Toxic Loans and the Rise of Populist Candidacies 

Get Complete Project Material File(s) Now! »

Toxic Loans and Local Economic Activity: Evi-dence on the Costs of Public Debt Shocks FIRST DRAFT


This paper provides first evidence on the effects of local public debt shocks on economic ac-tivity for highly indebted local governments. It exploits two exogenous shocks on local public debt affecting French municipalities exposed to the Swiss Franc: i) an information shock in 2011 signaling the degree of their debt toxicity, and ii) a sudden rise in their debt repayment following the 2015 unpeg of the Euro/Swiss Franc. Using a difference-in-difference strategy and relying on plant-level data, I first find that negative press coverage on local public debt is sufficient to temporarily decrease the total number of hours worked and the wage bill of plants located in CHF-indebted municipalities. Compared to an information shock, an actual increase in local public debt triggers strong and persistent consequences on local economic activity, by increasing plant closures in highly-indebted municipalities. Overall, these findings suggest that plants forecast a rise in local business taxes or a decrease in public investment in impacted municipalities.


Can public debt impact economic activity? The question has long been debated in the scientific literature, starting with Ricardo (1817) and Barro (1979). It has made the head-lines again during the world-wide Great Recession, when many countries experienced strong rise in public indebtedness and feared, as a result, consequences on their eco-nomic activity. While the relationship between government debt and economic growth has been extensively studied, both through theoretical and empirical studies, there is surprisingly scarce evidence on the potential impact of local public debt on economic activity (Sauvagnat and Vallée 2021).
Improving our knowledge on local public debt appears relevant for several rea-sons. First, among developed countries, many local governments are currently highly indebted, potentially affecting local economic activity. In the U.S. for instance, state and local governments have experienced a sharp increase in the level of debt securities and loans between 1990 and 2020 – from around $1.65 trillion in 1990 to more than $8.30 trillion in 2021 . In European countries, local government debt is non-negligible as well – with a total amount of 867.4 billion, representing more than 7% of the EU27’s general government gross debt in 2020 . Second, as most of the literature has focused on central government debt, it has missed the particular features of local government debt and their potential implication for local economic activity. Unlike central govern-ments, local governments often have to balance their operating budgets every year, as it is the case in France or in the U.S. In presence of budget shortfalls, they could be force to react rapidly – either by increasing their local taxation, or by decreasing their expenditure or even by renegotiating their debt. As a result, local economic condi-tions could be directly impacted by an increase in public debt. Last, by focusing on local public debt, empirical studies can better identify exogenous public debt shocks, control for time-variant country characteristics and deal with a large number of obser-vations. Here, the main empirical challenge is to find an institutional setting where the increase in debt burden does neither depend on local government characteristics nor on their local policies.
This paper overcomes these challenges by exploring a unique setting of exogenous local debt increase. Between 1996 and 2011, around 1,500 French municipalities con-tracted more than 3,000 structured loans with Dexia bank for a total amount of 8.94 billion. Through these financial contracts, municipalities have to repay fixed interest rates for the first years (on average between 2 and 7 years) before relying on variable interest rates indexed on international underlying assets for a longer period of time. Among municipalities indebted with Dexia structured loans, 17% are exposed to the Swiss Franc foreign exchange rates (usually, the Euro or the Dollar). This study exploits this design by comparing the local economic conditions of CHF-exposed municipali-ties with non-CHF-exposed municipalities, indebted with other underlying assets. It relies on two exogenous events that differently impacted municipalities depending or not on their CHF-exposure: i) the disclosure of the Toxic Loan scandal in September 2011, and ii) the unpeg of the Swiss Franc from the Euro in January 2015. The disclo-sure of the Toxic Loan scandal can be viewed as an information shock on the debt stock of CHF-exposed municipalities – independently from their characteristics or their local policies. It released the name of involved municipalities, the name of the contracts (in-dicated their exposure or not to the Swiss Franc), and the excess interests theoretically incurred by borrowing municipalities. Importantly, excess interests were announced as particularly large for CHF-exposed municipalities, due to the strong appreciation of the Swiss Franc between mid-2008 and mid-2011. Nevertheless, the 2011 disclo-sure remains purely informational. In other words, there is no significant difference in terms of actual debt stock or even actual debt repayment between CHF-exposed and non-CHF-exposed municipalities, as many municipalities postpone their debt repay-ment in the immediate aftermath by renegotiating their loans or challenging them in court. In addition to the 2011 disclosure, this paper explores another event, which is also orthogonal to municipal decisions or policies. In January 2015, the Swiss National Bank decided to unpeg the Swiss Franc from the Euro leading to a strong and sudden appreciation of the Swiss Franc. This event negatively impacts the debt of French mu-nicipalities exposed to the CHF – leaving unchanged debt repayment and debt stock in other municipalities. While the 2011 disclosure should be seen as information shock on public debt that negatively exposed CHF-indebted municipalities among municipal-ities contracting with Dexia, the unpeg of the Swiss Franc in 2015 could be interpreted as an actual debt increase on CHF-exposed municipalities.
To identify and compare the economic impact of an adverse information shock on public debt from an actual increase in public debt, I rely on the hypothesis that con-ditional on contracting toxic loans, municipalities indebted with the Swiss Franc are similar to municipalities exposed to other underlying assets. Two points are consistent with this hypothesis. First, there is no evidence suggesting that Dexia bank selected particular municipalities to contract with on the Swiss Franc. As shown by Cori and Le Gall (2013), structured loans were instead considered as a sort of insurance against financial volatility – with municipalities themselves insuring Dexia on a variety of fi-nancial assets. Second, CHF-loans were not seen ex-ante as riskier financial products among other structured loans. The Swiss Franc was particularly stable before the mid-2008, meaning that neither the bank nor the municipalities had any prior when they selected the Swiss Franc instead of another underlying asset. In line with the identifi-cation assumption, I do not find any large difference on observable variables between CHF-exposed municipalities and non-CHF-exposed municipalities.
Combining a Difference-in-Difference design with an Instrumental Variable strat-egy used by Sauvagnat and Vallée (2021), I rely on administrative and collected panel data at the plant and at the municipality level, to explore respectively the causal impact of negative press coverage on public debt and, the causal impact of an actual debt in-crease on local economic activity. In this paper, I make the two following contributions. First, I find that a negative information shock on local public debt can be sufficient to trigger slight change on local economic activity. In municipalities adversely impacted by the information shock, I observe a slight decrease in the number of hours worked and in the total wage bills in 2012, suggesting that plants located in these municipali-ties temporarily react to the information by adjusting their intensive margin. Second, I show that a sudden increase in local public debt leads to a significant rise in plant closures in affected municipalities. In terms of magnitude, a rise by 100 in munici-pal debt per inhabitant is associated with an approximate increase in plant closures by 2.8% per 10,000 inhabitants. The effect is strong and persistent – with CHF-exposed municipalities still experiencing relatively more plant closures in 2019 compared to their counterparts. Importantly, a rise in local public debt is sufficient to impact local economic activity – even in the absence of an increase in local tax rates or a decrease in municipal investment. Despite municipal efforts to mitigate the economic impact of local public debt, plants still forecast worse economic conditions in impacted mu-nicipalities. Local public debt appears therefore as an important driver of economic activity.
This paper is related to several strands of the literature. First, it relies on the large literature studying public debt and its economic impact. Earlier studies either focus on Ricardian equivalence (Leiderman and Razin 1988; Evans 1988, 1991) or on the channels through which public debt can negatively affect economic activity – such as sovereign risk spillovers (Corsetti et al. 2013), higher distortionary taxation (Barro 1979; Dotsey 1994), lower public expenditure (Aizenman, Kletzer, and Pinto 2007) or higher inflation (Cochrane 2011). Since then, several empirical studies have emerged on the impact of public debt on economic activity. One strand of this literature relies on cross-section variation to estimate local fiscal multipliers (Cohen, Coval, and Malloy 2011; Chodorow-Reich et al. 2012; Nakamura and Steinsson 2014; Suárez Serrato and Wingender 2016). Other papers focus on the effects of public debt on economic growth using cross-country comparison with usually mixed results; such as Schclarek (2004) on industrial and developing countries or Reinhart and Rogoff (2010) on 20 developed countries. At the local level, evidence is particularly scarce – even if it would provide better identification settings by controlling for endogenous shock at the national level. To the best of my knowledge, there are only three empirical studies focusing on local public debt: Cornaggia, Cornaggia, and Israelsen (2018) study the impact of credit rat-ings on municipal bond prices; Sauvagnat and Vallée (2021) use the toxic loan setting to explore how a local debt increase affects municipal budgets and electoral chances of incumbent mayors; and Adelino, Cunha, and Ferreira (2017) look at how Moody’s recalibration expands debt capacity and local government spending to improve local economic condition. This paper contributes to this literature by investigating instead for the first time, using plant-level data, how a high level of indebtedness incurred by local government can negatively affect local economic activity.
Second, this paper is related to the recent literature using quasi-experimental set-tings to identify the causal impact of indebtedness on various economic agents. Agar-wal et al. (2017), Di Maggio et al. (2017), Ganong and Noel (2019) explore exogenous debt payment reductions to show that borrowing households are more likely to ex-perience a lower probability of default and to increase spending on durable goods. Gilje (2016) uses quasi-random shocks to identify the causal impact of an increase in firm leverage on risk-shifting.2 Closer to my empirical setting, Verner and Gyöngyösi (2020) exploit variation in exposure to household foreign currency debt as an exoge-nous shock on household debt burden. They find a decline in local demand, with negative spillover effects. This article contributes to this literature by exploring for the first time a quasi-experimental design disentangling an exogenous information shock on public debt from an actual exogenous increase in public debt.
Finally, this paper relies on an emerging literature investigating the aftermath of highly-risky financial innovation. Pérignon and Vallée (2017), Gyongyosi and Verner (2020a), Sartre, Daniele, and Vertier (2020) explore the political aftermath of financial innovation – studying its impact on populism or its consequences on the likelihood of reelection for the incumbent. Sauvagnat and Vallée (2021) analyze its effect on mu-nicipal budgetary outcomes. This article focuses instead on the effects of highly-risky financial innovation on local economic activity.
The remainder of the paper is organized as follows. Section 2 presents the insti-tutional setting and the data. Section 3 describes the empirical strategy. Section 4 investigates the economic impact of an information shock on local public debt. Section 5 explores the economic aftermath of an actual increase in local public debt. Section 6 concludes.

READ  Function to smooth the RODTOX signal and to obtain the first and second derivative

Institutional setting and Data

Institutional setting

Toxic Loans and Dexia Bank
Many European local governments had relied on structured products before the Global Financial crisis (Sauvagnat and Vallée 2021). In France, structured loans were par-ticularly widespread, involving several layers of local governments (such as regions, French departments, establishments of inter-municipal cooperation and municipali-ties). In 2011, 15% of the debt of French municipalities were contracted through struc-tured loans. At that time, Dexia bank was the main source of funding for the public sector (Cour des Comptes 2013) and granted a large share of structured products. Ac-cording to an investigation report written by the National Assembly (Bartolone and Gorges 2011), Dexia had a 70% market share for structured loans in 2011.
Before being labelled as “Toxic Loans” by the media, structured loans were specula-tive instruments. Local governments enjoyed low fixed interest rates for the first years before relying on variable interests, indexed on financial underlying assets. The av-erage maturity was around 19 years but could sometimes go up to 50 years. There were various types of underlying assets, such as inter-bank offered rates, constant maturity swap, inflation or foreign-exchange rates. Among municipalities indebted with structured products, around 17% were exposed to CHF exchange rates (usu-ally the Euro/Swiss Franc or the Dollar/Swiss Franc exchange rates). As detailed by Sauvagnat and Vallée (2021), given the structure of the contracts, an appreciation of the CHF against another currency of 0.01 usually raises the interest rate by 0.5% to 1% when the option is in the money. Due to the long maturity of their loans, CHF-indebted municipalities have a long-lasting exposure to the CHF appreciation.
At the time the contracts were signed, there was no evidence suggesting that CHF-exposed municipalities differed from non-CHF exposed municipalities, conditional on contracting toxic loans. First, CHF exchange rates were considered as particularly sta-ble before the financial crisis. Thus the subsequent appreciation of the Swiss Franc was not forecast ex ante, neither by the bank nor by the municipalities. Importantly, a large part of CHF-exposed municipalities contracted on the Swiss Franc before 2008. Only 12% signed for a loan exposed to CHF after 2008. It confirms that CHF-exposed municipalities were mostly unaware of their riskier position – compared to their coun-terparts. Second, there is no evidence that Dexia bank itself selected particular munic-ipalities on the Swiss Franc. According to journalists Cori and Le Gall (2013), Dexia bank used instead these structured products as a way to be insured against financial volatility, once the options were in the money. Figure 1.1 displays the map of the 1,586 municipalities which contracted toxic loans before 2011 with Dexia bank. Municipal-ities exposed to the CHF are in darker blue while other municipalities are featured in green. CHF-exposed municipalities are located all over the territory and there is no evidence suggesting geographic clustering according to their exposure.
The Toxic Loan Scandal disclosure (2011)
In September 2011, the national newspaper Libération released on its website a confi-dential file detained by Dexia. They disclosed for the first time the name of all French municipalities which had contracted structured loans with Dexia bank. As shown by Appendix Figure 1.13, the map features various colored dots indicating for each mu-nicipality the degree of ex-post toxicity of the loans. The rare dots in green represent municipalities that benefit from structured loans (i.e., their interests are lower than the original fixed interest rate for which the contract was signed). The darker the dots are, the higher the amount of excess interests is announced. By clicking on each dot, citizens easily have access to detailed information concerning all the loans that their municipalities had contracted: such as the borrowed amount, the estimated excess in-terests, the date of contracting, the end date of the contract and the counterpart bank.
The disclosure of the Toxic Loan scandal occurred at a time where negotiations on a bail-out of Dexia were being held. However, the information was really unlikely to be anticipated by municipalities or economic actors. First, mayors were not required to disclose their loan contract during municipal councils (Pérignon and Vallée 2017). The only publicly-available information was the total amount of debt and the annual re-payment that the municipality incurred. Second, mayors had no particular incentives to reveal the type of loan contracts they selected, if they want to be reelected (Sartre, Daniele, and Vertier 2020). Third, looking at Google trends, there is no virtual request for the French translation of the word “Toxic Loan” before the disclosure by Libération (Figure 1.2). In September 2011, when the information was revealed, a large spike of requests occurred before decreasing to a level higher than before. It seems therefore convincing that economic actors were mostly unaware of these loans before the 2011 disclosure.
CHF-exposed municipalities were highly-profiled in the press. Once the map of Libération was online, it was relatively easy for journalists and for an informed audi-ence to disentangle CHF-exposed municipalities from other indebted municipalities. The names of the toxic-loan contracts are indeed particularly explicit. CHF-linked loans are named either after the word “CHF” or after the word “Swiss” (such as “Tofix Digi Swiss Flexi” or “Tofix Dual EUR-CHF”). As a result, there is a large number of journalistic articles published in 2011 on Toxic Loans and the Swiss Franc exchange rate. Figure 1.3 displays two bar charts: on the right, the frequency of each underlying asset among Dexia toxic-loan contracts; on the left, the number of articles mention-ing the word “Toxic Loans” and the name of at least one underlying asset, among the 1,242 articles published in 2011 on “Toxic Loans”. Surprisingly, structured loans on CHF only represent 10% of all the toxic-loan contracts granted by Dexia. However, compared to other underlying assets, there are particularly well exposed in the media – with 256 press articles mentioning the French translation for the words Toxic Loans and Swiss Franc (or CHF) in 2011. CHF-indebted municipalities were also more fea-tured in the press than non-CHF-indebted municipalities, with other exposures. Table 1.1 displays the average number of press articles published in 2011 mentioning the name of one municipality. While most municipalities were not mentioned in the press (i.e., the average is equal to 0.18), there was around 1 press article published on average per CHF-indebted municipality.
CHF-indebted municipalities were well covered in the press for two reasons. First, their debt was perceived as particularly “toxic” in 2011. Table 1.1 features the overhead ratio announced by Libération based on the files detained by Dexia. The overhead ratio is a measure of the ex-post toxicity of the debt. It is defined as the sum of excess in-terests divided by the total amount of the structured products contracting with Dexia. During the financial crisis, there was a strong appreciation of the Swiss Franc – with the EUR/CHF going from 1.65 in November 2007 to 1.13 in July 2011. As the result, the overhead ratios computed by Dexia for CHF-exposed municipalities are definitely higher on average compared to other municipalities (26% vs 9%). Second, the Swiss National Bank announced on September 6, 2011 (i.e. 14 days before the Libération dis-closure) a floor on the EUR/CHF exchange rate of 1.2. This event contributes to the high-profile of CHF-indebted municipalities in the media. It does not remove how-ever the high toxicity of their debt. As most municipalities contracted CHF-toxic loans before 2008, the EUR/CHF is fixed at a lower bound, which implies in theory very high interest rates (unfortunately stable) for the upcoming years in these municipalities.

Table of contents :

General Introduction
Introduction Générale 
Toxic Loans and Local Economic Activity: Evidence on the Costs of Public Debt Shocks 
1 Introduction
2 Institutional setting and Data
2.1 Institutional setting
2.2 Data
3 Identification strategy
3.1 Sample of analysis
3.2 Information shock vs. Debt shock on local public debt
3.3 Specifications and Identifications
4 Negative press coverage on public debt and its limited impact
4.1 Plant creation and Plant closure
4.2 Hours worked, Number of employees and Total wages
5 Rise in local public debt and its economic impact
5.1 Public debt shock and Plant closure
5.2 Mechanisms
6 Conclusion
Toxic Loans and the Rise of Populist Candidacies 
1 Introduction
2 Institutional Setting
2.1 Dexia bank and Toxic loans
2.2 The scandal disclosure by Liberation
2.3 In the aftermath of the scandal
2.4 French Municipal Elections and Populist political parties
2.5 Data description
2.6 Descriptive Statistics
3 Empirical strategy
3.1 Endogeneity issues
3.2 Instrumental variable strategy
3.3 Main specification
4 Results
4.1 Electoral Results
4.2 Candidate Entry
4.3 Heterogeneity results
4.4 Robustness tests
5 Potential Mechanisms
5.1 The economic aftermath of the scandal
5.2 Information, Press coverage and Populist rhetoric
6 Conclusion
The Increase in Partisan Segregation in the United States 
1 Introduction
2 Institutional setting and data
2.1 Partisan registration in the U.S
2.2 Data
2.3 Units of analysis
3 The rise in partisan segregation
3.1 Increase in partisan segregation across geographical units
3.2 Increase in partisan segregation within geographical units
4 Characteristics of areas driving the rise in partisan segregation
4.1 Change in partisan segregation in Democratic vs. Republican areas
4.2 Geographical and sociodemographic correlates
4.3 Change in partisan segregation across groups of citizens
5 Drivers of the increase in partisan segregation
5.1 Explaining factors
5.2 Decomposition of the change in D/(D+R) into explaining factors
5.3 Results of the decomposition
6 Conclusion


Related Posts