Is the Global Financial Crisis an outlier? Analyzing inflation dynamics during financial and non-financial crises

(1)

1 Is the Global Financial Crisis an outlier?

Analyzing inflation dynamics during financial and non-financial crises

Floortje Merten 10379975 December 15, 2017

Abstract

The past decade has been characterized by extraordinary inflation dynamics. Low inflation rates since the Global Financial Crisis can be attributed not only to cyclical factors, but also to structural factors exhibiting a permanent downward effect. By analyzing almost 150 recessions since 1980 in both advanced and emerging countries, this thesis provides evidence that part of inflation dynamics observed during this financial crisis, is in line with historical precedents. But the disinflationary pressures associated with the recent financial crisis, characterized by long and severe recessions, overrule a general financial crisis effect. The Global Financial Crisis is unique, since the widespread sharp and persistent inflation drop is unprecedented over a horizon of decades.

University of Amsterdam

Faculty of Economics and Business Master in Economics

Specialization Monetary Policy and Banking

Superviser: Aerdt Houben

(2)

2 Preface

This research is part of a DNB workstream on inflation that focuses on inflation dynamics. The key question is whether the inflation process has changed, and what this implies for the ECB’s monetary policy – and in particular for its current strategy. This thesis is inspired by research by Homar and Van Wijnbergen (2016) that focuses on the duration of recessions during financial crises conditional on bank recapitalization. I am grateful to Aerdt Houben for being a critical but extremely motivating supervisor. Besides, I thank Gabriele Galati and Sweder van Wijnbergen for sharing their perspectives and ideas, and for supporting my research. Lastly, I thank Maurice Bun for sharing econometric knowledge, and Martin Admiraal for statistical support. I hereby declare that this thesis is my own work. To the best of my knowledge, it contains no sources other than the ones mentioned in the thesis.

(3)

3 The economics textbook taught you that a drop in aggregate demand leads to a recession, during which output falls below potential, meaning: the usual level of production given the

economy’s resources and technology. This is however a temporary effect. The recession is followed by a recovery period, during which output goes back to its potential level. The

recession does not significantly affect potential output.

(4)

4

1. Introduction

Inflation remains extraordinary low for too long. The world has been experiencing a downward trend towards lower inflation rates since the Global Financial Crisis. Economists make large efforts to understand the macroeconomic dynamics of the global economy over the past decade. Despite an elaborate program of conventional and unconventional monetary measures, including more than doubling the central bank’s balance sheet, the ECB has not been able to reach its medium term inflation target within the euro area. The ECB has vowed to maintain its unconventional monetary program until it reaches a “sustained adjustment in the path of inflation which is consistent with the aim of achieving inflation rates below, but close to, 2 percent over the medium term”. Persistently low inflation since the financial crisis underlines the necessity of research focused on the evolving inflation process and its drivers.

Taking a broader perspective, this raises the question whether recent inflation dynamics are specific to the Global Financial Crisis. The discussion on the financial crisis and associated low inflation involves two main views. The first view considers the financial crisis to be part of a financial supercycle. Believers claim the very steep decline of output and subsequent slow recovery to be characteristic to the Global Financial Crisis. They also emphasize the extreme magnitude of the financial cycle, huge levels of leverage, and a drop in employment that was way more persistent following this financial crisis than it would have been following a usual recession (Rogoff, 2015). The second view relates the current stance of the global economy to secular stagnation, and emphasizes fundamental changes within the economy.

The question whether this crisis is extraordinary, is a key question for the ECB, since the answer may have significant implications for monetary policy. Currently, the ECB is trying to steer inflation towards its medium term objective. However, some propose that the ECB should alter its main objective of price stability. According to those, the effects of real factors like globalization and technology on inflation are being underestimated. A possible consequence is greater tolerance for deviations of inflation from its target, and larger weight on financial stability (Borio, 2017b).

This thesis consists of a literature review and an outlier analysis, both focused on the question whether inflation dynamics post-2007, a period characterized by the Global Financial Crisis, are different. The literature review covers inflation dynamics during the Global Financial Crisis, the discussion on the evolving inflation process and its temporary and structural drivers, and general financial crisis characteristics. The outlier analysis refers to the event study in which inflation dynamics during the Global Financial Crisis are compared to

(6)

6 historical inflation dynamics during both financial and non-financial crises. The main goal of this analysis, split up into a graphical data analysis and a supporting empirical analysis, is to distinguish between inflation dynamics that can be explained by the typical characteristics of a financial crisis, and inflation dynamics that cannot and are specific to the recent decade.

The panel dataset underlying the outlier analysis is unique. Quarterly GDP time series subtracted from underlying sources are aggregated to create an extensive panel dataset. The dating of recessions, constituting the foundation of the event study, is based on an accurate analysis of country specific GDP growth rates. This results in 149 events, covering a sample of 33 advanced and emerging economies over almost 40 years. In this thesis, a non-financial crisis refers to a normal recession that does not coincide with a financial crisis, while a financial crisis refers to a recession that does. Including both event types facilitates the isolation of a consistent financial crisis effect. The dating of financial crises is based on the Systemic Banking Crisis Database by Laeven & Valencia (2012). Event specific inflation rates form the final panel dataset, and are subject to both indexation and calculations in the outlier analysis.

The literature review clarifies that structural factors, like globalization and technology, have a downward effect on worldwide inflation since decades. Recent cyclical factors and the severity of the Global Financial Crisis clarify to a large extent why inflation since 2012 has been unexpectedly low in much of the developed world. The outlier analysis illustrates that an important part, but not all, of the recent disinflation is in line with inflation dynamics during preceding financial crises. Inflation declines on average more during a recession that coincides with a financial crisis, than during a recession that does not. However, the widespread extraordinary inflation drop and subsequent slow recovery is unprecedented, and specific to the Global Financial Crisis.

The set-up of this thesis is as follows. The literature review is split up. Paragraph 2 focuses on recent inflation dynamics, its drivers, and the risk of de-anchoring inflation expectations. Paragraph 3 outlines financial crisis characteristics. Subsequently, the outlier analysis follows in paragraph 4, consisting of both the graphical data analysis and the supporting empirical analysis. Paragraph 5 concludes.

(7)

7

2. Studying the inflation process

This first part of the literature review focuses on the inflation process and its drivers. Paragraph 2.1 elaborates on inflation in the aftermath of the Global Financial Crisis. Paragraph 2.2 focuses on the short-term drivers of inflation and presents a compact timeline of the financial crisis. Paragraph 2.3 sums up important long-term drivers of inflation. Finally, paragraph 2.4 underlines the risk of de-anchoring inflation expectations.

2.1 Investigating recent inflation dynamics

“If we can’t get inflation back up, trouble lies ahead” (Adam Posen, as cited in Giles, 2017, p.2).

Inflation dynamics characterizing the past decade have been introducing much uncertainty about monetary policy. A twin puzzle emerged in the aftermath of the Great Recession, and characterizes inflation dynamics in the euro area, the U.S., and other advanced economies.

At the trough of the Great Recession, inflation did not fall as much as predicted by economic models, according to which a prolonged and potentially disastrous period of disinflation should have taken place. Despite the severity and duration of the recession, and rising levels of unemployment, inflation remained relatively stable. ‘Missing disinflation’ characterizes inflation dynamics in advanced economies from 2009 until 2011, and can be explained by stable inflation expectations due to successfully established inflation targeting regimes and decreased sensitivity of inflation to economic slack (Friedrich, 2014).

Gilchrist, Schoenle, Sim and Zakrajšek (2017) support the ‘missing disinflation’ puzzle in the U.S. by estimating a model with financial frictions and a customer-market. Financial frictions create incentives to increase prices in reaction to adverse shocks. The customer-markets theory makes firms’ pricing decisions responsive to customer acquisition and retention. By studying U.S. good-level prices, firm income and balance sheet data between 2005 and 2012, they find that firms with limited internal liquidity significantly increased their prices in 2008, characterized by the financial turmoil. Counterparts not constrained by liquidity problems decreased their prices during the same period. The model thus explains the absence of disinflationary pressures following a drop in output, as observed during the Great Recession.

(8)

8 Since 2012, inflation in advanced economies has been following a downward trend, and has been well below pre-crisis levels for years, referred to as the ‘missing inflation’ period. Even though the global economy has lived up and recently reached its highest growth levels since years, inflationary pressures as aimed for by central banks remain absent. Inflation persistence increased, implying a delay in the convergence towards a long-term objective. Fundamental dependencies underlying economic mechanisms seem to have broken down (Ciccarelli & Osbat, 2017).

Inflation dynamics throughout the past decade can be accounted for on the one hand by short-term or cyclical factors, and on the other hand by long-term or structural factors that lead to a decrease in trend inflation. Although globalization, technological progress and demographic changes are consistent with a downward inflation trend, adverse cyclical factors explain the largest part of the ‘missing inflation’ in the euro area since 2012 (Ciccarelli & Osbat, 2017). This fits the traditional assumption that long-run inflation is a monetary phenomenon, and is not permanently affected by real influences (Deroose & Stevens, 2017). In any case, low inflation rates moving away from target come with significant risks to monetary policy and financial stability.

2.2 Short-term drivers of low inflation

2.2.1 The weakened link between inflation and economic activity

At the core of many macroeconomic models lies the Phillips curve, relating inflation to its lag, inflation expectations, and a measure of economic activity, like unemployment or the unemployment gap. The Phillips curve predicts that inflation increases in reaction to economic growth, and drops in reaction to economic contractions. However, recent inflation dynamics cast doubt on the empirical value of the Phillips curve.

Following the Great Recession, output remained below its pre-crisis trend while inflation initially dropped less than predicted in most advanced economies. This suggests either a weakened link between inflation and economic activity or hysteresis, meaning that recessions may have a permanent effect on output relative to the pre-recession trend. Blanchard, Cerutty and Summers (2015) investigated the link between inflation and economic activity during 122 recessions within 23 advanced countries between 1965 and 2015. They find that about two thirds of the recessions are associated with hysteresis, which implies that they are followed by relatively low output compared to the pre-recession trend. In about one-half of the recessions, output growth also remains below pre-recession trend. As a result, the

(9)

9 differential between output and predicted output based on pre-recession trend increases. Supply shocks, like an increase in oil prices or a coinciding financial crisis, explain the characteristics of a recession and related low output or low output growth. To test if the correlations imply causality, other types of shocks are analyzed. Since almost two thirds of the recessions associated with intentional disinflations, a strong form of a demand shock, are followed by lower output, they claim that hysteresis may indeed be at play on a large scale. These findings provide insight into absent inflationary stimulus throughout the recovery from the Great Recession.

Phillips curve estimations by Blanchard et al. (2015) suggest that long-term inflation expectations instead of lagged inflation have become more important in predicting inflation, which may explain the absence of a deflationary spiral during the ‘missing disinflation’ period. They also found evidence for a weakened relationship between the unemployment gap and inflation between 1970 and 1990, associated with the shift towards inflation targeting. Summarizing, inflation predictions have become less dependent on historical inflation, and more dependent on the central bank’s inflation objective. Hence, inflation has become more persistent, and less sensitive to economic slack. Even though they did not find a significant shift in the Phillips curve since 1990, Kuttner and Robinson (2010) claimed that inflation became less sensitive to changes in output and unemployment several years before the Global Financial Crisis. They found that the flattening of the Phillips curve is mainly due to a change in the price-setting behavior of firms, related to globalization. However, the exact shift in the empirical relevance of the Phillips curve remains unclear, since literature on this matter is to a large extent contradictory.

2.2.2 A timeline of short-term inflation drivers during the Global Financial Crisis

Global inflation has been volatile over the past decade. The following analysis provides an overview of cyclical factors underlying the development of global inflation during the recovery from the Great Recession.

The global economic recovery lost momentum in 2012, reflecting weaker output growth in emerging countries, muted GDP growth in the U.S., and the Great Recession in the euro area (BIS Annual Report, 2013). Weak global economic activity, and the associated decrease in commodity prices, reduced global inflationary pressures. Low inflation prospects dominated worldwide. A weak medium-term growth outlook for the major advanced countries was prevailing. Besides, new risks for inflation came from energy prices. The increase in oil prices during 2010 and early 2011 stagnated and reversed (World Economic

(10)

10 Outlook, 2012). Trends in commodity prices, and in particular in oil prices, have been among the main drivers pushing down inflation worldwide (Berganza, del Río & Borrallo, 2016).

Figure 1: Global inflation dynamics around the Global Financial Crisis1

Inflation during this period was remarkably stable, as illustrated by Figure 1. This can be partially explained by the fact that inflation in advanced countries has become less sensitive to economic slack. The severity of the recession caused structural unemployment. This implies that unemployed cannot compete on the labor market and have less influence on wages, associated with muted disinflationary pressures. The extended duration of the Great Recession caused exceptionally low long-term employment rates. Long-term inflation expectations became better anchored as a result of central bank credibility, also explaining part of the ‘missing disinflation’ (World Economic Outlook, 2012). Summarizing, relatively stable inflation was consistent with ongoing economic slack and a subdued response to cyclical conditions. Future inflation was expected to remain low in spite of monetary easing.

The downside risks persisted in 2013 and 2014, and global growth, although slightly lifted, remained slow and below pre-crisis averages. Growth mainly stabilized in advanced countries, accommodated by financing conditions. Inflationary pressures remained subdued, reflecting not only slow domestic growth and relatively low usage of domestic resources, but also global factors (BIS Annual Report, 2014). Low inflation indicated that output gaps still existed. Inflation expectations were declining. Productivity remained too low, and debt burdens and financial risks too high (BIS Annual Report, 2015). In the major advanced

1_{Data source: IMF and OECD}

-2 0 2 4 6 8 10 12 C P I in flatio n

(11)

11 countries, central banks remained concerned about inflation below objectives, highlighting an incomplete understanding of the drivers of inflation.

Instead of cyclical drivers, global drivers started to gain importance. The increased sensitivity of inflation to global factors reflects a greater integration between product and factor markets, influencing the pricing power of domestic producers and the bargaining power of workers. The correlation between domestic unit labor costs across countries increased, suggesting that inflation has become less sensitive to developments in domestic capacity (BIS Annual Report, 2015).

Between July 2014 and March 2015, the third largest oil price drop of the last half-century took place, strongly adding to global disinflationary pressures (BIS Annual Report, 2015). Inflation rates in the main advanced countries dropped to extremely low, sometimes negative levels. Despite disinflationary pressures, inflation rates remained high in several of the main emerging countries due to currency depreciations (Barganza, del Río & Borrallo, 2016).

Over the course of 2015, labor markets improved, which was however not associated with better economic performance in general. Output growth remained moderate and productivity growth weak, continuing its long-term decline mainly observed in advanced countries. Inflation remained subdued, except in some emerging countries. In the advanced countries, core inflation picked up, although it remained below target. Headline inflation was still low. The drop in commodity prices clarifies the gap between headline and core inflation around this time (BIS Annual Report, 2016). Note that significant variation in exchange rates explains the large differences in inflation rates across countries. The exchange rate pass-through has however weakened over the past decades due to better anchoring of inflation expectations and globalization (Barganza, del Río & Borrallo, 2016).

In 2016 and 2017, as unemployment rates and output gaps were shrinking, and as oil prices were picking up, demand pressures caused a modest increase in global inflation momentum. Even though economic growth picked up, a substantial and robust revival of inflation remained unlikely. Besides the decline in headline inflation, which can be accounted for by temporary factors, a decline in inflation has also been observed in core inflation rates, reflecting disinflationary pressures associated with structural drivers (Berganza, del Río & Borrallo, 2016).

(12)

12 2.3 Long-term drivers of low inflation

2.3.1 Globalization

It seems naïve to adhere to past frameworks of the inflation process in the context of a globalizing world in which the Soviet bloc and China have entered into the global economy, and in which other emerging markets have opened up to international trade (Giles, 2017). According to the globalization of inflation hypothesis, increased international economic interrelatedness over the past decades can account for part of the changes in the inflation process. The effect of globalization on inflation can be split up into a symmetrical and an asymmetrical effect. The symmetrical effect implies that, besides domestic slack, global slack should also be considered as a disinflationary force. The asymmetrical effect implies that low-cost producers and relatively cheap labor affect inflation once they enter the global economy. Especially in advanced economies, this asymmetrical effect is present, at least until costs converge (Borio, 2017b). As a consequence of globalization, worldwide inflation contains a common factor, which results on the one hand from common shocks spread out over global value chains that have become more complex over time, and on the other hand from converging monetary policy frameworks around the globe (Ciccarelli & Osbat, 2017).

Today’s global value chains include large international manufacturers that do not own all of their intermediate manufacturing plants. This challenges the original view where value chains are purely or mainly domestic, and where inflation is a domestic phenomenon closely linked to country-specific demand pressures. The enlargement of global value chains results from increasing international trade in intermediate goods and services, facilitated by advances in technology. Globalization implies that excess demand can be offset by redirecting part of the production to other countries that are experiencing economic slack. Hence, demand and supply conditions in one country influence output prices in another country, which in turn also affect wages and prices of other factor inputs. The result is that global slack becomes important for the domestic inflation process, and that trade in both intermediate goods and final goods strengthens this effect of global slack (Auer, Borio & Filardo, 2017).

The importance of low-cost producers in global production is increasing, and puts downward pressures on global prices. Importing countries experience lower import prices, hence low-cost production impacts domestic inflation. The negative effect of import penetration by lower-price Asian and Chinese producers on domestic inflation in the euro area is estimated to be 0.3 percentage point per year between 2000 and 2005. For the U.S., this effect is estimated to be 0.1 percentage point per year. Note that the final effect of import

(13)

13 penetration by low-cost producers depends on the reaction of competitors and domestically generated inflation, which in turn depends on the monetary policy response. The finding that the effect of import prices on domestic prices in 22 OECD countries is significantly larger than the ratio of imported goods and services over total domestic demand is striking. This implies that increased competition from low-priced imports puts pressure on domestic producers that have to compete with import industries, who lower the mark-up on domestic costs (Pain, Sollie & Koske, 2008). Moreover, mark-ups have become less responsive to domestic cyclicality in general, as a result of the increased importance of international prices relative to domestic prices (Cicarelli & Osbat, 2017).

Ciccarelli and Mojon (2010) conducted research on the co-movement of inflation rates across 22 OECD countries between 1960 and 2008, and show that the international co-movement has been extraordinarily high. Average inflation over the 22 countries can account for almost 70 percent of the variance in inflation within the sample. Besides, they emphasize that inflation rates in the individual countries converge towards the global inflation average, and that individual deviations from global inflation are being reverted. According to the globalization of inflation hypothesis, tighter co-movements of national inflation rates mirror increasing structural integration of goods and labor markets (Auer et al., 2017).

2.3.2 Technological progress and e-commerce

Other structural factors that potentially have a downward effect on inflation relate to behavioral changes due to technological innovation, such as the increase in online sales, at both consumer and business-to-business level. The growing importance of e-commerce puts downward pressure on prices through different channels. Firstly, both producers and consumers are able to compare different offers and hence save costs. Secondly, cost savings due to reductions in operating costs can be passed on to the consumer. Thirdly, e-commerce increases transparency, resulting in increased competition and downward pressure on prices and profit margins. The importance of e-commerce has been substantially increasing over the past decade; in the euro area, where 30 percent of the people searched online for information about a purchase in 2002, 65 percent of the people searched online for this information in 2014 (Ciccarelli & Osbat, 2017).

However, the actual pass-through of increased e-commerce to lower prices is not that strong, since it only explains a limited share of recent low inflation. Considering EU countries between 2003 and 2015, an increase in the share of people that looks online for goods and services of one percentage point is estimated to reduce non-energy industrial goods (NEIG)

(14)

14 inflation by 0.025 percentage point. Regarding the time span covered in this estimation, this implies that the increased share of people looking online for their purchase led to a decrease in NEIG inflation of about 0.1 percentage point per year (Ciccarelli & Osbat, 2017).

2.3.3 Demographic changes

A radical shift in the composition and size of the world population is about to materialize, which will affect the global economy. The main drivers of global demographic change are declining average fertility and mortality rates. During the 1950s and 1960s, fertility rates were elevated. Combined with an increase in life expectancy, this caused the total population to grow. But total population growth has been following a decreasing trend since then. For the world as a whole, the population growth rate is expected to remain positive. However, population growth for the OECD as a whole is expected to become negative around 2050. In Japan and Germany, the total population already started to decline (Yoon, Kim & Lee, 2014).

The working-age share of the total population in OECD countries followed an increasing trend until 2000. Since then, decreasing fertility combined with increasing longevity has been causing the working-age population share to diminish. The elderly dependence ratio, mirroring the working-age population, indicates a significant change in the population structure around the 2000s. The dependence ratio has been increasing since then, and is expected to remain increasing over time (Yoon et al., 2014).

An aging population is generally associated with lower aggregate demand, shifts in relative prices due to consumption preferences, and declining asset prices that lead to a negative wealth effect. Overall, this is associated with disinflationary pressures. On the other hand, the demographic changes could reduce the effective supply of labor, putting an upward pressure on inflation. The overruling effect depends on how demographic changes affect aggregate supply and demand, inflation expectations and asset prices, which in turn depend on real and nominal frictions, institutional design and behavioral reactions (Yoon et al., 2014).

Yoon et al. (2014) investigated the relation between demographic change and inflation in 30 OECD countries between 1960 and 2013. They conclude that an increasing size of the population puts upward pressures on inflation, due to increased aggregate demand. When the dependence ratio is taken into account, this puts significant downward pressure on inflation, while population growth remains associated with inflationary pressures. This also holds when life expectancy is taken into account. Overall, the ongoing demographic changes are expected

(15)

15 to have a significant disinflationary impact over the following years. This effect is particularly strong in economies experiencing aging combined with a quick population decline.

2.4 The risk of de-anchoring inflation expectations

Praet (2016), Executive Board member of the ECB, underlines the risk of de-anchoring inflation and inflation expectations towards the lower end of the price stability range. This may be the consequence of a central bank being too tolerant with its policy throughout an extended period of low inflation, or a central bank being unable to steer short-term inflation. Once the public’s faith in the central bank’s inflation objective diminishes, the economy might settle into a lower inflation regime.

Increased inflation expectations usually lead to upward nominal wage pressures and subsequently to increases in prices for goods and services, once raised by firms in reaction to the cost increases (de Haan, Hoeberichts, Maas & Teppa, 2016). This effect remains absent in an environment with declining inflation expectations. The fact that inflation expectations are less anchored when inflation is persistently low, instead of inflation being near the central bank’s objective, is worrisome. Hence, in a persistently low inflation environment, inflation expectations are relatively more responsive to past inflation. Besides, inflation expectations decrease in reaction to lower than expected inflation, but do not get revised upwards in reaction to higher than expected inflation (Ehrmann, 2015). Overall, when inflation expectations decrease, more weight is attached to past inflation, and inflation is driven downwards (Ciccarelli & Osbat, 2017).

Galati, Poelhekke and Zhou (2011) investigated the anchoring of long-term inflation expectations around the onset of the Global Financial Crisis. Long-term inflation expectations, when perfectly anchored at the central bank’s inflation target, should not be influenced by relevant macroeconomic news. Inflation expectations based on survey measures remained fairly stable in the euro area between 2006 and 2009, but increased somewhat in volatility in the U.S., and even more so in the U.K. They conclude that long-term inflation expectations became less anchored in the U.S. and the U.K. This may reflect altered market agents’ views on the central bank’s commitment to control inflation. Perhaps market agents presume that, at the peak of the financial crisis, the monetary transmission mechanism and financial stability are the central bank’s primary focus. Especially when the economy recovers, central banks should be prudent when exiting expansionary strategies.

The ECB’s credibility has been affected by the sovereign debt crisis, according to Pagenhardt, Nautz and Strohsal (2015). They analyzed the anchoring of inflation expectations

(16)

16 in the euro area between 2006 and 2015. Euro area inflation expectations have remained anchored until fall 2011, and have been de-anchored since, suggesting that the sovereign debt crisis increased market agents’ uncertainty regarding the ECB’s monetary strategy.

In the euro area, the risk of de-anchoring inflation expectations was particularly present in 2013 and 2014, when hit by a series of supply shocks. Declining inflation expectations throughout this period may refer to either hampered central bank credibility or concerns regarding policy effectiveness. Hampered central bank credibility and decreased trust in the central bank’s ability to reach and maintain its price stability objective are reflected by shifts in the mean expected inflation rate, and represents a deterioration of the long run equilibrium. Concerns regarding policy effectiveness are reflected in an increase in the persistence of inflation rates below the implicit inflation target. A loss in credibility may reveal expectations of secular stagnation and low inflation, while a decline in policy effectiveness implies a slower recovery towards the long run equilibrium. Determining what precisely triggered the decline in inflation expectations in the euro area is difficult. Either way, de-anchored inflation expectations below the central bank’s price stability objective come with costs (Ciccarelli & Osbat, 2017).

The re-anchoring towards a lower inflation rate together with the continuing economic slack holding back core inflation, and hence the lacking convergence point for long-term inflation expectations (since core inflation is a good predictor of headline inflation over the medium term), is what prompted the Governing Council to upscale its asset purchase program over 2015. From then on, the ECB’s monetary policy was refocused from fixing the monetary policy transmission mechanism towards fighting persistently low inflation (Praet, 2016).

(17)

17

3. The financial crisis

This second part of the literature review focuses on financial crises in general and on the Global Financial Crisis in particular. Paragraph 3.1 summarizes characteristics of the recent financial crisis. Paragraph 3.2 elaborates on two main views on the recent financial crisis, being the secular stagnation hypothesis and the financial supercycle hypothesis. Paragraph 3.3 analyzes similarities across financial crises over time. Finally, paragraph 3.4 focuses on the comparison between financial and non-financial crises.

3.1 The Global Financial Crisis in perspective

In the years following the Global Financial Crisis, much literature focuses on the question: Is this crisis different? Many characteristics of the Global Financial Crisis itself, including financial indicators in the run-up, are claimed to be specific to this financial crisis. But according to others, the ‘this-time-is-different’ syndrome is somewhat naïve. They claim that we have been here before, and emphasize the similarities between financial crises over time.

The current low inflation environment is one phenomenon underlying the ‘this-time-is-different’ conviction. Despite unprecedented conventional and unconventional monetary easing, the ECB has still not achieved its main objective: an inflation rate that is below but close to 2 percent over the medium term. This specific definition of price stability exists since 1999, but price stability as a central bank objective has been more prevalent since the Great Moderation. This period, ranging from the mid-1980s up to 2007, is typically seen as a period of extraordinary economic performance and a drop in macroeconomic volatility (Coibion & Gorodnichenko, 2011). Low and stable inflation has been the main objective of central banks in many advanced economies since then. Central bank independence and inflation targeting were the means to keep inflation expectations in control (Summers, 2016). During the Great Moderation, economic growth was generally stable, and inflation was generally low.

“Gone are the illusory certainties of the Great Moderation; with us are the deeper, sometimes troubling, questions about how the economy really works” (Borio, 2017a, p.1).

The extended period of stable macroeconomic conditions and a flourishing economy facilitated financial innovation in advanced countries, first amongst which the U.S. According to many analysts, the build-up of financial imbalances prior to the Global Financial Crisis in the U.S., like the increase in housing and equity prices, can be explained by this financial

(18)

18 innovation, as well as by the steady inflow of capital from Asia and a fall in macroeconomic risk during the Great Moderation. The sustainability of the exceptionally large U.S. current account deficit can, likewise, be explained by the new characteristics of the global economy. The U.S. was expected to enjoy large increases in productivity for decades (Reinhart & Rogoff, 2009a). But now we know that financial innovation also facilitated trade in complex financial instruments, with dramatic consequences felt in the global economy once the U.S. started to experience contractions in wealth, jumps in risk spreads and deteriorations in credit market conditions from summer 2007 onwards.

Looking backwards, the Global Financial Crisis has several characteristics that support the view that this crisis was different, like the steep drop in output and the subsequent relatively slow recovery. The severity of the Global Financial Crisis can be explained by the build-up of financial imbalances caused by the housing boom and bust, the huge leverage underlying the bubble, the major run-up and subsequent fall in equity prices, and the persistent increase in unemployment rates. The large increases in public debt levels following the Global Financial Crisis are unprecedented, and hence interpreted as a phenomenon specific to this financial crisis, exacerbating economic problems (Rogoff, 2015).

In order to determine whether the Global Financial Crisis is an outlier, researchers should put the financial crisis in the right perspective. Taking a long enough time span facilitates capturing different types of rare events, notably including financial crises and depressions, which appear to exhibit more similarities than researchers often think. By putting the Global Financial Crises in context with financial crises both before and after World War II, Reinhart and Rogoff (2014) claim that the Global Financial Crisis is actually no outlier.

Shortly after the onset of the Global Financial Crisis, its dire consequences could have been predicted. Reinhart and Rogoff (2009b) already emphasized the consequences of the financial crisis for the U.S. back in 2009, based on historical comparisons of unemployment levels, output and sovereign debt. The fact that this financial crisis was global, implied that it was difficult to step out of financial distress by increasing imports or by consumption smoothing through borrowing from other countries, which holds for both advanced and emerging countries. Jordà, Schularick and Taylor (2011) also concluded in 2009 that the recovery from the Great Recession would become sluggish. This conclusion was based on the estimated large cost of recessions that coincide with financial crises relative to normal recessions, and the even higher cost of recessions that coincide with a financial crisis that is global.

(19)

19 According to some, global imbalances and financial crises are the result of common causes, meaning that both external and domestic factors underlie the boom that went bust in 2007-2008. This view is expanded by Obstfeld and Rogoff (2009), who claim, just like Bernanke (2009), that the global imbalances in trade and capital flows since the beginning of the latest half of the 1990s and the Global Financial Crisis are interconnected. Three unsustainable trends during the mid-2000s are claimed to be characteristic to the run-up to the Global Financial Crisis. Firstly, in many countries, including the U.S., real estate values were rising. Secondly, several countries were experiencing increasing current account deficits at the same time. Thirdly, the amount of leverage had been building up to high levels in many sectors around world. Growing global imbalances permitted the global economic system to become more fragile and interconnected over time.

3.2 Secular stagnation or financial supercycle?

If you had told a rational economist back in 2009 about the conventional and unconventional monetary policies that were about to be implemented, they would probably have started to worry about inflation. Observing the low long-term government bond rates ten years after the onset of the Global Financial Crisis, markets still expect low inflation as well as low real interest rates, and do not assume that market conditions will go back to normal anytime soon. Nowadays, it is common to doubt a strict convergence towards a predetermined trend.

“As surprising as the recent financial crisis and recession were, the behavior of the world’s industrialized economies and financial markets during the recovery has been even more so” (Summers, 2016, para.1).

The secular stagnation hypothesis assumes structural changes within the economy, and can be summarized in three assumptions. Firstly, according to this view, the global economy has been suffering from insufficient aggregate demand since years before the onset of the Global Financial Crisis, which was likely to keep economic growth sluggish. Explanatory factors supporting this view are population ageing, increasing inequalities in both income and wealth, and decreased tangible investment due to technological progress. Secondly, proponents claim that the bubble preceding the Global Financial Crisis was the only reason for pre-crisis full employment, and for output to reach potential (Borio, 2017a). Thirdly, the hypothesis holds that equilibrium real interest rates have been at such low levels that they cannot be achieved through conventional central bank policies. At that point, desired

(20)

20 levels of savings outgrow desired levels of investment, resulting in deficient demand and stunted growth. The imbalance between savings and investment puts downward pressure on real interest rates. Over the past years, we have been observing sluggish demand, and persistently low inflation rates in spite of ambitious monetary easing, which fits the secular stagnation hypothesis. For this hypothesis to seem plausible, there have to be sufficient reasons to assume that equilibrium real interest rates have been abnormally low, and that propensities to save and invest developed accordingly. Supporting literature found declines in equilibrium real interest rates of several percentage points, and suggests that the increased propensity to save has been driven by the increased income and wealth inequality, more uncertainty regarding retirement and the availability of benefits, worsened borrowing conditions, and the accumulation of assets by sovereign funds and central banks. The decreased propensity to invest can be explained by a deceleration in labor force growth, cheaper available capital goods and a tighter credit market with stricter regulation (Summers, 2016).

By contrast, the financial supercycle hypothesis refers in the first place to a global failure to suppress the accumulation of the financial cycle, and hence to hinder severe and protracted recessions, sluggish and long-lasting recoveries, and low levels of productivity growth. The buildup of a financial supercycle implies credit and asset prices growing at the same time, causing a risky boom that might once turn to bust. Secondly, according to this view, the boom in the run-up to the crisis causes output to reach levels above potential, meaning that a boom was not even necessary in order to achieve potential. Thirdly, the hypothesis argues that the equilibrium real interest rate is at a positive and higher level than claimed by the secular stagnation hypothesis, since the deficiency in demand has been exaggerated, and since global factors from the supply side that have been the driving forces of inflation dynamics have been underestimated (Borio, 2017a). Contradicting the secular stagnation hypothesis, the downturn of the financial supercycle is temporary (Rogoff, 2015). According to proponents of the financial supercycle, the economy will be ready for a new phase of increasing leverage once it starts to recover. Financial innovation circumvents financial regulation and induces real interest rates to increase again (Borio, 2017a).

Proponents of the financial supercycle claim that deficient global demand was not widely observed in the run-up to the Global Financial Crisis, and that pre-crisis output was above potential. This is not only a post-crisis conclusion, but one that could have been spotted in the run-up to the financial crisis by analyzing the buildup of financial imbalances, like extraordinary credit growth and increases in housing prices. Indicators of the buildup of

(21)

21 financial imbalances could have pointed out real time that the economic expansion was unsustainable. Besides, proponents of the financial supercycle hypothesis stress that severe banking crises in general lead to long-term economic harm, and relatively slow and protracted recoveries. Therefore, regarding the unprecedented depth and breadth of the Global Financial Crisis, a slow economic recovery could have been expected. Furthermore, research has pointed out that financial credit booms have the tendency to undermine productivity growth, as the impact of investment that is allocated towards relatively low-productivity growth sectors during the boom is very large in the aftermath of the crisis. Lastly, research has shown that there is only a weak link between deflation and output drops, which completely disappears once asset prices are taken into account. This again emphasizes the large role of the financial cycle in explaining economic failure (Borio, 2017a).

Differences in both hypotheses lie mainly in the weight they put on real versus financial explanatory factors, on aggregate demand versus aggregate supply, and hence on the interpretation of the equilibrium real interest rate and inflation dynamics (Borio, 2017a). Proponents of the financial supercycle hypothesis claim that the secular stagnation hypothesis does not account for the buildup of bubbles and busts and the dramatic unfolding of an extremely severe financial crisis felt worldwide (Rogoff, 2015). But according to proponents of the secular stagnation hypothesis, although excessive debt buildups and deleveraging did contribute to the Global Financial Crisis, these factors cannot by themselves account for a long-term downward trend in the equilibrium real interest rate and persistently low inflation.

3.3 Studying financial crises over time

Studying the run-up to and aftermath of financial crises over time clarifies that financial crises exhibit specific characteristics. Firstly, credit growth is a good predictor of financial crises, as demonstrated by Schularick and Taylor (2009). After investigating the relation between money, credit, real economic activity and financial crises in advanced economies over a period of 140 years, they conclude that the financial system itself can create economic instability via an endogenous lending boom, and that financial crises can hence be seen as a credit boom gone wrong. The instability risks increase with the size of the financial sector. This implies that monetary policy should take credit and monetary aggregates into account, while aiming for financial and economic stability.

In addition, Reinhart and Rogoff (2009a) conclude that the run-up to as well as the recovery from banking crises in both advanced and emerging economies have much in common. Based on comparisons between the recent U.S. financial crisis and 18 historical

(22)

22 banking crises in advanced economies, they conclude that the behavior of U.S. equity and housing prices exhibits similar patterns over time. Also, growth momentum typically falls at the onset of a financial crisis, and stays relatively low for two years. Hence, the so-called inverted v-shaped output growth is characteristic to the run-up to financial crises, which also holds for other industrialized economies. Looking at differences between financial crises, U.S. public debt was slightly below the historical average during the recent financial crisis, as well as the pre-crisis inflation rate.

The findings on recoveries from banking crises are based on a slightly larger sample consisting of 22 systemic banking crises in both advanced and emerging economies. The aftermath of deep financial crises generally share three main characteristics. First, asset market collapses are deep and long lasting. On average, real housing prices drop by 35 percent over six years, while equity prices drop by 55 percent over only three and a half years. In general, asset prices tend to be more volatile than house prices. Secondly, the years following a severe banking crisis are associated with large drops in employment and output. Unemployment increases on average by seven percentage points over four years, coinciding with the downward phase of the financial cycle. The effect of systemic banking crises on unemployment is remarkable. Output drops on average by nine percent over two years. Hence, the duration of the decline in output is only half that of the rise in unemployment. Note that this measure is based on absolute changes in GDP. If measures such as output gaps in relation to potential GDP are being used, the duration is even longer. Thirdly, the real value of sovereign debt increases to extraordinary high levels, generally not due to bailouts and recapitalization, but due to the drop in tax revenues associated with output declines and countercyclical fiscal policy (Reinhart & Rogoff, 2009b).

Comparing a long enough time span of financial crises, both before and after World War II, reveals that the aftermath of financial crises can also be quite different. Schularick and Taylor (2009) analyze the development of monetary aggregates, credit and macroeconomic indicators in a sample consisting of 14 developed economies over a period from 1870 until 2008. Analyzing almost 140 years of financial crises uncovers interesting pre- and post-World War II differences between the aftermaths of financial crises. Likely, those different dynamics reflect differences in monetary and regulatory frameworks.

Before World War II, growth rates of credit and money declined significantly following financial crises, and did not reach their pre-crisis growth rates within five years after the onset of a financial crisis. However, after World War II, an effect of financial crises on growth rates of credit and money is almost imperceptible. This is probably due to central

(23)

23 bank policy supporting money growth, preventing the collapse of broad money and so keeping bank lending at relatively high levels. But, even though monetary policy seems to be more accommodating after World War II, and the consequences of financial crises post-World War II are more muted in absolute numbers relative to the preceding era, the real impact of financial crises remains comparable in magnitude relative to trend. This can be explained by the fact that, despite government incentives to soften nominal and real consequences of financial crises, the financial sector has been expanding and leverage has been increasing. By consequence, the real risks associated with financial turmoil gained in magnitude (Schularick & Taylor, 2009).

Another striking difference between the aftermath of financial crises pre- and post-World War II concerns growth and price developments. The slowdown in growth is most outspoken in the post-World War II period. However, financial crises before World War II are associated with disruptive deflation and stagnant monetary aggregates, while financial crises after World War II are associated with upward inflationary pressures relative to trend. This is related to a more active monetary policy response, substantiated by Schularick and Taylor (2009) by an increase in narrow money. The analysis suggests that the Fisherian debt-deflation mechanism observed in the pre-World War II period, implying increasing levels of real debt because of deflation, was avoided after World War II. Positive inflation, complemented by higher growth in monetary aggregates and less deleveraging relative to a non-crisis trend, is characteristic for financial crises during the second half of the twentieth century.

3.4 Financial versus non-financial crises

Comparing normal recessions or non-financial crises to recessions coinciding with a financial crisis reveals significant different dynamics. Recessions that coincide with a financial crisis are generally deeper and more damaging, and are often accompanied by major jumps in sovereign debt. Normal recessions not coinciding with a financial crisis typically last even less than a year. Multiyear recessions most often take place in economies that depend upon deep restructuring, for example within the financial system (Reinhart & Rogoff, 2009b).

More specifically, by analyzing 140 years of financial and non-financial crises in 14 advanced economies, Jordà, Schularick and Taylor (2011) find that post-World War II recessions surrounded by a financial crisis are about one-third as costly in terms of GDP as recessions not surrounded by a financial crisis. The relatively slow recovery of growth following a recession that coincides with a financial crisis can be partly accounted for by its

(24)

24 negative impact on loan growth. Inflation tends to decelerate during normal recessions, but even more so during recessions coinciding with a financial crisis. In euro area countries specifically, inflation turns out to be more sensitive to economic slack during deep and extended recessions than during mild recessions, due to higher willingness to cut prices and maintain current market share (Ciccarelli & Osbat, 2017). Regarding credit growth, the slowdown is about four times larger during recessions surrounded by financial crises than during normal recessions.

Lastly, global financial crises, meaning synchronized crises across a large number of countries, differ from financial crises in individual countries. Booms and busts in both growth and investment are larger in magnitude during global crises. Growth rates are higher in the run-up to global financial crises, and subsequently experience a larger drop at the onset of the turmoil. This result is mainly driven by the drop in growth rates during the Great Depression in 1930-1931. Individual financial crises are typically not preceded by higher growth rates. Regarding price developments, Jordà et al. (2011) claim that major global financial crises over the past 140 years tend to be preceded by inflation undershooting country averages. Again, this does not hold for individual financial crises. Both credit and money growth are significantly higher than country averages in the run-up to global and individual financial crises (Jordà et al., 2011).

(25)

25

4. Outlier analysis

The outlier analysis is an event study that focuses on inflation and its statistical properties throughout the Global Financial Crisis and historical precedents. Paragraph 4.1 describes the methodology and data. Paragraph 4.2 presents the data analysis, providing a graphical representation of inflation dynamics. The empirical analysis in paragraph 4.3 contains the regression model, results and robustness checks, and clarifies whether the change in inflation over the duration of a recession is significantly influenced by a financial crisis.

4.1 Methodology and data description

To determine whether inflation dynamics during the Global Financial Crisis are different, recent inflation dynamics are compared to a sample of similar historical events. However, severe financial crises are rare. Also, it would be wrong to conclude that all inflation dynamics during these episodes are the result of a financial crisis. Financial crises and recessions often coincide, and inflation dynamics are influenced during normal recessions not surrounded by financial crises too. Comparing recessions that do not coincide with a financial crisis to recessions that do, clarifies to what extent inflation dynamics can be attributed to a financial crisis. Moreover, analyzing inflation dynamics during recessions increases sample size significantly.

This outlier analysis focuses on a sample of 33 countries over a time span of 37 years.2

For the 23 advanced countries, all recessions between 1980 and 2017 are taken into account. This time span is mainly characterized by the Great Moderation for the sake of comparability between monetary regimes. For the 10 emerging countries, only recessions since 1990 are taken into account, since both GDP and inflation data for preceding years are missing. Note that the graphical data analysis focuses on advanced countries due to the lack of comparability with the generally volatile inflation dynamics in emerging countries. The empirical analysis focuses on both advanced and emerging countries, since the use of control variables enhances comparability.

The first event type included in the outlier analysis is a normal recession that does not coincide with a financial crisis. The dating of normal recessions, or non-financial crises, is based on quarterly GDP taken from the BIS. In certain cases, BIS data are replaced by OECD data if this extends the time series. Note that in these cases, the growth rates of quarterly GDP

2_{See Table 1 of Appendix 1 for an overview of all countries and the associated number of recessions in the sample. Note that the}

(26)

26 calculated from BIS data and OECD data almost entirely match. The first of two subsequent quarters of negative GDP growth indicates the start of a recession; the first of two subsequent quarters of positive GDP growth indicates the end of a recession. Some of these recessions coincide with a financial crisis. Those recessions constitute the second event type included in this event study.

The dating of financial crises is based on the Systemic Banking Crisis Database by Laeven and Valencia (2012). This dataset consists of systemic banking crises, sovereign debt crises and currency crises. A glance at this dataset learns that systemic banking crises are regularly complemented by sovereign debt crises and currency crises that occur within one or two years from the onset of the systemic banking crisis. To avoid treating several interrelated financial crises as separate events, only systemic banking crises are considered as financial crises in this event study.

According to the definition given by Laeven and Valencia (2012), a systemic banking crisis meets the following two conditions. First, there have to be significant signs of financial distress in the banking system, which can be bank runs, losses in the banking system, or the liquidation of a bank. Second, there have to be significant intervention measures as a reaction to losses in the banking system. The Systemic Banking Crisis Database provides detailed information on start and end dates of the systemic banking crises included in the outlier analysis. According to Laeven and Valencia (2012), all systemic banking crises that started in 2007-2008 are still ongoing in 2012, marking the last year covered by the database. In this outlier analysis, it is assumed that those financial crises ended in that year. All financial crises in the sample are complemented by a recession, except for the financial crisis in the U.S. in 1988.

The final panel dataset consists of event specific inflation rates, which are subject to indexation and calculations in the graphical data analysis and empirical analysis, respectively.3 Data on inflation is taken from the BIS, and refers to the monthly year-on-year change in the Consumer Price Index, with base year 2010.

The total sample includes 149 recessions, of which 116 took place in advanced countries and 33 in emerging countries. Of these 149 recessions, 41 coincide with a financial crisis, and 108 do not. Of the 41 financial crises in the sample, 31 took place in advanced countries and 10 in emerging countries. Of the 108 recessions not coinciding with a financial crisis, 85 took place in advanced countries, and 23 in emerging countries. 92 recessions took

(27)

27 place pre-2007, and 57 post-2007. Of the recessions that took place pre-2007, 17 coincide with a financial crisis, and 75 do not. Of the recessions that took place post-2007, 24 coincide with a financial crisis, being the Global Financial Crisis, and 33 do not.

4.2 Data analysis - A graphical representation 4.2.1 The Global Financial Crisis

Inflation dynamics during the Global Financial Crisis are presented in Figure 2. The dating of financial and non-financial crises is not yet taken into account. The yellow area indicates the maximum and minimum inflation rate in all euro area countries. Since the dataset contains monthly inflation, the euro area country with the highest or lowest inflation rate may differ from month to month.

Figure 2: Inflation dynamics during the Global Financial Crisis in the euro area, U.S. and U.K.

Figure 2 illustrates particular patterns of inflation during the Global Financial Crisis. A sharp drop in inflation takes place around 2009, followed by a slow recovery. Over the course of 2012 to 2014, inflation falls again. Since 2015 it has been picking up. This pattern characterizes the euro area, the U.S., and the U.K., although the length and severity of the drop in inflation varies.

4.2.2 Recession analysis

The following graphs represent inflation dynamics during recessions over time. Figures 3.1 to 3.7 include all recessions since 1980 in a particular country, and do not yet take the dating of financial crises into account. The figures are indexed, meaning that the lines represent the

-10 -5 0 5 10 15 20 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 C P I in flatio n

(28)

28 change in the inflation rate when compared to the inflation rate at the start of the recession (t=0). Hence, the y-axis represents an increase or decrease in inflation, or the change in the inflation rate in percentage points. The x-axis represents months relative to the start of the recession. t=0 indicates the start of any recession between 1980 and 2017. The orange pre-2007 band indicates the maximum and the minimum level of monthly inflation around any recession between 1980 and 2007 in the specific country. Inflation dynamics around the 2008-recession, characterized by the Global Financial Crisis, are highlighted in red. The solid line indicates the duration of the recession. Note that in all countries within this selected sample, the 2008-recession coincides with a financial crisis. The legend indicates how many recessions are in the sample: R6 means sixth recession, hence 5 recessions took place in the pre-2007 period.

(29)

29

Figures 3.1-3.7: Inflation dynamics during recessions (1980-2017) in selected advanced countries  indexed Figure 3.1 Figure 3.2 Figure 3.3 Figure 3.4 Figure 3.5 Figure 3.6 Figure 3.7 -7 -4 -1 2 -6 -2 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58

Switzerland

Max-Min pre-2007 R6 ('08-'09) -4 -2 0 2 -6 -2 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58

Japan

Max-Min pre-2007 R5 ('08-'10) -6 -4 -2 0 2 4 -6 -2 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58

Germany

Max-Min pre-2007 R6 ('08-'09) -16 -12 -8 -4 0 -6 -2 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58

U.S.

Max-Min pre-2007 R6 ('08-'09) -10 -7 -4 -1 2 -6 -2 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58

Sweden

Max-Min pre-2007 R3 ('08-'09) -12 -8 -4 0 4 -6 -2 2 6 ₁₀ ₁₄ ₁₈ ₂₂ ₂₆ ₃₀ ₃₄ ₃₈ ₄₂ ₄₆ ₅₀ ₅₄ ₅₈

U.K.

Max-Min pre-2007 R3 ('08-'09) -4 -2 0 2 -6 -2 2 6 10 14 18 22 26 30 34 38 42 46 50 54 58

Netherlands

R1 ('80-'82) R2 ('08-'09)

(30)

30 Figures 3.1 to 3.7 point out that the 2008-recession exhibits a particular pattern. Firstly, in all countries in the selected sample, inflation drops immediately or shortly after the onset of the recession, and recovers fairly slowly. In Germany, Sweden and the Netherlands, a ‘double-dip’ recession is visible. The observation that inflation in Japan and the U.K. sharply increases following the onset of the 2008-recession is striking. Inflation during the 2008-recession moves below the pre-2007 band in almost all selected countries, meaning that the inflation drop is extraordinarily large in comparison to preceding recessions. In all selected countries except Switzerland and Japan, inflation several years after the onset of the 2008-recession moves close towards the upper bound of the pre-2007 band. Inflation thus seems relatively high for a post-recession trend. However, inflation remains below its starting rate for the entire time span covered in the graphs.

Figure 4 presents inflation around the 2008-recession in selected advanced countries, relative to inflation around preceding recessions. The pre-2007 band indicates the maximum and minimum monthly inflation rate around any recession between 1980 and 2007 in the selected countries. The y-axis represents the inflation rate, and the x-axis represents months relative to the start of a recession.

Figure 4: Summary of inflation dynamics during recessions between 1980 and 2017 in selected advanced countries  non-indexed

Euro area inflation drops following the onset of the recession, and reaches the lower bound of the pre-2007 band. The starting rate of inflation at the onset of the 2008-recession is relatively low compared to earlier recessions. Inflation several years after the onset of the 2008-recession moves towards the middle of the pre-2007 band, which implies that inflation

-5 0 5 10 15 20 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 C P I in flatio n Months

Max-Min pre-2007 US R4 ('08-'09) Germany R6 ('08-'09)

Switzerland R6 ('08-'09) Sweden R3 ('08-'09) Japan R5 ('08-'10)

(31)

31 does not remain extraordinarily low, relative to preceding recessions. The observation that inflation during the 2008-recession does not remain at the lower bound of the pre-2007 band does not reflect declining trend inflation.

Figure 5 is the indexed version of Figure 4. Hence, the sample is the same, but the y-axis now represents the increase or decrease in inflation compared to its starting rate, or the change in the inflation rate in percentage points.

Figure 5: Summary of inflation dynamics during recessions between 1980 and 2017 in selected countries  indexed

Again, inflation several years after the onset of the 2008-recession is relatively high compared to earlier recessions. After a large inflation dip, being especially sharp in case of the U.S., inflation slowly moves back towards its starting rate. However, inflation during the 2008-recession remains below its starting rate throughout the entire time span covered in the graph.

4.2.3 Financial crisis analysis

Case studies highlight the development of inflation dynamics during financial and non-financial crises over time in more detail, as presented in Figures 6.1 to 6.4. t=0 again indicates the start of a recession. The y-axis represents the inflation rate, and the x-axis indicates the distance in months relative to the start of a recession.

-15 -10 -5 0 5 -6 -4 -2 0 2 4 6 8 ₁₀ ₁₂ ₁₄ ₁₆ ₁₈ ₂₀ ₂₂ ₂₄ ₂₆ ₂₈ ₃₀ ₃₂ ₃₄ ₃₆ ₃₈ ₄₀ ₄₂ ₄₄ ₄₆ ₄₈ ₅₀ ₅₂ ₅₄ ₅₆ ₅₈ ₆₀ C P I in flatio n in p er ce n tag e p o in ts Months

Max-Min pre-2007 US R4 ('08-'09) Germany R6 ('08-'09)

Switzerland R6 ('08-'09) Sweden R3 ('08-'09) Japan R5 ('08-'10)

(32)

32

Figures 6.1-6.4: Case studies of inflation dynamics during financial and non-financial crises since 1980  non-indexed

Figure 6.1 Figure 6.2

Figure 6.3 Figure 6.4

The figures point out that inflation dynamics during the 2008-recession, which was in Japan not complemented by a financial crisis, are similar. This highlights the global nature of the recent financial crisis. In case of the U.S. (Figure 6.1) and the U.K. (Figure 6.2), the 2008-recession, coinciding with the Global Financial Crisis, is associated with a sharp inflation drop. The effect of a financial crisis on inflation is also visible for the Japanese 1998-recession (Figure 6.4), which is relatively short-lived, but associated with a sharp inflation drop. In case

0 4 8 12 16 20 -6 -4 -2 0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32

U.K.

U.K. `80 Q1 (recession) U.K. `90 Q3 (recession)

U.K. `08 Q2 (recession + Sep '07 financial crisis) -3 0 3 6 9 12 15 -6 -4 -2 0 2 4 6 8 ₁₀ ₁₂ ₁₄ ₁₆ ₁₈ ₂₀

U.S.

U.S. `80 Q2 (recession) U.S. `81 Q4 (recession) U.S. `90 Q4 (recession)

U.S. `08 Q3 (recession + Dec '07 financial crisis)

-3 0 3 -6 -3 0 3 6 9 12 15 18 21 24 27 30 33

Japan

Japan `93 Q2 (recession)

Japan `98 Q2 (recession + Nov '97 financial crisis) Japan `01 Q2 (recession + Nov '97 financial crisis) Japan `07 Q2 (recession) Japan `08 Q2 (recession) Japan `10 Q4 (recession) Japan `14 Q2 (recession) -1 1 3 5 7 -6 -3 0 3 6 9 ₁₂ ₁₅ ₁₈ ₂₁ ₂₄ ₂₇ ₃₀ ₃₃ ₃₆ ₃₉

Germany

Germany `80 Q2 Germany `91 Q2 Germany `92 Q2 Germany `95 Q4 Germany `02 Q4

Germany `08 Q2 (recession + Sep '08 financial crisis) Germany `12 Q4 (recession + Sep '08 financial crisis)

Is the Global Financial Crisis an outlier? Analyzing inflation dynamics during financial and non-financial crises

Table of Contents

1. Introduction

2. Studying the inflation process

3. The financial crisis

4. Outlier analysis

Switzerland

Japan

Germany

U.S.

Sweden

U.K.

Netherlands

U.K.

U.S.

Japan

Germany