Income Distribution and Aggregate Saving: A Non‐Monotonic Relationship

• Published date: 01 December 2019
• Author: Peter Bofinger,Philipp Scheuermeyer
• Date: 01 December 2019
• DOI: http://doi.org/10.1111/roiw.12376

© 2018 Internationa l Association for Res earch in Incom e and Wealth

872

INCOME DISTRIBUTION AND AGGREGATE SAVING:

A NON-MONOTON IC RELATIONSHIP

by Peter bofinger and PhiliPP Scheuermeyer*

University of Würzb urg

Drawing on a panel of advanced economies, this paper documents a concave and non-monotonic link

between inequality and the aggregate household saving rate. We find that, at a low level of inequality,

more inequality is associated with higher saving; but we also show that a negative relationship between

inequality and saving prevails where inequality is high. Using different empirical approaches, we locate

the turning point, where the marginal effect of inequality turns from positive to negative, at a net

income Gini coefficient of around 30. Moreover, we show that the relationship between inequality and

saving also depends on financial market conditions. While inequality increases saving, when credit is

scarce it tends to reduce saving at high levels of credit. This paper primarily focuses on household sav-

ing, yet we also find some evidence for a non-monotonic effect of inequality on private saving, national

saving, and the current account balance.

JEL Codes: C23, D31, E21, O5

Keywords: saving, income distribution, panel data, non-linearities

1. introduction

Is there an empirical link between income distribution and aggregate saving?

This paper suggests yes, but in a non-monotonic way. It suggests that at a low level

of inequality, more inequality is associated with higher saving, but it also shows

that a negative relationship between inequality and saving prevails at high levels

of inequality.

Given the secular rise in income inequality, economists increasingly focus on

the macroeconomic implications of this development. A link between inequality

and saving lies at the heart of this literature. For instance, the debate about secu-

lar stagnation has drawn new attention to the Keynesian idea that rising inequal-

ity increases the aggregate propensity to save and thus exerts a drag on aggregate

demand (e.g. Eggertsson and Mehrotra, 2014; Summers, 2015). Assuming the

same positive relationship between inequality and saving, but coming to a different

conclusion, the neoclassical growth literature suggests that inequality promotes

economic performance by fostering capital accumulation (Bourguignon, 1981).

Note: We thank the editor, Prasada Rao, and three anonymous reviewers, as well as Kai Daniel

Schmid and Sebastian Rüth, for helpful comments and suggestions.

*Correspondence to: Philipp Scheuermeyer, Department of Economics, University of Würzburg,

Sanderring 2, 97070 Würzburg, Germany (philipp.scheuermeyer@uni-wuerzburg.de).

Review of Inc ome and Wealth

Series 65, Numb er 4, December 2 019

DOI : 10.1111 /roi w.123 76

bs_bs_banner

Review of Income and Wealth, Series 65, Number 4, December 2019

873

© 2018 Internationa l Association for Res earch in Incom e and Wealth

Yet, with regard to global current account imbalances, some studies argue that an

increase in inequality lowers private saving and the current account (Al-Hussami

and Remesal, 2012; Ranciere et al., 2012; Behringer and van Treeck, 2013).

Although household saving constitutes a common transmission variable in

all these strands of literature, the link between income inequality and saving is

theoretically and empirically unclear. As richer households tend to have a higher

propensity to save than households at the lower end of the income distribution (e.g.

Dynan et al., 2004), an increase in income inequality may cause a rise in aggregate

saving (Keynes 1936, 1939). Yet, if households engage in upward-looking interper-

sonal comparison, middle- and low-income earners might lower their saving rate

in response to rising top incomes (Drechsel-Grau and Schmid, 2014; Bertrand and

Morse, 2016). Thus an increase in inequality could just as well trigger expenditure

cascades and a decline in aggregate saving (Alvarez-Cuadrado and El-Attar Vilalta

2012; Frank et al. 2014).

In line with the theoretical ambiguity, cross-country and panel-data studies

that investigate the effect of inequality on national or private saving rates often

remain inconclusive (Schmidt-Hebbel and Servén, 2000; Li and Zou, 2004; Leigh

and Posso, 2009). With regard to household saving, some studies find a negative

effect of inequality, albeit they rely on samples of only a few countries (Leigh and

Posso, 2009; Alvarez-Cuadrado and El-Attar Vilalta, 2012; Behringer and van

Treeck, 2013).

The present study is the first to primarily focus on household-sector saving

rates, which we prefer over national or private saving rates due to a more direct

connection to the theories of interest. By combining saving rates from OECD data-

bases with net income Gini coefficients from the Luxembourg Income Study, the

paper rests on a panel of highly consistent data. Moreover, the Standardized World

Income Inequality Database was used to generate a large alternative sample, with

792 observations from 29 advanced economies.

Consistent with the theoretical ambiguity and the inconclusiveness of the

empirical literature, we do not find a clear linear correlation between inequality

and saving. However, we reveal a highly significant hump-shaped relation between

inequality and saving that is robust to a large set of controls, including equity and

house prices, credit availability, and financial liberalization. We find that the impact

of inequality on saving is positive at low levels of inequality, whereas it becomes

negative after some turning point, which is located at a Gini coefficient of between

28 and 32. This hump-shaped pattern is robust to different data sources, estimation

techniques, measures of inequality, and sample compositions.

As the availability of credit financing might be a precondition for expendi-

ture cascades (see, e.g., Rajan, 2010; Frank et al., 2014; Bertrand and Morse, 2016)

we also test whether the impact of inequality interacts with credit availability and

financial market liberalization. We find that rising inequality tends to reduce sav-

ing if financial markets are widely liberalized or the ratio of credit to GDP is high.

Nonetheless, in both a low-credit and high-credit environment, the hump-shaped

relationship between inequality and saving prevails.

Review of Income and Wealth, Series 65, Number 4, December 2019

874

© 2018 Internationa l Association for Res earch in Incom e and Wealth

While we primarily focus on household saving rates, we find some evidence

that the hump-shaped effect of inequality also appears for private saving rates,

national saving rates, and the current account balance.

The paper proceeds as follows. Section 2 describes the theoretical background

to the analysis. Section 3 briefly reviews the recent empirical literature on the

household, state, and cross-country level. Section 4 describes the data, focusing on

measures of saving and income distribution. Section 5 reports our baseline regres-

sion results, followed by an extensive sensitivity analysis, an exploration of interac-

tion effects, and regressions for alternative dependent variables. Section 6 discusses

the results and concludes.

2. the theoretical link between income diStribution and houSehold

Saving

The link b etween income distribution and aggregate household saving is

ambiguous, as there are various opposing effects on the microeconomic level,

which might be offsetting in the macroeconomic aggregate. First, according to

Keynes (1939), the individual propensity to consume decreases with personal

income, which implies “[...] that the collective propensity for a community as a

whole may depend (inter alia) on the distribution of incomes within it.” Possible

explanations for higher saving rates of richer households are bequests or wealth

that enter the uti lity function as luxu ry goods (e.g. Carroll 1998). Moreover, asset-

based mea ns testing for social sec urity benefits (e.g. Hubbard et al., 1995; Gruber

and Yelowitz, 1999) and a subsistence consu mption level that lies above the

income of poorer households (Musgrove, 1980) can lower the saving rate s of

poorer households. Ray (1998), however, argues that middle-i ncome households

may have even higher saving rates than the r ich, as the midd le class aspires to

build wealth i n order to climb the social ladde r.1

Assuming that the relationship between individual incomes and saving rates is

positive, a rising concentration of income at the top should lead to a rise in the

aggregate saving rate. However, if consumption or saving decisions of different

households are mutually interrelated, the opposite can be true. According to the

relative income hypothesis, “[...] the frequency and strength of impulses to increase

expenditure for one individual depend entirely on the ratio of his expenditures to

the expenditures of those with whom he associates” (Duesenberry, 1949, p. 32).

Building upon such consumption externalities, Frank et al. (2014) propose a for-

mal model of “expenditure cascades.” Similarly, Alvarez-Cuadrado and El-Attar

Vilalta (2012) incorporate relative income considerations into an overlapping gen-

erations (OLG) model. In both models, increasing consumption of a reference

group encourages additional consumption by households further down the income

1Ray’s line of reasoning refers to developing countries in particular. It depends on the assumption

that only the rich engage in conspicuous consumption, following the consumption level of richer people

in developed economies.