The Original Home-Field Evidence
The Italian evidence gives the local bias of entrepreneurship its most useful empirical shape. The question is not only whether founders are more likely to be local than wage workers, but whether local start-ups receive different opportunities once they exist. Reading the study carefully lets us connect founder location to firm scale and finance while keeping its descriptive design—and its limits for explaining the later US decline—in view.
The gap is about new business creation, not just family succession
Claudio Michelacci and Olmo Silva studied Italy because it offered a useful combination of common national rules and large regional differences in income, technology, and financial development. Their data came from the Bank of Italy's Survey of Household Income and Wealth for 1991, 1993, and 1995. The sample retained working heads of household between 18 and 65. That is a narrower population than the US census-based analysis, but it contains unusually detailed information about occupational choice, business size, capital, debt, and birthplace.
The study defines a person as local when they work in the province where they were born. Italy had 103 provinces at the time, roughly comparable to US counties. Because the survey did not record work location separately from residence, the researchers assume that people work where they live and check the analysis using regions instead of provinces. That assumption is reasonable enough for their robustness exercise, but it matters: the Italian “local” measure is not identical to living in one's US birth state.
Entrepreneurship is also defined broadly. The category includes entrepreneurs, self-employed craftspeople, professionals, and managers or partners who actively work in their firms. Family-firm workers are excluded because holding a job in a family business does not necessarily involve leadership or innovation. To focus on new creation, the study uses a start-up definition that attempts to exclude people who simply took over a parent's activity. The classifications are imperfect, yet they allow the central question to be asked: is the local bias present among people who create businesses, rather than only among people who inherit them?
The answer is yes. In the 1991–1993 specifications, about 79 percent of entrepreneurs were local, compared with about 72 percent of dependent workers. The gap—roughly 7 percentage points—remained close to 6 or 7 points after controls for age, gender, education, marital status, children, region, and survey year. When the researchers included the 1995 wave, the estimate remained of a similar size. It also persisted when they focused on start-ups.
That persistence changes the interpretation. Family transmission may still influence who has an idea, resources, or confidence, and the survey cannot eliminate every form of family advantage. But the local bias is not simply an artifact of people inheriting a business that was already located at home. New business formation itself is more locally concentrated than dependent employment in the Italian sample.
The comparison with workers is the study's key discipline. A local share of 79 percent sounds large until we ask what share of workers is local. The LBE is the difference between those shares, not the entire rate at which entrepreneurs remain in their birthplace. It measures an entrepreneurial excess of locality after the common tendency of workers and entrepreneurs to remain near their origins is taken into account.
The local gap grows with the business behind it
The most revealing Italian result is that LBE is not uniform across firms. It becomes larger as business size increases. When the researchers compare start-ups by employment, the local bias is about 10 to 11 percentage points for larger businesses. When they split firms by market value, the bias is also higher among the most valuable ventures, reaching roughly 10 points for the top decile in their sample.
A second comparison looks directly at local and non-local start-ups rather than at the probability of being local. After the study's controls, local start-ups have an average employment advantage of about 37 percent. They have about 40 percent more capital and a capital-labor ratio roughly 38 percent higher than non-local firms. These are large descriptive differences. They suggest that local embeddedness is associated not merely with a greater chance of opening a shop or practice, but with the resources and production choices of the business that gets opened.
The financing result points in the same direction. Local start-ups obtain about 45 percent more bank financing per unit of invested capital than non-local start-ups. The measure is based on business bank debt relative to capital, and the regressions account for personal and household characteristics, city size, education, occupation, region, and other controls. The result is consistent with the mechanism described earlier: a local founder can use relationships and reputation to obtain more external funding for a given capital base.
It is not proof that locality causes the larger firm. Local entrepreneurs may differ in wealth, experience, sector, family resources, or unmeasured knowledge. The firm measures are partly self-reported; the authors specifically caution that business value may be understated because respondents may fear information reaching tax authorities. Employment is less vulnerable to that particular concern, but it too is measured in a household survey rather than a complete administrative census of firms. The evidence is strong enough to establish a pattern and motivate a mechanism, not strong enough to turn the pattern into a personal ranking of local and non-local founders.
The relationship between locality and finance becomes clearer when the study compares regions with different access to credit. The researchers construct a capital-availability index from the conditional probability that potential borrowers are rejected or discouraged from applying for a loan. They estimate it at both the province and region level. In their province-level models, once local credit conditions are included, the raw local-entrepreneur gap is no longer statistically significant, but the interaction between entrepreneurial status and capital availability remains positive and significant.
The implied contrasts are substantial, though they must be read as model-based comparisons. Moving from the least developed provincial credit market to the sample's average capital-availability value would move the predicted LBE from zero to about 11 percentage points; the most financially developed province would imply about 14 points. At the regional level, the corresponding predicted range is about zero to 7 points. The study also instruments regional credit availability with banking structure shaped by Italy's 1936 banking regulation. The instrumental-variable estimates imply a range from zero to about 11 points between the least and most developed markets.
That historical instrument is an attempt to separate credit supply from the possibility that prosperous, entrepreneurial regions simply generate more demand for loans. It is a stronger design than comparing raw regional averages, but it still depends on the assumption that banking structure inherited from the 1936 rules affects present-day LBE primarily through financial development. The researchers also correct for measurement error in the credit index and still obtain an estimated LBE of about 6.5 points in the most dynamic credit markets. These results strengthen the finance interpretation without making the identification assumptions disappear.
This map should be read as an evidence structure, not as a diagram of proven causation. Founder locality is associated with start-up location and firm outcomes; regional credit availability is associated with the local advantage; and dependent workers provide the benchmark that makes LBE a relative measure. The reason the pattern is persuasive is that the relationships line up: local status is linked to entry, financing, capital intensity, and scale. The reason it remains qualified is that the same regional prosperity may influence several of those relationships at once.
The study finds a similar interaction when it asks who becomes an entrepreneur. Regional financial development is positively related to the probability of starting a business when entered alone. Once the model includes the interaction between credit availability and local status, the standalone credit coefficient loses significance while the local interaction remains positive. For an average individual, moving from the least to the most developed local financial market raises the predicted probability of starting a business by about 35 percent. The interaction also appears in firm employment, capital stock, and capital intensity: better local finance favors local entrepreneurs in particular.
Richer regions, weaker jobs, and the learning alternative
The Italian regional pattern helps distinguish finance from a more familiar explanation. Perhaps people become local entrepreneurs because they want to remain near home but cannot find good dependent work. In that case, LBE should be strongest where unemployment is high and wage opportunities are poor. The study finds the opposite direction. LBE is higher in regions with lower unemployment and higher GDP per capita, especially in the more developed north. The authors report that a five-percentage-point increase in unemployment is associated with an approximately 1.5 percent decrease in LBE for start-ups.
This does not mean that unemployment never pushes anyone into self-employment. It means that poor local jobs do not fit the overall regional pattern as the main explanation. The same conclusion follows from the firm outcomes. If local entrepreneurship were chiefly a last resort for attached workers, we would not necessarily expect local start-ups to be larger, more valuable, and more capital-intensive than the firms created by people who moved. The evidence instead points toward a place that gives local founders productive resources, not only a place they are unwilling to leave.
The other alternative is learning. A specialized industrial district can teach local people how a sector works. They may observe suppliers, customers, production methods, and business opportunities that outsiders do not see. If this mechanism explains LBE, the bias should be stronger where entrepreneurial activity is dense and where local firms are concentrated in a few sectors.
The study finds limited support for that prediction. Entrepreneurial density has a positive relationship with LBE in some specifications, but the effect is modest and becomes statistically non-robust after controls. Sectoral specialization has no significant role in accounting for the local bias. That is evidence against learning and spillovers as the complete explanation, not evidence that local knowledge is irrelevant. Learning may matter for particular industries or people even if it does not explain the broad cross-region pattern.
The evidence therefore favors a layered interpretation. Local finance and social capital appear to account for more of the observed combination—local entry, larger firms, higher capital intensity, and greater leverage—than either weak local employment or entrepreneurial clustering alone. But selection remains possible, and the mechanisms can coexist. A local person may know the industry, prefer to stay, and receive better credit. The study's results narrow the explanation; they do not reduce it to a single observable variable.
The contrast with wage workers makes that point sharper. In the Italian sample, local dependent workers earn less than migrants, including a reported wage disadvantage for local white-collar workers of about 5 percent after controls. At the same time, non-local entrepreneurs create smaller and less valuable businesses than local entrepreneurs. Moving appears to carry a positive return in wage work and a possible financing cost in entrepreneurship. The comparison is not a contradiction. Workers enter organizations that already possess capital and market relationships; founders must assemble those resources themselves.
The Italian study thus establishes the original home-field pattern and gives it an economic mechanism: local ties can improve access to credit, and credit can change the scale of the business. It also tells us what not to claim. The findings come from an older Italian household survey, use a province-level locality assumption, and compare a broad occupational category with wage workers. They are not direct evidence that American lending became more portable or that the US decline after 1970 was caused by the same financial gradient.
What they provide is a testable expectation for the US case. If the hometown advantage is strongest when self-employment needs substantial capital and lenders rely on local information, then the advantage should weaken when fewer self-employed people operate capital-intensive businesses—or when finance becomes less tied to the place where the founder is known. The next step is to assemble these observations into a working model before asking whether the American business mix moved in that direction.