AdobeStock_471050977.jpeg
VoxEU Column Monetary Policy

The zero lower bound on inflation expectations

Inflation expectations affect the decisions of households, firms, and policymakers. Expectations of negative inflation can be particularly harmful and lead to deflationary spirals when nominal interest rates are near zero. This column uses survey evidence to show that households and firms almost never expect deflation, even when it is a clear possibility. This apparent zero lower bound on inflation expectations has important implications for macroeconomic dynamics and the effectiveness of monetary policy. Unconventional policies, such as forward guidance, which aim to increase inflation expectations may be less effective when expectations are stuck at the zero lower bound.

Inflation expectations – the pace of price changes that people anticipate in the future – affect households’ desire to consume, firms’ incentives to revise prices, and policymakers’ willingness to change policies. Expectations of negative inflation (i.e. deflation) increase the perceived real interest rate and, according to a conventional macroeconomic view, become contractionary when the central bank nominal interest rate is at its effective lower bounds (ELB). Such a contraction can further depress inflation expectations and lead to a deflationary spiral.

The Great Depression and the almost three-decade long liquidity trap episode in Japan are often used to exemplify this intuition and had a profound impact on policymakers around the world. This experience may potentially rationalise why the central bankers used everything at their disposal – including unconventional policies such as quantitative easing, forward guidance, and negative nominal interest rates – to prevent deflationary spirals during the Global Recession and the Covid-19 crisis. These policies have strong support in standard macroeconomic models with full information rational expectations (FIRE) (Krugman 1998, Eggertsson and Woodford 2003, Eggertsson 2008). However, recent evidence points that people’s beliefs violate the FIRE assumption (Coibion and Gorodnichenko 2012, Bordalo et al. 2020). Hence, one might ask whether policymakers should fear deflationary expectations as much as the standard models imply.

Deflation expectations in the data

In our recent paper (Gorodnichenko and Sergeyev 2021), we document a new fact: households and (tentatively) firms, unlike professional forecasters and financial markets, do not expect deflation even when deflation is a clear possibility. We call this observation a zero lower bound (ZLB) on inflation expectations.

Figure 1 Realised inflation and inflation expectations

 

 

Notes: The top panel is the histogram of realised consumer price index (CPI) inflation (in red) and the Michigan Survey of Consumers inflation expectations (in grey) between 2008:01 and 2020:04. The bottom panel shows the histogram of realised inflation and inflation expectations in Japan.

To illustrate this fact, the top panel of Figure 1 plots the histogram of realised US inflation (a red line) and expected inflation from the Michigan Survey of Consumers (grey bars) between January 2008 and March 2020, a period with several deflationary episodes in the US. Two observations stand out: the inflation expectations distribution is skewed to the right, and virtually nobody expects negative inflation during this period despite a significant fraction of respondents expecting near-zero inflation.

The bottom panel of Figure 1 demonstrates an even more striking example of Japan, where inflation was negative in more than 40% of quarters between 2004 and 2020. Even in this deflationary environment, virtually no consumers expect deflation, and a significant fraction has zero-inflation expectations, consistent with the zero lower bound on inflation expectations.

While Figure 1 is informative, it masks the evolution of the inflation expectations distribution over time. To uncover these dynamics, Figure 2 plots the time series of the inflation rate (top panel) and the fraction of consumers who expect deflation, zero inflation, and positive inflation over time in Japan. One can see that both consumer price index (CPI) and core CPI inflation were often negative between 2003 and 2020. Nevertheless, this inflation evolution has been primarily reflected in the variation of the number of people who expect positive and zero inflation in the bottom panel of Figure 2, which often mirror each other. The fraction of consumers who expect negative inflation is relatively stable. The only exception was during the Global Crisis, when additional factors, such as a sharp decline in oil prices and a rapid appreciation of the yen, may have affected the expectations.

Figure 2 Realised inflation and expectations of positive, zero, and negative inflation

 

Notes: The top panel of the figure shows Japanese realised inflation (bold black line), Japanese realised core inflation (dashed black line). The bottom panel presents shares of Bank of Japan’s Opinion Survey participants who expect positive inflation (green line), zero inflation (yellow line), and negative inflation (red line).

With the zero lower bound on inflation expectations assumption, consumers expecting zero inflation can only revise their beliefs up, while those with positive initial inflation forecasts can lower them. Figure 3 demonstrates that this is indeed the case in the Michigan Survey of Consumers. The figure plots five histograms of changes in inflation expectations for five values of initial inflation forecasts. For example, the red histogram demonstrates that the consumers who expected zero inflation revise their beliefs almost only upwards. The yellow histogram represents consumers who initially believed that inflation would be 5%. These consumers lower their forecasts downward up to 5% but no more.

Figure 3 Inflation expectations revisions

 

Notes: The figure plots five histograms of inflation expectations revisions conditional on five values of the initial inflation expectations: 0, 1, 2, 3, 5 %. The horizontal axis is the size of the revision. The data are from the Michigan Survey of Consumers. The sample period is January 1978 – December 2019.

Our paper (Gorodnichenko and Sergeyev 2021) presents more evidence using alternative surveys of consumers (e.g. the NY Fed Survey of Consumers Expectations, the European Commission Business and Consumer Survey, the Dutch National Bank Household Survey), surveys of professional forecasters and firms, and forecasts of other variables. These additional pieces of evidence reveal that the zero lower bound applies only to inflation expectations of households and firms.

Implications of the zero lower bound on inflation expectations

What are the macroeconomic implications of the zero lower bound on inflation expectations? While remaining agnostic about the source of this constraint (which can be due to, for example, experience-based learning or ambiguity aversion), we explore the implications of the ZLB on inflation expectations in a standard New Keynesian model. The constraint qualitatively changes the model’s predictions. Here are some of them. First, following a negative aggregate demand shock that leads to a severe deflation due to the deflationary spiral mechanism under the FIRE assumption, deflation is limited in our model because inflation expectations do not fall below zero. It can potentially explain a missing deflation during the Great Recession in the US and the lack of significant deflation in Japan. Second, unconventional policies, such as forward guidance, that aim at lifting inflation expectations during liquidity traps are less effective. Intuitively, policies that would increase full information rational expectations (FIRE) inflation expectations are infra-marginal when expectations are stuck at the zero lower bound. Third, in liquidity traps, negative aggregate supply shocks are no longer expansionary, and fiscal multipliers do not become infinite when inflation expectations are stuck at zero, which is consistent with recent empirical evidence (Miyamoto et al. 2018, Wieland 2019).

Conclusion

In contrast to professional forecasters, households (and firms) do not anticipate deflation in the future even when deflation is likely. Households’ inflation expectations with a floor at zero affect macroeconomic dynamics and the effectiveness of unconventional policy tools that rely on the management of expectations in a standard New Keynesian model.

References

Bordalo, P, N Gennaioli, Y Ma and A Shleifer (2020), “Overreaction in Macroeconomic Expectations”, American Economic Review 110(9): 2748–82.

Coibion, O and Y Gorodnichenko (2012), “What can survey forecasts tell us about information rigidities?”, Journal of Political Economy 120(1): 116–159.

Eggertsson, G and M Woodford (2003), “The zero bound on interest rates and optimal monetary policy”, Brookings Papers on Economic Activity.

Eggertsson, G B (2008), “Great Expectations and the End of the Depression”, American Economic Review 98(4): 1476–1516.

Gorodnichenko, Y and D Sergeyev (2021), “Zero Lower Bound on Inflation Expectations”, NBER Working Paper 29496.

Krugman, P R (1998), “It’s baaack: Japan’s slump and the return of the liquidity trap”, Brookings Papers on Economic Activity 1998(2): 137–205.

Miyamoto, W, T L Nguyen and D Sergeyev (2018), “Government spending multipliers under the zero lower bound: evidence from Japan”, American Economic Journal: Macroeconomics 10(3): 247–77.

Wieland, J F (2019), “Are negative supply shocks expansionary at the zero lower bound?”, Journal of Political Economy 127(3): 973–1007.

2,730 Reads