Professional Documents
Culture Documents
The author thanks Christian Bjrnskov, Carlos Daz Vela, Kevin Grier, Ignacio Moral, Santiago Lago, Martin Rode and
Blanca Snchez-Robles for useful comments. He is also indebted to Vikram Nehru, from the World Bank, for data
supplying. Earlier versions of this article were presented at the 2010 Conference of the Public Choice Society held in
Monterey, CA, 11-14 March, and at the 2010 Conference of the European Public Choice Society held in Izmir, Turkey,
8-11 April.
1
Universidad de Cantabria, Departamento de Economa. Avenida de los Castros s,n., 39005 Santander (Spain).
(colls@unican.es), and Universidad Internacional Menndez Pelayo. Isaac Peral 23, 28040 Madrid (Spain).
(scoll@uimp.es).
2
Pages: 1
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
1. Introduction
The rapid increase in the size of the public sector has been one of the most salient
economic features of 20th century. According to Tanzi and Schuknecht (2000:6-7), public
expenditures as a share of GDP more than tripled during the century in a representative
sample of industrialized countries. The most rapid increase has corresponded to the
heading transfers and subsidies, indicative of the re-distributive action of government,
whose share in GDP has increased from about 1 per cent towards the end of 19 th century
to about 21 per cent one century later, according to the same source. As a consequence,
the public sector has increasingly become a re-distributor of wealth created by the
private sector of the economy.
From this it follows that the growth of the public sector should eventually come to a halt,
imposed by the limited extent of the privately created wealth available for redistribution. Prior to meet that limit, the distorsionary effects of taxes and other sources
of public revenue on resource allocation should likely lead to a reduction of the social
product, as compared to its potential value. This reflection has led to the notion of an
optimum size of the public sector from the point of view of economic growth, an idea
that was formulated by a number of authors towards the end of the 20 th century.
According to Armey (1995), the relationship between GDP and the government share
follows the pattern of an inverted U. About the same time, Barro, Rahn and Scully put
forward their hypotheses about another inverted-U pattern in the relationship between
the rate of growth of GDP and the government share.
These patterns would result from the interplay of two opposing forces. On the one hand,
public expenditure on those goods which are subject to market failure (i.e., on classical
public services) contributes to the generation of private output. This beneficial effect is,
however, decreasing on the margin due either to diminishing marginal returns of
classical public services, or to displacement of public expenditure to other, not so
beneficial targets. In turn, the public revenues necessary for the backing of public
expenditures have an opportunity cost in terms of foregone private production and
investment. As public revenues increase, the forgone private investment [consumption]
includes more and more profitable [valuable] allocations, with the result that the
opportunity cost of public expenditure increases on the margin. When we detract a cost
that increases more than proportionally with the size of the public sector from an output
that increases less than proportionally with that size, the resulting net effect should
follow the inverted-U pattern referred to above. Note, however, that the current practice
in national accountancy, consisting of valuing the product of the public sector through
its cost, tends to more or less mask this pattern.
The hypotheses put forward by Armey, Barro, Rahn and Scully have been submitted to
test many times. In the beginning, the most widely used technique was regression
analysis of cross-sections of countries. Nonetheless, as indicated in the next section, the
government share is but one determinant of GDP or of its rate of growth. As there are
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 2
many other determinants whose behavior changes from country to country, crosssectional analysis needs to control for too many variables. By contrast, to the extent that
some of these determinants tend to remain more or less stable within each country, time
series analysis seems to provide, in principle, a more suitable approach. On the other
hand, the relationship between GDP (or its rate of growth) and the size of the public
sector doesnt remain unidirectional in the course of time, a fact that poses a difficulty to
this type of analysis. In more recent time panel-data analysis has gained partisans, in the
belief that it reduces the seriousness of the problems affecting the other two
approaches. This belief is not without ground, so that the present state of the art heavily
rests on the results of panel-data analyses.
Be it as it may, this contribution follows the time-series approach, in an attempt to
overcome the endogeneity issue by means of the formulation of a simultaneous equation
model. Yet, this approach, in combination with the nature of the data, allows to
estimating only the short-term effect of the government share on GDP. Since this effect is
necessarily less visible than the long-term effect, our test is rather strict. The model
devised here is applied in succession to 24 advanced economies during the period that
runs from 1975 to 2007. In about half those countries the results strongly suggest that
the net growth effect of the increase in public expenditure is indeed negative. In the rest
of cases the evidence is less compelling, although it points in the same direction. All in
all, it can be said that our results are in line with those of prior time series analysis, as
well as with the mainstream of all other empirical approaches.
The remaining of this contribution is organized as follows. Section 1 reviews the
literature. Section 2 develops the model. In section 3 the benchmark estimates are
presented. Section 4 extends the model to include more variables, and checks the
robustness of the estimates obtained for each country. Section 5 concludes.
Review articles were published by Temple (1999), Tanzi and Shuknecht (2000), Poot (2000), Easterly and Levine
(2001), Leach (2003), and Nijkamp and Poot (2004). More recently, Facchini and Melki (2012) have offered a good
deal of references to individual titles. Berg and Henrekson (2011) focus on recent articles only.
3
Pages: 3
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
government share, that has an effect on GDP growth and presents different values in
different countries causes the exact location of the Armey or Scully curves to differ on
the plane among countries. Cross-sectional analysis, however, estimates one single curve
for all of them. At best, this oversimplifies the results, and, in the worst scenario, it may
bias them. For this reason, some authors have expressed serious doubts that this type of
analysis cannot help to clarifying the point at issue. This is, for instance, the position
taken by Slemrod (1998:102), and this authors contention is shared by Arcand and
Dagenais (1995) and by Mueller (2003:550). As indicated by the latter, and illustrated in
figure 1, the absence of a common curve for the countries included in the cross section
could possibly lead us to conclude that the relationship between income and public
expenditure is positive where it is in fact negative in every country, or the opposite.
Case b.
Y [grY]
Y [grY]
G/Y
G/Y
Note: The dotted lines indicate the functions of Y [grY] to G/Y as estimated via cross sectional analysis.
The list is available on request, and includes as well those panel-data analyses which use very few time periods.
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 4
positive ones. The corresponding proportions for OECD or industrial countries alone are
74.4% and 25.6% respectively. This small difference among the two samples suggests
that the customary assumption according to which a small public sector amounts to
provision of genuine public goods only is perhaps violated in practice. The proportions
dont show significant changes when we restrict our examination to the results of the
best practice contributions, defining this concept as the use of the production-function
approach. However, the proportion of negative coefficients increases when one looks at
the significant ones only. When doing this, the share of coefficients exhibiting minus
signs increases to 88.6% for the whole sample, and to 93.8% when we consider
developed countries only. But the fact remains that 11.4% of significant results in the
former case and 6.2% in the latter exhibit the wrong sign according to the hypotheses
submitted to test. Moreover, less than one estimate out of three (27.3%) succeeds in
yielding both significant and negative coefficients for the variable of interest in the
whole sample, and the corresponding figure for the subsample of developed countries is
just 36.6%.
Nijkamp and Poot (2004), and Facchini and Melki (2012) have analyzed comparable
samples of contributions, although they dont distinguish between cross-sectional and
other types of analyses. Their distributions of negative and positive coefficients, as well
as of significant and insignificant ones are, however, similar to those reported above.
This similarity is higher in the case of Facchini and Melki (2012), whose criteria to
classify results as favorable, unfavorable or ambiguous is closer to mine. Given the
problems that surround cross-sectional analysis on this topic, a possible conclusion to
be drawn is that the positive signs that are sometimes obtained correspond to defective
specifications or data samples. However, other interpretations are also legitimate.
Since one of the potential sources of divergence is the set of control variables included in
each estimation attempt, it is interesting to check how much influence these differences
may exert. This idea has led to the use of robustness tests consisting of systematically
altering the set of control variables until exhausting their possible combinations. The
robustness test currently applied to the relationship between the government share and
economic performance is the so-called extreme bound analysis (EBA)5. In the earliest and
most cited application of this test to the proposed determinants of economic growth,
Levine and Renelt (1992) discarded as non-robust all the proposed ones with the sole
exception of the investment rate, and another similar exercise carried out by Levine and
Zerbos (1993) came to similar results. An improved version of EBA devised by Sala i
Martin (1997) found that several among the determinants proposed by growth theorists
turned to be robustly related to the latter; however, no fiscal variable entered this group.
In this type of analysis, those explanatory variables whose inclusion seems to be well grounded compose an I-set,
which is included in all equations to be estimated. The rest of independent variables are included by turns, in groups
of two or three at a time. The test thus runs a number of regressions which can become really high. Attention is
focused on the lowest and highest s among those estimated for any given variable. To the highest one is added a
term, equal to two times its own standard deviation (as given by the computer output), and from the lowest one two
times its standard deviation are deduced. If the two resulting figures lie to the same side of zero, the variable in
question is deemed to be robustly related to the dependent one; else, their relationship is termed as non-robust.
5
Pages: 5
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
These results seem to run counter the hypotheses of Armey, Barro, Rahn and Scully, but
several reservations can be expressed. First, EBA attributes the same value to all
specifications imaginable, provided that they contain the I-set of variables and
irrespective of any other consideration. It is evident, however, that any regression that
includes among its arguments variables that represent different links in the same causal
chain could yield, by this very fact, insignificant and unreliable coefficients for these
variables. The same may happen when we include, together with a given variable, any
other which is co-determined or otherwise correlated with it 6. Thus, one should be
prepared to obtain for each variable a number of coefficients that exhibit insignificant
and/or wrong signs, even when it is significantly related to the dependent one in
properly specified models. The (excessive) precaution of adding two standard
deviations to the extreme bounds of the coefficient distribution, the ensuing comparison
of the signs of the resulting figures and the verdict of no robust relationship applied to
any variable that fails to produce equal signs for the two compose a test apt to yield a lot
of false negatives. Therefore, EBA analysis doesnt provide, in my opinion, a good way to
deal with the problem of control variables selection. But, in the absence of a better
solution to this problem to which cross-sectional analysis is particularly sensitive, it is
impossible for authors to agree on a definite conclusion.
This outcome invites to lend special attention to time-series analysis, focused on
individual countries. The underlying idea is that, as each country presents its own
inverted-U curve which is depending on the behavior of other variables, and as the latter
tend to change less over time than they do across the space, the issue of control
variables may be less of a problem. This, of course, doesnt mean that all variables other
than the government share have to remain constant. In addition, in time-series analysis
we have to deal with the problem of reverse causation. Where the dependent variable is
income per person, one cause for this problem is the working of the Wagners Law, the
long-run tendency of the government share to increase with GDP per capita. Next, no
matter whether the dependent variable is GDP or its rate of growth, another motive for
reverse causation stems from the tendency of public expenditure to remain more stable
than is private production during the business cycle. Due to this, the government share
tends to present short-term fluctuations contrary to those of GDP and (foremost) to
those of its growth rate.
As far as I know, no less than twenty academic contributions have tried to check the sign
of the relationship between GDP or its growth rate and the size of government, by means
Regarding the control variables to be included, one question to bear in mind is that the size of the public sector may
exert its purportedly negative effect on growth via two different mechanisms. First, if the marginal productivity of the
production factors is lower in the public sector, then their diversion from the private one should have an adverse
effect on growth. Second, the increased tax burden may impose a disincentive to savings, investment and, perhaps,
effort. Since the production function approach takes investment as given, it ignores the latter effect, and may
downplay the estimation of the negative impact of an increase in the size of the public sector. On the other hand,
dispensing with investment when one is trying to explain the growth rate or the level of GDP doesnt seem to be
advisable. There is no easy way out in this dilemma, however, when one is using single-equation models.
6
Pages: 6
It is fair to add that, in a prior contribution due to Ram (1986), this author obtained a positive sign for 100 out of 115
countries analyzed. However, his explanatory variable was not the government share, but the rate of growth of public
expenditures as such. The behavior of this variable is surely driven by the force of the Wagners Law, and hence it
cant be said that Rams results contradict the conclusion stated above. The few negative signs obtained in Rams
estimations corresponded mainly to the most advanced economies. In two later contributions Grossman (1987 and
1988) also employed the absolute size of the public sector as a regressor, instead of its size relative to GDP. Yet, his
results, referred to the USA during the 1929-to-1982 period, situated the optimal size of government for this country
at 20 per cent of GDP, a figure that has been largely surpassed since the 1940s.
8
This is done by Grossman (1987, 1988), Peden and Bradley (1989), Peden (1991), Vedder and Gallaway (1998), and
Facchini and Melki (2012).
9
10
As in Carlstrom and Gokhale (1991), or in Branson and Lowell (2001). The estimates of Dar and AmirKhalkhali
(2002) for individual countries apparently stem from a later transformation of the results of their panel-data analysis.
11
Pages: 7
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
previous year. The rationale underlying this particular choice is never stated, apart from
signaling that the optimum government share can be readily calculated from the
regression coefficients of these two variables; on the contrary, the validity of the
function is implicitly assumed to be a matter of fact. Scullys papers deal with the USA,
with the exception of Scully (1996) which refers to New Zealand, and economists of the
latter country seem to have been particularly attentive to possible flaws in the model.
Thus, Chapple (1997) pointed at an inconsistency between the theoretical model and
that estimated in practice, and Sieper (1997) and Kennedy (2000) criticized the model
itself on the grounds that it makes the implicit assumption that all investment of a given
year is consumed in the next one. Later on, Hill (2008) has demonstrated that the results
obtained with the Scully model are sensitive to the correction indicated by Chapple as
well as to slight changes in the estimation period. Thus, whereas Scullys estimates for
the optimal government share in the US closely cluster between 19.3 and 22 per cent of
GDP, Hills stretch from 9 to 29 per cent.
The preceding comments arent intended to dismiss the importance of the cumulated
evidence in favor of the hypotheses of Armey, Barro, Rahn and Scully, as achieved by
time-series analyses and the bulk of cross-sectional ones. On the contrary, I believe that
the basic convergence of results obtained from such different approaches can hardly be
attributed to mere chance. Thus, my own contribution to the time-series approach, as
presented in the ensuing sections, should be seen as no more than a test of these results,
after taking strong precautions facing the issues of endogeneity and the indirect effects
via investment. Prior to that, however, a reference to the most recent panel-data
analyses has to be made.
Lastly, a number of authors have used panel-data analysis for long periods of time,
focusing on developed countries only and using measures for public sector size that
include all public expenditures. In a recent review, Bergh and Henrekson (2011)
mention seven such articles, published in international academic journals with peer
reviewing12. With the exceptions of Agell et al. (2006) and Colombier et al. (2009), all
other studies indicate that the growth effect of public-sector size is indeed negative for
developed countries. Moreover, by adding new control variables Bergh and Karlsson
(2010) and Bergh and hrn (2011) have managed to reverse the conclusions of the two
dissenting papers. Thus, it can be said that the last wave of studies reasserts the negative
role of too a big government share.
An innovative feature in the article of Bergh and Karlsson (2010) is that is uses a new
type of robustness analysis, the Bayesian averaging of classical estimates (BACE)
developed by Doppelhoffer, Miller and Sala i Martin (2004) 13. By applying this type of
Their list includes Flster and Henrekson (2001), Dar and Amir Khalkhali (2002), Agell et al. (2006), Romero vila
and Strauch (2008), Colombier (2009), Afonso and Furceri (2010), and Bergh and Karlsson (2010). Were it enlarged
to include non-published papers, it could also mention, for instance, Heitger (2001), Chobanov and Mladenova (2009),
Josheski et al. (2011), Bergh and hrn (2011), or Afonso and Tovar Jalles (2011).
12
This method calculates a coefficient for each explanatory variable as the mean of all its estimates, as provided by the
different regressions. Contrary to EBA, BACE weights these coefficients by the goodness of the fitting of the
13
Pages: 8
analysis to a panel of rich countries during the 1970-to-1995 and 1970-to-2005 periods,
and using 21 variables (in all of their 7-digit possible combinations), Bergh and Karlsson
(2010) conclude that the ratio of tax revenue to GDP and (for the 1970-to-2005 period)
the ratio of government expenditure to GDP are indeed robust determinants of
economic growth.
At the end of the day, it can thus be said that the results of panel-data analyses confirm
the results obtained by time-series analyses as well as by most cross-sectional ones.
Pages: 9
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
which all or most observations are located on one and the same side of the Armey
curve, it is dubious that the corresponding half of the latter presents a constant
elasticity.
Once these comments have been made, we can proceed to present the model, which
assumes an aggregated production function of the Cobb Douglas type:
Y = A K L1-
(1)
When dividing the two sides of this expression by L, we get the intensive or per-worker
from of the function:
Y/L = A (K/L)
(2)
Taking growth rates (indicated here by means of lower case letters), the preceding
expression gives:
yl = a + (kl)
(3)
where, by want of the necessary information, we are forced to take a as a constant.
To the latter expression we may now add other possible determinants of y/l, as well as
an error term. To start with, we can write:
yl = a + (kl) + (g-y) +
(4)
where g-y indicates the growth rate of public expenditures relative to GDP. Of course,
other arguments may be added to the right-hand side of this expression.
Since we need to economize on symbols, we will rewrite the last expression as:
yl = a0 + a1 (kl) + a2 (gy) +
(5)
It is apparent in this expression that, since y enters the calculation of the rate of growth
of the government share in GDP (gy), we have a problem of reverse causality. For this
reason we will formulate a second equation, in which gy is a function of yl, as well as
of a vector of other determinants (R):
gy = b0 + b1 (kl) + b2 (R) +
(6)
Another point deserving attention in (5) is the assumption that, in the relationship
between product and capital, the causation runs from the latter to the former, as stated
in all growth models. However, many models in macroeconomics assume that the
investment rate is a function of (among other variables) the rate of growth of the
economy. This assumption is most clearly found in the Keynesian theory of the
accelerator, which takes investment to be dependent on the increase in output, but
macroeconomics also stresses the close dependence of consumption on income, and the
refined version of this hypothesis considers consumption to be a function of permanent
income. If we put apart the permanent and non-permanent components of income, and
given the identity I+C = Y, it follows from this that investment closely depends from the
transitory component of income, which is tantamount to say from income fluctuations.
The assumption that investment depends on the rate of growth of the economy is also in
accordance with common sense, which suggests that the capacity-utilization rate and,
perhaps, the rate of depreciation of the capital stock, also depend on the cycle. With all
this, it seems safe to assume that the rate of growth of the stock of capital per worker is a
function of yl. Moreover, if there is some truth in the idea that an increase in the size of
the public sector affects investment, the rate of growth of the capital stock per worker
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 10
should be also a function of gy. Allowing for the existence of a full host of other
determinants of investment (S), we should then write:
kl = c0 + c1 (yl) + c2 (gy) + c3 (S) + e
(7)
With this, the expressions (5), (6) and (7) become the structural equations of a system of
simultaneous equations, in which yl, gy and kl are endogenous variables. The
exogenous or predetermined variables of the system are all those comprised in the
vectors R and S, plus those pertaining to a third vector (T) that we could add to (5) if
we consider it convenient.
From the structural equations we have to formulate those of the reduced form, in order
to estimate the model. Since the rate of growth of capital per worker and that of the
share of government influence each other (directly the latter to the former, and via the
rate of growth of product per worker in the opposite case), the three equations of the
reduced model have to include, to the right of the equal sign, all variables that make up
the vectors R and S. Namely:
yl = d0 + d1 (R) + d2 (S) + e
(8)
kl = f0 + f1 (R) + f2 (S) + u
(9)
gy = h0 + h1 (R) + h2 (S) + u
(10)
Our next task is to decide which variables should be included in the vectors R and S
and, if the case comes, in vector T. This is necessary not only to make the model
operational, but also to determine which estimation technique is adequate to our case.
Depending on the number of these predetermined variables, if the number of
parameters in the reduced-form equations is equal to that in the structural ones, the
model is exactly identified, and the adequate estimation method is indirect least squares.
If, on the contrary, the reduced-form equations contain more parameters than do the
structural equations, the model is over-identified, and the appropriate technique is twostage least squares. With more than two variables in each of the vectors R and S, our
system becomes over-identified, and the appropriate method turns to be 2SLS. By the
way, it is worth noting that 2SLS also provides a way out the annoying dilemma of the
inclusion of investment in the growth equation: in the second-stage growth regression,
the instrument for investment represents only that part of the increase in the capital
stock that is free from the influence of changes in the size of the public sector. Therefore,
the regression coefficient of the instrument for the government share should catch the
full impact of the latter on growth, consisting of its direct effect (via resources
productivity) plus the indirect one (via how many resources are allocated).
The selection of the predetermined variables is a task to be performed with great care,
since inclusion of any one that is not really exogenous would render the whole exercise
useless. Here we are facing a tradeoff. The more numerous the variables included in the
sets R, S and T, the better is the fitting of the first-stage regressions and, at first sight,
the higher should be the confidence we put on the estimated instruments to be used in
the regressions of the second stage. On the other hand, however, any mistake incurred
into in this domain could ruin the whole construction. When reviewing the literatures on
the determinants of income growth, investment, and public sector growth, a number of
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 11
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
Pages: 12
Another possible candidate is the inflation rate of the economy which, according to
many authors, may deter capital accumulation.
Regarding those determinants of the rate of growth of output per worker (the
dependent variable in our growth equation) other than the increase in capital per
worker and changes in the size of the public sector, it would have been desirable to
include other usual arguments of the aggregate production function, as may be the stock
of human capital per worker. However, the extant series for this variable have been
calculated through several steps and, in each of these, heroic assumptions have been
made. The resulting series thus exhibit a smooth outline, apt to represent the trend but
insensitive to short term fluctuations. More often than not, their authors use only one
decimal point (the variable is usually measured in school years per person), so that the
series tend to remain stable for years, jumping to the next stair from time to time. In
either case, these series arent suitable for our work with growth rates.
All regressors used in the first-stage regressions have to be used also in one or more of
the second-stage ones. However, the second-stage regressions may include other
variables that we have refrained to use in the first one on the grounds that they cannot
be considered as truly exogenous. Variables that we could include in the growth
equation of the second stage only are the already mentioned changes in the proportion
of children within the population, the internal composition of the labor force by sector
and status, or the rural/urban distribution of the population, all of which tend to
accompany economic growth. Most probably, these variables arent real determinants of
growth, but here the target is to check the robustness of the influence of the government
share on growth. Given this objective, the presumed association of these variables with
economic growth converts them in good controls, as they make the test more stringent.
A final comment relating many variables of the system is that, on theoretical grounds, it
is unclear whether we should use their levels or, rather, their rates of growth in the
equations where they appear. Hence, where possible I have run different regressions
using either their levels or their growth rates.
In the two following sections I perform two different experiments. In section 4 I use a
rather reduced set of predetermined variables to estimate the first-stage regressions,
and no other exogenous variables in the second-stage ones. That set consists only of the
behavior of the growth rate in the rest of the countries that form the sample, the interest
rate, the ratio of service prices to those of material goods, and the openness of the
economy, all of which are available for the (almost) complete sample of countries. In
section 5, the list of predetermined variables is reinforced with others that arent
available for all countries. With the same purpose, in that section I employ other
available indicators, different from the ones used in section 4, for the same variables as
used in it, as well as new control variables to be included in the second-stage
regressions. Obviously, the use of all series is preceded by a test in order to ensure that
all of them are stationary; where not, the series in question have been discarded.
Going now to the data sources, the three endogenous variables of our system (i.e. the
rates of growth of product per worker, of the government share in the social product
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 13
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
and of the stock of capital per worker) have been calculated out of data taken from the
Penn World Tables (PWT 6.3). Output per worker is real output, expressed in PPPs,
divided by total employment (rgdpl2te in the source). The government share is that
corresponding to public consumption, calculated at current prices in order not to
interfere with the mechanism identified by Baumol (cg in the source). As for the growth
rate of the capital-stock per worker, its construction has required some more effort. One
possible indicator for this variable is the investment rate. Since the numerator of its
formula indicates the gross increase in the capital stock, and its denominator (Y) is a
function of (among other things) the number of persons working in the economy, the
investment rate could provide an acceptable proxy for the increase in capital per
worker. However, better indicators for this variable can be constructed from the Penn
World Tables. In order to calculate the rate of growth of the stock of capital, be it in
absolute terms or per worker, we need first a series on the capital stock as such. This is
hard to get. However, the stock of capital existing in any given year is no more than a
summation of past investment, in which the bulk corresponds to the decades
immediately prior to the year in question. Given that PWT provide data on investment
starting in 1950, the construction of an acceptable proxy for the capital stock since (say)
1975 onwards becomes thus possible. To get a more refined figure we can apply to the
simple sum a depreciation rate in the customary form. Through this process, I have
calculated two alternative proxies for the capital stock, with and without depreciation.
In the cases of those countries and periods for which I have found more careful
estimates for this variable, constructed by other researchers or teams of researchers, a
comparison of my series with theirs have been possible. The correlation coefficients
between them are very high and, surprisingly, higher for my gross measure. For
instance, in the case of Spain the r between my gross measure of the capital stock and
the best series available is 0.9966.
For the rest of variables, where they were not reported in PWT I have resorted to the
customary sources: OECD, IMF, World Bank and, in some cases, to national sources as
well. Where several alternative series were at hand, as is the case of the interest rates, all
series that I have been able to find have been used in section 5. The color of the
government has been assigned out of information provided by the webpage
Worldstatesmen.org. To this end I have examined only the party affiliation or political
background of the chief of the executive. Only in two cases (those of Switzerland and
Mexico) the color of government became too problematic to elucidate, obviously for
different reasons in each of these countries.
Finally, the selection of countries has been motivated by an interest in looking at what
happens where there are large public sectors, as is the usual case for rich countries.
Within OECD, and for evident reasons, I have dispensed with all former communist
countries. Germany has been also excluded, due in part to the same reason as well as to
the disruption of her time series as a consequence of reunification. As for the time span,
three different motives led me to restrict it to recent years. First, as stated before, the
use of linear models may be ill suited for the study of long periods, in which the
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 14
observations could be distributed almost certainly over the two halves of the Armey
curve. Second, I wanted to minimize the risk of using series that could exhibit changes in
their structure. Other things being equal, the shorter the period, the lower this risk
would be. The period stretching from 1975 to 2007 looks long enough for not getting
run of degrees of freedom; at the same time, it is not too long, and, running from the
aftermath of the first oil shock of 1973 to the eve of the 2008 crisis, it includes no major
economic breakthrough. Last, a period neither too long nor too short was convenient, as
stated above, for the calculation of a series for the stock of capital.
Throughout the preceding presentation it should become evident that the test
undertaken in this article is purposely restrictive, and biased against the hypotheses
submitted to check:
- The use of growth rates implies that we are estimating short-term elasticities
only.
- Certain control variables included in second-stage regressions may improperly
detract from the levels of significance of the variable of interest.
- Government expenditures are represented by consumption expenditures, which
is not the type of public expenditures that has increased the most during the
period.
And, the post-2008 years have been excluded, as the possible oversizing of the public
sector during these may have been magnified due to shrinking private production.
Pages: 15
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
countries) these coefficients are statistically significant at the 0.1 or lower levels.
Moreover, according to the relevant statistics our estimated instruments for (the rate of
growth of) the government share are weak in the cases of Finland, Italy, Japan, New
Zealand, Portugal, the United Kingdom and the United States 15. Moreover, the
corresponding first-stage regressions are only marginally significant at the 0.1 level in
the additional cases of Austria, Canada, and the Netherlands. In all of these cases,
therefore, one shouldnt put too much confidence in our coefficients for the variable of
interest, as estimated in this baseline model.
There is no reason, however, to expect any such bias in the other cases, which include
nine significant results. Up to this point, then, our baseline results support the
hypothesis of excess government for roughly one third of the countries that form our
sample. In seven additional cases the coefficients are also negative but insignificant, and
in the rest of countries but one the coefficients are still negative, although suspect.
The persistence of a possible bias, the lack of significance of other coefficients of interest
and the sometimes low R 2s of other regressions (although significant in all cases but
two, as indicated by the F-test), invite to see what happens when using other alternative
indicators for some of the variables included in these baseline regressions, and, more
important, when other control variables are incorporated to the model.
The F-statistics have been calculated according to Angrist and Pischke (2009:218) with their correction later posted
at http://www.mostlyharmlesseconometrics.com/2009/10/multivariate-first-stage-f-not/.
15
Pages: 16
Country
Constant
Australia
-4.411***
1.225***
-0.626***
0.49
<0.01
Austria
-4.310**
1.142***
-0.690
0.25
<0.01
Belgium
-5.209*
1.360**
-0.114
0.10
<0.10
Canada
-2.325
0.671
-0.356*
0.26
<0.05
Denmark
-5.616
1.355
-0.477**
0.11
<0.01
Finland
2.666
-0.102
-0.361***
0.29
<0.01
France
-12.702**
-0.261**
0.20
<0.05
Greece
4.691
-0.160
-0.05
<0.05
Iceland
-3.175
1.257*
-0.660**
0.19
<0.01
Ireland
0.236
0.485
-0.416*
0.02
>0.10
Italy
0.799
0.074
-0.481*
0.649***
0.35
<0.01
Japan
5.406
-0.710
-0.471
0.504**
0.22
<0.01
Korea
4.112
-0.091
-0.589***
0.36
<0.01
Luxembourg
5.739
-0.922
-0.149
-0.02
<0.01
Mexico
-3.773
1.088
-1.038
0.02
<0.01
Netherlands
-0.605
0.314
-0.287
-0.06
<0.01
New Zealand
-5.848***
1.377***
-2.489**
0.74
<0.01
Norway
-2.945
1.127**
-0.090
0.14
<0.05
Portugal
-2.907
0.920
-0.196
0.03
>0.10
Spain
-0.711
0.484**
0.021
0.17
<0.01
Sweden
-3.494*
1.081***
-0.294*
0.20
<0.01
Switzerland
-3.604**
1.110***
-0.731***
0.57
<0.01
Turkey
12.008
2.387*
-0.100
0.09
<0.05
U. Kingdom
-3.047
0.857
-0.498*
0.05
<0.10
-0.383
-0.033
0.15
<0.01
United States
Notes:
a
b
3.216*
GrKpw
gr G/Y
0.547**
-0.745
AR term
0.373*
-0.625***
Adjusted R2
Prob(F)b
The start and ending years may differ slightly from country to country, depending on data availability.
As calculated according to Angrist and Pischke (2009:140).
Pages: 17
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
Pages: 18
for the government share. This notwithstanding, the numbers of useful estimates in
table 3 is higher than those in table 2 for all countries (except for New Zealand, for
which the two are zero). On average, they are higher in proportion of four to one.
Regressions
run (Ljung
& Box Q
significant
= 0.05 )
With
negative
sign for
^
gr(G/Y)
of which
significant
(at least)
at 0.1
With
positive
sign for
^
gr(G/Y)
of which
significant
(at least)
at 0.1
Mean
of the
coefficient
for ^
gr(G/Y)
SD of the
coefficient
for ^
gr(G/Y)
Australia
18
All
None
-0.426
0.127
Austria
35
All
12
None
-0.700
0.183
Belgium
All
None
None
-0.290
0.172
Canada
All
None
-0.323
0.099
Denmark
All
None
-0.379
0.131
Finland
14
All
All
None
-0.368
0.021
France
All
None
-0.346
0.101
Greece
None
None
-0.013
0.072
Iceland
None
-0.371
0.063
Ireland
All
All
None
-0.485
0.048
Italy
15
All
14
None
-0.783
0.187
Japan
Korea
17
All
All
None
-0.681
0.198
None
-0.157
0.197
Mexico
16
All
None
None
-1.063
0.235
Netherlands
12
All
None
None
-0.332
0.048
Luxembourg
N. Zealand
Norway
44
42
19
None
-0.103
0.041
Portugal
None
None
-0.285
0.244
Spain
11
None
None
-0.006
0.125
Sweden
18
All
11
None
-0.370
0.176
Switzerland
57
All
All
None
-0.682
0.101
Turkey
None
None
0.063
0.105
U. Kingdom
All
None
-0.278
0.179
U. States
None
None
-0.044
0.150
Note: The exact period under analysis varies from one estimate to another, depending on the starting and ending
years of the series employed in each of them.
a
Pages: 19
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
Regressions
run (Ljung
& Box Q
significant
= 0.05 )
With
negative
sign for
^
gr(G/Y)
of which
significant
(at least)
at 0.1
With
positive
sign for
^
gr(G/Y)
of which
significant
(at least)
at 0.1
Mean
of the
coefficient
for ^
gr(G/Y)
SD of the
coefficient
for ^
gr(G/Y)
Australia
32
All
27
None
-0.540
0.121
Austria
60
All
19
None
-0.886
0.345
Belgium
32
27
None
-0.137
0.107
Canada
32
All
23
None
-0.285
0.087
Denmark
102
All
84
None
-0.416
0.058
Finland
44
All
35
None
-0.349
0.082
France
168
157
112
11
None
-0.274
0.161
Greece
27
24
None
None
-0.114
0.138
Iceland
11
All
None
-0.849
0.356
Ireland
45
34
18
11
None
-0.335
0.563
Italy
88
75
13
13
None
-0.223
0.218
Japan
60
All
49
None
-0.441
0.042
Korea
44
42
34
-0.467
0.172
102
All
51
None
-0.251
0.112
-0.641
0.978
102
83
None
19
None
-0.187
0.178
Luxembourg
Mexico
Netherlands
New Zealand
Norway
76
56
22
20
None
-0.066
0.121
Portugal
22
All
19
None
-0.564
0.067
102
23
None
79
None
0.068
0.091
Sweden
89
All
60
None
-0.329
0.028
Switzerland
75
All
71
None
-0.721
0.266
Turkey
29
26
None
None
-0.064
0.047
U. Kingdom
29
All
23
None
-0.588
0.175
United
States
51
44
18
None
-0.175
0.126
Spain
Note: a The exact period under analysis varies from one estimate to another, depending on the starting and ending
years of the series employed in each of them.
A detailed comment on the results presented in tables 2 and 3, country by country, is out
of question in this article. Thus, the following comments are restricted to the signs of the
coefficients of the variable of interest, their values, their significance levels, their means
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 20
and their standard deviations within each country. The similarity between tables 2 and 3
in all of these respects is another point to be commented.
Regarding the signs of the coefficients for the government share, as they appear in tables
2 and 3, these are predominantly negative, in the proportions of 18 to one in table 2, and
seven to one in table 3. Almost all positive coefficients differ from zero starting in the
second or third decimal only. Furthermore, they stem from specifications that include as
controls so powerful explanatory variables as are the rate of growth in other countries,
and/or the inflation rate. The aggregate figures, however, conceal differences among
economies. In ten countries all coefficients estimated are negative, these being Australia,
Austria, Canada, Denmark, Finland, Iceland, Sweden, Switzerland and the United
Kingdom (plus Japan, for which table 2 provides no information). The coefficients are
also overwhelmingly negative in the cases of Belgium, France, Italy, Korea, Luxembourg,
Mexico and Portugal, and predominantly negative though not so much in the rest of
cases, with the sole exception of Spain. Even in this country the positive coefficients
correspond to estimates that include the late 1970s, whereas the positive ones pertain
to estimates that, by want of long series for the corresponding variables, start around
1980. This is not so strange, as Spain emerged from an authoritarian regime in 1976
with a very small public sector that quickly increased in subsequent years. A replication
of the Spanish regressions starting in the early 1980s instead of 1975 causes all positive
coefficients to change to negative. Thus, a likely interpretation is that Spain surpassed
her optimum size for the public sector at a later date than did other countries.
Something similar, though far less evident, seems to have also happened in the parallel
cases of Portugal and Greece. In any event, the coefficients for Spain are close to zero
(0.06 or -0.07 on average, depending on the starting year), and those of the other two
countries are mainly negative all the time.
The next question is that of the significance of the coefficients obtained. Nowhere in the
tables do we find positive coefficients that turn to be significant. But there is no country
for which all coefficients are both negative and significant in the two tables, either. Four
countries are in this situation in table 2 (Namely: Finland, Ireland, Korea and
Switzerland), but none in table 3. In the aggregate, in one country (namely: Switzerland)
the proportion of negative and significant coefficients is over 90 per cent (97%), and in
other four (Denmark, Finland, Japan and Korea) it is over 80 per cent. It is over 70 per
cent in other three countries (Australia, Canada and the United Kingdom), and over 60
per cent in four more (France, Iceland, Portugal, and Sweden).
Thus far, then, we can say to have found rather strong indications of a negative
relationship between the rates of growth of the government share in more than half the
countries, and no similar indication of a positive one for any one. Since in no country all
coefficients are both negative and significant, the partisans of the extreme bound
analysis are free to interpret from this exercise that the inverse relationship between
the rates of growth of output per worker and of the government share is not robust in
any country whatsoever. My own interpretation is different, however. In a prior section I
have compared the extreme bound analysis to a clinical test that yields a lot of false
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 21
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
positives. Focusing on the trees of the extreme bounds is perhaps the best way to lose
sight of the wood of the whole distribution of the coefficients estimated 17.
In the two last columns of tables 2 and 3 we can find the means and SDs of the
distributions corresponding to the coefficients estimated for each country, and their
comparison is always interesting. As a matter of fact, in those countries for which we
have many estimations (which are almost all in table 3) we could made the assumption
that the estimated coefficients compose random samples, and calculate the confidence
level of their means in the manner that is usual in random sampling. This procedure is
also analogous to the one used to calculate the level of significance of any coefficient in
any particular regression, with the main difference that it can be estimated directly
instead of being inferred from the distribution of the errors.
Generally speaking, the mean of our coefficients is high relative to the SD, in both tables,
for Australia, Austria, Canada, Denmark, Finland, France, Iceland, Korea, Japan,
Luxembourg, Sweden, Switzerland, and the UK: 13 countries in total. Were we to
formally calculate the analogous to the significance levels of the mean coefficients
corresponding to these countries, we would place them at the 0.01 (in both tables) for
the cases of Australia, Austria, Finland, Korea and Switzerland, and above the 0.05 level
for Sweden. Whereas the number of regressions reported in table 2 is too small to allow
for such calculations, the significance levels arousing from table 3 would be 0.01 for
Canada, Japan, Luxembourg, and the United Kingdom, and 0.05 for Denmark, France, and
Iceland.
Thus far we have employed three different criterions to evaluate our results. Namely:
whether all (or almost all) coefficients exhibit the same negative sign; whether a
majority of them are significant, and whether their means compare favorably with their
SDs. There are seven countries that accomplish with all three criteria to a large extent,
these being Australia, Canada, Denmark, Finland, Japan, Switzerland and the United
Kingdom. Other five countries (Iceland, France, Korea, Luxembourg and Sweden) also
accomplish with these three criterions, although not as well as the former ones.
Generally speaking, then, we can say to have found strong indications that an increase in
the government share is bad for economic growth in at least twelve out of 24 countries
during the period under examination. Regarding the rest of the countries, the conclusion
is less firm, although in Austria, Belgium, Ireland, Italy, Mexico, Norway, Portugal, the
United States the results clearly point in the same direction. In Greece, the Netherlands,
Turkey and Spain after 1980, the indications also point to this direction, albeit timidly.
Is the assessment of a negative effect of the government share equivalent to a
confirmation of the Armey hypothesis? For this we should compare the ranking of
Furthermore, we have to take into account that, even where all components of any given distribution exhibit the
same sign, almost by definition one of the extreme bounds will be not too far from zero and this implies a high
probability of its non-being significant. There is another possible reason that may cause one coefficient not to be
significant where others in the same distribution are: this is that the error term in that precise specification (which
provides the denominator for the expression of the t-values) is higher than those of the rest of others. But this usually
means that the former specification is less efficient than are the other ones, so that we shouldnt put much stress on
that particular result while being oblivious of the rest.
17
Pages: 22
countries that we can establish according to our chosen criteria with the one we can
construct from the relative size of the public sector throughout the period under
analysis.
The results of this comparison are mixed. It is true that all OECD countries tend to
exhibit large public sectors, and that Sweden, Denmark, Finland and the United Kingdom
hold the four top posts in this respect. According to our estimates, they have in fact too
large public sectors. But increases in public expenditures also seem to have had clear
negative economic effects in the cases of Australia, Canada, Japan, Korea, Luxembourg or
(especially) Switzerland, all of which exhibit small public sectors. On their part, Norway
or the Netherlands are close to be in the opposite case.
One possible interpretation is that, when countries increase the size of their public
sectors over time, they dont do this in the orderly manner implicitly assumed in
Armeys hypothesis, i.e., starting with the most growth-enhancing public activities and
going down the scale up to those activities with no positive effect that could counteract
the necessary increase in the fiscal burden. This could allow for negative effects in
countries with comparatively small public sectors, as well as to the opposite case, due to
differences in the internal composition of expenditures and taxes. Other possible
explanations could point to the relative efficiency of the national bureaucracies and
decision making processes, as determined by the quality of both formal and informal
institutions of each country. These points have been stressed by the recent literature in
order to explain anomalies not dissimilar from the ones we have found here. Thus, Dar
and AmirKhalkhali (2001), as well as Bergh and Henrekson (2011), employ similar
arguments to explain the more than acceptable economic record of Scandinavian
countries. Afonso and Tovar Jalles (2011) also find that institutional quality may
counteract the effect of the government share in countries that rate differently in these
two respects, or reinforce it in countries that score similarly in the two of them.
Finally, a different interpretation is that it is not so much the absolute levels of public
expenditures, as their changes, that matters for GDP growth. Our method of estimation
of the elasticity of GDP to changes in the size of public consumption, through regression
of the rate of growth of the former on that of the latter, estimates indeed short-run
elasticities, thus allowing for this interpretation. If the explanatory variable doesnt exert
a durable effect on the dependent one, then the long-run elasticity of the latter to the
former would be lower than the short-run elasticity, and perhaps much lower. However,
it is not easy to imagine why and how this may happen.
In any case, the results obtained dont seem to be an artifact of inefficient estimation.
Care has been taken with problems of reverse causality, specification selection and
elimination of weak instruments. As a further precaution, the results of the whole
exercise have been validated by one additional test that can be described as follows. Our
ultimate reason to use a simultaneous-equations model and two-stage least squares is
that the explanatory variable, G/Y, depends to some extent on changes in the dependent
one, which provides the denominator for its expression. This dependence happens
because, although the movements of G along the business cycle follow those of Y, they
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 23
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
are smoother than these. To put it in another way, the elasticity of G to Y is positive in all
countries, though lower that unity. On the contrary, the elasticity of the denominator of
G/Y to Y is one by definition. Now, if we proceed to replacing the Ys in expression G/Y by
their trend values, then the elasticity of its numerator to GDP turns to be higher than that
of its denominator. Thus, when we recalculate the G/Y ratios using a smoothed series for
Y (as are the one that results from passing the original series through the HendrickPrescott filter), and we use the series of G/Y thus modified as a regressor with which to
explain the behavior of the growth rate of product per worker in a single-equation
model using OLS, the results turn to be biased towards the positive, instead of towards
the negative as it happens when the same regression is run using the original GDP
figures. Hence, this exercise provides us with an upwards-biased figure for the coefficient
that we are trying to estimate. Thus, when this upwards-biased coefficient turns to be
negative for any given country, the conclusion to be drawn is that, in that country, the
rate of growth of G/Y indeed exerts a strong negative impact on that of GDP. In
regressions not shown here I have run this test. The results still yield negative
coefficients for the variable of interest in about half the countries for which we have
obtained clearly negative coefficients throughout the paper, and about zero in most
other countries pertaining to this group. Incidentally, this test also indicates a negative
growth effect of increases of the public sector in New Zealand, a country for which I
didnt manage to get efficient instruments for the variable in the 2SLS exercise, and is,
therefore, absent from tables 2 and 3 18.
5. Concluding remarks
All academic contributions aimed at estimating the growth effects of big public sectors
by means of time-series analysis unanimously conclude that these effects are negative,
and this is also the conclusion of most panel-data and (to a lesser extent) of most crosssectional analyses addressed to the same question. Recently, Bergh and Henrekson
(2011) have stated a strong case in favor of the negative role of big public sectors,
relying exclusively in the results of recent panel-data analyses. Given the unanimity
reached through the analysis of time series, however, it seemed to be worthwhile to
check whether their conclusions hold after correcting for possible weaknesses in their
underlying models or, on the contrary, were caused by those purported flaws.
Three key points are at issue here: those of reverse causation, control-variables
selection, and the indirect effect of the government share through the investment rate.
This article tackles with these issues (especially with 1 and 3) through the formulation
of a simultaneous-equation model and the use of 2SLS. Due partly to the characteristics
of the available data, the choices we have made have resulted into a rather strict test of
the conclusions of prior contributions:
18
Use of the weak instruments obtained for this country also point in the same direction, however.
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 24
The use of growth rates implies that we are estimating short-term elasticities only.
Certain control variables included in second-stage regressions may bias toward zero
the values and the levels of significance of the elasticities to be estimated.
- Government expenditures are represented by public consumption expenditures,
which is not the heading that has increased the most during the period.
- And the post-2008 years have been excluded, during which the possible oversizing of
the public sector may have been magnified by shrinking private production.
Moreover, the nature of our exercise imposes the restriction of estimating single archelasticities throughout the whole period under analysis. Strictly speaking, this
restriction may be hardly compatible with the notion of a functional relationship that
mimics the shape of an inverted U. It may be admissible for those countries that have
remained on the descending branch of their curves during the whole period under
analysis, but it is less adequate when applied to countries that may have moved around
their tops. This may help to explain why, in certain countries where the vast majority of
the estimated coefficients are negative (though close to zero), these fail to be significant
at conventional levels.
The exercise performed here provides with estimates of the elasticity of GDP per worker
to changes in the relative size of public consumption expenditure in 24 OECD countries.
The number of estimates varies from country to country, from a minimum of 23 (for
Mexico) to a maximum of 176 (for France). In all, we have made 1758 valid estimations.
In none of these a positive and statistically significant elasticity of GDP to the share of
government has been obtained. On the contrary, 872 estimates (that is, almost exactly
50 per cent) are both negative and significant at least at the 0.1 level.
In an exam country by country, the elasticities estimated are all negative for 10 out of 24
economies, and overwhelmingly negative in most others. Since there are no countries
for which all estimates are both negative and significant at conventional levels, skeptics
have the right to remain unconvinced. Nonetheless, other indications such as the
proportion of negative coefficients estimated for each country, that of significant
coefficients, and the degree of coincidence of the estimates provide strong evidence in
favor of the conclusion of a negative elasticity in roughly half the countries that form the
sample. In the rest of these, the evidence also favors the conclusion of a negative
relationship, with the possible exception of only a few cases. A likely conclusion is, thus,
that most OECD countries increased their public expenditures during the period under
analysis beyond that level that was optimum for their economic growth, and remained
above that optimum at least during most years of the period under analysis.
This conclusion serves, in my opinion, as confirmation that the inverse relationships
between the two variables estimated by other authors who have performed prior-time
series analysis, referred to the same countries, are valid in general terms. The
conclusions of time-series analyses thus join those of panel-data analyses and, for the
case, those of most cross-sectional analyses published since the 1980s. The elasticities
estimated here are generally higher than those reported by most panel-data analyses,
-
Pages: 25
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
but similar to the ones estimated by Dar and AmirKhalkhali (2001) for roughly the same
sample of developed countries.
The results are broadly consistent with Armeys hypothesis, although we have detected
several seemingly deviant cases. These may be due to the action of other variables not
controlled for, to a preference of certain small-sized governments for non-productive
expenditures, or to our focus in the short run. When regressing rate of growth on rate of
growth we are reproducing almost literally the expression of the elasticity, but the use of
yearly observations implies that we are estimating short-run elasticities. If an increase
in the government share fails to exert a durable effect on GDP, then the long-run
elasticity would be lower that the short-term one, and the absolute level of G/Y would be
irrelevant for GDP growth whereas its growth rate is. However, it is difficult to see why
the long-term effect of public sector size could be lower than is the short-term one.
This question aside, the conclusion of the article is that, with a degree of certainty that
varies from country to country, a reduction in their current levels of public consumption
expenditures could have enabled most OECD countries to enhance GDP growth during the
period under analysis. This conclusion refers to a period in whose final years the
countries GDPs were inflated by the boom of residential construction and production of
other durables. This boom was the result of expansive monetary policies. However, once
the sound demand for those goods became satiated, the crisis followed in both the real
and the financial sector of the economy. If the booming national economies prior to 2008
already had problems with their overgrown public sectors, it is not difficult to imagine
what may be happening in the subsequent period.
References
Afonso, A., Furceri, D., 2010. Government size, composition, volatility and economic
growth., European Journal of Political Economy 26: 517-532.
Afonso, A., Schuknecht, L., Tanzi, V., 2003. Public sector efficiency: An international
comparison., European Central Bank Working Paper No. 242.
Afonso, A., Tovar Jalles, J., 2011. Economic performance and government size.,
European Central Bank Working Paper No. 1399.
Agell, J., Ohlsson, H., Skogman Thoursie, P., 2006 Growth effects of government
expenditure and taxation in rich countries: A comment., European Economic
Review 50: 211-219.
Angelopoulos, K., Philoppopoulos, A., 2007. The growth effects of fiscal policy in Greece
1960-2000., Public Choice 131: 157-175.
Angrist, J.D., Pischke, J.-S., 2009. Mostly Harmless Econometrics: An Empiricists
Companion. Princeton: Princeton University Press. See also the authors tag
posted at www.mostlyharmlesseconometrics.com/2009/10/multivariate-firststage-f-not/
Pages: 26
Arcand. J.-L., Dagenais, M.-G., 1995. The empirics of economic growth in a section of
countries: Do errors in variables really not matter?, Montral: Universit de
Montral. Facult des Arts et des Sciences. Dpartament de Sciences
conomiques, CAHIER 9546.
Armey, R., 1995. The Freedom Revolution. Washington, D.C.: Regnery Publishing Co.
Barro, R.J., 1989. A cross-country study of growth, saving and government.
Washington: National Bureau of Economic Research Working Paper No. 2855.
Barro, R.J., 1990. Government spending in a simple model of endogenous growth.,
Journal of Political Economy 98 (5, part 2): S103-S125.
Barro, R.J., 1991. Economic growth in a cross section of countries., Quarterly Journal of
Economics 1991: 407-443.
Baumol, W.J., 1972. Welfare Economics and the Theory of the State. Cambridge, Mass.:
Harvard University Press.
Bergh, A., Henrekson, M., 2011. Government size and growth: A survey and
interpretation of the evidence. Journal of Economic Surveys 25: 872-897.
Bergh, A., Karlsson, M., 2010. Government size and growth: Accounting for economic
freedom and globalization., Public Choice 142: 195-213.
Bergh, A., hrn, N., 2011. Growth effects of fiscal policies: A critical appraisal of
Colombiers study., Stockholm: Research Institute of Industrial Economics IFN
Working Paper No. 858.
Branson, J., Lowell, C.A.K., 2001. A growth maximising tax structure for New Zealand.,
International Tax and Public Finance 8: 129-146.
Carlstrom, C.T., Gokhale, J., 1991. Government consumption, taxation, and economic
activity., Economic Review Federal reserve Bank of Cleveland 27 (3): 18-29.
Chao, J.C.P., Grubel, H., 1998. Optimal levels of spending and taxation in Canada., In H.
Grubel (ed.) How to Use the Fiscal Surplus. What is the Optimal Size of
Government? Vancouver: The Fraser Institute: 53-68.
Chapple, S., 1997. One for the X-files: A critical assessment of Professor Scullys
Taxation and economic growth in New Zealand. Report to the Periodic Report
Group. Wellington: New Zealand Institute of Economic research.
Chobanov, D., Mladenova, A., 2009. What is the optimum size of government? Sofia:
Institute for Market Economics.
Colombier, C., 2009. Growth effects of fiscal policies: An application of robust-modified
M-estimator., Applied Economics 41: 899-912.
Dar, A., Amir Khalkhali, S., 2002. Government size, factor accumulation, and economic
growth: Evidence from OECD countries., Journal of Policy Modeling 24: 679-692.
Doppelhoffer, G., Miller, R., Sala i Martin, X., 2004. Determinants of long-term growth: A
Bayesian averaging of classical estimates (BACE) approach., American Economic
Review 94: 813-835.
Easterly, W., Levine, R., 2001. Its not factor accumulation: Stylized facts and growth
models., World Bank Economic Review 15: 177-219.
Pages: 27
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
Easterly, R., S. Zerbos, 1993. Looking at the facts. What we know about policy and
growth from cross-country analysis., Washington: The World Bank, Policy
Research Working Paper WPS 1115.
Facchini, F., Melki, M., 2012. Optimal government size and economic growth in France
(1871-2008): An explanation by the state and market failures. Paper presented
at the 2nd World Congress of Public Choice Societies. Miami, 8-11 March 2012.
Flster, S., Henrekson, M., 2001. Growt effects of government expenditure and taxation
in rich countries., European Economic Review 45: 1501-1520.
Grossman, P.J., 1987. The optimal size of government., Public Choice 53:131-147.
Grossman, P.J., 1988. Growth in government and economic growth: The Australian
experience., Australian Economic Papers 27: 33-45.
Heitger, B., 2001. The scope of government and its impact on economic growth in OECD
countries., Kiel: Institute of World Economics, Kiel Working Paper No. 1034.
Hill, R., 2008. Optimal taxation and economic growth: A comment., Public Choice 134:
419-427.
Josheski, D., Lazarov, D., Koteski, C., 2011. Analysis of the optimal size of the
government consumption., Munich: MPRA Paper No. 32983.
Kennedy, P.E., 2000. On measuring the growth-maximizing tax rate., Pacific Economic
Review 5:89-91.
Leach, G., 2003. The Negative Impact of Taxation on Economic Growth. London:
Reform.
Levine, R., Renelt, P., 1992. Cross-country studies of growth and policy. Methodological,
conceptual and statistical problems., World Bank Policy Research Working
Paper WPS 608.
Levine, R., Zerbos, S., 1993. Looking at the facts. What we know about policy and growth
from cross-country analysis., Washington: World Bank Policy Research Working
Paper WPS 115.
Magazzino, C., 2008. Modeli dinterpretazione della dinamica delle spese publiche ed
curva de Armey: Il caso italiano 1862-2001, Notizie di Politeia XXIV n 92: 4560.
Magazzino, C., Forte, F., 2010. Optimal size of government and economic growth in EU27., Munich: MPRA Paper No. 26669.
Mueller, D.C., 2003. Public Choice III. Cambridge: Cambridge University Press.
Nijkamp, P., Poot, J., 2004. Meta-analysis of the effect of fiscal policies on long-run
growth., European Journal of Political Economy 20: 91-124.
Peden, E.A., 1991. Productivity in the United States and its relationship to government
activity: An analysis of 57 years, 1929-1986., Public Choice 69: 153-173.
Peden, E.A., Bradley, M.D., 1989. Government size, productivity, and economic growth:
The post-war experience., Public Choice 61: 229-245.
Pevcin, P., 2004. Does optimal size of government spending exist?, Paper presented to
the European Group of Public Administration (EGPA) Annual Conference.
Ljubljana, Slovenia, 1-4 September 2004.
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 28
Poot, J., 2000. A synthesis of empirical research on the impact of government on longrun growth., Growth and Change 31: 516-546.
Rahn, R., Fox, H., 1996. What is the Optimum Size of Government. Denver, CO: Vernon K.
Krieble Foundation.
Ram, R., 1986. Government size and economic growth: A new framework and some
evidence from cross-country and time-series data., American Economic Review
76: 191-203.
Raymond Bara, J.L., 1992. Gasto pblico y crecimiento. Un anlisis de los efectos del
tamao del sector pblico en Espaa y en la Europa comunitaria., Papeles de
Economa Espaola 52/53: 180-196.
Romero vila, D., Strauch, R., 2008. Public finances and long-term growth in Europe:
Evidence from a panel data analysis., European Journal of Political Economy 24:
172-191.
Sala i Martin, X., 1997. I just ran two million regressions., American Economic Review
87 Papers and Proceedings: 178-183.
Scully, G.W., 1989. The size of the state, economic growth and the efficient utilization of
national resources., Public Choice 63: 149-164.
Scully, G.W., 1991. Tax rates, tax revenues and economic growth. Washington and
Dallas: National Center for Policy Analysis, Policy Report 98.
Scully, G.W., 1994. What is the optimal size of government in the United States?
Washington and Dallas: National Center for Policy Analysis, Policy Report 188.
Scully, G.W., 1995. The growth tax in the United States., Public Choice 85: 71-80.
Scully, G.W., 1996. Taxation and economic growth in New Zealand., Wellington: Inland
Revenue, Working Papers on Monitoring the Health of the Tax System 14.
Scully, G.W., 2003. Optimal taxation, economic growth and income inequality., Public
Choice 115: 299-312.
Scully, G.W., 2008. Optimal taxation, economic growth and income inequality in the
United States., Washington and Dallas: National Center for Policy Analysis, Policy
Report No. 316.
Sieper, E., 1997. Review by E. Sieper of Gerald W. Scully, Taxation and economic
growth in New Zealand., Wellington: Revised IRD Working Paper No. 14. Study
Commissioned for the Treasury. Government of New Zealand.
Slemrod, J., 1995. What do cross-country studies teach about government involvement,
prosperity, and economic growth?, Brookings Papers on Economic Activity 1995:
373-431.
Slemrod, J., 1998. How costly is a large, redistributive public sector?, Swedish Economic
Policy Review 5: 87-105.
Tanzi, V., Schuknecht, L., 2000. Public Spending in the Twentieth Century: A Global
Perspective. Cambridge: Cambridge University Press.
Taylor, J., 1993. Discretion versus policy rules in practice., Carnegie-Rochester Series on
Public Policy 39 : 195-214.
Pages: 29
S.Coll Is There Too Much Government In Developed Countries? A TimeSeries Analysis Of 24 OECD-Economies
Temple, J., 1999. The new growth evidence., Journal of Economic Literature 37: 112156.
Vedder, R.K., Gallaway, L.E., 1998. Government size and economic growth. Prepared for
the Joinf Economic Committee. Washington, DC: Joint Economic Committee.
Pages: 30
JEL Classification:
Z12; A13.
West University of Timisoara, Faculty of Economics and Business Administration, Blvd. Pestalozzi 16, Timisoara
300115, Romania, Email: bogdandima2001@gmail.com
19
East European Centre for Research in Economics and Business (ECREB), West University of Timisoara, Faculty of
Economics and Business Administration, Blvd. Pestalozzi 16, Timisoara 300115, Romania, E-mail:
ciprian.preda@feaa.uvt.ro
20
Vasile Goldi Western University of Arad, Faculty of Economic Sciences, Mihai Eminescu St. 15, Arad, Romania,
Email: stefana.dima@gmail.com
21
Pages: 31
1. Introduction
The choices we made in regard to this papers motto are designed to reflect the
complexity of religion as societal process. Firstly, the Dominican medieval distich
presented above represents a pledge for a polysemic interpretation of the revelation;
whereas The Letter to Hebrews views the function of faith as a guarantee for the
believer. On the other hand, the I Ching approaches the ethic imperative of the sincerity
that it is perfect in itself and of itself.
So, we view religion as dealing with a variety of individual and social life related aspects:
the position of an individual in the web of social interaction networks, or in relation with
the universe and the Divinity; the formal structures and the content of the dominant
ethics; the ultimate goals of personal and social evolutions as these could derive from
the assessment of a set of existence meanings. Thus, religion could influence a wide
range of beliefs and social behaviors and outcomes. It could affect the societal
architecture, institutions and mechanisms through a multitude of economic, politic and
attitudinal variables. In particular, religion could affect the social preference for a
democratic societal framework. There can be identified two major channels for such an
influence: one based on the impact exercised by religion on economic development and,
consequently, on the infrastructure of democratic institutions and mechanisms of
human development; the other is via the shape of beliefs, attitudes and behaviors
reverberating into the social dynamics. As Eilinghoff (2003:2) notes: A religious
concept in total can lead an individual to infer a certain course of action towards one or
several specific goals that are usually regarded as the meaning of life.
In regard to the first transmission channel, the research was initiated by Smith ([1776]
1965), Weber ([1905] 1958, 1988, 2002), Azzi and Ehrenberg (1975). Currently, there is
an emerging body of literature studying the connections between religion and economy.
Iannaccone (1998:1466) identifies three major analytical directions in this body of
literature: 1) the interpretation of religion from an economic perspective, seeking to
explain patterns of religious behavior among groups and cultures; 2) the study of
religions economic consequences; 3) the use of the religious doctrinal corpuscle to
promote or criticize different economic policies - religious economics. These studies view
Pages: 32
religion both as independent and dependent variable and conclude that religious activity
could affect economic performance on various levels.
The research field corresponding to the second transmission channel is an
interdisciplinary one with various approaches inspired by sociology, psychology,
anthropology or institutional economics. This research direction was initiated by de
Tocqueville ([1835; 1840] 2001) who suggested that as the American experience shows
religion is compatible with democracy (if the public and religious spheres are kept
separately). Moreover, Durkheim ([1912] 2001) focuses on religion as a factor able to
contribute to social cohesion. Currently, this field is dealing with issues like
secularization and post modernization and their impact on social and politic status.
The present paper is an attempt to empirically analyze the direct impact of religion on
democracy, by recognizing that religion could shape believes, behaviors, ethic structures
and institutional frameworks with consequences on social preference for a democratic
status. The first step of our research consists in defining religion and democracy in order
to provide an operational framework. A particular emphasis is placed on the view of
religion as a cultural artifact and as a generator of meanings, values and purposes of
human existence. Democracy is understood as a consequence of the dominant collective
position regarding the fundamental social equalities and freedoms and hence as a
complex system of entitlements and their borderlines. The second section is searching
for the linkages between these variables. Both direct and indirect transmission channels
of religion influence on democracy via the social ethics and economic development are
briefly discussed.
Furthermore, we test our hypothesis by constructing some measures of different
religious variables based on World Values Surveys data; variables dealing with aspects
such as religious concentration, de facto religious behavior, the perception of the social
role of religion and of its involvement in political life, the approach toward the spiritual
life of the societys members, and the religious denomination, and by taking into
account various measures of democracy (and some specific aspects such as electoral
processes and political participation, the functioning of the government, political culture
and civil liberties) - all these analyzed on cross-national level for a set of 41 countries.
To preview the output of the analysis, it appears that religion matters for democracy, but
the impact exercised by the individual religious variables is non-uniform. Empirically,
the religious variables are correlated with the dependent measures of democracy, but
the robustness of these correlations varies according to the involved methodologies of
estimating the democratic status. Still, overall there is a significant predictive power of
religion in respect to democratic status.
The remainder of the paper is organized as follows: Section 2 seeks to define the key
concepts. Section 3 is searching for possible linkages between the involved variables.
Section 4 tests for empirical evidences, while Section 5 comments on results. The
findings are summarized and some research limits are revealed in the final section.
Pages: 33
Pages: 34
Pages: 35
Pages: 36
Thus, several of the differences between religions emerge from the characteristics of the
nature, level and channels of rewarding / applying of the rewards / costs and from the
mechanisms of their confirmation as well. In order to be efficient, these systems of
rewards and costs which form the structure of the religious ethics should be: a)
consistent; b) self-explanatory; c) appropriate for the socio-economic current
environment; d) able to provide answers and action guidelines for larger set of problems
and actions.
The religious acts influence the social and economic conditions through the associated
ethics. If such acts are fostering traits - like a pronounced work ethics, social and
economic performances (as a sign of being a community of chosen ones or part of such
community), trust, thrift and decisional prudence, charity, tolerance, openness and
acceptance of the specific cultural, sexual or racial differences and so on -, these could
contribute to an increase in the degree of societal openness and also could spur
economic development in support in the adoption and evolution of the democratic
institutions.
In other words, religion can influence, via its specific ethics, the social framework by
stimulating / inhibiting different types of social behavior, in order to promote the
corresponding values of religiosity. It must be noticed that this line of argumentation
does not mainly emphasize the impact of religion on formation of networks, of social
clubs, with their capacity to congregate and mobilize a large number of citizens, but
more on the influence on social beliefs and, hence, behaviors. As McCleary and Barro
(2006:51) explain: A key concept is salvific merit which connects the perceived
probability of salvation to a persons lifetime activities. In some religions, salvific merit
can be earned in this life to enhance the chances for a better outcome in the next life.
For religion to boost up the democratic evolution of a society, it ought to have the
capacity to influence both the positive and negative conditions of democracy. Minimally,
it should: 1) agree that there is a reason for taking individual and collective actions (to
view this world as more than transitory to a better condition which can not be achieved
here by its believers) or, at least, not to discourage such actions (by insisting that these
are futile); 2) encourage the right actions through a well articulated description of
rewards and costs; 3) not promote the negative actions against the non-chosen or
non- believers - actions translated in anti-social acts, such as violence - in addition
recommending the avoidance of such actions; 4) stimulate the social mobility by not
sustaining any type of caste system on the argument that each individual social status is
predetermined (this request can be controversial in the initial stages of democracy
emergence, when religious institutions take position in favor of more rigid social
structures; however, this could change in time due to the emergence of more solid and
fully functional democratic structures and social faith doctrine may be lost in the
process); 5) perceive some actions (charity, self-sacrifices for the good of others and so
on) as active instruments of salvation (assuming, of course, that, in a way or another,
such salvation is possible) and, correlatively, to not promote the idea that extremely
spiritual of physical pain is a correct instrument of salvation; 6) promote a doctrine from
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 37
which it could be concluded that neither autocracy nor anarchy are social systems in
accordance with the Divine will; 7) perceive the democratic social order as a natural
one (given by God) and so any attempt to change such order is not according to the
Divine objectives.
Summarizing, the transmission channels are more complex than the issue of religions
position in respect to the secular authority. Different religions are giving different
answers to the issue of the correct political ideas and consequently are placing their
believers in a wide range of political opinions.
Generally, the main idea is that: By spiritually rewarding networks of mutual aid and
charitable acts, religion lowers the uncertainties of daily life. That is, charity is a form of
communal insurance, which can be efficient if the society has a lot of uncertainty Barro and Cleary (2007:193).
Summarizing: In caeteris paribus conditions, religion could influence both the positive and
negative conditions of the democratic status of a society. Such influence may vary
according to the particular conditions of religious and democratic evolutions, the nature of
religious institutions, the contents of its fundamental dogma, the religious behavior of
individuals as well as to the stance of the relationships between political structures and
mechanisms and civil society. Briefly, at a certain point in time, the relative democratic
performances of a country can be explained among a variety of factors by the nature and
relevance of its religious institutions.
4. Empirical evidences
The purpose of this section is to provide some empirical evidences about the interlinkages between the religious behavior characteristics in modern societies and their
degree of democracy acceptance.
In order to provide such evidences, there is necessary to involve some operational
measures of the religion and democracy variables. For instance, it could be noticed that
various aspects of religion such as religion concentration, the religious practices and
behaviors (including religious devotion), the degree of acceptance of a certain social role
of religion and of its involvement in political life, the approach toward the spiritual life
of the society members, and the predominant religious denomination (and implicitly
the predominant dogmatic structures), all these and others are susceptible to induce
different effects on societal status and on level of human development. So that a first
task of such a study should be the advance of some quantifiable measures of such
variables. Also for characterizing empirically the democratic status is necessary to
decompose the democratic architecture in some of its constituents (e.g. the nature and
mechanisms of electoral processes and political participation, the efficiency of
government functioning, the global levels of political culture and civil liberties) and to
provide a synthetic measure.
Pages: 38
Pages: 39
Pages: 40
The index is defined as the average of five variables from WVS reflecting the dominant
views in a society in relation to the role that can be attributed to religion in political life:
(3)
The shares sr denote the proportion of the respondents which a) are believing that the
religious institutions are able to provide adequate answer to the social problems and b)
which agree to the involvement of these institutions in political life by c) rejecting the
politicians who dont believe in God, d) accepting the influence of the religious leaders
on vote and public governance processes and e) requiring more people with strong
religious beliefs in public offices. These variables are chosen in order to focus as much as
possible on the construction of the index toward the space that religious institutions can
have in public life. There is no specific question concerning the implication of the
respondent themselves in active politics and, thus, these variables are reflecting the
overall attitude of society in respect to the political activism of believers.
The index places an equal weight to each component so that the views about religions
capacity to provide answers to the social issues are considered equally important as the
involvement of religious institutions in the public decisions processes. This implies that
the index should reflect both the legitimacy of the religiously motivated social actions as
well as connections between religious and political spheres.
In order to complete the picture, a fourth index is computed to describe the spiritual
life. This index reflects the shares of the respondents frequently meditating at the
meaning and purpose of life:
(4)
The mo, ms variables denote the shares of respondents thinking often and, respectively,
sometimes, to the sense of life. The weights 1, 2 are arbitrarily chosen to discriminate
between different levels of intensity for such meditations.
In order to reflect more synthetically the religious state, an aggregate index could be
constructed based on the four indexes:
(5)
Since we are not ex-ante imposing any condition about the relative importance of
individual indexes, the ij weights are not predetermined. Several solutions are possible
in order to set the values of the weights. We are considering two of them. The first is to
simply compute an equiponderate version of the global index (ij = 0.25). The second
consists in a factor analysis and in the use of score coefficients as weights.
Furthermore, we grouped adherence in seven categories: Protestant (including
Lutheran, Anglican, Methodist, Presbyterian, Pentecostal, Evangelical, and others);
Catholic; Orthodox; Muslim (this could be roughly breakdown into Sunni, Shiite, and
other types but such delimitation is not taken into account); Buddhist (including Shinto
and Taoist); Hindu (including Jains and Sikhs) and other religions (including Jew,
Jehovah witnesses, Native, as well as other denominations). From the total sample
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 41
representative for the countries selected in the analysis data set, 17.33% of the
respondents had identified themselves as Protestant, 23.59% as Catholic, 14.30% as
Orthodox, 17.20% as Muslim, 4.44% as Buddhist, 3.26% as Hindu and 4.85% as other
religions. The sample represents 69.23% of the total 2005-2008 World Values Survey
sample.
For assessing the direct impact of religion on democracy we are using as dependent
variable the Democracy Index compiled by The Economist which captures various formal
aspects of the democratic status. This index includes several elements relevant to the
adopted definition of democracy, since it seeks to reflect not only the stance of formal
architecture of democratic mechanisms, but also the effectiveness of political
arrangements susceptible to ensure the autonomy of civil society from political
inference and the equality of participation to political processes.
Pages: 42
Democracy
Index
Electoral
Process
Functioning
of
Government
0.952
(0.086)*
Political
Participation
Political
Culture
Civil
Liberties
0.622
(0.064)*
1.140
(0.043)*
0.904
(0.044)*
-0.085
(0.056)
0.052
(0.056)
-0.112
(0.037)*
0.001
(0.029)
-0.034
(0.116)
0.265
(0.174)
0.171
(0.093)***
0.211
(0.054)**
0.169
(0.035)**
0.007
(0.043)
0.149
(0.107)
0.187
(0.066)**
-0.006
(0.071)
-0.073
(0.0836)
-0.153
(0.032)**
-0.035
(0.020)**
-0.046
(0.756)
-1.676
(0.690)**
-0.783
(0.609)
0.609
(0.422)
0.073
(0.010)*
0.082
(0.004)*
0.409
(0.001)*
0.409
(0.001)*
0.013
(0.001)***
0.009
(0.005)***
0.007
(0.003)***
0.01
(0.005)***
Notes: a. The Democracy, Electoral Process, Functioning of Government, Political Participation, Political Culture and Civil
Liberties indexes as dependent variables in Columns 1 through 6 figures are the ones reported by the Economist
Intelligence Unit in 2008 Report. b. The religious indexes are computed according to the methodology described by
the relations (1) to (5). For all the variables, the figures represent the corresponding ranks (1 for the highest value
and 41 for the lowest). Each system treats the log of per capita GDP ranks as endogenous and uses as instruments:
the general government final consumption expenditure; gross capital formation; total value added by services sector;
the Index of Economic Freedom scores for 2007 and the other explanatory variables.
c. The economic variables are from UNCTAD Handbook of Statistics 2008.
d. The Index of Economic Freedom is provided by Heritage Foundation- http://www.heritage.org/Index/.
The most relevant result consists in that religion has a negative impact on democracy.
Both global indexes of religion are negatively and significantly correlated with
democracy indexes, but positively with all the other variables. This finding supports
approaches as the life-cycle religious participation profile in a rational-choice
perspective based on Azzi and Ehrenberg (1975). Since we do not use time-varying data
we cannot conclude directly that this is an effect of a secularization process. However,
we tend in a certain extend to agree with McCleary and Barro (2006:62) that
secularization can be seen as a gradual tendency.
Particularly, it could be noticed that the religion concentration affects statistically
significant all the dependent variables with the exception of the political participation
and civil liberties. One possible explanation is that the absence of religious pluralism
does affect the religious beliefs and behaviors and their translations in political
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 43
Pages: 44
Civil
Liberties
(Freedom
House)
Direct
Democracy
(IDEA)
Turnout
(IDEA)
0.110
(0.004)**
-0.077
(0.018)*
0.44
(0.046)*
0.0001
(0.001)
0.283
(0.028)***
-0.047
(0.031)
0.041
(0.003)**
-0.044
(0.016)*
-0.006
(0.003)**
-0.164
(0.049)*
-0.034
(0.003)**
0.126
(0.032)*
0.177
(0.070)*
**
-0.233
(0.082)*
**
0.400
(0.072)*
0.014
(0.028)
2.682
(0.271)*
1.878
(0.413)*
0.007
(0.003)**
0.227
(0.018)*
0.341
(0.004)*
0.010
(0.005)**
0.206
(0.003)*
0.350
(0.011)*
Notes:
a. Polity IV Institutionalized Democracy is an index provided by the Center for Systemic Peace-Polity IV Project .For
both indexes the figures used in the ranks computation are the ones corresponding to year 2006.
b. The Vanhanen's Index of Democracy represents the data constructed by T. Vanhanen (http://www.prio.no/
CSCW/Datasets/Governance/Vanhanens-index-of-democracy/Polyarchy-Dataset-Manuscript/) in his study of
polyarchy. The data used here are the ones for the last electoral cycle in each country (generally from the 2000
electoral processes).
c. The Political Rights and the Civil Liberties indexes are reported by Freedom House The figures are corresponding
to the 2007 values.
Pages: 45
d.
The Direct Democracy index values represents own computation based on information supplied by the Institute
for Democracy and Electoral Assistance.
e. Turnout represents the average political participation rate computed based on the voting age population as well
as the number of registered voters as these are reported by the Institute for Democracy and Electoral Assistance
(www.idea.int).
All the regression have been carried out by adding to the instrumental variable already
used a dummy for the ex-communist countries in order to control for the heterogeneity
of the methodologies involved in the estimation of democracy. While the communist
period had affected all the components of the religious life, this effect appears to be only
a temporary one, since there could be found a considerable rebound in religion in the excommunist countries. However, from the institutional point of view, the impact of the
communist regime could be more persistent due to the necessity of institutional
reconstruction of religion as well as of democratic infrastructure. It could be argued that
for the ex-communist countries the framework of the interrelations between religion
and democracy was substantially modified and so the consistency of the use of different
democracy measurement methodologies could be altered.
The results are quite puzzling. For instance, there could be find a negative and statistical
significant association between the religious global indexes and Polity IV project
measures but a positive and yet significant one between these indexes and Vanhanen's
Index of Democracy and respectively Political Rights and Civil Liberties indexes provided
by Freedom House. Several explanations could be advanced for such output. A short list
of them could include:
The Vanhanen's Index of Democracy is focused on political participation and electoral
processes so that it appears that country with medium to high level of religiosity such as
Italy or Turkey but also others like the ex-communist Bulgaria, Poland, Romania or
Moldova displays more or almost the same degree of democracy that United States,
Canada or Japan;
Despite their more complex methodology, with an important emphasis on institutional
elements of democracy, the Political Rights and Civil Liberties displays more or less
similar features with almost the same outliers for countries like Italy, Bulgaria, Poland,
Slovenia, Ukraine or Ghana and Trinidad and Tobago which are reported to have close
values of democratic characteristics close to the mature Western democracies;
From such observations, it could be argued that the mentioned measures of democracy
are able to capture the institutional aspects of political architecture and the electoral
processes but is still an open question about how much these measure are reflecting the
de facto functional infrastructure of a democratic society.
Secondly, it appears that there is a strong and stable negative correlation between the
Index of Religion Social Role and the various measures of democracy: the more able is
religion considered to supply answers to the social problems in a country, the less this
country democratic appears to be regardless the methodology used to estimate the level
Pages: 46
of democracy. Through, it could be presumed that the religion monopole on societal life
acts like an inhibitor for the development of the democratic institutions and
mechanisms. Of course, this result could not be use for answering to the question if such
loss in representative democracy quality is not counterbalanced in religious countries by
the adoption of some alternative forms of civil society participation to social and
political issues.
Thirdly, a correlative positive and robust relationship is the one manifested between the
Index of Religious Behavior and the involvement measurements for the level of
democracy: the more the citizens of a country are involving themselves in organized
religious practices, the more are they are willing to accept the specific institutional
arrangements of an representative democracy. Such an output should be correlated with
the less clear finding that the extent of religious concentration is negative correlated
with the democracy indexes (except for the Vanhanen's Index of Democracy and Freedom
House Civil Liberties ones). Thus, is there is achieved a certain degree of religious
competition in the conditions of a sufficient demand for religious products it is less
possible that a single denomination with possible non-democratic doctrines to influence
a large number of civic behaviors. Of course, one crucial aspect concerns the nature of
the doctrinal corpuscle and practices of a significant or dominant religious
denomination. For instance, a country like Norway with a 62.6% share of Protestant
denomination subsequently displays a high level of democratic status regardless the
methodology used to describe this status. The religious concentration as a single
variable has only a limited capacity to explain to relative preference for democratic
systems without a more detailed description of the religious demand and supply.
Fourthly, there is a certain negative interrelationship between the Index of Spirituality
and democracy indexes (again with the exception of Vanhanen's Index of Democracy):
the more spiritual issues are included in the cultural paradigm, the less a society prefers
the representative democracy. This statement does not imply that democracy is only a
pragmatic social arrangement rejected by more spiritual societies. There could not be
a democracy without a social acceptance of its fundamental values directly incorporated
in paradigm up to a certain threshold and as a consequence without a conscientious
effort to provide senses for personal and social existence . Such result could be more
related to a shift toward individualistic points of view and more fragmentation of the
societal coagulation processes with the increase in the focus in here and now
approaches of life. Again, such a thesis should not be considered in an absolute sense. On
one hand, peoples could be stimulated to reflect more on the meaning of life in
conditions of a short life expectation, a low level of rational education and in situations
when they feel that there are few opportunities for personal fulfillment. On the other
hand, is religion itself which could provide answers on this topic so that if there is a high
level of religiosity it compensates for the personal efforts in attributing sense to our
lives. However, this last argument is less supported by data since almost all countries
with high level of religiosity have also at least medium to high levels for Index of
Spirituality (maybe with the exception of Mali).
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 47
Fifthly, one of the most striking results insensitive to the changes in the methodology of
measuring the democracy is that Protestantism is a clear ground for both representative
and direct types of democratic regimes and that the relative preference for democracy
diminish in non-Christian denominations countries. Of course, such outcome should be
corrected with the unequal weight of different religions in the data sample but still there
seems to be a certain evidence for the Weberian argument.
In the context of these results, there appears that there are two other particular issues.
The first issue is related to the fact that all mentioned indexes are measuring, in a way or
other, the representative type of democracy. Or it is interesting to see if the direct
democratic forms are also interrelated with religion. Column 6 of Table 4 reports the
linkages between the religion indexes and a direct democracy index build on data
concerning the legal provisions about referendums, citizen initiatives and recalls at
different levels, the topics which could be subject of such referendums and the usage of
referendum mechanisms since 1980 provided by Institute for Democracy and Electoral
Assistance. By direct democracy we understood a system in which the social decisional
sovereignty is lodged to the civil societies in all the relevant matters for the social
processes. The associated mechanisms are: (a) referendums, which are votes on a specific
single issue or piece of legislation (; (b) citizen initiatives, whereby citizens can propose
new legislation or constitutional amendments; and (c) recall, under which citizens can
force a vote on whether to oust an incumbent elected official. The common characteristic
of these mechanisms is that they all place more power directly in the hands of voters, as
opposed to elected representatives. But it should be noticed that in practices this pure
type of democracy is in the best case intent but not realized as a fully operational
societal framework. Rather the societies which emphasis such democratic design are
deliberative democracy which incorporates elements of both direct democracy and
representative democracy. So that, the Direct Democracy Index should be seen as a
measure of the degree in which a society accepts to incorporate direct democracy
elements into its political mechanisms.
The output of the system suggests that there could be found a negative interaction
between economic development and the relative preference for direct democracy and,
correlatively, a positive one with the religiosity. The first finding could be partially
explained by some historical factors specific for mature economies and societies. More
exactly, it could be argued that the traditional democracies has been evolving like
representative democracy projects and the outcome was stable enough over time in
order to be preserved (with various adjustments). In the mean time, we could presume
that the emergent societies and / or the not fully consolidate democracies could find in
the direct democracy mechanisms a way to compensate and / or to reduce the costs of
the institutional construction for the infrastructure of a representative democracy but
we do not have data to support this idea.
The second finding could be enlighten by the observation that, since that Switzerland is
not included in the dataset, countries with high Direct Democracy Index are represented
by Orthodox and Catholic dominant denomination ones (like Peru, Serbia, Guatemala,
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 48
Italy or Poland and Slovenia). A possible explanation could subsist in the emphasis that
both Orthodoxy and Catholicism are placing on communion a an especially close
relationship of Christians, as individuals or as a Church, with God and with other
Christians, relationship which could be translated into a strong communitarian sense
which serve as ground to the preference for direct form of the community consultation
in a larger spectrum of social and political issues. Of course, this argument is not
complete since a medium to high level of index could be found for non-Christian
Buddhist dominant countries like Thailand or Korea. However, such line of
argumentation could be supported by the observation that according to our results a
higher level of religious concentration stimulates the adoption of direct democracy
elements. Thus is could be argue that this concentration is a stimulus for the
communitarian spirit and for a larger autonomy of local structures. Supplementary, the
Index of Religious Behavior and the Index of Religion Social Role appear to be negative
correlated with the Direct Democracy Index which could be explained by the importance
of individual actions in more social secularized societies. In other words, there is a
certain consistency in assuming that factors as destiny or predetermination are less
important in explaining the output of personal acts and also that religion could not
provide answers to the social problems and simultaneous to give up only with
limitations the control over the social decisions to representatives eventually perceived
to form a political aristocracy (or a benevolent technocracy or a omniscient
bureaucracy or a similar term designed to referrer to an exogenous specialized
structure).
Overall, it seems that a religious ethic which emphasis systematic correct interactions
between individuals and thus shorter societal interspaces is a strong support for the
emergence of direct democracy mechanisms. But the analysis is only partial since it not
includes a balanced comparative approach of the relative efficiency of religious and nonreligious networks through which the citizens gain and exchange information and
engage themselves in social actions.
A second particular issue is one about the effectiveness of the democratic mechanisms,
effectiveness which in a narrow sense could be synthesized by the degree of
participation to the political processes. Column 7 of Table 4 shows the correlations
between the religious indexes and the average political participation rate based on the
data provided by the Institute for Democracy and Electoral Assistance. It results that an
increase in the religiosity is associated with a higher participation to the electoral
processes. Of course, this outcome should be more clearly explained. For instance, Djupe
and Grant (2001: 311) are arguing that to the extent that religion is positively associated
with political participation is much more related to the role of churches in recruit[ing]
parishioners to participate in politics and in creating the perception among members of
a common set of political norms and expectations that would encourage participation.
Indeed, churches are civil (i.e. non-political) structures and there could be through
their specific channels a mobilization for the political life together with other nonreligious organizations of the civil society. And some churches have significant potential
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 49
for social mobilization. Supplementary, it could be argued that if there are some
monopolistic characteristic of the religious supply (i.e. a high denominational
concentration) it is much easier for a particular denomination to mobilize a large share
of the citizens and to support their involvement in politic life (of course, if the utility of
such denomination is increased by the maximization of the degree of participation for its
believers). Our data does not support such an argumentation since there appears to be a
negative (and not yet consistent) relationship between the Index of Religious
Concentration and the participation. But there is only a point. Another one is the extent
to which a certain set of religious beliefs and practices are encouraging the individuals
for self-engagement in the political processes. As Patterson (2005:149) finds in the
examination of political life in Latin America: the primary cleavage in political
participation was not between Catholics and Protestants but between the devout and
the not devout. Still, in our data appears a manifestation of both type of cleavages since
the denomination variable is positively and statistical significant correlated with the
political participation. In highly developed countries which largely are Protestants or
Catholics (or a combination between these two denominations), the electoral apathy is
(up to a certain level) a wide phenomena. Au contraire, in the new emergent
democracies (which have a lower share of Christian denominations) the electoral
processes are still attracting the citizens. Since in our opinion the electoral absenteeism
is the output of a multi-periodic process, a cross-section analysis of religion and
democracy could not explain why the attractiveness of democracy fades out in the
mature societies. We could only status that: 1) there is such a process and 2) this affects
especially the historical democracies (based in Protestants or Catholics countries).
It is interesting to note that there was found a positive and relevant connection between
the religious behavior and the tendency to vote: the more somebody acts in a religious
formal way the more is likely to participate in electoral processes. This result is
somehow different from other findings in literature. For instance, Thornton and Kent
(2009:9) conclude for a set of Latin America countries that: Church attendance was not
a predictor of the other forms of participation (voting, contacting, or campaigning).
Such a difference could be explained by both the facts that our dataset incorporate also
non-Christian countries with a different pattern of religious behavior (more exactly,
with a different content of what it means a religious life) and also that there are
significant differences among the countries in the content of political participation
beyond the existence of eventually formal similar electoral mechanisms.
Relatively surprising, the participation decrease with the increase of Index of Religion
Social Role: despite the fact that citizens believe that religion could provide answers to
social problems and that they are requiring more members of public authorities with
religious beliefs, they are involving themselves less in politic occasions. But such a
paradox could be only a partial one since it could be recalls that a higher level of the
belief in the social role of religion could have an adverse effect on democracy and thus
could reduce the opportunities to exercise the right to vote.
Pages: 50
The Index of Spirituality seems to act on the same way on political participation as on
direct democracy: an increase in the concerns about the meaning of life stimulates a
higher involvement in the political life as a result of a proactive attitude toward the
social problems.
Pages: 51
Third, there is a certain negative relationship between our Index of Spirituality and
democracy but this is less clear and stable among the various measures of the
democratic aspects;
Fourth, there is a certain cleavage between the societies with a dominant Protestant (and
for certain aspects Catholic) denomination and the others in respect of democratic
achievements.
These findings require a deeper analysis for a better understand of the involved
mechanisms. However, these suggest that there is a case for substantial inter-linkages
between religion and democracy.
References
Acemoglu, D., Johnson, S., Robinson, J.A., Yared, P., 2005. Income and Democracy, NBER
Working Paper Series, 11205, Available at: http://www.nber.org/papers/w112
05.
Azzi, C., Ehrenber, R.G., 1975. Household Allocation of Time and Church Attendance.
Journal of Political Economy, 83 (1): 2756.
Bader, V., 2003. Religious Diversity and Democratic Institutional Pluralism, Political
Theory, vol. 31, no. 2 (April 2003): 265-294.
Barro, R.J., 1996. Democracy and Growth, Journal of Economic Growth, vol. 1 (March):
1-27.
Barro, R.J., McCleary R.M., 2007. Political Economy and Religion in the Spirit of Max
Weber, in Nee, Victor and Swedberg, Richard (Eds.) On Capitalism, Stanford:
Stanford University Press, ISBN 978-0-8047-5664-8
Berger, Peter L., 1967. The Sacred Canopy: Elements of a Sociological Theory of
Religion. Garden City: Doubleday, 1967; New York: Anchor Press, 1990.
Bruce, S., 2006. Did Protestantism Create Democracy, in Anderson, J. (Ed.) Religion,
Democracy and Democratization, New York, NY: Routledge.
Djupe, P.A., Grant J.T., 2001. Religious Institutions and Political Participation in
America. Journal for the Scientific Study of Religion, 40(2): 303-314.
Driessen, M., 2010. Religion, State, and Democracy: Analyzing Two Dimensions of
Church-State Arrangements, Politics and Religion, vol. 3, no. 1 (April 2010): 5580.
Durkheim, D.., [1912] 2001. The elementary forms of religious life. (Cosman, Carol
transl.) New York: Oxford University Press, ISBN 0-19-283255-7.
Eilinghoff, C., 2003. Religious Information and Credibility. German Working Papers in
Law and Economics, Paper 8, Volume 2003.
Hofstede, G., 1997. Cultures and Organizations (revised edition). New York: McGraw Hill.
Iannaccone, L.R., 1988. A formal model of Church and Sect. American Journal of
Sociology, Supplement, 94:241-268.
Pages: 52
Kunin, S.D., 2003. Religion. The modern theories. Edinburgh University Press: Edinburgh
McCleary, R.M., Barro R.J., 2006. Religion and Economy. Journal of Economic
Perspectives, 20(2):4972.
Shen, J.-G., 2002. Democracy and growth:An alternative empirical approach, BOFIT
Discussion
Papers,
No.
13,
Available
at:
http://129.3.20.41/eps/mac/papers/0303/0303008.pdf.
Smith, A., [1776] 1965. An Inquiry into the Nature and Causes of the Wealth of Nations.
New York: Modern Library.
Smith, K.M., 2009, Differentiated Trust in Democratic Institutions Among Religious
Minorities: Does the Size of the Largest Religious Group Matter?, International
Journal of Sociology, 39(1): 30-48.
Stark, R., Bainbridge W.S., 1996. Theory of Religion. Rutgers University Press, ISBN
0813523303.
Tocqueville, A. (Charles-Henri Clrel) de, [1835 /1840] 2001.Democracy in America.New
York: Signet Classic / New American Library, ISBN 0-451-52812-3.
Weber, M. [1905]. Die Protestantische Ethik und der Geist des Kapitalismus. English
translations: The Protestant Ethic and the Spirit of Capitalism. Unwin Hyman,
London & Boston, 1930; New York: Scribner's Press, 1958; Gloucester, Mass.:
Peter Smith , 1988; New York: Penguin Books, 2002; American Studies archive
from
University
of
Virginia,
Retrieved
November,
20,
2009
(http://xroads.virginia.edu/~HYPER/WEBER/ cover.html); Max Weber Reference
archive, Retrieved November, 24, 2009 ( http://www.marxists.org/reference/
archive/weber/index.htm); Abukum Academy, Max Weber archive, Retrieved
November,
21,
2009
(http://www.ne.jp/asahi/moriyuki/abukuma/weber/world/ethic/pro_eth_fram
e.html); Max Weber archive at the Pfeiffer University, Retrieved November, 24,
2009 (http://media.pfeiffer.edu/lridener/DSS/Weber/PECAP. HTML); Max
Weber
texts
at
SocioSite,
Retrieved
November,
25,
2009
(http://media.pfeiffer.edu/lridener/ DSS/Weber/PECAP.HTML).
Verba, S., Schlozman, K.L., Brady, H.E., 1995. Voice and Equality: Civic Voluntarism in
American Politics. Cambridge: Harvard University Press, ISBN 0-674-94293-0.
World Values Survey 2005 Official Data File v.20090901, 2009. World Values Survey
Association , Retrieved October, 15, 2009 (www.worldvaluessurvey.org).
Aggregate File Producer: ASEP/JDS, Madrid.
Pages: 53
Department of Architecture, Built Environment and Construction Engineering ABC, Politecnico di Milano, Piazza
Leonardo da Vinci 32, 20133 Milano, Italu.
22
Department of Architecture, Built Environment and Construction Engineering ABC, Politecnico di Milano, Piazza
Leonardo da Vinci 32, 20133 Milano, Italu.
23
Pages: 54
Keywords:
JEL Classification:
1. Introduction
Globalisation is certainly not a new phenomenon, and in many periods of the last
century it reached very high levels, ones even comparable with those of today. What is
new is the long-term, contemporary acceleration of many parallel integration processes
which reinforce and integrate each other in multiple ways. For almost thirty years,
international trade has been steadily growing at a rate which is double that of world
GDP. Foreign direct investments (FDI), in their turn, have grown at rates which are
double that of international trade, and four times higher than world GDP. Most of these
investments are directed towards developed countries (80 percent in the years 19861990, around 60 percent in 1993-97) and seem particularly attracted by accelerations in
economic integration processes: in fact, the EU countries which led the process of
creating the Single Market in 1991-92 received up to 50 percent of world FDI (UNCTAD,
1997; Camagni, 2002).
Globalization is not an unequivocally defined process, directly measurable through
official statistics like GDP or international trade, or indirectly computable through single
figures on migration and population ageing; it is a multifaceted synthesis of a vast
number of factors of different nature economic, social, technological, institutional
difficult to find in official data. Moreover, globalization is not a state of the world
economy, but a process involving social, institutional, economic and technological
changes bundled together in such a way that a clear distinction between causes and
effects is difficult to draw (CEC, 2009).
In the context of this paper, globalization is mainly interpreted as a process of
internationalization of production and markets which can take various forms like
increasing international trade or increasing foreign direct investments all of which
give rise to the growing integration and interdependency of European economies with
other main world economies. According to this definition of globalization, its impacts are
mainly of an economic nature and associated with long-run structural changes in the
economy caused by the integration and internationalization of production and markets.
Much theoretical and empirical work has been developed on globalization, trying to
capture different effects of the quali-quantitative changes imposed by the integration of
markets through either multilateral or regional liberalization policies (Panagariya,
2000); new international trade patterns which see more and more developing and
emerging countries as exporters of manufacturing goods, thus forcing industrialized
countries to change their specialization towards high quality goods and, mainly services
(Bergoeing et al. 2004; Kucera and Milberg, 2003), new composition of intermediate vs.
final goods traded at international level, also as a result of multinational firms new
strategies (Yi, 2004; Hummels et al., 1998 and Hummels et al. 2001; Hanson et al. 2005),
new location patterns of foreign direct investments and consequent new growth
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 55
opportunities for developing economies (Hansen and Rand, 2006; Lall and Narula, 2004;
Moran et al., 2005), migration trends and international trade flows (Soubbotina, 2004;
Lucas, 2008), represent some of the main issues treated in the recent literature. From
the perspective of the above mentioned studies, though, globalisation can be regarded as
neutral for what concerns its spatial effects: opportunities and threats may look
equivalent and specular. A number of good reasons exists, however, for claiming that a
regional perspective is instead fundamental in order to understand the real economic
effects of globalization, and that conceptual and empirical analyses at regional level are
fundamental (Cooper et al., 2007; Capello et al., 2011).
Globalization provides greater access to other countries markets and resources, while
granting other countries greater access to the European market. Overall, this process is
mutually beneficial. However, the benefits are not evenly distributed across the
European territory and economic sectors and the consequence of increasing
globalization is the creation of additional pressure on local economies, obliged to face
tougher competition (Cooper et al., 2007). The research question of this paper is how
regions most exposed to globalization face stronger competition, and whether specific
success factors explain their growth patterns with respect to closer local economies.
Open regional economies are theoretically more dependent on innovation, required to
face competition, and at the same time generated by linkages with international firms
(Gorodnichenko et al., 2008); on the presence of high-value functions, as important
factors to attract additional high-value functions (Kenney and Florida, 2004); on high
quality human capital, that allows to keep control over processes of tasks unbundling at
the international level, that de-localize mostly low-value tasks (Baldwin, 2006); on the
attraction of FDI, expected to be growth-enhancing by allowing the incorporation of new
inputs and foreign technologies in the production function of the recipient economy and
by increasing the productivity of already existing input factors of the recipient economy
through labour training and skill acquisition (Beugelsdijk et al., 2008; Borensztein et al,
1998; De Mello, 1999).
Whether the presence of these specific factors (e.g. innovation, high-value functions,
high quality human capital and FDI) is more important for open regional economies than
for closed economies is investigated in this paper through an empirical analysis. The
latter is conducted on the entire European territory using a uniform dataset for 259
NUTS 2 regions of the 27 European member countries, excluding the overseas French
departments (Guadeloupe and Martinique), the Azores, Madeira and the Canaries. At the
empirical level, the main difficulty is the availability of a reliable dataset which
comprises all NUTS 2 regions of the 27 member countries, and of the identification of
global regions, i.e. those regions with an economic structure more open, and therefore
more sensible, to international trade.
The structure of the paper is as follows. The paper first presents an operational way to
identify different kinds of regions according to their degree of openness to the global
world (sec. 2.1). The taxonomy will turn to be useful for our empirical analysis given the
different growth patterns that the different groups of regions show (sec. 2.2). The paper
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 56
then presents an interpretative analysis, run with the aim to test whether regions with
different degrees of openness to the world economy perform differently. In particularly,
the aim of the interpretative analysis is to determine whether the role played by each
growth factor changes across regions, and in particular across regions with different
degrees of openness to the rest of the world (sections. 3 to 5).
The first result evidences that the globalization process per se does not explain regional
performance, since regions with different degrees of openness have no particular
regional dynamics, ceteris paribus, once structural factors are taken into account.
Moreover, our results show that the impact of each success factor on regional
differential growth does not normally change among groups of regions. A higher average
regional growth rate in global players with respect to regional and local ones seems
hence to be mainly dependent on the regional endowment of success factors especially
those with high impacts on growth rather than by differentiated marginal effects
among groups of regions (sec. 6).
Pages: 57
rather than a national growth pattern. The feature shared by all these concepts is the
idea that one way to be integrated into the global economy, and to gain advantages from
it, is to comprise international high-value functions, qualified human capital, increasing
returns in production activities, and physical accessibility.
The second dimension on which to measure a local economys degree of integration i nto
the world market is a pure economic dimension captured by the degree of that local
economys specialization in activities that are particularly open to international markets.
This dimension explains the capacity of a region to grow by virtue of the presence in it of
dynamic open sectors. It captures a MIX effect of a traditional shift-share analysis
(Perloff, 1957; Perloff et al., 1960).
On the basis of these two approaches, global players are identified as:
regions with high functional/territorial integration with global processes;
regions with high market integration, i.e. specialized in competitive and dynamic
open sectors (sectors in search of new markets, more open to competition, and
better able to gain advantages from world competition).
Only those regions well endowed with physical connections and possessing the
appropriate specialization in competitive and dynamic sectors have the potential to be
global players, these being defined as regions where globalizations impact is felt first
and most strongly. Global players are able to benefit from globalisation if they can
exploit the opportunities offered by globalization, minimize the risks associated with it,
and turn threats into opportunities.
1 Global players
2 Regional players
4 Pure gateways
3 Local players
On the basis of these two dimensions, four main theoretical regional types are identified:
1. Global players. These are regions at the core of globalization processes: they are
structurally open and have all the necessary physical and functional linkages with the
rest of the world; moreover, they are specialized in sectors which are open and
growing, so that their role in world trade flows and FDI attractiveness is maximum.
Pages: 58
These regions are therefore expected to be able to lead Europe and drive patterns of
response to globalization also for the other regions of the EU.
2. Regional players. These regions are specialized in open growing sectors but have
below-average physical and functional connectedness with other areas in the world.
These regions are therefore expected to take advantage of their specialization, but
they are also expected to be somewhat penalised with respect to global regions
because their good sectoral mix does not take advantage of a strong and efficient
territorial settlement structure, and does not exploit the agglomeration advantages
guaranteed by a city-region. The economic dynamics of these areas are expected to be
due to a MIX effect deriving from the presence in the region of sectors that are more
dynamic and more open than average at regional level because of increasing demand
in those sectors. The label regional is attached to these players because their
sectoral specialization would allow them to play a worldwide role, but, given their
lack of an urbanised settlement structure, they normally have to resort to global
regions as gateways to world markets. The term regional is hence to be understood
in its trade literature meaning, which interprets Europe as a region of the world. At
the same time, the term recalls the limited physical accessibility to and from the
world.
3. Local players. This category consists of regions which have neither the
functional/territorial elements to connect with the world nor the appropriate
specialization in open growing sectors. These regions are rather peripheral to
globalization processes and will hence be used as a control category by all the
analyses conducted in the following sections. Trends that pertain to globalization
forces are expected to be limited in this category. We label them local players
because their markets are expected to be local, i.e. normally limited to their own
region and, possibly, country;
4. The last category, gateway regions, are regions with a puzzling behaviour, i.e. regions
with structural openness but specialized in closed sectors. This strange behaviour
does not appear to exist in the reality, as evidenced in Capello et al. (2011) and in
Fratesi (2011), where the statistical analysis leading to the empirical identification of
the category of actual regions is presented in details.
The three first categories will be used in the following sections for the empirical
analysis.
Pages: 59
Since in the years 2000s the Eastern European Countries (i.e. the New 12 Member
States of the EU) significantly outperformed on average their western counterparts, and
this could induce a bias in the analysis, we chose to present table 2 also for the two
groups separately24.
In the first period of time, i.e. 1999-2002, global players significantly outperformed the
other types of regions in terms of real GDP performance. This was the case of both
regions in the Old 15 member countries and in the New 12 ones. Interestingly, in
Western regions regional players are the second performers, close to global players,
whereas in Eastern regions global players by far outperform local players (the second
best performers) as well as regional players. In the second period of time (2002-2005),
global players were again the best performers among European regions, but not
significantly so overall and in Western countries. In Eastern countries, by contrast, the
growth rate of global players was still significantly higher.
National effects were controlled for once regional growth had been analysed with
respect to its national average. The results show that global players have been leading
their respective countries in terms of growth rates; being a global player appears
significantly to increase the possibility of being a region benefiting from a period of
rapid globalization and to lead the country in terms of growth. In Eastern countries, the
differential of global players with respect to their countries is high and significantly
different from that of the other regions in both periods. In Western countries the
differential growth rate is larger in both periods, but significant only in the first period.
These results show that global players have a higher capacity on average to pro-act and
re-act to global trends. A more in-depth analysis allows us to determine whether this is
true for all global regions, and to identify which endogenous success factor positively
affects the endogenous capacity for growth in a period of globalization.
In an aggregate analysis, global regions record higher performance rates and
outperform all other regions in the country. Two main questions arise in this regard:
first, whether all global regions have high performance rates and, by the same token,
whether all local players have low performance rates; second, which local assets explain
these performances and especially whether local success assets differ among the groups
of regions. The answers to these questions have important policy implications because
they can help in devising ad-hoc, place-specific (in the words of the Barca Report, Barca,
2009) policy recommendations intended to reinforce those elements of the territorial
capital on which the competitiveness of each single regional type depends.
24
Notice that, with the inclusion of national dummies, this separation is no longer necessary in the following sections.
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 60
Global players
3.17
2.76
2.47
2.09
2.06
0.37
-0.39
-0.83
7.33***
0.16
-0.15
-0.41
4.39**
Global players
2.54
F
8.21***
F
4.04***
1.91
1.79
1.74
0.22
0.12
-0.16
-0.89
5.08***
0.03
-0.13
-0.24
0.97
Global players
6.22
F
24.28***
6.78
4.43
3.93
9.28***
1.54
-1.08
-0.48
9.89***
0.77
-0.21
-1.30
6.64
Pages: 61
Since the globalization index includes, among other things, also extra-European FDI, the latter has been excluded
from FDI variables in order to avoid multicollinearity.
25
From a technical point of view, this requires the inclusion in the regression equations of two sets of dummy
variables: the first set consists of one dummy variable for each typology of EU region, while the second made up of
country dummy variables, as well as possible interacted effects.
26
Pages: 62
recently be termed territorial capital, which consists of material and intangible, private
and public, soft and hard elements (Camagni, 2009).
In the vast literature on regional growth, a high number of factors are identified and
come from different paradigm shifts that took place over time:
from development (or even location) factors to innovation factors (Cappellin and
Nijkamp, 1986; Crescenzi and Rodriguez-Pose, 2011; Boschma and Lamboy,
1999; Boschma, 2005; Boschma and Martin, 2010);
from hard to soft factors consisting of either intangible, atmosphere-type, local
synergy and governance factors (Becattini, 1990; Camagni, 1991), or human
capital and knowledge assets (Foray, 2000).
For this reason, a rather differentiated set of local assets were chosen, namely: 27
the degree of innovation of regions (inno), expected to affect positively the
regional growth rates, as a large body of literature suggests (e.g. Howells, 2005;
Johannson and Karlsson, 2009; de Groot et al., 2009; Audretsch and Aldridge,
2009). Innovation was proxied by the share of human resources in science and
technology;
physical density (den), measured by the endowment of roads per square
kilometre as a proxy for the intensity of use of regional land. Measured as such,
this variable is not thought to capture infrastructural endowment and is
expected to have a negative sign when congestion effects prevail;
the endowment of human capital (humcap), measured by share of university
professors reported by the labour force survey (Park et al, 1925; Jacobs, 1961;
Thompson, 1965; Lucas, 1988; Karlsson et al., 2009);
a balanced urban system, with the presence of cities (Beguin, 1988), as the
tradition of the Christaller and Loschs models suggest. This is proxied by a
series of dummies on the settlement structure of regions, developed within the
ESPON programme; 28
FDI penetration in a region as a measure of regional attractiveness (fdi). We
only considered FDI originating from within Europe, in order to avoid
endogeneity with the globalization index built with extra-European FDI
(Baldwin and Martin, 1999; Casi and Resmini, 2011; OECD, 2007);
an exogenous but very important variable explaining regional differential
growth is the presence of public funds (pol) which, because they are aimed at
either demand-side support or supply-side development, should yield positive
27
All independent variables were lagged in order to reduce problems of endogeneity and reverse causation.
More precisely, rural regions are those regions with a population density < 100 / km sq. and a centre > 125,000 inh.
or a population density < 100 / km sq. with a centre < 125,000; urban regions are those regions with a city of between
150,000 and 300,000 inhabitants and a population density of 150 300 inhabitants / km sq.; or a lower population
density (100-150 inh. /km) with a larger centre (>300,000). Agglomerated regions, instead, are those regions with a
city of > 300,000 inhabitants and a population density of > 300 inhabitants / km sq. or a population density of 150
300 inhabitants / km sq.
28
Pages: 63
growth effects if funds are wisely spent. We used structural funds expenditure
per capita as a proxy for this factor, since national transfers are not available.
Moreover, the model includes national dummies which account for the fact that some
countries (e.g. the New Member countries of the EU) have grown considerably more
than the others. Since some countries do have just one region at Nuts2 level, these
countries, were the dummy would explain everything, are excluded from the analysis.
The model estimated was therefore the following, where the only coefficient with
expected negative sign is the one for excess physical density:
regrowthr 0 1natgrowthr 2innor 3denr 5 polr
(10)
6humcapr 7 fdir 8city r
The dependent variable for the empirical model is the real GDP growth rate at Nuts2
level for all European regions in the period 2002-2005. This period was characterized by
rapid globalization and was far from the big economic crisis which started in 2007.
Estimations with a longer time-span (i.e. 2000-2005) were also attempted and provided
results consistent with those presented here, but were not chosen because for many
independent variables only the value in 2000-2001 was available and the regressors
needed to be lagged in order to reduce the possible endogeneity problems.
In order to test the role played by globalisation on regional growth, as previously
mentioned we will first introduce a dummy accounting for globalisation stance into eq.
(1) (i.e. a dummy for the typologies) and then interacted this dummy with the other
explanatory variables in order to test whether the estimated coefficients varied across
types of regions. 29 This strategy makes it possible to assess, on the one hand, whether
global regions are, ceteris paribus, more dynamic than other kinds of regions, and, on the
other hand, whether success factors have different impacts on growth according to the
degree of world integration of regional economies.
The beta coefficients are the regression coefficients obtained by first standardizing all variables to have a mean of 0
and a standard deviation of 1, and are presented here in order to allow some comparison among coefficients of
variables with very different scales.
30
Pages: 64
regions thrive in an age of globalization. Physical density has a negative and significant
coefficient; the excessive density of some areas, which are consequently subject to
congestion diseconomies, is captured. Public policy support has a positive but slightly
insignificant standardized coefficient (0.08), which is in line with an ample literature on
structural funds where debate is ample and still not conclusive on their effectiveness
(Boldrin and Canova, 2001; Rodrguez-Pose and Fratesi, 2004; DallErba and LeGallo,
2008; Mohl and Hagen, 2010). Human capital has a positive and significant coefficient,
stable in other regressions.
Obs
R2
F
Moran's I
Spatial Error
Lagrange multiplier
Robust Lagrange multiplier
Spatial Lag
Lagrange multiplier
Robust Lagrange multiplier
0.177
0.025 * *
-0.177
0.081
0.129
-0.099
Included
246
0.7104
84.26
0.719
sig.
0***
0.219
0.085 *
Model1 (complete)
stand.
coeff.
p-value
0.141
-0.198
sig.
0.085 *
0***
0.075
0.134
0.254
0.07 *
0.070
0.009 * * *
0.041 * *
-0.097
0.045 * *
***
0.001 * * *
significant
Included significant
0.472
246
0.7134
59.15
0.578
0.563
1.99
0.734
0.158
0.392
2.26
0.703
0.133
0.402
1.301
0.044
0.254
0.834
1.689
0.132
0.194
0.717
Pages: 65
growth but, at the same time, regional growth can be an attractor for FDI, hence posing
the possibility of endogeneity. The variable is hence included lagged with respect to the
period of estimation, but if this reduces possible endogeneity, it does not necessarily
wipes it out if there is serial correlation. Since the inclusion or exclusion of FDI does not
alter the results for the other coefficients (Table 3), and in the impossibility to have time
series of regional FDI for European NUTS2 regions, we chose to keep FDI in the
regressions, trusting the literature which finds positive impacts of lagged FDI on
regional growth (Borensztein et al., 1998; de Mello, 1999; Mullen and Williams, 2005;
Beugelsdijk et al., 2008), though other literature is critical and this result is still debated
(Carkovic and Levine, 2005).
The final explanatory variable is the settlement structure, which was used as a proxy for
agglomeration economies. Between the various settlement structure dummies, the one
that has the highest explicative power is the rural dummy, which comes out negatively
related to regional growth and strongly significant; this means that the lack of large
cities is of detriment for regional growth.
The results were tested for spatial effects using various matrices, including a
standardized distance matrix and a standardized distance matrix with a threshold. All
tests rejected the presence of spatial autocorrelation in the regressions, and the need to
use a spatial lag or spatial error model, probably because the inclusion of national
dummies already captures all those spatial effects which are more country-dependent
than proximity dependent.
Pages: 66
A second question is whether the different success factors have different impacts on
regional growth for regions with different degrees of world integration. Two analyses
were run, in sequential order.
The first analysis split the region sample into two groups, global and regional players on
the one hand - being regions with an integrated economy (although with different
intensity) - and local players on the other, and determined whether the estimates were
statistically different using a Chow test. Table 5 presents the results of a regression
analysis performed separately on the two groups: global and regional players in the
middle, and local players on the right, while the first column retains the same basic
model of eq. (1) in order to allow comparisons. Model 7 on the local players is presented
twice, since the first model shows the existence of a spatial dependence in the error
terms and a SEM model is required. The last column contains the SEM for the local
players. Coefficients are not standardised and therefore a direct comparison is not
possible. When the first three columns are analysed, most standardized coefficients
appear to be similar between the two groups and the general regression model,
suggesting that the differences with respect to in growth models are not very high.
The Chow test, performed on the different models of Table 5, produced a p-value .95,
and is not conclusive on whether the null hypothesis that the two regressions actually
have the same coefficients cannot be rejected. An implied ambiguity therefore remains.
To overcome this doubt, and to test whether some specific coefficients differ across
types of regions, spatial heterogeneity was measured on each single coefficient,
multiplying each variable by a typology dummy, namely local players, i.e. estimating eq.
(1). This technical procedure made it possible to disentangle the differential effects of
each success factor in local players, and compare them against global and regional
players. Table 6 sets out the results of a regression in which each variable was crossed
with the local players dummy.
The general results (first column in Table 6), valid for global and regional players,
remain very similar to those of Table 3, with lower significance in some regressors,
probably because of the reduced number of degrees of freedom. Inspection of the
differential effects of success factors on local players, presented in the second column of
Table 6, highlights a general result: spatial heterogeneity is present only in the case of
physical density. All other marginal effects are insignificant, and in all but one case they
are highly insignificant.
The main conclusion to be drawn is that for most success factors spatial heterogeneity
does not hold: success factors impact in the same way on regional growth, despite the
degree of openness of regional economies. This finding raises the following question: if
the success factors are by and large the same for global, regional and local players, and if
their impact on growth is the same among these groups of regions, what is it that
explains the higher regional growth that, on average, global regions achieve? This is the
subject of the next section.
Pages: 67
R.Capello, U. Fratesi Growth Patterns in Global Regions: Do Specific Success Factors Make a Difference?
Table 4. Success factors for European regions: the effects of regional types
Model1
stand.
coeff.
Obs
R2
F
Moran's I
Spatial Error
Lagrange multiplier
Robust Lagrange multiplier
Spatial Lag
Lagrange multiplier
Robust Lagrange multiplier
0.141
-0.198
p-value
sig.
0.085 *
0 ***
Model2
stand.
coeff.
0.111
-0.208
0.075
0.134
0.254
0.07 *
0.076
0.139
0.070
0.009 * * *
0.066
-0.097
0.045 * *
-0.090
0.034
p-value
sig.
Model3
stand.
coeff.
0.183
0 ***
0.248
0.059 *
0.01 * *
0.058 *
0.394
0.146
-0.197
p-value
sig.
0.086 *
0 ***
Model4
stand.
coeff.
0.138
-0.201
246
0.7134
59.15
0.578
0.202
0.071 *
0.071
0.009 * * *
0.069
0.009 * * *
0.069
0.009 * * *
-0.097
0.001 * * *
significant
Included
0.128
0.385
1.681
0.114
0.195
0.736
0.133
0.402
1.689
0.132
0.194
0.717
0 ***
0.084
0.134
2.322
0.755
2.26
0.703
-0.201
0.089 *
0.202
0.071 *
0.587
0.563
0 ***
0.138
sig.
0.084
0.134
246
0.7135
58.56
0.597
0.714
46.5
0.543
0.089 *
p-value
0.241
0.075 *
0.046 * *
-0.092
0.058 *
-0.092
0.058 *
0.821
0.028
Included
sig.
0.078
0.133
0.009
0.001 * * *
significant
Included
p-value
Model5
stand.
coeff.
0.004 * * *
significant
Included
0.55
246
0.7141
52.64
0.598
2.24
0.706
0.134
0.401
1.654
0.12
0.198
0.729
0.482
-0.028
0.482
0.002 * * *
0.001 * * *
significant
Included
significant
0.55
246
0.7141
52.64
0.598
0.55
2.236
0.774
0.135
0.379
2.236
0.774
0.135
0.379
1.541
0.079
0.215
0.779
1.541
0.079
0.215
0.779
Pages: 68
Obs
R2
F
Moran's I
Spatial Error
Lagrange multiplier
Robust Lagrange multiplier
Spatial Lag
Lagrange multiplier
Robust Lagrange multiplier
0.141
-0.198
sig.
0.085 *
0 ***
0.154
-0.208
0.059 *
0 ***
0.075
0.134
0.254
0.07 *
0.095
0.091
0.199
0.256
0.070
0.009 * * *
0.065
0.018 * *
-0.097
Included
246
0.7134
59.15
0.578
0.146
0.36
1.600
0.851
-0.367
0.039 * *
-4.929
0.009 * * *
-0.043
0.092
0.713
0.423
0.000
75.729
0.183
0.256
0.258
0.212
0.003
0.161
0.045 * *
-0.127
0.043 * *
0.028
0.734
0.001 * * *
0.001 * * *
0.808
significant
Included
significant
Included
significant
0.563
175
0.7607
82.2
0.488
71
0.7599
0.626
-1.367
2.26
0.703
0.133
0.402
2.398
0.591
0.121
0.442
6.445
4.445
0.011 * *
0.035 * *
1.689
0.132
0.194
0.717
2.088
0.281
0.148
0.596
2.743
0.743
0.098 *
0.389
0.168
0.478
2.644
0.103
Included
significant
71
Squared correlation
Sigma
0.779
0.76
1.828
Pages: 69
stand.
coeff.
Obs
R2
F
Moran's I
Spatial Error
Lagrange multiplier
Robust Lagrange multiplier
Spatial Lag
Lagrange multiplier
Robust Lagrange multiplier
p-value
0.141
-0.198
sig.
0.085 *
0***
0.075
0.134
0.254
0.07 *
0.070
-0.097
Included
246
0.7134
59.15
0.578
0.102
-0.211
0.212
0 ***
0.257
p-value
sig.
0.106
-0.108
0.028 * *
-0.013
0.160
0.863
0.036 * *
0.100
-0.092
0.206
0.19
0.009 * * *
0.065
0.01 * *
0.030
0.53
0.045 * *
0.001 * * *
significant
-0.123
0.058 *
0 ***
0.053
-0.204
0.454
0.269
0.563
246
0.7248
55.74
0.415
0.678
2.26
0.703
0.133
0.402
2.485
1.323
0.115
0.25
1.689
0.132
0.194
0.717
1.163
0.001
0.281
0.982
Pages: 70
Local players only have the highest policy support, whose influence on regional growth is
positive but weaker.
Figure 1. Endowment of success factors, for global, regional and local players
3
Global players
Regional players
Local players
2.5
1.5
0.5
Innovation (Share of
science and technology
employment 2000)
Physical density (Total km Policies (Structural funds Human capital (1999-2001) FDI (number of FDI per
City effect (dummy for
of infrastructure on sqm
per capita 1994-1999)
million people 1999-2001) rural regions with no large
2000)
city)
Table 7 reports the non-standardized coefficients of the estimated eq. (4), which are measures
of the weight of each variable on growth and, for each type of region, the average values of
regressors in each group of regions and the average effect that each success factor generates
in terms of growth, obtained by multiplying the raw coefficient for the average value.
Table 7 shows that the three success factors that on average are more frequently present in
global regions namely innovation, human capital and FDI are also those that weight most
on growth. Thanks to both the high endowment and the high coefficient, these generate a
large part of regional growth in global player regions. The relatively high endowment of
human capital in global regions has a decisive positive effect on growth (0.98 points).
Regional players are similar to global players in their features, but with lower absolute values,
and only have a bit more of policy support.
It is the high number of rural regions (proxy for the lack of agglomeration economies) that
distinguishes local players from the others; their relatively negative important weight on
growth, the lack of agglomeration economies produce -0.22 point increase less in regional
player GDP growth. Local players are only relatively more positively affected by policies (.14)
and less characterised by congestion effects.
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 71
As a summary of and conclusion to the analysis of success factors, it is possible to state that
regional success factors for European regions are consistent with the theory and very similar
between global and regional players and local players, so that only one statistically significant
difference arises. However, the endowment of success factors differs across types of region; in
particular, global players are better endowed with those factors that have a high impact on
growth.
Pages: 72
Global Players
Effect
Growth
0.11
0.76
0.14
0.98
0.10
0.72
0.09
0.65
0.23
-0.36
0.37
-0.58
0.21
-0.34
0.15
-0.23
369'212
0.08
159'566
0.03
297'151
0.06
666'291
0.14
0.00
0.33
0.01
0.44
0.00
0.31
0.00
0.27
192.34
0.06
465.95
0.14
127.74
0.04
99.16
0.03
0.40
-0.14
0.17
-0.06
0.37
-0.13
0.61
-0.22
Constant
1.45
1.45
1.45
1.45
1.45
Effect
Growth
on Average
Value
Effect
Growth
Local Players
Average
Value
80.02
on Average
Value
Regional Players
on Average
Value
Effect on
Growth
2.17
2.41
2.11
2.09
2.43
2.76
2.47
2.09
Pages: 73
7. Conclusions
This paper has conducted an interpretative analysis of the success factors for growth at
regional level, with the aim to highlight whether the assets that guarantee an economic
performance are different for regions that are more open to integration processes than
for regions that are more isolated.
A first result is that more internationally integrated European regions record GDP
performance rates on average higher than those of the other kinds of regions. Their
higher general positive growth rates amid globalization highlight their capacity to turn
threats generated by a global economy into opportunities; their competitive advantages
are strong enough to enable their local economies to compete on a world market.
Moreover, global players lead their national economies, showing consistently positive
endogenous growth rates.
A second important result is that whilst global regions outperform the others on
average, a their trend is heterogeneous, once accounting for structural factors in a
multivariate regression, regional typologies do not explain higher-than-average GDP
growth, ceteris paribus.
The analysis of this paper shows that success factors explaining regional growth
differentials are fundamentally common to global, regional and local players. This shows
that openness to a global economy per se does not give rise to economic growth;
innovation, human capital, policy support and national effects make a difference in
explaining regional growth differentials in global, as well as regional and local, players. If
this is a reasonable result, a counter-intuitive one is that these factors appear to have the
same impacts on growth across space, for example the marginal effect of innovation on
growth does not vary between global, regional and local players.
If this is the case, the reasons why global players grow, on average, more than the other
groups of regions reside in their greater endowment of the success factors that play an
important role in growth, as shown in the last part of the paper.
This result is important for the development of ad-hoc intervention policies, that should
be devoted to the reinforcement of those regional success factors able to increase intersectoral productivity, namely through innovation where innovation is not merely
intended as the degree of R&D produced by a region, but in a more general sense as all
efforts devoted to increase knowledge, to foster industrial transformation, to develo p
local capabilities in order to cooperate synergically with other regions, and to invent
new organizational solutions at both the firm and public governance levels.
All this requires a change in policy style; integrated, inter-industry, pervasive policies
have to be devoted to prepare territories for innovation and global competition,
enhancing their adaptability to a changing external context, promoting their openness
and receptivity to new business ideas and organisational styles, rather than forcing the
Page: 74
locational decisions of single firms, and to negotiate the terms for fruitful cooperation
between territories and firms, rather than merely supplying favourable location factors.
References
Anselin, L., 2010. Thirty Years of Spatial Econometrics, Papers in Regional Science, vol.
89, n. 1, pp. 1-25
Audrestch, D., Aldridge, T., 2009. Knowledge Spillovers, Entrepreneurship and Regional
Development, in Capello R. and Nijkamp P. (eds.), Handbook of Regional
Dynamics and Growth: Advances in Regional Economics, Edward Elgar, pp. 201210
Baldwin, R., 2006. Globalisation: the Great Unboundling, Prime Ministers Office,
Economic Council of Finland
Baldwin, R., Martin, P., 1999. "Two waves of globalisation: Superficial similarities and
fundamental differences in Siebert H. (ed.), Globalisation and Labour, Chapter 1,
pp 3-59, J.C.B. Mohr for Kiel Institute of World Economics, Tubingen.
Barca, F., 2009. An Agenda for a Reformed Cohesion Policy: A Place-Based Approach to
Meeting European Union Challenges and Expectations, Independent Report
Prepared at the Request of Danuta Hbner, Commissioner for Regional Policy,
DGRegio, available on-line
Becattini, G., 1990. The Marshallian Industrial District as a Socio-Economic Notion in
Pyke F., Becattini G. and Sengenberger W. (eds.), Industrial Districts and Interfirm Cooperation in Italy, ILO, Geneva
Beguin, H., 1988. La Rgion et les Lieux Centraux, in Ponsard C. (eds.), Analyse
Economique Spatiale, Presse Universitaire de France, Paris, pp. 231-275
Bergoeing, R., Kehoe, T., Strauss-Kahn, V., Yi, K., 2004. Why Is Manufacturing Trade
Rising Even as Manufacturing Output Is Falling?, American Economic Review, vol.
94, n. 2, pp. 134-138
Beugelsdijk, S., Smeets, R., Zwinkels, R., 2008. The impact of horizontal and vertical FDI
on hosts country economic growth, International Business Review, vol. 17, pp.
452472
Boldrin, M., Canova, F., 2001. Inequality and Convergence in Europes Regions:
Reconsidering European Regional Policies, Economic Policy, 16, 32, 205-253.
Borensztein, E., De Gregorio, J., Lee, J-W., 1998. How does foreign direct investment
affect economic growth?, Journal of International Economics, vol. 45, pp. 115135
Boschma, R., 2005. Proximity and Innovation: a Critical Assessment, Regional Studies,
vol. 39, n. 1, pp. 61-74
Boschma, R., Lambooy, J., 1999. Evolutionary Economics and Economic Geography, vol.
9, n. 4, pp. 411-429
Boschma, R., Martin, R., 2010. Handbook of Evolutionary Economic Geography, Edward
Elgar
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 75
Brcker, J., Rietveld, P., 2009. Infrastructure and Regional Development, in CAPELLO R.
and NIJKAMP P. (eds.), Handbook of Regional Dynamics and Growth: Advances in
Regional Economics, Edward Elgar, pp. 152-181
Camagni, R., 2002. On the Concept of Territorial Competitiveness: Sound or
Misleading?, Urban Studies, vol. 39, n. 13, 2395-2412
Camagni, R., 2009. Territorial Capital and Regional Competitiveness: Theory and
Evidence, in Capello R. and Nijkamp P. (eds.), Handbook of Regional Dynamics
and Growth: Advances in Regional Economics, Edward Elgar, pp. 118-132
Camagni, R., (ed.) 1991a. Innovation Networks: Spatial Perspectives, Belhaven-Pinter,
London
Capello, R., Fratesi, U., Resmini, L., 2011. Globalization and Regional growth. Past Trends
and Future Scenarios, Springer-Verlag.
Cappellin, R., Nijkamp, P., (ed.) 1986. Technological Change, Employment and Spatial
Dynamics, Berlin, Springer-Verlag
Carkovic, M., Levine, R., 2005. Does foreign direct investment accelerate economic
growth?, in T. H. Moran, E. Graham, And M. Blomstrom (eds.), Does foreign direct
investment promote development? New methods, outcomes and policy
approaches, Peterson Institute, pp. 195-220
Casi, L., Resmini, L., 2011. Globalization, FDI and growth in European regions: an
empirical assessment, in Capello R. and Dentihno T. (eds.), Globalization Trends
and Regional Development: New Directions in Foreigner Direct Investments and
Human Capital Flows, Edward Elgar, forthcoming
Clark, C., Wilson, F., Bradley, J., 1969. Industrial Location and Economic Potential in
Western Europe, Regional Studies, vol. 3, n. 2, pp. 197-212
Cooper, R., Donaghy, K., Hewings, G., (eds.) 2007. Globalization and Regional Economic
Modeling, Springer Verlag
Crescenzi, A. Rodriguez-Pose, A., 2011. Innovation and Regional Growth in the European
Union, Springer-Verlag, Berlin.
Dall Erba, Legallo, J., 2008. Regional convergence and the impact of European structural
funds over 19891999: A spatial econometric analysis, Papers in Regional Science,
87-2, 219-244
De Mello, L. r., 1999. Foreign Direct Investment-Led Growth: Evidence from Time Series
and Panel Data. Oxford Economic Papers, 51(1), 133-151.
De Groot, H., Poot, J,. Smit, M., 2009. Agglomeration Externalities, Innovation and
Regional Growth: Theoretical Perspectives and Meta-Analysis, in Capello R and
Nijkamp P. (eds.), Handbook of Regional Dynamics and Growth: Advances in
Regional Economics, Edward Elgar, pp. 256-281
Fratesi, U., 2011. A globalization-based taxonomy of European regions. Regional Science
Policy and Practice. doi:10.1111/j.1757-7802.2011.01052.x
Friedmann, J., 1986. The World City Hypothesis, Development and Change, vol. 17, n. 1,
pp. 69-83
Page: 76
Gorodnichenko, Y., Svejnar, J., Terrell, K., 2008. Globalization and Innovation in
Emerging Markets, NBER Working Paper Series, wp n. 14481
Hansen, H., Rand, J., 2006. On the Causal Links Between FDI and Growth in Developing
Countries, The World Economy, vol. 29, n. 1, pp. 21-41
Hanson, G., Mataloni, R., Slaughter, M., 2005. Vertical Production Network in
Multinational Firms, The Review of Economics and Statistics, vol. 87, n. 4, pp.
664678
Howells, J., 2005. Innovation and Regional Economic Development: A Matter of
Perspective?, Research Policy, vol. 34, n. 8, pp. 1220-1234
Hummels, D., Ishii, J., Yi, K., 2001. The Nature and Growth of Vertical Specialization in
World Trade, Journal of International Economics, vol. 54, n. 1, pp. 75-96
Hummels, D., Rapoport, D., Yi, K., 1998. Vertical Specialization and the Changing Nature
of World Trade, Federal Reserve Bank of New York Economic Policy Review, vol. 4,
n. 2, pp. 79-99KENNEY M. and FLORIDA R. (2004), Locating Global Advantage:
Industry Dynamics in the International Economy, Stanford University Press,
Stanford Ca.
Jacobs, J., 1961. The Death and Life of Great American Cities, Random, New York
Johansson, B., Karlsson, C., 2009. Knowledge and Regional Development, in Capello R.
and Nijkamp P. (eds.), Handbook of Regional Dynamics and Growth: Advances in
Regional Economics, Edward Elgar, pp. 239-255
Karlsson, C, Johannson, B., Stough, R., 2009. Human Capital, Talent and Regional
Growth paper n. 191, CESIS Electronic Working Paper Series
Kucera, D., Milberg, W., 2003. Deindustrialization and Changes in Manufacturing Trade:
Factor Content Calculations for 1978-1995, Review of World Economics, vol. 139,
n. 4, pp. 601-624
Lall, S., Narula, R., 2004. Foreign Direct Investment and its Role in Economic
Development: Do We Need a New Agenda?, The European Journal of Development
Research, vol. 16, n. 3, pp. 447 464
Lesage, J., Pace, R., 2010, Introduction to Spatial Econometrics, Springer Verlag
Lesage, J. P., 1998. Spatial Econometrics, The Web Book of Regional Science, Regional
Research Institute, West Virginia University, Morgantown, WV
Lucas, R.E., 1988. On the Mechanics of Economic Development, Journal of Monetary
Economics 22, 3-42
Lucas, R., 2008. International Labour Migration in a Globalizing Economy, Carnegie
paper n. 92
Mohl, P, Hagen, T, 2010. Do EU structural funds promote regional growth? New
evidence from various panel data approaches, Regional Science and Urban
Economics, 40, 353-365.
Moran, T., Graham, E., Blomstrm, M., (eds.) 2005. Does Foreign Direct Investment
Promote Development?, Institute for International Economics, Washington D.C.
Mullen, J. K., Williams, M., 2005. Foreign Direct Investment and Regional Economic
Performance. Kyklos, 58(2), 265-282.
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 77
Page: 78
Keywords:
JEL Classification:
B51; E12.
Mumbai,
Vidyanagari,
Mumbai
400
098,
INDIA,
Business Environment Group, Indian Institute of Management Lucknow, Lucknow 226, 013- Uttar Pradesh, INDIA,
dtripatirao@yahoo.com
32
Pages: 79
1. Introduction
We critique received wisdom about financial reform. In developing countries, interest
rates were kept low in order to induce credit into sectors that were regarded as pivotal
from a social and long-term perspective and that would otherwise be deprived. Financial
liberalization has meant lifting these caps so that people will be tempted to save more
and resources for productive investment would thereby be released. The evidence does
not support the neoclassical justification. Empirically, the interest elasticity of savings
has always been in question anywhere and the analytical basis for the route from high
savings to high investment is far from impeccable. In the case of poor countries, the
general level of incomes is so low that sheer survival takes precedence over higherorder motives (Rosenzweig, 2001). Besides, there are strong and volatile inter-annual
fluctuations in income because rainfall is unpredictable. Maintaining consumption in the
face of low and variable income is the sole occupation of the people in general. With
agricultural transformations incomplete, markets guide finance away from agriculture
to industry and services. There is an urgent need to open the agrarian question, closed,
in particular, after structural adjustment programmes (SAPs) in poor economies.
Following Henry Bernstein, the peasantry as a category is problematic and attention
must be paid to the social relations of labor (Ellis, 2006). SAPs were introduced into subSaharan Africa in the seventies to rid agriculture there of exploitative parastatals.
Accordingly, governments dismantled upper and lower bounds on prices and pushed the
economies toward market exchange in order to induce the evolution of competitive
private trade in rural areas. The outcome, overall, has been disastrous. Private traders
have not rushed into spaces created by fleeing parastatals. The characteristics that
remain include a food security crisis which the upheavals of liberalization only
worsened. In other words, since the vast majority cannot access markets to purchase
food at affordable prices during the lean season, it makes sense to store food. The
outcome is limited exchange of food and cash. Actually, the dialectic of State intervention
and withdrawal is more subtle (Das, 2007). The State has stepped into agribusiness and
supported the production of luxury farm products like flowers and shrimps. At the same
time, it cannot ignore smallholder interests. Peasants are part of the commodityproducing process providing capital a market for manufactures. They continue to be an
important part of the reserve army providing industrialists a huge workforce.
Secondly, the technological revolution that has swept the financial services industry has
led to the conviction that it is unhelpful to think in terms of different financial entities.
Banks, for instance, are not regarded as possessing a special status anymore. All
financial entities are Universal Banks operating both the in the commodities market as
well as in markets for exotic financial products. A pivotal moment in contemporary
history, in this regard, is the repeal of the Glass-Steagall Act in 1999 in the USA. The Act
of 1933 distinguished between commercial banks and investment banks on the ground
Year 2013, Volume 1, Issue 1
www.jheec.com
Page: 80
that their domains of operation should not intersect in the interest of financial stability.
In the case of developing countries it is well established that the proximity of banks
increases financial savings and crowds out the merchant-money lender route to the
provision of insurance. The process of financial intermediation is facilitated. The
absence of well-functioning credit markets is reflected in inefficient asset stocks and
compositions. The mechanism of financial intermediation is critical in extending credit
to informationally-opaque small borrowers (Berger et. al., 2001). The current wave of
mergers and acquisitions is likely to create large banks that are adjuncts to capital
markets and that are oriented towards large corporate clients. Pure transactions lending
describes the relationship between lender and clients where due diligence and contract
terms are based on information that is available at the time of signing of the contract.
The information set might contain no more than the data that is available in balance
sheets. Their remoteness from the particularities of rural markets implies that the
abilities of large banks to process local-level information is limited. In the absence of
repeated interactions over time, lenders resort to class-based assessment of
probabilities rather than case-by-case assessment procedures (Runde, 2002). In the case
of small bank lending, on the other hand, information is of a qualitative kind and is
garnered from the borrowers suppliers and customers and also from the borrowers
interactions with the local community. Relationship banking entails a unique, one-toone, face-to-face interaction between borrower and lender on the basis of which
decisions concerning additional funding and monitoring strategies are arrived at. The
generation of information is costly and the costs are likely to be passed on to the
borrower. The conjecture is that borrowers without a credit history would be willing to
bear these costs in order to renegotiate contracts on favorable terms. More to the point
in the aftermath of the last financial crisis in the USA, an important element in the
frequency of crises is the input-output structure of the banking system, whether
extensive branching exists, whether the system consists of a few large banks or many
small banks (Gorton, 2012).
Our task is to distinguish the intervention of banks and nonbank financial
intermediaries in a classical-Marxian framework. We provide a non neoclassical
formalization of these intuitions below.
2. A commodities circuit
A common starting point of any structuralist account of a developing economy would be
the demarcation of the economy into an industrial and an agricultural sector. For the
purposes of the present study, the division cuts across the division of the economy into
Departments following Marxs schemes of reproduction (Kaleki, 1976). Our Sector I
includes Marxs Department II producing luxury goods for capitalists, Sector II is Marxs
Department III producing basics for workers and Department I producing investment
goods is divided into two parts, investment goods produced in Sector I and investment
Pages: 81
goods produced in Sector II. This classification is resorted to in order to make non basics
and basics output correspond with the output of Sector I and Sector II respectively.
The pricing rule for the industrial sector takes the form:
(1)
Pn (t ) (1 )wn (t )bn (t )
where Pn is Sector Is producer price; is the markup rate; wn is the nominal wage rate
and b is the inverse of average labor productivity. Output Xn is divided in the familiar
way between consumption Cn and investment In. Consistent with the division into
Departments above is the assumption that the agriculture sector is a food sector
providing basics to workers in both sectors. Similarly, the industrial sector output is
consumed by capitalists from both sectors. While food is the quintessential basic, it is a
catchall for the consumption by workers as might haute cuisine be a metaphor for
luxury consumption. Jack Goody (2006) has extended and modified the work of the
Marxist historian Gordon Childe in proposing this thesis. He makes the case that the
starting point of historical research into either European or Asiatic exceptionalism, as
Marx put it, must lie in the convergences of urban civilizations of the Bronze Age rather
than the divergences of the nineteenth century. Both the town and the country were
stratified by class based on economic differentiation in the Bronze Age. The continuous
series of evidence on towns in Asia indicate that they did not vanish to reappear later as
engines of early capitalist enterprise but were vibrant centers of manufacture and
exchange. Mass production and an increasing role for finance, as exchange intensified,
were not confined to Europe. Under feudalism, the history of urbanization is different
outside Europe. The urban economy with its distinction between haute cuisine and basic
consumption continuously evolved. Clearly, a class analysis is indispensable.
Modern non neoclassical macro models add a Keynes-Wicksell turn in the assumption
that firms make their investment plans on the basis of the divergence between the rate
of profit in industry, rn, and the real rate of return on bonds, i , with bonds and
equities being regarded as perfect substitutes. Distinguishing the agriculture sector by
the subscript a, excess demand in the non-food sector in nominal terms is given by
(2)
[ Pn (t )Cn (t ) Pa (t )Ca (t ) Pn (t ) I n (rn (i ))] Pn (t ) X n (t )
where, in an abuse of notation, Cn is luxury consumption by urban capitalists and Ca is
luxury consumption by their rural landlord-capitalist-merchant counterparts. In words,
the term on the right-hand side outside the brackets is the total output, in value terms, of
the output of Sector I. The demands of it are depicted in the square bracket and consist
of two parts; one part, subscripted by n, is the sum of consumption and investment
demands by the sector (the time argument in the case of the investment function is
implicit), the other, subscripted by a is the demand for luxury commodities coming from
Sector II.
The agricultural sector is assumed to be flexprice and we assume, without loss of
generality, that financial portfolio choices are unavailable to the kulaks. Investment in
the sector depends only on the rate of profit there. Excess demand in Sector II is
(3)
[wn (t ) Ln (t ) wa (t ) La (t ) Pa (t ) I a (r a )] Pa (t ) X a (t )
Year 2013, Volume 1, Issue 1
www.jheec.com
Page: 82
where wi Li is the wage bill in sector i. Once more, the term on the right-hand side outside
the brackets is the total output, in price terms, of the output of Sector II. The demands of
it are depicted in the square bracket and consist of two parts; one part, subscripted by a,
is the sum of consumption and investment demands by the sector (the time argument in
the case of the investment function, again, is implicit), the other, subscripted by n is the
demand for luxury commodities coming from Sector II.
The traditional Keynesian adjustment process posits an inversion of the Walrasian
ttonnement adjustment process. Quantities are assumed to adjust to quantity
discrepancies on the assumption of chronic excess capacity. We have
dX n
Pn (t )C n (t ) Pa (t )C a (t ) Pn (t )( I n (rn (i )) Pn (t ) X n (t )
(4)
dt
The agricultural sector, on the other hand, is resource-constrained. An increase in the
supply of food can only come from investment activities like irrigation works and so on.
Prices take on the brunt of adjustment for the agricultural sector. In the situation of
excess demand described above,
dPa
wn (t ) Ln (t ) wa (t ) La (t ) Pa (t ) I a (ra ) Pa (t ) X a (t )
(5)
dt
(6)
I a X a
0
Call the matrix A and determinant and trace are denoted by the usual det and tr
respectively. Now, since, by definition, Ia Xa < 0, detA > 0, and the other standard
condition for stability, trA < 0, is naturally met. The more stringent requirements to
prevent bifurcations are the following (Medio & Lines, 2001).
1+ trA + detA > 0
(7)
1 trA + detA > 0
(8)
1 detA > 0
(9)
The inequalities, in sum, neatly encapsulate a tension in the debate on the optimal route
to capitalism (Byres, 1996). A high value of the single north-west element is sufficient
for stability and validates the Preobrazhensky thesis underscoring the necessity of the
terms of trade to move against agriculture in order to facilitate accumulation outside
agriculture. The function of agriculture is not only to generate a real surplus but also a
financial surplus. The surplus of interest is the marketed surplus which represents a
command over real resources which can be transferred from agriculture. In addition,
small farmers are net buyers of food. Kaleki assumed that there would be no
inflationary price increases of necessities, particularly of staple foods. Any policy stance
of that kind militates against any sense of social justice (Kaleki, 1976, p.18). It is
necessary, then, to expand agricultural output in the short run (Xa in the south-east
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 83
corner of the matrix). The higher supply of food will enable the system to feed those who
transfer to non-agricultural output. A substantial increase in agricultural output is
feasible in a short period of time without heavy investment by the introduction of smallscale irrigation projects, double-cropping and the like. Any energies along these lines are
hampered by class divisions in the countryside. The mass of small peasants is bonded to
merchants and moneylenders. Farms operate under a system of disguised tenancy
without security of tenure. Government polices must be targeted at breaking these
institutional shackles. Service cooperatives could be set up for the purpose of credit
disbursement and sale of produce. It is worth noting, for completeness, that addressing
effective demand failure through an increase in the consumption of luxury commodities
by the landed aristocracy, (the north-east corner of the box), is irrelevant to the stability
of the model. Indeed, it is the task of Marxian political economy to critique growth
models which are oriented towards the consumption of non-basics (Perraton, 2007).
The task is to articulate conditions under which real wages can grow consistent with the
growth of productivity. The market-led model of land reform has been unsuccessful in
pushing a productivity-led redistribution of wealth in contrast to the State-led model
(Borras & McKinley, 2006). There can be no overall growth without accelerated rural
development and the latter cannot result without serious land reform. The efficiency of
the post-reform agrarian system in the State-led model which includes Japan, The
Republics of Korea and Taiwan, Bolivia, Chile, Cuba, Mexico, was obtained by massive
complementary public investments, credit and technical assistance.
At the same time, a critique of the Prussian road or what Lenin called capitalism from
above is implied, according to which landlords are the motor to capitalist
transformation. One regressive vertex of that path has been the continuous
impoverishment of the peasantry resulting in a significantly shrunken home market. The
impact on capitalist industry, as well, cannot be salutary. Department II branches (in the
traditional sense) would be constricted by the narrow home market. Lenin was
concerned with the sluggish mechanization of agriculture. The consequence was a
constricted market, once again, for the output of capitalist manufacturing industry
especially the products of Department I branches (in the traditional sense) like chemical
fertilisers, farm implements and machinery.
Lenin made the case for the American path or capitalism from below. The dynamic here
is from a differentiating peasantry with vigorous class-for-itself action pursued by rich
peasants and capitalist farmers. As the differentiation proceeds, the capital-labour
relation evolves. The transition is clearly plausible in a milieu where the landlord class is
weak but is not ruled out in a regime where the class is strong, provided certain
conditions are in place. The American path was progressive in two ways (Byres, 1996).
The forces of production were developing in the countryside. Modern inputs were being
applied and mechanization was proceeding apace. Semi-feudal relations of production
were not fetters. Secondly, the path entailed a massive growth of the home market.
Department I industries supplying agricultural inputs were faced with an expanding
home market. In addition, the rise in the standard of living of the peasantry provided an
Year 2013, Volume 1, Issue 1
www.jheec.com
Page: 84
0
di C n I n X n
An increase in the interest rate raises the price of non basics. In standard one-sector
models, Pn would be the general price level. In like manner, what is the impact on output
of a flexible interest rate?
dI
n
dX n
(11)
di 0
di
Pn
An increase in the interest rate reduces the output of non basics. Once again, in a onesector model Xn would be output. In familiar terms, the outcome is stagflation (Weller,
2001). The consequences for agriculture are no less deleterious. It is unnecessary to
posit a regime of excess demand to conclude that the impact of rising interest rates is a
rise in the price of agricultural output.
dI
n
dPa
(12)
di 0
di
Ca
A final consequence to consider is the impact of the interest rate on the markup. The
result is once again familiar. The degree of monopoly increases. A regime of excess
demand benefits the capitalist class.
dI
n
d
di
0
(13)
di dPn
(C n I n X n )
di
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 85
The signs of the derivatives are a corollary of the empirical observation that financial
liberalization in developing countries has had an adverse impact especially on farmers
and workers, including during so-called financial booms (Ghosh, 2005). The proposition
is not revolutionary. From the highest positions of establishment orthodoxy comes the
theorem, derived from exacting statistical testing, that inflation, inequality, and class
conflict are correlated (Crowe, 2006). The inflation tax is regressive not the least
because workers and capitalists have differing access to inflation-proof assets.
Developing countries are more unequal and therefore experience higher inflation. The
class nature of the political system and inequality interact to generate inflation.
Our outcomes are perfectly consistent with the unity of Keynesian, structuralist, and
Marxist models of growth and distribution (Foley & Taylor, 2006). The common core
resides in the study of the distribution of national income between workers and
capitalists; the eschewing of model closures that imply full employment of the labor
force; differential modeling of the consumption and savings decisions of workers and
capitalists; the adoption of an investment demand function independent of savings
decisions; and a separate treatment of the firm as an agent independent of its owner
households. With Marx, as with Kaleki later, less-than-full utilization of capacity is the
norm in advanced capitalism. Variations in investment will be met with variations in
capacity utilization with no effect on distribution. Income shares, on the other hand, are
determined by the principle of markup pricing on unit labor costs in noncompetitive
commodity markets. The markup, in its turn, is determined by the degree of monopoly
in the goods market and by the class struggle in the labor market.
The task ahead is further enlightened the interplay between the real and financial
components of the capitalist system. We suggest below that the circuit approach to
monetary macroeconomics, active in France and Italy, is a natural complement to the
real circuit enunciated here. Comprehensive treatments of the former will be found in
Cencini (2005), of the latter in Graziani (2003).
Page: 86
money commodity. In that case, Marxs theory of money is freed of theoretical metallism.
Rather, money is the socially acceptable representation of the universal equivalent
which has to be underpinned by social institutions. The modern credit system, which is
founded on promises to pay, conforms nicely. The practice of circuit theory is to reason
in terms of moments. Scrupulous accounting principles connect the moments. To
positive items must correspond negative items of equal magnitude. For Marx as well, the
price of labor power is identically equal to the variable capital that has been advanced to
purchase it. Recent internal critiques of Marxs circulation schemes are founded on the
discipline of accounting (Cartelier, 2007). The notion is that the form of circulation
corresponding to a commodity division of labor is M-C-M (with M = M). The difference
between simple commodity production and a capitalist economy is founded on a wage
relationship in the latter with a derived labor-money-commodity (L-M-C) circulation.
The argument hinges on Marxs emphasis on both the independence of activities and the
fact that each producer acts on her own account. The latter implies that each individual
is in command of her own process of production, purchasing necessary inputs and the
means of reproduction. The former means that commodity production is for sale. A
consequence is that producers can only be part of a commodity division of labor if they
own the means of payment permitting them to produce autonomously of the decisions
of others. This is M-C of M-C-M. Producers decide anarchically to spend a certain fraction
of money. These expenditures might be displayed along the rows of a payments matrix,
each producer associated with a row. But the social accounting matrix imposes an
interdependence among the producers. They must validate their activity by selling their
output. They cannot earn, from the market, less than they have spent. The constraint is
C-M of M-C-M. One producers receipts are another producers expenditures. Receipts
and expenditures are simultaneously determined. The circuit closes when the initial
disbursement of M by banks is returned to them by producers.
When individuals who constitute the commodity division of activities succeed in making
those who are excluded because they do not possess the means of payment join them,
the former become entrepreneurs, the latter workers. Workers spend a proportion of
their incomes. While they are free to decide the fraction of income saved, they are not in
command of their resources. Thus, L-M-C is a derivative of the circulation of
commodities. Workers have nothing to sell but their labor power because they are
neither private nor independent. They acquire commodities but these commodities are
simply use values. The consumption bundle of workers does not enter the technique of
production. Businessmen consider money wages as a cost not as an expenditure outlet.
On the other hand, in the case of simple commodity production, in equilibrium the
consumption of the producer is a socially necessary as other material inputs. It is only in
capitalist society with M-C-M buttressed by L-M-C, that net value may appear.
In a monetary production economy, the production and circulation of commodities
between the Departments is mediated by banks. The role of banks is unique in that they,
and only they, can lend out claims on their own debt (Bossone, 2001). Only banks
possess the ability to add to the existing stock of money by lending promises to pay.
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 87
Thereby the total credit in the economy can exceed what is possible if credit has to be
fully recovered. Banks create money and generate purchasing power in anticipation of
the production of commodities. In Capital Vol.III, as well, Marx does not assume that the
credit supply of commercial banks is limited by private savings but shows that
commercial banks can create credit without limits which circulates as credit money.
According to the circuit approach applied to our framework, the sequence consists of
three points: 1. the creation, 2. transfer and 3. destruction of income. An illustration on
the lines of Bossone in terms of the model sketched in the previous section is provided
in an Appendix. The italicized items are the balancing counterparts, in the balance sheet
sense, of identical primary entries whose numbers depict the three moments defined
above. At the start of the circuit banks negotiate with capitalists (industrial and
agricultural) the conditions for working capital loans, waLa and wnLn, respectively. The
banks credit the deposit accounts of the capitalists with the loan amounts, denoted by
the italicized entries, loan claims and loan deposits, respectively, in both Departments.
The firms produce, using the loans to pay wages to workers (industrial and agricultural).
Entries numbered 2 in the flow of funds distinguish this moment. Deposits are
transferred from the bank accounts of the capitalists to the accounts of the workers. The
latter is wages in the balance sheet of workers matched by the italicized wage payments
in the balance sheets of the former. At the second point of the sequence, workers spend
their incomes on basics. Capitalists spend their incomes on non basics. Sales across
classes and commodities, depicted by 3 and 3 are identically equal to purchases 3. The
circuit ends when capitalists use the proceeds from their sales to discharge their debt to
the banks. The money that was created is destroyed. Banks do not create value. The
freshly issued money assumes value only in the process of production. All money
transfers and payments for commodities and labor services take place through deposit
transfers across bank accounts. No cash circulates. At circuit end capitalists must secure
enough money to pay off their initial debt. No transfer of real resources is entailed from
them to the banking system. In case of the example worked out in the Appendix, adding
up over the two sides of the balance sheets of capitalists, their financial obligations to
the banks equals their deposits with the bank, both being equal to waLa + wnLn. Money
disappears in equilibrium as a shadow of the goods and services it symbolizes (Shubik,
2006).
The credit system introduced above works on a hierarchy of promises to pay with
stronger social validation and liquidity beginning with the base of bank-borrower
relationships. Graziani closely works out the details of a transaction between two parties
being effected by means of a promise to pay of a third party and so on. The pyramid
peaks with fiat money because only Central Banks can guarantee final payment
(Thornton, 2008). Other scholars, as well, work out the close connection between th e
bank issue of money to finance the production and circulation of commodities and
sovereign control of the mint (Goodhart, 2004; Wray, 2006). Before the evolution of
Central Banks, countries operated within low-level equilibrium traps. Production and
exchange was rudimentary and fragmented, mediated by local moneys. It is only when
Year 2013, Volume 1, Issue 1
www.jheec.com
Page: 88
the unifying force exercised by an instrument that bore the imprimatur of the State
evolved, that the capitalist mode of production could grow in leaps and bounds. Fiat
money rounded up private monies, destroyed the antimonies of time and space, and
enabled a manifold increase in the quantity and variety of goods and services.
Commercial banks, again, are not intermediaries between borrowers and lenders but
rather permit creditors and debtors to clear accounts with a third party. The market is
not a place where commodities are exchanged but is a clearing house for debts and
credits. Banks deliver this service. The Central Bank, after all, is a member of the club of
banks. Historically, some banks were accorded this special status because their notes
assumed a special quality. The agents who control the money and credit structures can
influence the constraints that are either slack or have positive shadow prices (Shubik,
2006). The decision to change the intensity of the constraints lie with the borrowers and
lenders and their assessments of risk. The collapse of the link between Central and
commercial banks, according to the Chartalists, would lead to the degrading of the mint
and the collapse into barter. The defining character, according to them, is the power of
the State to levy taxes. In the first place, absent money, taxes could only be imposed on
commodities, since only they can be delivered. If taxes were paid in goods or labor, the
balance of goods and services obtained would not be the quantum required for public
sector expenditure. Money, thus, reduces the transaction costs of governments. For our
purposes, it is worth noting that the ability to impose taxes payable only in money has
been used on numerous occasions in colonial history to coerce taxpayers out of a nonmonetary subsistence economy into a cash economy. The receipt of revenue was often a
subsidiary motive. In other words, public and private finance are organically related
(Merton & Bodie, 2007).
Our categories, so far, only include consumer goods markets and firms producing
consumer goods. In an extended Marxist circuit of capital, the capitalist class subdivides
into money capitalists and functioning capitalists. The expansion of the credit system
requires the establishment of interest-bearing capital with the interest rate as a claim on
a portion of the surplus value produced by productive labor. In what will be called the
second moment of circuit theory, capitalist production process requires money
advances. Each quantum of such money assumes the role of interest-bearing capital
which can be sold for interest. Functioning capitalists are willing to borrow from
money capitalists and pay interest because money has the potential to generate money
profits. In order for contracts to be signed, the Keynes-Wicksell criterion must be met.
The rate of profit has to exceed the rate of interest. Thus, the interest rate in Marxs
schema is a monetary category determined by the relative powers of industrial and
money capital. Reverting to circuit theory, households, typically, will not invest their
entire incomes on consumption but will avail of opportunities provided by financial
markets. At least we can assume that the propensity to save by the capitalist class
exceeds that of the working class. The savings are availed of by investing enterprises
which use these resources to purchase the output of capital goods-producing firms. In
identical fashion, the conclusion of this parallel stage 2 ends with the sales of their
Year 2013, Volume 1, Issue 1
www.jheec.com
Pages: 89
output by firms in capital goods markets. For the French school following Bernard
Schmitt and Alvaro Cencini, accounting purity must be maintained and the activity of
banks in the first moment must not be confused with the activity of nonbank financial
intermediaries (NBFIs) in the second moment. The recording of production, sales and
profits must be kept separate.
The problem, eloquently articulated by Keynes, is that fundamental uncertainty
underlies the decision to make long-term investments. Risk-management techniques he
would describe as no more than pretty, polite techniques to deal with the future. We
simply do not know. The temptation presses on private savers and investors to devise
contracts, beginning with first-order correspondences with the underlying real assets
but then moving to complex, contingent, claims where the connection with the
underlying assets is lost. The best of intellectual resources are at work today in writing
arcane models for packaging and repackaging assets far removed from the informational
idiosyncrasies of particular financial entities. Contemporary financialization attenuates
the circuit of the flow of capital goods and services by the focus on short-term gains at
the expense of what might be termed the long-term social surplus (Blackburn, 2006).
Balanced multi-Departmental growth requires infrastructure and investment in an
educated working class, a strategy that does not pay off for decades. The roller coaster
stock market ride of the nineteen eighties and nineties lacked credibility and foresight,
the hallmark of the plan of Keynes and other architects of the postwar boom. Developing
countries have not been exempt. In a panel data study of countries including Argentina,
Mexico and Turkey, financial liberalization is shown to channel real savings into
financial churning rather than long-term capital investments (Femir, 2007). Thus,
financialization is intimately linked with deindustrialization. Financial innovations carry
the further embellishment that, unlike developments in manufacturing and agriculture,
they enrich a negligible portion of the population and cannot form part of a broad-based
plan to increase demand. The vaunted superior risk-management techniques referred to
have been of virtually no consequence for production. Financial profits, instead, have
been ploughed back to increase the turnover of non basics. In Marxist terms, the current
phase of the division of labor is accompanied by the crisis of the law o f value-labor and
the return of mercantile and financial mechanisms of accumulation that is reminiscent of
the formal subsumption of labor under capitalism (Vercellone, 2008). In an economy
based on the production of knowledge by means of knowledge, the directly necessary
labour time for production is reduced and the monetary value of production falls
sharply. As a result, in order to keep the level of exchange value in place in order to
guarantee profits, capital is led to devise mechanisms of revenue extraction based on the
rarefaction of supply. Marxist scholars summarize the potential for crisis in the M-C-M
(Demir, 2007) or M-C-M' (Vercellone, 2007) general formula. In both cases, M or M'
denotes flexibility and freedom of choice. C is an interruption denoting materialization
and rigidity. Capital would rather be in liquid form.
Page: 90
Pages: 91
with the defining character of banks which, for decades, funded their assets through
retail deposits. Banks in the sixties and seventies were concerned with the quality of
credit assessment and monitoring of the borrowers actions during the tenure of loans.
In recent times, there has been a greater reliance on wholesale funding and the short
end of the credit market. Earlier, while in principle deposits were withdrawable on
demand, assets were relatively stable even when the market reputation of the bank in
which they were placed was in question. Now, the liabilities are of a maturity of one to
six months. In sum, liquidity has declined dramatically calling for frequent Central Bank
bailouts.
5. Class-based policy
Keynes, in the Treasury memoranda of 1943-1944, proposed the socialization of
investment as a strategy to address the problem of unknowledge that plagues the
inducement to investment (Seccareccia, 1994; Smithin, 1989). Through the thirties and
the forties, Keynes supported the establishment of a National Investment Board. The
mandate of the board was to move the economy towards full employment by regulating
the aggregate flow of investment expenditures through the control of long-term
financing. He did not advocate fiscal fine-tuning or deficit financing. Keynes made a
distinction between the ordinary or current budget and the capital budget with the
former balanced or in surplus in the short run. The distinction between investment and
capital spending and consumption or current spending has to be scrupulously
maintained. A policy of direct public investment, if accounted for by the capital budget
would, other things being equal, generate persistent surpluses in the ordinary budget.
Thereby deadweight debt would be gradually replaced by productive or semiproductive debt. The State was to use its surplus not to extinguish its debts but to
expand capital expenditures further. Deficit financing of capital spending is politically
more defensible than deficit financing of undifferentiated State expenditure. Productive
State investment would be immune to the charge of crowding out which might be made
of deficit-financed current spending. Specifically, the expectation was that two-thirds to
a third of total investment would be directly influenced by semi-public bodies whose
activities would not be antithetical to the traditional motive of private exchange but
would also include technical social motives that would justify investment in social
infrastructure. The socialization of investment would reduce interest rates through
capital saturation. It is clear that, nomenclature notwithstanding, this principle of capital
budgeting has little to do with the State ownership of the means of production. The
proposal has more to do with the composition of government expenditure than the
prognosis that a growing proportion of total investment expenditures in the economy
would become socialized.
Secondly, Bossone and Sarr (2005) have imaginatively extended circuit theory to the
plight of desperately poor economies. The proposal is to construct a firewall between
the lending and the deposit-creating functions of banks. In a parallel first step 1, DepositYear 2013, Volume 1, Issue 1
www.jheec.com
Page: 92
Pages: 93
M = waLa + wn Ln
(14)
The State must ensure that the growth of M is consistent with the production of basics.
The money distributed to workers would allow firms to capture liquidity as revenues.
Capitalists would use non DCIs to purchase inputs and start production. The Central
Bank would issue reserves and ensure that the reserves stock is consistent with its
projection of the production of basics. Reserve injections and withdrawals would be
effected through open market operations with non DCI intermediaries. Recent
scholarship supports our thesis (Duprey, 2013). Individual bank balance sheet data over
1990-2010 for 93 countries covering 459 public banks shows that public bank lending
decreases significantly less during economic downturns. Their balance sheets are less
vulnerable, their access to stable financing sources superior, and they are more inclined
to extend loan maturities. In general, public banks provide smaller loans to new
customers even at the peak of the cycle and cut back less on existing loan relationships
during downturns due to a limited dependence on short-term finance and secure public
support.
6. Conclusion
The production of basics by means of basics is financed by banks. The Central Bank, as an
important organ of the State, is able to generate and retain confidence in the monetary
circuit. The flow of new capital goods, on the other hand, depends on the expectations of
the future which are not stationary. The State, with its non-private decision-making
processes and time horizons, can effect long-term investments in basics investment.
Secondly, the time-honored practice of building firewalls between different financial
entities must be restored. In our case, the money-basics subsystem and hedge fundnonbasics subsystem must be separated. The dizzying pace of financial innovation can
continue without brakes in the latter case, throwing up winners and losers. Recalling
Keynes in conclusion, speculation could do no harm as bubbles on a steady stream of
enterprise. However, the situation is serious when enterprise becomes the bubble on a
steady stream of speculation.
References
Afonso, A & Furceri, D 2010, Government size, composition, volatility and economic
growth, European Journal of Political Economy vol. 26, no. pp. 517-532.
Berger, AN, Klapper, FL & Udell, GF 2001, The ability of banks to lend to informationally
opaque small businesses, Journal of Banking and Finance, vol. 25, no. 3, pp. 21272167.
Blackburn, R 2006, Finance and the Fourth Dimension, New Left Review, vol.39, pp. 3970.
Page: 94
Borras Jr, SM & McKinley, T 2006, The Unresolved Land Reform Debate: Beyond StateLed or Market-Led Models, UNDP International Poverty Research Centre, Policy
Research Brief No.2.
Bossone, B 2001, Do banks have a future? A study on banking and finance as we move
into the third millenium, Journal of Banking and Finance, vol. 25, no. 12, pp.
2239-2276.
*** 2002, Should Banks be Narrowed?, Levy Economics Institute Working Paper 354.
*** 2004, Financial Development and Industrial Capital Accumulation, World Bank
Policy Research Working Paper No. 2201.
Bossone, B & Sarr, A 2005, Non-Credit Money to Fight Poverty, in Fontana, G. &
Realfonzo, R (eds.), The Monetary Theory of Production, Palgrave, Macmillan
London, pp. 187-201.
Byres, T 1996, Capitalism from above and Capitalism from below, Macmillan Press,
London.
Cartelier, J 2007, The hypostasis of money, Cambridge Journal of Economics,
vol. 31, no. 2, pp. 217-233.
Cavalieri, D 2013, A theory of capital as value in progress, MPRA paper No.47197, May
2013
Cencini, A 2005, Macroeconomic Foundations of Macroeconomics, Routledge, London.
Crowe, C 2006, Inflation, Inequality, and Social Conflict, IMF Working Paper/06/158
Das, RC 2007, Introduction: Peasant, State and Class, The Journal of Peasant Studies,
vol.34, nos. 3&4, pp. 351-370.
Duprey, T 2013, Bank Ownership and Credit Cycle: the lower sensitivity of public bank
lending to the business cycle, Paris School of Economics Working Paper 2013-5
Ellis, F 2006, Agrarian Change and Rising Vulnerability in Rural Sub-Saharan Africa,
New Political Economy, vol.11, no.3, pp. 387-397.
Femir, F 2007, The Rise of Rentier Capitalism and the Financialization of Real Secgors in
Developing Countries, Review of Radical Political Economics, vol.39, no.3, 351359.
Foley, D & Taylor, L 2006, A heterodox growth and distribution model, in N. Salvadori
(eds.) Economic Growth and Distribution, Cheltenham: Edward Elgar, pp. 75-99.
Ghosh, J 2005, The Economic and Social Effects of Financial Liberalization: A Primer for
Developing Countries, United Nations, Department of Economic and Social
Affairs Working Paper No.4
Goodhart, CAE 2004, Financial Stability and Economic Growth: Explaining the links, in
Goodhart, CAE (ed.), Money, Stability and Growth, Palgrave, London, pp. 183-206.
*** 2008, The background to the 2007 financial crisis, Journal of International
Economics and Economic Policy, vol.14, no. 4, pp. 331-346.
Goody, J 2006, Gordon Childe, the Urban Revolution, and the Haute Cuisine: An
Anthrpro-archaeological View of Modern History, Comparative Studies in Society
and History, vol.48, no.3, 503-519.
Pages: 95
Gorton, G, 2012, Some reflections on the recent financial crisis, NBER Working Paper
18397
Graziani, A 2003, The Monetary Theory of Production, Cambridge University Press,
Cambridge.
Hein, E 2008, Money, Distribution, Conflict and Capital Accumulation, Palgrave
Macmillan, Hampshire.
Kaleki, M 1976, Essays on Developing Economies, The Harvester Press, Sussex.
Medio, A & Lines, M 2001, Nonlinear dynamics: a primer, Cambridge University Press,
Cambridge.
Merton, RC & Brodie, Z 2007, The Design of Financial Systems: Towards a Synthesis of
Function and Structure, NBER Working Paper 10620
Moseley, F 2008, The Macro-monetary Interpretation of Marxs Theory: A Reply to
Rvagnanis Critique, Review of Radical Political Economics, vol. 40, no. 1, pp. 107118.
Perraton, J 2007, Evaluating Marxian contributions to development economics, Journal
of Economic Methodology, vol. 14, no. 1, pp. 27-46.
Rosenzweig, MR 2001,Savings Behaviour in Low-Income Countries, Oxford Review of
Economic Policy, vol.17, no. 1, pp. 40-54.
Runde, J 2002, Assessing creditworthiness and small-firm bank lending, in Arestis, P,
Desai. M & Dow, S (eds.), Methodology, Microeconomics and Keynes, Routledge,
London, pp. 203-214.
Seccareccia, M 1994, Socialization of investment, in Arestis, P & Sawyer, M (eds.), The
Elgar Companion to Radical Political Economy, Edward Elgar, Aldershot, pp. 375380.
Smithin, J 1989, The Composition of Government Expenditure and the Effectiveness of
Fiscal Policy, in Pheby, J (ed.), New Directions in Post-Keynesian Economics,
Aldershot,
Edward Elgar, pp. 209-227.
Shubik, M 2006, Theory of Money and Financial Institutions, Cowles Foundation
Discussion Paper No.1572
Thornton, DL 2008, Monetary Policy: Why Money Matters and Interest Rates Dont,
Federal Reserve Bank of St. Louis Working Paper 2008-011A
Toporowski, J 2006, Michal Kalecki, in Clark, DA (ed.), The Elgar Companion to
Development Studies, Cheltenham, Edward Elgar, pp. 304-308.
Vercellone, C 2007, From Formal Subsumption to General Intellect: Elements for a
Marxist Reading of the Thesis of Cognitive Capitalism, Historical Materialism,
vol.15, pp.13-36.
*** 2008, The new articulation of wages, rent and profits in cognitive capitalism, Queen
Mary University School of Business and Management Working Paper No.2
Volcker, P 2012, Unfinished Business in Financial Reform, International Finance, vol.15,
no.1, pp. 125-135.
Page: 96
Appendix
The Monetary Circuit
A. BASICS
AGRICULTURAL CAPITALIST (AC)
Deposit with bank
1 + loan deposit
2 - wage payments
3 - payments to IC
3' + sales to AW
3 + sales to IW
Financial Obligations
+waLa
-waLa
-PaCa
+waLa
+wnLn
+waLa
0
+waLa
-waLa
Pages: 97
B. FINANCIAL INTERMEDIATION
BANK
Loan Account AC
Deposit Account AC
1 + loan claim
+waLa
+waLa
2 - debt payout
-waLa
2 - payment to AW
-waLa
3 + payment from AW
+waLa
3 + payment from IW
+wnLn
3 - payment to IC
-PaCa
Deposit Account AW
2 + payment from AC
+waLa
3 - payment to AC
-waLa
Loan Account IC
Deposit Account IC
1 + loan claim
+wnLn
+wnLn
2 - debt payout
-wnLn
2 - payment to IW
-wnLn
3 + payment from IC
+PnCn
3 + payment from AC
+PaCa
3 - payment to IC
-PnCn
Deposit Account IW
2 + payment from IC
+wnLn
3 - payment to IC
-wnLn
Page: 98
C. NON BASICS
INDUSTRIAL CAPITALIST (IC)
Deposit with bank
Financial Obligations
1 + loan deposit
+wnLn
2 - wage payments
-waLa
3 - payments to AC
-PnCn
3 + sales to AC
+PaCa
3 + sales to IC
+PnCn
+wnLn
2 + wages from IC
wnLn
3 - payments to AC
-wnLn
Pages: 99