Determinants of Total Factor Productivity in the Italian Regions *
|
|
- Brett Chandler
- 5 years ago
- Views:
Transcription
1 Determinants of Total Factor Productivy in the Italian Regions * Guido Ascari Universy of Pavia Valeria Di Cosmo Universy of Pavia April 2005 Abstract This paper investigates the determinants of TFP for Italian regions. We find strong evidence in favour of the factors commonly suggested by the theoretical lerature. In particular, our results points to research activy, human capal, social capal, infrastructures and agglomeration spillovers as the main determinants of TFP differences across Italian regions. Furthermore, we investigate the difference between Northern and Southern regions, finding significant differences, particularly regarding the effects of social capal and crime. JEL Classification: O47, C23, R11 Keywords: Total factor productivy, Italian regions, panel data * We particularly thank Carluccio Bianchi, Carolina Castagnetti, Roberto Golinelli, Emanuela Marrocu and Eduardo Rossi for advices and useful discussions. We are grateful to Giacomo Degli Antoni who kindly made available to us his measure of social capal for Italian regions. Corresponding author: Department of Economics and Quantative Methods, Via San Felice 5, Pavia, Italy. gascari@eco.unipv..
2 1 1. Introduction The role of technological progress has been the subject of an increasing attention in the lerature to understand the differences between developed and undeveloped countries. Prescott (1998), Prescott and Parente (2004) highlight the role of total factor productivy (TFP) in explaining international income differences (see also Easterly and Levine, 2001). As shown in Parente and Prescott (2004), TFP, together wh human capal, can explain a large part of the income differences across countries. As Prescott (1998) wres, the standard economic growth theory first needs to analyse the TFP determinants to become also a theory of international income differences. The same is true if one wants to explain the different degree of development whin a country. This seems particularly crical for countries whose areas or regions are characterised by very different degree of development, as is clearly the case of Italy. In this paper we look at the main determinants of TFP across the twenty Italian regions from a macroeconomic perspective. The novelty of the paper is twofold. First, to our knowledge we are the first to employ a panel data approach in estimating TFP determinants in Italian regions. Second, we will also divide the sample in two sub-groups (i.e., Northern regions and Southern regions) and investigate the difference between TFP levels and s determinants in the North and in the South of Italy. Our paper is close in spir to Aiello and Scoppa (2000), that is, to the best of our knowledge, the only paper in the lerature looking at TFP determinants in Italian regions from a macroeconomic perspective. Aiello and Scoppa (2000) first make a growth accounting exercise for the year 1997 showing that a large part of regional disparies are due to differences in TFP level across regions; then estimate a model of TFP determinants in a cross section referred to the 20 Italian regions for the year As we said above, our paper expands this analysis to a panel data approach, using a sample period ranging from 1985 to We do so to allow for heterogeney across regions and time. Taking into account the time series dimension provides a much richer information base to investigate the relationships among the data. This is particularly important in analysing the Italian economy since Italian regions present a great variabily in income levels and almost oppose, if not divergent, development paths of two macro-regions: the North and the South. For this reason, we will also look at the two macro-regions separately.
3 2 Other papers in the applied macroeconomic lerature investigate Italian regional TFP. Marrocu et al. (2000) estimate the TFP across Italian regions and economic sectors, relaxing the assumption of constant return to scale, wh a panel cointegration approach. The authors find the presence of constant return to scale at a national level, while the estimates of individual production functions for Italian regions show great differences in factor elasticies across regions and sectors. Di Liberto et al. (2004) employ a panel data approach (based on Islam, 1995) to study whether TFP is converging across Italian regions. This paper also highlights the heterogeney of TFP levels across regions. While linked to ours, Di Liberto et al. (2004) deal wh a different topic, focusing on TFP convergence and showing that only in the sub-period they can detect convergence in TFP levels across regions. Moreover, a series of papers investigate the role of public capal in explaining the economic growth of Italian regions. Picci (1999) estimate a production function including public capal, finding a significant role for public capal. The work by Bonaglia et al. (2000) is closer to ours since they analyse both the link between public capal and TFP, and the difference between North and South. They find a posive relationship between TFP and public infrastructure, especially for the South. More recently Petraglia (2003) cricises these studies, using a non-parametric stochastic frontier approach to conclude that public capal played a significant role in contributing to productivy gains not directly entering as an input in the production function but rather as a posive externaly in enhancing both technical efficiency and technological progress. Finally, Coppola et al. (1998) investigate the role of human capal in the development of Italian regions. On the basis of a cross-section regional regression of a production function that includes human capal as an input, they find a posive contribution of human capal to the growth of Italian regions. Many of the paper above emphasize the large differences in TFP levels across Italian regions, coherently wh the well-known dualism between Northern and Southern Italy. None of them however investigates what are the factors determining such differences. This is instead clearly an important task also from a policy perspective. In this paper we therefore explore if the TFP determinants, as suggested by the theoretical lerature, could explain the different regional TFP estimates. One important caveat is to be highlighted. The empirical lerature on productivy, growth and R&D is very large and varied, and studies differ greatly in terms of the level of aggregation (macro,
4 3 industry, firm or plant level). We share wh all the papers above a macroeconomic perspective in looking at productivy and growth. When we talk about productivy we basically mean TFP, that is, the so-called Solow residuals obtained from a tradional growth accounting analysis. We are definely aware of the well-known limations of the Solovian growth-accounting method in calculating total factor productivy (see, e.g., Atella and Quintieri, 2001). This method is indeed based on strong assumptions, such as constant returns to scale and competive output market, which are likely not to be strictly valid for Italian regions. Nonetheless, since this is the most commonly employed method in the economic lerature to calculate TFP from in macroeconomics. It seems therefore to us worth investigating, for the first time for Italian regions, the determinants of TFP, as largely defined in the macroeconomics lerature. The structure of the paper is as follows. Section 2 introduces the measure of TFP at national level and describes the behaviour of regional TFP across time and regions. Section 3 presents our benchmark econometric analysis and results. Section 4 modifies the TFP measure to take into account the possibily of human capal being an input to the production process. Section 5 concludes. 2. Measurement and data description 2.1. Measurement of inputs and output We measure output as the value added at factor costs, in real Italian lire, at constant prices of 1995, source ISTAT. Labour input is simply the labour workforce, that is, we include both of autonomous and subordinate workers. The only measure of labour input available from ISTAT is the number of workforce, so there is no labour qualy measure. As a measure for capal input we use the gross capal stock data constructed from gross investment data, by Bonaglia and Picci (2000) and Picci (1999). Since this dataset ends in 1995, we calculate the capal stock for the sample employing the perpetual inventory method and using a fixed depreciation rate calculated to be the average depreciation rate of the years TFP: measurement and description We calculate TFP applying the standard Solovian growth accounting methodology. Denote regions by i =1,...N. Regional GDP (Y) at time t is produced wh labour (L) and physical capal (K) according to a standard neoclassical production technology, that is
5 4 Y = A F(K, L ) (1) where A is the TFP. Consider a tradional Cobb-Douglas production function wh constant return to scale, Y = A K 1 α i L α i (2) TFP is then simply defined by: A Y =. To calculate the TFP levels we hence just need a 1 K L α i α i value for α i. Under the (heroic) assumption of perfect factor markets, the exponent α i equals the wid Li labour share of output, hence α i =, where wid is the nominal wage of subordinate workers at Y i constant prices. 1 Note that we allow α i to vary across regions, but not across time. This is because of the high variabily across time of the labour share whin the different regions, reflecting forces that are most probably not linked wh technology. α i instead is a technological parameter and should not vary too much in a sample of just 15 years. We therefore calculate α i as the average of the labour share across the sample period for each region. 2 Before proceeding wh the econometric analysis, is instructive to describe the main features of the TFP levels of the Italian regions. As noted in the introduction, the lerature suggests a large degree of heterogeney among the Italian regions. Our results confirm such finding. Figure 1 plots the time series behaviour of TFP levels for Italian regions. Some features are evident: 1) Different regions exhib very different TFP levels: the highest level (Lazio) is almost 3 times as large as the lowest one (Basilicata) along the sample period. 1 Since for Italian regions, there are no data available for the wages of autonomous workers, we consider the subordinate workers nominal wage w id as a proxy for the autonomous one. 2 Admtedly our results are not robust to the alternative assumption of α being the same across regions. We thank a referee to make us check this robustness argument. However, given the well-known strong heterogeney among Italian regions, one of the main goals of the paper is to analyse the difference among them and the peculiary of regional TFP determinants in a panel data approach. It seems to us then a natural hypothesis to let α varies across regions.
6 Piemonte Valle d'aosta Lombardia Trentino Alto Adige Veneto Friuli Venezia Giulia Liguria Emilia Romagna Toscana Umbria Marche Lazio Abruzzo Molise Campania Puglia Basilicata Calabria Sicilia Sardegna Figure 1. TFP levels of Italian regions Piemonte Valle d'aosta Lombardia Trentino Alto Adige Veneto Friuli Venezia Giulia Liguria Emilia Romagna Toscana Umbria Marche Lazio Abruzzo Molise Campania Puglia Basilicata Calabria Sicilia Sardegna average TFP Figure 2. Average TFP levels for Italian regions
7 standard deviation of TFP across regions Figure 3. Standard deviation of TFP across Italian regions, Center-Northern regions Southern regions Figure 4. Average TFP levels for Central-Northern and Southern regions, correlation between TFP and Y/L correlation between K/L and Y/L Figure 5. Correlation between TFP, output per worker and capal per worker,
8 7 2) There is also a very high degree of persistence in the ranking of TFP levels across regions, which basically does not change along the sample. 3) There is mixed evidence on convergence. Figure 3 shows the standard deviation of TFP levels across regions: while the 80 s seem to be a period in which the gap between rich and poor regions widened, in the second part of the 90 s some convergence took place. 4) As expected, TFP levels of Northern and Central regions (North) are substantially bigger than the ones of Southern regions (South), as evident from Figure 2 and Figure 4. 3 Figure 2 plots the average of TFP across the sample period for each region, showing that the Northern and Central regions are basically all above the average (roughly 0.15), while the Southern ones are below. Figure 4 also visualises the same fact across time periods. Finally, note that from equation (2) we can wre: Y L K = A L 1 α i, which shows that the level of labour productivy is explained only by two factors: TFP and the capal-labour ratio. It follows that the regional differences in labour productivy can be explained through disparies in these two factors. Figure 5 shows that most of the heterogeney in output-per-worker among different regions is explained by TFP rather than by the capal-labour ratio. Indeed, the correlation between TFP and output per worker is always above 0.9 in each year of the sample, while the correlation between capal-labour ratio and output per worker is rather low. This last evidence urges for a better understanding of the TFP determinants. This is what we move to next. 3. Empirical results In this section we present the results of the estimation of TFP determinants for the panel of Italian regions. We then need to detect which are the main factors affecting total factor productivy The empirical model Wh respect to the possible determinants of TFP level, the theoretical lerature provides us wh a good guidance. First, from the seminal contributions of Romer (1986, 1990), Grossman and Helpman (1991) and Aghion and Howt (1992), an immense lerature on endogenous growth stresses the role of R&D activies as the main engine of growth and TFP. Many models (e.g., 3 We define North the macro-region composed of Northern and Central regions, that is, Piemonte, Valle d Aosta, Lombardia, Trentino Alto Adige, Friuli, Liguria, Emilia Romagna, Toscana, Umbria, Marche and Lazio. Abruzzo, Campania, Puglia, Basilicata, Calabria, Sicilia and Sardegna instead make up the macro-region South.
9 8 Romer, 1990, Aghion and Howt, 1992) imply also a direct relationship between the number of people employed in the R&D sector and growth in TFP. This scale effect associated to the research process was empirically cricised by Jones (1995) and was the subject of an intense debate in the growth lerature. From an empirical point of view, a vast macroeconomic lerature investigating the national sources of economic growth (e.g., Cameron, 2003, Griffh et al., 2000, 2003) confirmed the linkage between R&D expendures, TFP, and growth. In the ISTAT database two variables are readily available to catch these effects at regional level: the level of private enterprises spending in R&D and the number of researcher on total workforce. Second, at least from Lucas (1998), growth models cannot abstract from human capal. A sufficient level of knowledge in the workforce is necessary to acquire and explo higher level of technology. Hence, the level of human capal is considered a natural necessary condion for high TFP level. There is however a potential problem: human capal can be thought as an input in the production process rather than a source of higher TFP (see, e.g., Mankiw et al., 1992 and Parente and Prescott, 2004). In the paper, we will tackle this issue constructing two different measures of TFP for Italian regions (one wh and one whout human capal as production input) and testing for human capal effects. Measuring human capal is not by any means a simple task, but there is a widespread agreement on using some measures of the average education attained by the labour force. Fortunately, again the ISTAT database provides us wh the average years of schooling in the workforce at the regional level. Third, public capal is also often indicated as a possible source of growth (e.g., Barro, 1990). As discussed in the introduction the role of public capal in affecting TFP levels and growth in Italian regions has recently been the subject of an interesting debate in the lerature. Both Picci (1999) and Bonaglia et al. (2000) find that public infrastructure (i.e., roads, railways, etc.) is the most important (if not the only one) component of public capal affecting TFP and growth. It would therefore be desirable to introduce some measures of in our estimates. More recently, a fast growing lerature stresses the importance of social capal. Roughly, the idea is that uny and trust whin the social communy creates the right environment for the development of economic activy. As stressed by Temple (2001), is difficult to find a suable variable for social capal. However, we tried to capture by two variables. The first one is provided again by ISTAT and measures the number of voluntary homicides in the regions. The second one has been constructed and used by Degli Antoni (2004). It is the main common factor in
10 9 a principal component analysis of three indicators of crime and lawsus (see Appendix for details). 4 We think that while the first variable is just a proxy for the amount of crime, the second one can work as a proxy for trust and cohesion whin the local communy. The fact that the two variables provide two different information is demonstrated by their very low correlation (-0.125). The theoretical and empirical lerature on economic geography and regional economics stresses the possible existence of spillover effects due to the agglomeration of firms in a particular location. A higher population densy could in principle allow firms and researchers to realise economies of scale among them. Following Aiello and Scoppa (2000), we hence use population densy as a proxy for agglomeration economies. The lerature proposes other possible sources of TFP. Cameron (2003) includes in his analysis the proportion of manual male workers covered by collective agreements multiplied by the proportion of manual males in the workforce, to understand the role played by labour unions in the productivy growth. Senhadji (2000) and Rivera Batiz (2002), instead, investigate the social components of TFP, as external shocks, polical stabily and democracy. Many of these effects however have a country dimension so that they are common to regions in the same country. Given the discussion above, we run the following panel data regression: log TFP = β 1 R&D + β 2 RICPOP + β 3 EDU +β 4 SOCIALK + + β 5 CRIME + β 6 DENSITY + β 7 INFRASTR + β 8 trend + v (3) where: - v = e i + u is the panel data error term and e i indicate the presence of unobserved cross section effects in our model; - R&D, private enterprises spending in R&D divided by the total fixed investments; 5 - RICPOP, the number of researcher on total workforce; - EDU, the average years of schooling of the labour force; - SOCIALK, the variable in Degli Antoni (2004) used as a proxy for social capal; - CRIME, the number of homicides divided by the population; - DENSITY, population densy, - INFRASTR, spending in total infrastructure. 4 We gratefully thank Degli Antoni, who kindly provided us wh this variable. The way Degli Antoni (2004) builds this variable is described in the Appendix. Sources of all the data series are also presented in the Appendix. 5 We decided to normalise R&D spending by total fixed investment, in order to have a variable that measures the R&D spending as a share of total investment. Results are however qualative robust to other normalisation (e.g., GDP).
11 10 We estimate the above equation for Italian regions in the sample Note that we take the natural logarhm of TFP to try to control the potential non-linear relation between the dependent variable and regressors. Moreover, we introduce a time trend to allow for growth in TFP that is exogenous wh respect to the variables included in (3). Furthermore, the two variables R&D and RICPOP show some degree of collineary. 7 Indeed, the correlation between the two variables is and a simple regression of one variable on the other exhibs a very high t-value of Given the debate following Jones (1995), would however be que interesting to disentangle the effects of these two variables, to assess the presence or not of scale effects on TFP. In order to cope wh this potential problem, instead of RICPOP, we use the variable RICPOP_R&D that is simply given by the residual of an OLS regression of RICPOP on R&D. 8 The significance of this variable would signal that RICPOP adds something in the explanation of the behaviour of TFP that is not already enclosed in R&D. We will hence use the variable RICPOP_R&D instead of RICPOP in all the estimations that follow below. Finally, we stress that a multicollineary problem instead does not exist between the variables CRIME and SOCIALK, as one may be inclined to think. 9 This supports our belief that these two variables measure different things: the amount of crime the former, and cohesion whin the local communy the latter. As we want to estimate such a model we firstly have to verify the presence of unobserved effects, secondly (if these effects are present) we have to verify which estimation technique is consistent and (possibly) the most efficient. Assuming E(u x i, e i ) = 0, where x i is the matrix of regressors for the i un, the general form of the block diagonal element of the variance-covariance matrix, if there are no serial autocorrelation or heteroskedasticy, is given by 10 6 We have to restrict the sample wh respect to the Figures in Section 2, because some of the regressors are available only from We thank a referee for pointing this possibily to us. 8 We obtained very similar results when we used the residuals of this regression obtained wh a panel data random effect model. 9 Recall that the correlation between CRIME and SOCIALK is equal to ( 0.125). 10 If there is no heteroskedasticy in stochastic residual, the diagonal of the Ω matrix is given by E(v v ) = E(e 2 i ) + E(u 2 ) + 2E(e i u ) = σ 2 2 e + σ u As there is no autocorrelation in stochastic term, the other elements of this matrix are given by E(v v is ) = E[(e i + u is )(e i + u )] = σ 2 2 e + E(e i u ) + E(e i u is )+ E(u u is ) = σ e
12 σν σ e Ω= 2 2 σe σ ν (4) 2 If e is constant between uns, there are no unobserved effects. Then, testing whether σ e =0 is statistically equivalent to say that there are no unobserved effects. We perform the Breush-Pagan test assuming a normal distribution of ν. The test performed wh our data rejects the null hypothesis at 1% level, and wh a p-value of 0. It is well known that GLS estimator is consistent only if we can demonstrate that regressors and unobserved effects are unrelated. Moreover, asymptotically, the GLS estimator is more efficient than the OLS estimator, since the maximum likelihood estimator has the minimum variance in the class of asymptotically normal estimators. We then perform the Hausman test in order to catch differences between fixed and random effects estimates. We cannot reject the null hypothesis wh a p-value of , and therefore the test suggests estimating the model wh GLS. We then perform the Whe test that strongly rejects the null hypothesis of absence of heteroskedasticy in residuals, while the Wooldridge (see Wooldridge, 2002, p. 275) test rejects the hypothesis of absence of first order autocorrelation whin uns. In other words, the hypothesis: E( u ) = 0 2 E(u is, u ) = 0 t = 1, 2 T all t s fail. In this case, the variance-covariance matrix is still a block diagonal matrix, whose elements on the principal diagonal are now given by ρi ρi... σ ε i 2 ' ρi 1 ρi... σνi 1 ρ + ee i i (5) i ρ 2 i ρ i 1... where u = ρ I u -1 + ε ; ε ~ WN and e i is the uny vector. Under the (not rejected) null hypothesis of the Hausman test, a consistent estimator for Ω is given by N 1 N i= 1 Ω= ˆ ˆˆ νν ' (6) i i
13 where the ν ˆ i are the OLS residuals. This is the estimator we will use in all our following estimates Italy Table 1 shows the results for the Italian regions. Five variables (i.e., R&D, EDU, SOCIALK; DENSITY; INFRASTR) are very significant and wh the expected posive sign. Therefore the variables suggested by the lerature above are indeed que important in explaining the behaviour of TFP in the Italian regions. Hence, we can conclude that: (i) R&D activy is an important determinant of Solow residuals; (ii) other forms of capal (i.e., human, social and public) significantly affect TFP; (iii) there are posive spillover effects of agglomeration on TFP. Table 1. TFP determinants in Italian regions 12 logtfp Italy R&D.3168** (.1573) RICPOP_R&D.7551 (.5003) EDU.0786** (.0063) CRIME - SOCIALK.028** (.005) DENSITY.0001** ( ) INFRASTR.00002** (6.54*10-6 ) Wald χ 2 (6) = Log-Likelihood = 662 p- value Hausman test for endogeney 0381 The coefficient on RICPOP_R&D is also posive, but not significant, even if s t-value (i.e., 1.51) is not too far from the 10% significance level threshold. 12 CRIME has the expected negative sign, but is highly not significant so is not included in the final regression we report in Table 1. Finally note that also the trend variable is not included since not significant. The trend variable however is very collinear wh the variable EDU, which shows an evident upward trend for all the regions. Correlation between EDU and a trend is very high (0.9) and a regression of EDU on a trend gives an extremely high t-value. If we do not include EDU in the regression the trend is then very significant showing that the TFP is characterised by an upward trend as was evident from Figure 1. When the variable EDU is included in the regression, however, the trend is not significant any 11 Note that we allow the first order autocorrelation coefficients in the error term to be different across uns. 12 That is why we decided to keep in the reported Table 1.
14 13 longer (which is therefore not included in our final regression in Table 1). Moreover, to demonstrate that the relationship between EDU and TFP is not spurious, we include in the regression the trend and the de-trended EDU, that is, the residuals of an OLS regression of EDU on a trend. Even in this case EDU exhibs a very significant coefficient (t-value equals 3.48) wh the expected posive sign. 13 We are therefore confident that the relationship between EDU and TFP is not at all spurious and we hence include EDU in the final regression, while excluding the trend (see also footnote 16, p. 16 in the next section). A potential and important problem in our estimates is the endogeney of some of the regressors. 14 In particular, one may think that R&D spending, the number of researchers and the level of human capal are influenced by the dynamics of TFP. We do test for endogeney of these three variables by using the Hausman test for endogenous variables. That is, we estimate our equation using lagged R&D, RICPOP and EDU as instruments for those three variables and then, we perform an Hausman test between the instrumented an the not instrumented estimates. Hausman test strongly accepts the null hypothesis of no systematic difference between the coefficients of the two estimates (p-value = 0.38), indicating that we can treat these variables as exogenous. Note that we run this test for all the estimations presented in what follows and in none of them the Hausman test detects the presence of endogenous variables (see Table 2 and 3). 3.3 North and South In this section, we analyse the well-known dichotomy between the North and the South among Italian regions. Our data allow us to separately investigate the TFP determinants for the Northern regions (i.e., Piemonte, Valle d Aosta, Lombardia, Trentino Alto Adige, Friuli, Liguria, Emilia Romagna, Toscana, Umbria, Marche and Lazio) and the Southern ones (i.e., Abruzzo, Campania, Puglia, Basilicata, Calabria, Sicilia and Sardegna). The estimation method is the same as above, since again the Breush-Pagan test rejects the null hypothesis, while the Hausman test suggests the use of the random effect estimator. Moreover, both the Whe test and the Wooldridge test reject the null hypothesis The trend was obviously very significant too wh a coefficient of and s.e Again we thank a referee for drawing our attention to this problem. 15 However, now we do allow the unobservable components to be also correlated across uns, given that we are analysing regions very close and similar across them. Note that: (i) this was not possible in the case of Italy as a whole, since the number of estimated covariances increases exponentially wh uns; (ii) estimations are que robust to assuming the absence of cross-sectional correlation among error terms.
15 14 The results for the North are very similar to the one for Italy, both from a qualative and quantative point of view. Also for the Northern regions CRIME is highly not significant, while all the other variables exhib a very significant coefficient wh the expected posive sign. The significance of RICPOP_R&D is then the only difference between North and Italy as a whole. Moreover, the values of the coefficients of the significant variables are very close to the ones of Italy (except for the social capal variable). Table 2. TFP determinants in Italian regions: North and South logtfp North South R&D.1242** (.0225) RICPOP_R&D.8127** (.0663) EDU.0573** (.0006) ** (.7306) ** (2.3609).12327** (.0105) CRIME ** ( ) - SOCIALK.0027** (.001) DENSITY.00053** ( ) INFRASTR ** (8.95*10-7 ) ** ( ) Wald χ 2 (6) = Wald χ 2 (5) = Log-Likelihood = 241 Log-Likelihood = p- value Hausman test for endogeney Results are instead a b different for the Southern regions. First, also for the South both R&D and RICPOP_R&D are strongly significant, but the values of their coefficients are much higher than in the case of Italy and the North. One possible interpretation for this result lies in a sort of law of decreasing returns in affecting TFP. The Southern regions exhib relatively lower both R&D spending levels and number of researchers such that an increase in those factors would simply be more effective on TFP, coeteris paribus, in the South rather than in the North. Note that the same argument can be applied to the other significant variables (i.e., EDU and INFRASTR) since both show a coefficient higher than the equivalent for Northern regions. Besides, since DENSITY is not significant, agglomeration effects are not important in the Southern regions, as validated by the relative absence of productive districts in the South compared wh the Centre-North of Italy. More notably, oppose to Italy and the North, CRIME is now significant wh a posive sign, while SOCIALK is not significant. This is a striking result that is possibly due to the strong peculiary of the Italian Mezzogiorno and the substantial presence of organised crime. Since many productive
16 15 activies are controlled by organised crime, is not so surprising that the variable CRIME has a posive effect on TFP, while SOCIALK is spoiled and therefore not significant. 4. TFP and human capal as an input to production After the seminal works of Lucas (1988) and Mankiw et al. (1992), is common in the lerature to consider human capal as an input of the production function. In this case, we can recalculate the TFP as A Y = (7) 1 α i α i K ( LH ) where H is the human capal measured by the variable EDU. Figure 6 plots the time series behaviour of the TFP for the sample The main regional TFP features described in Section 2 are qualatively still valid, especially regarding the North vs. South difference. The main, and important, change is that now the TFP does not show any evident trend Piemonte Valle D'Aosta Lombardia Trentino Alto Adige Veneto Friuli Venezia Giulia Liguria Emilia Romagna Toscana Umbria Marche Lazio Abruzzo Molise Campania Puglia Basilicata Calabria Sicilia Sardegna Figure 6. TFP levels of Italian regions, considering human capal as an input to production We then perform the same analysis of section 3, obviously removing EDU from the regressors. We employ the same econometric technique, since the results of the various tests were the same as in the previous section. The results are summarised in Table 3.
17 Italy The estimation for the panel of all the Italian regions confirms the previous results, revealing their robustness. In particular, both RICPOP_R&D and CRIME are still not significant as in the previous estimates. RICPOP_R&D is now highly not significant so that we om from our preferred regression showed in Table 3. We can therefore conclude that number of researchers is not a determinant of regional TFP once one controls for R&D spending in the panel of the 20 regions. Moreover, interestingly enough, the trend variable is not significant, suggesting the possibily that TFP and EDU were cointegrated in the model of the previous section. 16 This is a particular nice result that we think corroborates even more our estimates and their coherence between the two methods of calculating TFP levels. Finally, note that also the values of the significant coefficients are que similar. Table 3. TFP determinants, considering human capal as an input to production logtfp Italy North South R&D.3048** (.1464).2986** (.0524) RICPOP_R&D ** (.21) CRIME ** ( ) SOCIALK.0392** (.0054) DENSITY.0016** ( ) INFRASTR ** (7.37*10-6 ).00925** (.0017) ** ( ) ** (2.91*10-6 ) TREND * ( ) 2.94** (.9179) 10.94** (3.1857).0018** (.0004) ** ( ).00595** (.0016) Wald χ 2 (4) = 1420 Wald χ 2 (7) = Wald χ 2 (5) = Log-Likelihood = Log-Likelihood = Log-Likelihood = p- value Hausman test for endogeney North and South Wh respect to the estimates regarding the Northern regions, results are very similar both qualatively and quantatively to the ones in Table 2 in the previous section. In particular, all the significant variables in Table 2 are still very significant wh the expected posive sign and analogous coefficients. There are however two differences. First, the CRIME variable is now significant wh the expected negative sign, while was not significant in the estimates in Table Apart the inclusion of a time trend, we chose to ignore the non-stationary issue in the econometric treatment of the previous section for two reasons: (i) 15 time observations do not seem to be enough to allow any sensible panel cointegration analysis; (ii) the analysis in this section does not seem to suffer of this problem, while anyway confirms our result. Cointegration analysis is then left to further research.
18 17 Second, the trend variable is significant wh a negative sign, showing a downward trend for the TFP in Northern regions. The value of the coefficient however is very low, meaning a very weak trend. 17 Moreover, while the log-likelihood does worsen excluding the trend variable, the estimates of the coefficients (and their significance) on the other regressors are not very much affected by s exclusion. Wh respect to the Southern regions, the results in the previous section are again confirmed. The significant variables are still the same ones and wh the same signs, while again SOCIALK and DENSITY are still not significant. The main difference wh Table 2 is the significant coefficient on the trend that shows an upward trend, contrary to the North that instead exhibs a downward one. On the one hand, also in the case of Southern regions, the coefficient signals a very slight trend (as evident from Figure 6), whose exclusion from the regression basically changes neher the loglikelihood nor the coefficients. On the other hand, taken together wh the results for the Northern region, this could explain why the trend is not significant when the panel data estimation includes all the regions: TFP in the South seems to have risen during the sample period due to forces exogenous to our equation, while the contrary seems true for the North of Italy. Particularly notable is again the posive and significant coefficient on CRIME that corroborates the previous finding, possibly pointing to the importance of organised crime in the management of the productive activies in the South of Italy. This result stresses even further, and possibly in a more worrying way, the Italian North-South dichotomy. 5. Conclusions A large lerature by now recognizes TFP as the main determinant of income differences across countries and regions whin countries. In this paper, we try to empirically assess the determinants of TFP levels for Italian regions. This seems a que important task given the great dispary in the level of development between the different Italian regions. To our knowledge, we are the first to use a panel data approach whin the standard growth accounting framework for Italian regions. Our results do confirm the predictions of the theoretical lerature, pointing to research activy, human capal, social capal, infrastructures and agglomeration spillovers as the main determinants of TFP differences across Italian regions. In 17 Note that the coefficient on trend variable when TFP is calculated as in Section 2 and when the variable EDU is detrended, is.006, that is, posive and in absolute value three times bigger than the coefficient in Table 3.
19 18 particular, going back to debate stimulated by Jones (1995), R&D expendures rather than the number of researchers is more important in affecting regional TFP. Finally, when we divide the Italian regions in two macro-regions (North and South), the notorious dualism of Italian development clearly emerges, wh respect to both the TFP levels and the estimation results, most notably wh respect to the variables related to social capal and crime.
20 19 Data Appendix Real Output: Data on real output are taken from the Office of National Statistics (ISTAT). As in 1995 a new accounting methodology (Sec95) was imposed by the new European classification, we recalculate the old series (Sec79) as follows. We assume that the proportion existing between the new 1995 series and the old one also exist between the old series ( ) and the new ones. That is, we impose the condion 1995Sec95 *1994Sec Sec 95 = 1995Sec79 Value added :Data are taken from the CRENoS database NewRegio02 available on the web se Labour input : Total employment is from the Office of National Statistics (ISTAT). Capal input: Data for individual capal are supplied by Bonaglia and Picci (2000) and Picci (1999). Total Factor Productivy: This is calculated from data on real output, labour input and capal input as above. As explained in the main text, the measure of TFP is based on growth accounting methodology. R&D: data on enterprise R&D spending are provided by ISTAT, as enterprise total fixed investments. RICPOP: the source is again ISTAT and is just the number of researchers over the total workforce. EDU: We define the total stock of human capal of the labour force as Di Liberto and Symons (1998), that is: EDU i = AS HK J Ji Ji where J is the schooling level, AS Ji is the number of years of schooling represented by level J, and HK J is the fraction of the labour force for which the Jth level of education represents the highest level attained. Whin the Italian system, primary level includes eight years of schooling, secondary level is attained after 5 years, and we consider that universy courses include four years of attendance. As explained by Di Liberto and Symons (1998), this indicator represents a measure of the average years of schooling of the labour force. Data are again taken from ISTAT. SOCIALK : This is a proxy for social capal, obtained by Giacomo Degli Antoni, Universy of Parma, through the statistical analysis of principal component of three variable (all provided by ISTAT): (i) the number of bill protests, (ii) reports to the police, (iii) labour lawsu proceedings. SOCIALK is then the first principal component. Degli Antoni (2004) shows that this common factor explains around 75% of the comovement of the variables. Degli Antoni (2004) then uses this variable to study the determinants of social capal (at regional and provincial level) and whether social capal posively influences economic growth, finding a posive and significant effect in the Italian regions. CRIME is the number of voluntary homicides and the source is ISTAT DENSITY is the regional population densy, the source is ISTAT INFRASTR is the amount of executed public works; the source is ISTAT.
21 20 References Aghion, P. and Howt, P. 1992, A model of growth through creative destruction, Econometrica, 60, Aiello, F; Scoppa, V. 2000, Uneven regional development in Italy: explaining differences in productivy levels, Giornale degli Economisti e Annali di Economia, 59, Atella, V., and Quintieri, B. 2001, Do R&D Expendures Really Matter for TFP?, Applied Economics, 33, Barro, R.J. 1990, Government spending in a simple model of endogenous growth, Journal of Polical Economy, 98, Bonaglia, F; Picci, L Il capale nelle regioni aliane, DSE,Universà di Bologna,374. Bonaglia, F; La Ferrara, E., and Marcellino,M., 2000, Public capal and economic performance: evidence from Italy, Giornale degli Economisti e Annali di Economia, 59, Cameron, G Why did UK manufacturing productivy growth slow down in the 1970s and speed up in the 1980s?, Economica, 70, Coppola, G., De Blasio, G., and Gallo, M. 1998, Development of Italian Regions: the Role of Human Capal, RISEC, 3, Degli Antoni, G., 2004, Capale Sociale e Cresca Economica: la Teoria ed il Caso Italiano, Tesi di Dottorato in Economia Polica, Universà di Pavia. Di Liberto, A; Symons, J Human capal stocks and the development of alian regions: a panel approach, Quaderni di ricerca CRENoS Di Liberto, A; Mura, R and Pigliaru, F How to measure the unobservable: a panel technique for the analysis of TFP convergence, Quaderni di ricerca CRENoS Easterly, W; and Levine, R It is not factor accumulation: stylized facts and growth models, The World Bank Economic Review,15, Griffh, R; Redding, S and Van Reen, J. 2000, Mapping the two faces of R&D: productivy growth in a panel of OECD industries, CEPR discussion paper, Griffh, R; Redding, S and Van Reen, J R&D and absorptive capacy: theory and empirical evidence, Scandinavian Journal of Economics, 105, Grossman, H.I., and Helpman, E. 1991, Innovation and growth in the global economy, Cambridge, MA: MIT Press. Islam, N.1995.Growth empirics : a panel data approach, Quarterly Journal of Economcis, 110,
22 21 Jones, C. I. 1995, Time series tests of endogenous growth models, Quarterly Journal of Economics, 110, Mankiw, N; Romer, D and Weil, N A contribution to empirics of economic growth, The Quarterly Journal of Economics, 107, Marrocu, E; Paci, R and Pala,R Estimation of total factor productivy for regions and sectors in Italy. A panel cointegration approach Quaderni di ricerca CRENoS Lucas, R On the mechanics of economic development, Journal of Monetary Economics, 22, Parente, S. and Prescott, E A unified theory of the evolution of international income levels, Federal Reserve Bank of Minneapolis Staff Reports, 333. Petraglia, C Total factor productivy growth and public capal: the case of Italy, Studi Economici, 78, Picci, L Productivy and infrastructure in the Italian regions, Giornale degli Economisti e Annali di Economia, 58, Prescott, E Needed: a theory of total factor productivy, International Economic Review, 39, Romer, P. M Increasing returns and long run growth, Journal of Polical Economy, 94, Romer, P.M. 1990, Endogenous technological change, Journal of Polical Economy, 98, S71-S102. Rivera-Batiz, F; Rivera Batiz, L Democracy, partecipation and economic development: an introduction, Review of Development Economics, 6, Senhadji, A Sources of economic growth: an extensive growth accounting exercise, IMF Staff Papers, 47. Temple, J Growth effects of education and social capal in the OECD countries, CEPR Discussion Paper Wooldridge, J.M. 2002, Econometric analysis of cross section and panel data, Cambridge: MA, MIT Press.
Productivity patterns in the Italian manufacturing sector
Productivity patterns in the Italian manufacturing sector Rosa Bernardini Papalia Dipartimento di Scienze Statistiche, Università di Bologna bernardini@stat.unibo.it Silvia Bertarelli Dipartimento di Economia
More informationESTIMATION OF TOTAL FACTOR PRODUCTIVITY FOR REGIONS AND SECTORS IN ITALY. A PANEL COINTEGRATION APPROACH
Emanuela Marrocu University of Cagliari and CRENoS (emarrocu@unica.it) Raffaele Paci University of Cagliari and CRENoS (paci@unica.it) Roberto Pala University of Cagliari (rpala@amm.unica.it) ESTIMATION
More informationThe Impact of the Economic Crisis on the Territorial Capital of Italian Regions.
The Impact of the Economic Crisis on the Territorial Capital of Italian Regions. Cristina Brasili, Annachiara Saguatti, Federica Benni, Aldo Marchese, Diego Gandolfo Department of Statistics P.Fortunati,
More informationABSORPTIVE CAPACITY IN HIGH-TECHNOLOGY MARKETS: THE COMPETITIVE ADVANTAGE OF THE HAVES
ABSORPTIVE CAPACITY IN HIGH-TECHNOLOGY MARKETS: THE COMPETITIVE ADVANTAGE OF THE HAVES TECHNICAL APPENDIX. Controlling for Truncation Bias in the Prior Stock of Innovation (INNOVSTOCK): As discussed in
More informationSub-national PPPs: Country case studies. Publications, experiments and projects on the computation of spatial price level differences in Italy
3rd Meeting of the Country Operational Guidelines Task Force Sub-national PPPs: Country case studies Publications, experiments and projects on the computation of spatial price level differences in Italy
More informationA multilevel strategy for tourism development at regional level The case of the Marche Region
Tourism and Biodiversity in Protected Areas 8th European Charter Network Meeting and Charter Awards (Brussels 6 november 2013) A multilevel strategy for tourism development at regional level The case of
More informationNon-compensatory Composite Indices for Measuring Changes over Time: A Comparative Study
Non-compensatory Composite Indices for Measuring Changes over Time: A Comparative Study Matteo Mazziotta and Adriano Pareto Italian National Institute of Statistics Introduction In the recent years a large
More informationExport and innovative performance of Italian firms
Export and innovative performance of Italian firms preliminary results work in progress Emanuela Marrocu e Stefano Usai University of Cagliari and CRENoS Raffaele Brancati, Manuel Romagnoli MET-Economia
More informationInterest rates and convergence across Italian regions
DIVISION OF ECONOMICS STIRLING MANAGEMENT SCHOOL Interest rates and convergence across Italian regions Sheila Dow Alberto Montagnoli Oreste Napolitano Stirling Economics Discussion Paper 2009-13 May 2009
More informationModeling GARCH processes in Panel Data: Theory, Simulations and Examples
Modeling GARCH processes in Panel Data: Theory, Simulations and Examples Rodolfo Cermeño División de Economía CIDE, México rodolfo.cermeno@cide.edu Kevin B. Grier Department of Economics Universy of Oklahoma,
More informationResponse Errors and Interviewer Characteristics: A Multidimensional Analysis
Proceedings of Q2008 European Conference on Quality in Survey Statistics Response Errors and Interviewer Characteristics: A Multidimensional Analysis Massimo Greco, Matteo Mazziotta, Adriano Pareto 1 Keywords:
More informationEvaluating total factor productivity differences by a mapping structure in growth models
Evaluating total factor productivity differences by a mapping structure in growth models Rosa Bernardini Papalia * and Silvia Bertarelli ** Abstract: The paper aims at providing a suitable measure of total
More informationGrowth and sectoral dynamics in the Italian regions
Growth and sectoral dynamics in the Italian regions Raffaele Paci and Francesco Pigliaru University of Cagliari and CRENoS Via S. Ignazio 78, 09123 Cagliari, Italy email: paci@unica.it pigliaru@unica.it
More informationSpecification testing in panel data models estimated by fixed effects with instrumental variables
Specification testing in panel data models estimated by fixed effects wh instrumental variables Carrie Falls Department of Economics Michigan State Universy Abstract I show that a handful of the regressions
More informationComparing Two Non-Compensatory Composite Indices to Measure Changes over Time: a Case Study
ANALYSES Comparing Two Non-Compensatory Composite Indices to Measure Changes over Time: a Case Study Matteo Mazziotta 1 Italian National Institute of Statistics, Rome, Italy Adriano Pareto 2 Italian National
More informationA note on the empirics of the neoclassical growth model
Manuscript A note on the empirics of the neoclassical growth model Giovanni Caggiano University of Glasgow Leone Leonida Queen Mary, University of London Abstract This paper shows that the widely used
More informationGeographic concentration in Portugal and regional specific factors
MPRA Munich Personal RePEc Archive Geographic concentration in Portugal and regional specific factors Vítor João Pereira Domingues Martinho Escola Superior Agrária, Instuto Polécnico de Viseu 0 Online
More informationSTOCKHOLM UNIVERSITY Department of Economics Course name: Empirical Methods Course code: EC40 Examiner: Per Pettersson-Lidbom Number of creds: 7,5 creds Date of exam: Thursday, January 15, 009 Examination
More information1 Motivation for Instrumental Variable (IV) Regression
ECON 370: IV & 2SLS 1 Instrumental Variables Estimation and Two Stage Least Squares Econometric Methods, ECON 370 Let s get back to the thiking in terms of cross sectional (or pooled cross sectional) data
More informationapplications to the cases of investment and inflation January, 2001 Abstract
Modeling GARCH processes in Panel Data: Monte Carlo simulations and applications to the cases of investment and inflation Rodolfo Cermeño División de Economía CIDE, México rodolfo.cermeno@cide.edu Kevin
More informationGeneral motivation behind the augmented Solow model
General motivation behind the augmented Solow model Empirical analysis suggests that the elasticity of output Y with respect to capital implied by the Solow model (α 0.3) is too low to reconcile the model
More informationESTIMATING FARM EFFICIENCY IN THE PRESENCE OF DOUBLE HETEROSCEDASTICITY USING PANEL DATA K. HADRI *
Journal of Applied Economics, Vol. VI, No. 2 (Nov 2003), 255-268 ESTIMATING FARM EFFICIENCY 255 ESTIMATING FARM EFFICIENCY IN THE PRESENCE OF DOUBLE HETEROSCEDASTICITY USING PANEL DATA K. HADRI * Universy
More informationEconometrics. Week 8. Fall Institute of Economic Studies Faculty of Social Sciences Charles University in Prague
Econometrics Week 8 Institute of Economic Studies Faculty of Social Sciences Charles University in Prague Fall 2012 1 / 25 Recommended Reading For the today Instrumental Variables Estimation and Two Stage
More informationAdvanced Economic Growth: Lecture 8, Technology Di usion, Trade and Interdependencies: Di usion of Technology
Advanced Economic Growth: Lecture 8, Technology Di usion, Trade and Interdependencies: Di usion of Technology Daron Acemoglu MIT October 3, 2007 Daron Acemoglu (MIT) Advanced Growth Lecture 8 October 3,
More informationMETHODOLOGY AND APPLICATIONS OF. Andrea Furková
METHODOLOGY AND APPLICATIONS OF STOCHASTIC FRONTIER ANALYSIS Andrea Furková STRUCTURE OF THE PRESENTATION Part 1 Theory: Illustration the basics of Stochastic Frontier Analysis (SFA) Concept of efficiency
More informationEconometrics of Panel Data
Econometrics of Panel Data Jakub Mućk Meeting # 1 Jakub Mućk Econometrics of Panel Data Meeting # 1 1 / 31 Outline 1 Course outline 2 Panel data Advantages of Panel Data Limitations of Panel Data 3 Pooled
More informationQuaderni di Dipartimento. On the potential pitfalls in estimating convergence by means of pooled and panel data
Quaderni di Dipartimento On the potential pitfalls in estimating convergence by means of pooled and panel data Carluccio Bianchi (Università di Pavia) Mario Menegatti (Università di Parma) # 37 (02-04)
More informationIntermediate Macroeconomics, EC2201. L2: Economic growth II
Intermediate Macroeconomics, EC2201 L2: Economic growth II Anna Seim Department of Economics, Stockholm University Spring 2017 1 / 64 Contents and literature The Solow model. Human capital. The Romer model.
More informationresearch paper series
research paper series Globalisation, Productivy and Technology Research Paper 2002/14 Foreign direct investment, spillovers and absorptive capacy: Evidence from quantile regressions By S. Girma and H.
More informationChapter 6 Analyzing Intra-Industry and Inter-Industry Technology Spillover of Foreign Direct Investment across Indian Manufacturing Industries
Chapter 6 Analyzing Intra-Industry and Inter-Industry Technology Spillover of Foreign Direct Investment across Indian Manufacturing Industries 161 6.1 Introduction Foreign direct investment (FDI) is believed
More informationApplied Econometrics (MSc.) Lecture 3 Instrumental Variables
Applied Econometrics (MSc.) Lecture 3 Instrumental Variables Estimation - Theory Department of Economics University of Gothenburg December 4, 2014 1/28 Why IV estimation? So far, in OLS, we assumed independence.
More informationMaking sense of Econometrics: Basics
Making sense of Econometrics: Basics Lecture 4: Qualitative influences and Heteroskedasticity Egypt Scholars Economic Society November 1, 2014 Assignment & feedback enter classroom at http://b.socrative.com/login/student/
More informationThe Process of Spatial Data Harmonization in Italy. Geom. Paola Ronzino
The Process of Spatial Data Harmonization in Italy Geom. Paola Ronzino ISSUES Geospatial Information in Europe: lack of data harmonization the lack of data duplication of data CHALLENGES Challenge of INSPIRE:
More informationTHE LONG-RUN DETERMINANTS OF MONEY DEMAND IN SLOVAKIA MARTIN LUKÁČIK - ADRIANA LUKÁČIKOVÁ - KAROL SZOMOLÁNYI
92 Multiple Criteria Decision Making XIII THE LONG-RUN DETERMINANTS OF MONEY DEMAND IN SLOVAKIA MARTIN LUKÁČIK - ADRIANA LUKÁČIKOVÁ - KAROL SZOMOLÁNYI Abstract: The paper verifies the long-run determinants
More informationApplied Econometrics. Lecture 3: Introduction to Linear Panel Data Models
Applied Econometrics Lecture 3: Introduction to Linear Panel Data Models Måns Söderbom 4 September 2009 Department of Economics, Universy of Gothenburg. Email: mans.soderbom@economics.gu.se. Web: www.economics.gu.se/soderbom,
More informationKnowledge assets and regional performance
Knowledge assets and regional performance Raffaele Paci and Emanuela Marrocu Universy of Cagliari, CRENoS Abstract Regional competiveness, especially in the industrialised countries, is increasingly reliant
More informationIdentifying the Monetary Policy Shock Christiano et al. (1999)
Identifying the Monetary Policy Shock Christiano et al. (1999) The question we are asking is: What are the consequences of a monetary policy shock a shock which is purely related to monetary conditions
More information4- Current Method of Explaining Business Cycles: DSGE Models. Basic Economic Models
4- Current Method of Explaining Business Cycles: DSGE Models Basic Economic Models In Economics, we use theoretical models to explain the economic processes in the real world. These models de ne a relation
More informationWooldridge, Introductory Econometrics, 3d ed. Chapter 16: Simultaneous equations models. An obvious reason for the endogeneity of explanatory
Wooldridge, Introductory Econometrics, 3d ed. Chapter 16: Simultaneous equations models An obvious reason for the endogeneity of explanatory variables in a regression model is simultaneity: that is, one
More informationEstimation of Panel Data Models with Binary Indicators when Treatment Effects are not Constant over Time. Audrey Laporte a,*, Frank Windmeijer b
Estimation of Panel ata Models wh Binary Indicators when Treatment Effects are not Constant over Time Audrey Laporte a,*, Frank Windmeijer b a epartment of Health Policy, Management and Evaluation, Universy
More informationCultural Globalization and Economic Growth
17 Cultural Globalization and Economic Growth Nuno Carlos Leão 1 This article investigates the relationship between cultural globalization and economic growth for the Portuguese experience for the period
More informationINTRODUCTION TO BASIC LINEAR REGRESSION MODEL
INTRODUCTION TO BASIC LINEAR REGRESSION MODEL 13 September 2011 Yogyakarta, Indonesia Cosimo Beverelli (World Trade Organization) 1 LINEAR REGRESSION MODEL In general, regression models estimate the effect
More informationChapter 15 Panel Data Models. Pooling Time-Series and Cross-Section Data
Chapter 5 Panel Data Models Pooling Time-Series and Cross-Section Data Sets of Regression Equations The topic can be introduced wh an example. A data set has 0 years of time series data (from 935 to 954)
More informationDEPARTMENT OF STATISTICS
Tests for causaly between integrated variables using asymptotic and bootstrap distributions R Scott Hacker and Abdulnasser Hatemi-J October 2003 2003:2 DEPARTMENT OF STATISTICS S-220 07 LUND SWEDEN Tests
More informationECON 402: Advanced Macroeconomics 1. Advanced Macroeconomics, ECON 402. New Growth Theories
ECON 402: Advanced Macroeconomics 1 Advanced Macroeconomics, ECON 402 New Growth Theories The conclusions derived from the growth theories we have considered thus far assumes that economic growth is tied
More informationBeyond the Target Customer: Social Effects of CRM Campaigns
Beyond the Target Customer: Social Effects of CRM Campaigns Eva Ascarza, Peter Ebbes, Oded Netzer, Matthew Danielson Link to article: http://journals.ama.org/doi/abs/10.1509/jmr.15.0442 WEB APPENDICES
More informationStructural VAR Models and Applications
Structural VAR Models and Applications Laurent Ferrara 1 1 University of Paris Nanterre M2 Oct. 2018 SVAR: Objectives Whereas the VAR model is able to capture efficiently the interactions between the different
More informationA Non-Parametric Approach of Heteroskedasticity Robust Estimation of Vector-Autoregressive (VAR) Models
Journal of Finance and Investment Analysis, vol.1, no.1, 2012, 55-67 ISSN: 2241-0988 (print version), 2241-0996 (online) International Scientific Press, 2012 A Non-Parametric Approach of Heteroskedasticity
More informationCO INTEGRATION: APPLICATION TO THE ROLE OF INFRASTRUCTURES ON ECONOMIC DEVELOPMENT IN NIGERIA
CO INTEGRATION: APPLICATION TO THE ROLE OF INFRASTRUCTURES ON ECONOMIC DEVELOPMENT IN NIGERIA Alabi Oluwapelumi Department of Statistics Federal University of Technology, Akure Olarinde O. Bolanle Department
More informationGrowth and Ergodicity: Has the World Converged?
Growth and Ergodicity: Has the World Converged? John S. Landon-Lane The University of New South Wales J. Arnold Quinn University of Minnesota January 7, 2000 Keywords: Growth, Convergence, Markov chain,
More informationSPATIAL HIERARCHICAL ANALYSIS OF ITALIAN REGIONS
The Regional Economics Applications Laboratory (REAL) of the University of Illinois focuses on the development and use of analytical models for urban and regional economic development. The purpose of the
More informationA Robust Approach to Estimating Production Functions: Replication of the ACF procedure
A Robust Approach to Estimating Production Functions: Replication of the ACF procedure Kyoo il Kim Michigan State University Yao Luo University of Toronto Yingjun Su IESR, Jinan University August 2018
More informationThe Productivity of Transport Infrastructure Investment: A Meta- Analysis of Empirical Evidence
The Productivy of Transport Infrastructure Investment: A Meta- Analysis of Empirical Evidence Patricia C. Melo, Daniel J. Graham, Ruben Brage-Ardao Centre for Transport Studies Department of Civil and
More informationMore on Roy Model of Self-Selection
V. J. Hotz Rev. May 26, 2007 More on Roy Model of Self-Selection Results drawn on Heckman and Sedlacek JPE, 1985 and Heckman and Honoré, Econometrica, 1986. Two-sector model in which: Agents are income
More informationOn Returns to Scale Assumption in Endogenous Growth
International Journal of Sciences: Basic and Applied Research (IJSBAR) ISSN 2307-453 (Print & Online) http://gssrr.org/index.php?journaljournalofbasicandapplied ---------------------------------------------------------------------------------------------------------------------------
More informationAugmented and unconstrained: revisiting the Regional Knowledge Production Function
Augmented and unconstrained: revisiting the Regional Knowledge Production Function Sylvie Charlot (GAEL INRA, Grenoble) Riccardo Crescenzi (SERC LSE, London) Antonio Musolesi (University of Ferrara & SEEDS
More informationDepartment of Economics, UCSB UC Santa Barbara
Department of Economics, UCSB UC Santa Barbara Title: Past trend versus future expectation: test of exchange rate volatility Author: Sengupta, Jati K., University of California, Santa Barbara Sfeir, Raymond,
More informationDeriving Some Estimators of Panel Data Regression Models with Individual Effects
Deriving Some Estimators of Panel Data Regression Models wh Individual Effects Megersa Tadesse Jirata 1, J. Cheruyot Chelule 2, R. O. Odhiambo 3 1 Pan African Universy Instute of Basic Sciences, Technology
More informationGrowth: Facts and Theories
Notes on Growth: Facts and Theories Intermediate Macroeconomics Spring 2006 Guido Menzio University of Pennsylvania Growth In the last part of the course we are going to study economic growth, i.e. the
More informationNew Notes on the Solow Growth Model
New Notes on the Solow Growth Model Roberto Chang September 2009 1 The Model The firstingredientofadynamicmodelisthedescriptionofthetimehorizon. In the original Solow model, time is continuous and the
More informationSixty years later, is Kuznets still right? Evidence from Sub-Saharan Africa
Quest Journals Journal of Research in Humanities and Social Science Volume 3 ~ Issue 6 (2015) pp:37-41 ISSN(Online) : 2321-9467 www.questjournals.org Research Paper Sixty years later, is Kuznets still
More informationRecent Advances in the Field of Trade Theory and Policy Analysis Using Micro-Level Data
Recent Advances in the Field of Trade Theory and Policy Analysis Using Micro-Level Data July 2012 Bangkok, Thailand Cosimo Beverelli (World Trade Organization) 1 Content a) Endogeneity b) Instrumental
More informationAppendix A: The time series behavior of employment growth
Unpublished appendices from The Relationship between Firm Size and Firm Growth in the U.S. Manufacturing Sector Bronwyn H. Hall Journal of Industrial Economics 35 (June 987): 583-606. Appendix A: The time
More informationThreshold effects in Okun s Law: a panel data analysis. Abstract
Threshold effects in Okun s Law: a panel data analysis Julien Fouquau ESC Rouen and LEO Abstract Our approach involves the use of switching regime models, to take account of the structural asymmetry and
More informationEconometrics in a nutshell: Variation and Identification Linear Regression Model in STATA. Research Methods. Carlos Noton.
1/17 Research Methods Carlos Noton Term 2-2012 Outline 2/17 1 Econometrics in a nutshell: Variation and Identification 2 Main Assumptions 3/17 Dependent variable or outcome Y is the result of two forces:
More informationGraduate Macro Theory II: Notes on Quantitative Analysis in DSGE Models
Graduate Macro Theory II: Notes on Quantitative Analysis in DSGE Models Eric Sims University of Notre Dame Spring 2011 This note describes very briefly how to conduct quantitative analysis on a linearized
More informationDealing With Endogeneity
Dealing With Endogeneity Junhui Qian December 22, 2014 Outline Introduction Instrumental Variable Instrumental Variable Estimation Two-Stage Least Square Estimation Panel Data Endogeneity in Econometrics
More informationHigh growth firms, firm dynamics and industrial variety - Regional evidence from Austria
High growth firms, firm dynamics and industrial variety - Regional evidence from Austria Seminar IEB Barcelona, November 2015 Klaus Friesenbichler & Werner Hölzl Overview of the presentation 1. Motivation
More informationRepeated observations on the same cross-section of individual units. Important advantages relative to pure cross-section data
Panel data Repeated observations on the same cross-section of individual units. Important advantages relative to pure cross-section data - possible to control for some unobserved heterogeneity - possible
More informationSTOCKHOLM UNIVERSITY Department of Economics Course name: Empirical Methods Course code: EC40 Examiner: Lena Nekby Number of credits: 7,5 credits Date of exam: Saturday, May 9, 008 Examination time: 3
More informationECNS 561 Multiple Regression Analysis
ECNS 561 Multiple Regression Analysis Model with Two Independent Variables Consider the following model Crime i = β 0 + β 1 Educ i + β 2 [what else would we like to control for?] + ε i Here, we are taking
More informationEconometrics (60 points) as the multivariate regression of Y on X 1 and X 2? [6 points]
Econometrics (60 points) Question 7: Short Answers (30 points) Answer parts 1-6 with a brief explanation. 1. Suppose the model of interest is Y i = 0 + 1 X 1i + 2 X 2i + u i, where E(u X)=0 and E(u 2 X)=
More informationClub Convergence: Some Empirical Issues
Club Convergence: Some Empirical Issues Carl-Johan Dalgaard Institute of Economics University of Copenhagen Abstract This note discusses issues related to testing for club-convergence. Specifically some
More informationApplied Microeconometrics (L5): Panel Data-Basics
Applied Microeconometrics (L5): Panel Data-Basics Nicholas Giannakopoulos University of Patras Department of Economics ngias@upatras.gr November 10, 2015 Nicholas Giannakopoulos (UPatras) MSc Applied Economics
More informationCREN S. Centro Ricerche Economiche Nord Sud Università degli Studi di Cagliari EXTERNALITIES AND LOCAL ECONOMIC GROWTH IN MANUFACTURING INDUSTRIES
CREN S Centro Ricerche Economiche Nord Sud Università degli Studi di Cagliari EXTERNALITIES AND LOCAL ECONOMIC GROWTH IN MANUFACTURING INDUSTRIES Stefano Usai Raffaele Paci CONTRIBUTI DI RICERCA 01/13
More informationFirst revision: July 2, 2010 Second revision: November 25, 2010 Third revision: February 11, Abstract
The Economic Impact of Capal-Skill Complementaries on Sectoral Productivy Growth New Evidence from Industrialized Industries during the New Economy Dr. Thomas Strobel Ifo Instute for Economic Research
More informationEconomics 582 Random Effects Estimation
Economics 582 Random Effects Estimation Eric Zivot May 29, 2013 Random Effects Model Hence, the model can be re-written as = x 0 β + + [x ] = 0 (no endogeneity) [ x ] = = + x 0 β + + [x ] = 0 [ x ] = 0
More informationThe Convergence Analysis of the Output per Effective Worker and Effects of FDI Capital Intensity on OECD 10 Countries and China
Middle Eastern Finance and Economics ISSN: 1450-2889 Issue 8 (2010) EuroJournals Publishing, Inc. 2010 http://www.eurojournals.com/mefe.htm The Convergence Analysis of the Output per Effective Worker and
More informationPublic Infrastructure and Economic Growth in Mexico
Public Infrastructure and Economic Growth in Mexico Antonio Noriega Matias Fontenla Universidad de Guanajuato and CIDE April 15, 2005 Abstract We develop a model where investment in infrastructure complements
More informationThe Ramsey Model. (Lecture Note, Advanced Macroeconomics, Thomas Steger, SS 2013)
The Ramsey Model (Lecture Note, Advanced Macroeconomics, Thomas Steger, SS 213) 1 Introduction The Ramsey model (or neoclassical growth model) is one of the prototype models in dynamic macroeconomics.
More informationThe Contribution Rate of Thrice Industrial Agglomeration to Industrial Growth in Ningxia The Calculate Based on Cobb-Douglas Function.
International Conference on Economics, Social Science, Arts, Education and Management Engineering (ESSAEME 2015) The Contribution Rate of Thrice Industrial Agglomeration to Industrial Growth in Ningxia
More informationThe Augmented Solow Model Revisited
The Augmented Solow Model Revisited Carl-Johan Dalgaard Institute of Economics University of Copenhagen February, 2005 Abstract This note briefly discusses some recent (re-)investigations of the (augmented)
More informationAssumption 5. The technology is represented by a production function, F : R 3 + R +, F (K t, N t, A t )
6. Economic growth Let us recall the main facts on growth examined in the first chapter and add some additional ones. (1) Real output (per-worker) roughly grows at a constant rate (i.e. labor productivity
More informationLinear Models in Econometrics
Linear Models in Econometrics Nicky Grant At the most fundamental level econometrics is the development of statistical techniques suited primarily to answering economic questions and testing economic theories.
More informationWill it float? The New Keynesian Phillips curve tested on OECD panel data
Phillips curve Roger Bjørnstad 1 2 1 Research Department Statistics Norway 2 Department of Economics University of Oslo 31 October 2006 Outline Outline Outline Outline Outline The debatable The hybrid
More informationSmooth transition pollution-income paths
Smooth transion pollution-income paths Nektarios Aslanidis and Anastasios Xepapadeas Universy of Crete Department of Economics Universy Campus 74 00, Rethymno Greece n.aslanidis@econ.soc.uoc.gr, xepapad@econ.soc.uoc.gr
More informationTHRESHOLD EFFECTS IN THE OPENNESS-PRODUCTIVITY GROWTH RELATIONSHIP: THE ROLE OF INSTITUTIONS AND NATURAL BARRIERS
THRESHOLD EFFECTS IN THE OPENNESS-PRODUCTIVITY GROWTH RELATIONSHIP: THE ROLE OF INSTITUTIONS AND NATURAL BARRIERS By SOURAFEL GIRMA, MICHAEL HENRY AND CHRIS MILNER (GEP and School of Economics, Universy
More informationEquating output per worker to GDP per capita, the growth rate of GDP per capita
3 Homework 3 1. We have seen in class Kaldor s stylized facts of growth in developed countries. The Cobb-Douglas production function is used to replicate fact a. In this exercise, you are asked to show
More informationBASELINE STUDY ON THE PREVALENCE OF SALMONELLA IN LAYING FLOCKS OF GALLUS gallus IN ITALY
BASELINE STUDY ON THE PREVALENCE OF SALMONELLA IN LAYING FLOCKS OF GALLUS gallus IN ITALY FINAL REPORT According to Annex I of the technical specifications (SANCO/34/2004), in Italy 431 holdings of laying
More informationBayesian Econometrics - Computer section
Bayesian Econometrics - Computer section Leandro Magnusson Department of Economics Brown University Leandro Magnusson@brown.edu http://www.econ.brown.edu/students/leandro Magnusson/ April 26, 2006 Preliminary
More informationEconometrics of Panel Data
Econometrics of Panel Data Jakub Mućk Meeting # 6 Jakub Mućk Econometrics of Panel Data Meeting # 6 1 / 36 Outline 1 The First-Difference (FD) estimator 2 Dynamic panel data models 3 The Anderson and Hsiao
More informationEconometric Analysis of Cross Section and Panel Data
Econometric Analysis of Cross Section and Panel Data Jeffrey M. Wooldridge / The MIT Press Cambridge, Massachusetts London, England Contents Preface Acknowledgments xvii xxiii I INTRODUCTION AND BACKGROUND
More informationEstimates of the Sticky-Information Phillips Curve for the USA with the General to Specific Method
MPRA Munich Personal RePEc Archive Estimates of the Sticky-Information Phillips Curve for the USA with the General to Specific Method Antonio Paradiso and B. Bhaskara Rao and Marco Ventura 12. February
More informationWooldridge, Introductory Econometrics, 4th ed. Chapter 15: Instrumental variables and two stage least squares
Wooldridge, Introductory Econometrics, 4th ed. Chapter 15: Instrumental variables and two stage least squares Many economic models involve endogeneity: that is, a theoretical relationship does not fit
More informationRecent Advances in the Field of Trade Theory and Policy Analysis Using Micro-Level Data
Recent Advances in the Field of Trade Theory and Policy Analysis Using Micro-Level Data July 2012 Bangkok, Thailand Cosimo Beverelli (World Trade Organization) 1 Content a) Classical regression model b)
More informationEconometrics Summary Algebraic and Statistical Preliminaries
Econometrics Summary Algebraic and Statistical Preliminaries Elasticity: The point elasticity of Y with respect to L is given by α = ( Y/ L)/(Y/L). The arc elasticity is given by ( Y/ L)/(Y/L), when L
More informationEstimating efficiency spillovers with state level evidence for manufacturing in the US
Loughborough Universy Instutional Reposory Estimating efficiency spillovers wh state level evidence for manufacturing in the US This em was submted to Loughborough Universy's Instutional Reposory by the/an
More informationBusiness Cycles: The Classical Approach
San Francisco State University ECON 302 Business Cycles: The Classical Approach Introduction Michael Bar Recall from the introduction that the output per capita in the U.S. is groing steady, but there
More informationPanel Data. March 2, () Applied Economoetrics: Topic 6 March 2, / 43
Panel Data March 2, 212 () Applied Economoetrics: Topic March 2, 212 1 / 43 Overview Many economic applications involve panel data. Panel data has both cross-sectional and time series aspects. Regression
More informationresearch paper series
research paper series Globalisation, Productivy and Technology Research Paper 2003/32 Threshold and Interaction Effects in the Openness-Productivy Growth Relationship: The Role of Instutions and Natural
More information