Business Politicians and Fiscal Consolidation

What explains the variation in countries’ propensity to engage in austerity policy? Economic and political country-level factors are the paramount explanations in the literature. Nevertheless, the variation in fiscal preferences at the executive level remains under-explored, except for ideology. Moreover, budget decisions are endogenous to the state of the economy, thus casting doubts on standard measures based on the debt and/or deficit ratio. This paper contributes to the literature in three ways. First, I turn to the individual level of analysis and suggest that leaders with business experience are more likely to pursue a balanced budget and tend to implement fiscal consolidation policies based on spending cuts. Second, I ease concerns on individuals’ self-selection into office by relying on fiscal adjustments that are weakly orthogonal to the economic cycle. Third, I employ an instrumental variable approach to account for self-selection into business. I empirically test the relationship between business leaders and fiscal consolidation on a panel of 17 OECD countries between 1978 and 2014.


Introduction
"I was in the world of business for 25 years.If you didn't balance your budget, you went out of business."(Mitt Romney, 2011, Presidential Debate). 1ethodologically, I move beyond previous studies on the effects of leaders on public debt in two ways.First, I take advantage of the newest developments in macroeconomic research and utilize a dataset on fiscal consolidations that are weakly exogenous to the business cycle (Devries et al. 2011;Alesina et al. 2020).By doing so, the empirical analysis takes into account the primary endogeneity concern that has affected previous studies, i.e. the selection of former businesspeople into politics as a function of the economic cycle.Second, I rely on an instrumental variable (IV) approach to isolate exogenous variation in the (future) leaders' decision to enter the business world at the time when they first entered the job market.To the best of my knowledge, this is the first attempt in the literature on leaders to account for individuals' self-selection into business experience.In a nutshell, I uncover tow main findings.First, politicians with prior business experience are more likely to pursue fiscal consolidation for the purpose of restoring a balanced budget.Second, they tend to implement fiscal consolidation policies based on spending cuts rather than tax increases.
The paper is organized as follows.In Section 2, I will review the literature on the effect of individual leader characteristics on public policy, with a particular emphasis on fiscal policy.In so doing, I will underline the main methodological weaknesses of previous studies.Then, I will draw from an interdisciplinary literature on the individual traits of businesspersons to derive a set of testable implications.Section 4 describes the research design.First, I will introduce the dataset on weakly exogenous fiscal adjustments and explain how and why it helps us overcome some of the methodological challenges faced in previous studies.Then, I will describe the IV strategy adopted to isolate exogenous variation in business experience.Next, I will describe the coding of business experience and the modeling strategy.Section 5 contains the main empirical results and Section 6 lists the robustness checks to be found in the Online Appendix.A conclusion follows.
"In the real world, individuals, as such, do not seem to make fiscal choices.They seem limited to choosing 'leaders,' who will, in turn, make fiscal decisions" [Buchanan 1967, p. v].
Why should individual-level characteristics matter for policy outcomes?After all, according to standard Downsian models, individual-level traits should not matter at all.Candidates respond to the median voter's preferences in order to maximize their chances of remaining in power (Downs 1957).If this is the case, policy outcomes should be independent of leaders' characteristics since all competing candidates will converge on similar policy preferences (Beach and Jones, 2016).Indeed, this is the implicit premise underlying most political economy explanations of fiscal consolidation.By emphasizing the role of economic and political conditions (Grilli et al., 1991;Hübscher, 2016) and/or domestic institutions (Roubini and Sachs, 1989;Hallerberg et al., 2007), leaders are implicitly modeled as having no independent stance towards fiscal consolidation, but only a (strong) preference for those economic policies that would solidify their political power.
Under this persepctive, leaders are "dispensible actors" in the policy process (Greenstein 1969, p. 50).Nevertheless, alternative models are not so restrictive.By relaxing the assumption of politicians as simply driven by vote-maximization and/or office-seeking goals, these models allow for the possibility that policy-makers would enact their personally preferred policies.For instance, in "citizen-candidate" models, candidates cannot credibly commit to enact policies that are inconsistent with their own personal preferences (Besley and Coate, 1997).Thus, under these models, a leader's traits matter in determining their preferences and would translate into actual policy outcomes (Beach and Jones, 2016).Nevertheless, fiscal policy preferences at the executive level have been rarely explained, with the partial exception of ideology (e.g.Moessinger 2014;Hübscher 2016).
Recognizing the potential role of individuals, a growing empirical literature has been connecting leaders' personal traits to the public policies they enact once in office (e.g.Alexiadou 2016;Gift and Krcmaric 2017).These studies tend to emphasize leaders' socializing experiences, such as education and occupation (Krcmaric et al., 2020).Within the broader literature on leaders' bi-ographical characteristics, we can situate a narrower set of studies focusing on fiscal policy and public debt.Dreher et al. (2009) were among the first to study the effect of leaders' occupational and educational background and to find some evidence that former entrepreneurs are more likely to undertake liberalizing reforms. 8Similarly, Mikosch and Somogyi (2009) explore the variation in public deficits across countries as a function of leaders' occupational and educational background.
At the sub-national level, Beach and Jones (2016) zoom in city councils in California to investigate the public policy effects of business politicians.Leveraging close elections in a regression discontinuity framework, they find no evidence that the election of a business-candidate has an impact on city expenditures or revenues.Using a similar research design, Kirkland (2021) finds that mayors with business background shift the allocation of expenditures by curtailing spending for redistributive policies.In the German context, Jochimsen and Thomasius (2014) investigate how several characteristics of regional finance ministers affect the Laenders' public deficits, but they find no effect of leaders' (non-finance) business sector experience.By contrast, Szakonyi (2020) uncovers more pernicious effects in the case of Russian sub-national governments.The author shows that former businesspersons prioritize policies that would bring immediate benefits to private firms rather than the general public, succinctly concluding that these actors run the local government "for business" rather than "like a business".Finally, Hayo and Neumeier (2016) find that leaders with higher socioeconomic status exhibits more fiscal discipline.
The above mentioned studies clearly contribute to our understanding of how individual leaders matter even in fiscal policy.Unfortunately, though, they suffer from a number of methodological and conceptual weaknesses.
First of all, some of the previous studies are quite a-theoretical in nature (with some notable exceptions, e.g.Hayo andNeumeier 2012, 2016).For example, while both Dreher et al. (2009) and Mikosch and Somogyi (2009) argue that policy-makers' distinct preferences may lie in professional (or educational) socialization, they do not theoretically elaborate on which backgrounds might be more relevant and the specific mechanisms linking individual backgrounds to fiscal pref-erences.Second, many studies have been predominantly focused at the sub-national level (e.g.Jochimsen and Thomasius 2014).While it is often desirable to study a phenomenon at a lower level of analysis, looking at sub-national units is problematic for a few reasons.To begin with, in many countries fiscal policy is still highly centralized, thus leaving little discretion to sub-national political bodies and little variation to be explained (Balassone et al., 2002).Moreover, even in the most decentralized and/or federal states (e.g.USA, Germany) there is unlikely to be as much leeway in terms of discretionary spending at the sub-national level as there is at the national level.9 In addition, sub-national public budgets are often affected by negotiations between different levels of governments (both horizontally and vertically).If politicians with business experience are particularly effective at bargaining -a claim often made in the business literature (e.g.Kerr et al. 2018) -they might end up with a larger portion of national funds than their counterparts without business experience for reasons other than their baseline fiscal preferences.Third, as several authors have noted, endogeneity concerns loom large in individual-level studies (Krcmaric et al., 2020).Two major issues arise in this context.First, (future) leaders are not randomly selected into their respective occupations.Second, the selection mechanism into political office is confounded by a host of factors that may also impact a countries' fiscal policy.To be true, some scholars have carefully adopted research designs that account for leaders' selection into political office, often leveraging close elections in a regression discontinuity framework (e.g.Beach and Jones 2016;Szakonyi 2020;Kirkland 2021).Nevertheless, all these studies focused on the sub-national level.
At the same time, the literature on individual-level traits has often ignored the first selection mechanism or, at best, has subsumed it into the theoretical explanation for the phenomenon of interest (Krcmaric et al. 2020).A good example of the latter approach is in a recent study on leaders' business background and free riding behavior in NATO countries (Fuhrmann 2020).The author suggests two mechanisms to explain why individuals with business background may be more likely to free ride on their partners: first, a business environment socializes individuals to value their own self-interest; second, individuals predisposed towards utility maximization are more likely to opt for a business career.While both mechanisms may be at play, only the first one supports a causal interpretation of business experience.
These weaknesses, in turn, might explain the abundance of null or even counter-intuitive results in the literature.For example, in the previously cited study by Mikosch and Somogyi (2009), some of the findings even contradict the authors' hypothesis that individuals with an economics degree or experience in business should correlate with a reduction in public debt.Likewise, Jochimsen and Thomasius (2014)'s work on German Laenders finds that leaders with business experience in non-finance sectors are associated with higher deficit levels.10At the same time, Beach and Jones (2016) find no evidence that the election of business-candidates has an impact on city expenditures or revenues in the case of Californian cities.Finally, Hayo and Neumeier (2016) show that upperclass leaders are generally associated with lower deficit-to-GDP ratio.Nevertheless, once they disaggregate the analysis by specific professions, businesspeople are not found to be more deficit averse.Likewise, in his book-length research of the causes and consequences of businesspersonsturned-mayors in contemporary Russia, Szakonyi (2020) finds no evidence that such policy-makers run more balanced budgets.

Theory
Why should a former businessperson have systematically different fiscal policy preferences relative to their non business counterpart?In a nutshell, I propose three main channels: first, socialization effects from working in the business sector will positively affect the individual's beliefs regarding the benefits of a balanced budget in general and expenditure-based (as opposed to tax-based) fiscal consolidation in particular; second, shared material interests with their previous professional network are likely to predispose these leaders to favor the type of fiscal consolidation that is least likely to damage business interests, i.e. expenditure based fiscal consolidation; third, former businesspeople are likely to exhibit a distinct perception of their own competence and skills, which in turn would influence their (perceived) ability to sustain the fiscal consolidation process as well as its electoral consequences.While it is not possible to disentangle these effects at the country level, they do provide a useful analytical framework that is in line with the previous literature on individual leaders' characteristics (Krcmaric et al., 2020).I now discuss each channel in turn.

Socialization
A vast body of research in social psychology has shown how individual beliefs spread through inter-group and inter-personal relations (Pettigrew, 1998).It has long been observed that the workplace affects one's own attitudes and behaviors even after accounting for self-selection, a phenomenon known as "workplace socialization" in sociology (Peterson, 1992).Such formative experiences are unlikely to be forgotten once an individual enters politics (Szakonyi, 2020).It is not just a matter of factual knowledge acquired in the workplace.Any kind of (non-trivial) occupational experience implies the internalization of the fundamental values that occupation is based on (Mikosch and Somogyi, 2009).These beliefs come to constitute individuals' cultural imprints and worldviews and, either consciously or unconsciously, inform their preferences once in a position of (political) power.In other words, during their life, individuals acquire a set of dispositions, which partly reflects their cumulative life experiences (Hayo and Neumeier, 2016).
Not unlike education, occupational experiences serve as a template for understanding and acting in the social world; experiencing a similar set of incentives, conditions, and ideational exposure will have an homogenizing effect on preferences within the same (occupational) class.In particular, working at a firm is likely to heighten an individual's perception regarding the benefits of avoiding persistent structural deficits.Former businesspeople are likely to show an overall drive towards efficiency (Szakonyi, 2020).Indeed, excessive government spending can be seen as a symptom of bad political management and lack of coherent leadership (Alesina and Perotti, 1995).As running a deficit is the public-sector equivalent to a company making a loss, fiscal responsibility is likely to become a key element of an administration led by a former businessperson.As Romney's quote in the introduction nicely summarizes it, the rule in the private sector is quite simple: if you don't balance your budget, you go out of business in the long run.
To be true, taking on a reasonable amount of debt is a precious financing tool for firms (as well as governments), but businesses are generally more sensitive to the demands of their shareholders who require sustainable profits and to shy away from uncertain investments (Szakonyi 2020).
While former President Trump has often boasted being the 'king of debt', running long-term unsustainable deficits is hardly a winning strategy for a typical firm.Indeed, the finance literature offers two main models of firms' financing decisions and its relationship with a firm's profitability -the trade-off and the perking-order model (Fama and French, 2002).Conventionally, the tradeoff theory of capital structure has been interpreted to suggest a positive relationship between firms' profitability and the leverage ratio, i.e. the amount of debt relative to equity.If that was the case, it would be hard to suggest that business experience socializes individuals to value budget balancing.If anything, business leaders may be tempted to opt for a more aggressive debt financing strategy in line with their prior knowledge on how to run a successful company.Nevertheless, the trade-off theory's prediction regarding the profitability-debt financing relationship runs counter to the established empirical fact that more profitable firms actually have a lower leverage ratio (less debt) (Abel 2018;Fama and French 2002).Indeed, a negative relationship between firms profitability and leverage ratio has been found in starkly different contexts, such as several European countries in a comparative perspective (Gebauer et al., 2018) as well as in single-country studies (e.g.Filipovic and Demirovic 2016).According to financial economist Stewart Myers "the strong inverse correlation between profitability and financial leverage" constitutes "the most telling evidence against the static trade-off theory" (Myers 1993, p. 9).In an attempt to solve the "capital structure puzzle", Myers and others developed an alternative theory known as the pecking order model of financing decisions.While a comprehensive description of this alternative model is beyond the scope of this paper, it should suffice to say that it establishes a negative relationship between profitability and leverage ratio.The key aspect is the introduction of future-oriented concerns from profitable firms, which refrain from incurring excessive debt in the present to avoid foregoing future investments and/or having to finance them at a higher borrowing cost.11Moreover, it is worth noticing that firms' financing strategies have evolved over time (Graham et al., 2015).First, there is evidence that the negative relationship between firm profitability and debt leverage has only strengthened between the 20s and the 80s before plateauing since then.Second, usage of corporate debt as a financing strategy became as prominent as it is today starting in the 70s (and only in the most advanced countries early on).Recall that, while the present study covers the 1978-2014 period, the relevant business experiences of (future) heads of the executive must have taken place earlier by definition.Indeed, assuming that individuals entered the job market upon graduation, 80% of leaders in the dataset started their career prior to 1975, which roughly coincides with the rise of corporate debt as we know it today. 12aving set the goal of a structural balanced budget in the medium-long term, how would a businessperson go about reaching that aim?Clearly, the main tool for budget balancing in the private sector is to contain costs, which is the firm-level equivalent of restraining government spending.
On the other hand, a fiscal consolidation centered around tax increases has no obvious equivalent in the private sector.While a government has the coercive capacity to increase revenues first and then (maybe) reduce spending, such luxury is rarely available to private enterprises.In addition, businesspersons are likely to have experienced the (perceived or real) constraining effects of large bureaucracies.Such experiences, in turn, might result in an overall skeptic attitudes towards big governments, in general, and large bureaucratic apparatus in particular (Witko and Friedman 2008;Albertos and Kuo 2018).For example, Italian business tycoon and former PM Silvio Berlusconi has frequently discussed the "ills" of bureaucracy (and of big government more in general) and featured the need for "less bureaucracy and friendlier fiscal policy" to stimulate business activity and economic growth. 13Likewise, in his party's manifesto, former Finnish Prime Minister and businessman Juha Sipila also proposed an economic program revolving around "tightened public sector expenditures, improvements in productivity, the removal of excessive regulations and, if necessary, a savings of two billion in spending."14

Material interest
Moving on to the interest mechanism, scholars studying individual-level characteristics have suggested that politicians are more likely to favor policies that benefit, or at least do not harm, their former industry.Indeed, the connection makes intuitive sense considering the potential for shared frames of reference, common backgrounds, experiences, and interests.Adolph (2013), for example, finds strong evidence that central bankers from the financial sector are more inflation averse.
More specifically on business experience, Witko and Friedman (2008) argue that because of different material interests former business persons legislate more favorably towards business in the US Congress.Likewise, Szakonyi (2020) and Baccini et al. (2018) show that Russian business leaders' approach to regional government intervention in the economy is strongly pro-business, as in defending the interests of existing firms.
Moreover, the effects of material interests are likely to be compounded by psychological factors that have been found to be associated with a person's background in business (Fuhrmann, 2020).Prior research categorizes two ideal-types of individuals: pro-selfs and pro-socials (e.g., Bogaert et al. 2008).While pro-selfs are rational egoists who care mostly about maximizing their own gains, pro-socials tend to exhibit more altruistic behavior.Experimental evidence has repeatedly shown how individual with business experience tend to exhibit pro-self tendencies.Such individuals are likely to display more selfish behavior than their counterparts with different occupational experiences (Baumol 1996).More recent research in economic psychology supports this view, leading Urbig et al. (2012) to conclude that selfishness is one of businesspeople's "distinctive personality characteristics."Similarly, well-known research in behavioral economics shows that business majors play ultimatum games in a more selfish fashion that their colleagues from other departments (Kahneman et al. 1986).
The interest channel -in both its material and psychological variants -is likely to play a role in former businesspersons' stance towards fiscal policy.While fiscal consolidation policies may not be directly beneficial to a country's business, at least in the short-run, its negative effects are likely to be felt in the short run.As long as tax-based and expenditure-based adjustments have a differential effects on businesses, the interest mechanism will affect the type of fiscal adjustment pursued.In particular, former businesspersons may be less likely to favor a fiscal adjustment policy based on tax hikes, which would negatively affect their former industry in the short term, at least considering the less costly alternative of reducing government spending.This line of reasoning is consistent with the previously cited studies as well as with the available survey evidence on business' preferences towards fiscal policy.For example, in the case of Spain, Albertos and Kuo (2018) find that most firms' executives support fiscal consolidation efforts, but only if focused on expenditure reduction.At the same time, the interviewees consistently rank (high) taxes as their primary concern.Likewise, survey responses of legislative candidates in US states also reveal starkly different attitudes of business owners on a variety of issues pertaining public spending, such as welfare, regulation, economic inequality, and healthcare.Business owners indicate a strong preference for a smaller role of the state in all these key policy areas (Carnes, 2018).

Self-perception
Finally, it is relevant to notice that individuals with business experience may have a distinctive way to perceive themselves as highly competent and particularly skilled at manipulating their environment (Fuhrmann 2020;Szakonyi 2020).Indeed, the psychological literature has emphasized how individuals with high levels of self-efficacy believe that they have an ability to "produce and to regulate events in their lives" (Bandura, 1982).Such individuals with a heightened sense of power have been found to be more optimistic, more risk-taking, and less concerned about loss of status (Anderson and Galinsky 2006).Importantly, individuals with business experience have been repeatedly found to display higher levels of self-efficacy and feelings of power than those without business experience (e.g., Markman et al. 2002).Indeed, we need not to go far to find anecdotal examples of such behavior.For example, former US President Trump famously claimed that he "could stand in the middle of 5th Avenue and shoot somebody and [he] wouldn't lose voters".On the other side of the pond, his British counterpart, Cameron, confidently stated his belief that he "will get what [he] want[s]" 15 in terms of a new relationship with the EU. 16  Why should a heightened sense of self-efficacy affect fiscal policy decisions?The main reason stems from the inherent uncertainty about the economic and political effects of fiscal adjustment.
To begin with, while most economists agree on the long-term beneficial effects of fiscal consolidation, there is a voluminous and unsettled debate regarding its short-term effects.If fiscal tightening dampens aggregate economic output in the short-term, it will hurt politicians' popularity (Bojar et al., 2022).Second, aggregate considerations aside, fiscal consolidation plans tend to be contentious policy issues, given their perceived distributional effects (Woo et al., 2017).To be true, the literature on the political and electoral costs of fiscal adjustments is remarkably inconclusive (Bojar, 2018).Some scholars suggest that the electoral costs of fiscal consolidation are weak or even non-existent (e.g.Alesina et al. 1998), while others find a strong association between fiscal consolidation and government support (Hübscher and Sattler 2017;Jacques and Haffert 2021).
While a thorough discussion of these debates is beyond the scope of this paper, 17 it seems reasonable to suggest that there is some degree of uncertainty surrounding the economic and electoral effects of fiscal consolidation.Moreover, it should be noticed that what really matters is the policymakers' perceptions of their own ability to withstand the short-term political and economic costs of fiscal consolidation.In that respect, anecdotal evidence also suggests that policy-makers generally perceive austerity as a political liability.As former Prime Minister of Luxembourg Jean-Claude 15 Available at https://www.theguardian.com/uk-news/2014/jan/05/cameron-cap-european-migrants-uknegotiations-eu 16Both examples offer a sober reminder that this mechanism rests on feelings and perceptions of one's own ability, which might be at odds with reality. 17The issue is particularly hard to settle because of endogeneity concerns, i.e. the possibility that only strong governments will engage in fiscal consolidation plans.Notice, though, that such methodological concern alone also implies that there should be some costs associated with fiscal consolidation: why should strong governments pursue such policies if not because they are better placed to sustain the costs of adjustment relative to weak governments?Juncker once quipped, "We all know what to do, but we don't know how to get re-elected once we have done it."18Finally, different types of fiscal consolidation policies also have heterogeneous effects in terms of economic and political costs.In particular, expenditure-based adjustments are more likely to affect a narrower constituency which is, in turn, more capable to overcome collective action problems and mount a political challenge (Alesina et al., 2020).Moreover, expenditurebased plans are harder to design, slower to implement, and less likely to generate immediate revenue benefits (Alesina et al., 2020). 19Finally, expenditure-based fiscal consolidation are more likely to increases economic inequality (Woo et al. 2017).Overall, there are strong reasons to suspect spending cuts to be more of a political liability than tax-based adjustments.Empirically, in one of the rare studies on the type of fiscal adjustments and government approval, Jacques and Haffert (2021) find exactly that: spending cuts reduce government approval, while the impact of tax increases remains minimal. 20Given the inherent uncertainties surrounding the costs of fiscal consolidation in general, and of expenditure-fiscal adjustment in particular, biographical factors that may affect the perception of such costs are important.Specifically, if leaders with business backgrounds exhibit a heightened sense of power and self-efficacy, they should generally have greater confidence in their ability to control the economic and political processes set in motion by unpopular fiscal policies, thus making fiscal consolidation and, more specifically, expenditure cuts a more attractive option.To paraphrase Juncker, they should be more prone to believe they know not only "what to do" but also "how to get re-elected" once they have done it.
To summarize, three main mechanisms -socialization, self-interest, and psychological -suggest that former business-people will be more likely to implement fiscal consolidation policies and, in particular, expenditure-based plans.As such, two observable implications can be derived: • Hypothesis 1: Compared to leaders without business experience, leaders with business experience are more likely to opt for fiscal adjustment policies.
• Hypothesis 2: Compared to leaders without business experience, leaders with business experience are more likely to opt for expenditure-based fiscal adjustment policies.

Weakly exogenous fiscal consolidations and self-selection into office
As mentioned before, virtually all previous studies on individual level characteristics and public debt have focused on raw measures of deficit or debt-to-GDP ratios or their sub-national equivalent (e.g.Mikosch and Somogyi 2009;Jochimsen and Thomasius 2014;Moessinger 2014;Hayo and Neumeier 2016).The obvious problem with this approach is that the economic business cycle might affect both the country's fiscal position (and, hence, its baseline probability of engaging in fiscal consolidation) and its electorate's preference for business politicians.Worse still, several Western countries rely on automatic stabilizers (e.g.unemployment schema, food stamps, changes to income tax base etc) designed to offset the negative welfare effect of the business cycle by temporarily increasing the government deficit and without additional authorization by policymakers.
As such, policymakers in power at the trough of the business cycle might correlate with changes in government finances without having done anything at all.
Leveraging close elections in a regression discontinuity framework is by far the most common approach to account for the political selection endogeneity (Beach and Jones, 2016;Szakonyi, 2020;Kirkland, 2021).Unfortunately, though, a regression discontinuity design requires a large number of observations as well as detailed biographical information about both winners and losers in each election.Moreover, it is not exactly clear how to leverage the close election designs in parliamentary systems and, especially, in non-majoritarian contexts (e.g.Italy, Germany).As such, it is ill-suited for broad cross-country comparisons. 21The recent macroeconomic literature, though, suggests an alternative strategy.Instead of directly modeling selection into office, it is possible to focus on a subset of fiscal consolidation policies that are orthogonal to the business cycle.
With this goal in mind macroeconomists have proposed the so-called "narrative approach".
The idea is to rely on a careful assessment of historical documents to identify fiscal shocks that are plausibly exogenous to the economic cycle.Such historical approach is based on the following steps (Devries et al. 2011;Guajardo et al. 2014;Alesina et al. 2020).First, verifying that the policy documents do not discuss a desire to respond to current or prospective economic conditions.
Second, within that subset of policy changes, the authors classify the official motivations stated (e.g.reducing the budget deficit; to spur long-run growth; to restrain an overheating economy etc).The events are classified qualitatively based on the most relevant policy documents in each country and/or international organizations, including Budget Reports, Budget Speeches, OECD Economic Surveys and IMF Staff Reports.The documents must 'explicitly provide evidence of what policymakers believed at the time that the decisions were taken' (Devries et al. 2011, p.4).
Specifically, for the purpose of this paper, I will focus on (weakly) exogenous changes motivated by a desire to reduce the budget deficit.Finally, the authors track the policy changes to make sure that they were not reverted by subsequent governments and run formal econometric tests to assess the extent to which the policy changes can be predicted by current macroeconomic factors (Alesina et al., 2020).
The above discussion notwithstanding, the narrative approach is no panacea.Indeed, it is important to notice that this approach identifies fiscal plans that are weakly exogenous. 22Naturally, a desire to reduce public deficit stems from the past deterioration of public finances.In turn, a deteriorated financial position is likely to affect other macroeconomic factors, which may be correlated with the probability of having a business politician as the Head of the Executive (Guajardo et al., 2014).For this reason, it will still be necessary to control for the past realizations of relevant macroeconomic variables to account for the remaining sources of endogeneity.Nevertheless, as the models later will show, most macroeconomic variables will remain statistically insignificant in virtually all models, while the variables capturing the country's (past) deficit and debt will remain significant.This is exactly what we would expect if the fiscal consolidation was motivated by a desire to reduce previously accumulated public debt, but its timing was orthogonal to the business cycle.Indeed, this is consistent with the findings in Alesina et al. (2020) where the authors discuss at length the degree of predictability of the fiscal plans.This limitation notwithstanding, the narratively identified fiscal policies constitute a major improvement relative to previous studies not only on methodological grounds, but also conceptually.Indeed, unlike continuous measures of budget deficits, these discrete events reflect the discretionary and deliberate decisions by governments on fiscal policy.Uncorrelated with macroeconomic contemporaneous factors that may confound causal identification, such a measurement is ideal to explore the actual impact of non-economic factors on policy changes (Hübscher, 2016).
In particular, I rely on the most comprehensive and updated available dataset, which Alesina et al. ( 2020) constructed based on the previous work of Devries et al. (2011).The dataset comprises 17 OECD countries for the period 1978-2014. 23In order to select (weakly) exogenous fiscal consolidation measures, the authors searched for "clear sentences [...] that attributed measures either to the aim of correcting the dynamics of some budgetary item (such as pension reforms aimed at reducing outlays), or to the aim of addressing the dynamics of the debt over GDP ratio, or the deficit" (Alesina et al. 2020, p.79).As mentioned above, I focus on fiscal changes motivated by a desire to reduce the budget deficit or, to use the authors' terminology, to address the dynamics of public finances.The authors code both fiscal "plans" and "episodes".The former refers to the initial government's decision to fiscally consolidate in the future (fiscal plans can be multi-year), while the latter refers to each year of fiscal adjustment as it actually took place. 24Moreover, the authors code the size of fiscal consolidation that a given plan/episode entails, broken down by type (taxes or spending).Based on the relative amount of spending cuts and tax hikes involved, each fiscal consolidation plan/episode is then coded as expenditure-based (EB) or tax-based (TB) (the two are mutually exclusive). 25he empirical analysis will focus on fiscal episodes for three main reasons (see Online Appendix for the results using plans).First, multi-year fiscal plans can start in one phase of the economic cycle (e.g.recession) and end in another phase of the economic cycle (e.g.expansion).
By focusing on episodes, instead, the analysis encompasses the complete phases of the business cycle, something that would be missed with plans (Alesina et al. 2020).Second, the coding of fiscal plans is a bit ambiguous since it includes both genuinely new plans as well as substantial modifications to previously announced plans (see Alesina et al. 2020, p. 74).Unfortunately, it is not immediately obvious how to distinguish between the two in the dataset.Since we cannot exclude the possibility that business leaders are systematically more (or less) likely to modify previous plans relative to their non business counterpart, relying on plans might bias the results in an unknown (and unknowable) direction.Third, and most importantly, relying on episodes seems more consistent with the data generating process.Fiscal authorities decide not only on the initiating fiscal adjustment, but also on its continuation, modification or ceasing on a year-by-year basis.The decision to focus on fiscal episodes is also consistent with previous studies on (weakly exogenous) fiscal consolidation policies in political science (e.g.Hübscher 2016).

Instrumental Variable approach and self-selection into occupation
Clearly, (future) leaders self-select into their respective occupations.This makes it difficult to determine whether the observed relationship is driven by the individuals' actual experiences or whether individuals simply select into occupations that are consistent with their personal characteristics (Krcmaric et al., 2020).If the latter, a leader's occupation might be helpful to predict behavior, but is are not the true source of causality.While this problem is pervasive in leaders' studies, it is often only briefly acknowledged and/or assumed away suggesting that both self-selection and socialization are at work (e.g.Fuhrmann 2020).
A possible solution is to rely on a plausibly exogenous source of variation that affects a (future) leader's decision to start a business career without directly affecting a country's fiscal policy when the former businessperson actually becomes the Head of the Executive.This is, essentially, the realm of instrumental variable estimation.To search for possible instruments, I turn to the historical macrofinancial record.The intuition underlying the strategy detailed below is that economic factors at the time when (future) leaders first entered the job market directly affected the individuals' decision to purse a career in business.At the same time, though, these factors are unlikely to affect the country's fiscal policy positions decades after, thus satisfying the exclusion restriction.
In particular, I rely on a recent macrofinancial historical database that covers 17 countries over the last 150 years (Jordà et al., 2017). 26In particular, I propose to use the following two instruments -the gross fixed capital formation (as a % of GDP) and the total loans to the non-financial private sector (as % of GDP). 27The former (also known simply as investment-to-GDP ratio) shows how much of the new value added in the economy is being re-invested (rather than consumed).
Fluctuations in this index are often used to infer future business activity, business confidence, and the pattern of economic growth.A rise in investment should contribute towards higher aggregate demand and more dynamic economic activity.Similarly, the total private sector borrowing is a major determinant of domestic financial stability which, in turn, increases business confidence and economic growth.It also proxies for the the banking system's liquidity which, in turn, affects business decisions.Given the lack of detailed and systematic data on when (future) political leaders first entered the job market, I rely on two versions of the variables described above: first, I use a 10-year average of each measure, starting from the year when the (future) leader turned 20 years old.The problem with such a long window, though, is that it reduces variation in the instruments (and, hence, their strength).For example, two leaders from the same country who were born five years apart share 50% of variation in the instruments, even if they may have faced fairly different job market conditions.As a second version of the instruments, I manually collected data on each leader's graduation year (high school or college) and compute the 5-year average of the two measures, starting from the year when the (future) leader graduated. 28hree conditions must be met for IV estimation.First, the instruments need to have a causal effect on the treatment.It seems reasonable to expect both instruments to be significantly and positively correlated with the probability of starting a business career.The investment-to-GDP ratio captures the dynamism of the economy, which is likely to affect the starting of new businesses or of new employment opportunities in existent business activities.Likewise, the private loans to GDP ratio captures the degree to which business and would-be businesspeople may face liquidity constraints.On both concepts, and how they relate to business development and employment opportunities, there is a well developed literature in economics (Uusitalo 2001).Moreover, the first condition can be verified empirically: section 5.2 will show that the instruments pass standard tests of relevance, while Appendix G shows the first-stage relationship between the instruments and business experience to be positive, as theory would suggest.Second, the instruments should be exogenous, i.e. uncorrelated with the leaders' unobservable characteristics that may affect their fiscal preferences.This condition is also likely to be met as individuals cannot choose where they were born -which is usually where they start their careers -and clearly cannot affect domestic financial conditions at the macro level.This requirement would be violated if individuals in their 20s selfsorted in different countries as a function of the instruments and with the goal of starting a business career.Nevertheless, this behavior is unlikely to be prominent.To begin with, the transaction costs of starting a business abroad (e.g.regulation concerning loans to non residents) tend to be high and are unlikely to be offset by moderate variation in the domestic financial environment of each country.Moreover, as shown before, most leaders in the dataset entered the labor force before the mid 70s, when labor mobility was much lower than today (for example, the Schengen Agreement, which guarantees EU citizens the right to move to and work in any EU country, was signed in 1985).The third and most demanding condition for IV estimation is the exclusion restriction, i.e. the instruments can affect the outcome only via the treatment.This condition is also likely to be met.To be sure, investments as well as private loans can be contemporaneously related with the government deficit and debt which, in turn, affects the likelihood of fiscal consolidation.An increase in the central government deficit may discourage investment decisions and increase the cost of borrowing for the private sector due to the higher risk of a government default (de Mendonc ¸a and Brito, 2021).Nevertheless, the key aspect of my IV strategy is that the instruments capture macrofinancial country-level factors at the time when the leader first entered the job market, hence much prior to their decision to run for office.Indeed, the leaders in the dataset became Heads of the Executive 32.7 years after first entering the job market, on average (assuming leaders entered the job market upon graduation).Considering that the average business cycle in the mecrofinancial dataset lasts 6.6 years (see Jordà et al. 2017, Table 5), this implies that the typical country will have experienced five complete phases (expansions and recessions) by the time the leader takes office.
It seems extremely unlikely that the two variables -which capture macrofinancial aspects from five business cycles prior -could be directly related to a country's fiscal position more than three decades later.In theory, they might be related indirectly if the two variables were highly persistent over time, so that their realizations at t-30 may still contain information about their realization at time t.To check the plausibility of this counterargument, I separately regress the contemporaneous and past realizations of investment-to-GDP and private borrowing ratios on the deficit-to-GDP ratio controlling for all the other variables in the complete model (see Section 4.4).Then, I run a partial F test under the null that the two coefficients are jointly zero.Unsurprisingly, when the two variables enter the equation contemporaneously, they are jointly statistically significant at the 5% level, confirming the known empirical results in the literature (e.g.Jordà et al. 2017).By contrast, when I use the realization of the two variables at the time when the leaders entered the job market, the two coefficients are indistinguishable from zero regardless of whether I use the the 10-year averages upon turning 20 years old or the 5-year averages upon graduation.

The measurement of Business Experience
For practical and theoretical reasons, I concentrate on the Heads of the Executive.These are the most individually powerful decision-makers in the executive branch of government and, as shown in the literature review section, exert the most significant influence on government's policy.While the individual characteristics of Finance Ministers may also have an effect on fiscal consolidation policies, the Prime Minister or the President is the one proposing (or directly appointing) the Ministers (Dewan and Hortala-Vallve 2011).Indeed, Finance Ministers usually make policy proposals that align with the the guidelines set by the leader of the executive.Consistent with this view, the whole literature on the "strength" of the Finance Minister assumes that a strong finance minister is the one that can reliably act as a "faithful" agent of the Head of the Executive (Hallerberg et al. 1999;Hagen 2002;Hallerberg et al. 2007). 29As Hallerberg et al. (1999) put it, "If the prime minister prefers that the party's ideal budget be reached, which should usually be the case, she will have identical preferences on the budget as the finance minister" (p.217).30Indeed, the strength of the minister is usually conceptualized in terms of proximity to the the Head of Executive and operationalized accordingly.For example, Jochimsen and Thomasius (2014) consider a finance minister to be strong if he/she belongs to the same party as the prime minister.Moreover, spending priorities are often viewed as a leading example of how the leader provides direction to individual ministers (Dewan and Hortala-Vallve, 2011).Focusing on the leader of the executive also makes the results more comparable with the previous studies on individual leaders and fiscal policy (Dreher et al., 2009;Mikosch and Somogyi, 2009;Hayo and Neumeier, 2016).
Hence, the main explanatory variable is a dummy that takes the value of 1 for each country-year observation where the head of the Executive has business experience.All mechanisms suggested in the theory section implicitly rest upon the assumption that the individual has a high-level business experience.A non-trivial position in the company seems necessary for the occupational experience to influence the individual's fiscal policy preferences.Moreover, public enterprises tend to perform less efficiently and less profitably.Committed to a broader mission than profit maximization, they are also likely to differ in terms of goals, business approaches, and, consequently, socialization processes.As such, two conditions must be met for the leader to be coded as having business experience: first, the (future) politician must have played a key role in running and/or managing a firm; second, they mush have worked at a private company.With this goal in mind, I rely on Fuhrmann (2020) who recently coded all NATO leaders' occupational experiences up to 2014.
A leader is coded as such if he/she "established or owned a company, worked as a firm's chief executive officer, served on a corporate board of directors, or worked as a senior manager or an executive" (Fuhrmann 2020, p.8).As a starting point for the countries that are not in Fuhrmann's sample (i.e.non-NATO countries), I rely on the LEAD dataset, which provides biographical information up to 2004 (Ellis et al., 2015).Unfortunately, the LEAD dataset does not provide the most appropriate coding of business experience for the purpose of the present study.For example, some leaders in the dataset worked for state-owned public companies, while others did not hold high-level positions.As such, I re-code the original business experience variable according to Fuhrmann's definition. 31To do so, I rely on the original sources consulted by Ellis et al. (2015). 32  Where I could not find the information needed, I complemented the search with additional primary as well as secondary sources (academic books and articles, newspaper articles, obituaries, online encyclopedias, and national government websites).
Figure 1 below shows the final result of the data-collection phase.The red squares indicate the presence of a leader with executive/managerial business experience in the private sector in a given country and year. 31Fuhrmann also distinguishes between private and public sector experience.Hence the coding scheme is the same. 32I thank Michael Horowitz for sharing the detailed bibliographical sources used for the original dataset.
The main dependent variable for Hypotheses 1 is a dummy variable that takes the value of 1 for a given country-year observation if a fiscal episode took place.For hypotheses 2, the dummy takes the value of 1 if the fiscal episode was expenditure-based.α i is a country-specific intercept, µ t are year-fixed effects, and ε i,t the error term.X i,t−1 is a vector of lagged macroeconomic variables, and X i,t is a vector of contemporaneous political variables.The coefficient of interest is β 1 .
While the narrative approach's goal is to disentangle fiscal consolidation from the business cycle, it is still desirable to control for a battery of past macroeconomic variables that may affect the probability and type of fiscal consolidation.Not only this allows us to account for the remaining endogeneity, but also to check if the weak orthogonality assumption is met.While fiscal adjustments may be predicted by past economic data, mostly the deficit and the debt, most macroeconomic variables capturing the business cycle should be at best weak predictors of fiscal consolidation (Alesina et al. 2020, ch. 12).Hence, I control for standard macroeconomic variables that are known to affect public finances (e.g.Hayo and Neumeier 2016).In particular, I include the deficit and the interest payments on government bonds (both as % of GDP) from the OECD Economic Outlook n. 97, the debt-to-GDP ratio from the IMF Historical Public Debt Database, the log of real GDP per capita (OECD Historical population data and projections 1950-2050), the unemployment rate, the inflation rate, and the real GDP growth rate (all from the OECD Economic Outlook n. 97), and trade openness as % of GDP (Armingeon et al., 2020). 33hen, I add a battery of political variables that may act as confounders.First, I include a variable capturing the government's ideology of the government (Database of Political Institutions).
Second, I control for election years accounting for the potential influence of the political business cycle (Alesina et al., 1992).Importantly, the literature on fiscal consolidation has often emphasized the importance of "government strength" to explain governments' self-selection into fiscal adjustments (e.g.Alesina et al. 2020).Moreover, given the nature of the research question, it is important to account for possible constraints on the leader's power to affect the public budget (Hayo and Neumeier, 2016).As such, I add a dummy to control for whether the head of government's party has a majority in all houses and a variable that captures the degree of government fractionalization (Database of Political Institutions); a variable capturing the seat share of government parties (Seki and Williams, 2014); and a veto-player index (Henisz, 2000).Finally, I include a dummy variable that takes on the value of 1 from the year a country committed to the Maastricht criteria and a dummy for proportional electoral systems, that have been found to be more prone to deficit spending (Grilli et al., 1991).
5 Empirical Results

Baseline results
To test my two hypotheses, I estimate a set of logit fixed-effects models.To ease concerns about suppression effects of the main variable of interests due to the inclusion of control variables, I include the covariates sequentially.Moreover, to account for temporal dependence between events, I include the cubic polynomial approximation of spell-time (Carter and Signorino, 2010).Model As Table 1 shows, hypothesis 1 is by and large confirmed.Business experience remains positive and statistically significant across all specifications.Substantively, the full model predicts an increase of 27 percentage point in the probability of a fiscal episode for countries with a business leader as their Head of the Executive.The effect is sizeable.For comparison, consider that an increase of the deficit ratio by 3 percentage point (the limit explicitly mentioned in the Maastricht Treaty in the 90s for joining the eurozone) increases the probability of fiscal consolidation by 24 percentage points.Importantly, we should notice the lack of statistical significance of most economic variables capturing the business cycle.This nicely shows the strength and validity of the narrative approach to identify fiscal consolidation policies.The political variables, instead, have considerable predictive power in all models.A similar pattern arises in the rest of the models.34Clustered Standard Errors in parenthesis.* p < 0.10, * * p < 0.05, * * * p < 0.01 One possible objection is that, while business leaders consolidate more often than their nonbusiness counterpart, they may also enact smaller fiscal adjustment packages.This, in turn, would cast doubts on the substantive implications of these findings.As I show in Appendix A, though, this is not the case.Actually, business leaders tend to enact larger fiscal consolidation packages and deeper government spending cuts, although the effect is small and not statistically significant.

Instrumental Variables
While the results above show a solid and substantively meaningful effect, they may not have a causal interpretation.In particular, individuals endowed with unobserved characteristics that may affect fiscal preferences in a conservative direction might self-select into a business career.If that is the case, the experience of business would not be causing fiscal preferences, while having a business career would still be a strong predictor of fiscal policy.To disentangle the effect of self-selection into occupation due to unobservable characteristic from the effect of experiencing a business career, I rely on quasi-exogenous variation in the probability of starting a business career.Table 3 and 4 show the results using the instrumental variable approach described in Section 4.2 for fiscal consolidation episodes and expenditure-based adjustments, respectively.To save space, I only show the full models with each instrument separately (columns 1-4) and combined (columns 5-6).Appendix I shows the models without year-fixed effects. 35As we can see, the results indicate a strong and significant effect of the exogenous part of business experience induced by the instruments.The Kleibergen-Paap F statistics is above 10, thus satisfying the Staiger and James (1997) rule of thumb value.The Hansen J statistics further suggest the over-identifying restrictions to be valid.As it is often the case, the IV-2SLS estimates are larger than the comparable marginal effects from the main models.Indeed, we should bear in mind that IV-2SLS yields a weighted-average local causal effects (LATE) across all instrument-specific compliant sub-populations.Loosely speaking, then, the results in Table 3-4 refer to those individuals who chose a business career because of the liquidity and/or investment conditions in their countries at the time when they entered the job market and who would have chosen a different career path had those macrofinancial characteristics been different.By assumption, the individuals in this sub-populations are less likely to exhibit the unobserved characteristics that may correlate with pursuing a business career.Thus, for these (future) leaders the actual experience of a business career should have a larger effect on their attitudes and preferences.Consistent with this view, the coefficients capturing the local treatment effect for the compliant populations are larger and may not be representative of the average treatment effect in the overall population.Bearing these limitations in mind, though, such approach is superior to ignoring the issue of self-selection into occupational experience.In summary, the estimation results support the two hypotheses.Leaders with business experience are less debt-tolerant and more likely to achieve a balanced budget via spending cuts rather than tax increases.

Robustness checks
To check the robustness of the results, I modify the baseline specification in several ways (all results in the Online Appendix).
First, I repeat the analysis using fiscal plans instead of episodes as the dependent variable (Appendix B).Second, fiscal years begin and end at different times across countries (see the documentation in Alesina et al. 2020).In some cases, fiscal adjustments are packaged within the budget laws prepared in the Fall for the next year.Thus, I rerun the main models without lagging the independent variable (Appendix C).Third, fiscal adjustments tend to cluster in time.While I have been controlling for temporal dependence in all models, the lagged dependent variable may have a further direct effect on the probability of adjustment.As such, I augment the main models with the past realization of fiscal episodes (Appendix D).Fourth, since unconditional logit mod-the models using the conditional logit estimator (Appendix E).Fifth, I follow previous studies on individual-level characteristics and cluster at the leader-level (Gift and Krcmaric, 2017).Moreover, if global/regional economic crises affect multiple countries' fiscal position at the same time, the unobserved components of outcome might be correlated not only within countries, but also within years.As such, I use Cameron et al. (2011)'s multiway clustering variance estimator at the country-and year-level (all the models with different standard errors are in Appendix F).Finally, in Appendix H, I re-run the instrumental variable regressions using the Limited Information Maximum Likelihood method, which is known to be more robust in the presence of weak instruments (Hahn et al., 2004).36

Conclusion
Political economists often assume that the only genuine preference of politicians is to behave opportunistically to maximize their probability to remain in power.As a result, executive-level fiscal preferences have received scant attention in the literature, with the partial exception of ideology.
This paper employs a different approach.Combining insights from sociology, political science, psychology, and business studies, I draw a connection between politicians' experience in the business sector and fiscal preferences.I hypothesize that businesspoliticians may be more prone to enact fiscal consolidation -and to rely more heavily on expenditure cuts than tax hikes -due to socialization effects, material connections, and psychological factors.I test the two hypotheses using data from 17 OECD countries over the 1978-2014 period.Methodologically, I ease the endogeneity concerns that have plagued previous studies in two ways.First, I focus on a subset of fiscal consolidations that are weakly exogenous to the business cycle.Second, I rely on an instrumental variable (IV) approach to isolate exogenous variation in the (future) leaders' decision to enter the business world at the time when they first entered the job market.The results are theory-consistent

Figure 1 :
Figure 1: Heads of the Executive with Business Experience

1
shows the simple bivariate relationship with time and country fixed effects.Model 2 includes the economic variables and Model 4 adds the political variables.Given the recently debated limitations regarding the estimation and interpretation of two-way fixed effects with staggered and discontinuous treatment (e.g.Kropko and Kubinec 2020), I also show the complete model without year fixed effects (Model 3).To save space, I will not show the covariate coefficients in subsequent tables, although I will always show the same 4-model structure.

Table 1 :
Logit Fixed Effects Models -Fiscal Consolidation p < 0.10, * * p < 0.05, * * * p < 0.01 Moving on to Table2, we see that hypothesis 2 is also confirmed.Substantively, Model 4 predicts a 9 percentage points increase in the probability of enacting an expenditure-based fiscal adjustment when the country's leader is a former businessperson.The effect is roughly equivalent to an increase in the deficit ratio by 2 percentage points. *