Monetary Policy and Inequality Links: Should Central Banks be Concerned?

Abstract

In this article, the link between monetary policy and inequality is investigated using the vector auto regression based impulse response functions. The study was motivated by persistent high income inequality in South Africa and the COVID-19 pandemic has renewed the debate on the links between monetary policy and inequality. South African annual time series data from 1990 to 2021 was used to examine the relationship between monetary policy and inequality. The impulse response functions show that the overall income distribution is briefly improved following a monetary policy shock. The findings suggest a fluctuating effect of monetary policy on inequality of which after period six a shock to monetary policy has no effect to inequality since the relationship reaches a stable state. The policy implication is the countercyclical use of monetary policy to reduce inequality is only effective in the early phases and in the long term the effect of monetary policy on inequality is reduced. The policy suggestion is that monetary policy alone cannot tackle inequality in an economy with other structural challenges that are outside of the monetary system. Hence policy coordination between monetary policy and fiscal policy is essential.

Introduction

The COVID-19 pandemic and the preceding global financial crisis have raised crucial interrogations on the rising global inequality. COVID-19 has laid bare and exacerbated global inequalities between and within countries with severe economic and social consequences (International Monetary Fund [IMF], 2020). Anecdotal evidence shows that the economic costs of COVID-19 were mainly concentrated on the least fortunate people in society. Distributional issues on income are increasingly becoming topics of discussion in central banks’ (Bank of International Settlement [BIS], 2021a; IMF, 2020). The World Inequality Report (2022) noticed pronounced inequality levels since the past three decades with most income held by private individuals than governments. Over 100 million people were thrown into poverty due to the COVID-19 pandemic with excessive financial gain of $3.7 trillion for the global billionaires who constitute only 0.1% of the world’s population (World Inequality Lab, 2022). During the COVID-19 pandemic, central banks globally played a role in ensuring that the financial system supports households and firms through the crisis and measures are still applied to ensure economic recovery (see IMF, 2020; Marozva & Magwedere, 2021). This was achieved by lowering benchmark interest rates and following unconventional monetary policy rules to maintain favourable financial conditions and support credit flow (IMF, 2020). However, low interest rates and extensive use of balance sheets to support economic activity and lower unemployment has been argued to benefit the rich more than the poor (BIS, 2021a).

The primary mandate of the South African Reserve Bank is to protect the value of the currency in the interest of balanced and sustainable economic growth (South African Reserve Bank [SARB], n.d.). Furthermore financial stability is also the statutory mandate of SARB following an inflation targeting monetary policy rule of 3–6% since the year 2000.

South Africa has been battling structural inequality since post-apartheid and is one of the most unequal society in the world with a Gini coefficient of 0.63 (World Bank, 2020). The legacy of exclusion during Apartheid has contributed to structural inequality with bare mobility of intergenerational income (World Bank, 2020). Since 1994, inequality is South Africa has prominently remained a vital social-economic and policy challenge. Regardless of policy interventions aimed at reducing inequality, high levels of inequality have persisted in South Africa (Leibbrandt & Díaz Pabón, 2021). Such resilience in inequality incited debates for better understanding of the mechanisms that reproduce and create inequalities. The role of central bank in income distribution hasn’t been spared from these debates. Although there are muted policy discussions that monetary policy has a role on income distribution outcomes, monetary policy is not directly used in South Africa as a tool to tackle inequality. Inequality remains a global concern and questions have been raised if central banks have a role in addressing inequality challenges (BIS, 2021a). Inequality is not a monetary phenomenon; its evolution and scale is affected by variations in business cycles and globally it has been on the increase (BIS, 2021a). Furthermore, by fulfilling their mandates, central banks enhance an equitable society over shorter time horizons by ensuring stable prices and sustainable economic activity (BIS, 2021a; Saiki & Frost, 2014). In developed economies quantitative easing has been blamed for deteriorating inequality levels for these economies (BIS, 2021b; Mumtaz & Theophilopoulou 2017; Saiki & Frost 2014).

The International Monetary Fund (2020) opined that income disparities affect the effectiveness of monetary policy in both developed and emerging markets. If the effects of contractionary and expansionary monetary policy net out during a business cycle, inequality is a secondary phenomenon to monetary policy (Kronick & Villarreal, 2019). Rate cuts in an economy with a larger poor population tend to increase consumption expenditure than building wealth and it is argued that lower interest rate increases inequality (Mumtaz & Theophilopoulou, 2017). The pandemic has given rise to research interest on the nexus between monetary policy and inequality (El Herradia & Leroy, 2021). In the endogenous growth model Jin (2009) opined that money growth increases inequality despite frictionless markets. Households that are more vulnerable to inflation peaks and hold cash and cash equivalent assets during the period of low interest rates tend to have lower capital growth compared to their richer counterparts (Greenwood & Jovanovic, 1990). Furthermore, interest rate cuts contribute to capital gains in long-term assets increasing inequality, as assets holders gain more by interest rate cuts as compared to non-asset holders (Saiki & Frost, 2014).

According to the World Inequality Lab (2021) asset allocation in South Africa continues to shape income disparities. Piketty (2018) argued that the return on capital which is greater than economic growth, is the main driver of inequality and discontent towards democratic values. Persistent income disparity among individuals is linked to socio-economic costs which include poverty, low levels of education and employment (Stiglitz, 2012). Furthermore it is argued that income inequality hinders economic growth and it can be a source of instability within societies (Taghizadeh-Hesary et al., 2020). Inequality within a country responds to differences in skill and technological advances (Dabla-Norris et al., 2015). Heterogeneity in household income affects the impact of monetary policy on income distribution (Coibion et al., 2017). The transmission mechanism of monetary policy to inflation is influenced by the level of inequality within the economy (Colciago et al., 2019). Income redistribution in emerging and developing markets is essential not only for equality but for faster economic growth and poverty reduction (IMF, 2018).

However, it remains unclear among researchers whether monetary policy decisions by central bank should include income distribution or they also play a role in addressing or amplifying inequality. Bernanke (2015) opined that direct distributional policies in an economy should be done through the fiscal policy than monetary policy. Literature is inconclusive on the links of monetary policy on inequality as there are different channels of transmission. Debate on the role of monetary policy in distribution of income has become a high profile topic in recent central bank speeches (see BIS, 2021a; 2021b). This study contributes to the debate on the nexus between monetary policy and inequality. The results of the study provide an understanding to policymakers on the ways in which its policies affect inequality.

The article is organized as follows: the second section discusses theoretical and empirical literature of the study. The third section discusses the methodology, while the fourth section presents the results and a discussion on the results. The study is concluded in the fifth section.

Literature Review

Several channels on the effects of monetary policy on equality which includes inflation, employment, assets prices and borrowing costs channels are suggested in literature (see Coibion et al., 2017). The employment channel suggested that looser monetary policy increases employment (Coibion et al., 2017; Draghi, 2016). The asset price channel argues that looser monetary policy benefits richer households more as they hold riskier and more cyclical assets than the lower income households (see Bernanke, 2015). Lower interest rate boosts the prices of assets such that it benefits more for the household who own assets already. In the income composition channel if changes in monetary policy increase business profit more than wages, income is accumulated by the already richer households (Coibion et al., 2017). The heterogeneity in sources of households’ income (business profits, financial income and labour income) is affected differently by the monetary policy (Saiki & Frost, 2014).

Furceri et al., (2018) documented that monetary policy affect income inequality but the magnitude of the effect depends on the redistributive policies and the share of labour income. Mumtaz and Theophilopoulou (2020) opined that monetary policy effect on inequality is greater for countries with limited redistributive policies and large labour share of income. Tightening monetary policy increases income and consumption inequality (Coibion et al., 2017; Mumtaz & Theophilopoulou, 2017; 2020). El Herradia and Leroy (2021) found that tightening monetary policy reduces the income of the top 1%, however, the findings were not the same for the overall income distribution as tightening monetary policy did not reduce the overall inequality. Easing monetary policy indirectly reduces inequality via its effects on reducing unemployment of low income households (Mumtaz & Theophilopoulou, 2017). Duarte and Schnabl (2019) found that accommodative monetary policy reduces inequality through the redistribution of income among the different income groups in the society. Upsurge in asset prices as a result of the unconventional monetary policy has a potential of widening inequality according to the findings of Domanski et al. (2016).

Using administrative data in Sweden, Amberg et al. (2021) reported that low interest rates benefited the richest and the poorest more, with little or no effect on the middle income. However, the finding of Andersen et al. (2021) using Danish data found that low interest rates increases inequality, as the initial income of the individual plays a role on the overall effect of low interest rate. There is no consensus in empirical literature on the effect of low interest rate to inequality. Low interest rate in a bullish stock market increases the asset prices creating more income and wealth for asset holders (Vereckey, 2021). Alternatively in low interest rate periods with larger unbanked low income earners, accommodative monetary policy creates more inequality. Heterogeneity in households’ occupations and balance sheets implies that they are affected by the monetary policy shocks differently (Coibion et al., 2017). Bernanke (2015) suggested that the degree of inequality is a long run phenomenon resulting from structural changes including globalization, technological changes, demographic trends and changes in the labour markets. Accelerated returns for higher and new skills more than the less skilled workers worsen inequality (Bernanke, 2015). O’Farrell et al. (2016) found a limited effect of monetary policy on inequality while also suggesting that high inequality has an effect on the effectiveness of monetary policy. Economic growth is argued to be the major driver or reducing inequality within and between countries (IMF, 2018). Higher inflation and interest rate cuts are thought to be beneficial to poor households as it create more jobs and reduce the cost of debt for low-income earners (Bhorat & Oosthuizen, 2005; Romer & Romer, 2000). However, heterogeneity of income, debt composition and sources among household makes the interaction of monetary policy and inequality complex (BIS, 2021a).

Data and Methodology

This section presents and discusses the data and the research design employed in this study. The data, description of variables, population and sample is done in the following section. The research design/approach is presented and discussed in the next section 3.2.

Data and Variable Description

South Africa’s time series data from 1990 to 2021 was used to examine the relationship between monetary policy and inequality. The study utilized annual time series data on Gini coefficient, repo rate, inflation and unemployment. The description, definition of variables and the data sources is summarized in Table 1. The choice of the variables used in the study was mainly guided by data availability for the time series.

Table 1.

Variable Description.

Variable	Definition of Variables	Data Source	Expected Sign
GINI coefficient (INEQ)	Uneven distribution of income across the entire households. The measure of inequality derived from the Lorenz curve.	SWIID, 2022
Repo Rate (INT)	The rate at which the central bank of a country lends money to private banks	SARB, 2022	Positive/Negative (+/−)
Unemployment (UNEMP)	Total number of unemployed labour force as a percentage of the total labour force.	World Economic Outlook (IMF), 2022	Positive (+)
Inflation (INF)	Changes in prices of goods and services that households consumes	World Economic Outlook, 2022	Positive (+/−)

Source: The authors.

Model Specification and Estimation Techniques

To examine the links between monetary policy and inequality, the Gini coefficient was incorporated in a standard macroeconomic VAR-framework consisting of, monetary policy variable, inflation and unemployment. The estimated model for the study is as follows

$$\text{INEQ}=f(\text{INT},\text{INF},\text{UNEMP})$$

(1)

The variables of interest in Equation (1), include interest rate (INT), inflation (INF) and unemployment (UNEMP).

The repo rate is the main variable of interest as it is the tool that is used for monetary policy in South Africa. The finding of the effect of the monetary policy in South Africa is inconclusive, as some studies find that the effect of monetary policy on inequality is subject to the existing redistributive policies (Furceri et al., 2018; Mumtaz & Theophilopoulou, 2020). Others find that monetary policy has no effect on inequality (El Herradia & Leroy, 2021), while others have found a negative relationship (Duarte & Schnabl, 2019).

The second important traditional determinant in empirical studies is inflation and the findings have also been inconclusive. South Africa uses an inflation targeting monetary policy regime; hence inflation variable was considered for the determinant on inequality. There are mixed empirical studies on the effect of inflation on income inequality. The inflation targeting policy aims at managing inflation between 3–6% with intention of keeping inflation low to ensure economic stability. Empirical studies are inconclusive on the links between inequality and inflation (Albanesi, 2007; Bhorat & Oosthuizen, 2005; Mumtaz & Theophilopoulou, 2020).

Unemployment is another driver of inequality considered in this study. The effect of unemployment on inequality is also inconclusive (Bhorat & Oosthuizen, 2005; Mumtaz & Theophilopoulou, 2020; Romer & Romer, 2000). Unemployment in South Africa is persistently high and the majority of South African households are excluded from the labour market income (BIS, 2021b).

Estimation Techniques

The links between monetary policy and inequality in this study is explained using the vector autoregressive framework (see Sims, 1980). The standard Sims (1980) vector auto regression is an unrestricted reduced-form approach in which a common lag length for each variable in each equation is applied. The baseline model assumes that inequality as measured by the Gini coefficient reacts to changes in monetary policy as measured by repo rate, unemployment and inflation.

The dynamic links between the focus variables are examined using the vector auto regressive based impulse response function. Following Sims (1976) and Saiki and Frost (2014) restrictions are imposed on coefficient matrices to be null, and the same lag length. Selecting the lag length that fits the data structure in a VAR process is very essential for the power of the test (Gujarati & Porter, 2012; Liew, 2004). The optimal lag length for the VAR process of the variables used in the study was determined using the Akaike information criteria (AIC) (see Akaike, 1973). Liew (2004) argued that for smaller sample (60 observations and less) the AIC is most accurate in estimating the optimal lag length. The AIC minimize the chance of under estimation while maximizing the chance of recovering the true lag length. The data structure of the study is time series hence the unit root was employed to determine the order of integration of the variables. Following Choi and Chung (1995) the data frequency used for this study is low (annual span) hence the stationarity properties were tested using Phillips and Perron (1988). The Philips and Perron unit root test is more appropriate unit root test for low frequency data (Choi & Chung, 1995; Gries et al., 2009).

The impulse response function was employed to determine the inequality response to ‘shocks’ in the monetary policy, inflation and unemployment within the VAR system. In other words, this approach is designed to determine how each variable responds over time to an earlier ‘shock’ in that variable and to ‘shocks’ in other variables. A weakness of the IRF is that it depends on the ordering of variables in a VAR system. Inorder to control for the sensitivity to the ordering the generalised IRF which is not contingent to ordering was employed in consistence with Pesaran and Shin (1998) and Saiki and Frost (2014). The parameters in Equation (1) were consistently estimated using the VAR framework, and turned to the generalized impulse response functions with respect to one standard error shock to inequality. This section specifies the models of the VAR framework. Furthermore the diagnostic test of autocorrelation, heteroscedasticity, normality and stability test were used to determine the accuracy of fitting the VAR model to the data (Gujarati & Porter, 2012).

The VAR-based impulse response functions were estimated representation of the empirical model. Equation (1) can be expressed as follows:

$${\text{INEQ}}_t={\alpha }_1+{\displaystyle {\sum }_{j=1}^k{\beta }_j{\text{INEQ}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\gamma }_j{\text{INT}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\psi }_j{\text{INF}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\varphi }_j{\text{UNEMP}}_{t-j}}+{\mu }_{1t}$$

(2)

$${\text{INT}}_t={\alpha }_2+{\displaystyle {\sum }_{j=1}^k{\beta }_j{\text{INT}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\gamma }_j{\text{INEQ}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\varphi }_j{\text{INF}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\psi }_j{\text{UNEMP}}_{t-j}}+{\mu }_{2t}$$

(3)

$${\text{INF}}_t={\alpha }_3+{\displaystyle {\sum }_{j=1}^k{\beta }_j{\text{INF}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\gamma }_j{\text{INEQ}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\psi }_j{\text{INT}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\varphi }_j{\text{UNEMP}}_{t-j}}+{\mu }_{3t}$$

(4)

$${\text{UNEMP}}_t={\alpha }_4+{\displaystyle {\sum }_{j=1}^k{\beta }_j{\text{INF}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\gamma }_j{\text{INEQ}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\varphi }_j{\text{INT}}_{t-j}}+{\displaystyle {\sum }_{j=1}^k{\psi }_j{\text{UNEMP}}_{t-j}}+{\mu }_{3t}$$

(5)

Where INEQ is the Gini coefficient

INT is the repo rate

μ is the stochastic error terms which in the VAR system are impulses or shocks

k is the lag length

An impulse response function measures the time profile of the effect of shocks at a given point in time on the expected future value of the dynamic system (Pesaran & Shin, 1998). The following section presents the results of the study and a discussion.

Results and Discussion

The results of the unit root test are summarized in Table 2. The Phillips and Perron (PP) as stated earlier is the appropriate method for the low data frequency used for the study. The Dicky–Fuller is reported for robustness of the test results. All the variables are first difference stationary I(1).

Table 2.

Results of the Unit Root Test.

Variable	Stationarity at Levels			Stationarity at First Difference
	Intercept	Intercept and Trend	None	Intercept	Intercept and Trend	None
Panel A: Augmented Dickey-Fuller
INEQ	−1.721347	−1.732022	0.346621	−5.584433***	−5.628496***	−5.625475***
INT	−1.409496	−1.600844	−1.191683	−6.764273***	−6.684617***	−6.691255***
INF	−1.536222	−1.832519	−1.338316	−5.287584***	−5.612515***	−4.970213***
UNEMP	−1.552924	−1.146014	0.612071	−2.793349**	−2.674036	−2.595478**
Panel B: Phillips and Perron (PP)
INEQ	−1.721347	−1.770406	0.364923	−5.584433***	−5.629959***	−5.625475***
INT	−1.079087	−1.263726	−1.572980	−10.56893***	−11.70873***	−7.663106***
INF	−2.849965	−2.507838	−1.054992	−5.611142***	−8.796913***	−5.256127***
UNEMP	−1.954659	−3.073907	2.146229	−10.51217***	−10.26999***	−6.595935***

Source: The authors.

Note: *** p < 0.01, ** p < 0.05 and * p < 0.1.

The Breusch–Pagan LM test and White’s test for heteroscedasticity confirm the rejection of null hypothesis; the model is homoscedastic. Also, the results from unit root test confirmed that all variables included in the examination of the linkage between these variables were integrated of order one. The next step was to test for the existence of a cointegration relationship between inequality (INEQ), unemployment (UNEMP), interest rates (INT) and inflation rate (INF). For this purpose, the study uses the Johansen–Juselius (maximum likelihood) cointegration test procedure. The results of the Johansen–Juselius cointegration tests are presented in Table 3.

Table 3.

Maximum Likelihood Cointegration Test.

Trace Test				Maximum Eigen Value Test
Null	Alternative	Trace Statistic	Critical Value @5%	Null	Alternative	Max-EigenStatistic	Critical Value @5%
Cointegration between INEQ INF UNEMP and INT
r = 0	r ≥ 1	55.5207	47.21	r = 0	r = 1	28.2497	27.07
r ≤ 1	r ≥ 2	27.2710	29.68	r ≤ 1	r = 2	17.7426	20.97
r ≤ 2	r ≥ 3	9.5285	15.41	r ≤ 2	r = 3	8.3800	14.07
r ≤ 3	r ≥ 4	1.1484	3.76	r ≤ 3	r = 4	1.1484	3.76

Source: The authors.

Note: Where r stands for the number of cointegrating vectors. The lag structure of VAR is determined by the highest values of the Akaike.

The results of Johansen–Juselius cointegration tests reported in Table 3 reveal the existence of cointegrating relationship between the variables. This implies a stable long-run relationship between inequality (INEQ), unemployment (UNEMP), interest rates (INT) and inflation rate (INF). Both the trace test and the maximum Eigen value statistics reject the null hypothesis of no cointegration. Specifically, the results indicate that there is a unique cointegrating vector between inequality, unemployment, interest rates and inflation rate. Before the impulse response function tests, the nexus was subjected to variance decomposition test. Nevertheless, the results presented in Table 4–Table 7 on variance decomposition were not discussed for brevity.

Table 4.

Variance Decomposition of INEQ.

Period	S.E.	INEQ	INF	UNEMP	INT
1	0.007338	100.0000	0.000000	0.000000	0.000000
2	0.009058	97.02336	0.007304	2.333894	0.635438
3	0.009686	94.67766	2.717399	2.044193	0.560750
4	0.010237	89.03619	8.545886	1.906089	0.511832
5	0.010460	86.62569	10.53057	2.262184	0.581553
6	0.010662	83.90031	10.54123	4.678483	0.879984
7	0.010830	81.38489	10.34916	7.098717	1.167227
8	0.010926	79.98390	10.43004	8.201748	1.384317
9	0.010987	79.19464	10.62384	8.653260	1.528263
10	0.011035	78.67815	10.65480	9.015024	1.652024

Source: The authors.

Table 5.

Variance Decomposition of INF.

Period	S.E.	INEQ	INF	UNEMP	INT
1	1.671962	1.590189	98.40981	0.000000	0.000000
2	1.945150	9.113552	87.94138	2.513092	0.431975
3	2.337871	7.977418	61.37532	28.46402	2.183243
4	2.688935	6.764018	47.87829	43.40191	1.955788
5	2.796921	7.818011	44.26368	45.98084	1.937465
6	2.838636	8.909266	43.75218	45.45759	1.880960
7	2.866853	9.703563	43.19220	45.26012	1.844116
8	2.898022	9.980688	42.28157	45.92157	1.816174
9	2.927454	10.06762	41.54402	46.60227	1.786099
10	2.940990	10.20067	41.17642	46.85319	1.769712

Source: The authors.

Table 6.

Variance Decomposition of UNEMP.

Period	S.E.	INEQ	INF	UNEMP	INT
1	2.023854	8.165052	0.044117	91.79083	0.000000
2	2.559213	17.66278	0.901825	81.41980	0.015599
3	2.759676	19.97840	0.870826	79.13674	0.014032
4	2.882989	21.27029	0.882205	77.57562	0.271889
5	2.979907	21.58500	0.850082	77.26000	0.304922
6	3.055604	21.25802	0.870176	77.49242	0.379385
7	3.105012	21.05982	0.867122	77.61464	0.458426
8	3.134809	20.92647	0.852042	77.67166	0.549831
9	3.156396	20.79203	0.847849	77.72480	0.635317
10	3.174917	20.62801	0.840013	77.83862	0.693355

Source: The authors.

Table 7.

Variance Decomposition of INT.

Period	S.E.	INEQ	INF	UNEMP	INT
1	1.432089	14.63218	21.11951	2.120505	62.12780
2	1.663518	25.26981	15.88382	9.297754	49.54861
3	2.007911	17.37181	12.45562	32.54491	37.62766
4	2.283459	13.59801	11.65688	43.52828	31.21682
5	2.398905	12.33815	10.56300	47.26020	29.83865
6	2.476779	11.58434	10.48135	48.29854	29.63577
7	2.552067	10.91148	10.17214	50.17067	28.74571
8	2.642179	10.18585	9.536402	53.01565	27.26210
9	2.726649	9.564971	8.978650	55.60727	25.84911
10	2.791652	9.128647	8.672667	57.36553	24.83316

Source: The authors.

Variance Decomposition Results

Impulse Response Function Result

The monetary policy shocks denote an unanticipated or non-systematic change in nominal interest rates. The results from the impulse response function are presented in Figures 1 and 2. From the IRF, the study chose three innovations that influence the income disparity: (a) monetary shock, (b) inflation (c) unemployment.

Figure 1.

Source: The authors.

Figure 2.

Source: The authors.

To trace the time path of the response of inequality to shock in policy rate, inflation and unemployment, the VAR-based impulse response function were used. Initially a shock in the policy rate reduces inequality for two periods, there after any shock to interest rate increases inequality until period three. This period in South Africa coincided with the transitioning period from the apartheid to a democratic South Africa. It is not clear whether the temporary reduction in inequality was due to a monetary shock or the political dimensions influenced this relationship. A standard deviation shock to repo rate has a fluctuating negative response to income distribution before period four. The impulse response function shows a sharp decline in inequality before period two increasing between periods two and three and then a sharp decline until period four. Between periods four and seven, there is a sharp increase in inequality due to a monetary policy shock until period seven. Thereafter inequality is in a stable state to a shock in monetary policy. The results of an increase in inequality due to a monetary policy shock are consistent with Kronick and Villarreal (2019) findings for Canada. After period six there is not so much response of inequality to a shock in monetary policy. Furthermore before period six, the monetary shock effect to inequality is negative. It is opined that lower interest rates in South Africa tend to worsen income distribution ceteris paribus since inequality in South Africa is structural and persistent (BIS, 2021b). It is argued that the majority of South African households have some form of high-cost debt from micro lenders which is mostly non inelastic to changes in the repo rate (BIS, 2021a). Thus influence of a change in monetary policy is stronger in those households with formal debt than the majority of the population that have high-cost debt from micro lenders.

Shock to monetary policy will have a negative effect to inequality both in the short run and in the long run. The study shows that in the long run it however reaches a stable state. Low income household face uninsurable risks more than their high income counterparts. Any shock to interest rate have indirect effects to macroeconomic aggregates (Kaplan & Violante, 2018; Kronick & Villarreal, 2019). The results of this study are in line with the empirical literature despite the study including the COVID-19 period. The model was tested for stability and confirmed that the model was stable. It is argued that the monetary and the fiscal policy interventions during the COVID-19 period brought stability in the financial markets. Queyranne et al. (2021) stated that the monetary and fiscal policy interventions muted the transmission channel of monetary policy in Morocco (Queyranne et al., 2021). Furthermore, Magwedere and Marozva (2021) found that in the long run COVID-19 did not significantly affect the liquidity available in the financial markets.

Furceri et al. (2021) opined that inequality respond to pandemics with a lag since previously large pandemics were later followed by income inequality. The study further suggested that in the midst of pandemic, inequality is only worsened in the absence of strong policy actions. Anecdotal evidence shows that during the COVID-19 pandemic, the policy interventions by government and central banks abated the severe socio-economic effects of the pandemic (IMF, 2020; Sandbu, 2020; Stiglitz, 2020). Hence the unconventional monetary policy tools were deployed during the COVID-19 period in South Africa like in many other countries; which included interest rate cuts and asset purchases by the central bank stabilized the financial and limited the severe socio-economic impacts of the pandemic.

The inflation channel argues that higher inflation is beneficial to the poor, as increase in inflation creates more jobs for the lower income groups (Mumtaz & Theophilopoulou, 2020; Romer & Romer, 2000). Figure 1 shows that a shock to inflation initially increases inequality for the first period. Thereafter inequality decreases with a shock on inflation until period four. After the period, a shock in inflation has a persistent increase in inequality. The findings of a persistent increase in inequality due to a shock to inflation are consistent with Albanesi (2007) findings. Low-income earners are more vulnerable to inflation shocks as they hold more cash and have very low bargaining power in the event of inflation shock (Albanesi, 2007; BIS, 2021b). The results show a decline in inequality after an inflation shock until period four; the level of inequality begins to rise although it is still negative. Lower-income earners are hard hit by inflation as they lack the means to protect their wages and savings. The result of increasing inequality due to a shock in inflation is consistent with the empirical finding of Kronick and Villarreal (2019). From period four, a shock to inflation increases inequality. This coincides with literature where an increase in the level of inflation can worsen income distribution (BIS, 2021b; IMF, 2018). Shocks to inflation have a positive effect to inequality in the short run. In the long run, the response of inequality to inflation shocks is negative. The results are consistent with the finding of Easterly and Fischer (2001) who concluded that inflation hurts the low income households more.

With an unemployment rate of 34.9%, more than a third of South Africans are not connected to the labour market (Statistics South Africa, 2021). A shock to unemployment tends to increase inequality until period two, declining gradually to reach a stable state in period three. Between periods three and four, there is a negative response of inequality to a shock on unemployment. After period four, there is a sharp increase in inequality due to a standard deviation shock to unemployment. In the long run, there is a positive response to inequality due to a shock in unemployment, thus inequality increases. This is inconsistent with the results in the Euro area, where a shock to monetary policy reduces inequality by creating jobs for low-income households (Vereckey, 2021). The increasing inequality to a shock to unemployment is consistent with the World Bank (2020) assertion that the quality of jobs created in South Africa is not sufficient to reduce inequality. This was further supported by Leibbrandt and Díaz Pabón (2021) findings that labour market in South Arica is the largest contributor of inequality. Figure 2 presents the results the other variables in the system to a shock in inequality.

Figure 2 summarizes the response of monetary policy, inflation and unemployment to a shock in inequality. A shock in inequality increases interest rate initially until period two. Between periods two and three, any shock to interest rate result in a sharp decline in interest rate until period three with a slight increase in period four. After period five, interest rate reaches a stable state with no response to any shock to inequality. The result shows that in the long run, inequality has no effect to monetary policy in South Africa. This is inconsistent with Kronick and Villarreal (2019) who found that inequality had an effect on the transmission mechanism of monetary policy. The Bank of International Settlement (2021a) reiterated that monetary policy alone has no effect on inequality, as other policy initiatives such as fiscal policy have more effect to income distribution. Response of inflation to an inequality shock is an increase in the short term until the second period, with cyclical movements between periods four and six. In the long run after period six, inflation gradually decreases. The response of inflation to inequality in both short term and long run is positive.

Although unemployment response to inequality is negative, the results suggest an increase in the long run after period two. Historically, even under economic boom the unemployment rate in South Africa hasn’t fallen below the 20% level making it appear as if it is inelastic to the shocks in the monetary policy system. The heterogeneity of income sources from labour is a complex and structural phenomenon in the context of South Africa (Leibbrandt & Díaz Pabón, 2021). Keeping interest rates low to favour employment has ripple effects to price stability in South Africa owing to dysfunctional labour markets. Rigidity in the labour markets is further isolated by Saiki and Frost (2014) as contributors of increasing inequality in Japan. The increasing inequality due to a shock to unemployment is consistent with Coibion et al. (2017) who found that inequality in the US widened over the 1980–2018 period largely because of the monetary policy effect in the labour market. Empirical differences of the effect of monetary policy in this study with previous studies can be largely explained by differences in the country context and methodological differences.

Conclusion

This article contributes to the monetary policy inequality debate by examining the links between income inequality and monetary policy in South Africa. The findings suggest a non-responsive inequality to monetary policy shocks in the long run, as inequality reaches a stable state after period six. This supports the composition of debt in South Africa where the majority of the households’ debts are from micro lenders, where the interest rate in the micro-lending sector is not tied to the monetary policy rate. In the short run, the results show that there is no clear consensus on the effect of monetary policy on inequality. Thus inequality in South Africa is propagated by other factors that are outside the realm of monetary policy. Additionally inequality and unemployment in South Africa have structural constraints outside the influence of monetary policy. This remains a challenging initiative for monetary policy alone to be used as a tool to overcome these structural phenomena.

Although inequality is not a monetary phenomenon, there are active discussions for the central banks to consider inequality when changing the repo rate (policy rate). The COVID-19 pandemic has ushered in policy disruptions, where active monetary and fiscal policy coordination during the pandemic appears to have stabilized economies from spiralling into deeper recessions. The study provides a message to policy making that, monetary policy can have short-term benefits in reducing inequality which can even vanish in the long run. It is recommended that in making monetary policy decisions, the central bank should consider to integrate monetary policy to fiscal policy strategies. Thus the central bank should choose a set of policy tools that safeguard their mandate while abating the potential distributional effects of monetary policy. Further research on fiscal and monetary policy coordination to tackle inequality is recommended. For further studies, it is recommended to contact an event study and identify the structural changes in the economy that affects the transmission mechanism of monetary policy.