Social security pension and the effect on household saving

• Published date: 01 July 2024
• Author: Elin Halvorsen,Zhiyang Jia,Herman Kruse,Trond C. Vigtel
• Date: 01 July 2024
• DOI: http://doi.org/10.1111/sjoe.12550

Scand. J. of Economics 126(3), 529–560, 2024

DOI: 10.1111/sjoe.12550

Social security pension and the effect

on household saving

Elin Halvorsen

Statistics Norway, NO-0131 Oslo, Norway

elin.halvorsen@ssb.no

Zhiyang Jia

Statistics Norway, NO-0131 Oslo, Norway

zhiyang.jia@ssb.no

Herman Kruse

Statistics Norway, NO-0131 Oslo, Norway

herman.kruse@ssb.no

Trond C. Vigtel

Statistics Norway, NO-0131 Oslo, Norway

trond.vigtel@ssb.no

Abstract

In this paper, we examine the substitution between pension wealth and household saving. To

identify the effect of reductions in social security pension wealth on household saving, we

utilize variations in changes in social security pension wealth induced by Norway’s 2011 reform

across different cohorts, time periods, and sectors. Our study focuses on the saving behaviors of

individuals between the ages of 57 and 61, and we f‌ind that the annual saving rate increased by

around 1.4 percentage points after the reform. When considering the overall life-cycle changes

in household saving, this corresponds to a crowd-out effect of about 50 percent of the total loss

in pension wealth.

Keywords: Difference-in-difference; household saving; pension reform; quasi-natural

experiment

JEL classif‌ication:D14; E21; H55

1. Introduction

How do less generous pensions affect households’ saving behavior? The

prospect of an aging population and increased f‌iscal burden has spurred pension

reforms in many European countries, lowering expected pension benef‌its, and

thus transferring more responsibility onto households to save privately for

retirement. Although less generous pension systems might provide f‌iscal

sustainability, they might also harm pension adequacy for the recipients.

Extensive attention has been given to how individuals respond to reduced

c

2023 The Authors. The Scandinavian Journal of Economics published by John Wiley & Sons Ltd on behalf of F¨

oreningen

f¨

or utgivande av the SJE.

This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License,

which permits use and distribution in any medium, provided the original work is properly cited, the use is non-commercial

and no modif‌ications or adaptations are made.

530 Social security pension and the effect on household saving

pension generosity by prolonging their working careers,1but less attention

has been given to whether individuals follow the consumption-smoothing

behavior often assumed in standard life-cycle theory.2We aim to provide

causal evidence on the saving responses to reduced pension generosity by

using a quasi-natural experiment setting from a Norwegian pension reform.

Theoretically, the effect of a social security pension system on households’

saving behavior is ambiguous. In a life-cycle framework, social security

pension systems can be viewed as mandatory saving that reduce individuals’

need to save for retirement on their own. This phenomenon is referred to as

the “crowd-out” effect, where an increase in one saving component crowds

out the other. In other words, more generous social security pensions lead to

a decrease in private saving, and vice versa. However, changes to expected

pension wealth/benef‌its can also have an effect on labor supply on both

the extensive and intensive margins. If the pension system causes people

to retire earlier, thus extending the retirement period, this might stimulate

saving instead. Additionally, social security pension wealth is typically an

illiquid asset, which further complicates theoretical predictions regarding the

substitution effect between mandatory social security pension accrual and

private saving. Thus, the question of whether a social security pension system

increases or decreases personal savings depends on the relative strength of

these factors and remains an empirical question.

In this paper, we utilize a pension reform implemented in Norway in

2011 to study the private saving responses to changes in accumulated social

security pension wealth. The 2011 Norwegian pension reform introduced

changes to the accrual system, resulting in varying impacts on individuals’

pension wealth based on their birth cohort and occupational sector. There

are two key features of the reform that we use for identif‌ication. First, the

pension reform introduced a new pension wealth accrual system with a cutoff

on birth cohort. Individuals born in 1953 or earlier had their entire pension

accrual from the National Insurance System (NIS) determined by the old

system and were not affected by the new accrual system. Individuals born

in 1954 and later were gradually transitioned to the new system, which is

generally less generous (see Section 2for details). This gives rise to the

“pre” (1953 and earlier) and “post” (1954 and later) cohorts. Second, public

sector workers experienced less exposure to the detrimental effects of the

1See, for example, Song and Manchester (2007), Haider and Loughran (2008), Brown (2013),

Vestad (2013), Hernæs et al. (2016,2021), Manoli and Weber (2016), and Brinch et al. (2015).

2However, following the inf‌luential paper by Feldstein (1974), there have been some occasional

studies such as Gale (1998), Attanasio and Brugiavini (2003), Attanasio and Rohwedder (2003),

Bottazzi et al. (2006), Aguila (2011), Engelhardt and Kumar (2011), Feng et al. (2011),

Lachowska and Myck (2018), and Lindeboom and Montizaan (2020).

c

2023 The Authors. The Scandinavian Journal of Economics published by John Wiley & Sons Ltd on behalf of F¨

oreningen

f¨

or utgivande av the SJE.

E. Halvorsen et al. 531

reform compared with their private sector counterparts. This is because public

sector workers had access to a def‌ined benef‌it occupational pension scheme,

ensuring an annual pension benef‌it payout equal to 66 percent of their f‌inal

year of earnings. This counteracted the less generous pension wealth resulting

from the reformed NIS. Therefore, we use the private sector workers as our

“treatment group” and public sector workers as our “control group”, and we

expect the treatment group to experience a more pronounced negative effect

on their expected pension wealth. A crowd-out effect through private saving

would therefore be indicated by an increase in saving among the private

sector workers relative to their public sector counterparts.3The structure

of the reform described above calls for a difference-in-differences (DiD)

design. To address potential endogeneity problems regarding the crowd-out

effect, we extend our study by employing the two-stage least-squares (2SLS)

method to estimate the magnitude of the saving response. In addition, we

correct for the re-optimization across the remaining life-cycle by using

Gale’s Q (Gale, 1998). Our unit of analysis is the household, and most of

our observations are individuals with a spouse. Only a few (approximately 35

percent) are not married, and even fewer are lifetime singles (never married).

We f‌ind that for those aged 57–61, the reform induced on average an

increase in the saving rate of 1.4 percentage points, corresponding to a

f‌ive-year increase in household saving of approximately NOK 21,400 (≈EUR

2,100). These f‌indings suggest a strong substitution effect between social

security pension wealth and private savings. After correcting re-optimization

with Gale’s Q, we obtain a crowd-out effect of about 50 percent, meaning

that households save privately approximately 50 percent of the lost pension

wealth. This falls within the mid-range of the estimates reported in the

literature, which span from 10 to 75 percent.4

For the time being there is no consensus regarding the magnitude of

such offsetting effects between pension generosity and private saving. The

contributions to the literature vary across country, time period, identif‌ication

3Although the public sector workers are systematically less exposed to the reform, the 66 percent

gross replacement rate from the public occupational pension scheme did not relieve public sector

workers entirely from any reform exposure. In the odd case that some public sector workers lost

expected pension wealth, this will pull the responses in our control group in the same direction as

our treatment group, and therefore our estimates should be interpreted as lower bound estimates.

4Feldstein (1974) reports that a one dollar increase in social security wealth depresses private

saving by about 40 cents, which is a crowd-out effect of 0.40; Gale (1998) estimates the effect

to be 0.50; Attanasio and Brugiavini (2003) report a range of effects between 0.30 and 0.70;

Attanasio and Rohwedder (2003) report estimates between 0.65 and 0.75; Bottazzi et al. (2006)

estimate it to be 0.65; Aguila (2011) reports it to be 0.50; Engelhardt and Kumar (2011) f‌ind it to

be between 0.53 and 0.67; Feng et al. (2011) estimate it to be between 0.10 and 0.16; Lachowska

and Myck (2018) report it to be 0.24; and Lindeboom and Montizaan (2020) estimate an overall

effect of 0.38.

c

2023 The Authors. The Scandinavian Journal of Economics published by John Wiley & Sons Ltd on behalf of F¨

oreningen

f¨

or utgivande av the SJE.