ONLINE APPENDIX Concentrated Powers: Unilateral Executive Authority and Fiscal Policymaking in the American States As noted in Note 13 of the manuscript document, discrepancies exist between using Thad Beyle s gubernatorial budget powers index and the authors own coding when it comes to determining whether the governor has full or shared budget making powers/responsibility. 1 Specifically, the Beyle Index contains several coding errors regarding whether the governor has full or shared budget-making responsibility. This is a critical matter since this distinction is the foundational component of Beyle s five-point ordinal Governors Budget Powers Index that is widely used by scholars conducting research on American state politics and policymaking. Table SA-1 lists all relevant state-year observations in our sample where this is an issue. In addition, a spreadsheet file is included in the replication materials that contain both measures for all fifty states for the FY 1987-2009 sample period for scholars to compare and utilize in their own research (gubernatorial budget making power comparison.krause & Melusky.05-16- 11.xlsx). In total, there are 156 state-year observations based on Beyle s index which are inconsistent with the authors own independent coding directly from the Book of the States for 1 The authors data comes from The Governors: Powers Budget Making Power Table in The Book of the States for various years, while Beyle s is generated from his five point ordinal gubernatorial budget powers index (Sources: for 1980, 1988, 1994, 1998, 2001 actual years in which this index was updated, http://www.unc.edu/~beyle/gubnewpwr.html; for 2007, Thad Beyle and Margaret Ferguson. Chapter 7: Governors and the Executive Branch. Table 7-5: Governor's Institutional Powers, Pages 212-213. In Politics in the American States: A Comparative Analysis. Virginia Gray and Russell L. Hanson, editors. Ninth Edition. Washington, D.C.: CQ Press. (2007)). 1
each year covered in their sample period (14.85%. of the overall sample). This discrepancy is nontrivial given that these measures are correlated at only 0.5064 in our overall sample. A subset of 118 state-year observations (11.24% of the overall sample) that were misclassified using the Beyle index occur in years in which this index was not updated (shaded as forest green cells in the corresponding spreadsheet). The remaining 38 state-year observations (3.61% of the overall sample) that were misclassified using the Beyle index occurred in years in which this index was, in fact, updated (shaded as orange cells in the corresponding spreadsheet). [Insert Table SA-1 About Here] A summary of the unilateral executive policy control variables are discussed on the bottom paragraph appearing on Page 10 of the manuscript. As noted in Note 14 of the manuscript, a more detailed descriptive breakdown of these measures appears in Table SA-2. State-year observations equal to 0 refer to an absence of unilateral executive policy control (i.e., the governor does not enjoy unilateral control over budget making or official revenue forecasts), 1 refers to state-year observations in which the governor only has unilateral (full) budget making responsibility, 1 refers to state-year observations in which the governor only has unilateral control over official revenue forecasts, and 2 refers to state-year observations where the governor has both unilateral (full) budget making responsibility and unilateral control over official revenue forecasts. [Insert Table SA-2 About Here] Table SA-3 lists the descriptive statistics for the dependent variable, as well as for the remaining covariates. 2 Information on each of these variables mean, standard deviation (SD), minimum (Min), and maximum (Max) are reported. Because of document space limitations in 2 The descriptive statistics for the unilateral executive policy control measures are discussed on the bottom portion of Page 10 in the manuscript. 2
the manuscript, this information is provided here. This information can be obtained in the STATA output file contained in the replication materials (concentrated powers.05-16- 11.JOP.smcl). [Insert Table SA-3 About Here] In Note 33 of the manuscript, we mention three sets of supplementary regression analyses that appear in the Online Appendix document. The results from these analyses appear in the STATA output file contained in the replication materials (concentrated powers.05-16- 11.JOP.smcl). Statistical analysis of the partisan governor distinctions disaggregating unilateral executive policy control into separate budget making and revenue forecast authority are presented in Table SA-4. These results indicate that Democratic governors possessing unilateral control over either official revenue forecasts or budget making produces significantly higher fiscal spending growth compared to when they have neither unilateral power at their disposal. Moreover, revenue forecasting authority offers Democratic governors a greater impetus for spending than compared to possessing full budget making responsibility (Model SA-1: 4.11 vs. 1.50). Democratic governors possessing both unilateral control over both fiscal policymaking tools result in trivial differences in fiscal spending growth compared to the baseline of an absence of unilateral control (Model SA-1: Full Effect: 4.11 + 1. 50-4.93 = 0.68). This inference is consistent with the results from the aggregated measure of unilateral executive policy control reported in Model 5 in Table 2 of the manuscript. This general pattern also holds in the trimmed sample of observations where seven extreme valued state-year observations are dropped from the analysis (Model SA-2). Similarly, the disaggregate results for Republican governors display a pattern that is quite consistent with aggregate results reported in Table 2. Only when governors unilaterally control both official revenue forecasts and budget making authority does one observe a significant increase in fiscal spending growth (Model SA-3: Full 3
Effect: 0.39 + 0.67 + 4.82 = 5.88). This pattern still holds for the trimmed sample model, albeit it is somewhat attenuated (Model SA-4: Full Effect: 1.58-0.21 + 2.94 = 4.31). [Insert Table SA-4 About Here] We further explore the issue of gubernatorial partisanship effects by estimating regression models that drop all 20 observations corresponding to 3 rd party or independent governors (see Note 26 in the manuscript). The logic here is that gubernatorial partisan effects are being obscured by governors whose preferences may not easily fit into a partisan-ideological space. These results appear in Supplementary Table SA-5. The evidence for both the aggregate and disaggregate measures of unilateral executive control over fiscal policymaking are substantively similar to the findings reported in Table 1 of the manuscript. [Insert Table SA-5 About Here] Finally, an additional robustness check is performed that allows for conditional unilateral executive policy control effects with five of the executive constraint and incentive covariates (see Note 33 in manuscript). These results appear in Supplementary Table SA 6. For the key results, the magnitude of the coefficients and level of statistical significance attained by Partial Unilateral Executive Policy Control, as well as for Dual Balanced Budget Restrictions, change very little between models. Yet, the impact of fiscal transparency on fiscal spending growth is strongly conditioned by whether the governor has full unilateral executive policy control. Otherwise, fiscal transparency has no tangible bearing on fiscal spending growth in these set of model specifications. The coefficients for Lame Duck Gubernatorial Term fail to reach a level of statistical significance in Models SA-7 & SA-8 compared to the results from our reported results based on Models 1 & 2 reported in Table 2 of the manuscript. The results obtained from this analysis appear to reveal significant conditional unilateral executive policy control effects in only a handful of instances. This should hardly be surprising 4
given that the relative paucity of cases in each particular case, coupled with problems arising from collinearity, should make it difficult to uncover significant conditional effects. One of these exceptions involves the interaction between Full Unilateral Executive Policy Control x Fiscal Transparency Index Score which attained statistical significance in Models SA-7 & SA-8 (p < 0.01). That is, governors possessing unilateral control over both official revenue projections and budget formulation decrease fiscal spending growth by an average of 3.02% in Model SA-7 (2.50% in Model SA-8) for each two standard deviation (0.34) increase in the fiscal transparency index score. Conversely, the conditional effect of lame duck governors serving in their final term in office substantially enhances their willingness to engage in myopic policymaking behavior when they have absolute control over fiscal policymaking. Lame duck governors under these circumstances will generate an extra 2.86% fiscal spending growth compared to when they remain viable for seeking re-election. It should be noted, however, that this inference is based on 41 observations (3.90% of the entire sample), and is estimated with much less precision in the trimmed sample model (Model SA-8). [Insert Table SA-6 About Here] The conditional effects results are not reported in the manuscript for three reasons. First, in the vast majority of instances, the interaction effects are not statistically discernible from zero, plus baseline effects are more pronounced (i.e., less conservative) to those reported in Table 1. Second, in the handful of instances when the interaction coefficients are statistically discernible, they are based on a relative paucity of observations, and thus should be interpreted with caution. Finally, the statistical models presented in Table 1 provide a more parsimonious fit to these data relative to the results reported in Table SA-6. For each comparative model, the Bayesian Information Criterion [BIC] statistic is lower for the estimated model reported in Table 1 (Models 1 & 2) compared to those reported in Table SA-6 (Models SA-7 & SA-8). 5
SUPPLEMENTARY TABLE SA-1 State Name (Fiscal Years) Governor Full Budget Governor Full Budget Responsibility (Authors) Responsibility (Beyle) Arkansas (2003-2008) 0 1 Colorado (1988, 1999-2002, 2007-2008) 0 1 Connecticut (2004-2008) 0 1 Florida (1994-2002, 2007-2008) 0 1 Georgia (2004-2006) 0 1 Hawaii (2007-2008) 0 1 Illinois (2003-2008) 0 1 Indiana (2003) 0 1 Iowa (2005-2008) 0 1 Kansas (1988) 1 0 Kentucky (1988, 1999) 0 1 Louisiana (1999-2001, 2008) 0 1 Maine (2007-2008) 0 1 Michigan (2003-2006) 0 1 Mississippi (1989-2001, 2008) 0 1 Montana (2003) 0 1 Nebraska (2003-2008) 0 1 New Mexico (1988) 0 1 New Mexico (1990-1998) 1 0 New York (2003-2008) 0 1 North Carolina (1989-2002) 0 1 Oklahoma (2003-2008) 0 1 Oregon (2007-2008) 0 1 Pennsylvania (1992-2002) 0 1 Rhode Island (2007-2008) 0 1 South Dakota (2005-2006) 0 1 South Dakota (2008) 1 0 Tennessee (2004-2007) 0 1 Texas (2002-2007) 0 1 Utah (2007-2008) 0 1 West Virginia (1990-1995) 0 1 Wyoming (2004-2008) 0 1 6
SUPPLEMENTARY TABLE SA-2 State Name Executive Fiscal Authority Fiscal Years Alabama 1 1988-2008 Alaska 1 1988-2008 Arizona 1 1988-2008 Arkansas 2 1988-2002 1 2003-2008 California 1 1988-2008 Colorado 1 1988 0 1989-2002, 2007-2008 1 2003-2006 Connecticut 1 1988-2003 0 2004-2008 Delaware 1 1988-2008 Florida 1 1988-1993, 2003-2006 0 1994-2002, 2007-2008 Georgia 2 1988-2003, 2007-2008 1 2004-2006 Hawaii 1 1988-2006 0 2007-2008 Idaho 1 1988-2008 Illinois 1 1988-2002 0 2003-2008 Indiana 1 1988-2002, 2004-2008 0 2003 Iowa 1 1988-2004 0 2005-2008 Kansas 1 1988-2008 Kentucky 1 1988-1993 0 1994-1999 1 2000-2008 Louisiana 0 1988-2008 Maine 2 1988-1992 1 1993-2006 0 2007-2008 Maryland 1 1988-2008 Massachusetts 1 1988-2008 Michigan 1 1988-2002, 2007-2008 0 2003-2006 Minnesota 2 1988-2008 Mississippi 1 1988-1992 0 1993-2008 Missouri 1 1988-2008 Montana 1 1988-2002, 2004-2008 0 2003 Nebraska 1 1988-2002 0 2003-2008 Nevada 1 1988-2008 New Hampshire 1 1988-2008 New Jersey 1 1988-2008 New Mexico 0 1988-1989 7
1 1990-2008 New York 1 1988-2002 0 2003-2008 North Carolina 0 1988-2002 1 2003-2008 North Dakota 2 1988-2008 Ohio 1 1988-2008 Oklahoma 1 1988-2002 0 2003-2008 Oregon 2 1988-2006 1 2007-2008 Pennsylvania 2 1988-1991, 2003-2008 1 1992-2002 Rhode Island 2 1988-1990 1 1991-2006 0 2007-2008 South Carolina 0 1988-2008 South Dakota 1 1988-2004, 2007-2008 0 2005-2006 Tennessee 2 1988-1992 1 1993-2003 0 2004-2008 Texas 0 1988-2008 Utah 1 1988-2006 0 2007-2008 Vermont 1 1988-2008 Virginia 2 1988-2008 Washington 1 1988-2008 West Virginia 2 1988-1989, 1996-2008 1 1990-1995 Wisconsin 1 1988-2008 Wyoming 1 1988-2003 0 2004-2008 Governor had an absence of UEC (neither budget making/formulation nor revenue forecast authority): coded 0, governor had UEC budget making/formulation only: coded 1, governor had UEC revenue forecast authority only: coded 1, and governor had full UEC (budget making/formulation and revenue forecast authority): coded 2. 8
SUPPLEMENTARY TABLE SA-3 Descriptive Statistics for Variables (Full Sample: N T = 1050) Variable Mean SD Min Max % Change Real Per Capita General Expenditure Growth 3.03 4.61-20.74 28.95 Dual Balanced Budget Restrictions 0.56 0.50 0.00 1.00 Fiscal Transparency Index Score 0.51 0.19 0.11 1.00 Divided Government 0.59 0.49 0.00 1.00 Legislative Veto Override Majority 0.13 0.34 0.00 1.00 Lame Duck Gubernatorial Term 0.27 0.44 0.00 1.00 Gubernatorial Election Year 0.25 0.43 0.00 1.00 State Citizen Ideology 49.93 14.90 8.45 95.97 Binding Forecast 0.41 0.49 0.00 1.00 Governor s Party -0.05 0.99-1.00 1.00 Partisan Control of the Legislature 0.14 0.84-1.00 1.00 Bienniel Budget Process 0.40 0.49 0.00 1.00 Real Average Legislative Salary 23765.34 20388.50 0.00 111678.90 Lagged % Per Capita Real Personal Income Growth 1.91 2.12-13.09 9.64 Lagged Unemployment Rate 5.37 1.65 2.30 13.40 Lagged Economic Growth Volatility 1.75 1.16 0.05 10.85 % Reliance on Sales Tax Revenues 0.23 0.11 0.00 0.50 Lagged % Per Capita Real Federal Aid Growth 3.04 10.12-142.20 156.03 9
SUPPLEMENTARY TABLE SA-4 The Impact of Robust Executive Authority on Fiscal Spending Growth in the American States (FY 1988 FY 2008) Fixed Effects Estimation (Robust Standard Errors Clustered by State) --- Disaggregating Unilateral Executive Policy Control by Partisan Governors Democratic Governors Model SA-1 [Full Sample] Democratic Governors Model SA-2 [Trimmed Sample] Republican Governors Model SA-3 [Full Sample] Republican Governors Model SA-4 [Trimmed Sample] Coefficient Coefficient Coefficient Coefficient Level of Unilateral Executive Policy Control Unilateral Executive Revenue Forecast Control (+) 4.11*** (1.40) 3.18** (1.31) 0.39 (2.35) 1.58 (1.28) Unilateral Executive Budget Control (+) 1.50** (0.68) Combined Unilateral Executive Revenue Forecast & Budget Control (+) -4.93*** (1.65) Executive Constraints on Fiscal Spending Dual Balanced Budget Restrictions [Balanced Budget Law & No Deficit Carry Over Law] (-) Fiscal Transparency Index Score (-) -3.66 (3.13) Divided Government (-) 1.69 (1.21) Governor Line-Item Veto Authority (-) 0.08 (1.26) Legislative Veto Override Majority (-) -0.59 (1.10) Executive Incentives for Fiscal Spending Lame Duck Gubernatorial Term (+) 0.32 (0.43) Gubernatorial Election Year (+) -0.63 (0.46) 1.67** (0.67) -3.05** (1.22) 0.67 (0.91) 4.82** (2.24) -2.10 + -3.71 + (2.79) 1.63 (1.23) 0.30 (1.27) -0.73 (1.26) 0.16 (0.42) -0.40 (0.49) (1.60) -0.95 (5.13) -0.76 (1.10) 1.54 (1.60) 0.33 (0.69) 1.42** (0.56) 0.30 (0.40) -0.21 (0.80) 2.94*** (1.09) -2.05 (1.62) 0.21 (4.38) -0.76 (1.09) 0.68 (1.32) 0.55 (0.83) 0.52 (0.52) 0.15 (0.29) State Citizen Ideology (+) -0.03-0.03 0.04 (0.04) 0.002 (0.03) Ancillary Controls Binding Revenue Forecast (-) 0.02 (0.89) -0.32 (0.94) 4.69*** (1.74) 3.80** (1.61) Partisan Control of the Legislature (+) 1.29 + (0.83) 1.10 + (0.82) 1.05 + (0.64) 1.25* (0.65) Bienniel Budget Process (-) -4.42*** (0.95) -3.67*** (0.80) 2.58*** (0.74) 1.18 (1.19) Real Average Legislative Salary (+) -0.24E-04 (0.39E-04) -0.12E-04 (0.42E-04) 0.33E-04 (0.42E-04) 0.29E-04 (0.41E-04) Lagged % Per Capita Real Personal Income Growth (+) 0.35** (0.13) 0.33** (0.12) 0.03 (0.13) 0.14 + (0.10) Lagged Unemployment Rate (-) -1.04*** (0.22) -0.82*** (0.20) -0.59*** (0.19) -0.54*** (0.20) Lagged Economic Growth Volatility (-) -0.39** (0.17) -0.39** (0.16) -0.41* (0.23) -0.41** (0.20) % Reliance on Sales Tax Revenues (+) 11.09 (11.72) 10.80 (9.99) 13.34 + (8.72) 12.93 + (9.54) Lagged % Per Capita Real Federal Aid Growth (+) 0.05*** (0.01) 0.05*** (0.01) 0.06*** 0.05* Constant 3.86 (4.22) 3.30 (3.84) -6.15 (5.11) -2.53 (3.83) Correlation (u i, Xβ) -0.68-0.65-0.90-0.88 Overall R 2 0.113 0.099 0.005 0.008 BIC 2841.05 2712.09 3225.90 2932.81 Effective Sample Size 490 483 540 527 Notes: Dependent variable is defined as the Annual Per Capita Real State Government General Expenditure Growth. Robust standard errors clustered by state appear inside parentheses. *significant at the 0.10 level **significant at the 0.05 level ***significant at the 0.01 level (two-tailed) + significant at the 0.10 level (one-tailed test). 10
SUPPLEMENTARY TABLE SA-5 The Impact of Robust Executive Authority on Fiscal Spending Growth in the American States (FY 1988 FY 2008) Fixed Effects Estimation (Robust Standard Errors Clustered by State) --- Excluding Third Party Governor Cases Aggregate Model SA-5 Disaggregate Model SA-6 Level of Unilateral Executive Policy Control Partial Unilateral Executive Policy Control (+) [Unilateral Executive Revenue Forecast Control] Coefficient 1.13** (0.46) Coefficient 1.72 (1.33) Full Unilateral Executive Policy Control (+) [Unilateral Executive Budget Control] Combined Unilateral Executive Revenue Forecast & Budget Control (+) -1.07 (1.32) Executive Constraints on Fiscal Spending Dual Balanced Budget Restrictions [Balanced Budget Law & No Deficit Carry Over Law] (-) 1.31 (1.23) -2.62*** (0.68) Fiscal Transparency Index Score (-) -3.13 + (1.95) Divided Government (-) -0.23 (0.34) Governor Line-Item Veto Authority (-) 0.68 (0.76) Legislative Veto Override Majority (-) -0.46 + Executive Incentives for Fiscal Spending (0.33) Lame Duck Gubernatorial Term (+) 0.66* (0.35) Gubernatorial Election Year (+) -0.14 (0.29) 1.06** (0.48) -2.61*** (0.68) -3.03 + (1.97) -0.21 (0.34) 0.70 (0.76) -0.44 (0.34) 0.65* (0.36) -0.15 (0.29) State Citizen Ideology (+) 0.002 0.0003 Ancillary Controls Binding Revenue Forecast (-) -0.36 (0.72) Governor s Party (+) 0.27 + (0.18) Partisan Control of the Legislature (+) 0.51 + (0.31) Bienniel Budget Process (-) -1.10 + (0.67) Real Average Legislative Salary (+) 0.34E-04* (0.19E-04) Lagged % Per Capita Real Personal Income Growth (+) 0.21** (0.10) Lagged Unemployment Rate (-) -0.71*** (0.14) Lagged Economic Growth Volatility (-) -0.29** (0.13) % Reliance on Sales Tax Revenues (+) 16.95** (6.83) Lagged % Per Capita Real Federal Aid Growth (+) 0.05*** (0.01) Constant 0.06 (2.18) -0.19 (0.68) 0.27 + (0.18) 0.51 + (0.31) -1.13 + (0.68) 0.34E-04* (0.19E-04) 0.21** (0.10) -0.72*** (0.14) -0.29** (0.13) 17.00** (6.80) 0.05*** (0.01) -0.03 (2.23) Correlation (u i, Xβ) -0.83-0.83 Overall R 2 0.035 0.036 BIC 6068.19 6074.86 Effective Sample Size 1030 1030 Notes: Dependent variable is defined as the Annual Per Capita Real State Government General Expenditure Growth. Robust standard errors clustered by state appear inside parentheses. *significant at the 0.10 level **significant at the 0.05 level ***significant at the 0.01 level (two-tailed) + significant at the 0.10 level (one-tailed test). 11
SUPPLEMENTARY TABLE SA-6 The Impact of Robust Executive Authority on Fiscal Spending Growth in the American States (FY 1988 FY 2008) Fixed Effects Estimation (Robust Standard Errors Clustered by State) --- Allowing for Various Conditional Relationships Aggregate Model SA-7 [Full Sample] Aggregate Model SA-8 [Trimmed Sample] Level of Unilateral Executive Policy Control Coefficient Coefficient Partial Unilateral Executive Policy Control (+) 3.10*** (1.11) Full Unilateral Executive Policy Control (+) 7.38** (3.50) Executive Constraints on Fiscal Spending Dual Balanced Budget Restrictions [Balanced Budget Law & No Deficit Carry Over Law] (-) -2.14** (0.85) Fiscal Transparency Index Score (-) -0.94 (1.83) Divided Government (-) 0.30 (0.81) Governor Line-Item Veto Authority (-) 0.44 (0.85) Legislative Veto Override Majority (-) -0.40 (0.35) Executive Incentives for Fiscal Spending Lame Duck Gubernatorial Term (+) 0.06 (0.52) Gubernatorial Election Year (+) 0.52 (0.78) State Citizen Ideology (+) -0.001 2.89*** (1.07) 6.78** (2.76) -1.89** (0.81) 0.15 (1.65) 0.14 (0.80) -0.08 (0.69) -0.50 (0.42) -0.13 (0.59) 0.33 (0.81) -0.01 Conditional Effects Partial Unilateral Executive Policy Control Dual Balanced Budget Restrictions Full Unilateral Executive Policy Control Dual Balanced Budget Restrictions Partial Unilateral Executive Policy Control Fiscal Transparency Index Score Full Unilateral Executive Policy Control Fiscal Transparency Index Score Partial Unilateral Executive Policy Control Divided Government Full Unilateral Executive Policy Control Divided Government Partial Unilateral Executive Policy Control Lame Duck Gubernatorial Term Full Unilateral Executive Policy Control Lame Duck Gubernatorial Term Partial Unilateral Executive Policy Control Gubernatorial Election Year Full Unilateral Executive Policy Control Gubernatorial Election Year -0.40 (0.83) -2.25 (2.97) -1.97 (1.73) -8.87*** (2.85) -0.60 (0.84) -0.28 (1.07) 0.50 (0.58) 2.86*** (0.95) -0.76 (0.86) -1.05 (0.94) Constant -1.25 (2.17) -0.45 (0.67) -2.68 + (2.00) -2.31 + (1.65) -7.36*** (2.46) -0.45 (0.83) -0.33 (1.02) 0.19 (0.66) 2.25 + (1.34) -0.64 (0.84) -0.05 (0.89) -0.12 (1.77) Correlation (u i, Xβ) -0.81-0.79 Overall R 2 0.041 0.049 BIC 6236.72 5798.34 Effective Sample Size 1050 1030 Notes: Dependent variable is defined as the Annual Per Capita Real State Government General Expenditure Growth. Robust standard errors clustered by state appear inside parentheses. *significant at the 0.10 level **significant at the 0.05 level ***significant at the 0.01 level (two-tailed) + significant at the 0.10 level (one-tailed test). 12