AdobeStock_310462858.jpeg
VoxEU Column Macroeconomic policy Poverty and Income Inequality

Macroeconomic indicators with real incomes: From the poorest to the richest Americans

Economists have long debated the most effective metrics for measuring poverty and inequality. This column presents analysis of the relative importance of three prominent macroeconomic indicators – the rate of unemployment, the inflation rate, and the growth rate of GDP per capita. Using evidence from the US, the author argues that higher unemployment rates unambiguously increase poverty measures, but that inflation matters more in the middle and upper-middle of the distribution than in the tails.

The relative importance of unemployment versus inflation as macroeconomic indicators has long been debated. Arthur Okun’s famous ‘misery index’ (OMI) – developed when Okun was an advisor to the Johnson Administration in the US during the 1960s – weighted the two indicators equally. Some observers put more weight on unemployment, while others emphasise inflation.

At the time of writing (mid-2021), there is a public debate in the US stemming from fears that the policies introduced in the wake of the Covid-19 pandemic are stimulating inflation. Concerns have also been raised about the impacts of macro policy choices on poverty and inequality.

Economic theory has offered some insights. It is widely appreciated that market frictions can entail that wages and other income sources are unlikely to be fully indexed and so short-term real effects of inflationary shocks are likely. There is evidence that higher inflation rates have been caused poverty levels to increase in developing countries, though often with higher inflation rates than in rich countries (such as the US in recent decades). And in rich countries many income sources (such as Social Security in the US) are indexed to inflation. Even so, the concern about inflation remains present, and has a ‘distributional dimension’. Since around 2019, the Federal Reserve has been interested in understanding the distributional consequences of monetary policy and has explicitly included promoting “broad-based and inclusive outcomes” as a policy goal (Board of Governors 2020).

Turning to unemployment, this too can be expected to matter differently at different levels of income. Theoretical support for this view is found in the model-based simulations done by Krusell et al. (2009). These indicate a ‘U-shape’ in the welfare gains from eliminating business cycles, with larger gains for the poorest and the richest strata (though for different reasons). It has been argued that the poorest will incur the largest proportionate impact on their incomes from a rise in the overall unemployment rate, under the (plausible) assumption that they depend most on relatively unskilled and more casual labor – hit hardest by a recession (e.g. Mukoyama and Şahin 2006).

What does the evidence suggest? 

Microeconomic evidence can tell us, for example, whether poorer families have higher rates of unemployment. But macroeconomic evidence can also reveal indirect effects on real incomes. For example, a higher economy-wide unemployment rate may reduce the wage-bargaining power of workers in poor families or reduce (public or private) transfer payments to those families. There is also likely to be heterogeneity within any given income group, such as due to differences in dependence on the labour market, wage setting, discrimination by race or gender, and wealth portfolios. As a result, it is interesting to see if systematic differences in the mean impacts of these macro variables are evident at different levels of income. 

In a new paper, I ask how the relative importance of three prominent macro indicators –the rate of unemployment, the inflation rate, and the growth rate of GDP per capita – depends on whether one is talking about the real incomes of the poor, middle-income groups, or the rich (Ravallion 2021). 

The new paper explores these issues using real income distributions assembled from almost 30 years of survey data since the 1980s, spanning a wide range of settings, including the Great Recession of 2009-10. The literature does not suggest that any single summary measure of ‘inequality’ or ‘poverty’ could capture adequately the nature of the distributional changes induced by these variables. So, I look at data on incomes from the poorest to the richest Americans. In addition to various quantiles of the distribution, the paper draws on the measure of the ‘floor’ – interpretable as the lower bound to the distribution of time-mean income – as proposed in Ravallion (2016) and estimated for the US by Jolliffe et al. (2019). Recognising the existence of transient factors and measurement errors influencing the lowest observed incomes in surveys, the floor is measured by a weighted average of incomes among the poorest 20%, with higher weight on lower incomes. 

The regression specifications aim to isolate the short-term effects of the macro variables at each income level. The regressions for log income are dynamic (including the lagged dependent variable) and they include a time trend. The current-year values of the macro indicators are used (though the paper notes some implications of lagging the unemployment rate).

The results reveal a systematic non-linear pattern in how the key macroeconomic indicators affect real incomes in America. Consider unemployment first. Figure 1(a) gives the estimated semi-elasticities of each income level on the unemployment rate. A significant negative effect is indicated at all income levels, from the poorest to the richest. A higher unemployment rate unambiguously increases poverty measures. This re-affirms earlier results in the literature, including Blank and Blinder (1986). Echoing the theoretical argument of Krusell et al. (2009), I find a striking ‘inverted U-shape’ in the effects of unemployment, as can be seen in Figure 1a. This implies that a higher unemployment rate reduces the skewness of the distribution. This more complex distributional pattern found in the study cannot be seen if one only looks at a summary statistic of overall inequality, such as the popular Gini index. Indeed, the paper finds no significant effect on the Gini index.

The results for the inflation rate are summarised in Figure 1b. Inflation matters more in the middle and upper-middle of the distribution than in the tails. Indeed, for the floor and those living at the 20th percentile, I find no significant effect of inflation, when controlling for the unemployment rate and the GDP growth rate. 

Figure 1 Semi-elasticities of real incomes by percentile and 95% confidence interval

a) Unemployment rate

 

b) Inflation rate

 

Note: The floor is a weighted mean of the incomes of the poorest 20%, with higher weight on lower incomes. (Weights declining with squared income are used here.) The graph gives semi-elasticities (regression of the log income on the unemployment or inflation rates) by the income quantiles, e.g., “p=0.99” refers to the income of the richest 99th percentile from the bottom and “p=0.50” is the median. Robust standard errors used in constructing the confidence intervals.

The study also found that GDP growth rates matter at all levels of income, and especially for the poorest. But this effect is largely attributable to the impact of growth on the unemployment rate. 

The restrictions implied by Okun’s Misery Index – namely equal weights on unemployment and inflation and excluding the GDP growth rate – are rejected statistically across the bulk of the distribution, and strongly so for the poorer strata. Indeed, the index appears to only be defensible for the high-income groups. The unemployment rate should have a higher weight than inflation in a ‘misery index’ calibrated to real incomes across the whole distribution, at least in this time period.

The paper does not go into the political economy implications of these findings, but some points are notable. If it is the middle and upper-middle strata that have the greatest political influence, then macroeconomic policies will tend to attach greater importance to controlling inflation than would be in the interest of either tail of the distribution, especially the poorest. 

This might well change if those in the richest stratum exercise the greatest political power, although that is less clear from this paper’s results. Granted, inflation is not a statistically significant predictor of the real incomes of the richest stratum, but this could well reflect divergent effects within the stratum. It should be noted that the point estimate of the effect of inflation on real incomes of the rich is still high; it is the standard error that blows up at the very top, as can be seen in Figure 1a. 

Nor should the paper’s finding that the unemployment rate is a strong predictor of the real incomes of the rich be taken to imply that the rich would support any policies to reduce the unemployment rate. The statistical significance could well reflect correlated aspects of the business cycle rather than direct impacts of reducing the unemployment rate. This merits further investigation.

References

Blank, R M and A Blinder (1986), “Macroeconomics, Income Distribution and Poverty”, in S H Danziger and D H Weinberg (eds), Fighting Poverty: What Works and What Doesn't, Cambridge, MA: Harvard University Press.

Board of Governors (2020), “Guide to Changes in the 2020 Statement on Longer-Run Goals and Monetary Policy Strategy”, Board of Governors, US Federal Reserve.

Jolliffe, D, J Margitic and M Ravallion (2019), “Food Stamps and America’s Poorest”, NBER Working Paper 2605.

Krusell, P, T Mukoyama, A Şahin, and A A Smith (2009), “Revisiting the Welfare Effects of Eliminating Business Cycles”, Review of Economic Dynamics 12: 393-404.

Mukoyama, T and A Şahin (2006), “Costs of Business Cycles for Unskilled Workers”, Journal of Monetary Economics 53: 2179-2193.

Ravallion, M (2016), “Are the World’s Poorest Being Left Behind?”, Journal of Economic Growth 21(2): 139-164.

Ravallion, M (2021), “Macroeconomic Misery by Levels of Income in America”, NBER Working Paper 29050.

2,835 Reads