05-07-2013, 02:05 PM
Private vs Public Ownership: Labor Outcomes in the Banking Industry
Private vs Public Ownership.pdf (Size: 754.02 KB / Downloads: 18)
Abstract
This paper examines the eects of working in a private rm instead of a state-owned rm for the Portuguese
banking industry. Our goal is to identify the eects on median employment duration, hazard rate, type of
employment contract and wages. A second contribution of the paper is to analyze a privatization process and
identify the winners and losers of that process.
We nd that there are not distinguishable dierences on wages per hour by working in a private bank
instead of a public bank. On the other hand, employment in a private bank implies a lower median employment
duration but increases the likelihood of an open-ended contract. We also nd that the marginal treatment
eects on log of wages per hour have the expected shape, since they are decreasing in the likelihood of working
in the public bank. Workers with a higher likelihood of working in a private bank have a very high wage
premium by working in a private bank.
Privatization improved workers compensation, particularly in the banks that remained public. Further-
more, it increase the job stability on public banks relatively to private banks.
Introduction
The dierences between private and public rms1 and the eects of privatization on labor, rm and economic
outcomes are frequent themes in the political debate. The debate gained a new importance with the wave of
privatization processes which started in the early 1980's in several countries.2 Despite the importance of the
question, the economic literature about this topic until the 1990's was relatively scarce and, often, the results
obtained were questionable by the limitations of the data and by the incapacity to control in a proper way
for the endogeneity of ownership and workers' choice. Recently, the access to new databases and the wave of privatizations in the former communist countries created new opportunities for researchers interested in the
analysis of the ownership eects.
This paper tries to contribute to the existent literature by analyzing the eects of ownership on labor outcomes.
Our goal is to identify and to measure the gains or losses on median employment duration, hazards rates, type
of employment contract and wages of working in a private rm instead of a public rm. A second contribution of
the paper is to analyze a privatization process and identify the winners and losers of that process.
We think that our paper has two main novelties relatively to the existent literature. First, we analyze ownership
eects on the type of employment contract (i.e, the choice between xed-term and open-ended contracts), median
duration and hazard rates, while the existent literature is usually focused in wage dierentials. Second, we identify
the marginal treatment eect of ownership on labor outcomes.
Literature Review
Our paper is related with the branch of economic literature that studies the existence and magnitude of ownership
eects3. This literature has examined, for example, dierences in performance, eciency4 and labor outcomes
due to ownership structure. In our paper we are essentially interested in the latter point. The study of ownership
eect on labor outcomes has been essentially focused in the analysis of wage dierential between the public and
the private sector. Nevertheless, some papers also analyze the eects of ownership on non-wage compensations. 5
Along the years several papers have analyzed the public wage dierentials for the US. However, there is not a
consensus in the results. For example, papers as Smith (1977) and Gyourko and Tracy (1988) have found that
federal employees earnings are nearly 20 percent higher than their otherwise equivalent private sector counterpart.
According to Belman and Heywood (2004), the wage dierential remains after control for occupational dierences,
but it is slightly lower. On the other hand, papers as Hartman (1983) have shown that, when properly dened
positions are compared, federal workers are underpaid. Such con
icting results were also found at the state
and local level. For example, Moore and Newman (1991) found a positive wage dierential for the Houston
metropolitan transit workers while Moore and Raisian (1991) and Belman and Heywood (1995) found very small
dierentials at the state level and negative dierentials at the local level.
Privatization of the banking industry in Portugal
On 25th April of 1974 a military coup put end to the dictatorship that ruled Portugal during 48 years. The
collapse of the dictatorship led to struggles among the former opposition groups and parties as to who would
exercise power and what short of regime would be created. This created an economic and political turmoil in
the eighteen months after the revolution. During that period, the movement that took power followed an antimonopolistic
policy, where monopolies were understood as large private enterprises. One of the measures taken
within this anti-monopolistic policy was a nationalization process that aected several sectors in the economy,
including the banking sector (Nunes, Bastien and Valerio 2005).
The heirship of the policies taken in the post-revolutionary period was persistent and it had long-term eects.
In the 1980s the Portuguese banking industry was almost exclusively in public hands and there were several
controls that constraint nancial operations. In fact, the liberalization of banking industry only started in 1983,
with the adoption of the rst laws to allow freedom of entry for private banks8 (law 11/83 of 16th August 1983,
decree-law 406/83 of 19th November 1983 and decree-law 51/84 Of 11th February 1984).
A consequence of this liberalization was an increase in the number of banks in the market and subsequent rise in
competition, which stimulated nancial innovation. This liberalization process cannot be disentangled from the
goal of joining the European Community. This goal served also as a catalyst for nancial markets reforms whose
objectives were to increase eciency, improve macroeconomic management and the international competitiveness
of Portugal (OECD 1999, Monteiro 2009, Bac~ao 1997). The improvement in eciency was seen as essential for
the sustainability of the industry in a common market, since, prior to 1984, the industry was almost exclusively
composed by a small number of public rms which were overstaed and inecient and by a large number of small
and inecient mutual and cooperative banks. Thus, the banking industry in Portugal was in weak conditions to
compete with foreign rms. (OECD,1999)
Model for the outcomes
In this section we will try to motivate a particular functional specication for each of the outcomes analyzed in
the paper.
First of all, we should notice that only the MTE estimators require a functional specication of the outcomes in
order to obtain identication. Nevertheless, a functional specication of the outcomes is useful to identify the
groups that win and lose with the treatment.
In our paper the outcomes analyzed are wages, type of employment contract (i.e. choice between xed-term and
open-ended contract), mean employment duration and hazard rates. The two latter outcomes deserve particular
attention by two main reasons. First, duration will be a censored variable which creates several challenges in the
estimation of the eects of ownership on median employment duration and on employment hazard rates. Second,
we do not observe these two outcomes and so in a rst stage we need to estimate them. In subsection 6.1 we
propose a strategy to deal with these problems
Data and Outcomes
In our study we use the data of Quadros de Pessoal (henceforward QP), a survey conducted by the Portuguese
Ministry of Employment and Social Security. This survey is done since 1985 and covers the workforce of all rms
employing paid labour in Portugal. Moreover, its longitudinal nature, where a unique number identies both
rms and workers, makes it possible to follow them over time.
This data source is valuable to analyze dierences between labor outcomes in the private versus the public sector
because it allows to identify in which sector each rm is in each year and the respective labor outcomes. The
access to raw data les enabled us to compute labor outcome variables ourselves. The QP survey is mandatory
for all rms, public and private, therefore the data should be immune to bias resulting from dierences in survey
questionnaires and from no compliance with the survey.