Monetary policy and inequality

Prepared by Maarten Dossche, Jiří Slačálek and Guido Wolswijk

Published as part of the ECB Economic Bulletin, Issue 2/2021.

1 Introduction

The issue of economic inequality began to receive increased attention after the global financial crisis. During that period, the increase in unemployment, the heterogeneous evolution of house and stock prices, and the fall in interest rates all affected households in very different ways. Income and wealth inequality has risen in most advanced economies since the early 1980s, with some countries now seeing levels comparable to those recorded at the start of the 20th century, raising increasing concerns regarding the political and economic consequences of that trend.^[1]

More recently, there has also been a greater focus on the interaction between monetary policy and inequality. In response to the global financial crisis, monetary policy embarked on a prolonged period of monetary accommodation, with unconventional measures (such as forward guidance and asset purchases) being used to lower and flatten the yield curve. As such measures tend to have a larger impact on the price of long-term assets than changes in short-term interest rates, this has given rise to concerns that monetary policy is mainly benefiting wealthier households.^[2] In addition, given the growing recognition that the pass-through of monetary policy is dependent on the distribution of income and wealth, central banks have begun to pay more attention to the heterogeneity of households.^[3] These developments have been supported by a wealth of new academic research on the role that household heterogeneity plays in the transmission of macroeconomic shocks and policies.^[4]

This article reviews the latest evidence on economic inequality and its interaction with monetary policy, with a particular focus on the euro area. Section 2 analyses both secular trends and cyclical fluctuations in the distribution of income and wealth in the euro area (a distinction that matters, since the interaction between monetary policy and inequality is arguably fairly cyclical in nature). Section 3 then considers the channels through which monetary policy may affect the distribution of income and wealth, looking at them from both a theoretical and an empirical perspective. Section 4 looks at how heterogeneity in household income and wealth affects the transmission of monetary policy to household spending. And Section 5 summarises the implications of these findings for monetary policy, as well as considering a number of aspects that require further research.

2 Analysis of trends and cycles in inequality

Secular trends in income and wealth inequality

In most advanced economies, income inequality has increased since the 1980s. This trend is illustrated by Chart 1, which shows the pre-tax national income of the top 10% of households as a percentage of the pre-tax national income of the bottom 50% of households. We can see that income inequality is substantially higher in the United States than it is in the European countries in that chart. What is more, the United States also saw the strongest increase over the period 1980-2019. By contrast, the three Nordic countries (Finland, Norway and Sweden) have relatively low levels of income inequality. It is also noticeable that only one country – Spain – saw a decline in income inequality over the period in question.

Income taxation and government transfers have a dampening effect on income inequality, with the precise nature of that impact varying across countries on the basis of their tax systems. If we compare the distribution of disposable income with that of market income for the four largest euro area countries, the United Kingdom and the United States (Chart 2), the dampening effect of direct taxation (e.g. via progressive income taxes) and transfers is evident. The degree of income inequality is indicated by the Gini coefficient, which measures the extent to which the distribution of income among individuals/households deviates from a perfectly equal distribution, with a value of 0 indicating absolute equality and a value of 100 signalling full inequality (whereby the top income group receives all income). Most notably, we can see that the amount of income redistribution carried out by the government is far higher in the European countries (average decline of 20 basis points in the Gini coefficient) than it is in the United States (decline of 14 basis points). However, it is not clear, a priori, to what extent the higher degree of redistribution in Europe can explain the more limited increase in income inequality in the European countries in question (as shown in Chart 1). Two recent studies concluded that the difference between Europe and the United States in terms of the increase in income inequality is driven mainly by market income – i.e. income before redistribution.^[5]

Consumption is substantially less concentrated than net wealth, which may suggest that economic well-being is more evenly distributed than wealth. Consumption inequality is sometimes regarded as a better indicator of the standard of living and welfare than income or wealth-based measures.^[7] As Chart 3 shows, the Gini coefficient of consumption expenditure is typically lower than that of disposable income, reflecting the fact that higher-income households have a higher saving rate. The recent rise in wealth inequality has been shown to be a result not only of higher (gross) saving rates, but also of higher returns for wealthier households, suggesting that greater concentration of capital income may be an important driver of wealth inequality.^[8] The Gini coefficient of net wealth, which is defined as assets (housing, deposits, bonds, equity, etc.) minus liabilities (mortgages, personal loans, etc.), is almost double that of consumption.

The main factors driving increases in inequality include globalisation, technological progress and changes in taxation. Globalisation and skill-biased technological change have adversely affected the wages and employment of lower‑skilled labour and benefited higher-skilled labour and owners of capital.^[9] At the same time, declines in the progressivity of the tax system have contributed to increases in post-tax inequality.^[10] A recent study summarised potential drivers of inequality at the various economic stages – i.e. policies in the pre-production stage focusing on people entering the workforce (e.g. as regards access to education or the job market), production policies affecting workers (e.g. as regards unionisation or the minimum wage), and post-production policies redistributing income.^[11] Its findings suggest that the rise in income and wealth inequality has been driven mainly by structural policies (e.g. policies on education, taxation and market regulation).

Excessive inequality may entail macroeconomic costs and dampen economic growth. For example, to the extent that it reflects inequality of access to education or finance, economic growth may suffer as a result of the economy failing to use its full potential. Moreover, aggregate demand may be depressed as a result of higher‑income households having a lower propensity to consume.^[12] Indeed, empirical studies seem to conclude that very high levels of inequality may well curb economic growth in advanced economies.^[13] However, while redistributive policies (e.g. higher tax rates) may limit the degree of inequality, they may also cause distortions that reduce overall welfare. Thus, the socially optimal degree of economic inequality is not easy to determine, reflecting the complex nature of the interaction between its various drivers.

Monetary policy is not likely to be a substantial driver of increases in inequality, but it should not ignore them either. Advanced economies have tended to adopt fairly similar monetary policy strategies in the period since the 1980s, so it seems unlikely that monetary policy is helping to drive the cross-country variation that has been observed. Nevertheless, monetary policy may still need to take those developments into account, especially if they consistently depress aggregate demand or place downward pressure on the natural rate of interest.^[14] Governments have instruments at their disposal that are better suited to addressing excessive levels of inequality, such as taxation (e.g. progressive taxes on income, wealth and inheritance), transfers, market regulation, and access to education and health services.

Inequality and the business cycle

The cyclicality of income inequality differs across countries, while wealth inequality tends to be procyclical. While lower-income households tend to be more affected by recessions as a result of the greater sensitivity of their labour income (see below), higher-income households tend to be more exposed to the business cycle as a result of capital income (e.g. profits) making up a larger percentage of their total income. Since the distribution of labour and capital income differs across countries, the cyclicality of income inequality can also differ.^[15] Wealth inequality, on the other hand, tends to be mostly procyclical. The procyclicality of profits and equity prices means that equities are not conducive to smooth consumption over the business cycle. With only wealthier households being willing to shoulder such risk (in exchange for a risk premium), that procyclicality helps to explain both the limited levels of participation in the stock market and the substantial equity premium.^[16] Consequently, the cyclical properties of certain asset prices can become a longer-term determinant of wealth inequality.

The cyclical sensitivity of earnings differs across households, which gives macroeconomic stabilisation policies a role to play in averting longer-term increases in inequality. Chart 4 reports “worker betas” for the euro area, which measure the elasticity of labour income in relation to changes in aggregate GDP growth.^[17] As the chart shows, the labour income of workers in lower-income households is more sensitive to changes in GDP growth (see also Box 1, which focuses specifically on the coronavirus (COVID-19) crisis). For those workers, the welfare costs of business cycles are likely to be substantial, especially as they are less able to smooth consumption as a result of their tighter liquidity and credit constraints. Overall, the cyclical behaviour of earnings suggests that macroeconomic stabilisation policies – whether monetary or fiscal – can play an important role in dampening cyclical increases in income inequality. To the extent that cyclical job losses lead to persistent or even permanent scars on people’s labour market opportunities, these policies can also help to mitigate income and wealth inequality in the longer term.^[18]

3 Monetary policy and household inequality

How does monetary policy affect income and wealth inequality?

The economic literature has identified three main channels through which a change in the monetary policy stance can have distributional consequences for household wealth and income. The first is the savings remuneration (and cost of debt) channel, which stems from differences in the size and composition of household balance sheets. Because of those differences, a change in interest rates will have opposing effects on the economic conditions of net borrowers and net savers. The second is the asset price channel, which stems from the heterogeneity of household portfolios and the diverse capital gains (or losses) that are produced when the monetary policy stance changes. In this respect, asset price movements induced by expansionary policies are more likely to benefit the wealthy (and, in some cases, the middle class) to the extent that they hold longer-term assets (as explained in more detail in Box 3). The third channel relates to household income and arises because the elasticity of employment relative to the business cycle is heterogeneous across individuals and dependent on individual characteristics.

Monetary policy and inequality: empirical evidence from the euro area

Net interest income reacts differently across households in response to a reduction in the interest rate. This mainly depends on the extent to which households’ interest-bearing assets have short maturities and their mortgages have adjustable rates.^[23] As the estimates in Chart 5 show, the net interest income of poorer households has not been greatly affected by the fall in interest rates, since they tend to have low levels of both debt and interest-bearing assets. In contrast, the net interest income of middle-class households has increased as interest rates have declined, mainly because they have relatively high levels of mortgage debt. Finally, richer households have recorded a net loss of interest income, since they tend, on average, to be less indebted.^[24] Thus, the direct impact of lower interest rates does not seem to increase income inequality.

An easing of monetary policy affects household labour income via two main channels. The first involves unemployed people finding jobs – and generally experiencing a substantial increase in income as a result (the earnings heterogeneity channel).^[25] The probability of this outcome depends on people’s demographic characteristics (age, level of education, marital status, number of children, etc.). The second involves increases in the wages of employed individuals (the income composition channel).

The easing of monetary policy through the ECB’s asset purchase programme (APP) has substantially reduced the unemployment rate in the lower part of the income distribution.^[26] Panel a of Chart 6 estimates the decline in the unemployment rate which can be attributed to the APP for each of the five quintiles of the income distribution, calculating those declines four quarters after the APP shock (which is represented here by an unanticipated drop in the interest rate term spread – the difference between long and short-term interest rates – of 30 basis points). The aggregate decline in the unemployment rate (which totals around 0.7 percentage points) affects individuals very differently, mainly benefiting households with incomes in the lowest quintile. Their unemployment rate falls by more than 2 percentage points, while the unemployment rates of all other income quintiles fall by less than 0.5 percentage points.

The decline in unemployment is found to be a substantial driver of wage increases across the income distribution, particularly in the lower quintiles. Panel b of Chart 6 breaks the overall increase in mean income down into (i) the part that is due to the decline in unemployment (the earnings heterogeneity channel) and (ii) the part that is due to the increase in wages for all workers (the income composition channel). The earnings heterogeneity channel has a particularly strong impact in the bottom quintile, where wage growth plays a very small role. However, it also accounts for the bulk of the total impact on income in three of the other four quintiles (with the top quintile being the exception).

Overall, the labour market impact of the APP is estimated to result in some reduction in income inequality. Changes in unemployment rates substantially affect household incomes, with incomes increasing considerably on account of households either starting to or continuing to earn wages (instead of receiving unemployment benefits). The mean income of the lowest quintile rises by more than 3%, while those of other quintiles increase by around 0.5%.

The APP also has an impact on the distribution of wealth via the portfolio composition channel. The APP boosts the value of some household assets (stocks, bonds and housing), which affects around two-thirds of all households who hold them. Because stocks are mostly held by wealthier households, an increase in stock prices tends, by itself, to lead to an increase in wealth inequality. However, that increase is offset by the parallel rise in house prices. Housing is fairly evenly distributed across euro area households, with a large percentage (around 60%) owning their main residence.^[27] Moreover, housing accounts for around 70-80% of the total value of all household assets.^[28] When the effects on stock prices and house prices are combined, the Gini coefficient of net wealth – a broad measure of inequality – remains broadly unchanged.

These findings are consistent with the growing body of literature estimating the distributional effects of monetary policy.^[29] Two recent studies, for example, concluded that the overall effect that unconventional monetary policy measures have had on income and wealth inequality has been fairly small.^[30] The Deutsche Bundesbank, meanwhile, has concluded that non-standard monetary policy measures have probably reduced income inequality, while their impact on the distribution of wealth is less clear.^[31] For the Portuguese economy, the Banco de Portugal has found that while increases in house prices tend to reduce wealth inequality, rises in the value of self-employment businesses and marketable financial wealth increase it.^[32] Meanwhile, recent work focusing on cross-racial differentials in the United States has found that while an accommodative monetary policy stance tends to reduce cross-racial differences in unemployment (and thus earnings), it exacerbates wealth differentials.^[33]

4 Heterogeneity and the transmission of monetary policy to household spending

Whereas the previous section focused on the impact that monetary policy has on income and wealth inequality, this section looks at the ways in which the distribution of income and wealth shapes the transmission of monetary policy to households. Specifically, it estimates the manner in which the transmission of monetary policy to consumption varies across individual households on the basis of the structure of their income and wealth, their marginal propensity to consume (MPC) and the ways in which their income responds to aggregate shocks.

Direct and indirect transmission channels for monetary policy

The effects that monetary policy has on individual households can be grouped together in two broad categories: direct and indirect.

Direct effects are the immediate, partial-equilibrium consequences of the change in interest rates. These include the impact that the different interest rate paths have on households’ net financial income (net interest rate exposure). This is often the main channel highlighted by commentators. It is heterogeneous across households, depending on the composition of their asset and liability portfolios. For example, a reduction in policy rates will reduce interest payments for households with outstanding debt (as explained above), especially if their loans have variable rates. It will also reduce the financial income of households that are not indebted, but hold short-maturity assets (whose real returns will temporarily fall). A second direct effect of monetary policy involves changes to households’ saving incentives (intertemporal substitution). This effect is also heterogeneous, since it mostly applies to households that have a stock of liquid savings and are therefore able to temporarily adjust them without paying large transaction costs.

Indirect effects operate through the general equilibrium responses of prices and wages (and thus labour income and employment). When policy rates are reduced, the resulting direct increase in household expenditure (and investment by firms) leads to an increase in output and exerts upward pressure on employment and wages. This indirect effect, which works via the increase in labour income (especially in the lowest quintile, as described in the previous section), leads to additional increases in aggregate demand. It, too, will have heterogeneous consequences to the extent that different sources of earnings (e.g. wages versus income from private businesses) or different pools of unemployed workers (e.g. low versus high‑skilled individuals) display differing degrees of elasticity in their response to the change in aggregate expenditure.

Empirical estimates for the euro area

The economic literature is increasingly finding evidence that individual households behave differently in terms of the ways in which their consumption responds to income shocks – i.e. that households differ in their marginal propensity to consume.^[40] Authors have found it useful to classify households on the basis of the amount of liquid assets that they hold. Households with few liquid assets tend to have large MPCs, and they are often described as “hand-to-mouth”, because they tend to consume all of their income.^[41] They can be either “poor” (if they do not own any assets) or “wealthy” (if they have positive illiquid wealth (e.g. they own their home), but have very limited liquid assets and large spending commitments (e.g. a sizable mortgage)).^[42] In contrast, unconstrained (or “non-hand-to-mouth”) households behave in line with the permanent income hypothesis: their spending patterns remain essentially unchanged in response to a transitory increase in income. On the basis of that classification, 10% of euro area households are “poor hand-to-mouth”, 12% are “wealthy hand-to-mouth” and 78% are “non-hand-to-mouth”.

In order to estimate the size of the various transmission channels to consumption, we need both aggregate and micro-level data on household portfolios and income.^[43] As regards aggregate data, vector autoregressive models can be used to estimate the manner in which aggregate earnings and asset prices respond to monetary policy. In addition, micro-level data are necessary in order to quantify households’ exposure to interest rate risk, inflation risk, asset prices and labour income risk.

As Chart 7 shows, spending by hand-to-mouth households increases more strongly than spending by non-hand-to-mouth households in response to an easing of monetary policy. A 100 basis point cut in real interest rates increases consumption by poor hand-to-mouth households by almost 1% and increases consumption by wealthy hand-to-mouth households by 1.6%, while consumption by other households increases by just 0.5%.

When monetary policy is eased, consumption by hand-to-mouth households is mostly stimulated via indirect effects, while consumption by non‑hand‑to‑mouth households is mostly stimulated via the intertemporal substitution channel. The impact of indirect effects is heavily skewed towards hand-to-mouth households, as they tend to have lower incomes and benefit disproportionately from the new jobs and the corresponding employment income. The effect that this channel has on consumption is amplified because those households have larger MPCs than other households. The intertemporal substitution channel plays a major role for non-hand-to-mouth households that have significant stocks of savings, while the other transmission channels (operating through changes in house prices and, in particular, changes in net financial income) play a smaller role in quantitative terms.^[44]

Accounting for differences across households is important when assessing the impact that monetary policy has on aggregate consumption. In general, monetary policy has only a temporary impact on the economy, with its effects tending to dissipate in the long run. Thus, other factors, such as globalisation or the ways in which individual tax systems redistribute income and wealth (e.g. by means of progressive taxation), will tend, in the long run, to be more important drivers of inequality than monetary policy. However, our results indicate that monetary policy can indeed be effective in supporting household consumption during downturns.

5 Conclusion

In many advanced economies, inequality has been on the rise for several decades, mainly as a result of factors other than monetary policy. While income and wealth inequality have tended to rise in most advanced economies, there is significant cross-country heterogeneity, both in the extent to which inequality has increased and in the current levels of inequality. In the euro area, for example, income and wealth inequality are both generally lower than in the United States. While some drivers of rising inequality are common to most countries (e.g. globalisation), policies other than monetary policy have been key in explaining those cross-country differences. After all, most advanced economies have, since the 1980s, tended to adopt fairly similar monetary policy frameworks, with inflation remaining low and stable. What is more, to the extent that cyclical job losses lead to permanent scars on people’s employability, monetary policy has actually helped to prevent longer-term increases in income and wealth inequality on account of the business cycle.

Indeed, the easing of monetary policy would seem, overall, to have dampened economic inequality in recent years. The direct effects of such easing are heterogeneous as a result of differences in households’ ownership of housing and, accordingly, the prevalence of mortgages. More importantly, the easing of monetary policy clearly has an inequality-reducing impact via its indirect effects, resulting in increased employment for lower-income households in particular. While the ECB has neither the mandate nor the instruments that it would need to specifically target the distribution of income and wealth, its current policies generally seem to have an equalising impact through their contribution to macroeconomic stabilisation.

The heterogeneity of households plays a key role in the transmission of monetary policy. As households differ substantially in terms of the composition of their wealth, the sensitivity of their income to the business cycle, and their propensity to consume, the distribution of income and wealth plays a key role in shaping the transmission of monetary policy to economic activity and inflation. While recent improvements to models and data have contributed to a better understanding of this part of the monetary transmission channel, several puzzles remain (as Box 3 showed). Consequently, the ECB continues, like many other central banks, to invest in new macroeconomic models and data sources in order to improve its understanding of the ways in which its policies affect inequality and the manner in which household heterogeneity shapes the transmission of its policies.