This paper aims to empirically test for Ireland the “granular hypothesis” (Gabaix 2011) which posits that firm-level productivity shocks can explain a sizable portion of aggregate productivity fluctuations.
The Irish case is particularly relevant as Ireland has been experiencing increasing economic concentration in recent years to the point that micro shocks to a few selected firms in 2015 led to significant level shifts in aggregate variables like GDP (+34 per cent) and, particularly, labour productivity (+23 per cent) and total factor productivity (-12 per cent).
Combining macro data from the CSO and the OECD with micro data from the Annual Business Survey of Economic Impact (ABSEI), the granular hypothesis is tested in Ireland for the period 2000-2016.
Research findings confirm that productivity shocks to the 5 largest firms (in terms of value added) in Ireland account for a large fraction (about one-third) of aggregate productivity growth. These empirical results shed light on the origins of Irish productivity fluctuations, the consequences of economic concentration on resilience and the importance of diversification policies aimed at broadening Ireland’s enterprise base of productive firms.
VIP Independent Call Girls in Andheri 🌹 9920725232 ( Call Me ) Mumbai Escorts...
What is behind aggregate productivity in Ireland? A granular approach
1. What is behind aggregate productivity growth in Ireland?
A granular approach
Dr Javier Papa
(Competitiveness & Productivity Unit)
6th March 2019
NERI Seminar Dublin
This research paper has been prepared by an IGEES senior economist working in the
Competitiveness and Productivity Unit of Ireland’s Department of Business, Enterprise and
Innovation (DBEI). The views presented here are those of the author and do not
necessarily represent the official views of this Department.
2. 2 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Outline
• Introduction and literature review
• Data and methods
• Empirical findings
• Conclusions and policy implications
3. 3 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Introduction (1)
• Mainstream view: fluctuations of macroeconomic variables (GDP, employment, productivity) are caused by
economy-wide or aggregate shocks (inflation, strikes, policy changes)
• Unlikely that uncorrelated individual shocks will have an aggregate impact as they would cancel out on average…
• However, recent evidence shows that economy-wide shocks (Cochrane 1994) are not as important as
industry-specific shocks (Acemoglu et al 2012) to explain aggregate fluctuations.
• Granular hypothesis (Gabaix 2011): since modern economies are dominated by large firms, then
idiosyncratic (productivity) shocks to these firms may not cancel out on average and thus lead to sizable
aggregate shocks on GDP, employment and productivity.
• In fact, the author showed that productivity shocks to the 100 largest firms in the US explain about one-third
of variations in GDP p/c and aggregate productivity growth between 1952 and 2008.
• The granular effects is expected to be larger in economies less diversified than the US…
4. 4 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Introduction (2)
• This granular hypothesis is particularly relevant for small open economies like Ireland, where large firms are
known for having a large impact on macroeconomic performance…
• Economic concentration (as measured by the HHI) in Ireland was the highest in 2011 compared to other
advanced economies (e.g. France, Germany, Italy, Japan, Denmark, Netherlands) and is growing over time…
• Firm-level research (Papa et al 2018) showed that top (10th decile) performing firms account for a large
share of productivity in manufacturing (70 per cent) and services (45 per cent) over the period 2006-2014.
• MNEs’ relocation of balance sheets in 2015 has allegedly led to the unprecedented GDP growth of 34 per
cent, labour productivity growth of 23 per cent and total factor productivity decline of 12 per cent.
5. 5 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Introduction (3)
• Importance of micro-to-macro links for the Irish economy…
• Main objective: test Gabaix’s granular hypothesis on Ireland and thus examine to what extent
(unexpected) productivity shocks to the largest firms have an impact on macroeconomic variables,
particularly on aggregate (total factor) productivity.
• Why productivity is important?
• Key factor of national competitiveness;
• Main driver of improvements in living standards;
• Engine of economic growth over the medium to longer term
6. 6 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Data and methods (1)
• Macroeconomic data:
• Real GDP, GDP per capita and GDP deflator from OECD.
• Total factor productivity (VA-based Solow residual) from CSO.
• Firm-level data:
• Output (sales), value added, employment, 2-digit sector from Annual Business Survey of Economic Impact ABSEI
• ABSEI main features:
• Survey of the client companies of Enterprise Ireland, IDA Ireland and Údarás na Gaeltachta.
• Coverage: 4,200 agency client firms employing ten or more employees in Ireland between 2000 and 2016.
• Enterprises surveyed over the period 2000-2014 represent about 75 per cent of total Irish exports.
• Unbalanced panel data, though most of client firms remain in the survey over time.
7. 7 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Data and methods (2)
• Data cleaning:
• Missing, negative and zero valued observations are removed from the micro data.
• Sample size reduced to 2,000 – 2,600 firms. Large enterprises not affected.
• Outliers treatment:
• Outliers with growth rates larger than 20 per cent (absolute terms) are capped, affecting 40 percent of top 100 firms
• Robustness check including outliers with up to 50 per cent growth rates.
• Other data adjustments:
• Following Gabaix’s approach, growth rates are demeaned by year and industry.
• Unlike Gabaix’s approach, financial firms are not excluded from the sample as they are mostly ‘fintech’ companies
• Unlike Gabaix’s approach, firm level labour productivity is calculated as VA per worker (rather than sales p/worker)
• Sales in Ireland are decoupled from GVA, GDP and TFP given the large share of (imported) intermediates into exports
8. 8 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Data and methods (3)
• Granular residual calculations:
• Impact of granular residual on aggregate TFP growth:
gTFP(t)
• Assessment of R2 statistics…
9. 9 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Empirical findings (1)
Descriptive statistics:
• Ireland’s aggregate productivity growth (Solow Residual)
has been falling from 2000 up until the 2008 crisis
• Then it bounces back to positive growth rate before
levelling off around near zero growth rates (on average)
• The volatility of productivity fluctuations has increased
substantially after the 2008 crisis
Note: Inflation is expressed as an index 2000=1; Real GDP is expressed in euros; and both change in GDP per capita and the
Solow Residual are expressed in rates of change.
Sources: CSO for TFP (Solow Residual), OECD for GDP and World Bank for Inflation
Figure 1. Key macroeconomic data (inflation, real GDP, GDP p/capita and TFP), 2000-2014
10. 10 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Empirical findings (2)
Descriptive statistics:
• Value added of the 100 largest firms account for 20 per
cent of GDP (on average) over the period 2000-2014.
• In terms of sales the proportion goes up to 50 per cent
• The results are very similar to the top 50 firms thus
highlights the substantial contributions that a few large
firms make to the Irish economy.
Figure 2. Value-added of top 100 & 50 firms as proportion of GDP, 2000-2014
Source: Author’s calculations based on ABSEI
11. 11 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Empirical findings (3)
Descriptive statistics:
• Declining trend of (mostly negative) productivity shocks to
the 100 and 5 largest firms between 2001 and 2008.
• Rebound effect in 2009 after the crisis…
• Marked negative shocks in 2012/13 as a results o the
‘patent cliff’ in the Pharma-Chemical sector.
• These shocks are also reflected in aggregate TFP (figure 1)
Figure 3. Granular residual fluctuations (micro productivity shocks), 2000-2014
Source: Author’s calculations based on ABSEI
-2.0%
-1.5%
-1.0%
-0.5%
0.0%
0.5%
1.0%
1.5%
2.0%
2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Granular Residual (top 5 firms) Granular Residual (top 100 firms)
12. 12 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Empirical findings (4)
Regression results:
• Productivity shocks to the top 100 (and even 5 top) largest
firms in Ireland have a significant and sizable impact on
aggregate TFP growth.
• Productivity shocks in previous periods do not seem to
have a lasting effect on TFP growth – no spillover effects…
• Explanatory power (R2) of granular residual is high for the
top 100 firms (24%) and higher for the top 5 firms (33%)
• Sectors: Pharma-Chemical; ICT and Food & drinks.
• The effect size in Ireland (3.6) is larger than the US (3.3)
Table 1. Explanatory power of the (productivity) granular residual
Dependent variable: TFP Solow Residual
Top 100 firms Top 5 firms
(1) (2) (3) (4)
GranularResidualUsingYearAndIndustryDemeaning_Contemporaneous 3.649**
3.685*
3.626**
4.123**
(1.527) (1.859) (1.303) (1.626)
GranularResidualUsingYearAndIndustryDemeaning_OnePeriodLag 0.401 0.817 -0.122 0.044
(1.479) (1.643) (1.300) (1.378)
GranularResidualUsingYearAndIndustryDemeaning_TwoPeriodsLag 1.928 1.431
(1.941) (1.703)
Constant 0.039*
0.058*
0.016*
0.022*
(0.021) (0.028) (0.008) (0.011)
Observations 13 12 13 12
R2
0.369 0.423 0.439 0.486
Adjusted R2
0.242 0.206 0.327 0.293
Residual Std. Error 0.024 (df = 10) 0.025 (df = 8) 0.023 (df = 10) 0.024 (df = 8)
F Statistic 2.919 (df = 2; 10) 1.952 (df = 3; 8) 3.911*
(df = 2; 10) 2.520 (df = 3; 8)
Note 1: Significance levels at: *
p<0.10 **
p<0.05 ***
p<0.01
Note 2: Standard errors in parentheses
Note 3: Period covered 2000-2014; results still robust for 2000-2016.
13. 13 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Empirical findings (5)
Limitations, robustness checks, further steps:
• Key limitation – small sample size of 15 year-observations
• Unlike the US study (55 obs), micro-data limitations also feature other countries like Germany and the UK (20 obs)
• Reassurance – contemporary granular effect (3.6) and adj. R2 (0.33) are similar to the US (3.3 and 0.33)
• The results are robust for the extended period 2015-2016 with the inclusion of a year dummy...
• Impact of non-agency clients...?
• The results are robust when more outliers (with demeaned growth rates of up to 50 percent) are allowed in.
• In fact, the explanatory power (R2) of the granular effect increased from 33% up to 43%.
• Further research should include longer time series with broader coverage and KBC investment data....
14. 14 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Conclusions & policy implications
• The empirical findings confirm the granular hypothesis that productivity shocks to very few large firms seem to
account for a large fraction (about one-third) of aggregate productivity fluctuations, at least for 2000-2016….
• Understanding the origins of aggregate fluctuations (and improving its forecasting) requires researchers / policy makers
to not focus exclusively on macro and exogenous shocks but on concrete micro shocks to large firms in key sectors.
• The granular effect seem to be larger for Ireland than for more diversified economies (US) due to high economic
concentration, which caused the increased volatility in productivity growth, particularly after the 2008 crisis…
• The NCC recommends broadening and diversifying Ireland’s enterprise and export base which is critical to ensuring
the Irish economy is resilient and adaptable to both large firm- and industry-specific shock
• Limited evidence of spillovers (diffusion effect) to the wider economy from lasting shocks to large firms
• This Department’s Future Jobs initiative focuses on actions to maximize the contribution of large multinational
enterprises to the Irish economy and increase the productivity of SMEs to bridge productivity gaps, among others.
15. 15 An Roinn Gnó, Fiontar agus Nuálaíochta | Department of Business, Enterprise and Innovation
Thank you
javier.papa@dbei.gov.ie