State Fiscal Reserves and the Great Recession

ABSTRACT

Governments around the world accumulate fiscal reserves to prepare for exogenous shocks such as recessions. Using data from 2004 to 2016 for state governments in the United States, we estimate empirical models controlling for time-variant and -invariant state characteristics. Our results show that savings increase at the expense of spending in specific categories and that fiscal reserves also support only certain expenditure functions. Further, we find that the relationship between savings and spending differs over the business cycle as fiscal reserves increase during times of economic prosperity and decrease during economic declines.

KEYWORDS:

• Budgeting
• financial management
• budget stabilization

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1. See for earlier studies: (Gold, 1983; Navin & Navin, 1997; Pollock & Suyderhood, 1986; Vasche and Williams 1987).

2. Some cutback management research on local governments in the United States also finds that expenditure cuts are unevenly distributed during fiscal stress (e.g., Arapis et al., 2017; Kim & Chen, 2020).

3. Our first two strategies are consistent with the prior fiscal savings literature at the state (e.g., Hou, 2003; Wei & Denison, 2019) and local levels (e.g., Jones et al., 2021).

4. We state vectors and their coefficients in bold letters.

5. In the local savings literature, authors have similarly examined changes in UUBs over the business cycle. For example, multiple articles have split the panel and estimated regression models prior to, during, and/or after periods of economic decline (Arapis & Reitano, 2018; Arapis et al., 2017; Stewart, 2009, 2011; Stewart et al., 2017, 2013). For this article, we use interaction effects, similar to Barrett et al. (2019).

6. This modification to the state sample is consistent with Hou (2005) and Wagner and Elder (2005).

7. Note that the classification of BSFs differs between studies (Hou & Brewer, 2010; Knight & Levinson, 1999; Rodriguez-Tejedo, 2012; Wagner, 2003). Our classification is closest to that of the Pew Charitable Trusts, which also maintains the data used in the analysis. Alabama and Oregon are included in the sample after 2007.

8. We assign a value of 50% for states electing legislators without party designation.

9. In the case of a unified government, as evidenced by the three aforementioned measures, a dichotomous variable receives a value of one.

10. See the following website: https://rcfording.wordpress.com/

11. For the sake of comprehensiveness, the first coefficient in Model (1) is interpreted as follows: A $1 per capita increase in BSFs is associated with a $0.461 decrease in total expenditures per capita. To calculate the relationship between BSFs and total expenditures for the average state in our sample, we multiply 0.461 by the $27 per capita from Table 2, which results in a $12.5 per capita decrease in total expenditures for a $27 per capita increase in BSFs.