Where firms diversify may matter more than how many products they export

Adding products to an export catalogue does not necessarily signal stronger growth. Research on Italian firms suggests that the distinction between familiar product areas and activities outside the core can reveal patterns that a simple product count misses.

Reading growth signals in export baskets

A preprint by Maddalena Mula, Valerio De Stefano, Manuel Sebastian Mariani and Andrea Zaccaria asks whether tools used to study national economies can help explain corporate performance. Its central finding is that both the income profile of a firm’s exported products and their position relative to its core export area are associated with subsequent growth.

The researchers combine export records and financial information supplied by Italy’s national statistics institute, ISTAT. Their study covers more than 57,000 firms that exported every year from 2015 through 2021, with export records distinguishing 5,389 product categories. The sample therefore focuses on persistent exporters, rather than all Italian businesses.

For the main comparison across firms, the authors average export characteristics and other explanatory variables over 2015–2017 and examine performance in 2019–2021. Growth is measured through changes in three-year averages of log value added. Labour productivity measures value added per employee. The regressions account for initial productivity, employment, firm age, sector and region.

Products associated with wealthier exporters

One indicator, called EXPY, adapts an approach originally developed for countries. Each product receives a score based on the income per person of the countries exporting it, weighted by their revealed comparative advantage—a measure of relative export specialization. The authors use logarithms of national income and standardize the resulting product scores.

A firm’s EXPY is then the average score of its exported products, weighted by export volume. This concerns the countries that export those goods, not the countries buying the Italian firm’s exports.

Higher EXPY is positively and statistically significantly associated with both subsequent value-added growth and future labour productivity in the main cross-sectional models. The pattern also remains when the authors substitute a product score calculated without taking logarithms of national income.

EXPY should not be read as a direct measure of technological difficulty. It captures an income association that the authors interpret as a proxy for product productivity and market value. A product can score well without requiring exceptionally rare capabilities. Indeed, two alternative measures of product complexity have opposite associations with firm performance in an additional analysis, underscoring that these indicators are not interchangeable.

Inside and outside the core

The second contribution separates diversification into products inside and outside a firm’s core export cluster. The authors construct a network linking firms to products and apply the BRIM community-detection algorithm to identify groups with concentrated export connections.

The procedure identifies eight blocks, including clusters associated with machinery and metalworking, apparel and textiles, and automotive and transport products. A firm’s in-block diversification counts the products it significantly exports within its assigned block. Out-of-block diversification counts those belonging to other blocks. Significant exports are identified using a revealed-comparative-advantage threshold.

Across firms, the two counts have contrasting associations with performance. Greater in-block diversification is associated with lower subsequent growth and labour productivity, while greater out-of-block diversification is associated with higher outcomes, holding the other included variables constant.

How the clusters are defined matters. When the researchers replace network-derived blocks with the Harmonized System’s standard product sections, diversification outside the core section is no longer statistically significantly associated with either outcome. A separate measure of coherence—the similarity among products in a firm’s basket—also lacks a statistically significant association in the main models.

A signal, not a demonstrated growth strategy

The distinction between comparing different firms and following the same firm over time is crucial. In additional panel regressions with firm and year fixed effects, EXPY is positively associated with growth at two- and three-year horizons, but not statistically significantly at one year in the untreated data. In-block diversification remains negatively associated with growth at every examined horizon.

Out-of-block diversification is less robust within firms. Its association is not statistically significant at any horizon in the untreated panel. A positive three-year association appears after the authors cap extreme growth observations at the first and ninety-ninth percentiles.

The authors therefore describe outside-core diversification as a possible indicator of growth prospects, not necessarily a driver. Their panel models explain less than 1% of within-firm growth variation. Establishing causality remains an open question: the findings do not demonstrate that adding products outside the core will make a particular company grow faster.