Accumulation of cultural capital at the intersection of socio-demographic features and productive specializations

Abstract

This paper aims at introducing a conceptual framework and assessing how features of local systems combine with high levels of cultural capital. This framework encompasses the local productive specializations and socio-demographic characteristics, as well as their interplay. A review on related concepts and contributions helps to generate three hypotheses on place-based cultural capital. The paper works under the three hypotheses and applies the framework to an original dataset based on the Italian local systems. The results show how urban areas and made-in Italy local productive systems tend to associate with high levels of cultural capital. Moreover, the interplay between local productive specializations and socio-demographic characteristics highlights the role of place specificities. Such relations should be considered in the elaboration of culture-based policies of territorial development, and in further researches over the accumulative forces of cultural capital.

Appendices

Appendix 1

See Figs. 4 and 5.

Fig. 4

Source: Authors’ elaboration on ISTAT dataset

Classification and spatial distribution of Made-in Italy LMAs by Prevailing Productive Specialization (PPS).

Fig. 5

Source: Authors’ elaboration on ISTAT dataset

Classification and spatial distribution of LMAs by socio-demographic characteristics (SDEM).

Appendix 2

The observed variable for LMAs cultural vocation level $\varvec{CV}_{i}$ is related to the corresponding latent variable $\varvec{CV}_{i}^{\varvec{*}}$ as:

$$\varvec{CV}_{i} = \left\{ {\begin{array}{*{20}l} {1\; = \;{\text{Peripheral}}\,{\text{Areas}}} \hfill & {{\text{if}}\;\varvec{CV}_{i}^{\varvec{*}} \le \mu_{1} ,} \hfill \\ {2\; = \;{\text{Tourism}}\,{\text{Driver}}} \hfill & {{\text{if}}\;\mu_{2} \le \varvec{CV}_{i}^{\varvec{*}} \le \mu_{1} ,} \hfill \\ {3\; = \;{\text{Cultural}}\,{\text{Entrepreneurship}}} \hfill & {{\text{if}}\;\mu_{3} \le \varvec{CV}_{i}^{\varvec{*}} \le \mu_{2} ,} \hfill \\ {4\; = \;{\text{Cultural}}\,{\text{Heritage}}\;{\text{Strength}}} \hfill & {{\text{if}}\;\mu_{4} \le \varvec{CV}_{i}^{\varvec{*}} \le \mu_{3} ,} \hfill \\ {5\; = \;{\text{Great}}\;{\text{Beauty}}} \hfill & {{\text{if}}\,\varvec{CV}_{i}^{\varvec{*}} \ge \mu_{4} .} \hfill \\ \end{array} } \right.$$

(2)

$\varvec{CV}_{i}$ is treated as a categorical variable, with $j$ numerical values assigned to each outcome as in (2). In our ordered probit model, the categorical variable $\varvec{CV}_{i}$ is estimated as a linear function of the control variables plus a vector of cut points. Because there are $Z = 5$ alternatives, we calculate $Z - 1 = 4$ cut points ($\mu_{1}$ to $\mu_{4}$, with: $- \;\infty = \mu_{1} < \cdots < \mu_{4} = + \;\infty$).

The underlying model consists of an equation relating the latent cultural vocation ($\varvec{CV}_{i}^{\varvec{*}}$) to LMA categories (LMA), their socio-demographic (SDEM) and background characteristics of the LMA, the latter represented by the vector ${\mathbf{X}}_{i}$.

$$ \varvec{CV}_{\varvec{i}}^{\varvec{*}} = \varvec{ }\beta_{0} + \beta_{1} {\text{LMA}} + \beta_{2} {\text{SDEM}} + \beta_{3} {\mathbf{X}}_{i} + \eta_{i} + \varepsilon_{j} $$

where $\beta_{1} , \beta_{2}$ and $\beta_{3}$ are the unknown parameters estimated using the maximum likelihood technique, $\eta_{i}$ are the regional fixed effects, $\varepsilon_{j}$ is the normally distributed error term. The model does not contain a constant term as it would be exactly collinear with the cut points. We can represent the probabilities of these outcomes if we assume a particular probability distribution. The probability of observing outcome $i$ (Eq. 2) is equal to the probability that the estimated linear function, plus random error, is within the range of the cut points estimated for the outcome:

$$ {\mathbf{Pr}}\left( {\varvec{CV}_{j} = j} \right) = {\mathbf{Pr}}(\mu_{j - 1} < \varvec{CV}_{i}^{*} \le \mu_{ij} ). $$

(3)

Assuming the errors being distributed like a standard normal distribution $N\left( {0,1} \right),$ the ordered probit model leads to the following probability function (4), that holds for $\varvec{CV}_{j} \left[ {2, 4} \right]$

$$ {\mathbf{Pr}}\left( {\varvec{CV}_{j} = j} \right) =\upphi\left( {\mu_{j} - {\mathbf{X}}_{i} \beta_{i} } \right) -\upphi\left( {\mu_{j - 1} - {\mathbf{X}}_{i} \beta_{i} } \right), $$

(4)

with $\upphi\left( . \right)$ representing the cumulative distribution function of a standard normal distribution. Equations (5) and (6) yield, respectively, the probabilities for $j$ = 5 (Great Beauty) or $j$ = 1 (Peripheral Areas):

$$ {\mathbf{Pr}}\left( {\varvec{CV}_{j} = 5} \right) = 1 -\upphi\left( {\mu_{4} - {\mathbf{X}}_{i} \beta_{i} } \right). $$

(5)

$$ {\mathbf{Pr}}\left( {\varvec{CV}_{j} = 1} \right) =\upphi\left( {\mu_{1} - {\mathbf{X}}_{i} \beta_{i} } \right), $$

(6)

Appendix 3

See Tables 4, 5 and 6.

Table 4 Ordered probit full estimates.

Table 5 Link test.

Table 6 Robustness checks to CV scale variations.