The Implications for Productivity of Financial Constraints: a Firm-Level Investigation of Italy
1. The Implications for
Productivity of Financial
Constraints: a Firm-Level
Investigation for Italy
Productivity Forum - OECD
July, 7 - 8 2016
2. 2
The issues at stake
The structural problems that the Italian economy faces date back to well
before the onset of the great financial crisis. These structural difficulties are
well summarized by the weak pattern of TFP over the last twenty years
The crisis has negatively impinged on the weak pattern of productivity
causing a further deterioration of it
Several explanations have been proposed. We focus on the role of access
to finance for firms and the their difficulties in attracting external funds
We investigate whether the crisis has amplified the severity of the financial
restraints and whether the sensitivity of productivity to financial constraints
has changed in the aftermath of the crisis
We focus on firm-level data investigating the implications of the use of
finance by firms for their own productivity developments
3. 3
Background
A large body of empirical and theoretical literature points to the existence of a
positive relationship between financial development and growth …and negative
impact of financial constraints on productivity
1. High levels of productivity are generally detected in firms characterized by a
high incidence of: a) innovative investment projects that often need a long
horizon to yield returns, b) intangible assets such as those pertaining to
R&D activities and c) human capital
2. If financial restraints bind and thus affect the investment decision and the
demand for the other inputs, then the input combination diverges from the
one that would prevail for an unconstrained firm.
3. Investment in intangible assets are more subject to financial constraints
due to a) their low value as collateral compared to standard tangible assets
and b) the higher uncertainty surrounding their expected returns
4. 4
A quick survey of the literature
We review the literature related to the following topics:
the relationship between the use of finance by firms and their productivity
the role of investment on intangibles
Misallocation and its impact on productivity
Disruption in the diffusion of knowledge and technology
Underlying questions:
How does financially constraints feed into the previous issues?
How did crisis interplay with the role of financial constraints?
5. 5
Empirical work on Italian manufacturing sector
Use of Bureau Van Dick data (Orbis and Amadeues) from 2005 to 2015
Preliminary analysis:
Cursory view of financial conditions of corporate sector
Construction of Financial Condition Index (FCI) as Pal & Ferrando (2010)
Regression analysis on the role of Financial Constraints
First, we use a direct approach to appraise the impact of financial constraints on
productivity levels. We augment the Olley and Pakes’ method as in Fernandes
(2007) and Ferrando-Ruggieri (2015): FCI is an “input” of the production function
Second, we use an indirect approach where - after estimating a production function
equation to derive productivity – we estimate an equation to assess the impact of
the indicator of financial constraints on productivity growth e.g. Fernandes (2007)
Moreover, we use a productivity equation at the sectoral level to investigate the
technology diffusion mechanism described earlier and role played by financial
constraints in disrupting diffusion; e.g. Dan Andrews et al. (2015)
6. 6
The rate of change of TFP in manufacturing -
comparing 4 measures: Istat, weighted mean firm’s TFP growth rate
under different hypotheses on financial constraints
TFP dynamics on different specifications (y-o-y percentage changes)
Source: MEF-DT elaborations on ORBIS and AMADEUS microdata as well as on ISTAT NA data from I.STAT.
Generally FCI Index contributes to weaken TFP dynamics.
-11.0%
-6.0%
-1.0%
4.0%
9.0%
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
TFP annual growth rates by ISTAT
TFP estimation without FCI index
TFP estimation with FCI index both in polynomial term both as production factor
TFP estimation with FCI Index only in polynomial term
7. 7
The dispersion across firms of the TFP rate of growth
TFP standard deviation on different specifications (y-o-y percentage
changes)
Source: MEF-DT elaborations on ORBIS and AMADEUS microdata.
TFP dispersion was very high before 2008. Instead, it diminished
during the economic crisis. The inclusion of FCI Index in TFP
estimation generally contributes to increase variability.
0.15
0.20
0.25
0.30
0.35
0.40
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
TFP estimation without FCI index
TFP estimation with FCI index both in polynomial term both as production factor
TFP estimation with FCI Index only in polynomial term
8. 8
The production function parameters with the augmented Olley-Pakes’
method (direct approach): the estimated effect of financial constraints
table
FCI neither in
production function
no in polynomial term
FCI only on
polynomial term
FCI both in
production
function and in
polynomial term
Coefficient of labour 0.673 0.702 0.702
Coefficient of capital (separate estimation on sub-sector sub-
samples)
0.114 0.161 0.119
Coefficient of capital (single estimation on whole sample) 0.101(.0017) 0.101 (.002) 0.105 (.002)
Coefficient of age (separate estimation on sub-sector sub-
samples)
0.000 0.001 0.000
Coefficient of age (single estimation on whole sample) -0.002(.0003) -0.002 (.0004) -0.001 (.0004)
Coefficient of FCI Index (separate estimation on sub-sector
sub-samples)
-0.385
Coefficient of FCI Index (single estimation on whole sample) -0.292 (.008)
Note: standard errors could not be provided for the coefficients of labor as well for the other coefficients obtained on a separate estimation on sub-
sector samples. They have been estimated separately on sub-samples defined on the basis of sector and size.
Source: MEF-DT elaborations on ORBIS and AMADEUS microdata as well as on ISTAT NA data from I.STAT.
9. 9
A simple counterfactual exercise (the impact on TFP of a generalized
10 per cent reduction of the degree of financial constraints)
TFP levels (2005=100) with FCI both as in polynomial term as in
production function: actual trend and trend with a 10% decrease of FCI.
Source: MEF-DT elaborations on ORBIS and AMADEUS microdata.
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
94
96
98
100
102
104
106
108
110
112
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
TFP estimation with FCI index both in polynomial term both as production factor
TFP estimation with FCI index both in polynomial term both as production factor (alternative hypothesis with a decrease of 10 per cent of the FCI- index)
Percentage cumulative gap (left axis)
10. 10
The productivity equation:
the effects of financial constraints on productivity
Dependent variable: logarithm of TFP in level under the hypothesis of FCI
only in the polynomial term
Coef.s (standard
errors in
brackets)
Lagged value of dependent variable 0.333
(0.013)
Lagged value of Financial Constraint Index (FCI) -0.754
(0.035)
Constant 4.150
(0.212)
Year effects Yes
Sector specific effects Yes
Size specific effects Yes
Arellano-Bond test for AR(1) in first differences -0.600
(0.552)
Note: The index is constructed as in Ferrando and Ruggieri (2015).
Source: MEF-DT elaborations on ORBIS and AMADEUS microdata.
11. 11
The productivity equation:
the effects of crisis on productivity
Note: The index is constructed as in Ferrando and Ruggieri (2015).
Source: MEF-DT elaborations on ORBIS and AMADEUS microdata.
Dependent variable: logarithm of TFP in level under the
hypothesis of FCI only in the polynomial term
Coef.s (standard errors in
brackets)
Lagged value of dependent variable 0.208
(0.015)
Lagged value of Financial Constraint Index (FCI) -0.155
(0.035)
Crisis (after 2008)#FCI -0.207
(0.038)
Constant 4.349
(0.241)
Year effects Yes
Sector specific effects Yes
Size specific effects Yes
Arellano-Bond test for AR(2) in first differences -1.390
(0.165)
12. 12
Disruptions in the diffusion of technology from the frontier
firms: comparing average TFP growth of top 5% performers with
that of the remaining firms
TFP dynamics of top 5% performer and other 95% (2007=100)
Source: MEF-DT elaborations on ORBIS and AMADEUS microdata.
TFP estimation without FCI index both in plynomial term both as
production factor
TFP estimation with FCI Index only in polynomial term
169.5
113.3
90
100
110
120
130
140
150
160
170
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Top 5 per cent Olter 95 per cent
151.8
113.9
90
100
110
120
130
140
150
160
170
2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
Top 5 per cent Olter 95 per cent
13. 13
The productivity equation:
the effect of the distance from the frontier (best 5%)
Note: The index is constructed as in Ferrando and Ruggieri (2015).
Source: MEF-DT elaborations on ORBIS and AMADEUS microdata.
Dependent variable: logarithm of TFP in level under the hypothesis
of FCI only in the polynomial term
Two Stage IV Panel
regression (standard errors
in brackets)
Lagged value of dependent variable 0.831(0.167)
Productivity gap (ECT) -0.804(0.116)
Lagged value of Financial Constraint Index (FCI) -0.102(0.035)
Growth of best 5% firms in each sector 0.298(0.041)
Constant 0.158(0.069)
Year effects Yes
Sector specific effects Yes
Size specific effects Yes
Arellano-Bond test for AR(2) in first differences 0.300(0.767)