Workers’ bargaining power, business cycle fluctuations, and the Phillips curve
Understanding the Phillips Curve through the lens of workers’ bargaining power and business cycle fluctuations
Three stylised structural changes have been underway since the end of 1980s. First, the Phillips curve flattened. Figure 1 reports rolling estimates of the slope of the Phillips curve based on a panel of six advanced economies. The large swings in output and unemployment since the Global Crisis put the debate on the ‘elusive’ Phillips curve under the spotlight, as a key ingredient to the ‘missing deflation’ and ‘missing inflation’ puzzles (Hall 2013, Coibion and Gorodnichenko 2013, Constancio 2015, Williams 2010 among many others). Yet the flattening was already well underway when the Global Crisis hit – it is widely acknowledged that it dates back to the end of the 1980s (Blanchard 2016, Del Negro et al. 2020).
Figure 1 Slope of the Phillips curve
Notes: The y-axis displays rolling estimates (based on 20-years windows) of the slope β of the Phillips curve πit=πit*+βyit+uit, based on a panel of six advanced economies (France, Germany, Italy, Japan, United Kingdom and United States); πit* is a measure of trend inflation, computed as the four-quarter moving average of past core inflation, and yit is the output gap.
Second, workers’ bargaining power progressively declined. Quoting Alan Kruger’s 2018 lecture at the Jackson Hole Symposium, “over the last decades there has been a proliferation of practices that enhance monopsony power in labour markets and weaken workers bargaining power. Their main effects have been to shrink the slice of the pie going to workers and increase the one going to employers”. The decline in workers’ bargaining power is well visible in different indicators. In general, all the forces that traditionally counterbalanced firms’ monopsony power and boosted workers’ bargaining power have receded in recent decades. Employment protection laws have become looser, minimum wage has decreased, trade union density and collective bargaining coverage has fallen, globalisation has made workers more vulnerable to threats of job loss due to delocalisation (Stansbury and Summers 2020). To illustrate the extent and the dynamics of this phenomenon, Figure 2 reports a synthetic index of workers’ bargaining power for the same set of countries considered in Figure 1, built by taking the first principal component of several of these indicators. The weakening of workers’ bargaining power since the end of 1980s was substantial and broad-based.
Figure 2 A synthetic indicator of workers’ bargaining power
Notes: Each line is a synthetic measure of workers’ bargaining power obtained by taking the first principal component of several indicators from the OECD dataset, including union density and coverage, employment protection legislation and the coverage of collective bargaining agreements.
Third, the contribution of the number of workers (extensive margin) relative to hours worked per employee (intensive margin) in the cyclical adjustment of the labour input increased. Figure 3 illustrates this fact by comparing three samples for the whole euro area – 1970-1990, 1991-2008 and 1991-2016. The increase in the contribution of the extensive margin to variation in total hours worked would be striking if one compared the first and second sample only – the share of fluctuations in total labour input due to the number of workers went from an average 56% between 1970 and 1990 to 90% between 1991 and 2008. Using the third sample (i.e. including the years after 2008) downsizes this increase, mostly due to Germany, where short-time work was used to an exceptional extent in order to secure jobs in the aftermath of the Global Crisis. Still, even including these years, the share of fluctuations in the total labour input due to the extensive margin has strongly risen.
Figure 3 Relative contribution of intensive and extensive margin to the changes in total hours worked for the euro area
Notes: Denoting with TH the cyclical component of the Hodrick–Prescott filter of the log of total hours, with N the one of the log of the number of workers and with H the one of the log of hours per employee, the contributions of employment and hours per worker to fluctuations in total hours worked are given by ((cov(TH,N))/(var(TH))) and ((cov(TH,H))/(var(TH))), respectively. Data are from the updated Ohanian and Raffo (2012) dataset.
In our recent research paper (Lombardi et al. 2020), we show that these three stylised facts are tied.
Unveiling their ties
To understand the structural ties between these three facts, consider an economy where the labour market is made of workers transitioning between employment and unemployment and firms reacting to changes in demand by varying hours worked per employee and/or the number of workers, for which posting a new vacancy is costly. In this context, both firms and workers are willing to continue their relationship as long as wages are within a given band. The bargaining powers of firms and workers determine how the surplus generated by an employment relationship is split between them. When firms want to increase their output following a demand shock, their marginal benefit of hiring a new employee (instead of using overtime) is larger the smaller the share of the surplus going to workers (i.e. the lower is workers’ bargaining power). Hence, when firms can secure a larger share of the surplus, they would react to changes in demand by adjusting more the extensive (the number of workers) rather than the intensive margin (hours per worker – see Figure 4 and Lombardi et al. 2020, for details about the model and the simulations). National accounts data for Germany, France and Italy paint a picture that is consistent with this story. Figure 5 points to a significant negative relationship between the ratio of changes in employment over changes in the hours worked and the index of bargaining power presented in Figure 2.
Figure 4 Employment-hours per worker multiplier
Notes: See Lombardi et al. (2020). The x-axis displays the time horizon, while the y-axis reports the employment-hours per worker multiplier, defined as the ratio between the cumulated impulse response functions of employment and hours per worker, obtained in the theoretical model by simulating a demand shock. Workers’ bargaining power is defined in a range [0,1].
Figure 5 Correlation between bargaining power and the adjustment of the extensive margin relative to the intensive one
Notes: The figure displays the ratio between the percentage change in employment relative to changes in the intensive margin (net of changes in value added) and the measure of bargaining power reported in Figure 2. We use national account data for Germany, France and Italy (private sector only), over the sample 1996-2018
Why does the relative adjustment of the intensive and extensive margins affect cyclical fluctuations of inflation? The reason is that the marginal wage, i.e. the variation in wage per employee due to a marginal change in hours worked, is typically increasing with hours per employee.1 Therefore, following a given variation in the output gap, the smaller is the change in hours per worker relative to the change in employment, the smaller the change in unit labour costs and hence that of inflation (Figure 6).
Figure 6 Inflation-output gap multiplier
Notes. See Lombardi et al. (2020). The x-axis displays the time horizon, while the y-axis reports the inflation-output gap multiplier, defined as the ratio between the cumulated impulse response functions of inflation and output gap, obtained in the theoretical model by simulating a demand shock. Workers’ bargaining power is defined in a range [0,1].
Micro and macro evidence support this mechanism
To test whether this story finds support in the data, we look at both micro- and macro-evidence. We first use firm-level data drawn from a yearly survey on Italian firms (INVIND).2 Participating firms are asked to report, among other things, the share of their employees who are members of a union (i.e. firm-level union density) and we use it as a proxy for workers’ bargaining power. We find robust evidence that in firms with a higher degree of unionisation, the reaction of the extensive margin to demand shocks is smaller while the one of the intensive margin is larger. We also find that, for a given demand shock, firms’ price changes are significantly smaller when workers are less unionised. We obtain similar results by looking at a cross section of firms in different European countries from the ECB’s Wage Dynamic Network (WDN) survey. We find a significant negative relationship between bargaining power (proxied by the share of workers covered by collective agreements) and firms’ (relative) propensity to adjust along the extensive margin.
As for the macro evidence, we estimate, with Bayesian methods, the model using data for the euro area over two samples, 1970-1989 and 1990-2017. Estimates point to a sizable decline in the bargaining power of workers – while they obtained more than half of the joint surplus in the earlier subsample, they end up earning their reservation wage in the more recent subsample. As per Figure 6, such a decline in workers’ bargaining power has produced a contraction of the sensitivity of inflation to changes in the output gap as the relative contribution of adjustments through employment has increased.
The erosion of workers’ bargaining power is a long-term trend (see also BIS 2017). Looking ahead, it is still unclear whether a possible de-globalisation accompanied by the reshoring of supply chains might dampen or reverse that trend (see Goodhart and Pradhan 2020). Yet our results suggest that, unless workers’ bargaining power recovers, a flare up of inflation due to tight labour markets is unlikely. This implies that more monetary accommodation might be needed to close the gap between inflation and its target. The inflation-targeting framework itself hinges on the existence of a positive trade-off between inflation and economic activity, hence the structural labour market shifts should be key concerns in the monetary policy debate.
Bank for International Settlements (2017), 87th Annual Report, Box IV.A.
Blanchard, O J (2016), “The Phillips Curve: Back to the ’60s?”, American Economic Review 106(5): 31-34.
Coibion, O, and Y Gorodnichenko (2013), “Is the Phillips curve alive and well after all? Inflation expectations and the missing disinflation”, VoxEU.org, 15 November.
Hall, R E (2013), “The Routes into and out of the Zero Lower Bound”, Prepared for the Federal Reserve Bank of Kansas City’s Jackson Hole Symposium, Global Dimensions of Unconventional Monetary Policy,” August 23.
Constancio, V (2015), “Understanding Inflation Dynamics and Monetary Policy,” Panel remarks at the Jackson Hole Economic Policy Symposium, 29 August.
Del Negro, M, M Lenza G Primiceri and A Tambalotti (2020), “What’s up with the Phillips curve”, NBER working paper 27003.
Goodhart, C, and M Pradhan (2020), “Future imperfect after coronavirus” VoxEU.org, 27 March.
Krueger, A B (2018), “Reflections on Dwindling Worker Bargaining Power and Monetary Policy”, Luncheon Address at the Jackson Hole Economic Symposium, August 24.
Lombardi, M J, M Riggi, and E Viviano (2020), “Bargaining power and the Phillips curve: a micro-macro analysis”, Bank of Italy, Temi di discussion, 1302.
Ohanian, L E, and A Raffo (2012), “Aggregate Hours Worked in OECD Countries: New Measurement and Implications for Business Cycles”, Journal of Monetary Economics 59(1): 40-56.
Stansbury, A, and L Summers (2020), “Declining worker power versus rising monopoly power: Explaining recent macro trends”, VoxEU.org, 2 June.
Williams, J C (2010), “Sailing into Headwinds: The Uncertain Outlook for the U.S. Economy”, Presentation to Joint Meeting of the San Francisco and Salt Lake City.
 This is true under the realistic assumption that labour supply elasticity is less than infinite, that is workers require a higher compensation to work longer.
 The survey is conducted by the Bank of Italy and covers non-financial firms with at least 20 employees.