Using simple metrics to quantify development capacity.

• Author: Zipper, Riley

Recent large-scale economic development (ED) initiatives like Amazons search for "HQ2" have underscored the need for ED practitioners and policymakers to be able to measure their community's development capacity. The immensity of available data that could be used to do so, however, can be quite overwhelming.

The Indiana Business Research Center's (IBRC) work with Metrics for Development (M4D) involves collecting relevant data that can be used to gauge a county's capacity for economic development and packaging it so policymakers, ED practitioners and the public can derive meaningful insights for their communities. M4D, along with its umbrella project Regional Economic Development (RED), will build upon the functionality of StatsAmerica.org, which already provides vast amounts of data.

Methods

The underlying data contain a variety of different scales and sizes, so using the raw numbers could produce misleading results because it would mean measures on a larger scale would, in effect, be weighted more heavily than those on a smaller scale.

A simple min-max method of feature scaling was used to represent all data items on a scale from 0 to 1, where 0 = worst and 1 = best in the United States for any given measure. For those measures that would, in theory, negatively impact development (e.g., poverty rate), the inverse was used. This process ensures the "0 = worst, 1 = best" dichotomy is upheld and makes comparison between the 3,110 U.S. counties straightforward and intuitive.

Given the vastness of the data set (over 70 variables for each U.S. county), 13 indexes were created by adding the relevant measures together and then applying min-max rescaling again. Indexes were built around topics that represent a conventional accounting of development capacity, including human capital, health, industries, population and occupations. More novel indexes that considered less-obvious aspects of capacity, such as commuting and urban sprawl (e.g., means of transportation to work), civil society (e.g., charitable organizations), natural amenities and school financing, were also constructed to paint a comprehensive picture of each county's development capacity.

Table 1 indicates the variables within each index. Finally, using the 13 indexes, the M4D Index was created for each county to show its overall capacity for social and economic development. When the M4D Index is standardized using the same scheme as above, interpreting which counties are capable of development and which need some work is simple: the county with the greatest development capacity has a standardized M4D value of one, and the county with the least development capacity has a value of zero.

View county-level nationwide maps of these indexes in the appendix.

Nationwide findings: A rising tide lifts all boats

A good way to start exploring these data is to look at the differences between the highest-and lowest-scoring counties in the United States. Figure 1 shows the makeup of the raw (i.e., non-standardized) M4D indexes for the five highest- and lowest-scoring counties. You can see that the five counties at the low end of the spectrum are penalized for their scarcity of job opportunities, low earnings and productivity, poor health, and lack of school systems. Ziebach County, South Dakota, also suffers from remarkably high crime: Its Crime Index is 0, meaning it has the highest crime per capita in the country. For 2014 (the latest year of data available), the FBI reported 284 property crimes per 1,000 people and an astounding 215 violent crimes per 1,000 people. For context, the next closest county is Lincoln County, West Virginia, with 218 property crimes and 39 violent crimes per 1,000 people.

What exactly is so different between the counties at the low end and the high end? To investigate this question, the counties with M4D scores above and below two standard deviations from the mean--a total of 158 counties--were isolated and statistics were compared for nearly 200 variables, some of which were included in the indexes and some of which were not. Table 2 shows a small selection of these.

The variables with the largest discrepancies between the lowest- and highest-scoring counties include net migration rate, population growth from 2010-2016, all of the natural amenities variables, employment growth from 2001-2016, the percent of residents with obesity and diabetes, crime, venture capital investments, poverty, the percent of residents with a bachelor's degree, and the percent of residents who reported not working over the last year.

The lowest-scoring counties exhibit several characteristics that are contributing to the divergence, including:

* Low human capital...