Medium and long run prospects for UK growth in the aftermath of the - - PowerPoint PPT Presentation

Medium and long run prospects for UK growth in the aftermath of the financial crisis

Nicholas Oulton

Centre for Economic Performance London School of Economics

Email: n.oulton@lse.ac.uk

Centre for International Macroeconomics and Finance Conference “The Causes and Consequences of the Long UK Expansion: 1992 to 2007” 19th ‐ 20th September, Clare College Cambridge

The issue

Up till the end of the boom in 2008Q1, the UK’s productivity growth rate was excellent. But then during the Great Recession productivity (GDP per hour) fell by 4.5%. In 2013Q1 it was 5.1% lower than at the peak in 2008Q1. So how has the Great Recession affected our long‐run prospects? Will the growth rate be permanently lowered? Or will we eventually get back to something like the growth rate of 1990‐ 2007?

2

Hypothetical paths for GDP per hour during recession and recovery

3

GDP per hour (log scale) time GDP per hour trend

Optimistic

GDP per hour (log scale) time GDP per hour trend

Pessimistic

GDP per hour (log scale) time GDP per hour trend

Very pessimistic

OUTLINE

• UK growth in the run‐up to the crisis

1. UK performance 1970‐2007 2. Projected growth of productivity using only pre‐crisis data

• What has happened since 2007?

1. The UK productivity puzzle 2. Explanations

• Will UK growth in the aftermath of the crisis be faster
• r slower?

1. Long run effects of a financial crisis: theory 2. Long run effects of a financial crisis: empirics

4

OUTLINE

• UK growth in the run‐up to the crisis

1. Performance 1970‐2007 2. Projected growth of productivity using only pre‐crisis data

[This section draws heavily on N. Oulton (2012). “Long term implications of the ICT revolution: applying the lessons of growth accounting and growth theory”. Economic Modelling, vol. 29, 1722‐1736, 2012.]

• What has happened since 2007?
• Will UK growth in the aftermath of the crisis be

faster or slower?

5

UK productivity growth was poor 1970‐1990 …

6

Source: EU KLEMS, November 2009 release (www.euklems.net)

… but improved greatly over 1990‐2007

7

Source: EU KLEMS, November 2009 release (www.euklems.net)

A growth‐model‐based approach to projecting UK productivity growth

• One‐sector model (textbook neoclassical)
• Two‐sector model: necessary since we need to

take account of the ICT revolution

8

ICT (computers, software & comms.): a unique product group

9

One‐sector model Long run growth rate:

1

[ ] , 1 Y BK hH

 





  

ˆ , 1 ˆ / , (TFPgrowth), :growth of skill ( ), 1- : labour share

h h

g y y Y H B g h    

 

   

10

PARADOX

If a country does not produce ICT goods and services, then it gains nothing from the ICT revolution! Or does it?

11

Two‐sector model

• Two sectors
• 1. Non‐ICT: produces consumption goods and

non‐ICT capital goods

• 2. ICT: produces ICT capital goods
• Productions functions are identical in the two

sectors, except for TFP which grows faster in ICT

• This implies:

Rate of decline of relative price of ICT goods = TFP growth in ICT minus TFP growth in non‐ICT

12

Long run growth in a two‐sector model (no ICT output)

↗ ICT use effect

Growth of GDP per hour Growth of TFP in non-ICT sector Growth of skill Labour share ICT income share Rate of decline of ICT relative price Labour share plus plus  

13

Long run growth, with some ICT output

↗ ICT output effect

Growth of GDP per hour Growth of (non-ICT) TFP Growth of skill Labour share ICT share Rate of decline of ICT relative price Labour share ICT share Rate of decline of ICT relative plus income plus plus

• utput

   price

14

Difference between TFP growth rates in ICT and non‐ICT sectors

15

Projected growth rates of GDP per hour for the UK market sector derived from one‐sector and two‐sector models (parameters values calibrated on pre‐crisis data from EU KLEMS)

16 Description Parameter Method Value TFP growth rate (whole economy) Mean, 1990‐2007 0.0114 TFP growth rate (consumption sector) Mean, 1990‐2007 0.0087 ICT income share HP trend 0.0641 ICT output share HP trend 0.0183 Labour share Mean, 1990‐2007 0.7301 Rate of decline of ICT relative price HP trend ‐0.0590 Growth rate of skill Mean, 1990‐2007 0.0057 Projected growth rates of GDP per hour One‐sector model 0.0213 Two‐sector model 0.0261 ln B  ln

C

B 

ICT

K

v

ICT

w

L

v ln p 

h

g ln y  ln y 

Bottom line Actual growth of GDP per hour in the market sector 1990‐2007: 2.87% p.a. Projected growth rates after 2007: One‐sector model: 2.13% p.a. Two‐sector model: 2.61% p.a.

17

OUTLINE

• UK growth in the run‐up to the crisis

1. UK performance 1970‐2007 2. Projected growth of productivity using only pre‐crisis data

• What has happened since 2007?

1. The UK productivity puzzle 2. Explanations

• Will UK growth in the aftermath of the crisis be faster
• r slower?

1. Long run effects of a financial crisis: theory 2. Long run effects of a financial crisis: empirics

18

Hypothetical paths for GDP per hour during recession and recovery

19

GDP per hour (log scale) time GDP per hour trend

Optimistic

GDP per hour (log scale) time GDP per hour trend

Pessimistic

GDP per hour (log scale) time GDP per hour trend

Very pessimistic

GDP has yet to regain its previous peak in 2008Q1

20

GDP per hour has fallen since the recession began and is still below its previous peak

21

But hours worked have surpassed their previous peak …

22

… and the same is true of jobs

23

Explanations

1. Distortion due to hard‐to‐measure or otherwise problematic sectors 2. Reallocation of labour to sectors where productivity is lower 3. Mis‐measurement of GDP due to mis‐measurement of banking output 4. Overheating in the boom 5. Lower physical capital input 6. Lower human capital (skill) 7. The impact of austerity 8. Crippled banks and zombie firms 9. Labour hoarding

24

Productivity growth in the market economy (exc. Government services)

25

Productivity growth looks similar if we also exclude “problematic” sectors

26

• 2. Reallocation of labour to sectors where productivity is lower?

Calculate GDP per hour assuming labour shares were the same as in 2007. Result: virtually no difference.

27

The productivity puzzle is widespread …

28

29

30

31

• 3. Mis‐measurement of banking output during the boom?

“Bankers sold toxic rubbish during the boom. So banking output was overstated during the boom. Therefore UK performance was not as good as we thought. And this helps to explain the productivity puzzle since GDP has not fallen as much as the

• fficial figures say.”

WRONG: The growth of real GDP is measured from the expenditure side [GDP(E)]. And most of the “toxic” part of banking is not part of final expenditure. So banking output may have been mis‐measured but this does not imply that GDP was mis‐measured (Oulton, NIER, 2013)

32

• 5. Lower physical capital per hour worked?

Business investment fell substantially during the Great Recession and has not recovered. But this does not necessarily imply lower capital intensity (capital per hour worked). A Perpetual Inventory Model using business investment shows that capital per hour worked is now higher than at the peak (4‐ 7% higher in 2012Q4 than in 2007Q4).

33

• 7. Austerity?

“Savage cuts to government spending have crippled the UK economy, just when a Keynesian fiscal stimulus was needed”. BUT: Interest rates have been very low (0.5% since March 2009). The banks were bailed out. The exchange rate was 23% lower in 2013Q1 than at its peak in 2007Q1. QE has resulted in the BoE owning 29% of the stock of gilts.

34

Austerity in practice

35

90 95 100 105 2 7 Q 1 2 7 Q 3 2 8 Q 1 2 8 Q 3 2 9 Q 1 2 9 Q 3 2 1 Q 1 2 1 Q 3 2 1 1 Q 1 2 1 1 Q 3 2 1 2 Q 1 2 1 2 Q 3 2 1 3 Q 1

GDP Source: ONS, Quarterly National Accounts, June 2013 and own calculations. Note: Seasonally adjusted. Grey bar marks Great Recession (2008Q2-2009Q3). 2008Q1=100

Government expenditure on goods and services and GDP

Austerity in practice

36

90 95 100 105 2 7 Q 1 2 7 Q 3 2 8 Q 1 2 8 Q 3 2 9 Q 1 2 9 Q 3 2 1 Q 1 2 1 Q 3 2 1 1 Q 1 2 1 1 Q 3 2 1 2 Q 1 2 1 2 Q 3 2 1 3 Q 1

• Gov. exp. (official)

GDP Source: ONS, Quarterly National Accounts, June 2013 and own calculations. Note: Seasonally adjusted. Grey bar marks Great Recession (2008Q2-2009Q3). 2008Q1=100

Government expenditure on goods and services and GDP

Austerity in practice

37

90 95 100 105 110 2007Q1 2007Q3 2008Q1 2008Q3 2009Q1 2009Q3 2010Q1 2010Q3 2011Q1 2011Q3 2012Q1 2012Q3 2013Q1

• Gov. exp. (official)
• Gov. exp. (K1)
• Gov. exp. (K2)

GDP Source: ONS, Quarterly National Accounts, June 2013 and own calculations. Note: Seasonally adjusted. Grey bar marks Great Recession (2008Q2-2009Q3). K1: deflated by GDP deflator; K2: deflated by AWE 2008Q1=100

Government expenditure on goods and services and GDP

• 8. Crippled banks and zombie firms

Crippled banks may cut lending to profitable

• projects. If so, we have already accounted for

this effect in looking at investment and capital. But crippled banks may also exercise forbearance and so create zombie firms. This requires micro‐data analysis to assess ‐‐‐ results not yet available.

38

Labour hoarding Labour hoarding is very likely during the recession because of hiring and firing costs. But after the recession is over, why should firms keep on staff who are surplus to requirements? A possible answer: capital utilisation falls due to fall in demand and real wages fall (labour market flexibility), so employment is maintained but utilised capital per worker falls.

39

Hypothetical reaction of a retail chain to a fall in sales

40

Shops 1 2 3 4 5 Total Pre‐crisis Sales 10 10 10 10 10 50 Employees 10 10 10 10 10 50

Assume that the breakeven level of sales is 9 for each shop. Then each shop is profitable pre‐crisis.

Hypothetical reaction of a retail chain to a fall in sales

41

Shops 1 2 3 4 5 Total Pre‐crisis Sales 10 10 10 10 10 50 Employees 10 10 10 10 10 50 Crisis Sales 8 8 8 8 8 40 Employees 10 10 10 10 10 50

Sales fall by 20%. Assume that the breakeven level of sales is 9 for each shop. Then each shop is now making a loss, so:

Hypothetical reaction of a retail chain to a fall in sales

42

Shops 1 2 3 4 5 Total Pre‐crisis Sales 10 10 10 10 10 50 Employees 10 10 10 10 10 50 Crisis Sales 8 8 8 8 8 40 Employees 10 10 10 10 10 50 After firm’s reaction Sales 10 10 10 10 40 Employees 11 11 11 11 44

The firm closes one shop (shop 5). Because of lower wages, 4 out of the 10 workers in the closed shop are reallocated to the other 4 shops. Utilised K/L and Y/L now lower.

OUTLINE

• UK growth in the run‐up to the crisis

1. Performance 1970‐2007 2. Projected growth of productivity using only pre‐crisis data

• What has happened since 2007?

1. The UK productivity puzzle 2. Explanations

• Will UK growth in the aftermath of the crisis be slower?

1. Long run effects of a financial crisis: theory 2. Long run effects of a financial crisis: empirics

43

This section draws heavily on:

• N. Oulton and M. Sebastiá‐Barriel (2013). “Long and short‐term

effects of the financial crisis on labour productivity, capital and

• utput”. Bank of England Working Paper no. 470.

[http://www.bankofengland.co.uk/publications/Documents/workingpapers/wp470.pdf]

44

Effect of financial crises on GDP: empirics

1. High debt‐GDP and growth Reinhart and Rogoff (2010) find that based on data for 44 countries spanning about 200 years, GDP growth rates fall as the gross central government debt‐GDP ratio rises. The growth effects are similar in advanced and emerging economies. Reinhart and Rogoff (2012) argue that the negative association between debt‐GDP ratios and growth cannot be entirely due to cyclical effects (recessions causing high debt) since low growth is highly persistent in highly‐indebted countries (so high debt is causing low growth). My interpretation takes into account the critique of the published results by Herndon et al. (2013) and the subsequent response by Reinhart and Rogoff (2013).

• 2. Post WWII banking crises

Cerra and Saxena (2008); Furceri and Mourougane (2009); IMF (2009), Barrell et al. (2010), Papell and Prodan (2011) : These studies find large effects of banking crises on GDP but do not distinguish clearly between short run and long run effects; recovery is slow. They also claim that advanced countries suffer less than developing ones. 3. 1870‐2008 Jordà et al. (2012) : based on a study of nearly 200 recession episodes in 14 advanced countries between 1870 and 2008, they find that more credit‐intensive booms tend to be followed by deeper recessions and slower recoveries.

45

Effect of financial crises on level of GDP per hour: theory

1. Interest rates were very low in the boom, reflecting a mispricing of risk. If these are higher post‐recovery, then K/L will be lower and consequently Y/L will be lower too. 2. In short/medium run higher unemployment causes loss of human

• capital. This effect disappears in the very long run (through death,

retirement or emigration). But if new entrants to the labour market are particularly affected by unemployment, the effect could last 50 years. 3. The recession probably reduces innovative activity of all types (entry of new firms, mobility of workers between jobs, experimentation with new business models, also R&D etc). This lowers TFP level: due to cumulative nature of knowledge and inelastic supply of innovators, this loss is

• permanent. If TFP lower, then desired K/L will fall too.

4. The financial crisis will lead to a rise in the debt/GDP ratio which requires efficiency‐reducing taxes to finance and/or reductions in productivity‐improving public expenditure (Reinhart and Rogoff, 2010 and Reinhart et al. 2012).

46

Model

47 1 1 2 1

(1) ( ) (1 ) , 1, 0 1, 1, : actuallog labour productivity

• f -th country in year

: growth of labour productivity ( ) : long

it it it it it it it it it it it it

q q q q q crisis q level i t q q q q           

     

                     run log labour productivity

• f -th country in year

: Reinhart-Rogoff crisis dummy for -th country in year : measures effect of crisis on productivity

it

level i t crisis i t short run growth 

Model, cont.

48 1

(2) , 0; : zero-mean error [Maintained hypothesis: no long run effect of crisis on ] : country fixed effect : year effect : me

it it i t it it it i t

q q a a crisis growth a a     

  

      asures effect of crisis on productivity long run level

Regression equation

49 1 1 2 3 1 4 1 5 2 6 3

Combining (1) and (2) leads to the regression equation: , : year dummies; : fixed effects from which the long-run level ef

T it i u t u it it it it it it u t i

q D crisis crisis q q q D         

      

           



2 3 4 5

fect of a crisis, measured by , can be backed out: 3( ) 3 2          

Data

• Labour productivity (GDP per worker) from The Conference Board’s Total

Economy Database (TED) for 2011.

• Crisis dummies (0/1) from website of Reinhart and Rogoff’s (RR) book, This

Time Is Different (2009).

• Merging of TED and RR yields annual data for 61 countries over 1950‐2010.

Includes OECD, Asian, Latin American and African countries.

• Capital estimated by PIM using underlying investment series from the PWT.

We use mainly the RR banking crisis dummies: Banking crisis defined by RR as “(1) bank runs that lead to the closure, merging or takeover by the public sector of one or more financial institutions and (2) if there are no runs, the closure, merging, takeover or large‐scale government assistance of an important financial institution (or group of institutions) that marks the start of a string of similar outcomes for other financial institutions.”

50

The Reinhart‐Rogoff Database

51

Crisis type 1950‐1979 1980‐2010 Currency 12.6 22.1 Inflation 10.3 17.8 Stock market 19.2 21.4 Domestic debt 1.0 3.1 External debt 8.4 15.4 Banking 0.9 19.8 Proportion of country‐years spent in financial crisis (61 countries), %

Regression results: estimated values of θ for banking crisis (61 countries, 1950‐2010)

(1)

Variable affected by crisis All countries and years GDP per worker 100 x θ ‐1.096*** (s.e.) (0.356) 52 ***: significant at 1% ; ** significant at 5%

Regression results: estimated values of θ for banking crisis (61 countries, 1950‐2010)

(1)

Variable affected by crisis All countries and years GDP per worker 100 x θ ‐1.096*** (s.e.) (0.356) Capital per worker 100 x θ ‐1.137*** (s.e.) (0.411) TFP 100 x θ ‐0.813** (s.e.) (0.340) GDP per capita 100 x θ ‐1.794*** (s.e.) (0.372) 53 ***: significant at 1% ; ** significant at 5%

Regression results: estimated values of θ for banking crisis (61 countries, 1950‐2010)

(1) (2) (3) (4) (5) (6) (7) (8)

Variable affected by crisis All countries and years

• Exc. Great

Recession

• Exc. Asia
• Exc. Latin

America Developed countries

• nly

Developing countries

• nly

1950‐79 1980‐2010 GDP per worker 100 x θ ‐1.096*** ‐1.005*** ‐1.112*** ‐0.550 0.362 ‐1.258*** 0.831 ‐1.231*** (s.e.) (0.356) (0.380) (0.417) (0.382) (0.278) (0.467) (1.954) (0.382) 54 ***: significant at 1% ; ** significant at 5%

Regression results: estimated values of θ for banking crisis (61 countries, 1950‐2010)

(1) (2) (3) (4) (5) (6) (7) (8)

Variable affected by crisis All countries and years

• Exc. Great

Recession

• Exc. Asia
• Exc. Latin

America Developed countries

• nly

Developing countries

• nly

1950‐79 1980‐2010 GDP per worker 100 x θ ‐1.096*** ‐1.005*** ‐1.112*** ‐0.550 0.362 ‐1.258*** 0.831 ‐1.231*** (s.e.) (0.356) (0.380) (0.417) (0.382) (0.278) (0.467) (1.954) (0.382) Capital per worker 100 x θ ‐1.137*** ‐1.419*** ‐0.997** ‐0.755* ‐0.0423 ‐1.677*** ‐0.832 ‐1.229*** (s.e.) (0.411) (0.462) (0.405) (0.392) (0.379) (0.397) (1.122) (0.418) TFP 100 x θ ‐0.813** ‐0.718** ‐0.807** ‐0.373 0.394 ‐0.864** 1.019 ‐0.854** (s.e.) (0.340) (0.341) (0.379) (0.329) (0.288) (0.437) (1.916) (0.358) GDP per capita 100 x θ ‐1.794*** ‐1.574*** ‐1.849*** ‐1.444*** ‐0.789** ‐1.773*** 0.390 ‐1.875*** (s.e.) (0.372) (0.386) (0.406) (0.396) (0.386) (0.495) (2.170) (0.410) 55 ***: significant at 1% ; ** significant at 5%

Do banking crises reduce the long run growth rate?

The cross‐country panel analysis assumes that the long run growth rate is not affected (the “pessimistic” case is a maintained hypothesis). How plausible is this hypothesis? Two case studies suggest opposite conclusions: 1. The US after the Great Depression of the 1930s

The growth rate returned to its previous trend. Long run loss of level: 17% (calculation based on Perron (1989)).

2. Japan after the bubble burst in 1990

No fall in GDP level (annual basis) after the bubble but growth of GDP per hour and of TFP slowed a lot. Japan has used extreme Keynesian fiscal policy: the debt‐GDP ratio reached 226% in 2010.

56

57

100 150 200 250 300 1970 1975 1980 1985 1990 1995 2000 2005 2010 GDP per head GDP per worker GDP per hour Source: The Conference Board, Total Economy Database (http://www.conference-board.org/data/economydatabase) log scale, 1970=100

Japan: GDP per head, GDP per worker and GDP per hour

Conclusions

• Assuming the UK’s banking crisis lasts 5 years, the

long run hit to the level of GDP per worker is likely to be about 4‐5% and the long run hit to the level of GDP per capita will be about 7‐9%.

• When recovery is complete, GDP per hour in the

UK will grow at about the rate projected before the crisis: 2.61% p.a. in the market sector.

• These conclusions are predicated on the UK

following reasonable good policies in other respects, in particular stopping the debt‐GDP ratio from rising too high.

58