Health politics? Determinants of US states’ reactions to COVID-19

View author details View Less
  • 1 University of Lille, France and CIRANO, , Canada
  • | 2 University of Lille, , France
  • | 3 University of Lille, , France, Polytechnique Montréal and CIRANO, , Canada
Full Access
Get eTOC alerts
Rights and permissions Cite this article

Were policy responses of the US states to the pandemic driven by partisan politics or by budgetary reasons? We show that balanced-budget rules also had an impact, mediated by the possibility of benefiting from the funds previously stored in budget stabilisation funds. State policymakers tried to square the circle by simultaneously respecting budget rules, limiting the economic impact of the social distancing measures, combating the pandemic and pandering to their political bases. Some fiscal rules have induced a trade-off between health and public finance, which may reignite the debate on the pro-cyclicality of fiscal rules.

Abstract

Were policy responses of the US states to the pandemic driven by partisan politics or by budgetary reasons? We show that balanced-budget rules also had an impact, mediated by the possibility of benefiting from the funds previously stored in budget stabilisation funds. State policymakers tried to square the circle by simultaneously respecting budget rules, limiting the economic impact of the social distancing measures, combating the pandemic and pandering to their political bases. Some fiscal rules have induced a trade-off between health and public finance, which may reignite the debate on the pro-cyclicality of fiscal rules.

Introduction

Today people started losing their jobs because of … Do Nothing Democrats, who should immediately come back to Washington and approve legislation to help families in America. End your ENDLESS VACATION! (Donald J. Trump, Twitter, 15:00, 17 April 2020)

According to Adolph et al (2020), faced with the COVID-19 pandemic, Republican governors and governors in states with more Trump supporters reacted more slowly, adopting social distancing measures more reluctantly. Such a partisan, politically induced reaction would clearly be in line with President Trump’s rants (as exemplified by the opening quote, in which lockdowns and shelter-in-place orders are essentially compared to imposed vacations). However, the decisions also entailed huge potentially negative health consequences.

It must be acknowledged that governors had to take decisions in a highly uncertain and politically fraught environment. The opening quote from President Trump confirms the political stance, itself confirming a previous view in which he insisted that the ‘cure [that is, lockdowns and social distancing] cannot be worse than the problem itself’.1 Hence, it may be true that the political context had an impact on the adoption of policy measures to fight the new epidemic. Yet, is this the whole story?

In fact, adopting quarantine, shelter-in-place lockdowns or other forms of social distancing measures is de facto to impose a cost on the economy. Closing businesses is economically costly but shutting schools also has a strong impact as parents have to choose between working or staying at home to take care of their children, the latter decision implying that firms have to face a labour shortage and thus disruptions in the production process. Closing businesses obviously affects sales- and profit-based tax revenues, while closing schools and other places of congregation has second-round impacts as households’ revenues dwindle and income-based tax revenues shrink correspondingly.

Moreover, not only is it hard to decide what restrictive measures to take, but deciding when to announce them is also problematic. From an epidemiological perspective, the earlier the containment measures are taken, the shorter they need to last. Later adoption of containment measures can lead to harsher consequences for government finances because later adoption will induce a longer period of economic freezing. Hence, policymakers are confronted with a twinned trade-off: not only does the adoption of measures on social distancing and restrictions on economic activity save lives at the cost of lost economic activity (and induced public finance consequences); but the timing of the decisions can also be important as adopting too late or too fast also makes a difference, in both the health and economic dimensions.

Measuring the economic costs of the pandemic is still a daunting challenge but the first estimates draw a landscape of pain and sorrow, with probably the worst recession the US may have ever known.2 Concerning the US states, Clemens and Veuger (forthcoming) estimate that state government sales and income tax revenues will drop by approximately US$106 billion in the fiscal year 2021, representing 0.5 per cent of their gross domestic product (GDP), with a loss of 11.5 per cent in expected revenues. As expenditures may increase too, the states’ budgets may be widely affected. However, most US states face the binding constraint of balanced-budget requirements. Balanced-budget rules (BBRs) force states to balance their books every year, generally forbidding their governors and legislatures from passing, executing or reporting any deficit.3

Hence, as the pandemic spread in early 2020, governors were faced with the threat of falling expected revenues, the possibility of having to increase expenditures to support their population and the illegality of running a deficit. If most of the states also benefit from the presence of rainy-day funds (RDFs), also called budget stabilisation funds (see, for example, Zhao, 2016), in which previous surpluses may have been ‘stored’ in case of harsh circumstances, these funds cannot easily be raided and exit rules are often stringent. In other words, given that the budgetary process is constrained by the presence of fiscal rules, governors have had to face the pandemic, with its induced costs (economic and fiscal), while they were preparing the next fiscal year’s budget. Moreover, the stabilisation funds may have been expected to moderate the unexpected shock to government expenditure and revenue, if only they could easily be raided. A question thus emerges: given that it is the announced health-related measures that are likely to create the fiscal shock, have the funds played their counter-cyclical vocation?

Previous research has shown that restrictions on the possibility of carrying deficits from one year to the other induce states to implement adjustments (that is, spending cuts or tax increases) in the face of economic shocks (for a theoretical appraisal, see Poterba, 1994; Clemens and Miran, 2012; Azzimonti et al, 2016). As a consequence, BBRs have been accused of creating volatility by inducing pro-cyclical responses (which was particularly noticeable during the Great Recession, as Jonas [2012] and Campbell and Sances [2013] have shown). Stringent fiscal rules may impede policymakers’ reactions to shocks for fear of breaching the balanced-budget requirements. In short, BBRs reduce the possibility of smoothing out the impact of economic shocks. In some ways, the COVID-19 pandemic is no different from other shocks, and governors have been caught between a rock and a hard place: how can they support the population and deal with the economic consequences of the shock while ensuring a balanced budget?

One cannot rule out the possibility that policymakers in the US states may have been fearful of the fiscal impacts of the adoption of health policy measures that were, essentially, driving the economy to a halt, bringing with them large reductions in revenues. In this research, we thus analyse if and how partisan politics and fiscal institutions correlated in US states’ reactions to the health crisis. We analyse how fiscal rules and the rules governing the use of budget stabilisation funds correlate with the policy measures taken to combat the epidemic in the US. In terms of health measures, we first consider the determinants of the number of social distancing measures announced by US states (up to 7 April). Then, we analyse the length of time between the rise of the epidemic and the announcement of the social distancing measures. Finally, we look at the probability of having a shorter length of reaction before the adoption of each social distancing measure.

To our knowledge, there are two articles that are closely related to what we explore here. First, Adolph et al (2020) explore how the interplay between the spread of the pandemic, political partisanship and policy diffusion explains the timing of governors’ decisions to close businesses and schools, and to impose quarantines. They perform an event history analysis of several social distancing policies implemented in the US states. Their main conclusion is that Republican governors and governors from states with more Trump supporters were slower to adopt social distancing policies. As delays in the adoption of such measures are likely to trigger serious adverse public health outcomes, this result is important. However, as their analysis does not include the legal constraints of BBRs, it is important to examine if budgetary considerations may have affected the governors’ decisions. The second article is by Baccini and Brodeur (2020), who show that Republican governors were also less likely to implement a stay-at-home order. They also focus on the term limits that some governors may face, and reveal that governors without a term limit were significantly quicker to adopt statewide orders than those with a term limit. However, in their estimates, they too do not control for the presence of BBRs. In view of their importance in the previous crisis (the Great Recession) and of the size of the fiscal adjustment induced by social distancing measures, it is important to complement their analysis.

Our results reveal that both partisanship and fiscal institutions have played a role in the adoption of social distancing measures. However, it appears that fiscal rules may have induced a trade-off between health and the economy, as well as some pro-cyclical behaviours. In other words, we show that budgetary constraints have been critical in responding to the pandemic.

We present the literature on the cost of the pandemic and on the measures taken to address it, as well as the relation between the latter and the fiscal situation of the states. We then turn to the data. After that, we discuss the results on the number of measures adopted and on the timing of their adoption. The final section concludes.

Literature review: optimal policies to fight a pandemic and their real impacts

The COVID-19 epidemic has spurred an intense effort by researchers around the world, and not only in virology. Concerning our question, we classify it in two strands: one related to the theoretical optimal policy design to deal with the disease; and one measuring the transmission mechanisms, as well as the consequences of the implemented policies. We also add the literature on fiscal institutions (that is, BBRs and RDFs).

In the literature concerned with optimal policy design, Barnett et al (2020) and Kempf (2020) provide frameworks where the uncertainties related to the health impact are embedded, showing how the optimal mitigation response depends on the fatality rate and reproduction rate of the disease, as well as the response by policymakers confronted with polarised populations, as is the case in the US. Kempf’s (2020) analysis helps to understand the delays in response across US states as policymakers must weigh the health benefits of, say, quarantine measures against the economic damages they inflict. Nevertheless, an optimal response to uncertainty should lead to harsher policy measures in order to reduce the cost of underestimating the dangers of the disease, at greater economic cost (Barnett et al, 2020). Such an analysis can be backed by the computation of the shadow cost of infection risks (Collard et al, 2020), which lies at the basis of the trade-off between health- and economic-related costs.

Lockdowns, quarantines and social distancing measures have been part of the arsenal deployed by policymakers to fight the pandemic. Although they are probably the closest policy to the optimal one (Piguillem and Shi, 2020), lockdowns have led to large economic costs, causing a furore among sceptical politicians (see the previous discussion and the Trump quotation). Alvarez et al (2020) analyse the optimal lockdown policy, in terms of intensity and duration, and show how much it depends on the proportion of infected and susceptible people in the population, as well as the extent of testing. A by-product of their analysis is that, under their calibration, lockdowns represent only 25 per cent to 30 per cent of the welfare cost of the disease – thus appearing as a necessary ill, rather than a remedy worse than the cure. Jarosch et al (2020) reveal that the optimal policy should be deployed fast and, though it should not be a complete lockdown, should involve social distancing for a long period. The length of the lockdown is also a focus of the study by Lee et al (2020), who show the risks of an early lifting of the lockdown. In an analysis related to that of Alvarez et al (2020); Gonzales-Eiras and Niepelt (2020) put figures on the optimal lockdown for the US, in the form of economic activity reduced ‘by two thirds for about 50 days’, which would amount to a deep recession, with a 9.5 per cent GDP loss and the implied increase in unemployment.4

Guerrieri et al (2020) wonder if the epidemic is a supply or demand shock, and show that in a model with incomplete markets and liquidity-constrained consumers, ‘a 50% shock that hits all sectors is not the same as a 100% shock that hits half the economy’, demonstrating that the shock will have the properties of a supply shock in such a framework. This thus reduces the relevance of fiscal stimuli, except that full insurance payments to workers will retain their desired impact.5 Mitman and Rabinovich (2020) also find that a large unemployment-related transfer is optimal, at least as a first policy reaction, to compensate for the shock.

All this literature points to COVID-19 leading to large economic costs, and points to the policy measures that should be implemented. Offsetting the induced costs would require large fiscal measures, financed by debt, thereby obviating the respect of any BBR. Compared with these theoretical recommendations, how have the real measures fared?

This second strand of the literature can itself be separated into two: ‘Why would such policies be efficient?’; and ‘How efficient are the policy measures?’ On the why side, a strong mechanism seems to be information, as evidenced by Gupta et al (2020). Moreover, the fact that the removal of a policy does not induce a relapse, as shown in the case of the repeal of the governor’s order by the Supreme Court in Wisconsin (analysed by Dave et al [2020a]), tends to support this information channel. However, the same information can be processed differently, and there may be a feedback loop between the underlying health condition of an agent and the response to the disease (Velasco and Chang, 2020). Barrios and Hochberg (2020) and Wright et al (2020) show that partisanship, as much as income, is a predictor of compliance with the quarantine policies. This reveals that to be effective, a politician’s decision will need obedient people. While Wright et al (2020) or Fan et al (2020) show that such behaviour may vary along party lines, Gitmez et al (2020), taking this feature into account, show that a person’s behaviour in a pandemic context is an externality on any other’s. Agents thus need public information to be biased (in some ways, overestimating the danger) to correct for the externality, and for information to influence behavioural responses. This theoretical result, however, does not include the possibility that some partisan voters may actually hold virus-related information in disdain. Barrios and Hochberg (2020) indicate that such disdain characterises Trump voters, while Allcott et al (2020) show that agents living in Republican areas adopt less social distancing.

On the how side, Friedson et al (2020) look at California’s ‘shelter-in-place order’ (also termed ‘confinement’ or ‘quarantine’, de facto implying a lockdown as workers should refrain from going to their jobs). They reveal that, being the first state to adopt such a policy, the prevalence of the disease was reduced and many deaths were avoided, at an induced cost equal to 400 job losses per life saved. Dave et al (2020b) look at the impact of the same policy for all the states that have implemented it.6 They confirm the beneficial impact of the lockdown in terms of avoided deaths and reduced prevalence of the disease, though ‘early adopters and high population density states appear to reap larger benefits’, a conclusion shared by Desmet and Warzciag (2020).

Workers will be affected differently by the types of policy measures implemented. Mongey et al (2020) describe those most susceptible to being affected as being ‘in low-work-from-home or high-physical-proximity jobs’. These are less-educated workers (as also established by Aum et al [2020]), who have a lower income on average, have less liquid assets and are more likely to be renting their housing.7 These categories of workers experienced greater declines in their employment levels during the lockdown period, if only due to lower spending by high-income individuals (Chetty et al, 2020). The increase in unemployment that could be expected from the lockdown and social distancing measures quickly became evident, jumping to a high of 15 per cent in the US (starting from a very low level before the crisis), and even to 26.5 per cent according to some estimates (Couch et al, 2020). The increase is unprecedented, as well as the record high level. Yet, the upsurge was not uniformly distributed as black people and Latinos suffered even more (Couch et al, 2020).

Even if the lockdown and other measures can be considered as responsible for the job losses, Aum et al (2020) show that around half of them would have been incurred anyway, if only due to reduced hiring by the sectors most affected, or by the increased uncertainty that precludes new investments. One mechanism is that business owners have seen their numbers reduced by almost a quarter (and by 41 per cent for African-American ones) across almost all sectors and industries (Fairlie, 2020), even though small businesses in more affluent ZIP codes appear to have borne a more than proportional share of the brunt of the adjustment (Chetty et al, 2020).

All in all, the literature surveyed points to heavy costs of the pandemic, to partisan degrees of recognition of the severity of the crisis and to the importance of the measures implemented to address it. Sauvagnat et al (2020) estimate that, by May 2020, state-mandated business closures might have cost more than 3 per cent of 2019 US GDP and saved 1 per cent of the US population. Some of the huge costs generated by the pandemic are to be found on the fiscal side. As of July 2020, the federal government has accumulated a US$2.7 trillion deficit (representing more than 10 per cent of GDP), and is considering the adoption of a new coronavirus-relief bill. Our own analysis aims at understanding how policymakers in the US states have faced the crisis.

The third literature we rely on describes how fiscal institutions (that is, BBRs and rules surrounding the use of budget stabilisation funds, or RDFs, can constrain policymakers. The literature on these fiscal institutions has shown that they are, in fact, complementary mechanisms, whose objective is the control of debt. Battaglini and Coate (2008) recall that Barro’s (1979) fiscal smoothing argument relies on the assumption that governments are benevolent. In this model, public spending has to fluctuate over time, with budget surpluses and deficits being used as a buffer to prevent tax rates from changing too rapidly and abruptly (Battaglini and Coate, 2008).

However, when the government is not benevolent, which can happen if politicians have either a partisan bias (Hibbs, 1977) or an opportunistic tendency (Nordhaus, 1975; Rogoff and Sibert, 1988; Rogoff, 1990), these fluctuations may not be random or optimal. In such cases, decision-makers are subject to debt and deficit biases, and indebtedness can increase without being checked. While in Barro’s (1979) model, the benevolent planner makes decisions and creates equitable transfers between citizens, in the model of Battaglini and Coate (2008), the governing body is biased towards patronage and spending inflation (in an archetypal ‘tragedy of the commons’ issue). Based on this, it can then be shown that the political bias leads to distortions in taxes (proportional to the candidates’ winning margin), to levels of public goods that are inferior to the optimal level and to extremely high levels of debt compared to optimal levels (see, for example, Angeletos et al, 2016).

The public finance problems arising from high debt are essentially twofold: (1) an increased risk of default, with the resulting financing difficulties; and (2) the reduction in the government’s leeway associated with the size of the debt service (Ball et al, 1998). For the US states, two instruments have therefore been identified to control the level of debt. Fiscal rules are the principal one. Although their origin can be traced back to the period during which the US states wrote their constitution, they have been enforced from the 1980/1990s to the present day in many more countries, and both at the national and sub-national levels. Their spread has been so large that, according to Asatryan et al (2018: 107), it can be said that ‘one of the main policy measures to prevent governments from running persistent deficits and to ensure the long-term sustainability of public finances, and thus the level of debt, has been the use of fiscal rules’. To achieve such debt targets, fiscal rules will not only control the debt, but also impose constraints on the components of the budget (Fernández and Parro, 2019).

Hou and Smith (2006) provide a synthesis of the rules present in the budget process in the US states, and discuss the various indicators available in the US sub-national case, where the rules are deemed binding. While the rules appear to meet their objectives in the US states, they have also been accused of inducing pro-cyclical variations of the budget (a view questioned by Clemens and Miran [2012]).

The second instrument is much more specific to the US states, and has been more recently designed: budget stabilisation funds (or RDFs). RDFs are designed to cover revenue shortfalls and respond to unforeseen events by setting aside money for general purposes (Pew Charitable Trusts, 2014). Their purpose is thus clearly counter-cyclical as they are meant to smooth budgets over multiple years and across different phases of the business cycle; however, their operations are also governed by more or less restrictive rules (Pew Charitable Trusts, 2014; 2017). This instrument would reduce the potential pro-cyclical effects generated by balanced-budget requirements.

In the context of the unprecedented health crisis that began in January 2020, at what is, in fact, the middle of the fiscal year for most US states, these two instruments (balanced-budget requirements and budget stabilisation fund rules) must be considered as they have probably influenced both the speed with which health measures have been announced and the number of measures announced. A first reason (as explained, for example, by Bohn and Inman [1996]) is that the budgetary process is modified by the presence of such rules. In the face of the pandemic, when preparing the budget for the next fiscal year, governors in US states can only have been confronted with the squaring exercise of preparing a budget that could only be expansionary (due to the fall in fiscal revenues and potential expenditures associated with the health crisis), while having to respect their state’s balanced-budget requirement.

A second reason is that the rules governing the operation of RDFs can only have played a significant role in allowing or forbidding the use of these funds to cushion the unexpected coronavirus exogenous shock to government expenditure and revenue. Given that, contrary to a standard recession, it is the announced health-related measures that are likely to create the fiscal shock, have the funds played their counter-cyclical role or have they reinforced the shock?

Data

We build our analysis on three sets of variables of interest, plus a set of control variables.8 First, we study states’ social distancing measures and prevalence of COVID-19 cases. The policy measures are examined along eight dimensions, announced over the period from the first reported case of transmission in the US in January 2020 up to 7 April 2020. Sources of data are Fullman et al (2020) and the Center for Systems Science and Engineering at Johns Hopkins University.9 The policy measures considered are gatherings restrictions, school closures, restaurant restrictions, non-essential and other business closures, stay-at-home orders, travel restrictions and curfews. As can be seen from Tables 1a and 1b, we compute the number of policy measures taken by each state and by each of the census regions, as well as the number of days between the appearance of the first COVID-19 case and the announcement of each measure. We also include the number of cases in each state and in each region.

Table 1a:

Descriptive statistics: US states’ social distancing measures

I: 7 April 2020, based on eight policy measuresObsMeanStd DevMinMax
Number of measures announced505.020.912.007.00
Log_number of measures announced501.590.210.691.95
Policy 1: Gatherings restrictions501.000.001.001.00
Policy 2: School closures500.960.200.001.00
Policy 3: Restaurant restrictions500.940.240.001.00
Policy 4: Non-essential business closures500.920.270.001.00
Policy 5: Stay at home500.880.330.001.00
Policy 6: Quarantine500.240.430.001.00
Policy 7: State curfew500.040.200.001.00
Policy 8: Travel restrictions500.040.200.001.00
II: Average number of measures adopted by other states in the region on the day of the announcement of the last measure by state i
Region average number of measuresObsMeanStd DevMinMax
505.030.563.505.67
III: Delay in announcing a policy in days from the date of first COVID-19 case in the state
Variables: _TimeAfter1stCase_ policy pObsMeanStd DevMinMax
–Policy 1: Gatherings restrictions5011.1813.07–6.0053.00
–Policy 2: School closures4811.5412.70–5.0051.00
–Policy 3: Restaurant restrictions4713.9113.30–1.0054.00
–Policy 4: Non-essential business closures4615.6514.18–1.0056.00
–Policy 5: Stay at home4421.9314.175.0064.00
IV: Yr = 1 if the state has a length of time for announcing the adoption of a measure inferior to the average observed by the other states in the region; 0 otherwise
Variables ‘Policy p_Yr’ObsMeanStd DevMinMax
–Policy 1: Gatherings restrictions500.660.480.001.00
–Policy 2: School closures500.620.490.001.00
–Policy 3: Restaurant restrictions500.580.500.001.00
–Policy 4: Non-essential business closures500.540.500.001.00
–Policy 5: Stay at home500.500.510.001.00
V: Share of other states in the region having adopted Policy p on the day of the announcement by state i
Variables ‘Share of states in region of state i with Policy p adopted’ObsMeanStd DevMinMax
–Policy 1: Gatherings restrictions500.440.350.001.00
–Policy 2: School closures500.440.370.001.00
–Policy 3: Restaurant restrictions500.320.330.001.00
–Policy 4: Non-essential business closures500.410.360.001.00
–Policy 5: Stay at home500.450.330.001.00
Table 1b:

Descriptive statistics: US states’ COVID-19 cases and policy measures

I: Number of COVID-19 cases in the state at the announcement of the last decided measure
State i COVID-19 casesObsMeanStd DevMinMax
50941.941547.1319.007954.00
II: Number of COVID-19 cases in other states in the region when the last measure is announced
State i’s region COVID-19 casesObsMeanStd DevMinMax
508513.6613073.23126.0083871.00
III: Number of COVID-19 cases in the state at the announcement of Policy p by state i
Variables ‘State i COVID-19 cases Policy pObsMeanStd. Dev.MinMax
–Policy 1: Gatherings restrictions5072.54132.580.00727.00
–Policy 2: School closures4898.38190.260.00967.00
–Policy 3: Restaurant restrictions47113.91187.830.00967.00
–Policy 4: Non-essential business closures46210.13299.530.001083.00
–Policy 5: Stay at home44974.141605.6211.007954.00
IV: Number of COVID-19 cases in the state at the announcement of Policy p by state i or at 7 April if the state has not announced the measure
Variable ‘State i COVID-19 cases Policy p’ BISObsMeanStd DevMinMax
–Policy 1: Gatherings restrictions5072.54132.580.00727.00
–Policy 2: School closures50460.562441.790.0017309.00
–Policy 3: Restaurant restrictions50298.381179.370.008333.00
–Policy 4: Non-essential business closures50261.80373.150.001746.00
–Policy 5: Stay at home50920.581515.5411.007954.00
V: Number of COVID-19 cases in other states of the region at announcement of Policy p by state i
Variables ‘State i’s region COVID-19 cases Policy pObsMeanStd DevMinMax
–Policy 1: Gatherings restrictions50777.981763.348.009700.00
–Policy 2: School closures48827.001843.8051.0010281.00
–Policy 3: Restaurant restrictions47844.601043.2140.006088.00
–Policy 4: Non-essential business closures462292.804858.1340.0023731.00
–Policy 5: Stay at home447820.6413440.50310.0083871.00
VI: Number of COVID-19 cases in other states of the region at announcement of Policy p by state i or at 7 April if the state has not announced the measure
Variable ‘State i’s region COVID-19 cases Policy p’ BISObsMeanStd. Dev.MinMax
–Policy 1: Gatherings restrictions50777.981763.348.009700.00
–Policy 2: School closures502172.568305.1251.0057350.00
–Policy 3: Restaurant restrictions502055.687084.5340.0050014.00
–Policy 4: Non-essential business closures502642.644807.9240.0023731.00
–Policy 5: Stay at home508730.5614424.55310.0083871.00

Second, we include BBRs and information on the states’ budget stabilisation funds. BBRs constitute a system of legal provisions and requirements covering the states’ budget process. Some of the provisions are embedded in the states’ constitutions; others are part of lower-level types of regulation. Budget stabilisation funds, or RDFs, allow states to set aside a surplus for times of unexpected revenue shortfall or budget deficit (Randall and Rueben, 2017). As can be seen from Table 1c, most states have some type of RDF but their relatively recent spread across US states has led to different rules (on how much and when to contribute to the RDF, whether it should be capped, and, importantly in our context, under what conditions the funds can be spent).

Table 1c:

Descriptive statistics: fiscal rules and control variables

I: BBRsObsMeanStd DevMinMax
ACIR (1987): degree of stringency508.082.630.0010.00
Hou and Smith’s (2010) classification (T = technical; P = political)
BBR #1: ‘Governor must submit a balanced budget’ (P)500.800.400.001.00
BBR #2: ‘Own-source revenue must match (meet or exceed) expenditures’ (T)500.220.420.001.00
BBR #3: ‘Own-source revenue and general obligation (or unspecified) debt (or debt in anticipation of revenue) must match (meet or exceed) expenditures’ (T)500.720.450.001.00
BBR #4: ‘Legislature must pass a balanced budget’ (P)500.720.450.001.00
BBR #5: ‘A limit is in place on the amount of debt that may be assumed for the purpose of deficit reduction’ (T)500.420.500.001.00
BBR #6: ‘Governor must sign a balanced budget’ (P)500.040.200.001.00
BBR #7: ‘Controls are in place on supplementary appropriations’ (T)500.380.490.001.00
BBR #9: ‘No deficit may be carried over to the next fiscal year (or biennium)’ (T)500.140.350.001.00
II: Budget stabilisation fundsObsMeanStd DevMinMax
Budget stabilisation withdrawal conditions (Pew Charitable Trusts, 2017): no fund500.060.240.001.00
RDF restrictive rules500.320.470.001.00
RDF soft rules500.780.420.001.00
RDF both types of rules500.160.370.001.00
RDF/GDP500.00510.00820.000.0420
III: Political and economic variablesObsMeanStd DevMinMax
Republican governor500.520.500.001.00
Trump voters (%, 2016)500.490.100.290.69
Polarisation index (2016)502459.681209.071659.8110000.00
Log(GDP per capita)5011.00280.187210.594411.3967
Expenditure forecast/GDP50.0469586.015388.0199704.1020465
Revenue forecast/GDP50.0466631.0155593.0195457.102949
Budget balance forecast/GDP50.0019332.00253340.0123652

The standard measure for BBRs is the one built by the US Advisory Commission on Intergovernmental Relations (ACIR, 1987). We use it for comparison with the literature that relies on it. As can be seen in Table 1c, the index reveals a relatively high degree of constraint, with an average score of 8.08/10 for the 50 states. Yet, since its publication in 1987, it has not been updated. Hence, we will also include the classification proposed by Hou and Smith (2010), which we have updated, hand-picking modifications of the fiscal regulations in each state. This classification differentiates between nine types of balanced-budget characteristics, and is based on an analysis that distinguishes between the technical rules (T) and the political ones (P) along the budget process (executive preparation, legislative review and implementation).

Among political rules, two directly target the governor. Table 1c shows that policy rule BBR #1 (‘Governor must submit a balanced budget’) is adopted in 80 per cent of the states, while BBR #6 (‘Governor must sign a balanced budget’) is adopted in only two states (California and Massachusetts). Concerning technical rules, BBR #2 (‘Own-source revenue must match (meet or exceed) expenditures’) is operational in 11 states. The last technical rule is BBR #9 (‘No deficit may be carried over to the next fiscal year (or biennium)’), which concerns seven states.

We also include data from the Pew Charitable Trusts (2017) report on each state’s budget stabilisation fund (or RDF), which can be used as a way to smooth out the negative effects of recessions. Their presence has often been overlooked but we believe may have an impact on policymakers’ reactions. Thus, we first include the fund’s 2019 amount (more precisely, we scale it by each state’s GDP).10 However, the states that have an RDF (the exceptions being Colorado, Illinois and Montana [see Pew Charitable Trusts, 2017]) are confronted with two types of rules in the use of funds: on the one hand, the rules we will call ‘RDF restrictive rules’, where the withdrawal of funds is allowed if the reason is explicitly ‘related to volatility’ (of revenues and/or economic); and, on the other, the rules we will classify as ‘RDF soft rules’, where the reason is not linked to this definition of volatility (but to a forecast error or a budget variance, or even to no conditions). Table 1c shows, in particular, that only 16 states have rules explicitly linked to the restrictive criterion, including eight strictly. As several states hold different types of RDFs, we also include information on the differences between rules, if they diverge: the variable ‘RDF both kind of rules’ reveals this to be the case for 16 per cent of the states, for which one fund can have strict rules, while another fund has laxer ones.

Third, we include political variables: a dummy signalling a Republican governor in state i; the percentage of Trump voters in the 2016 presidential election; and a measure of opinion polarisation in each state.11 The latter is built from the American National Election Study (ANES), by considering a Herfindahl-Hirschman (HH) index of the shares of respondents declaring themselves conservative, liberal or moderate.12 This accounts for the possibility that Americans are increasingly divided along moral or economic issues, as confirmed by, for example, Baldassari and Park (2020) and Barrios and Hochberg (2020).

Finally, we include GDP per capita (from the Bureau of Economic Analysis [BEA]) as a catch-all control variable. Given the disparities among US states, we take the log of this variable.13 We also include three variables that could influence a governor’s behaviour: the budget balance forecast; the revenue forecast; and the expenditure forecast (obviously, we do not include simultaneously expenditure and revenue forecasts as they are correlated). These will control for the impact that the expected budget for the current year (that is, before the crisis started) might have as states that have approved more balanced or prudent budgets should have more fiscal space to withstand the consequences of a lockdown or a temporary freezing of economic activities.

Results on the adoption of social distancing measures

Table 2 displays the results of our analysis on the determinants of the number of social distancing measures announced by US states. The estimated equation is:
M1

where i = 1, …, 50 states, is a constant, an error term, BBR is a set of fiscal rule variables, RDF is a set of RDF variables, POLITICAL is a set of political variables and X is a set of control variables (namely, LogGDP, Region average number of measures, State i COVID-19 cases, and State i’s region COVID-19 cases). The estimation technique we use is the Log-Ordinary Least Squares (OLS). Given the count nature of the dependent variable, an expected option would have been to use Poisson or negative binomial models. However, the conditions for using a Poisson model are that the considered events should occur randomly over a fixed period of time, which is not met in our context. More importantly, the probability of occurrence should be very small while the number of incidences should be very large. This limiting condition of Poisson is not fulfilled in our context, and implementing a Poisson procedure would deliver unreliable estimates. Moreover, the Poisson regression is estimated by maximum likelihood estimation, and thus usually requires a large sample size (see Hutchinson and Holtman, 2005), which is another condition not met in our case. A negative binomial model would run into the same issues, plus the fact that the distribution of our dependent variable is not over-dispersed. Finally, we have performed the skewness-kurtosis (sk) test on the original variable, and the distribution is close to a normal one. These reasons have led us to use the Log-OLS procedure. As we have no zeros in the dependent variable, we do not lose data due to undefined values, and we keep the benefits of simple OLS while considering the specificities of the dependent variable. (Moreover, the normality test is even improved.)

Tables 2a and 2b display our results for this first regression. We find that the percentage of Trump voters in the 2016 presidential election tends to reduce the number of policy measures taken in US states to face the epidemic. However, neither this variable nor the one indicating the presence of a Republican governor, nor their interaction, appear as robustly significant determinants. Also, the degree of political polarisation in the state is not significant. Hence, contrary to what the literature suggests, partisan considerations do not appear as a strong determinant of the adoption of social distancing measures in the US states.

Table 2a:

Determinants of the number of social distancing measures announced by US states (7 April, based on eight possible measures)

12345678910
BBRs stringency (ACIR)–0.0166–0.00431–0.00221–0.0104–0.0122–0.0145–0.0134–0.0157–0.0158–0.0168
(–1.44)(–0.35)(–0.16)(–0.82)(–1.08)(–1.23)(–1.19)(–1.33)(–1.21)(–1.26)
Trump voters (%, 2016)–0.608–0.451–0.392–0.131–0.0625–0.0923–0.0251–0.0339–0.00748
(–1.54)(–0.78)(–0.94)(–0.34)(–0.16)(–0.24)(–0.06)(–0.06)(–0.01)
Republican governor–0.04630.0811–0.0287–0.0262–0.0193–0.0260–0.0190–0.0265–0.0388
(–0.76)(0.24)(–0.47)(–0.48)(–0.34)(–0.47)(–0.34)(–0.08)(–0.12)
Republican governor * Trump voters–0.2650.01560.0438
(–0.38)(0.02)(0.07)
Polarisation index–0.000093
(–0.89)
Polarisation index squared6.99e–09
(0.72)
RDF/GDP9.795**9.952**8.740**4.3874.2754.1764.0654.0504.148
(2.45)(2.45)(2.19)(1.11)(1.07)(1.05)(1.02)(0.99)(0.84)
RDF restrictive rules0.01750.0209–0.00402–0.0244–0.0316–0.0289–0.0359–0.0362–0.0285
(0.25)(0.29)(–0.06)(–0.38)(–0.48)(–0.45)(–0.55)(–0.54)(–0.41)
RDF soft rules–0.0347–0.0339–0.0657–0.0554–0.0686–0.0578–0.0709–0.0710–0.0696
(–0.47)(–0.46)(–0.88)(–0.84)(–0.99)(–0.87)(–1.03)(–1.02)(–0.94)
Log(GDP per capita)–0.329–0.317–0.279–0.366*–0.340*–0.373*–0.347*–0.347*–0.307
(–1.57)(–1.48)(–1.34)(–1.94)(–1.76)(–1.98)(–1.80)(–1.76)(–1.49)
Region average number of measures0.05820.06340.07690.07860.08600.07950.08690.08660.0791
(0.95)(0.99)(1.24)(1.41)(1.51)(1.43)(1.53)(1.48)(1.28)
Log(1 + state i COVID-19 cases)0.0412*0.04020.0464*0.0686***0.0691***0.0691***0.0697***0.0698***0.0568**
(1.71)(1.65)(1.94)(2.95)(2.95)(2.98)(2.98)(2.92)(2.03)
Log(1 + state i’s region COVID-19 cases)0.02710.02520.0249–0.0003250.000584–0.00285–0.00182–0.001740.00639
(1.05)(0.94)(0.98)(–0.01)(0.02)(–0.11)(–0.07)(–0.07)(0.23)
End balance forecast 2020/GDP18.178.5838.6548.6498.302
(1.51)(0.74)(0.75)(0.74)(0.68)
Revenue forecast 2020/GDP6.789***6.305***
(3.21)(2.83)
Expenditure forecast 2020/GDP7.091***6.597***6.605***6.384**
(3.25)(2.88)(2.82)(2.45)
Constant1.728***4.777*4.559*4.0624.707**4.375*4.779**4.438*4.451*4.228*
(17.67)(1.88)(1.73)(1.60)(2.06)(1.87)(2.10)(1.90)(1.84)(1.69)
Observations50505050505050505050
R-squared0.0410.4110.4140.4450.5370.5440.5400.5470.5470.559
Adjusted R-squared0.0210.2600.2440.2840.4030.3960.4060.4000.3830.364

Notes: t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

Table 2b:

Determinants of the number of social distancing measures announced by US states (7 April, based on eight possible measures)

12345678910
BBR #6 (Hou and Smith)–0.0227–0.180–0.180–0.189–0.282*–0.281*–0.277*–0.277*–0.279*–0.286*
(–0.15)(–1.19)(–1.17)(–1.25)(–1.99)(–1.96)(–1.97)(–1.94)(–1.93)(–1.92)
BBR #7 (Hou and Smith)0.144**0.136**0.135**0.123**0.08960.08780.0930*0.09140.08750.0898
(2.39)(2.32)(2.22)(2.03)(1.62)(1.55)(1.69)(1.62)(1.48)(1.49)
Trump voters (%, 2016)–0.808**–0.785–0.703*–0.436–0.422–0.417–0.405–0.303–0.262
(–2.17)(–1.41)(–1.80)(–1.21)(–1.13)(–1.15)(–1.09)(–0.55)(–0.46)
Republican governor–0.0376–0.0208–0.0249–0.00844–0.00646–0.00867–0.006910.06440.0777
(–0.65)(–0.07)(–0.42)(–0.16)(–0.12)(–0.16)(–0.13)(0.23)(0.26)
Republican governor * Trump voters–0.0351–0.148–0.168
(–0.06)(–0.25)(–0.28)
Polarisation index–0.0000709
(–0.71)
Polarisation index squared4.28e–09
(0.46)
RDF/GDP11.38***11.39***10.73***6.1466.1206.1236.1026.0967.139
(3.00)(2.96)(2.79)(1.59)(1.57)(1.59)(1.56)(1.54)(1.48)
RDF restrictive rules0.04310.04340.03330.01940.01790.01620.01490.01590.0271
(0.63)(0.63)(0.48)(0.31)(0.28)(0.26)(0.23)(0.25)(0.40)
RDF soft rules0.02000.01980.002040.007260.003900.006370.003380.00179–0.00137
(0.28)(0.27)(0.03)(0.11)(0.06)(0.10)(0.05)(0.03)(–0.02)
Log(GDP per capita)–0.403*–0.402*–0.351*–0.364*–0.354*–0.370**–0.361*–0.353*–0.309
(–2.02)(–1.97)(–1.69)(–2.00)(–1.86)(–2.04)(–1.89)(–1.81)(–1.51)
Region average number of measures0.02940.03010.04770.07160.07440.07170.07420.07790.0662
(0.50)(0.49)(0.77)(1.29)(1.28)(1.29)(1.28)(1.29)(1.04)
Log(1 + state i COVID-19 cases)0.0621**0.0618**0.0645**0.0859***0.0859***0.0863***0.0863***0.0853***0.0749**
(2.60)(2.51)(2.69)(3.73)(3.68)(3.74)(3.69)(3.55)(2.68)
Log(1 + state i’s region COVID-19 cases)0.007170.007030.00650–0.0171–0.0168–0.0193–0.0189–0.0196–0.0104
(0.28)(0.27)(0.26)(–0.70)(–0.68)(–0.78)(–0.76)(–0.77)(–0.38)
End balance forecast 2020/GDP10.712.1521.9272.4090.763
(0.95)(0.20)(0.18)(0.22)(0.06)
Revenue forecast 2020/GDP6.351***6.224***
(3.01)(2.79)
Expenditure forecast 2020/GDP6.485***6.368***6.369***6.473**
(3.02)(2.80)(2.77)(2.52)
Constant1.540***5.741**5.716**5.007*4.725**4.598*4.790**4.675*4.536*4.202
(41.22)(2.38)(2.30)(1.98)(2.13)(1.97)(2.17)(2.01)(1.87)(1.69)
Observations50505050505050505050
R-squared0.1090.4940.4940.5060.5930.5940.5940.5950.5950.607
Adjusted R-squared0.0710.3470.3300.3460.4610.4470.4630.4480.4330.416

Notes: t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

Much more significant are the results related to BBRs and RDFs. Although the ACIR index of the stringency of budget rules is never significant (see Table 2a), some rules definitely are, according to the Hou and Smith (2010) classification scheme. As shown in Table 2b, BBR numbers 6 and, especially, 7 are decisive. BBR number 7, in particular, has led to a higher number of policy measures being adopted. This rule is a technical one and indicates that ‘Controls are in place on supplementary appropriations’ – and supplementary appropriations are exactly what could have been needed to face the consequences of stay-at-home orders, as well as of business closures. Hence, it appears that since the negative economic consequences of the lockdown measures were expected, as long as some legal controls were in place, governors have relied on policy measures, anticipating that the level of deficit would be reined in by the controls.14 This interpretation is reinforced by the fact that when we include the expected expenditure or the expected revenue, the rules lose their degree of significance, to the benefit of the fiscal variables. Higher expected revenues (or expenditures, as both are strongly correlated) have tended to increase the number of social distancing measures that the US states have announced.

The amount of funds in the RDFs is also positive and significant, with a large coefficient, indicating that governors anticipated that RDFs could be used to smooth out the consequences of social distancing and lockdowns. Here again, the inclusion of variables related to the budget forecast tends to reduce the significance of the RDFs, which indicates that these amounts may be considered when preparing the budget, and that this has been the case in the face of the epidemic.

Finally, wealthier states (in terms of GDP per capita) may have been more reluctant to adopt a higher number of policy measures against the epidemic. Although the coefficient is barely significant, this may be related to the fact that Democratic states are more usually urban and wealthier, as compared to Republican states. In addition, the number of measures adopted in the region to which the state belongs has a positive, though not significant, impact. Moreover, the number of declared COVID-19 cases in a state has a positive and significant impact on the adoption of a larger number of measures.

Results on the timing of adoption of social distancing measures

In Tables 3a3e, we look at the length of time (measured by the number of days between the first COVID-19 case declared in the state and the adoption of the policy by the same state) it took to adopt each type of social distancing policy measure.15 We look separately at each policy measure because each kind of policy under analysis may have a different effect on government finance. For instance, closing schools does not directly reduce revenues (though it might reduce expenditure), while closing restaurants or other business activities can have a more direct effect on government finance, if only in terms of lost tax revenues.

Table 3a:

Determinants of length of announcement of gatherings restrictions (7 April)

12345678
Republican governor3.9943.4743.4833.5453.327–10.09–7.796–9.476
(1.17)(1.01)(1.03)(1.04)(0.96)(–0.59)(–0.43)(–0.51)
Trump voters (%, 2016)–24.16–27.04–30.86–31.14–31.65–49.65–44.76–45.32
(–1.30)(–1.44)(–1.63)(–1.63)(–1.64)(–1.67)(–1.41)(–1.41)
LogGDP (per capita)3.6781.8233.3863.3992.4180.9162.8603.842
(0.36)(0.17)(0.33)(0.33)(0.23)(0.09)(0.25)(0.32)
RDF/GDP–117.7–79.0715.309.35710.98–15.6429.43–4.286
(–0.60)(–0.40)(0.07)(0.04)(0.05)(–0.07)(0.12)(–0.02)
Log(1 + state i COVID-19 cases)3.596***3.590***3.175***3.186***3.247***3.231**3.065**2.773*
(3.15)(3.14)(2.72)(2.72)(2.73)(2.70)(2.43)(1.95)
Log(1 + state i’s region COVID-19 cases)–0.535–0.612–0.123–0.116–0.223–0.546–0.241–0.0983
(–0.41)(–0.47)(–0.09)(–0.09)(–0.16)(–0.38)(–0.15)(–0.06)
Share of states in region of state i with Policy p announced–1.360–1.835–2.423–2.569–2.643–0.529–0.898–0.740
(–0.18)(–0.24)(–0.32)(–0.34)(–0.34)(–0.06)(–0.11)(–0.09)
BBR #6 (Hou and Smith)23.63***23.87***26.23***25.99***25.75***25.85***25.48***26.08***
(2.82)(2.84)(3.09)(3.06)(3.00)(2.99)(2.91)(2.91)
RDF restrictive rules–1.648–1.463–1.646–1.585–1.492–1.533–1.396–1.427
(–0.44)(–0.39)(–0.45)(–0.43)(–0.40)(–0.41)(–0.37)(–0.37)
RDF soft rules–10.34**–9.762**–10.70**–10.66**–10.30**–9.828**–10.33**–10.12**
(–2.55)(–2.38)(–2.66)(–2.65)(–2.49)(–2.34)(–2.36)(–2.28)
End balance forecast 2020/GDP–551.7–304.6–384.0–425.4–363.6
(–0.96)(–0.49)(–0.61)(–0.66)(–0.55)
Revenue forecast 2020/GDP–156.4
(–1.40)
Expenditure forecast 2020/GDP–152.4–128.7–115.6–96.31–118.5
(–1.33)(–1.03)(–0.91)(–0.72)(–0.82)
Republican governor * Trump voters27.4823.2626.60
(0.80)(0.65)(0.72)
Polarisation index–0.000895–0.00353
(–0.47)(–0.59)
Polarisation index squared0.000000254
(0.46)
Constant–19.193.849–6.172–6.3744.35729.446.1381.135
(–0.16)(0.03)(–0.05)(–0.05)(0.04)(0.23)(0.05)(0.01)
Observations5050505050505050
R-squared0.5820.5920.6020.6000.6030.6100.6120.615
Adjusted R-squared0.4740.4730.4870.4850.4740.4690.4570.445

Notes: Policy p is where p refers to the same category of policy; t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

Table 3b:

Determinants of length of announcement of school closures (7 April)

12345678
Republican governor7.702**8.386**7.888**7.928**8.374**–12.15–11.82–8.648
(2.55)(2.65)(2.57)(2.59)(2.61)(–0.70)(–0.65)(–0.45)
Trump voters (%, 2016)–40.23**–37.77*–36.03*–34.96*–35.11*–66.55*–65.78*–62.31*
(–2.18)(–2.01)(–1.78)(–1.72)(–1.71)(–2.00)(–1.88)(–1.74)
Log(GDP per capita)0.4031.253–0.170–0.2930.647–1.180–0.847–2.410
(0.04)(0.13)(–0.02)(–0.03)(0.06)(–0.12)(–0.08)(–0.21)
RDF/GDP–3.846–21.56–48.75–58.15–50.63–72.79–65.33–26.79
(–0.02)(–0.11)(–0.23)(–0.28)(–0.24)(–0.35)(–0.28)(–0.11)
Log(1 + state i COVID-19 cases)5.479***5.699***5.757***5.814***5.854***5.973***5.918***6.285***
(4.23)(4.27)(4.09)(4.14)(4.12)(4.22)(3.75)(3.68)
Log(1 + state i’s region COVID-19 cases)–2.858**–2.769**–3.064**–3.136**–2.957*–3.297**–3.246*–3.422**
(–2.12)(–2.04)(–2.17)(–2.20)(–2.00)(–2.20)(–1.99)(–2.05)
Share of states in region of state i with Policy p announced2.5991.5851.9941.9641.420–0.005650.00716–0.153
(0.31)(0.19)(0.24)(0.23)(0.17)(–0.00)(0.00)(–0.02)
BBR #2 (Hou and Smith)–10.49***–11.47***–10.79***–10.86***–11.48***–12.12***–12.04***–12.43***
(–3.08)(–3.14)(–3.10)(–3.12)(–3.10)(–3.26)(–3.10)(–3.13)
RDF restrictive rules–1.084–1.452–1.328–1.410–1.570–1.695–1.658–1.676
(–0.33)(–0.44)(–0.40)(–0.42)(–0.47)(–0.51)(–0.48)(–0.48)
RDF soft rules–12.04***–13.03***–12.33***–12.41***–13.04***–12.83***–12.86***–13.52***
(–3.24)(–3.30)(–3.26)(–3.28)(–3.26)(–3.22)(–3.17)(–3.18)
End balance forecast 2020/GDP461.4361.4305.3292.8251.0
(0.77)(0.54)(0.46)(0.42)(0.36)
Revenue forecast 2020/GDP61.42
(0.54)
Expenditure forecast 2020/GDP75.7645.9245.7748.3477.11
(0.65)(0.35)(0.35)(0.36)(0.53)
Republican governor * Trump voters43.6242.9436.68
(1.19)(1.13)(0.92)
Polarisation index–0.0001550.00312
(–0.09)(0.54)
Polarisation index squared–0.000000307
(–0.60)
Constant31.7120.1134.0134.5024.3260.5656.6865.47
(0.29)(0.18)(0.30)(0.31)(0.21)(0.51)(0.44)(0.50)
Observations4848484848484848
R-squared0.6190.6250.6220.6230.6260.6410.6410.645
Adjusted R-squared0.5160.5100.5060.5080.4980.5040.4890.479

Notes: Policy p is where p refers to the same category of policy; t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

Table 3c:

Determinants of length of announcement of restaurant restrictions (7 April)

12345678
Republican governor4.3833.9243.7663.8313.652–21.44–16.11–17.35
(1.25)(1.08)(1.05)(1.07)(0.99)(–1.10)(–0.77)(–0.81)
Trump voters (%, 2016)–20.03–21.58–25.11–25.05–25.20–59.29*–51.31–51.64
(–0.98)(–1.03)(–1.17)(–1.16)(–1.15)(–1.75)(–1.44)(–1.43)
Log(GDP per capita)1.6080.3982.2272.1811.413–3.761–0.4860.719
(0.14)(0.03)(0.19)(0.19)(0.12)(–0.30)(–0.04)(0.05)
RDF/GDP–83.59–66.09–10.80–16.67–16.54–52.8346.7815.85
(–0.39)(–0.30)(–0.05)(–0.07)(–0.07)(–0.22)(0.17)(0.06)
Log(1 + state i COVID-19 cases)4.155***4.002***3.592**3.638**3.626**3.538**3.278**2.897
(3.07)(2.87)(2.38)(2.41)(2.37)(2.33)(2.09)(1.62)
Log(1 + state i’s region COVID-19 cases)0.3330.2600.5730.5830.5050.01120.5100.541
(0.20)(0.16)(0.34)(0.35)(0.29)(0.01)(0.27)(0.28)
Share of states in region of state i with Policy p announced–3.362–3.814–2.723–2.869–3.197–7.412–6.676–6.022
(–0.37)(–0.42)(–0.30)(–0.32)(–0.34)(–0.76)(–0.68)(–0.60)
BBR #6 (Hou and Smith)27.08***27.35***28.76***28.53***28.48***28.82***27.61***27.86***
(3.12)(3.12)(3.22)(3.20)(3.15)(3.22)(3.02)(3.00)
RDF restrictive rules–2.156–1.916–1.925–1.909–1.809–2.132–1.670–1.581
(–0.54)(–0.47)(–0.48)(–0.47)(–0.44)(–0.53)(–0.40)(–0.38)
RDF soft rules–12.44***–11.92**–12.13***–12.13***–11.89**–11.55**–12.20**–11.65**
(–2.89)(–2.69)(–2.80)(–2.80)(–2.66)(–2.60)(–2.68)(–2.45)
End balance forecast 2020/GDP–360.8–204.1–499.0–551.8–485.2
(–0.56)(–0.29)(–0.68)(–0.75)(–0.64)
Revenue forecast 2020/GDP–110.6
(–0.86)
Expenditure forecast 2020/GDP–103.4–88.27–61.13–28.92–54.61
(–0.79)(–0.62)(–0.43)(–0.19)(–0.34)
Republican governor * Trump voters52.8542.4544.37
(1.31)(0.99)(1.01)
Polarisation index–0.00164–0.00476
(–0.75)(–0.66)
Polarisation index squared0.000000289
(0.46)
Constant–4.32811.24–3.333–3.3705.29680.8941.2735.70
(–0.03)(0.08)(–0.03)(–0.03)(0.04)(0.55)(0.26)(0.22)
Observations4747474747474747
R-squared0.5560.5590.5650.5630.5640.5860.5930.596
Adjusted R-squared0.4320.4210.4280.4260.4110.4230.4150.400

Notes: Policy p is where p refers to the same category of policy; t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

Table 3d:

Determinants of length of announcement of non-essential business closures (7 April)

12345678
Republican governor3.5243.4993.4423.4933.482–8.965–5.558–7.569
(1.03)(1.00)(0.98)(1.00)(0.98)(–0.51)(–0.31)(–0.40)
Trump voters (%, 2016)–30.10–30.31–30.95–30.45–30.53–48.45–42.23–42.79
(–1.51)(–1.47)(–1.48)(–1.44)(–1.42)(–1.48)(–1.24)(–1.24)
Log(GDP per capita)–2.821–2.923–2.648–2.755–2.839–4.142–1.717–0.576
(–0.25)(–0.25)(–0.23)(–0.24)(–0.24)(–0.34)(–0.14)(–0.04)
RDF/GDP21.9023.1333.2826.2426.26–0.93897.3961.91
(0.10)(0.11)(0.15)(0.12)(0.11)(–0.00)(0.36)(0.22)
Log(1 + state i COVID-19 cases)4.501***4.489***4.388***4.457***4.458***4.397***4.110**3.700**
(3.67)(3.55)(3.05)(3.09)(3.04)(2.98)(2.68)(2.10)
Log(1 + state i’s region COVID-19 cases)–1.196–1.204–1.102–1.159–1.172–1.212–0.778–0.729
(–0.80)(–0.79)(–0.68)(–0.71)(–0.69)(–0.71)(–0.43)(–0.39)
Share of states in region of state i with Policy p announced9.2679.2659.0389.1759.1918.4837.8008.281
(1.13)(1.11)(1.07)(1.08)(1.06)(0.97)(0.88)(0.92)
BBR #2 (Hou and Smith)–6.505*–6.427–6.310–6.431–6.390–6.100–6.411–6.526
(–1.70)(–1.54)(–1.55)(–1.58)(–1.48)(–1.40)(–1.45)(–1.46)
BBR #6 (Hou and Smith)22.61**22.65**23.05**22.77**22.77**23.47**22.38**22.78**
(2.63)(2.59)(2.52)(2.50)(2.46)(2.50)(2.34)(2.34)
RDF restrictive rules–1.252–1.234–1.239–1.243–1.233–1.445–0.962–0.977
(–0.34)(–0.33)(–0.33)(–0.33)(–0.32)(–0.37)(–0.24)(–0.25)
RDF soft rules–12.16***–12.11***–12.09***–12.13***–12.10***–11.89**–12.53***–11.98**
(–3.01)(–2.84)(–2.92)(–2.93)(–2.80)(–2.72)(–2.80)(–2.56)
End balance forecast 2020/GDP–34.11–23.77–139.6–179.0–111.5
(–0.05)(–0.03)(–0.19)(–0.24)(–0.15)
Revenue forecast 2020/GDP–20.94
(–0.16)
Expenditure forecast 2020/GDP–8.135–6.556–4.22528.421.531
(–0.06)(–0.04)(–0.03)(0.18)(0.01)
Republican governor * Trump voters26.3219.5922.85
(0.73)(0.52)(0.59)
Polarisation index–0.00161–0.00512
(–0.75)(–0.69)
Polarisation index squared0.000000322
(0.50)
Constant55.2456.5854.5354.9855.9679.0150.2946.51
(0.42)(0.42)(0.41)(0.42)(0.41)(0.56)(0.34)(0.31)
Observations4646464646464646
R-squared0.6560.6560.6570.6560.6560.6620.6680.671
Adjusted R-squared0.5450.5310.5320.5310.5170.5100.5030.490

Notes: Policy p is where p refers to the same category of policy; t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

Table 3e:

Determinants of length of announcement of stay-at-home orders (7 April)

12345678
Republican governor4.0914.0454.0164.0984.058–6.247–4.261–7.828
(1.09)(1.04)(1.03)(1.06)(1.02)(–0.33)(–0.22)(–0.39)
Trump voters (%, 2016)–36.14–36.30–36.41–36.10–36.24–48.55–41.93–57.42
(–1.50)(–1.47)(–1.48)(–1.46)(–1.44)(–1.43)(–1.15)(–1.43)
Log(GDP per capita)–5.101–5.180–4.742–5.134–5.257–5.218–1.772–6.902
(–0.38)(–0.38)(–0.34)(–0.36)(–0.36)(–0.36)(–0.11)(–0.40)
RDF/GDP–336.0–334.3–315.8–337.9–338.4–340.4–70.38–681.0
(–1.03)(–1.01)(–0.82)(–0.88)(–0.87)(–0.86)(–0.11)(–0.75)
Log(1 + state i COVID-19 cases)3.427**3.402**3.321*3.437*3.423*3.402*3.0992.581
(2.36)(2.23)(1.84)(1.91)(1.86)(1.82)(1.57)(1.26)
Log(1 + state i’s region COVID-19 cases)–1.914–1.892–1.834–1.922–1.909–1.821–1.364–1.377
(–0.95)(–0.91)(–0.84)(–0.86)(–0.84)(–0.79)(–0.55)(–0.55)
Share of states in region of state i with Policy p announced3.8703.7263.6983.8883.7594.2023.9112.378
(0.38)(0.36)(0.36)(0.37)(0.35)(0.38)(0.35)(0.21)
BBR #2 (Hou and Smith)–7.208*–7.106–7.085–7.219–7.126–6.845–6.752–6.722
(–1.70)(–1.56)(–1.59)(–1.62)(–1.50)(–1.42)(–1.38)(–1.37)
BBR #6 (Hou and Smith)26.04***26.07***26.29**26.02**26.03**26.63**25.91**25.79**
(2.80)(2.76)(2.69)(2.68)(2.63)(2.65)(2.52)(2.51)
RDF restrictive rules–0.979–0.952–0.996–0.979–0.950–1.219–1.183–0.428
(–0.23)(–0.22)(–0.23)(–0.22)(–0.21)(–0.27)(–0.26)(–0.09)
RDF soft rules–11.31**–11.25**–11.28**–11.31**–11.25**–11.18**–11.72**–9.881*
(–2.47)(–2.37)(–2.42)(–2.42)(–2.33)(–2.29)(–2.32)(–1.82)
End balance forecast 2020/GDP–49.50–52.08–104.9–221.995.46
(–0.07)(–0.07)(–0.13)(–0.27)(0.11)
Revenue forecast 2020/GDP–19.15
(–0.10)
Expenditure forecast 2020/GDP1.8414.0261.824–9.548–30.19
(0.01)(0.02)(0.01)(–0.05)(–0.15)
Republican governor * Trump voters21.4917.1124.42
(0.55)(0.42)(0.59)
Polarisation index–0.00173–0.00978
(–0.53)(–1.08)
Polarisation index squared0.000000949
(0.95)
Constant98.4899.4995.4998.75100.1104.666.01147.1
(0.64)(0.63)(0.60)(0.62)(0.62)(0.64)(0.36)(0.73)
Observations4444444444444444
R-squared0.5850.5850.5850.5850.5850.5890.5930.606
Adjusted R-squared0.4420.4240.4240.4240.4050.3910.3750.373

Notes: Policy p is where p refers to the same category of policy; t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

The equation, estimated by standard OLS, is:
M2

where p = 1, …, 5 policies, i = 1, …, 50 states, is a constant, an error term, BBR is a set of fiscal rule variables, RDF is a set of RDF variables, POLITICAL is a set of political variables and X is a set of control variables (namely, LogGDP(per capita), End balance forecast 2020, Revenue forecast 2020, Expenditure forecast 2020, State i COVID-19 cases Policy p, State i’s region COVID-19 cases Policy p, and Share of states in region of state i with Policy p announced).

We discard the ACIR stringency index, which was never significant in the previous analysis, and focus on BBRs. Here again, if the percentage of Trump voters negatively influences the adoption of any policy measure, the coefficient is large but rarely significant, whatever the type of policy concerned.16 The number of declared cases in the state has a significant and strong influence on the length of adoption, with a positive coefficient. In other words, the more important the number of cases, the slower the adoption of social distancing measures.17 This is not the case for the number of cases in the surrounding states, in particular, for the closure of schools (see Table 3b): the more important the number of cases in the region, the faster the adoption of school closures. This may be due to the fact that schools host children from surrounding states, in particular, those close to the border of each state, and that governors wanted to reduce the number of infections coming from outside of their state.

The institutional and legal context also played a major role. Whatever the policy measure, the softer the rules on getting funds out of the RDFs, the fewer the number of days for adopting any type of policy. Hence, it clearly appears that the negative economic impacts of the fight against the epidemic have been considered, and that RDFs have been considered as essential to smooth out their financial consequences: the more funds are easily available, the easier it is to offset the losses in revenues (or increases in expenditures) induced by the restrictions on economic activity.

A set of fiscal rules has played a major role in the adoption of policy measures. In particular, BBR number 2, which stipulates that ‘Own-source revenue must match expenditures’, tends to reduce the number of days necessary to implement school closures (see Table 3b) and, less significantly, non-essential business closures and stay-at-home orders (see Tables 3d and 3e). As this rule means that any policy with an impact on revenues must have an offsetting change in expenditures, it is not surprising that its impact has been strong on school closures (essentially, school closures tend to reduce expenditures as furloughed teachers can benefit from federal support, while school buses no longer need to be fuelled or maintained). In other words, this policy measure has allowed governors to save money, which they needed in order to deal with the consequences of other policy measures.

On the contrary, BBR number 6, stipulating that ‘The governor must sign a balanced budget’, lengthened the period of adoption of many policy measures, with school closures being the exception (see Tables 3a, 3c, 3d and 3e). As the epidemic hit the US during the period of preparation of the next fiscal year’s budgets, it is not surprising that this recommendation led to some delays in the adoption of health measures as their impact on the budget could only be expected to increase it. This also points to the possibility that governors may have considered a trade-off between the health and economic dimensions, induced by the presence of fiscal rules. In other words, the fear of an unbalanced budget, and of breaking the commitment stipulated by fiscal rules, may have prompted governors to be more reluctant in adopting health measures. Remarkably, the fiscal requirements have more influence than the other variables related to budget preparation (forecasts of revenues, expenditures or the end balance).

What prompts a state to act faster? We consider if each state has acted faster than the average state in its region, and we do this for each type of policy measure (for descriptive statistics, see the five Policy p_Yr variables in Table 1a). Tables 4a and 4b synthesise our results. We still find that the percentage of Trump voters is never significant. However, in this set of regressions only, the variable attached to a Republican governor does become significant. More precisely, it is always negative, meaning that in states dominated by the Republicans, the speed of adoption of social distancing measures was slower (in comparison with Democrat-governed states). This is especially true for restaurant restrictions, non-essential business closures and stay-at-home orders, which are probably the measures with the largest consequences on the budget balance.

Table 4a:

Probability of a shorter time period before adoption of a policy (7 April)

Gatherings restrictionsSchool closuresRestaurant restrictions
BBR #1 (Hou and Smith)–3.412*–3.698**–4.441**
(–1.78)(–2.16)(–2.14)
BBR #4 (Hou and Smith)–4.679–2.803**–2.949
(–0.98)(–2.03)(–1.64)
BBR #9 (Hou and Smith)2.1892.638*2.339*6.7563.409**2.161**
(1.63)(1.96)(1.84)(1.60)(2.05)(2.03)
RDF restrictive rules1.6261.9124.286
(1.38)(1.42)(0.96)
RDF soft rules1.5422.662*1.040
(1.26)(1.81)(0.45)
Republican governor–1.116–0.668–0.518–1.289–0.843–0.945–5.716*–2.169**–2.155*
(–1.44)(–0.93)(–0.85)(–1.34)(–1.13)(–1.24)(–1.68)(–1.97)(–1.94)
Log(GDP per capita)1.2891.0700.2120.7112.0370.087129.862.8991.584
(0.54)(0.47)(0.10)(0.27)(0.78)(0.04)(1.45)(0.93)(0.45)
RDF/GDP–11.64–6.500–10.66–122.9**–130.1**–146.0**1214.5618.0**718.4*
(–0.42)(–0.21)(–0.33)(–2.18)(–2.05)(–2.04)(1.23)(2.13)(1.86)
Trump voters (%, 2016)5.679–3.55838.91
(1.54)(–0.51)(1.42)
End balance forecast 2020/GDP426.3155.4–407.7
(1.56)(1.04)(–1.50)
Revenue forecast 2020/GDP21.1739.00–83.42*
(1.04)(1.30)(–1.78)
Log(1 + state i COVID-19 cases policy p)–1.262***–1.227***–1.040***–1.996***–1.616**–1.588**–5.834–1.992***–2.013***
(–3.08)(–3.08)(–2.93)(–2.67)(–2.49)(–2.26)(–1.51)(–2.96)(–2.81)
Log(1 + state i’s region COVID-19 cases Policy p)0.1220.2450.01440.2700.3360.160–0.145–0.328–0.166
(0.39)(0.93)(0.06)(0.68)(1.03)(0.51)(–0.10)(–0.87)(–0.41)
Share of states in region of state i with Policy p announced0.1880.03090.07571.0730.119–0.082810.68*2.3571.044
(0.16)(0.03)(0.07)(0.47)(0.08)(–0.05)(1.74)(1.22)(0.52)
Constant–14.41–8.9470.5891.118–14.077.631–323.5–20.10–3.375
(–0.54)(–0.36)(0.03)(0.04)(–0.49)(0.27)(–1.41)(–0.60)(–0.09)
Observations505050505050505050
Pseudo R-squared0.5340.4890.4480.6920.6120.6250.8360.6860.722

Notes: Policy p is where p refers to the same category of policy; t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

Table 4b:

Probability of a shorter time period before adoption of a policy (April 7)

Non-essential business closuresStay-at-home orders
BBR #1 (Hou and Smith)–1.444*–1.438*–1.276*
(–1.65)(–1.71)(–1.66)
BBR #7 (Hou and Smith)1.708*2.287**1.315*
(1.68)(2.13)(1.74)
BBR #9 (Hou and Smith)3.046*3.841*2.839*1.4211.5851.813*
(1.81)(1.77)(1.92)(1.63)(1.63)(1.80)
RDF restrictive rules0.1931.316**
(0.19)(2.08)
RDF soft rules1.2750.594
(0.89)(0.91)
Republican governor–1.802*–2.384*–1.512*–1.376**–1.637**–1.469**
(–1.69)(–1.75)(–1.82)(–2.16)(–2.38)(–2.39)
Log(GDP per capita)1.2921.4771.309–2.161–1.743–1.104
(0.31)(0.48)(0.51)(–0.95)(–0.89)(–0.56)
RDF/GDP275.6841.9**362.1–13.49–0.650–4.996
(0.93)(1.99)(1.42)(–0.32)(–0.02)(–0.14)
Trump voters (%, 2016)2.988–1.316
(0.47)(–0.33)
End balance forecast 2020/GDP–619.7**299.6
(–2.11)(1.40)
Revenue forecast 2020/GDP–12.94–7.206
(–0.43)(–0.30)
Log(1 + state i COVID-19 cases Policy p)–1.443**–1.726**–1.346**–0.696**–0.580**–0.647*
(–2.23)(–2.25)(–2.54)(–2.51)(–2.07)(–1.95)
Log(1 + state i’s region COVID-19 cases Policy p)0.3760.4830.459–0.0815–0.109–0.111
(0.67)(0.73)(1.04)(–0.27)(–0.39)(–0.38)
Share of states in region of state i with Policy p announced–0.186–0.944–0.520–1.018–0.794–1.144
(–0.11)(–0.42)(–0.33)(–0.96)(–0.76)(–1.07)
Constant–13.26–12.45–11.4030.6025.3819.37
(–0.28)(–0.36)(–0.41)(1.18)(1.18)(0.90)
Observations505050505050
Pseudo R-squared0.6590.7230.6330.4790.4410.406

Notes: Policy p is where p refers to the same category of policy; t-statistics in parentheses. * p < .1; ** p < .05; *** p < .01.

More important than partisan considerations are the (deterioration in) health conditions and the legally binding fiscal requirements. First, for all the policy measures, it appears that the number of COVID-19 cases has played an important role: the higher the number, the slower a governor announces a policy. The effect is even more significant for gatherings restrictions, restaurant restrictions and school closures (compared to stay-at-home orders and non-essential business closures). The health authorities’ recommendations on limiting the spread of the pandemic by reducing the opportunities for contact between people are followed tardily, and only in proportion to the number of cases – which may nevertheless have been too late to stop the spread of the disease.

Second, BBR number 1 (‘Governor must submit a balanced budget’) reduced the speed of adoption of school closures and stay-at-home orders (though the effect is barely significant for the latter). Given that these measures may have only a second-order impact in a budget, this could be expected. The same interpretation applies to BBR number 4 (‘Legislature must pass a balanced budget’), significant only for restrictions on restaurants.

BBR number 7 (‘Controls are in place on supplementary appropriations’) tends to increase the probability of acting quickly: where adjustments to the budget are subject to audits or controls, governors have tended to act faster, knowing that any fiscal drift would be monitored. Finally, BBR number 9 (‘No deficits are allowed to be carried over into the next fiscal year or budget cycle’) increases the probability of acting quickly, and this is true with regard to all the policy measures, whether on limiting gatherings of people or on closing economic activities in the face of the spread of the disease, though with different degrees of significance. Theoretically, later adoption of containment measures could lead to harsher consequences for government finances as it might imply a longer period of economic freezing. In fact, from an epidemiological perspective, the earlier the containment measures are taken, the shorter those containment measures may need to last. Thus, even if governors and their administrations are only concerned about avoiding or reducing the length of the economic freeze, it makes less sense for them to adopt containment measures later, when the epidemic has already exploded.18 This is exactly what our results reveal: governors seem to have acted under the combined pressure of the need for information in the face of uncertainty (as confirmed by the importance of COVID-19 cases) and the institutional constraints they have to deal with (as confirmed by the importance of BBRs – in particular, here, the no-deficit-carryover rule).

Rules on budget stabilisation funds do not seem to have influenced the speed of decision-making. However, the coefficient attached to the level of the RDF is significant and positive for restaurant restrictions and non-essential business closures. While these measures are typically those with a potentially large effect on the budget, acting fast in this case means a greater impact, and the RDF is then even more useful in cushioning the shock. These funds would compensate for an unexpected deficit, and thus for facing the fiscal consequences of the pandemic. This interpretation is reinforced by the fact that the amount in the RDF has a negative coefficient for school closures, which have reduced consequences for the budget balance (compared to, say, non-essential business closures). Here, again, our results support the view that the institutional context (the fiscal rules and other financial regulations) have had a remarkable influence in the face of the pandemic.

Conclusion

BBRs have played a decisive role: rules of a political nature – in particular, that the governor balance the budget – increase the delay in decision-making, while those forbidding the carryover of a deficit prompt them to act faster. One explanation would be that rules of a political nature place a significant weight on the political responsibility of the governor, especially on the responsibility for the consequences of their actions in balancing the budget. This increases the time for reflection, and the probability of acting slowly, comparatively to neighbouring states. Conversely, technical rules tend to increase the number of measures, as well as the speed of announcement. Would technical rules only indirectly engage the governor’s responsibility to balance the budget, removing responsibility for policies in favour of painful future budgetary adjustments, at the price of generating more pro-cyclicality?

RDFs allow for faster decision-making and their amount favours the implementation of more social distancing measures. The absence of an RDF creates more uncertainty. While the National Association of State Budget Officers has in recent years signalled that the states’ problem is the management of surpluses (rather than budget cuts, as in the 2008 crisis), we show that higher reserves made it easier to adapt in face of the pandemic shock. Moreover, while rules explicitly linked to economic and/or revenue volatility (that is, more restrictive ones) favour the counter-cyclical role in face of an unanticipated economic shock, in the case of this health crisis, withdrawal rules not explicitly linked to this volatility (that is, softer ones) have allowed for better reactivity (unlike in a classic recession, the measures announced to confront the health crisis are likely to create the fiscal shock). Our results indicate that although politics can be an important determinant in the adoption of policy measures, in face of the pandemic, institutional economic rules or, more precisely, budgetary constraints have trumped politics.

Funding

We acknowledge financial support from the LEM (UMR 9221).

Acknowledgements

The authors would like to thank the three referees of the journal, as well as the editors, for useful comments and remarks. Support and comments by Marcelin Joanis and Jérôme Héricourt have also contributed to the realisation of this research. The usual disclaimer applies.

Notes

1

As reported, for example, by the New York Times, see: www.nytimes.com/2020/03/23/us/politics/trump-coronavirus-restrictions.html

2

For the US, first estimates can be found in, for example, Barro et al (2020) or Eichenbaum et al (2020), as well as in the literature review provided later.

3

Hou and Smith (2010) detail the institutional context surrounding fiscal decisions in the US states under the constraint of BBRs, while Hansen (2020) shows that fiscal rules are efficiently constraining the behaviour of policymakers because they are internalised by domestic political actors.

4

It can be shown that testing widely can reduce the economic costs as it would favour the possibility of some workers returning to work earlier (Ichino et al, 2020). Delivering ‘passports’ to tested workers lies at the core of the proposal by Eichenberger et al (forthcoming). Lee et al (2020) show that low-skilled workers would benefit most from such a policy, while Brotherhood et al (2020) insist on the gains for the younger workers.

5

In particular, the desired impact is to reduce the loss of consumption, traded off with the probability of COVID-related death. For an analysis across such a line, see Hall et al (2020).

6

For a more global analysis, involving 50 countries, which confirms the results obtained on and in the US, see Jinjarak et al (2020). For Germany, see Glogowski et al (2020). Askitas et al (2020) look at policies across 135 countries, confirming the importance of the restrictions on mobility in the arsenal deployed against the disease, while Lin and Meissner (2020) analyse the spillovers of the lockdown measures across 70 countries as well as the US states. Cronert (2020) focuses on the specific case of school closures in 167 countries, revealing that competitive elections may have prompted policymakers to react faster. This adds a nuance to Cepaluni et al’s (2020) results which show that democracies are disadvantaged when it comes to imposing measures that typically constrain civil liberties.

7

Although even wealthier agents have suffered, in so far as real estate is an important part of their wealth, given that the housing market has been severely hit (Yörük, 2020).

8

The still short-term perspective on the crisis forbids the use of sophisticated econometric techniques given the small number of observations. Adolph et al (2020) use event-studies techniques, while we will rely on standard OLS and probit analyses, as do Baccini and Brodeur (2020). Nevertheless, in such a context, correlations are more than telling, even if the techniques forbid going too far in terms of causal conclusions.

9

The 2019 novel coronavirus COVID-19 (2019-ncov) data repository can be found at: https://github.com/CSSEGISandData/COVID-19

10

Sources include NASBO’s (National Association of State Budget Officers) ‘Fiscal survey of the states’ for autumn 2019 (available at: www.nasbo.org/reports-data/fiscal-survey-of-states) and data from: https://gsfic.georgia.gov/revenue-shortfall-reserve-holdings-reports

11

For Republican governorship, data were from the National Conference of State Legislators, ‘State partisan composition’, for January 2020 (available at: www.ncsl.org/research/about-state-legislatures/partisancomposition); for the percentage of Trump voters in the 2016 election, data were from: https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/42MVDX

12

More precisely, in response to the question: ‘We hear a lot of talk these days about liberals and conservatives. When it comes to politics, do you usually think of yourself as extremely liberal, liberal, slightly liberal, moderate, slightly conservative, conservative, extremely conservative or haven’t you thought much about this?’ After aggregating the individual ANES data state-wise, we calculate an HH polarisation index for each state in 2016. (We disregard the option ‘don’t know’ or ‘haven’t thought much about it’.)

13

The small number of observations forbids the addition of too many control variables, and GDP per capita in many ways summarises an important number of differences among US states.

14

To save on space, we only reproduce results for the significant rules. Other results are available upon request.

15

Given the small number of observations for curfews and restrictions on travel, we neglect these two measures in the rest of the study.

16

The small number of observations means that one should be cautious about statistical significance.

17

The question of reverse causality (that is, states that took more time to adopt policies saw a higher increase in the number of cases) may be raised. To tackle this, we estimated the same regressions with a seven-, ten- and 14-day lag of the COVID-19 cases regressor. In this set of estimates, the instantaneous number of COVID-19 cases is no longer significant, while the lags are strongly significant. Yet, the sign associated with the respective coefficients is still always positive. Hence, we think it can be safely interpreted that the number of cases did indeed slow the decision process as the delay between the increase in cases and the announcement increases in the (lagged) number of cases.

18

We thank one of the referees for this interpretation.

Conflict of interest

The authors declare that there is no conflict of interest.

References

  • ACIR (Advisory Commission on Intergovernmental Relations) (1987) Fiscal discipline in the federal system: national reform and the experience of the states, Report No. A-107.

    • Search Google Scholar
    • Export Citation
  • Adolph, Ch., Amano, K., Bang-Jensen B., Fullman, N. and Wilkerson, J. (2020) Pandemic politics: timing state-level social distancing responses to COVID-19, https://doi.org/10.1101/2020.03.30.20046326.

    • Search Google Scholar
    • Export Citation
  • Allcott, H., Boxell, L., Conway, J.C., Gentzkow, M., Thaler, M. and Yang, D.Y. (2020) Polarization and public health: partisan differences in social distancing during the coronavirus pandemic, NBER Working Paper, No. 26946.

    • Search Google Scholar
    • Export Citation
  • Alvarez, F.E., Argente, D. and Lippi F. (2020) A simple planning problem for COVID-19 lockdown, NBER Working Paper, No. 26981.

  • Angeletos, G.M., Collard, F. and Dellas, H. (2016) Public debt as private liquidity: optimal policy, NBER Working Paper, No. 22794.

  • Asatryan, Z., Castellón, C. and Stratmann, T. (2018) Balanced budget rules and fiscal outcomes: evidence from historical constitutions, Journal of Public Economics, 167 (November): 10519.

    • Search Google Scholar
    • Export Citation
  • Askitas, N., Tatsiramos, K. and Verheyden, B. (2020) Lockdown strategies, mobility patterns and COVID-19, CesIfo Working Paper, No. 8338.

    • Search Google Scholar
    • Export Citation
  • Aum, S., Lee, S.Y.T. and Shin, Y. (2020) COVID-19 doesn’t need lockdowns to destroy jobs: the effect of local outbreaks in Korea, NBER Working Paper, No. 27264.

    • Search Google Scholar
    • Export Citation
  • Azzimonti, M., Battaglini, M. and Coate, S. (2016) The costs and benefits of balanced budget rules: lessons from a political economy model of fiscal policy, Journal of Public Economics, 136 (April): 4561.

    • Search Google Scholar
    • Export Citation
  • Baccini, L. and Brodeur, A. (2020) Explaining governors’ response to the Covid-19 pandemic in the United States, IZA Discussion Paper, No. 13137.

    • Search Google Scholar
    • Export Citation
  • Baldassari, D. and Park, B. (2020) Was there a culture war? Partisan polarization and secular trends in US public opinion, Journal of Politics, 82(3): 80927.

    • Search Google Scholar
    • Export Citation
  • Ball, L., Elmendorf, D.W. and Mankiw, N.G. (1998) The deficit gamble, Journal of Money, Credit and Banking, 30(4): 699720.

  • Barnett, M., Buchak, G. and Yannelis, C. (2020) Epidemic responses under uncertainty, NBER Working Paper, No. 27289.

  • Barrios, J. and Hochberg, Y.V. (2020) Risk perception through the lens of politics in the time of the COVID-19 pandemic, NBER Working Paper, No. 27008.

    • Search Google Scholar
    • Export Citation
  • Barro, R.J. (1979) On the determination of the public debt, Journal of Political Economy, 87(5): 94071.

  • Barro, R.J., Ursúa, J.F. and Weng, J. (2020) The coronavirus and the great influenza pandemic. Lessons from the ‘Spanish Flu’ for the coronavirus’s potential effects on mortality and economic activity, NBER Working Paper, No. 26866.

    • Search Google Scholar
    • Export Citation
  • Battaglini, M. and Coate, S. (2008) A dynamic theory of public spending, taxation, and debt, American Economic Review, 98(1): 20136.

  • Bohn, H. and Inman, R.P. (1996) Balanced-budget rules and public deficits: evidence from the U.S. states, Carnegie-Rochester Conference Series on Public Policy, 45(1): 1376.

    • Search Google Scholar
    • Export Citation
  • Brotherhood, L., Kircher, P., Santos, C. and Tertilt, M. (2020) An economic model of the Covid-19 epidemic: the importance of testing and age-specific policies, CesIfo Working Paper, No. 8316.

    • Search Google Scholar
    • Export Citation
  • Campbell, A.L. and Sances, M.W. (2013) State fiscal policy during the Great Recession: budgetary impacts and policy responses, The Annals of the American Academy of Political and Social Science, 650 (November): 25273.

    • Search Google Scholar
    • Export Citation
  • Cepaluni, G., Dorsch, M.T. and Branyiczki, R. (2020) Political regimes and deaths in the early stages of the COVID-19 pandemic, Cambridge Open Engage, doi: 10.33774/coe-2020-hvvns.

    • Search Google Scholar
    • Export Citation
  • Chetty, R., Friedman, J.N., Hendren, N. and Stepner, M. (The Opportunity Insights Team) (2020) How did Covid-19 and stabilization policies affect spending and employment?, A New Real-time Economic Tracker Based On Private Sector Data, NBER Working Paper, No. 27431.

    • Search Google Scholar
    • Export Citation
  • Clemens, J. and Miran, S. (2012) Fiscal policy multipliers on subnational government spending, American Economic Journal: Economic Policy, 4(2): 4668.

    • Search Google Scholar
    • Export Citation
  • Clemens, J. and Veuger, S. (forthcoming) Implications of the Covid-19 pandemic for state government tax revenues, National Tax Journal, 73(3): 61944. doi: 10.17310/ntj.2020.3.01.

    • Search Google Scholar
    • Export Citation
  • Collard, F., Hellwig, C., Assenza, T., Kankanamge, S., Dupaigne, M., Werquin, N. and Fève, P. (2020) The hammer and the dance: equilibrium and optimal policy during a pandemic crisis, CEPR Discussion Paper, No. 14731 (v. 2).

    • Search Google Scholar
    • Export Citation
  • Couch, K.A., Fairlie, R.W. and Xu, H. (2020) The impacts of Covid-19 on minority unemployment: first evidence from April 2020 CPS microdata, CesIfo Working Paper, No. 8327.

    • Search Google Scholar
    • Export Citation
  • Cronert, A. (2020) Democracy, state capacity, and COVID-19 related school closures, APSA Preprints, https://preprints.apsanet.org/engage/apsa/article-details/5ea8501b68bfcc00122e96ac, doi: 10.33774/apsa-2020-jf671-v4.

    • Search Google Scholar
    • Export Citation
  • Dave, D.M., Friedson, A.I., Matsuzawa, K., McNichols, D. and Sabia, J.J. (2020a) Did the Wisconsin Supreme Court restart a Covid-19 epidemic? Evidence from a natural experiment, NBER Working Paper, No. 27322.

    • Search Google Scholar
    • Export Citation
  • Dave, D.M., Friedson, A.I., Matsuzawa, K. and Sabia, J.J. (2020b) When do shelter-in-place orders fight COVID-19 best? Policy heterogeneity across states and adoption time, NBER Working Paper, No. 27091.

    • Search Google Scholar
    • Export Citation
  • Desmet, K. and Warzciag, R. (2020) Understanding spatial variation in Covid-19 across the United States, NBER Working Paper, No. 27329.

  • Eichenbaum, M.S., Rebelo, S. and Trabandt, M. (2020) The macroeconomics of epidemics, NBER Working Paper, No. 26882.

  • Eichenberger, R., Hegselmann, R., Savage, D.A., Stadelmann, D. and Torgler, B. (forthcoming) Certified coronavirus immunity as a resource and strategy to cope with pandemic costs, Kyklos, doi: 10.1111/kykl.12227.

    • Search Google Scholar
    • Export Citation
  • Fairlie, R.W. (2020) The impact of Covid-19 on small business owners: evidence of early-stage losses from the April 2020 current population survey, NBER Working Paper, No. 27309.

    • Search Google Scholar
    • Export Citation
  • Fan, Y., Orhun, A.Y. and Turjeman, D. (2020) Heterogeneous actions, beliefs, constraints and risk tolerance during the COVID-19 pandemic, NBER Working Paper, No. 27211.

    • Search Google Scholar
    • Export Citation
  • Fernández, J.G. and Parro, F. (2019) Fiscal rules and financial systems: complements or substitutes?, Oxford Bulletin of Economics and Statistics, 81(3): 588616.

    • Search Google Scholar
    • Export Citation
  • Friedson, A.I., McNichols, D., Sabia, J.J. and Dave, D. (2020) Did California’s shelter-in-place order work? Early coronavirus-related public health effects, NBER Working Paper, No. 26992.

    • Search Google Scholar
    • Export Citation
  • Fullman, N., Bang-Jensen, B., Amano, K., Adolph, Ch. and Wilkerson, J. (2020) State-level social distancing policies in response to COVID-19 in the US, www.covid19statepolicy.org/.

    • Search Google Scholar
    • Export Citation
  • Gitmez, A., Sonin, K. and Wright, A.L. (2020) Political economy of crisis response, CEPR Discussion Paper, No. 14778 (v. 3).

  • Glogowski, U., Hansen, E. and Schächtele, S. (2020) How effective are social distancing policies? Evidence on the fight against COVID-19 from Germany, CesIfo Working Paper, No. 8361.

    • Search Google Scholar
    • Export Citation
  • Gonzales-Eiras, M. and Niepelt, D. (2020) On the optimal ‘lockdown’ during an epidemic, CEPR Discussion Paper, No. 14612.

  • Guerrieri, V., Lorenzoni, G., Straub, L. and Werning, I. (2020) Macroeconomic implications of COVID-19: can negative supply shocks cause demand shortages?, NBER Working Paper, No. 26918.

    • Search Google Scholar
    • Export Citation
  • Gupta, S., Nguyen, T.D., Rojas, F.L., Raman, S., Lee, B., Bento, A., Simon, K.I. and Wing, C. (2020) Tracking public and private responses to the COVID-19 epidemic: evidence from state and local government actions, NBER Working Paper, No. 27027.

    • Search Google Scholar
    • Export Citation
  • Hall, R.E., Jones, C.I. and Klenow, P.J. (2020) Trading-off consumption and Covid-19 deaths, NBER Working Paper, No. 27340.

  • Hansen, D. (2020) The effectiveness of fiscal institutions: International financial flogging or domestic constraint?, European Journal of Political Economy, Apr 18: 101879.

    • Search Google Scholar
    • Export Citation
  • Heinemann, F., Moessinger, M.D. and Yeter, M. (2018) Do fiscal rules constrain fiscal policy? A meta-regression-analysis, European Journal of Political Economy, 51(January): 6992.

    • Search Google Scholar
    • Export Citation
  • Hibbs, D.A.J. (1977) Political parties and macroeconomic policy, American Political Science Review, 71: 146787.

  • Hou, Y. and Smith, D. (2006) A framework for understanding state balanced budget requirement systems: re-examining distinctive features and an operational definition, Public Budgeting and Finance, 26(3): 2245.

    • Search Google Scholar
    • Export Citation
  • Hou, Y. and Smith, D.L. (2010) Do state balanced budget requirements matter? Testing two explanatory frameworks, Public Choice, 145(1/2): 5779.

    • Search Google Scholar
    • Export Citation
  • Hutchinson, M.K. and Holtman, M.C. (2005) Analysis of count data using Poisson regression, Research in Nursing & Health, 28(5): 40818.

    • Search Google Scholar
    • Export Citation
  • Ichino, A., Favero, C.A. and Rustichini, A. (2020) Restarting the economy while saving lives under COVID-19, CEPR Discussion Paper, No. 14664 (v. 2).

    • Search Google Scholar
    • Export Citation
  • Jarosch, G., Farboodi, M. and Shimer, R. (2020) Internal and external effects of social distancing in a pandemic, CEPR Discussion Paper, No. 14670.

    • Search Google Scholar
    • Export Citation
  • Jinjarak, Y., Ahmed, R., Nair-Desai, S., Xin, W. and Aizenman, J. (2020) Accounting for global COVID-19 diffusion patterns, January–April 2020, NBER Working Paper, No. 27185.

    • Search Google Scholar
    • Export Citation
  • Jonas, J. (2012) Great Recession and fiscal squeeze at U.S. subnational government level, IMF Working Papers, No. 12/184.

  • Kempf, H. (2020) One foe, so many fights. Making sense of Covid-19 policies, CesIfo Working Paper, No. 8325.

  • Lee, S.Y., Aum, S. and Shin, Y. (2020) Inequality of fear and self-quarantine: is there a trade-off between GDP and public health?, CEPR Discussion Paper, No. 14679 (v. 9).

    • Search Google Scholar
    • Export Citation
  • Lin, Z. and Meissner, C.M. (2020) Health vs. wealth? Public health policies and the economy during Covid-19, NBER Working Paper, No. 27099.

    • Search Google Scholar
    • Export Citation
  • Mitman, K. and Rabinovich, S. (2020) Optimal unemployment benefits in the pandemic, CEPR Discussion Paper, No. 14915.

  • Mongey, S., Pilossoph, L. and Weinberg, A. (2020) Which workers bear the burden of social distancing policies?, NBER Working Paper, No. 27085.

    • Search Google Scholar
    • Export Citation
  • Nordhaus, W.D. (1975) The political business cycle, Review of Economic Studies, 42(2): 16990.

  • Pew Charitable Trusts (2014) Building state rainy day funds, report, the PEW Charitable Trusts, www.pewtrusts.org/-/media/assets/2014/07/sfh_rainy-day-fund-deposit-rules-report_artready_v9.pdf.

    • Search Google Scholar
    • Export Citation
  • Pew Charitable Trusts (2017) When to use state rainy day funds, report, the PEW Charitable Trusts, http://pew.org/2o7fCls.

  • Piguillem, F. and Shi, L. (2020) Optimal Covid-19 quarantine and testing policies, CEPR Discussion Paper, No. 14613 (v. 2).

  • Poterba, J.M. (1994) State responses to fiscal crises: the effects of budgetary institutions and politics, Journal of Political Economy, 102(4): 799821.

    • Search Google Scholar
    • Export Citation
  • Randall, M. and Rueben, K. (2017) Sustainable Budgeting in the States: Evidence on State Budget Institutions and Practices, Urban Institute, www.urban.org.

    • Search Google Scholar
    • Export Citation
  • Rogoff, K. (1990) Equilibrium political budget cycles, American Economic Review, 80(1): 2136.

  • Rogoff, K. and Sibert, A. (1988) Elections and macroeconomic policy cycles, Review of Economic Studies, 55(1): 116.

  • Sauvagnat, J., Barrot, J.N. and Grssi, B. (2020) Estimating the costs and benefits of mandated business closures in a pandemic, CEPR Discussion Paper, No. 14757.

    • Search Google Scholar
    • Export Citation
  • Velasco, A. and Chang, R. (2020) Economic policy incentives to preserve lives and livelihoods, CEPR Discussion Paper, No. 14614.

  • Wright, A.L., Sonin, K., Driscoll, J. and Wilson, J. (2020) Poverty and economic dislocation reduce compliance with COVID-19 shelter-in-place protocols, CEPR Discussion Paper, No. 14618 (v. 3).

    • Search Google Scholar
    • Export Citation
  • Yörük, B.K. (2020) Early effects of the Covid-10 pandemic on housing market in the United States, CesIfo Working Paper, No. 8333.

  • Zhao, B. (2016) Saving for a rainy day: estimating the needed size of U.S. state budget stabilization funds, Regional Science and Urban Economics, 61 (November): 13052.

    • Search Google Scholar
    • Export Citation
  • ACIR (Advisory Commission on Intergovernmental Relations) (1987) Fiscal discipline in the federal system: national reform and the experience of the states, Report No. A-107.

    • Search Google Scholar
    • Export Citation
  • Adolph, Ch., Amano, K., Bang-Jensen B., Fullman, N. and Wilkerson, J. (2020) Pandemic politics: timing state-level social distancing responses to COVID-19, https://doi.org/10.1101/2020.03.30.20046326.

    • Search Google Scholar
    • Export Citation
  • Allcott, H., Boxell, L., Conway, J.C., Gentzkow, M., Thaler, M. and Yang, D.Y. (2020) Polarization and public health: partisan differences in social distancing during the coronavirus pandemic, NBER Working Paper, No. 26946.

    • Search Google Scholar
    • Export Citation
  • Alvarez, F.E., Argente, D. and Lippi F. (2020) A simple planning problem for COVID-19 lockdown, NBER Working Paper, No. 26981.

  • Angeletos, G.M., Collard, F. and Dellas, H. (2016) Public debt as private liquidity: optimal policy, NBER Working Paper, No. 22794.

  • Asatryan, Z., Castellón, C. and Stratmann, T. (2018) Balanced budget rules and fiscal outcomes: evidence from historical constitutions, Journal of Public Economics, 167 (November): 10519.

    • Search Google Scholar
    • Export Citation
  • Askitas, N., Tatsiramos, K. and Verheyden, B. (2020) Lockdown strategies, mobility patterns and COVID-19, CesIfo Working Paper, No. 8338.

    • Search Google Scholar
    • Export Citation
  • Aum, S., Lee, S.Y.T. and Shin, Y. (2020) COVID-19 doesn’t need lockdowns to destroy jobs: the effect of local outbreaks in Korea, NBER Working Paper, No. 27264.

    • Search Google Scholar
    • Export Citation
  • Azzimonti, M., Battaglini, M. and Coate, S. (2016) The costs and benefits of balanced budget rules: lessons from a political economy model of fiscal policy, Journal of Public Economics, 136 (April): 4561.

    • Search Google Scholar
    • Export Citation
  • Baccini, L. and Brodeur, A. (2020) Explaining governors’ response to the Covid-19 pandemic in the United States, IZA Discussion Paper, No. 13137.

    • Search Google Scholar
    • Export Citation
  • Baldassari, D. and Park, B. (2020) Was there a culture war? Partisan polarization and secular trends in US public opinion, Journal of Politics, 82(3): 80927.

    • Search Google Scholar
    • Export Citation
  • Ball, L., Elmendorf, D.W. and Mankiw, N.G. (1998) The deficit gamble, Journal of Money, Credit and Banking, 30(4): 699720.

  • Barnett, M., Buchak, G. and Yannelis, C. (2020) Epidemic responses under uncertainty, NBER Working Paper, No. 27289.

  • Barrios, J. and Hochberg, Y.V. (2020) Risk perception through the lens of politics in the time of the COVID-19 pandemic, NBER Working Paper, No. 27008.

    • Search Google Scholar
    • Export Citation
  • Barro, R.J. (1979) On the determination of the public debt, Journal of Political Economy, 87(5): 94071.

  • Barro, R.J., Ursúa, J.F. and Weng, J. (2020) The coronavirus and the great influenza pandemic. Lessons from the ‘Spanish Flu’ for the coronavirus’s potential effects on mortality and economic activity, NBER Working Paper, No. 26866.

    • Search Google Scholar
    • Export Citation
  • Battaglini, M. and Coate, S. (2008) A dynamic theory of public spending, taxation, and debt, American Economic Review, 98(1): 20136.

  • Bohn, H. and Inman, R.P. (1996) Balanced-budget rules and public deficits: evidence from the U.S. states, Carnegie-Rochester Conference Series on Public Policy, 45(1): 1376.

    • Search Google Scholar
    • Export Citation
  • Brotherhood, L., Kircher, P., Santos, C. and Tertilt, M. (2020) An economic model of the Covid-19 epidemic: the importance of testing and age-specific policies, CesIfo Working Paper, No. 8316.

    • Search Google Scholar
    • Export Citation
  • Campbell, A.L. and Sances, M.W. (2013) State fiscal policy during the Great Recession: budgetary impacts and policy responses, The Annals of the American Academy of Political and Social Science, 650 (November): 25273.

    • Search Google Scholar
    • Export Citation
  • Cepaluni, G., Dorsch, M.T. and Branyiczki, R. (2020) Political regimes and deaths in the early stages of the COVID-19 pandemic, Cambridge Open Engage, doi: 10.33774/coe-2020-hvvns.

    • Search Google Scholar
    • Export Citation
  • Chetty, R., Friedman, J.N., Hendren, N. and Stepner, M. (The Opportunity Insights Team) (2020) How did Covid-19 and stabilization policies affect spending and employment?, A New Real-time Economic Tracker Based On Private Sector Data, NBER Working Paper, No. 27431.

    • Search Google Scholar
    • Export Citation
  • Clemens, J. and Miran, S. (2012) Fiscal policy multipliers on subnational government spending, American Economic Journal: Economic Policy, 4(2): 4668.

    • Search Google Scholar
    • Export Citation
  • Clemens, J. and Veuger, S. (forthcoming) Implications of the Covid-19 pandemic for state government tax revenues, National Tax Journal, 73(3): 61944. doi: 10.17310/ntj.2020.3.01.

    • Search Google Scholar
    • Export Citation
  • Collard, F., Hellwig, C., Assenza, T., Kankanamge, S., Dupaigne, M., Werquin, N. and Fève, P. (2020) The hammer and the dance: equilibrium and optimal policy during a pandemic crisis, CEPR Discussion Paper, No. 14731 (v. 2).

    • Search Google Scholar
    • Export Citation
  • Couch, K.A., Fairlie, R.W. and Xu, H. (2020) The impacts of Covid-19 on minority unemployment: first evidence from April 2020 CPS microdata, CesIfo Working Paper, No. 8327.

    • Search Google Scholar
    • Export Citation
  • Cronert, A. (2020) Democracy, state capacity, and COVID-19 related school closures, APSA Preprints, https://preprints.apsanet.org/engage/apsa/article-details/5ea8501b68bfcc00122e96ac, doi: 10.33774/apsa-2020-jf671-v4.

    • Search Google Scholar
    • Export Citation
  • Dave, D.M., Friedson, A.I., Matsuzawa, K., McNichols, D. and Sabia, J.J. (2020a) Did the Wisconsin Supreme Court restart a Covid-19 epidemic? Evidence from a natural experiment, NBER Working Paper, No. 27322.

    • Search Google Scholar
    • Export Citation
  • Dave, D.M., Friedson, A.I., Matsuzawa, K. and Sabia, J.J. (2020b) When do shelter-in-place orders fight COVID-19 best? Policy heterogeneity across states and adoption time, NBER Working Paper, No. 27091.

    • Search Google Scholar
    • Export Citation
  • Desmet, K. and Warzciag, R. (2020) Understanding spatial variation in Covid-19 across the United States, NBER Working Paper, No. 27329.

  • Eichenbaum, M.S., Rebelo, S. and Trabandt, M. (2020) The macroeconomics of epidemics, NBER Working Paper, No. 26882.

  • Eichenberger, R., Hegselmann, R., Savage, D.A., Stadelmann, D. and Torgler, B. (forthcoming) Certified coronavirus immunity as a resource and strategy to cope with pandemic costs, Kyklos, doi: 10.1111/kykl.12227.

    • Search Google Scholar
    • Export Citation
  • Fairlie, R.W. (2020) The impact of Covid-19 on small business owners: evidence of early-stage losses from the April 2020 current population survey, NBER Working Paper, No. 27309.

    • Search Google Scholar
    • Export Citation
  • Fan, Y., Orhun, A.Y. and Turjeman, D. (2020) Heterogeneous actions, beliefs, constraints and risk tolerance during the COVID-19 pandemic, NBER Working Paper, No. 27211.

    • Search Google Scholar
    • Export Citation
  • Fernández, J.G. and Parro, F. (2019) Fiscal rules and financial systems: complements or substitutes?, Oxford Bulletin of Economics and Statistics, 81(3): 588616.

    • Search Google Scholar
    • Export Citation
  • Friedson, A.I., McNichols, D., Sabia, J.J. and Dave, D. (2020) Did California’s shelter-in-place order work? Early coronavirus-related public health effects, NBER Working Paper, No. 26992.

    • Search Google Scholar
    • Export Citation
  • Fullman, N., Bang-Jensen, B., Amano, K., Adolph, Ch. and Wilkerson, J. (2020) State-level social distancing policies in response to COVID-19 in the US, www.covid19statepolicy.org/.

    • Search Google Scholar
    • Export Citation
  • Gitmez, A., Sonin, K. and Wright, A.L. (2020) Political economy of crisis response, CEPR Discussion Paper, No. 14778 (v. 3).

  • Glogowski, U., Hansen, E. and Schächtele, S. (2020) How effective are social distancing policies? Evidence on the fight against COVID-19 from Germany, CesIfo Working Paper, No. 8361.

    • Search Google Scholar
    • Export Citation
  • Gonzales-Eiras, M. and Niepelt, D. (2020) On the optimal ‘lockdown’ during an epidemic, CEPR Discussion Paper, No. 14612.

  • Guerrieri, V., Lorenzoni, G., Straub, L. and Werning, I. (2020) Macroeconomic implications of COVID-19: can negative supply shocks cause demand shortages?, NBER Working Paper, No. 26918.

    • Search Google Scholar
    • Export Citation
  • Gupta, S., Nguyen, T.D., Rojas, F.L., Raman, S., Lee, B., Bento, A., Simon, K.I. and Wing, C. (2020) Tracking public and private responses to the COVID-19 epidemic: evidence from state and local government actions, NBER Working Paper, No. 27027.

    • Search Google Scholar
    • Export Citation
  • Hall, R.E., Jones, C.I. and Klenow, P.J. (2020) Trading-off consumption and Covid-19 deaths, NBER Working Paper, No. 27340.

  • Hansen, D. (2020) The effectiveness of fiscal institutions: International financial flogging or domestic constraint?, European Journal of Political Economy, Apr 18: 101879.

    • Search Google Scholar
    • Export Citation
  • Heinemann, F., Moessinger, M.D. and Yeter, M. (2018) Do fiscal rules constrain fiscal policy? A meta-regression-analysis, European Journal of Political Economy, 51(January): 6992.

    • Search Google Scholar
    • Export Citation
  • Hibbs, D.A.J. (1977) Political parties and macroeconomic policy, American Political Science Review, 71: 146787.

  • Hou, Y. and Smith, D. (2006) A framework for understanding state balanced budget requirement systems: re-examining distinctive features and an operational definition, Public Budgeting and Finance, 26(3): 2245.

    • Search Google Scholar
    • Export Citation
  • Hou, Y. and Smith, D.L. (2010) Do state balanced budget requirements matter? Testing two explanatory frameworks, Public Choice, 145(1/2): 5779.

    • Search Google Scholar
    • Export Citation
  • Hutchinson, M.K. and Holtman, M.C. (2005) Analysis of count data using Poisson regression, Research in Nursing & Health, 28(5): 40818.

    • Search Google Scholar
    • Export Citation
  • Ichino, A., Favero, C.A. and Rustichini, A. (2020) Restarting the economy while saving lives under COVID-19, CEPR Discussion Paper, No. 14664 (v. 2).

    • Search Google Scholar
    • Export Citation
  • Jarosch, G., Farboodi, M. and Shimer, R. (2020) Internal and external effects of social distancing in a pandemic, CEPR Discussion Paper, No. 14670.

    • Search Google Scholar
    • Export Citation
  • Jinjarak, Y., Ahmed, R., Nair-Desai, S., Xin, W. and Aizenman, J. (2020) Accounting for global COVID-19 diffusion patterns, January–April 2020, NBER Working Paper, No. 27185.

    • Search Google Scholar
    • Export Citation
  • Jonas, J. (2012) Great Recession and fiscal squeeze at U.S. subnational government level, IMF Working Papers, No. 12/184.

  • Kempf, H. (2020) One foe, so many fights. Making sense of Covid-19 policies, CesIfo Working Paper, No. 8325.

  • Lee, S.Y., Aum, S. and Shin, Y. (2020) Inequality of fear and self-quarantine: is there a trade-off between GDP and public health?, CEPR Discussion Paper, No. 14679 (v. 9).

    • Search Google Scholar
    • Export Citation
  • Lin, Z. and Meissner, C.M. (2020) Health vs. wealth? Public health policies and the economy during Covid-19, NBER Working Paper, No. 27099.

    • Search Google Scholar
    • Export Citation
  • Mitman, K. and Rabinovich, S. (2020) Optimal unemployment benefits in the pandemic, CEPR Discussion Paper, No. 14915.

  • Mongey, S., Pilossoph, L. and Weinberg, A. (2020) Which workers bear the burden of social distancing policies?, NBER Working Paper, No. 27085.

    • Search Google Scholar
    • Export Citation
  • Nordhaus, W.D. (1975) The political business cycle, Review of Economic Studies, 42(2): 16990.

  • Pew Charitable Trusts (2014) Building state rainy day funds, report, the PEW Charitable Trusts, www.pewtrusts.org/-/media/assets/2014/07/sfh_rainy-day-fund-deposit-rules-report_artready_v9.pdf.

    • Search Google Scholar
    • Export Citation
  • Pew Charitable Trusts (2017) When to use state rainy day funds, report, the PEW Charitable Trusts, http://pew.org/2o7fCls.

  • Piguillem, F. and Shi, L. (2020) Optimal Covid-19 quarantine and testing policies, CEPR Discussion Paper, No. 14613 (v. 2).

  • Poterba, J.M. (1994) State responses to fiscal crises: the effects of budgetary institutions and politics, Journal of Political Economy, 102(4): 799821.

    • Search Google Scholar
    • Export Citation
  • Randall, M. and Rueben, K. (2017) Sustainable Budgeting in the States: Evidence on State Budget Institutions and Practices, Urban Institute, www.urban.org.

    • Search Google Scholar
    • Export Citation
  • Rogoff</