Comparison to Previous Indices of State-Level Economic Freedom

Comparison to Previous Indices of State-Level Economic Freedom

This project remains the only effort to code both economic and personal freedom in the 50 states. Other studies compare economic freedom or “competitiveness” in the states but do not treat other critical aspects of individual liberty or selectively subsume a few noneconomic issues within economic freedom concepts. For example, the Fraser Institute’s Economic Freedom of North America (EFNA) index does not deal with such interventions as gun control, homeschooling regulations, and marijuana laws.43 Meanwhile, the Pacific Research Institute’s U.S. Economic Freedom (USEF) index subsumes gun control and seatbelt laws under “Regulatory Sector” along with occupational licensing, recycling programs, and labor regulations.44 Lastly, Rich States, Poor States, a publication of the American Legislative Exchange Council (ALEC), creates a state ranking on Economic Outlook based on 15 fiscal and regulatory variables that are equally weighted. To be fair, economic freedom (or economic policy outlook) may be a valid concept unto itself, and these studies claim only to measure that concept. However, given that liberty and human flourishing encompass and require more than mere economic freedom, this study provides a more robust understanding of the overall condition of freedom in the American states.

We also believe our measurement of economic freedom improves on prior studies. In fact, this report includes component scores for both economic freedom (the sum of scores on our fiscal policy and regulatory policy categories) and personal freedom (the paternalism category) for those who wish to maintain the distinction.45 We note improvements under the following five headings:

  1. Number of variables. Our database includes far more variables than the EFNA and ALEC studies, which use 10 and 15 variables, respectively, while avoiding the pitfalls of double counting and variable interdependence in the USEF study, which includes multiple variables for tobacco and alcohol taxes, recycling requirements, total tax take, and various categories of government spending. We also have complete data on every variable.
  2. Standardization of variables. ALEC does not specify a standardization method. EFNA uses a 0–10 scale standardization of every variable, where 0 corresponds to “a low level of economic freedom” on the policy measure and 10 corresponds to “a high level of economic freedom,” with other states interpolated based on relative position on the raw variable. The problem with this approach is that it is extremely sensitive to outliers. If one state has much higher government spending than other states, for instance, then 49 states will cluster around the 10 value while the big-government state will take on the 0 value. USEF ranks the states 1–50 on each indicator variable and averages those indicators to create sector scores, then uses principal components analysis (PCA) to reduce the variance in their policy variables to a single dimension. This method of standardizing the indicator variables throws out important information, namely the size of the difference between states on continuous variables (such as government spending). By contrast, we standardize each variable by taking, for each state on each policy variable, the number of standard deviations better (freer) than the mean.46 This approach takes into account the size of differences among states on raw variables while moderating skew due to outliers.
  3. Weighting of variables. USEF averages standardized indicators for five components of economic freedom and then conducts PCA on those five components, extracting the first dimension as the summary measure of economic freedom. As a check of external validity the authors report that the overall economic freedom variable predicts net population migration, which we also find for our personal and economic indices. The problem with using PCA to create an index of economic freedom is that it uses correlations among variables to create the components. In essence, the procedure teases out the ways state governments tend to covary on public policies, a concept that political scientists refer to as “policy ideology.”47 Thus, liberal states tend to have high income taxes, low sales taxes, and recognition of same-sex domestic partnerships, for instance, while conservative states take the opposite tack on those policies. Policies that are highly ideologically charged will “load” heavily onto the first extracted principal components.48 The USEF measure of economic freedom is actually a measure of policy conservatism on economic issues. While USEF problematically uses PCA to weight the variables, EFNA and ALEC weight policy areas equally to create their overall indices. Although there is no objectively correct way to weight these variables, since every individual values different aspects of freedom differently, we have weighted variables roughly according to the number of people affected by the policy, the intensity of preferences on the issue, and the importance of state policy variation.
  4. Measurement issues. We improve on previous attempts to measure fiscal interventionism. For instance, USEF uses revenues and spending per capita, which are poor measures of government intervention that reward states for having low per-capita income (because states with poorer economies bring in less revenue for a given tax rate). Mississippi has low government spending per capita but high government spending as a percentage of the state economy. EFNA divides revenues and spending by state GDP, which is better but not ideal, since state GDP figures suffer from a “corporate headquarters bias” and the attribution of labor income solely to the state where it was earned (important for states that send or receive many interstate commuters). To figure out the best denominator for fiscal variables, we regressed total spending and revenues by state on state personal income, GDP, and “corrected GDP” (used in the last version of this index). We found that state personal income was by far the best predictor of the size of state and local government, and that once personal income was included, none of the other measures of economic size correlated with government size. Since we believe that this aspect of freedom is inversely related to the proportion of the economy coercively extracted by the state, and personal income appears to be the best measure of the resources available for such extraction, we now measure taxation and spending as a percentage of personal income.
  5. Variable relevancy. USEF includes variables that might not bear a direct relationship to freedom (e.g., number of state legislators and government units). ALEC includes an institutional measure: constitutional tax and expenditure limits. Our database includes only variables measuring public policies and their enforcement, rather than policy outcomes (growth, unemployment, etc.) or institutional rules (size of legislature, initiative and referendum, procedures for raising taxes or spending, etc.).

In conclusion, our report not only provides a broader framework for understanding the state of freedom in the American states, but also carefully measures the economic components of freedom.


Footnotes

43. Nathan J. Ashby et al., Economic Freedom of North America 2010 (Vancouver, BC: Frasier Institute, 2010).

44. Lawrence J. McQuillan et al., U.S. Economic Freedom Index: 2008 Report (San Francisco: Pacific Research Institute, 2008), http://www.pacificresearch.org/docLib/20080909_Economic_Freedom_Index_2008.pdf.

45. We nevertheless maintain that individual liberty is a seamless concept and that a rigid conceptual division between “economic” and “personal” freedoms is unsupportable. Singapore is not very free despite its pro-capitalist economic policy. Property rights are not secure when “unapproved” uses of property (or one’s own body) are punished.

46. For variables for which lower raw numbers are better, the formula for the standardized variable is STDVARi=-(RAWVARi-RAWVAR)/(stdev(RAWVAR)).For variables for which higher raw numbers are better, the formula for the standardized variable is STDVARi=(RAWVARi-RAWVAR)/(stdev(RAWVAR))

47. See Erikson, Wright, and McIver, Statehouse Democracy; and Sorens, Muedini, and Ruger, “U.S. State and Local Public Policies in 2006.”

48. For instance, occupational licensing is an important threat to freedom but does not load significantly onto the first component extracted from a PCA because it is not a liberal-conservative ideological issue. See Sorens, Muedini, and Ruger, “U.S. State and Local Public Policies in 2006.”

' '