Staying in Power: How Does Political Continuity Shape Debt

Jaime Bonet, Jhorland Ayala, and Jorge Guerra

XIII Jornadas Iberoamericanas de Financiación Local

The findings and opinions are those of the authors and do not reflect the views of Banco de la República or its Board of Directors

Introduction and motivation

Empirical strategy and data

​

Results

​

Conclusions

Content

Political leeway: Intensive and Extensive margin

Introduction and motivation

Political leeway: Intensive and Extensive margin

Empirical strategy and data Results

​

Results

Content

Summary

• What we do: relationship between political continuity and public debt at the

municipal level in Colombia.

• How: Uses a regression kink discontinuity design taking advantage of

exogenous variation given for the electoral win/loss margin.

• Key Findings:

– Municipalities governed by the same political party across successive elections tend to see a 0.25% increase in debt levels for every percentage point increase in their win margin.

4

Literature review and contribution

(1) Literature: How politicians change fiscal policy to remain in power:

• Theoretical models show that fiscal policies are oriented to affect electoral behavior to influence

electoral cycles at subnational levels (Drazen and Eslava, 2010; Raveh and Tsur, 2020) .

• Empirical work is the same line: political leaders manipulate fiscal (expenditure) to sway electoral

outcomes (Frey, 2021; Chortareas et al., 2016; Litschig and Morrison, 2010).

• Local political cycles in Colombia do not align with national cycles, affecting debt levels at

subnational levels (Hallerberg and Strauch, 2002).

​

(2) Main contribution: How fiscal policy changes once politicians hold in power

• Exploring not just the pre-election manipulation of fiscal policies but how these policies evolve post- election when the same party remains in power.
• This perspective is relatively unexplored in existing political economy literature.
• The results are robust to placebo test and reassignments of electoral outcomes of different years

​

5

The capacity for indebtedness is being mismanaged at the local levels

Source: Calculations by the authors based on information from the National Planning Department (DNP) and the Ministry of Finance.

6

0

20

40

60

80

100

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

Interest payments/operational savings

Departments

Municipalities

Legal Limit

Payment capacity Solvency indicator

120 100

90

0

10

20

30

50

40

60

70

80

1994

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

2008

2009

2010

2011

2012

2013

2014

2015

2016

2017

2018

2019

2020

debt/current fiscal revenue

Departments

Municipalities

Legal limit

However, some municipalities have substantially increased their indebtedness

Sources: Bogotá [1] [2] ; Barranquilla [1] [2] [3]

• Barranquilla:
• Governed by the same political party for five consecutive terms from 2008 to 2027, with the

same mayor serving three of those terms.

• At the beginning of 2024, the municipal council approved a debt increase of 3 trillion COP to complete the initiatives outlined in the 2020-2023 development plan.
• The development plan costs an estimated 13.2 trillion COP, accounting for 22.2% of the total plan.
• The Inter-American Development Bank issued a loan of 0.39 trillion COP (100 million USD) at the end of 2023, also endorsed by the council.
• Bogotá:
• The city council approved an increase in debt of 10.8 trillion COP in 2021 and 11.8 trillion COP in

2022.

• These amounts represent 26% of the district's development plan for the 2020-2024 period, totaling 87.7 trillion COP.

7

Conclusions

Content

Introduction and motivation

​

Political leeway: Intensive and Extensive margin

Empirical strategy and data

Results

Data

• Unit: 707 municipalities where the party/coalition that won in 2011 ran again in 2015.
• Electoral: victory/loss margin (v) : difference between the first/second and the second/first if won/lose.
• CEDE, Sanchez and Pachón (2014)
• Per capita municipal debt.
• National planning department
• Covariates: access to historic and contemporary institutions, geography.

​

11

Data: win margin vs vote share - 2015

The graph shows the total vote share for the repeater party plotted against the repeater win-loss margin—the difference between the repeater party’s vote share and the largest non-repeater party’s vote share—both in 2015. Observations within 2 percentage points of the threshold at zero are in black. The diagonal line is the hypothetical one-to-one relationship between the two variables in an election with only two parties.

12

Repeaters

Non-repeaters

Having won or lost by a narrow voting margin

Kink Discontinuity Regression Design: Identification

• Having won (T=1, v>0) or lost (T=0, v≤0) by a narrow voting margin is a product of chance/random.
• Comparing the slopes of the relationship between debt and the margin to the left (v<0) and to the right (v>0) gives us the effect of being reelected on debt.
• Key assumptions:
• No manipulation
• Continuity in v

​

13

Kink Discontinuity Regression Design: Estimation

Results

Content

Introduction and motivation

​

Political leeway: Intensive and Extensive margin

​

strategy

Empirical strategy and data

Municipalities governed by the same political party across successive elections have an

0.25% increase in debt levels for each pp in win margin

Note: The figure displays the log of 2016 per capita debt across eight percentage bins against the 2015-win margin. The left side illustrates the relationship between non-repeaters (who won in 2011 but lost in 2015). The right side illustrates the relationship between repeaters (who won both in 2011 and 2015).

16

Robust to:

• Bandwidths
• Kink placebo
• Electoral placebo

This debt increase is more pronounced during election years

It the effect high or low ? Barranquilla case

18

Debt per capita (million COP)

Introduction and motivation

Conclusions

Content

Political leeway: Intensive and Extensive margin

Empirical strategy and data Results

Conclusions & policy recommendations

• Municipalities governed by the same political party across successive elections tend to see a 0.25% increase in debt levels for every percentage point increase in their win margin.
• This debt increase is more pronounced during election years.
• The findings highlight the need for monitoring and managing fiscal policies to

prevent unsustainable debt accumulation due to political continuity.

• It is necessary do more research about if this higher indebtedness have efficiently used or not.

20

Thanks

More information about CEER

Notes: Columns 1 and 2 reports the mean of each variable for repeaters and non-repeaters respectively. The third column presents the difference between the two groups and the fourth has the p-value correspondent.

22

Assumptions: no manipulation – McCrary test

Note: The figure displays a histogram of the win-loss margin to visually examine whether the margin is continuous at the threshold. The bins are set at 1 percentage point (pp). Below is the density plot from the McCrary (2008) test, which assesses the presence of any discontinuity in the density of the win-loss margin. The Kernel used is triangular.

23

Assumptions: continuity across covariates

Note: Table presents regressions of the main control variables against the running variable. This graph displays local averages of the outcomes in 8-percent bins plotted against the win margin, complemented by overlaid smoothed linear regression lines based on raw data on each side of the cutoff

24

Distance to colonial road

Distance to dept. capital

Longitude

Assumptions: continuity across covariates

Note: Table presents regressions of the main control variables against the running variable. This graph displays local averages of the outcomes in 8-percent bins plotted against the win margin, complemented by overlaid smoothed linear regression lines based on raw data on each side of the cutoff

25

Population

Year of creation

Rurality index

Main results: bandwidths

Note: The table presents the parameter ϕ across different bandwidths for the study of political continuity's effect on municipal debt, including all covariates and fixed effects. The first column uses the entire range from

-1 to 1 as the bandwidth. The second column applies an optimal bandwidth (h^*) of 0.154 based on the minimum standard error method, covering 430 observations. The third column doubles this optimal bandwidth to include more observations. The fourth column explores a bandwidth of 0.27, positioned between the optimal and double bandwidth, to examine the sensitivity of the estimated effects.

26

Main results: placebo test (changing pseudorandom v)

Note: Figure 9 displays the results of the permutation test for the RDK analysis, where each estimate is derived from randomly assigning the dependent variables of one municipality to another while keeping overall debt levels constant. Panel (a) shows the probability density functions for all ϕ placebos in gray and the actual ϕ in black, with its confidence intervals marked by continuous and dotted lines. Panel (b) presents the cumulative density functions, illustrating that the bulk of placebo effects cluster below the lower confidence limit of the actual effect, reinforcing the non-random nature of electoral impacts on municipal debt, and affirming the methodological soundness of the RDK approach.

27

PDF CDF

Main results: changing electoral results (2011, 2019)

Note: Table 5 explores the impact of electoral results from different years (2011 and 2019) in 2016 municipal debt levels, holding debt constant to test the influence of political continuity. Results across all columns show no significant differences, confirming the null hypothesis that variations in electoral outcomes do not affect municipal debt levels in 2016, suggesting the debt observed is independent of past or future election results.

28

What if we switch 2015 v data for 2011 and 2019 v data

Should be no effect

Possible mechanism: council share

Note: Table 5 explores the impact of electoral results from different years (2011 and 2019) in 2016 municipal debt levels, holding debt constant to test the influence of political continuity. Results across all columns show no significant differences, confirming the null hypothesis that variations in electoral outcomes do not affect municipal debt levels in 2016, suggesting the debt observed is independent of past or future election results.

29