“Political Business Cycles in Local Employment”
Francisco José Veiga
Linda G. Veiga
NIPE WP 13 / 2005
“ “P Po ol li it ti ic ca al l B Bu us si in ne es ss s C Cy yc cl le es s i in n L Lo oc ca al l E Em mp pl lo oy ym me en nt t” ”
C Cé és sa ar r C Co oe el lh ho o
F Fr ra an nc ci is sc co o J Jo os sé é V Ve ei ig ga a
L Li in nd da a G G. . V Ve ei ig ga a
N NI IP PE E* * W
WP P 1 13 3 / / 2 20 00 05 5
* NIPE – Núcleo de Investigação em Políticas Económicas – is supported by the Portuguese Foundation for
Science and Technology through the Programa Operacional Ciência, Teconologia e Inovação (POCTI) of the
Quadro Comunitário de Apoio III, which is financed by FEDER and Portuguese funds.
Political Business Cycles in Local Employment
César Coelho(1), Francisco José Veiga(2), and Linda G. Veiga(2)*
(1) NIPE; (2) Universidade do Minho and NIPE
Using employment data for Portuguese municipalities, we find strong evidence of political
business cycles. Employment increases shortly before elections mainly in municipalities
where the mayor’s party has a majority of deputies in the municipal assembly and where she
is running for reelection.
Keywords: Political business cycles, local governments, employment, Portugal
JEL classification: D72, H7
In this paper we test for the existence of political business cycles in employment in
Portuguese municipalities. We use an unexplored dataset covering all mainland municipalities
from 1985 to 2000. Our results provide strong evidence of employment increases before
elections, particularly in municipalities where the mayor’s party has a majority in the
municipal assembly and where she is running for reelection. Employment increases are
notably higher in construction and public works and in community, social and personal
services. We argue that these employment variations are a consequence of increases in local
public expenditures that were documented by Veiga and Veiga (2004).
* Corresponding author: Linda G. Veiga, Escola de Economia e Gestão, Universidade do Minho, 4710-
057 Braga, Portugal. Tel.: +351-253604534; Fax: +351-253676375; Email: firstname.lastname@example.org.
2. Political business cycles in local employment
Although the literature on political business cycles is already extensive,1 most studies focus
on the behavior of central governments. However, Rogoff and Sibert (1988), Rogoff (1990)
and Harrington (1993) have developed political business cycle models that are applicable to
local as well as national governments. These models predict that incumbent politicians may
increase spending prior to elections to signal greater “competence.” This occurs when voters
are rationally, but imperfectly, informed. In contrast, Nordhaus’s (1975) well-known model of
the political business cycle assumes that voters are myopically responsive to national
economic conditions. Testing for the presence of local political business cycles may therefore
help to distinguish the applicability of models that would have similar empirical implications
at a national level.
In this paper, we use as our laboratory all the Portuguese mainland municipalities. In a
previous paper, Veiga and Veiga (2004) provided strong evidence of opportunistic cycles in
municipalities’ expenditures, particularly in investment items highly visible to the electorate,
such as those included in spending categories for Other Buildings and Miscellaneous
Constructions (especially in Overpasses, streets and complementary works, Rural road, and
Others). Pre-election increases in these expenditures are consistent with an effort by mayors
to signal greater competence. Building on these results, we now test for the existence of
political business cycles in municipal employment.
The dependent variable in our analysis is the change in the number of employees
working in the firms in the municipality (?Employment). We are able disaggregate this
variable according to the Portuguese Standard Industrial Classification (CAE) system.
Changes in national employment (?National Employment) are used to control for the impact
1 For surveys of this literature see Alesina, Cohen and Roubini (1997) and Drazen (2000).
of the macroeconomic situation of the country. The set of political variables used consists of
the following dummy variables:
- Election year (year before election): takes the value of one in election years (in the
year before elections), and of zero otherwise;
- Right: equals one for municipalities run by right-wing mayors, and zero otherwise;
- Majority (minority): equals one for municipalities where the mayor’s party has a
majority (minority) in the Municipal Assembly, and zero otherwise;
- Incumbent: takes the value of one when the mayor is running for another term in
office, and zero otherwise;
Three additional variables are included in all models as controls:
- %Pop<15: percentage of the population under 15 years old;
- %Pop>65: percentage of the population over 65 years old;
- PopDens: population density.
Data on the total number of employees in municipal firms, and according to the
Portuguese Standard Industrial Classification system, from 1985 to 2000, was obtained from
the “Quadros de Pessoal” database, of the Portuguese Ministry of Labour and Social
Solidarity (MTSS)2 Data on national employment comes from the IMF’s International
Financial Statistics. Political data, namely election dates and municipal electoral results were
obtained from the Technical Staff for Matters Concerning the Electoral Process (STAPE).
Finally, data on the three control variables was obtained from the Portuguese Institute of
It is worth noting that election dates are defined exogenously from the perspective of
the local authorities and that, during our sample period, there were no legal restrictions on the
2 The “Quadros de Pessoal” are a yearly mandatory employment survey that covers virtually all firms employing
paid labor in Portugal.
number of terms a mayor could serve. In the period subsequent to the re-establishment of
Democracy in Portugal in 1974, municipal elections have been held in 1976, 1979, 1982,
1985, 1989, 1993, 1997 and 2001, always in December.
The empirical model can be summarized as follows:
itit it iit
iTtNi ,...,1 ,...,1
where yit is the dependent variable (?Employment),
X is a vector of explanatory variables
(described above), β β β β is a vector of parameters to be estimated, νi is the individual effect of
municipality i, and εit is the error term. Given the presence of individual effects, νi, the model
above can be estimated assuming those effects as fixed or random. But, the lagged value of
the dependent variable would be correlated with the error term, εit, even if the latter is not
serially correlated. According to Arellano and Bond (1991), this implies inconsistent
estimates of the model, when, as in our sample, there is a clear dominance of cross sections
(275 municipalities) over time periods (16 years). These authors developed a Generalized
Method of Moments (GMM) estimator that solves the problems noted above. First
differencing (1) removes the individual effects (νi) and produces an equation that is estimable
by instrumental variables. In this paper, we use the extended version of the GMM estimator
proposed by Blundell and Bond (1998).3
Table 1 shows the results of estimations for changes in total municipal employment. In
the first two columns, we can see that employment increases during electoral years (and the
3 Difference Sargan tests indicate that, for our data, this system-GMM estimator, which combines the first-
differenced equations with the equation in levels, is preferable to that of Arellano and Bond (1991), which only
includes the first-differenced equations. In all estimations, the instruments used for the lagged dependent variable
were the following: levels lagged 2 and 3 periods were used in the equations in first differences, and once lagged
first differences were used in the equation in levels. Since the remaining explanatory variables are exogenous,
they were used as their own instruments.
year before), relative to the remaining years of the electoral cycle, and that right-wing mayors
increase local employment more than left-wing ones. Then, we tested for differences in
opportunistic behavior according to ideology (column 3), the level of support the mayor
enjoys at the Municipal Assembly (column 4), and the mayor’s decision to run or not for
another term in office (column 5). Results indicate that pre-election increases in local
employment occur only when the mayor is running for reelection and when the mayor’s party
has a majority of deputies in the Municipal Assembly. In these circumstances mayors have
both stronger incentives and greater power to manipulate spending. These effects are stronger
in municipalities led by right-wing mayors.
[Insert Table 1 about here]
When we perform the tests in employment series disaggregated by sector of activity
(see Table 2), we find that employment increases in electoral years are stronger, and more
significant, in Construction and Public Works and Community, Social and Personal Services.
This result is not surprising if we take into account Veiga and Veiga’s (2004) finding of
strong evidence of opportunistic cycles in Portuguese municipalities’ investment expenditures
in Other Buildings, Miscellaneous Constructions and Other Investments4.
[Insert Table 2 about here]
4 On these series there are increases in the election year, relative to the sample mean, of 14.9%, 10.4% and
Using an unexplored dataset describing employment in all mainland Portuguese
municipalities over a period of sixteen years, we provide strong evidence of political business
cycles in local employment.5 Although mayors do not control a large fraction of local
employment directly, they affect it indirectly through municipal expenditures that generate
jobs in local firms. By improving local economic conditions shortly before elections, mayors
may signal their competence to the electorate and improve their chances of reelection,
consistent with the rational political business cycle models introduced by Rogoff and Sibert
We acknowledge helpful comments from Henry Chappell. Financial support was provided by
the Portuguese Foundation for Science and Technology under research grant
POCTI/2001/ECO/37457 (partially funded by FEDER).
5 It is worth noting that results are basically the same when we use employment levels (employment as a
percentage of municipal population) instead of changes in employment. These results are available from the
authors upon request.
Alesina, A., G. Cohen and N. Roubini, 1997, Political Cycles and the Macroeconomy (MIT
Press, Cambridge, MA,).
Arellano, M. and S. Bond, 1991, Some tests of specification for panel data: Monte Carlo
evidence and an application to employment equations, The Review of Economic
Studies 58, 277-297.
Blundell, R. and S. Bond, 1998, Initial conditions and moment restrictions in dynamic panel
data models, Journal of Econometrics 87, 115-143.
Drazen, A., 2000, Political Economy in Macroeconomics (New Jersey, Princeton University
Harrington, J. Jr., 1993, Economic policy, economic performance, and elections, American
Economic Review 83(1), 27-42.
Nordhaus, W., 1975, The Political Business Cycle, Review of Economic Studies 42, 169-90.
Rogoff, K., 1990, Equilibrium political budget cycles, American Economic Review 80, 21-36.
Rogoff, K. and A. Sibert, 1988, Elections and Macroeconomic Policy Cycles, Review of
Economics Studies 55, 1-16.
Veiga, L.G. and F.J. Veiga, 2004, Political Business Cycles at the Municipal Level, NIPE
Working Paper, WP-4/2004, 1-26. Available in PDF format at:
TABLE 1: POLITICAL BUSINESS CYCLES IN EMPLOYMENT
1 2 3 4 5
Year Before Election
Sargan test (p-value)
Sources: IMF(IFS), INE, MTSS and STAPE.
Notes: - Estimations of system-GMM linear models for panel data (which combine the equations in first-
differences with the equation in levels), using the econometric software Stata 8.2.
- The coefficients and t-statistics for the control variables %Pop<15, %Pop>65, and PopDens
(included in all estimations) are not shown in order to economize space.
- Two-step results using robust standard errors corrected for finite samples.
- T-statistics are between parentheses. Significance level for which the null hypothesis is rejected:
**, 1% and *, 5%.
- Sargan is a test for the validity of the over-identifying restrictions for the GMM estimators,
asymptotically χ2. P-value is reported. The hypothesis of no second-order serial correlation in
the first-differenced residuals is never rejected.
TABELA 2: POLITICAL BUSINEES CYCLES IN EMPLOYMENT BY ECONOMIC ACTIVITY
Dep. Var. (-1) -.063
Year Before Election
Sargan test (p-value)
Sources: IMF(IFS), INE, MTSS and STAPE.
Notes: - Estimations of system-GMM linear models for panel data (which combine the equations in first-differences with the equation in levels), using the econometric
software Stata 8.2. Two-step results using robust standard errors corrected for finite samples.
- The dependent variable is the percentage change in municipal employment for the classification of economic activity indicated in the respective column.
- The coefficients and t-statistics for the control variables %Pop<15, %Pop>65, and PopDens (included in all estimations) are not shown in order to economize
space. T-statistics are between parentheses. Significance level for which the null hypothesis is rejected: **, 1% and *, 5%.
- Sargan is a test for the validity of the over-identifying restrictions for the GMM estimators, asymptotically χ2. P-value is reported. The hypothesis of no second-
order serial correlation in the first-differenced residuals is never rejected.
Most Recent Working Papers
Coelho, César, Francisco José Veiga and Linda G. Veiga, Political
Business Cycles in Local Employment, 2005.
NIPE WP 13/2005
NIPE WP 12/2005 Veiga, Francisco José and Ari Aisen, The Political Economy of
NIPE WP 11/2005 Silva, João Cerejeira, Searching, Matching and Education: a Note,
NIPE WP 10/2005 de Freitas, Miguel Lebre, Portugal-EU Convergence Revisited:
Evidence for the Period 1960-2003, 2005.
NIPE WP 9/2005 Sousa, Ricardo M., Consumption, (Dis) Aggregate Wealth and Asset
NIPE WP 8/2005 Veiga, Linda Gonçalves and Maria Manuel Pinho, The Political
Economy of Portuguese Intergovernmental Grants, 2005.
NIPE WP 7/2005 Cortinhas, Carlos, Intra-Industry Trade and Business Cycles in
NIPE WP 6/2005 Mourão, Paulo Reis, Que Critérios Redistributivos na Lei das
Finanças Locais?, 2005.
NIPE WP 5/2005 Mourão, Paulo Reis, Elasticities of Regional and Local
Administrations Expenditures – the Portuguese case, 2005.
NIPE WP 4/2005 Mourão, Paulo Reis, Incentivos à Localização em Trás-os-Montes e
Alto Douro (os séculos XII – XVI), 2005.
NIPE WP 3/2005 Alexandre, Fernando and Pedro Bação, Monetary Policy and Asset
Prices: the investment channel, 2005.
NIPE WP 2/2005 Conraria, Luís Aguiar and Yi Wen, Understandig the Impact of Oil
NIPE WP 1/2005 Bleaney, Michael and Manuela Francisco, Inflation Persistence and
Exchange Rate Regimes: Evidence from Developing Countries, 2005.
NIPE WP 15/2004
Ribeiro, José Cadima, José Viseu, Tânia Delalande and Cristina
Rodrigues, UEFA Euro 2004 Visitor Analysis, 2004.
NIPE WP 14/2004
Ribeiro, José Cadima, José Viseu, Tânia Delalande and Cristina
Rodrigues, UEFA Euro 2004 Tourism Impact Analysis, 2004.
NIPE WP 13/2004 Pereira, Nuno, J. Cadima Ribeiro and José Viseu , Sport Tourism:
Regional Promotion Stategies, 2004.