Quality of Life in Italy: Finding the Ideal Town

The Search for the Ideal Town to Live in Italy

I am on a quest to find the ideal town to call home in Italy. The concept of an “ideal town” is subjective, but the journey allows me to learn about the socioeconomic and geographical factors of my new country. With countless towns to consider, I needed to narrow down the list of options. While an objective ranking cannot provide the ultimate answer, it can help me define my preferences and simplify the decision-making process. To this end, I gathered data from Il Sole 24 Ore, an Italian newspaper, to classify Italian provinces.

Our current home, Bologna is great, but does not meet some of our needs. Last fall, my family and I made the move from San Francisco to Bologna in search of a new home closer to our families in Turkey and a chance to immerse ourselves in Italian culture. Bologna was our first choice because it is centrally located in North Italy, offers easy connections globally, including Turkey, and has a large enough English-speaking community to help ease the culture shock for our children. We planned to either settle down in Bologna or at least enjoy it while we planned our next step. While Bologna met most of our expectations, it also fell short in some areas. For example, the air pollution in the Po Valley was more significant than I had anticipated. Although Bologna is at the border of the valley, the air quality is quite poor. The city is lively, but also crowded and noisy, and it lacks the outdoor opportunities for kids that we took for granted in San Francisco. PM10 concentrations in the Po Valley on 18th December, 2021.

source: https://www.iconameteo.it/news/notizie-italia/allarme-smog-nel-nord-italia-pm10-e-pm2-5-in-aumento-le-zone-piu-inquinate/

We have begun to think about our next move and plan to take a more informed approach this time. However, with so many beautiful towns in Italy, we need to narrow down our options. I began searching for objective metrics to help us make our decision. To my surprise, Bologna ranks first among all 107 provinces in Italy, according to a ranking by the Italian newspaper Il Sole 24 Ore. The ranking is based on 90 criteria that cover a wide range of areas, such as crime rate, home prices, length of cycling paths, access to health specialists, and green spaces. I do believe Bologna is a great place, but it being the top-ranking province in Italy? That was surprising. Taking a closer look at the methodology revealed why our perception and the ranking were not aligned: In its defense, Il Sole 24 Ore’s ranking is quite comprehensive and they developed a great site where you can dive into the individual dimensions and see rankings for each separate indicator. You can check it out yourself. Even if you don’t understand Italian, Google Translate does a good job. One issue with the overall ranking system is that it averages all indicators, without considering how they might overlap in measuring the same latent factors. The goal is to provide one general-purpose ranking using all the information available. For example, halving the crime rate is not necessarily as significant as doubling the length of bike paths in a region, but in this ranking it is. While this method of ranking may work for general purposes, it doesn’t consider the specific needs, values, and expectations of our family. Therefore, I wanted to use the data provided by Il Sole 24 Ore and create a new ranking that better aligns with our preferences. As a first step, I reviewed the 90 indicators and selected the ones that are most relevant to our family’s needs and values. I used the following four dimensions as a guide for determining which indicators to include: Social capital: I looked for indicators that reflect the level of social capital in an area along with the ease of integration for newcomers. It’s notoriously difficult to measure social capital directly in an area or community, but the literature suggests some proxies such as education, newspaper readership, organizational memberships, and participation in civic life. I selected overall education level, turnout in elections, percentage of immigrants living there, and the percentage of the population that is within our desired age range. I also included metrics related to disconnected youth (which I reverse-coded by using the percentage of “connected” youth) and the number of civil cases per 100,000 residents (which I also reverse-coded by converting it to the number of residents per civil case). Material well-being: I wanted to find out if the area has an efficient local government, a reliable public transportation system, a healthy population, easy access to public services, etc. These can be mostly captured by wealth and material development. I selected life expectancy at birth, a composite Urban Ecosystem score based on public transportation, garbage collection, etc., bike paths (km per person), green areas in the urban areas, a composite quality of life of women index based on 12 indicators such as gender equality in STEM, occupations and participation in sports. Safety: Coming from San Francisco, we are particularly sensitive about property crime and would love to live somewhere we don’t have to lock our bikes with the strongest steel chain money can afford even if we leave it for two minutes outside the market. I selected the overall criminality index, snatch thefts, and street robberies. Outdoors and sports: This is a part of life we care a lot about and our new home has to provide decent opportunities. I looked for indicators related to air quality, climate, and opportunities for outdoor activities.

Some other uncategorized indicators that I selected were composite scores for the quality of life for the elderly, young, and babies, housing prices, number of cafes, bookstores, and restaurants per person. I made some changes to the data to make it easier to compare different provinces in Italy. For example, I flipped the numbers for house prices and crime rates to make higher numbers mean better outcomes. I also converted all the scores to standardized units to make them easier to compare. This way, I can see which provinces are doing better or worse in different areas, like crime or housing prices. To make the data easier to interpret, I used a method called exploratory factor analysis. This technique helps to identify patterns in a large set of data and condenses it into a smaller set of factors. For example, a factor that represents Quality of Life can be inferred from measurements like life expectancy and gender equality. This allows me to focus on the latent factors that drive the observations, without worrying about the overlap and redundancy across different indices. Based on the correlation structure in my data, I identified four latent factors that could be used to explain the observed measurements. I will provisionally call them Factor A, B, C, and D, and will go over them one by one. Factor A - North and South The primary factor that explained the variations across provinces is a measure of wealth and economic development in an area. It’s positively associated with the quality of life for women, connected youth, high election turnouts, long life expectancy, and a strong urban ecosystem. However, it’s negatively associated with civil cases per capita, climate index, and number of days without rain. The negative correlation with two exogenous variables (rain and climate) suggests that there is a strong spatial component to Factor A because human activity is unlikely to have an effect at these scales. Factor A is also negatively correlated with air quality, which is consistent with the idea that wealth comes with industrial activity that can affect air quality. In areas with high Factor A, house prices are also high. Table: Indicators positively associated with Factor A. Indicators marked with * have flipped signs.

• 40th out of 107 in Factor A (material well-being)
• 7th out of 107 in Factor B (negative urbanism)
• 53rd out of 107 in Factor C (retirees and college students?)
• 9th out of 107 in Factor D (education) This is a far cry from the top place in Il Sole 24 Ore’s ranking. However, because the factors themselves are mixed-valued, I cannot use them as is for ranking provinces. It’s time to introduce our values and priors to the ranking. Our Personal Ranking While exploratory factor analysis helped me explore the data, the factors it produced were positively correlated with both good and bad indicators. EFA can’t generate “pure” factors that align with our personal values, which makes it difficult to make trade-offs between competing factors that determine livability in a town for me. Based on the previous analysis, I created 6 sub-dimensions that capture what we care about in a province. Each sub-dimension includes indicators that are positively correlated with each other, allowing me to create sub-indices that directly reflect our priorities. Wealth / Material quality of life Quality of life for women, *Disconnected youth (reverse coded as connected youth) Electoral participation *Litigation index (reverse coded) Newspaper readership Kids Life expectancy at birth Regular resident immigrants Bike paths Urban ecosystem Fitness, pools, well-being Urban green areas

Sports and Outdoors: Scores for regional opportunities and infrastructure for Swimming Nature tourism Cycling Winter sports

Safety *Snatch thefts (reverse coded) *Crime index (reverse coded) *Street robberies (reverse coded)

Street-life Number of restaurants per resident Number of cafes per resident Number of bookstores per resident

Peers Population (log-scaled, if everything else is equal we don’t mind more people around) % of the population between ages 0 and 14 % of the population between ages 30 and 50

Education Average years spent in the educational system % population with at least a high school degree % of people with post-secondary (college, trade schools, etc.) education

A very important dimension that is missing here is air quality and the weather (e.g., average sunlight received in winter, the number of days that are too hot or too cold). These metrics are very location specific and can change from one town to another even in the same province, especially in mountainous areas. The dataset I have only reports air quality for the capital town of each province and I am not satisfied with its quality. In any case, once I determine a short list of towns to check out it will be an easy exercise to find relevant data for these qualities and I don’t have to track down a reliable air quality dataset for all cities in Italy. I converted all of the above individual indicators to z-scores (representing how many standard deviations a province is away from the national average) and took the average within each sub-dimension, giving me 6 scores for each province. The correlation table reveals that some indices are negatively correlated (e.g., Safety and Peers, Street Life and Peers), suggesting a trade-off between competing goals. Others are positively correlated (Education, Outdoors and Sports, and Quality of Life), indicating that they might not be differentiated enough and could be capturing a common factor. Nonetheless, the correlations are neither too high nor too low, giving me confidence that these six dimensions are not excessively redundant. Considering our preferences, the three dimensions - Wealth, Sports/Outdoors, and Safety - hold the most significance for us when ranking towns. We believe that higher material quality of life, better access to sports and outdoor opportunities, and lower crime rates are always desirable, regardless of their current levels. On the other hand, the remaining three dimensions - Education, Street Life, and Peers - offer diminishing returns. For instance, we don’t mind if the percentage of our age peers is 60% or 70%, as long as it’s not as low as 20%. Regarding safety, however, I was concerned that I was over-indexing on this dimension. I tried to compare San Francisco and Bologna (two cities we are familiar with) to calibrate the relative safety scores of Italy with our perceived safety in the States. Turns out Bologna, even though it’s among less safe provinces in Italy, is actually comparable to or safer than San Francisco with respect to burglary and murder. Bologna had less than half of the burglary rate in San Francisco (321 burglaries per 100K residents versus 722 per 100K residents) and almost a tenth of SF’s murder rates (0.7 murders per 100K versus 7 murders per 100K). Even allowing for wide discrepancies in specific crime type definitions and reporting rates, it’s clear San Francisco would rank as one of the least safe cities had it been placed in Italy. So I decided not to put too much weight on Safety either. As a result of these considerations, I decided to account for our non-linear preferences, by eliminating the bottom quintile of provinces in each of the 6 sub-dimensions first, and then ranking them according to the average score over the most salient 2 sub-dimensions (Quality of Life and Sports and Outdoors). The initial filter left me with 26 provinces that do not rank among the worst quintile in either of the sub-dimensions. The top 15-ranking provinces according to the average Quality of Life and Sports and Outdoors (and the quintile groups they are in):

The top 3 positions (Sondrio, Trento, and Aosta) are northern and Alpine provinces that offer many outdoor opportunities and the highest quality of life scores. Trento ranking the 2nd is a strong confirmatory evidence for me that the ranking is working as expected, as I lived in Trentino-Alto Adige while doing my Ph.D. and it’s one of those rare paradise-like places on Earth where you can do road biking, mountain biking, skiing, hiking, and wind-surfing all within less than one-hour driving distance from your home. Unfortunately, it is too far away from major airports which I do not include as a dimension yet - a common aspect it shares with Sondrio and Aosta. A photo from a biking trip I took in Alto Adige in 2019.A surprising finding was the region of Marche which is represented by three provinces in the top-15: Pesaro e Urbino, Ancona, and Ascoli Piceno. We’ve never been to Marche before and never thought about it as a potential destination. My next step will be to read about and plan some trips to Marche. Another finding that made me reconsider some of my positions was Siena and Verona. Instinctively, I tend to discard touristic places as serious candidates, but Siena seems to have relatively high scores with Safety being in the top quintile and may deserve serious consideration. The same goes for Verona. Conclusion I have confidence my rankings work. The one province where I would spend the rest of my life if travel logistics was not a constraint is ranking 2nd. Bologna would not be among the candidates if we relied only on this analysis. Mainly because it couldn’t make the cut for the Safety dimension. I am still on the fence about this. It will be difficult to leave the trauma of our bikes stolen in San Francisco behind and I’ll keep thinking about Safety for a while. Verona and Siena, while being touristic, may deserve a second chance. Marche seems like an interesting region that I need to look into more. My next step will be to expand this list to actual Italian towns (Comune) that are located in these provinces and look at the driving times to major airports that we want to stay nearby.