📈 economics

Cooperative Credit Banking and Local Economic Activity in Italy: A Finite Mixture Modelling Approach to Latent Spatial Regimes

Using a finite-mixture modelling approach on Italian provincial data from 2002 to 2022, this paper reveals that the positive impact of cooperative credit banking on local economic activity is confined to specific latent territorial regimes that transcend conventional macro-regional boundaries, highlighting the context-dependent nature of locally embedded financial intermediation.

Original authors: Giuseppe Terzo

Published 2026-07-14

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Giuseppe Terzo

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine Italy as a giant, bustling pizza shop with 103 different slices (provinces), each with its own unique flavor. For a long time, economists have been trying to figure out if the "cooperative credit banks" (let's call them the Community Chefs) are the secret ingredient that makes these slices taste better and the local economy grow.

Most people assumed the answer was the same everywhere: "Yes, Community Chefs help!" or "No, they don't." They tried to group the pizza slices by big regions, like "The Northern Slice" or "The Southern Slice," thinking the rules were the same for everyone in that big area.

But this paper, written by Giuseppe Terzo, says, "Wait a minute! Let's look closer." Instead of guessing the groups beforehand, the author used a special mathematical magnifying glass called a Finite Mixture Model. Think of this like a magic sorting machine that doesn't care about the map's borders; it just looks at the data and says, "Hey, these specific slices actually belong together because they react the same way, even if they are miles apart."

The Big Reveal: It's Not One Size Fits All

When the author looked at the data from 2002 to 2022, the "magic sorting machine" found three hidden groups of provinces, not just the usual North vs. South split.

The "Other Chef" Group (39% of provinces): In these places, the Community Chefs didn't seem to make a huge difference to the local economy. Here, the economy seemed to rely more on the big, standard banks and other factors like how open the area was to trade.
The "Teamwork" Group (41% of provinces): In this middle group, both the Community Chefs and the standard banks were helping the local economy grow. It was a balanced team effort.
The "Super Chef" Group (20% of provinces): This is the most exciting find! In this smaller, special group, the Community Chefs were superstars. Their presence was strongly linked to a booming local economy. In fact, in these specific spots, the standard banks didn't seem to matter much at all; the Community Chefs were doing the heavy lifting.

Why This Changes the Map

The paper argues that if you just look at the "average" effect (like taking a bite of the whole pizza and guessing the flavor), you miss the magic. The average suggests Community Chefs are helpful, but the Finite Mixture Model shows that this help is context-dependent.

The author explicitly rules out the idea that these differences are just about the old-fashioned "North vs. South" divide. The map of these three groups cuts right across the traditional lines. A province in the North might belong to the "Super Chef" group, while a neighbor in the South might be in the "Other Chef" group. The rules aren't about geography; they are about the hidden, local configuration of how banking and business work together in that specific spot.

How Sure Are We?

The author didn't just guess; they ran the numbers through several different tests to make sure the results weren't a fluke.

They checked the data using standard math (OLS) and found a positive link.
They used "time machines" (historical data from 1936) to make sure the banks weren't just growing because the economy was already doing well (this is called using an instrument).
They even ran a dynamic model that accounts for how economies change over time.

All these different methods pointed in the same direction: the link exists, but it's not uniform. The "Super Chef" effect in that 20% of provinces is statistically significant, meaning it's a real pattern, not just random noise. The paper suggests that the role of these local banks depends entirely on which of these three hidden "territorial regimes" you are looking at.

So, the next time someone says, "Local banks help the economy everywhere," you can smile and say, "Well, actually, they are the secret sauce in about one-fifth of the places, a team player in another 40%, and just another ingredient in the rest." The map is more complex, and much more interesting, than we thought.

Technical Summary: Cooperative Credit Banking and Local Economic Activity in Italy

Problem Statement
Cooperative Credit Banks (CCBs) in Italy are widely recognized as locally embedded financial institutions that may foster local economic development through relationship lending and the accumulation of soft information. However, existing empirical literature has largely addressed the spatial heterogeneity of the relationship between CCBs and local economic activity by imposing ex ante geographical partitions, such as the North–South divide or NUTS-1 macro-regions. This approach assumes that the role of CCBs is uniform within these administrative boundaries, potentially overlooking latent heterogeneity within conventional geographical areas. Consequently, it remains unclear whether the efficacy of CCBs varies strictly across administrative lines or follows unobserved ("latent") territorial configurations.

Methodology
The study utilizes a panel dataset of 103 Italian provinces (NUTS-3) spanning the period 2002–2022. The dependent variable is the natural logarithm of per capita value added, while the key independent variable is the density of CCB branches per 1,000 inhabitants.

The empirical strategy proceeds in two stages:

Benchmark Estimation: The author first estimates the average association using Ordinary Least Squares (OLS), Instrumental Variable (IV) 2SLS, and System-GMM models.
- Endogeneity Handling: To address potential endogeneity, the study employs two types of instruments:
  - An external historical instrument: The provincial density of rural and artisan banks in 1936 (predetermined prior to the 1936 banking reform).
  - Lewbel-generated instruments: Constructed using heteroscedasticity in the first-stage residuals.
- Controls: The models include lagged controls for population density, life expectancy, sectoral employment shares, dependency ratios, trade openness, and the density of non-CCB bank branches.
Finite Mixture Modelling (FMM): To move beyond average effects and identify latent spatial regimes, the paper employs a Finite Mixture Model. This approach probabilistically assigns provinces to unobserved classes ( $K$ ) based on the data rather than pre-defined geography. The model allows for class-specific intercepts ( $\alpha_k$ ), slopes ( $\beta_k$ ), and control coefficients ( $\gamma_k$ ). The optimal number of classes is determined by comparing information criteria (AIC, BIC), convergence stability, and substantive interpretability.

Key Results

Average Association: Benchmark estimates (OLS, IV, and Sys-GMM) consistently show a positive average association between CCB branch density and local economic activity. IV and GMM estimates suggest a higher coefficient than OLS, indicating potential attenuation bias in the latter.
Latent Heterogeneity: The FMM analysis reveals that the average positive association masks substantial latent heterogeneity. A three-class model is identified as the preferred specification:
- Class 1 (~39% of sample): CCB density is not statistically significant. Economic activity here is driven by non-CCB banks, secondary employment, and trade openness.
- Class 2 (~41% of sample): CCB density shows a positive and significant, but moderate, association. Both cooperative and non-cooperative banks contribute to economic activity.
- Class 3 (~20% of sample): A smaller, distinct regime where CCB density exhibits a substantially larger and highly significant coefficient. In this regime, non-CCB bank branches are statistically insignificant, suggesting that economic activity is driven specifically by the cooperative component of the banking system.
Spatial Distribution: The spatial mapping of these regimes demonstrates that they do not align with traditional macro-regional divisions (e.g., North vs. South). Instead, the regimes cut across administrative boundaries, exhibiting local spatial clustering that is not captured by standard geographical partitions.

Key Contributions
The paper's primary contribution is demonstrating that the estimated role of cooperative credit banking is contingent upon how spatial heterogeneity is modeled. By shifting from pre-defined geographical partitions to latent territorial regimes, the study reveals that the "local embeddedness" of CCBs is a context-dependent form of financial intermediation. The findings suggest that the North–South divide is an insufficient proxy for the variation in CCB effectiveness; rather, specific territorial configurations exist where CCBs are the dominant driver of local economic activity, distinct from areas where conventional banking structures prevail.

Significance and Claims
The paper claims to provide evidence that locally embedded banking relevance varies across latent territorial configurations. It argues that policy evaluations of CCBs should not rely on a uniform territorial lens, as their impact is highly dependent on the specific configuration of local banking systems and productive structures. The study concludes that broader conclusions about locally embedded financial institutions may change significantly when territorial heterogeneity is explicitly modeled through latent regimes rather than imposed through fixed administrative boundaries.

The Big Reveal: It's Not One Size Fits All

Why This Changes the Map

How Sure Are We?

Technical Summary: Cooperative Credit Banking and Local Economic Activity in Italy

More like this