The Risk of Second-Round Effects
- The surge in Euro area inflation—with headline expected to reach 9% in September—has raised the prospect of second-round effects, where increases in wage growth and long-run inflation expectations push up inflation persistently. We explore the risk of such effects by taking a historical look at the effects of oil price shocks on inflation, wage growth and inflation expectations with an empirical model of the Euro area back to 1970.
- Our results show that higher oil prices have a quick and large effect on headline inflation (the “first-round” effect), as energy prices directly affect headline inflation. Moreover, we find that rising energy prices typically push up wage growth notably for about 18 months and lift inflation expectations persistently (“second-round”). But we also find that higher oil prices tend to lead to tighter monetary policy and a weaker labour market (“third-round”).
- Our results therefore suggest that second-round effects via wages have historically not driven a persistent increase in inflation as third-round effects via higher unemployment have tended to dampen pay pressures after energy price shocks. But we find that second-round effects via rising inflation expectations can lead to persistently higher inflation, although the magnitude is typically modest. Monetary policy has historically played an important role in leaning against the second-round effects, but come at the cost of higher unemployment.
- We split our sample in the mid-1990s and find that the second-round effects via wage setting and inflation expectations have been materially weaker recently than in the 1970s and 1980s. While reassuring at first glance, it is unclear whether this reflects structural changes or macroeconomic conditions. Using a panel of 11 OECD countries we find that structural changes—including reduced worker unionisation and greater central bank independence—have limited second-round effects. But we also find evidence for nonlinearities—with bigger effects after large oil price increases and tight labour markets—which suggests that second-round effects now might be larger than in recent years.
- While our analysis suggests that notable second-round effects via higher wage growth are likely in coming quarters, our results imply that these are unlikely to be the main cause of concern for the ECB, as third-round effects should limit the second-round effects via pay pressures. Instead, our analysis suggests that second-round effects via higher inflation expectations are the key to watch.
Although these tend to be smaller in magnitude, they have historically been more persistent and thus require persistently tighter monetary policy. The recent rising trend in long-term inflation expectations thus points to risks that the ECB will need to hike more quickly than our baseline forecast and take policy into restrictive territory.
The Risk of Second-Round Effects
Euro area inflation surged to 7.5% in April and, given ongoing pipeline pressures from the rise in energy and food prices, we look for inflation to peak at 9% in September. The spike in inflation has raised the prospect of second-round effects, where increases in long-run inflation expectations and wage growth lead to persistently higher inflation in coming years, as workers get used to high inflation and demand higher wages to retain their income share. Such a wage-price spiral led to persistently high inflation after the oil shocks in the 1970s (Exhibit 1).
ECB officials have debated the risk of such second-round effects in recent communication, including the account of the March Governing Council meeting. Some Governing Council members worry that strong second-round effects will emerge given the high level of inflation and the possibility of non-linear effects, pointing to lags in the pass-through from high inflation to wage growth. Others have argued that structural changes since the 1970s make second-round effects less likely and that “third-round” effects (where growth and employment weaken in response to the energy shock) should be expected to mitigate the second-round effects.
Learning From History
Given the uncertainty around the prospects of second-round effects, we take a historical look at the effects of oil price shocks on inflation, wage growth and inflation expectations with an empirical model of the Euro area. To do so, we use the ECB’s Area-Wide Model database with area-wide quarterly data back to 1970Q1.
We start by estimating a statistical (vector-autoregression) model which looks at the interplay of oil prices, the unemployment rate, headline HICP inflation, wage growth, long-term inflation expectations and the short-term policy rate. We construct a composite measure of long-term inflation expectations using the ECB’s Survey of Professional Forecasters (back to 1999Q1), Consensus Economics (using EMU4 data back to 1992Q1) and an inflation trend (estimated back to 1970Q1 to proxy for long-term expectations). We then use our model to construct impulse responses to an oil price shock, which we scale to a 100% increase, roughly similar to the change in oil prices over the last year (or, alternatively, since the outbreak of Covid). We acknowledge that our statistical model is subject to a number of caveats, including uncertainty around the identification of oil shocks and the estimated pre-1992 inflation trend.
That said, our results help capture the three rounds of effects on inflation (Exhibit 2).First, a surge in oil prices has a quick and direct effect on headline inflation (the “first-round” effect), as energy prices enter headline inflation. In particular, we estimate that a doubling of oil prices has historically pushed up headline inflation by 2pp after one year.
Second, we see that higher oil prices push up both wage growth and long-term inflation expectations (the “second-round” effect). Historically, a 100% increase in oil prices has raised nominal wage growth by about 1pp over the subsequent year, with a delay of a couple of quarters given sticky wage setting. The effect on long-term inflation expectations is smaller at about 0.1pp but statistically significant and, importantly, highly persistent.
Third, higher oil prices tend to weaken the labour market (the “third-round” effect), with the unemployment rate rising around 0.3pp after about two years. While part of the weakening of economic activity is likely due to the negative effect of higher energy prices on real incomes and consumption, monetary policy has typically tightened in response to rising oil prices, amplifying the weakening in the labour market. The rise in the unemployment rate, in turn, helps to mitigate the second-round effects, contributing to a slowdown in wage growth in the second year after the oil price shock (and, ultimately, a lower policy rate in light of higher unemployment).
We then disentangle these effects by decomposing the inflation and wage responses into the three pass-through rounds above. For example, to isolate the first-round effect we simulate the oil shock holding constant wage growth, inflation expectations, the unemployment rate and monetary policy. We then calculate the second and third-round effects by endogenising the response of the other variables step by step. Exhibit 3 decomposes the response of headline inflation (left) and wage growth (right) into the three rounds discussed above.
Exhibit 3 offers a few insights. First, the first-round effect of higher oil prices into headline inflation dissipates after about a year, given rapid pass-through from wholesale to retail energy prices (dark blue bars). Second, the right chart shows that most of the second-round effect through wages occurs within the first 18 months after the oil shock, followed by more muted second-round wage effects in subsequent years (dark red bars). Our interpretation is that workers bargain to restore some of their lost income share in response to higher energy prices, but do not push for a permanent increase in wage growth. Third, the second-round effect via inflation expectations is smaller but much more persistent than the effect via catch-up wage setting (light red bars). In particular, our results suggest that higher inflation expectations have historically pushed up both headline inflation and wage growth by 0.1-0.2pp on a persistent basis. Fourth, the “third-round” effect via higher unemployment plays a notable role in mitigating the persistent effects from higher inflation expectations (green bars). Finally, we find that tighter monetary policy is a key factor in leaning against the second-round effects in the second year after the shock, slowing both wage growth and inflation significantly (grey bars).
Taken together, our results show that (1) second-round effects via wage bargaining tend to be sizeable but by themselves have historically not driven a persistent increase in inflation; (2) second-round effects via inflation expectations are smaller in magnitude but more likely to lead to persistently higher inflation; (3) third-round effects via higher unemployment have historically dampened the second-round effects; and (4) tighter monetary policy can play an important role in leaning against the second-round effects.
Given the many economic and institutional changes over time, we split our sample into an “early” period (1970Q1-1994Q4) and a “late” sample (1995Q1-2017Q4).3 Exhibit 4 summarises our results, focusing on the peak response of each variable. Given the different volatility of oil prices across the samples, we show the responses here for a one standard deviation oil price shock.
Our results confirm that the second-round effects via wage setting and inflation expectations were materially larger in the early sample than the more recent period. In particular, we find that the knock-on effect into wage setting was about three times larger in the early period, while the second-round effect into inflation expectations was almost eight times bigger in the first part of the sample. Exhibit 4 suggests that some of this can be traced to bigger third-round effects in the later sample (with a larger rise in the unemployment rate in response to energy increases) and a bigger response of monetary policy.
Back to the 1970s?
While it is reassuring that second-round effects have been more muted since the mid-1990s, it is unclear whether such reduced pass-through is here to stay. On the one hand, structural changes suggest that the risk of second-round effects might be more limited today. In terms of the labour market, central wage bargaining and wage indexation are now less widespread than in the 1970s (Exhibit 5, left), and labour market reforms have reduced the sensitivity of pay growth to labour market conditions.
Moreover, studies have shown that monetary policymaking has become more systematic and credible, as central banks have become more independent and transparent (Exhibit 5, right). As a result, inflation expectations have become more anchored and less sensitive to current inflation.
High inflation and tightening labour markets, on the other hand, could reignite second-round effects despite these structural changes. In particular, non-linearities could emerge in which higher energy prices have larger effects when inflation is already elevated and labour markets have limited slack.
We therefore explore the extent to which second-round effects tend to depend on macro-economic conditions versus institutional characteristics. To do so, we estimate the effect of lagged oil price changes on future wage growth in a quarterly panel of 11 OECD countries since 1970.4 We control for other determinants of wage growth, including the unemployment rate, inflation expectations and trend labour productivity. We estimate a similar equation for inflation expectations.
Exhibit 6 shows the estimated second-round effects for wage growth (left) and inflation expectations (right). Starting with wages, our results point to some nonlinearity in the size of the second-round effect, as the pass-through coefficient is larger for positive oil shocks, large increases in energy prices and during times when labour markets are tight. But we also find evidence that high rates of worker unionisation pushed up the second-round effect of wages in the 1970s and 1980s. Turning to inflation expectations, we likewise find evidence of non-linear effects. While worker unionisation has no effect on the second-round effect via inflation expectations, we see clear evidence that more central bank independence has dampened the effect of energy prices on expectations for future price increases.
Taken together, our results have mixed implications for the likely strength of second-round effects in current circumstances. While structural changes are likely to mitigate the risk of strong second-round effects, evidence of non-linear effects suggests that the effects might be larger in the current environment than in recent years.
Expectations Up, Waiting for Wages?
Evidence on second-round effects remains ambiguous so far (Exhibit 7). On the one hand, measures of inflation expectations have firmed notably, with inflation swaps converging to US levels and long-run inflation expectations in the ECB’s Survey of Professional Forecasters (SPF) now slightly above 2%. On the other hand, indicators of wage growth have not moved up yet, with our area-wide wage tracker still around 2%, in sharp contrast to the US and the UK. Continued labour market improvement, firming wage surveys and the incoming minimum wage increases all point to higher wage growth ahead. But the wage growth data are noisy and released with a long lag, creating uncertainty around the speed at which pay pressures are building.
Our analysis above suggests that sizeable second-round effects via higher wage growth are likely in coming quarters. But our results also imply that these are unlikely to be the main cause for concern, as these tend to be short-lived and third-round effects help limit the second-round effects via wages. That is, our findings suggest that wage growth might not be the deciding factor whether concerning second-round effects are setting in, providing little comfort that they have so far remained moderate.
Instead, our analysis suggests that second-round effects via higher inflation expectations are the key to watch. Although these tend to be smaller in magnitude, they have historically been more persistent and thus require persistently tighter monetary policy. Given the recent upward trend in long-term inflation expectations, we see risks that the ECB will need to hike more quickly than our baseline forecast and take policy into restrictive territory.