Macrofinancial History and the New Business Cycle Facts Discussion - - PowerPoint PPT Presentation

Macrofinancial History and the New Business Cycle Facts

Discussion by: Marco E. Terrones Worshop on Estimating and Interpreting Financial Cycles.

The views expressed in this discussion are those of the author and do not necessarily represent those of the IMF or IMF policy.

Marco E. Terrones Macrofinancial History Sep 2016 1 / 15

Questions and Main Findings

Q1: What phenomenon has the advanced economies (AEs) experienced over the past 50 years?

◮ The AEs have experienced the so-called financial hockey stick –

private credit (particularly mortgages) has reached record levels.

Q2: Has the financial hockey stick been associated with changes in the cyclical characteristics of these economies?

◮ Duration and amplitude. Expansions associated with rapid credit

growth are stronger and more durable. Yet, the recessions that follow are deeper.

◮ Volatility and skewness. High credit is associated with less volatility

in real variables; yet, it is also associated with worse tail events and lower growth.

◮ Synchronization. Real (real and financial) variables are more

synchronized in high credit economies.

◮ International synchronization. Advanced economies have become

both more "financialized" and synchronized.

Marco E. Terrones Macrofinancial History Sep 2016 2 / 15

Discussion

Very interesting paper on an important and under-researched

• area. It builds on previous work of the authors.

◮ Introduces a new macrofinancial dataset (17 AEs, 1870-2013). ◮ Corroborates the presence of financial hockey stick in AEs, which

signals a shift to a new "credit era".

◮ Examines the implications of the financial hockey stick for the

Business Cycle (BC) and Financial Cycle (FC) of these economies.

Very important findings; however, the issues of causality and the drivers of "financialization" need further research. Moreover, the BC and FC should be clearly identified.

◮ The links between BC and FC have been documented by others

(i.e. Claessens, Kose, and Terrones (2009, 2012)).

◮ Using various methods to identify cycles and document their

moments can be confusing.

◮ While using long time-series has pros, it also has cons. Marco E. Terrones Macrofinancial History Sep 2016 3 / 15

Roadmap

The financial hockey stick. What is it? Causality? Triggers? The financial hockey stick and the business cycle characteristics in the AEs.

◮ Are the BC and FC features robust to the methodology utilized to

define these cycles?

◮ Pros and cons of using 150 years of annual data.

Conclusion

Marco E. Terrones Macrofinancial History Sep 2016 4 / 15

The Financial Hockey Stick. What Is It?

Stylized fact whereby private credit to GDP in the AEs has risen markedly over the past 50 years (Schularick and Taylor, 2012).

20 40 60 80 100 120 140

Percent of GDP

1 8 7 1 8 8 5 1 9 1 9 1 5 1 9 3 1 9 4 5 1 9 6 1 9 7 5 1 9 9 2 5 2 1 3

Date

Credit to GDP Money to GDP

(Mean, 1870-2013)

20 40 60 80 100 120 140

Percent of GDP

1 8 7 1 8 8 5 1 9 1 9 1 5 1 9 3 1 9 4 5 1 9 6 1 9 7 5 1 9 9 2 5 2 1 3

Date

Credit to GDP Money to GDP

(Median, 1870-2013)

Source: JST, 2016

• Fig. 1. The Financial Hockey Stick

Private credit to GDP has been used as a proxy for financial development.

Marco E. Terrones Macrofinancial History Sep 2016 5 / 15

The Financial Hockey Stick. Causality

Development literature:

◮ Fin. development is

correlated with economic activity (Goldsmith (1969)). Does financial development spur income or vice versa?

Growth literature:

◮ Fin. development is related

to a country’s growth (King and Levine, 1993; Levine, 2005). Does financial development cause growth

• r vice versa?

New literature:

◮ The financialization of the

AEs is negatively related to growth and volatility. Which way does causality go?

• Fig. 2. Fin. Development and Income

Marco E. Terrones Macrofinancial History Sep 2016 6 / 15

What Factors Could Trigger Strong Credit Expansion?

Productivity gains:

◮ Technological progress is

typically financed with external sources (i.e. credit). Rapid progress could lead to strong credit expansion.

Financial liberalization:

◮ Measures to eliminate

financial repression and develop the financial sector have frequently spurred strong credit growth.

Financial openness:

◮ A surge in capital inflows

• ften leads to a rapid

expansion of credit.

• Fig. 3. Triggers of Credit Booms

Mendoza and Terrones (2008, 2012)

Marco E. Terrones Macrofinancial History Sep 2016 7 / 15

Triggers of Credit Booms: Regression Analysis

Tab.1. Triggers of Credit Booms.

Marco E. Terrones Macrofinancial History Sep 2016 8 / 15

The Blade of the Financial Hockey Stick

What factors could have helped form the blade of the financial hockey stick? Globalization and financial sector reforms

100 200 300 400 500 600 700

Percent of GDP

1870 1885 1900 1915 1930 1945 1960 1975 1990 2005 2013

Date

Credit to GDP Trade Openness Financial Openness

(Mean, 1870-2013)

5 10 15 20 25

Financial Reform Index

20 40 60 80 100 120 140

Percent of GDP

1870 1885 1900 1915 1930 1945 1960 1975 1990 2005 2013

Date

Credit/GDP Fin.Reform Index

(Mean, 1870-2013)

Source: MT, 2008, JST, 2016

Fig 4. The Blade of The Financial Hockey Stick

Indeed, there is a large literature examining how financial globalization might have affected the BC of a country.

Marco E. Terrones Macrofinancial History Sep 2016 9 / 15

Methodologies to Identify Business/Financial Cycles

• Classical. This is a non-parametric method that identifies the

turning points using algorithms subject to some censoring rule. Method was pioneered by Bry and Boschan (1971).

◮ Very simple to understand; and ◮ Utilizes the original data; not affected by new incoming data.

• Detrending. A trend model is fitted to a given series and the cycle

is determined as a deviation from this trend. Various detrending models: log-difference; log-linear; log-quadratic; HP filter; band pass filter; etc.

◮ Cyclical component is especific to the detrending method (Canova

1998).

◮ Cyclical component has, in some cases, a unit root or long memory. Marco E. Terrones Macrofinancial History Sep 2016 10 / 15

Business/Financial Cycles Are Method Dependent

1 2 3 4 5

Index, log

1870 1885 1900 1915 1930 1945 1960 1975 1990 2005 2013

Date

(Log level and Trend)

Real GDP per capita

1 2 3 4 5

Index, log

1870 1885 1900 1915 1930 1945 1960 1975 1990 2005 2013

Date

(Log level and Trend)

Real Credit Per Capita

• 25-20-15-10 -5 0

5 10 15 20 25

Percent

1870 1885 1900 1915 1930 1945 1960 1975 1990 2005 2013

Date

(Cyclical Component)

• 25-20-15-10 -5 0

5 10 15 20 25

Percent

1870 1885 1900 1915 1930 1945 1960 1975 1990 2005 2013

Date

(Cyclical Component) Note: Orange=Log level, Grey=HP trend, Red=Growth rate, Blue=HP cycle

Fig 5. Real and Credit Cycles in the US

Marco E. Terrones Macrofinancial History Sep 2016 11 / 15

Cycle Features: Classical vs Detrending Methods

Dating of recessions/expansions (downturns/upturns) vary. Cyclical component is sensitive to the detrending model. Confusing to use these methods simultaneously.

1.5 2 2.5 3 3.5

Index, log

1870 1885 1900 1915 1930 1939

Date (US Trend Component)

• 20
• 10

10 20

Percent

1870 1885 1900 1915 1930 1939

Date

(US Cyclical Component)

Note: Blue=Lambda 1, Yellow=Lambda 2, Light Blue=Lambda 6.5, Orange=Lambda 100, Red=Log-diff

Fig 6. Detrending Method and Different Smoothing Parameters

Marco E. Terrones Macrofinancial History Sep 2016 12 / 15

The Financial Hockey Stick and BC and FC

Why have the moments of consumption not changed as much to reflect the rapid expansion of mortgages in AEs?

◮ Why has risk sharing not improved significantly? ◮ How does an improvement in household financing lead to changes

in the BC characteristics of an economy?

Why and how has the financial hockey stick changed the underlying exogenous shocks driving the BC?

◮ Are these effects on the mean or variance or both? ◮ What are the channels? Marco E. Terrones Macrofinancial History Sep 2016 13 / 15

Using 150 Years of Data Has Pros and Cons

Pros:

◮ A long history helps

study the occurrence of rare events –i.e. financial and macroeconomic disasters.

Cons:

◮ Non-economic

convulsive episodes (i.e. WWI and WWII) can change the data generating process.

◮ Countries change

substantially in 150 years.

6 7 8 9 10 11

PPP, log

1870 1885 1900 1915 1930 1945 1960 1975 1990 2005 2013

Date

Note: Red=US, Purple=UK, Magenta=Germany, Yellow=Japan, Blue=Rest

Fig 7. Advanced Economies: Real GDP per Capita

Marco E. Terrones Macrofinancial History Sep 2016 14 / 15

Conclusion

Interesting paper based on a new macrofinancial database (17 advanced economies, 1879-2013). The advanced economies have experienced the financial hockey

• stick. But, what explains this development?

The financial hockey stick is associated with changes in the business cycle . . .

◮ Lower growth, less volatility, but more negative skewness. ◮ More synchronized real variables with credit. ◮ More synchronized economies.

To fully embrace these findings– the issues of causality and the relationship between financialization, financial globalization and financial sector reform need to be settled.

Marco E. Terrones Macrofinancial History Sep 2016 15 / 15