Cristiano Cantore, Filippo Ferroni and Miguel León-Ledesma.
How do monetary policy shocks affect the distribution of income between workers and owners of capital? Do workers benefit relatively more when policy changes? Tackling this question empirically requires technical econometric methods, but we are able to show that the share of output allocated to wages (the labor share) temporarily increases following a positive shock to the interest rate. This means that the slice of the pie enjoyed by those whose earnings are mostly made up of wages increases at the expense of profits and capital income. Strikingly, this redistribution channel that shows up in the data runs precisely in the opposite direction to the predictions of standard New Keynesian models commonly used to study the effects of monetary policy.
Despite its importance, there is no systematic empirical evidence on the effect of monetary policy shocks on the labor share. Yet, structural models widely used for monetary policy analysis (Galí (2015)) rely on a transmission mechanism that has clear implications for the evolution of the labor share following a monetary policy shock.
Empirical evidence of the effect of monetary policy on the Labor Share?
In our paper, Cantore et al. (2018), we provide new econometric evidence on the effect of monetary policy surprises on the labor share for five developed countries (Australia, Canada, EU, US, and the UK) from the mid-1980s until the Great Recession. As interest rates can vary for many reasons, identifying an exogenous variation in the rates that is attributable to monetary policy is a complex matter. To this end, we explored various approaches proposed in the literature.
We considered timing restrictions (Cholesky), assuming that macroeconomic aggregates do not respond to the policy variable within the quarter, “reasonable” and quantitatively loose restrictions (sign), implying that a monetary tightening depresses the price level and monetary base growth, and, only for the U.S., instrumental variables restrictions, imposing a correlation between the monetary policy shock and a narrative monetary series counterpart based on FOMC minutes (see Romer and Romer (2004)) or between the shock and the intraday variation of the Federal Fund rate in a narrow window around the monetary policy action (see Gürkaynak et al. (2005) and Miranda-Agrippino and Ricco (2017)).
Figure 1 shows the impulse response functions (IRFs). The labor share temporarily increases after an identified surprise monetary policy contraction in all the countries under study. In our paper we also show that it is robust to using different measures of the labor share, sample periods, different sectors of the economy and identification methods discussed above.
Figure 1: Responses over time of the labor share to a 1% increase in the short-term interest rate (using recursive Cholesky identification). Responses are reported in % deviations
We then dig deeper into the transmission mechanism of monetary policy on the labor share by looking into its different components. The labor share can be decomposed into the ratio between real wages and labor productivity. Therefore, an increase in the labor share following a monetary tightening can occur from a movement in either component. Figure 2 reports the IRFs for each of the two variables for five countries. In each case, the reason why the labor share increases is that real wages fall (or do not increase) but labor productivity falls more. The fall in labor productivity can be rationalized with labor market frictions that make employment decline less relative to output. At the same time nominal wage rigidities could explain why wages are falling (or at least not increasing) less than productivity.
Figure 2: Responses over time of the labor share components to a 1% increase in the short-term interest rate (using recursive Cholesky identification). Responses are reported in % deviations
What would simple New Keynesian models predict?
During the past few decades, New Keynesian models have been the principal theoretical means for economists to explore the transmission mechanism of monetary policy. In these models, monetary policy affects inflation and real economic activity through the effect of interest rate changes on the mark-ups that firms charge over their marginal costs of production. Changes in mark-ups redistribute income between labor and profits. A higher markup would imply a lower labor share.
The essence of that mechanism is as follows: a rise in interest rates will tend to reduce demand and when prices cannot adjust immediately, this implies that prices are too high relative to the new level of demand. They are not at their optimal lower level to equate demand and supply. Since prices are above optimal, firms are charging a higher mark-up. As mark-ups rise, the labor share of income falls and the profit share (mark-ups) increases.
But strikingly, the sign of this effect goes in the opposite direction relative to the empirical evidence shown above. This strong theoretical inverse relationship between the labor share and mark-ups has been used extensively in the literature debating the cyclicality of mark-ups (Nekarda and Ramey (2013), Bils et al. (2018)).
Our results indicate a tightening increases the labor share – precisely the opposite of what the standard models would say.
Our approach differs from studies focusing on the cyclicality of mark-ups because the labor share (unlike mark-ups) is directly observable from national accounts and so does not require theoretical models to derive implied measures. Hence we start from the data on labor share without any theoretical restriction and then compare the data behavior with their theoretical counterparts.
We then analyze whether extensions of the simplest New-Keynesian models can match the stylized fact that the labor share rises following a monetary tightening, we study different families of models commonly used in macroeconomics for the analysis of monetary policy.
Can a more elaborate model get the sign right?
As discussed, in the simplest version of the New-Keynesian model (Galí (2018)), the labor share is equal to the inverse of the price markup and hence pro-cyclical following a monetary policy shock, contrary to our empirical evidence. However, this direct correspondence between the price markup and the labor share does not necessarily hold in other versions of the model. We analyze four ways in which the link between the labor share and the markup can be broken: 1) models that consider fixed costs in production and a broader set of nominal and real frictions (Christiano et al. (2005)); 2) a cost channel of monetary policy which directly links the wage bill to interest rates (Phaneuf et al. (2018)); 3) complementarity between production inputs (Cantore et al. (2014)); and 4) models where wages are not set competitively but as a result of a bargaining process between workers and firms (Christiano et al. (2016)).
Can different families of models that break the relationship between mark-ups and the labor share generate the dynamics of the labor share that we’ve documented?
The size and complexity of these models means that it is not straightforward to answer this question analytically. Instead we explore numerically whether different combinations of behavioral and market structure parameters can generate the correct sign and magnitude for the response of the labor share. We derive measures of the labor share from the models and look at their response to a monetary policy shock. This is carried out using a three-step approach. We first look at the likelihood that these models can generate, at least a priori, the empirical responses from our SVAR using a Prior Sensitivity Analysis (Canova (1995)) approach. Secondly, we single out the key model parameters driving the response of the labor share, real wages, and productivity using Monte Carlo Filtering techniques (Ratto (2008)). Third, we estimate these key parameters by matching the models’ responses to a monetary policy surprise to those obtained in the data (Christiano et al. (2010)). The intuition behind this approach is to check across a wide range of parameters rather than just focusing on a baseline calibration. The IRFs are shown below:
Figure 3: In grey, the empirical response of a variable to a monetary policy surprise (with uncertainty bands). The colored lines are the estimated responses from the different theoretical models. US data 1959Q2-2008Q4
The key result is that standard models generate the “wrong sign” for the effect on labor share when compared to the empirical results. This is not just a feature of the basic New Keynesian model, but carries over in more elaborate versions commonly used for monetary policy analysis. Several of the models do a reasonable job at matching the responses of standard macroeconomic variables to an identified monetary policy shock, but they are unable to reproduce the response of the labor share.
Conclusions
Our results emphasize that macroeconomics needs to develop models that are able to replicate the cyclical behavior of the labor share and its components. So far, neither models with price and/or wage rigidities nor other relevant real frictions are able to match the dynamics observed in the data. Accordingly, using these models to analyze the distributional effects of monetary shocks could be misleading.
Cristiano Cantore works in the Bank’s Research Hub, Filippo Ferroni works for Chicago FED and Miguel León-Ledesma works for University of Kent.
If you want to get in touch, please email us at bankunderground@bankofengland.co.uk or leave a comment below.
Comments will only appear once approved by a moderator, and are only published where a full name is supplied. Bank Underground is a blog for Bank of England staff to share views that challenge – or support – prevailing policy orthodoxies. The views expressed here are those of the authors, and are not necessarily those of the Bank of England, or its policy committees.
