 2.9 billion of withheld consumption a year and 0.04 percent of GDP.
  4. Households have priced in less than half of what the reform rationally implies. A calibrated pension-wealth model puts the rational offset saving at 0.24 points of income (EUR 6.7 billion a year), 90 percent of it caused by the abolition of the deduction-free pension after 45 contribution years. Measured saving is about 43 percent of that benchmark, consistent with only 27 percent of Germans saying they know the reform package.
  5. Pension news moves saving intentions. In the two months dominated by pension news, the consumer survey's propensity to save rose about 3 points more than income expectations alone would predict; in September 2026 it reached 21.5 points, the highest level since the 2008 financial crisis.
  6. The cost lies in 2027 and depends on legislation. Pension-related saving would cost 0.04 percent of GDP in 2027 under a clear, grandfathered reform, 0.12 percent under drift and 0.24 percent (EUR 14.7 billion of consumption) under disorder. Each 1 percent of perceived across-the-board pension cuts adds 0.29 points to the saving rate.

1. Introduction: The Puzzle of the Optimistic Saver

Germany’s recovery in 2026 has been carried by the state. GDP grew by 0.3 percent in the second quarter and 0.2 percent in the third, and the government expects growth of 1.3 percent for the year, after three years of contraction or stagnation. The chamber of commerce has said openly that debt-financed public spending is doing most of the work. Private investment fell again in the third quarter. Exports, strong in the first half, weakened noticeably. In this configuration, private consumption, about half of GDP, is the component that would have to turn a public-spending recovery into a self-sustaining one. It has not done so.

The weakness of consumption is not, at first sight, a problem of income. Employment has fallen and unemployment is close to 3 million, but nominal disposable income grew by 3.0 percent in the year to the second quarter of 2026. Inflation, pushed up by the energy shock of the Iran war, rose to 3.3 percent in September, its highest since December 2023, before easing to 2.8 percent in October. That squeezes real income, but not by enough to explain a stall in real consumption from 1.6 to 0.1 percent within two quarters. The missing piece is saving. Households kept a stable share of their income aside at a moment when, by the logic of improving expectations, they should have been spending more of it.

At the same time Germany has been conducting its most consequential pension debate in two decades. In late 2025 the coalition’s pension package, which fixed the replacement rate at 48 percent until 2031, set off a controversial debate about the future of the pension system, one that the NIM consumer survey already cited in December 2025 as a reason for rising saving intentions. A pension commission was asked to propose reforms for the period after 2031. On 23 June 2026 it delivered 33 recommendations: the end of the deduction-free pension after 45 contribution years, an early retirement age raised from 63 to 64, a statutory retirement age linked to life expectancy from 2032, the return of the sustainability factor and a mandatory funded pillar financed by an extra 2 percent of gross wages. The chancellor’s government has said it wants to implement the package, but the labour minister resigned on 12 October 2026, no draft bill existed in November, and a parliamentary vote is not expected before the first quarter of 2027.

The coincidence has produced a ready explanation: households save because they fear for their pensions. It is plausible, but it has not been tested against the national accounts and its magnitude has never been estimated. Is pension anxiety a central driver of the consumption freeze, a marginal one, or a looming one?

Our central argument is threefold. First, the consumption freeze of 2026 is above all a saving freeze, and most of it is explained by fundamentals that have nothing to do with pensions: the end of ECB easing, the energy shock to real income and a softer labour market. Second, a pension-related component is nevertheless visible in the data, at about 0.1 percentage points of disposable income so far, and it is smaller than what the content of the reform would rationally justify, because most households have not yet worked out what the reform means for them. Third, the main macroeconomic risk is therefore ahead, in 2027, and it depends almost entirely on how the reform is legislated and communicated.

Section 2 sets out the household picture and the pension timeline; Section 3 decomposes the consumption slowdown; Section 4 estimates a saving-rate equation; Section 5 isolates the pension component; Section 6 builds a calibrated pension-wealth model; Sections 7 and 8 quantify costs and 2027 scenarios; Sections 9 and 10 draw implications and conclude.

2. The Household Picture in Autumn 2026

2.1 A saving rate that stopped moving

The household saving rate, measured as saving in percent of disposable income, is the central variable of this paper. According to the seasonally adjusted national accounts series published by the Bundesbank, it stood at 10.1 percent in each of the last three quarters for which data exist, the fourth quarter of 2025 and the first two quarters of 2026. In unadjusted terms it was 9.6 percent in the second quarter of 2026 and 12.6 percent in the first, the latter unchanged from a year earlier. Quarterly disposable income reached EUR 697.4 billion in the second quarter, of which households saved EUR 70.3 billion.

Table 1. Household disposable income, saving and saving rate, seasonally adjusted.

Quarter Disposable income (EUR bn, sa) Saving (EUR bn, sa) Saving rate (%, sa)
2024 Q2 659.8 73.2 11.1
2024 Q4 671.6 74.4 11.1
2025 Q1 672.2 67.1 10.0
2025 Q2 677.1 67.7 10.0
2025 Q3 684.3 70.7 10.3
2025 Q4 691.1 69.9 10.1
2026 Q1 692.5 69.6 10.1
2026 Q2 697.4 70.3 10.1

Source: Deutsche Bundesbank, seasonally adjusted economic statistics (Destatis national accounts).

Two features of the series matter. The first is the step down at the beginning of 2025, from 11.1 to 10.0 percent, which released about EUR 7 billion a quarter into consumption. The second is the subsequent flatness. A saving rate of 10.1 percent is not high by German standards: it is close to the 2015 to 2019 average and well below the pandemic peak of 19.9 percent in the second quarter of 2020. The often-quoted claim that Germans “save more than ever” refers to a different concept, the gross saving rate, which at 18.9 percent in 2025 and in the first half of 2026 is indeed high by international standards (against 14.3 percent in the European Union and 10.7 percent in the United States), but which has been falling, not rising, since 2024.

That distinction frames the analytical question correctly. The puzzle of 2026 is not that German households are saving an unusually large share of their income. It is that the saving rate stopped falling when, given improving expectations, it might have been expected to continue down.

2.2 Consumption at a standstill

Real private consumption grew by 1.3 percent in 2025 as a whole. Its year-on-year growth rate peaked at 1.6 percent in the fourth quarter, then dropped to 0.5 percent in the first quarter of 2026 and 0.1 percent in the second. Destatis reported that private consumption in the first quarter did not exceed its level of the preceding quarter, and its flash estimate for the third quarter showed only a slight increase in total consumption. The German Economic Institute (IW) expects real private consumption to grow by just 0.3 percent in 2026.

Table 2. The household sector in autumn 2026.

Indicator Latest value Period Comparison
Real private consumption, y/y 0.1% 2026 Q2 1.6% in 2025 Q4
Consumer price inflation 2.8% October 2026 3.3% in September
Energy prices, y/y 12.2% October 2026 14.9% in September
Core inflation (ex food and energy) 2.1% October 2026 Goods 3.0%, services 2.5%
Consumer climate minus 30.8 November 2026 minus 22.5 in November 2025
Propensity to save (survey) 17.2 October 2026 21.5 in September
Household financial assets EUR 9,831 bn 2026 Q2 plus EUR 335 bn on the quarter
ECB deposit rate 2.50% Since 16 September 2026 2.00% until June 2026
Statutory pension contribution rate 18.6% 2026 19.9% in 2007

Sources: Bundesbank, Destatis, NIM, ECB, BMAS.

The financial accounts tell a consistent story. Household financial assets rose by EUR 335 billion in the second quarter of 2026 to EUR 9,831 billion, but EUR 258 billion of the increase was valuation gains as markets recovered from the first-quarter turbulence. Net acquisition of financial assets and other changes amounted to about EUR 77 billion. Households are wealthier on paper and continue to accumulate, but they are not accelerating.

2.3 A wedge in the consumer survey

The monthly consumer climate survey conducted by the Nuremberg Institute for Market Decisions (NIM) with GfK provides the most timely view of household sentiment. Its components moved in opposite directions through 2026. Economic expectations rose for six consecutive months, from minus 13.7 points in April to plus 2.1 in October. Income expectations, by contrast, collapsed to minus 24.4 points in April after the war began, recovered to about minus 12 in May and June, spiked briefly to plus 1.7 in August and fell back to minus 15.0 in September and minus 15.7 in October. The propensity to save rose from 13.9 in June to 17.0 in July, eased in August and jumped by 6.0 points to 21.5 in September, a level the NIM describes as comparable to the 2008 financial crisis.

-30-20-100102030Apr 2026May 2026Jun 2026Jul 2026Aug 2026Sep 2026Oct 2026Index pointsCommission report, 23 JunePropensity to saveEconomic expectationsIncome expectations
Figure 1. Consumer survey components, April to October 2026. Source: NIM Konsumklima press releases; survey months.

The wedge between the economy and the wallet is the survey’s version of the saving freeze. In the NIM’s own words, hope for an economic upswing is there, but willingness to spend has not followed, and when consumers look at their own finances, scepticism still prevails. The NIM attributes the September collapse in income expectations mainly to high energy prices eroding purchasing power. But it also cited, in its December 2025 release, the controversial debate about the future of the pension system as one reason for a 5-point jump in the propensity to save to 18.7. The survey therefore contains both stories: an energy shock to current income and a policy shock to future income.

2.4 Timeline of the pension debate

The pension debate has unfolded in four phases, summarised in Table 3. The timing will be used in Section 5 to identify the pension component of saving.

Table 3. Timeline of the pension debate.

Date Event Relevance
Late 2025 Controversy over the 2025 pension package (48 percent level until 2031); NIM links higher saving to it First wave of attention
First half of 2026 Commission at work; press leaks on higher retirement ages and the end of the pension at 63 Uncertainty, no content
23 June 2026 Commission delivers 33 recommendations Concrete content
July to August 2026 Government signals implementation; IMK and NIM surveys show rejection and worry Content spreads
12 October 2026 Labour minister Baerbel Bas resigns Timetable in doubt
November 2026 No draft bill; cabinet target end-2026; Bundestag vote planned for Q1 2027 Prolonged uncertainty

Sources: ZDF, cash-online, t-online, vermoegenszentrum.de, NIM.

The commission’s recommendations matter for households because they change the expected value and the timing of pension income. Table 4 lists those with the largest direct effects on household wealth.

Table 4. Commission recommendations with direct effects on household pension wealth.

Recommendation Content Effect on household pension wealth
End of deduction-free pension after 45 years Used by 262,400 people in 2024, average pension EUR 1,677 a month Large, concentrated near retirement
Early retirement with deductions from 64 Earliest age raised from 63; 238,000 entries in 2024 Small in money
Statutory age linked to life expectancy From 2032, ratio 2 to 1; 67.5 by about 2041, 70 in the 2090s Moderate, cohorts from mid-1960s
Sustainability factor Back from 2032, weight 0.33; transition factor keeps 48 percent at entry Lower indexation in retirement
Funded statutory pillar 2 percent of gross wages from 2028, 0.5 points a year; total rate 18.6 to 20.6 percent Lower net income now, higher pension later
Altersteilzeit Minimum age raised from 55 to 58 Small

Sources: BMAS fact sheet of 23 June 2026, gegen-hartz.de, t-online, vermoegenszentrum.de.

Two points stand out. The reform is not, on average, a cut in the replacement rate at entry: the transition factor is designed to keep the level for new retirees at least at today’s 48 percent, against 46 percent under current law from 2031. But it is a large and concentrated loss for one group, the long-insured who planned to retire early without deductions, and it introduces uncertainty for everyone else. For a cohort born in 1964 or later, retiring two years before the statutory age with deductions of 0.3 percent a month costs 7.2 percent of the pension for life. That is the main channel through which the reform reaches household balance sheets.

3. Why Consumption Stalled: An Accounting Decomposition

Before modelling behaviour, it is useful to establish how much of the consumption slowdown is arithmetically attributable to saving. Nominal consumption equals disposable income times one minus the saving rate. Its growth can therefore be split into nominal income growth and a saving-rate contribution, (1 minus s(t)) / (1 minus s(t minus 4)) minus 1. Subtracting consumer price inflation gives an approximation of real consumption growth.

Table 5. Decomposition of consumption growth into income, saving and prices.

Quarter Nominal disposable income, y/y (%) Saving-rate contribution (pp) Nominal consumption, y/y (%) CPI inflation (%) Implied real consumption (%) Real consumption, national accounts (%)
2024 Q4 5.00 minus 0.56 4.42 2.30 2.12 2.4
2025 Q1 2.75 0.90 3.67 2.26 1.42 1.2
2025 Q2 2.61 1.24 3.88 2.10 1.78 1.5
2025 Q3 2.75 0.90 3.68 2.10 1.58 1.0
2025 Q4 2.90 1.12 4.05 2.30 1.75 1.6
2026 Q1 3.01 minus 0.11 2.90 2.23 0.67 0.5
2026 Q2 3.00 minus 0.11 2.89 2.60 0.29 0.1

Source: Bundesbank, Destatis, OECD via FRED; author's calculations. The implied real figure uses the CPI, not the consumption deflator, which explains small gaps to the national accounts.

The decomposition tracks the national accounts closely and leaves little doubt about the mechanism. Through 2025, the falling saving rate added between 0.9 and 1.2 percentage points a year to consumption growth. In 2026 that contribution turned slightly negative. Between the fourth quarter of 2025 and the second quarter of 2026 the saving contribution swung by 1.23 points, nominal income growth edged up by 0.1 point, and inflation rose by 0.3 points. Of the roughly 1.5 point slowdown in real consumption growth, about 1.2 points is the saving rate.

00.511.522.52025Q12025Q22025Q32025Q42026Q12026Q2% y/y or ppReal disposable income growthContribution of the saving rateReal private consumption
Figure 2. Real income growth, the saving-rate contribution and real consumption, 2025 to 2026. Source: Bundesbank, Destatis; author's calculations.

The figure makes a point that is often missed in the public debate. Real disposable income growth has not collapsed: it has hovered between 0.4 and 0.8 percent in every quarter since early 2025. What collapsed is the extra spending that came from households drawing down their saving rate. In 2025 German consumers spent more than their income growth alone would have allowed; in 2026 they stopped doing so. Any explanation of the consumption freeze must therefore be an explanation of why the decline of the saving rate came to an end.

4. A Saving-Rate Equation for Germany

4.1 Specification

We model the seasonally adjusted saving rate as a function of the variables that the empirical literature on household saving consistently identifies as relevant:

s(t) = a + b1 x inflation(t) + b2 x deposit rate(t) + b3 x unemployment(t) + b4 x real income growth(t) + b5 x log policy uncertainty(t) + pandemic dummies + e(t)

Inflation is the year-on-year change in consumer prices. The interest rate is the quarterly average of the ECB deposit facility rate, which captures the return on the bank deposits that dominate German household portfolios. Unemployment is the registered rate published by the Federal Employment Agency, interpolated from annual averages. Real income growth is the year-on-year growth of nominal disposable income minus inflation. Policy uncertainty is the logarithm of the news-based Economic Policy Uncertainty Index for Germany, averaged over each quarter. Eight quarterly dummies absorb the pandemic period from 2020 to 2021, when lockdowns made saving involuntary.

The equation is estimated by ordinary least squares from the second quarter of 2012 to the fourth quarter of 2024, 51 observations, with heteroskedasticity and autocorrelation consistent standard errors. The estimation deliberately stops before 2025, so that the model’s predictions for 2025 and 2026 are genuinely out of sample and cannot be contaminated by any pension effect.

4.2 Estimates

Table 6. Saving-rate equation, 2012 Q2 to 2024 Q4.

Variable Coefficient HAC standard error t-statistic
Constant 13.93 1.27 10.9
Inflation (pp) minus 0.076 0.031 minus 2.4
ECB deposit rate (pp) 0.213 0.040 5.3
Unemployment rate (pp) minus 0.463 0.094 minus 4.9
Real disposable income growth (pp) 0.112 0.028 4.0
Log policy uncertainty index minus 0.141 0.155 minus 0.9
Pandemic dummies (8 quarters) included
R-squared 0.986
Residual standard deviation, non-pandemic quarters 0.27 pp

Source: author's estimates on Bundesbank, Destatis, ECB, Federal Employment Agency and FRED data.

The coefficients are economically sensible and precisely estimated. A one-point increase in the ECB deposit rate raises the saving rate by about 0.21 points, the clearest single regularity in the data and the reason why the monetary tightening of 2022 to 2023 pushed saving up and the easing of 2024 to 2025 pulled it down. Faster real income growth raises saving in the short run, consistent with consumption smoothing: households save part of income surprises. Inflation lowers the measured saving rate, as households protect consumption volumes when prices rise, which is what happened in 2022. The negative unemployment coefficient reflects the strong downward trend of German unemployment in the 2010s, which coincided with a gradual rise in saving as employment incomes grew; it should be read as a reduced-form control rather than as a precautionary effect, and Section 5.2 shows that the results do not depend on it.

The most interesting coefficient is the one that is not significant. General economic policy uncertainty, as measured by newspaper coverage, has no detectable effect on the German saving rate over this period. The index spiked in 2022 with the energy crisis, in late 2024 with the collapse of the coalition and in April 2025 with US tariffs, reaching 1,502, and none of these episodes produced extra saving beyond what rates, income and inflation explain. This matters for our argument. If pension anxiety has an effect, it must work through a channel specific to lifetime wealth, not through generic uncertainty.

4.3 Out-of-sample predictions for 2025 and 2026

Table 7 compares the model’s predictions with the observed saving rate for the six quarters after the estimation window.

Table 7. Out-of-sample predictions of the saving rate. Author's estimates.

Quarter Observed (%) Predicted (%) 90% prediction interval Gap (pp)
2025 Q1 10.0 10.34 9.77 to 10.91 minus 0.34
2025 Q2 10.0 10.17 9.57 to 10.78 minus 0.17
2025 Q3 10.3 10.12 9.53 to 10.72 plus 0.18
2025 Q4 10.1 10.12 9.56 to 10.68 minus 0.02
2026 Q1 10.1 10.11 9.53 to 10.69 minus 0.01
2026 Q2 10.1 10.04 9.47 to 10.61 plus 0.06
9101112132014201620182020202220242026% of disposable incomeestimation ends 2024Q4pandemic quarters off scaleSaving rate, seasonally adjustedModel prediction (shaded: 90% band, out of sample)
Figure 3. Household saving rate: actual versus the model, 2014 to 2026. Source: Bundesbank; author's estimates. Pandemic quarters are absorbed by dummies and lie off the scale.

The model performs well. All six observations lie comfortably within the 90 percent interval. In the first half of 2025 households saved somewhat less than fundamentals implied, with an average gap of minus 0.26 points. From the second half of 2025 the gap turned slightly positive, averaging plus 0.08 points, and in the first half of 2026 it averaged plus 0.02 points. In short, the national accounts show no large excess saving in 2026. What they show is the disappearance of the under-saving of early 2025, and a slight upward drift since.

4.4 What moved the saving rate

The model also allows a decomposition of the 1.0 point fall in the saving rate between the fourth quarter of 2024 and the second quarter of 2026.

Table 8. Contributions to the change in the saving rate, 2024 Q4 to 2026 Q2. Author's calculations.

Driver Change in driver Contribution to saving rate (pp)
ECB deposit rate 3.27 to 2.04 percent minus 0.26
Real income growth 2.58 to 0.36 percent minus 0.25
Unemployment rate 6.61 to 6.90 percent minus 0.14
Inflation 2.30 to 2.60 percent minus 0.02
Policy uncertainty broadly unchanged 0.00
Unexplained (change in model gap) plus 0.45 to plus 0.06 minus 0.33
Total observed change 11.1 to 10.1 percent minus 1.00
minus 0.4minus 0.3minus 0.2minus 0.100.1Deposit rateminus 0.26Real income growthminus 0.25Unemploymentminus 0.14Inflationminus 0.02Policy uncertaintyplus 0.00Unexplainedminus 0.33Contribution to the change in the saving rate, 2024Q4 to 2026Q2, pp
Figure 4. Contributions to the fall of the saving rate from 2024 Q4 to 2026 Q2. Source: author's calculations.

Two thirds of the decline is accounted for by observable drivers, chiefly monetary easing and the slowdown in real income growth. The remaining third is the unwinding of an unusually high saving rate at the end of 2024, when households saved about 0.45 points more than fundamentals implied, probably in reaction to the collapse of the coalition and the snap election. That positive gap closed in early 2025 and briefly turned negative, giving consumption an extra boost. By 2026 the gap was close to zero.

The decomposition also anticipates the rest of 2026: the ECB’s hikes to 2.25 percent in June and 2.50 percent in September add about 0.1 point to the fundamental saving rate by the fourth quarter, independently of any pension effect (Section 7.3).

4.5 How much can the national accounts detect?

A model that predicts well is still a blunt instrument for detecting small effects. To measure its resolution, we re-estimated it five times, each time ending one year earlier, and computed the average error of the following year’s predictions.

Table 9. Rolling one-year-ahead forecast errors of the saving-rate equation. Author's calculations.

Forecast year Estimation ends Mean error (pp) Largest quarterly error (pp)
2017 2016 Q4 minus 0.29 0.55
2018 2017 Q4 plus 0.20 0.33
2019 2018 Q4 minus 0.39 0.76
2023 2022 Q4 minus 1.19 1.57
2024 2023 Q4 plus 0.58 0.87

The root mean squared annual error is 0.64 points, inflated by 2023, when the model had not yet seen a period of positive interest rates. Even in calm years, annual errors of 0.2 to 0.4 points are normal. The implication is important for the honesty of any claim about pension anxiety: an effect of 0.1 or 0.2 points cannot be established from the residual of a forecasting model alone. It can only be identified by combining timing restrictions, survey evidence and structural reasoning. That is what the next two sections do.

5. Isolating Rentenangst

5.1 Event-dummy estimates

The first identification strategy uses timing. If the pension debate raises saving, the saving rate should be higher, conditional on fundamentals, in the quarters in which the debate was intense. We re-estimate the equation over the full sample to the second quarter of 2026, adding a dummy that equals one from a given quarter onwards. Three start dates are tested: the fourth quarter of 2025 (pension package dispute), the first quarter of 2026 (commission at work, leaks) and the second quarter of 2026 (report).

Table 10. Pension-debate dummies in the full-sample saving-rate equation. Author's estimates.

Pension-debate dummy starting in Coefficient (pp) HAC standard error t-statistic p-value
2025 Q4 0.06 0.075 0.79 0.43
2026 Q1 0.07 0.069 1.06 0.29
2026 Q2 0.10 0.058 1.75 0.08

The estimates are small, positive in every variant, and rise as the window narrows toward the report. The 2026 Q2 dummy, at 0.10 points, is significant at the 10 percent level. With annualised disposable income of about EUR 2,790 billion, each 0.1 point of the saving rate corresponds to EUR 2.8 billion a year. The pension debate, on this estimate, was withholding about EUR 2.9 billion a year of consumption by mid-2026. The rising profile is what one would expect if the effect grows as the debate acquires concrete content.

5.2 Robustness

Table 11 shows how the two key quantities, the out-of-sample gap in the first half of 2026 and the 2026 Q2 dummy, change across six specifications.

Table 11. Robustness of the pension component across specifications. Author's estimates.

Specification R-squared Gap 2026 H1 (pp) Pension dummy from 2026 Q2 (pp) Standard error
Baseline (five regressors) 0.986 plus 0.02 0.10 0.06
With lagged saving rate 0.988 plus 0.12 0.13 0.07
Real rate instead of rate and inflation 0.985 minus 0.25 minus 0.03 0.09
Without policy uncertainty 0.986 minus 0.14 minus 0.03 0.05
Without unemployment 0.975 minus 0.18 0.00 0.07
Sample from 2014 Q1 0.987 plus 0.13 0.14 0.07

The range is instructive. In the specifications that fit the data best, the pension term lies between 0.10 and 0.14 points. In three restricted specifications it is zero. None of the specifications produces a large effect, and none produces a negative one of any size. The fair summary is that the national accounts are consistent with a pension-related saving component of between zero and about 0.15 points in the first half of 2026, with a central estimate of 0.10. That is the measured effect we carry forward. It is small, and the honest conclusion from the macro data alone is that the popular narrative overstates what has happened so far.

5.3 What households say

Two large representative surveys conducted around the publication of the commission’s report provide direct evidence on intentions. The IMK, the macroeconomic institute of the Hans Boeckler Foundation, surveyed about 4,000 people aged 18 to 75 between 17 June and 13 July 2026. The NIM surveyed 3,422 people in Germany between 20 May and 13 June 2026, with a follow-up of 1,000 people in early July.

Table 12. Survey evidence on pension worries and behaviour, mid-2026.

Survey result Share Source
Somewhat or very worried about financial security 81% IMK
of which very worried 34% IMK
Under-30s with at least some worry about 92% IMK
Reject the social-policy reform proposals 76% IMK
Reject a higher retirement age 73% IMK
Cutting spending because of the reform debate, all respondents 42% IMK
… among the very worried / somewhat worried / not worried 71% / 35% / 9% IMK
Confident they could live comfortably on the statutory pension 19% (US: 34%) NIM
Not confident 57% NIM
Would cut spending if pension contributions rise 35% NIM
Would cut private retirement saving if contributions rise 16% NIM
Say they know the reform package 27% NIM
Support a new funded pillar / a higher retirement age 31% / 16% NIM

Sources: IMK Policy Brief 222 (Boeckler Impuls 14/2026); NIM press release of 26 August 2026.

The survey picture is far more dramatic than the national accounts, but self-reported restraint is a poor guide to aggregate saving, for three reasons. The IMK debate covered health, long-term care and sick pay as well as pensions, so its 42 percent is not a pension-specific figure. Cutting spending at the margin, for instance postponing a purchase, may amount to a small fraction of income. And the respondents most likely to cut, the very worried, are disproportionately younger and lower-income households, whose share of aggregate saving is small. The IMK itself warns of possible panel bias. The NIM’s finding that only 27 percent know the reform package points in the same direction: worry is widespread, knowledge is not.

5.4 A monthly event study on the propensity to save

The NIM’s monthly survey allows a sharper test. The propensity to save moves for many reasons, chief among them changes in income expectations: when households expect lower income, they report more willingness to save. If pension news has an independent effect, months dominated by pension news should show a larger rise in saving intentions than the change in income expectations alone would predict.

We have monthly readings of the propensity to save, income expectations and economic expectations for November and December 2025 and for April to October 2026. We classify December 2025 (when the NIM explicitly cited the pension controversy) and July 2026 (the first survey after the commission report) as pension-news months. We estimate the relationship between monthly changes in saving propensity and income expectations on the five other months, then compute the abnormal change in the two pension-news months.

Table 13. Event study on the NIM propensity to save. Slope estimated on non-event months: minus 0.22 (standard error 0.12).

Survey month Change in propensity to save Change in income expectations Predicted change Abnormal change
December 2025 (pension news) plus 5.0 minus 6.8 plus 1.6 plus 3.4
May 2026 minus 2.2 plus 11.4 minus 2.4 plus 0.2
June 2026 0.0 plus 0.8 minus 0.1 plus 0.1
July 2026 (pension news) plus 3.1 minus 2.3 plus 0.6 plus 2.5
August 2026 minus 1.5 plus 16.2 minus 3.5 plus 2.0
September 2026 plus 6.0 minus 16.7 plus 3.7 plus 2.3
October 2026 minus 4.3 minus 0.7 plus 0.2 minus 4.5

Source: NIM press releases; author's calculations.

-4-20246minus 15minus 10minus 5051015Change in propensity to save2025-122026-052026-062026-072026-082026-092026-10Monthly change in income expectations, pointsPension-news monthsOther months (dashed: fit)
Figure 5. Monthly changes in the propensity to save against changes in income expectations. Source: NIM; author's calculations.

In the two pension-news months, the propensity to save rose by an average of 3.0 points more than income expectations predict. Among the other five months the abnormal changes average zero. The sample is tiny and the result is illustrative rather than statistically established: the non-event months have abnormal changes with a standard deviation of about 2.5 points, so each event month is roughly one standard deviation above normal. The September jump, the largest in the sample, is mostly explained by the collapse in income expectations that the NIM attributes to energy prices. But the pattern is consistent with the national accounts estimate: pension news raises the intention to save, by an amount that is noticeable but not dominant.

6. What Would a Rational Household Save?

6.1 The question

The previous section measured what households appear to be doing. This section asks what they should be doing, if they understood the commission’s proposals and responded to them as a standard life-cycle saver would. The comparison is informative in both directions. If measured saving exceeded the rational benchmark, Rentenangst would be an overreaction that clear communication could dispel. If it fell short, households would not yet have absorbed the reform, and further increases in saving would be likely as awareness spreads.

6.2 A calibrated pension-wealth model

The model computes, for each affected birth cohort, the expected loss of pension wealth implied by the reform, and the annual saving required to offset it by the time of retirement. The sum across cohorts is the rational offset saving flow in 2027. Two reform elements are modelled explicitly.

The first is the abolition of the deduction-free pension after 45 contribution years. In 2024, 262,400 people retired under this rule with an average pension of EUR 1,677 a month. For cohorts born from 1964, the rule allows retirement two years before the statutory age of 67. Without it, a person who keeps the planned retirement date faces deductions of 0.3 percent per month, 7.2 percent for life. Those who instead work two years longer lose leisure, not money, and need not save more. We treat the share who keep their date as uncertain, between 30 and 70 percent. We apply the loss to the ten cohorts retiring between 2027 and 2036, assuming no grandfathering, and weight it by the probability that the measure is legislated, between 50 and 90 percent given the political turbulence of October 2026.

The second is the link between the statutory age and life expectancy from 2032. Cohorts retiring between 2032 and 2041 face a rise of 0.6 to 6 months in the statutory age. About 856,800 people entered retirement in 2024, 500,400 early and 356,400 at the regular age. We value the pension of a typical new retiree at 30 to 40 earnings points at the 2026 pension value of EUR 42.52 per point. Those who keep their retirement date face deductions of 0.3 percent per month of the increase.

Each cohort spreads its offset over its remaining working years at a real return of 1 to 3 percent, and values pension wealth over a retirement of 18 to 22 years. A third calculation converts a perceived across-the-board pension cut of 1 percent, applied to the fifteen cohorts retiring between 2027 and 2041, into an annual saving flow. It does not represent the commission’s proposals; it measures what fear of broader cuts would do.

Table 14. Parameters of the pension-wealth model.

Parameter Distribution or value Basis
Users of deduction-free pension per cohort 262,400 Deutsche Rentenversicherung 2024, via t-online
Their average pension EUR 1,677 per month Same
Deduction if retirement date is kept 7.2% (24 months x 0.3%) Statutory deduction rules
New retirees per cohort 856,800 2024 entries, via t-online
Pension value EUR 42.52 per point (2026) Pension adjustment 2026
Earnings points of typical new retiree uniform 30 to 40 Calibration
Share keeping planned retirement date uniform 0.3 to 0.7 Calibration
Probability of legislation uniform 0.5 to 0.9 Calibration, after 12 October 2026
Retirement duration uniform 18 to 22 years Calibration
Real return on offset saving uniform 1% to 3% Calibration
Draws 200,000 Fixed seed

Sources as indicated; calibrations are the author's.

6.3 Results

Table 15. Rational offset saving implied by the commission's proposals. Author's calculations, 200,000 draws.

Component 10th percentile Median 90th percentile Median, pp of disposable income
Abolition of deduction-free pension (EUR bn per year) 3.9 6.0 8.8 0.22
Retirement-age link (EUR bn per year) 0.4 0.7 1.0 0.02
Total rational offset (EUR bn per year) 4.4 6.7 9.8 0.24
Per 1% perceived across-the-board cut (EUR bn per year) 6.8 8.0 9.3 0.29
Measured in national accounts, 2026 Q2 dummy 2.9 0.10
00.10.20.30.4MeasuredOffset: Rente mit 63Offset: age linkOffset: totalPer 1% perceived cutpp of disposable income0.100.220.020.240.29
Figure 6. Measured pension-related saving versus the calibrated offset model. Bars show medians, lines the 10th to 90th percentile range (for the measured value, the 90 percent confidence interval). Source: author's calculations.

Three results stand out. First, the rational response to the reform is modest in macroeconomic terms: about 0.24 points of disposable income, or EUR 6.7 billion a year, with an 80 percent range of 0.16 to 0.35 points. For a reform that the public debate has described as a fundamental break, the aggregate wealth effect is limited, because the transition factor protects the replacement rate at entry and the age link is gradual.

Second, the response is highly concentrated. Ninety percent of it comes from the abolition of the deduction-free pension, and 79 percent of that from the five cohorts closest to retirement, who must save a 7.2 percent lifetime loss in a few years. Before the reform, a full cohort of these retirees would lose pension wealth of about EUR 6.2 billion if all kept their planned date. That is the core of the pension wealth shock, and it falls on a group, long-insured workers in their late fifties and early sixties, that is visible, organised and politically salient.

Third, measured saving is below the rational benchmark. The 0.10 point estimate is 43 percent of the model’s median, with a range of 29 to 65 percent across the model’s distribution. This awareness ratio is strikingly close to the NIM finding that only 27 percent of Germans know the reform package, and it supports a clear interpretation: households are not overreacting to the reform; most have not yet translated it into personal consequences. As awareness rises, rational offset saving will rise toward the benchmark even if nothing else changes.

6.4 Expected loss or uncertainty? The precautionary premium

The word Rentenangst suggests that fear of the unknown, rather than a known loss, drives behaviour. A standard way to measure this is the precautionary premium: the extra saving a household with prudent preferences undertakes because a loss is uncertain, compared with a certain loss of the same expected size. We compute it in a two-period model with constant relative risk aversion, where retirement resources are hit by a loss with given probabilities.

Table 16. Precautionary premium: extra saving relative to a certain loss of equal expected value. Author's calculations.

Loss distribution Risk aversion 2 Risk aversion 4
Content risk: 7.2% loss with probability 0.7 plus 3% plus 5%
Fat tail: 7.2% with probability 0.6, 20% with probability 0.15 plus 9% plus 15%

The premium is small when uncertainty concerns only the content of the proposals: a 3 to 5 percent addition to the expected-loss response. It becomes material, 9 to 15 percent, only when households attach a meaningful probability to much larger cuts, such as the “pension at 70” headlines that circulated before the report. The implication is that most of what matters is the perceived expected loss, not uncertainty as such. That is good news for policy. Perceived losses can be corrected by information; pure uncertainty is harder to dispel. Each percentage point of perceived across-the-board pension cut adds 0.29 points to the saving rate, three times the currently measured effect, so the battle over perceptions is worth fighting.

6.5 Two channels

The concentration of the rational response on cohorts near retirement matters for its macroeconomic effect. These households are in their peak saving years, so offset saving is rational, limited and fairly predictable. The surveys, by contrast, find the most worry among the young: 92 percent of under-30s report at least some financial worry. Young households cannot save much, but they can postpone large purchases, as the persistently low propensity to buy suggests. The pension debate thus acts through rational offset saving among older workers and through perception-driven spending postponement among younger ones; the second channel is the more volatile.

7. The Cost to Consumption and Growth

7.1 From saving to GDP

Each percentage point of the saving rate corresponds to about EUR 27.9 billion of annual consumption at 2026 income levels. Not all of that is lost to German GDP: part of consumption is imported. Assuming an import content of consumption of 25 to 35 percent and a short-run multiplier on domestic value added between 0.8 and 1.3, centred on 1, one point of the saving rate is worth about 0.44 percent of GDP, with nominal GDP of roughly EUR 4,450 billion.

Table 17. GDP level cost of excess saving, percent of GDP, multiplier of 1. Author's calculations.

Excess saving (pp of income) Import share 25% Import share 30% Import share 35%
0.1 0.05 0.04 0.04
0.2 0.09 0.09 0.08
0.3 0.14 0.13 0.12
0.5 0.24 0.22 0.20
0.7 0.33 0.31 0.29
1.0 0.47 0.44 0.41

7.2 The cost so far

At the measured 0.10 points, the pension debate cost about EUR 2.9 billion of consumption and 0.04 percent of GDP at an annual rate by mid-2026. That is small relative to the forces discussed in Section 4: the end of the ECB easing cycle and the energy shock to real income each weighed on consumption five to ten times more heavily. Rentenangst is not, so far, the main reason why German consumption stalled.

7.3 The second half of 2026

The national accounts for the third quarter will be published in late November. Our model allows a conditional nowcast. The fundamental saving rate is 10.05 percent in the third quarter and 10.14 percent in the fourth, held up by the ECB’s two rate increases. If the pension component rises to about 0.15 points, in line with the July and September jumps in saving intentions, the observed saving rate would be about 10.2 percent in the third quarter and 10.3 percent in the fourth. A third-quarter reading of 10.4 percent or above would signal that the pension effect is growing faster than our central path; a reading of 10.0 percent or below would suggest that the September saving intentions were mostly energy-driven and transitory.

8. Scenarios for 2027

8.1 Three paths

How the pension component evolves in 2027 depends on legislation and communication. We define three scenarios, each anchored in the model of Section 6.

Table 18. Scenario definitions for 2027. Author's assumptions, calibrated on the model in Section 6.

Scenario Legislation and communication Path of pension-related saving Level end-2027 (pp)
Clarity Cabinet draft by end-2026, law in Q1 2027; cohorts retiring before 2032 grandfathered; 48 percent guarantee and funded pillar clearly explained Peaks at 0.15 in H2 2026, falls to the rational residual after grandfathering 0.08
Drift No draft until mid-2027, contested in coalition; content gradually understood; modest perceived broader cuts (0.25 percent) Rises linearly as awareness converges on the rational offset 0.31
Disorder Piecemeal and contested reform; retroactive changes, new leaks; perceived broader cuts of 1.5 percent Rises as awareness and perceived losses grow together 0.67

In the Clarity scenario, grandfathering the cohorts that would retire before 2032 removes the steepest part of the offset requirement, because those cohorts would otherwise have to accumulate a 7.2 percent lifetime loss in one to five years. With legislation certain and the near cohorts protected, the rational offset falls to about 0.10 points, which we treat as the long-run residual of a reform that genuinely asks later cohorts to work longer. In Drift, awareness converges on the full rational offset of 0.24 points by the end of 2027, plus a small perception premium. In Disorder, households additionally price in a 1.5 percent cut across the board, which at 0.29 points per percentage point adds about 0.43 points.

00.20.40.62026Q12026Q22026Q32026Q42027Q12027Q22027Q32027Q4pp of disposable incomeEstimated, 2026H1DisorderDriftClarity
Figure 7. Three paths for pension-related excess saving to the end of 2027. Source: author's calculations.

8.2 Costs in 2027

We run 200,000 draws for each scenario, varying the level of excess saving by plus or minus 30 percent around the path, the import share between 25 and 35 percent and the multiplier between 0.8 and 1.3.

Table 19. GDP cost of pension-related saving in 2027 under three scenarios. Author's calculations.

Scenario Average excess saving in 2027 (pp) Consumption withheld (EUR bn) GDP cost, median (% of GDP) 10th to 90th percentile
Clarity 0.10 2.7 0.04 0.03 to 0.06
Drift 0.26 7.2 0.12 0.09 to 0.15
Disorder 0.53 14.7 0.24 0.18 to 0.31
Weighted (35 / 45 / 20) 0.26 7.2 0.11
00.10.20.3ClarityDriftDisorderGDP level cost in 2027, pp0.04 pp, EUR 2.7 bn0.12 pp, EUR 7.2 bn0.24 pp, EUR 14.7 bn
Figure 8. GDP cost of pension-related saving in 2027 by scenario. Source: author's calculations.

The difference between Clarity and Disorder is 0.2 percent of GDP, about EUR 12 billion of consumption in a single year. It is a cost that can be avoided, or incurred, without any change in the budget. Weighting the scenarios by our subjective probabilities (35 percent Clarity, 45 percent Drift, 20 percent Disorder, reflecting the absence of a draft bill and the minister’s resignation) gives an expected 2027 cost of 0.11 percent of GDP.

8.3 What the scenarios do not capture

Three effects lie outside the scenarios. First, the funded pillar from 2028 adds 0.5 points of contributions a year; whether this raises measured household saving depends on whether the fund is classified with households or with social security, and a credible mandatory pillar could crowd out part of today’s voluntary precautionary saving. Second, longer working lives raise income and partly offset the consumption cost, mostly after 2027. Third, monetary policy remains the dominant driver: each percentage point on the ECB deposit rate moves the fundamental saving rate by about 0.2 points, more than any pension scenario.

9. Policy Implications

The analysis supports the substance of the commission’s proposals. The German pay-as-you-go system faces a contribution rate rising toward 21.5 percent by 2050 under current law, and the commission’s combination of a longer working life, a funded pillar and a broader contributor base is a coherent response. The aggregate wealth effect of the proposals on households is limited. The macroeconomic risk lies not in what is proposed but in how it is legislated. Five implications follow.

First, legislate quickly. Every quarter of delay allows awareness of losses to spread without the reassurance of a final, stable rule. Our Drift scenario costs three times as much as Clarity in 2027, mainly because uncertainty persists. The government’s own timetable, a cabinet decision by the end of 2026 and a Bundestag vote in the first quarter of 2027, should be treated as a ceiling, not a target, and the appointment of a new labour minister should not reopen the package.

Second, grandfather cohorts close to retirement. The abolition of the deduction-free pension after 45 years is the single largest source of rational offset saving, and almost four fifths of it falls on people within five years of retirement. A transition that protects those who would retire before 2032 would cost little in long-run sustainability, because the rule is phased out within a decade anyway, but it would remove most of the immediate saving shock and much of the political resistance.

Third, communicate the guarantee, not just the cuts. The transition factor that keeps the replacement rate at entry at least at 48 percent is the most reassuring element of the package, and it is almost invisible in public discussion. Our precautionary-premium calculation shows that perceived expected losses, not uncertainty as such, drive saving. Each percentage point of perceived across-the-board cuts costs 0.29 points of the saving rate. A personal digital pension overview, as recommended by the commission, should be accelerated so that every insured person can see the effect of the reform on their own entitlement.

Fourth, make the funded pillar an asset, not a tax. Only 31 percent of respondents support a new funded pillar. If contributions are perceived as an extra levy, they will depress consumption; if they are perceived as personal wealth, visible in individual accounts, they can replace voluntary precautionary saving. The Swedish model invoked by the commission works precisely because contributions are individually owned and reported annually.

Fifth, measure what matters. Germany lacks a timely indicator of pension-related saving. The NIM consumer survey could add a monthly question on saving for old age, and the Bundesbank could publish the quarterly flow into old-age provision products separately in its financial accounts. Without such data, the debate will continue to oscillate between survey alarms and national accounts that are too coarse to settle it.

10. Conclusion

German households in 2026 have behaved less irrationally than the word Rentenangst suggests. The consumption freeze of 2026 is real and severe: real consumption growth fell from 1.6 to 0.1 percent within two quarters. But it is above all a saving freeze driven by fundamentals. Monetary easing in 2024 and 2025 pulled the saving rate down from 11.1 to 10.0 percent and fuelled consumption in 2025; when the ECB stopped cutting and then began to hike again, and when the energy shock slowed real income growth, the decline in saving came to an end. Our saving-rate equation, estimated on data that end before the pension debate began, predicts the 2026 saving rate almost exactly.

Pension anxiety is nevertheless visible. A pension-debate term adds about 0.10 points to the saving rate from mid-2026, and saving intentions jump by about 3 points beyond normal in months dominated by pension news. The original contribution of this paper is to compare that measured effect with what a rational household would save in response to the commission’s proposals. The answer, 0.24 points of income or EUR 6.7 billion a year, concentrated on long-insured workers close to retirement, is more than twice the measured effect. Households have not yet priced in the reform.

That is the central message for 2027. The cost of Rentenangst so far, about 0.04 percent of GDP, is small. Its potential cost, if the reform drifts or becomes disorderly, is three to six times larger, up to 0.24 percent of GDP and EUR 14.7 billion of consumption in a single year. Germany needs this reform. Whether it costs the economy a few billion euros or several times that depends on choices that cost the budget nothing: speed, grandfathering for those near retirement, and a clear account of what the reform guarantees as well as what it takes away.

About this series. This paper is the seventh in a ten-part series on the German economy in 2026 published by Global Strategy Files. The other papers examine the absorption of the EUR 500 billion infrastructure fund, industrial energy elasticity in the second energy shock, the irreversibility threshold of German chemicals, a county-level map of the China shock in the automotive industry, rearmament and labour reallocation, the end of Bund scarcity, the municipal fiscal crisis, the insolvency wave, and the Beveridge curve paradox of a shrinking workforce.

Annex A. Methodology

Data. Quarterly seasonally adjusted disposable income, saving and the saving rate of private households (2011 Q2 to 2026 Q2) are taken from the Bundesbank’s seasonally adjusted economic statistics, which reproduce the Destatis national accounts. Consumer price inflation is the OECD consumer price index for Germany (FRED series CPALTT01DEQ659N) to 2025 Q1, extended with monthly rates from Destatis releases and published index values. The ECB deposit facility rate is converted into quarterly averages from the ECB’s table of key interest rates. The unemployment rate is the Federal Employment Agency’s annual average rate in percent of dependent civilian labour force (Destatis long series), interpolated linearly with each annual value placed at mid-year; the 2026 value scales the 2025 rate by the ratio of about 3.0 million unemployed to the 2025 average of 2.95 million. The Economic Policy Uncertainty Index for Germany is the monthly Baker, Bloom and Davis index (FRED series DEEPUINDXM), averaged by quarter. Real private consumption growth is taken from the Bundesbank’s key indicators table. Consumer survey indicators are taken from NIM press releases.

Saving-rate equation. Ordinary least squares, 2012 Q2 to 2024 Q4 (51 quarters), HAC standard errors with four lags, eight quarterly pandemic dummies (2020 Q1 to 2021 Q4). Out-of-sample prediction intervals are standard OLS prediction intervals at 90 percent. Pension dummies are estimated on the full sample to 2026 Q2. Rolling forecasts re-estimate the model on samples ending in the fourth quarter of 2016, 2017, 2018, 2022 and 2023 and predict the following four quarters.

Consumption decomposition. Nominal consumption growth is (1 + g) x (1 minus s(t)) / (1 minus s(t minus 4)) minus 1, where g is nominal disposable income growth over four quarters. Real consumption is approximated by subtracting CPI inflation.

Event study. Monthly changes in the NIM propensity to save are regressed on changes in income expectations over the five months without dominant pension news (May, June, August, September and October 2026). Abnormal changes are observed minus predicted values. November 2025 values are derived from the December 2025 release.

Pension-wealth model. For each affected cohort retiring in year T, the expected pension-wealth loss L is the number of retirees x monthly pension x 12 x deduction x annuity factor (retirement duration D, real rate r) x share keeping the retirement date x probability of legislation. The annual offset saving of a cohort with n years to retirement is L x r / ((1 + r)^n minus 1). Flows are summed over cohorts retiring 2027 to 2036 (deduction-free pension) and 2032 to 2041 (age link, 0.6 to 6 months). The grandfathered variant excludes cohorts retiring before 2032 and sets the probability of legislation to one. 200,000 draws, seed 2026.

Precautionary premium. Two-period model, u(c) = c^(1 minus g) / (1 minus g), first-period income 1, retirement resources 1 minus loss plus saving, zero interest. The premium is the ratio of optimal saving under the loss distribution to optimal saving under a certain loss of the same mean, minus one.

Scenarios and GDP cost. Quarterly paths of pension-related excess saving from 2026 Q3 to 2027 Q4. The GDP cost is excess saving x annual disposable income (EUR 2,790 billion) x (1 minus import share) x multiplier / nominal GDP (EUR 4,450 billion). Monte Carlo: excess level uniform plus or minus 30 percent, import share uniform 0.25 to 0.35, multiplier triangular (0.8, 1.0, 1.3). Scenario weights are the author’s judgement.

Limitations. The saving-rate equation is a reduced form on 51 quarters, with negative interest rates for much of the sample; its out-of-sample errors are of the same order as the effect sought, so the pension estimate is suggestive rather than conclusive. Unemployment is interpolated from annual data, the 2025 Q2 to Q4 CPI comes from a secondary source, and the event study rests on seven months. The pension-wealth model abstracts from labour supply, within-cohort heterogeneity and the funded pillar, and its parameters are calibrations. The scenarios are conditional projections, not forecasts. The qualitative conclusion, a real but small pension effect below the rational benchmark, is robust to these limitations.

Annex B. Data Tables

Table B1. Annual data used in the saving-rate equation (selected years).

Year Saving rate, annual average of quarters (%) CPI inflation (%) ECB deposit rate, average (%) Unemployment rate (%)
2012 9.8 2.0 0.13 7.6
2014 10.0 0.9 minus 0.09 7.5
2016 10.0 0.5 minus 0.38 6.8
2018 10.9 1.7 minus 0.40 5.8
2019 10.3 1.4 minus 0.43 5.5
2022 10.1 6.9 0.07 5.8
2023 10.2 6.0 3.30 6.2
2024 11.0 2.3 3.73 6.5
2025 10.1 2.2 2.26 6.8
2026 (H1) 10.1 2.4 2.02 about 6.9

Sources: Bundesbank, OECD via FRED, Destatis, ECB, Federal Employment Agency; author's averages.

Table B2. Additional data used in this paper.

Indicator Value Date Source
Saving rate, not seasonally adjusted 9.6% 2026 Q2 Destatis via cash-online
Saving rate, not seasonally adjusted 12.6% 2026 Q1 Destatis
Gross saving rate 18.9% 2025 and H1 2026 Destatis
Gross saving rate, EU / US / Switzerland 14.3% / 10.7% / 27.4% 2025 Destatis
Real private consumption plus 1.3% 2025 Bundesbank
GDP plus 0.2% q/q, plus 1.3% y/y 2026 Q3 Destatis
EPU index for Germany 749 (peak 1,502 in April 2025) October 2026 FRED
Pension contribution rate 18.6% (20.6% with funded pillar) 2026 BMAS
Total pension entries 500,400 early, 356,400 regular 2024 t-online

Annex C. A Rentenangst Dashboard

Because the pension component is small relative to the noise in the national accounts, it should be tracked with a combination of indicators. We propose five.

Table C1. Rentenangst dashboard. Thresholds: author's proposal.

Indicator Publication Warning threshold Current reading
Saving rate gap to model (this paper) Quarterly, Bundesbank Above plus 0.3 pp for two quarters plus 0.06 pp (2026 Q2): calm
Abnormal change in propensity to save Monthly, NIM Above plus 3 points in a pension-news month plus 2.5 (July 2026): watch
Propensity to save, level Monthly, NIM Above 20 points 17.2 (October), 21.5 (September): watch
Draft bill status Federal government No cabinet decision by end-2026 No draft (November 2026): warning
Share who know the reform Ad hoc surveys Below 50 percent after legislation 27% (mid-2026): warning

References

Baker, S., Bloom, N. and Davis, S. (2016). Measuring Economic Policy Uncertainty. Quarterly Journal of Economics 131(4). Data: Economic Policy Uncertainty Index for Germany, FRED series DEEPUINDXM, accessed November 2026.

Behringer, J., Dullien, S. and Endres, L. (2026). Sozialstaatsdebatte belastet privaten Konsum und Vertrauen in die Regierung. IMK Policy Brief 222; Boeckler Impuls 14/2026.

Bundesministerium fuer Arbeit und Soziales (2026). Faktenpapier zum Bericht der Alterssicherungskommission. Berlin, 23 June 2026.

cash-online (2026). Rentenkommission: Fahrplan; Konsumklima sinkt trotz besserer Konjunkturerwartung. October and November 2026.

Destatis (2026). Press releases on GDP (22 May 2026; 30 October 2026), consumer prices (September and October 2026) and the gross saving rate (Zahl der Woche 44/2026); long series on registered unemployment.

Deutsche Bundesbank (2026). Seasonally adjusted economic statistics: disposable income and saving of households; key economic indicators; financial accounts press release for the second quarter of 2026.

European Central Bank (2026). Key ECB interest rates.

Federal Reserve Bank of St. Louis (2026). FRED series CPALTT01DEQ659N (OECD consumer prices, Germany) and ECB key rates.

gegen-hartz.de (2026). 33 Aenderungen bei der Rente in der Uebersicht.

Institut der deutschen Wirtschaft (2026). Autumn forecast 2026, as reported in November 2026.

NIM, Nuremberg Institute for Market Decisions (2025, 2026). Konsumklima powered by GfK, press releases of December 2025 and May to October 2026; press release on the pension reform study, 26 August 2026.

t-online (2026). Rente mit 63 vor dem Aus: Rentenkommission empfiehlt Abschaffung. 22 June 2026.

vermoegenszentrum.de (2026). Rentenreform: Zeitplan und Inhalte. Updated 3 November 2026.

ZDFheute (2026). Rentenreform: Das sind die 33 Vorschlaege der Rentenkommission. June 2026.

Driss EFFINA, PhD
Independent Researcher, Global Strategy Files

Free preview
The Q4 2026 Global Outlook

The opening pages of our four country reports: United States, United Kingdom, Japan and China. A free PDF, sent to your inbox.

You will also receive occasional analysis from Global Strategy Files. Unsubscribe in one click, at any time.