tá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Colombia - Bogotá - Col
Downward Wage Rigidities and Other
Firms' Responses to an Economic
Slowdown: Evidence from a Survey of
Ana María Iregui B.
Ligia Alba Melo B.
María Teresa Ramírez G.
Downward Wage Rigidities and Other Firms’ Responses to an Economic Slowdown:
Evidence from a Survey of Colombian Firms♣
Ana María Iregui B.
Ligia Alba Melo B.
María Teresa Ramírez G.
This paper uses a wage setting survey of 1,305 Colombian firms to explore the nature
and sources of wage rigidities. This is the first study of a non-European emerging
economy that uses evidence from a survey of firms to analyse this topic. The survey
was carried out during the first half of 2009, when the Colombian economy was
showing signs of a slowdown in economic activity and increasing unemployment. The
sample is fully representative of the population under study. The results provide
evidence of nominal and real downward wage rigidities in the country. The most
important factor in not reducing base wages during an economic slowdown is to avoid
the loss of more experienced and productive workers, which is related to the efficiency
wage theory in its adverse selection version. In addition, ordered logit regressions were
used to determine what factors are related to wage rigidities. The findings indicate that,
in general, permanent contracts, workforce composition, labour intensity and the
presence of collective agreements play an important role in explaining wage rigidities
in the country.
Keywords: wage rigidities, survey evidence, efficiency wages, Colombia, labour market,
JEL Classification: C25, J30, J50
♣ The authors are senior researchers with the Research Unit of the Deputy Governor’s Office at Banco de la República
(Central Bank of Colombia). They would like to thank Juan Carlos Guataquí for his comments and suggestions and Cindy
Moreno for her research assistance. The participation of the firms that agreed to complete the survey also is appreciated.
The opinions expressed herein are those of the authors and do not necessarily reflect the views of Banco de la República
or its Board of Directors.
It is important to understand the nature and causes of wage rigidities, since such rigidities
partly determine the persistence and volatility of inflation, as one of the main components
of the marginal cost. Also, wage rigidities offer a microeconomic explanation to a
macroeconomic phenomenon: voluntary unemployment. As Tobin (1972) and Akerlof et
al. (1996) state, when nominal wages are downwardly rigid, a certain level of inflation
allows for a greater flexibility in real wages, thereby helping adjustments in the labour
The reduction of inflation and the adoption of an inflation targeting regime, which took
place in several countries during the past two decades, have renewed interest in the study
of wage rigidities, due to the impact they can have on the labour market.1 The Colombian
case is no exception. Colombia has experienced a gradual fall in inflation since the
beginning of the nineties; however, it was only after 1997 that inflation came close to the
announced target.2 The main decline in inflation took place between 1998 and 1999,
when it went from 16.7% to 9.3%. Since then, inflation has remained in the single digit
level. On the other hand, unemployment increased, reaching a peak in 1999-2000, when
the economy faced a deep recession. By 2009, the scenario in Colombia was one of low
inflation, high unemployment and signs of an economic slowdown (see Figure 1). Since
1 See, for example, European Central Bank, “Wage Dynamics in Europe: Final Report of the Wage
Dynamics Network (WDN),” December 2009.
2 In Colombia, explicit inflation targets have existed since 1991. For details on the implementation of an
inflation targeting regime in Colombia, see Gómez et al. (2002).
the aim of this paper is to study wage rigidities, the economic conditions prevailing in the
country offer a suitable context for applying a unique survey to Colombian firms.
Inflation, unemployment and GDP growth in Colombia: 1991-2009
Source: DANE and Banco de la República
To explore wage setting mechanisms, analyse the nature and sources of wage rigidities
and test different theories of wage rigidities in the country, we designed and applied a
survey to Colombian firms. The survey allows us to obtain answers directly from those
who set wages in a firm and helps us to understand the behaviour of firms and the labour
market. In addition, it provides evidence for the micro-foundation of the Central Bank’s
wage and price models, by incorporating real and nominal rigidities, and offers elements
for monetary policy decisions.
19911992 19931994 199519961997 199819992000 200120022003200420052006200720082009
Anual variation (%)
TargetInflation (CPI)Unemployment rate GDP growth
A study of wage rigidities in an emerging country, such as Colombia, is also important
because the country’s institutions and labour market could have characteristics that
differentiate it from developed countries where this type of study has been concentrated.
For instance, Colombia has high levels of informality. In fact, informal workers
accounted, on average, for 58% of the total number of workers during the period 2001-
2007. Unlike the situation in Europe, union density in Colombia is very low: less than
5% in recent years (Guataquí et al., 2009). As a result, we would expect the role of unions
in explaining wage rigidities in Colombia to be less important than in Europe.
Furthermore, the legal minimum wage in Colombia plays a very important role in setting
wage increases (Iregui et al. 2009b). Another aspect to highlight is the presence of high
non-wage labour costs, which come to nearly 40% of base wages.
To the best of our knowledge, this is the first study for a non-European emerging
economy that uses evidence from a survey of firms to identify and analyse downward
wage rigidities. The literature on downward wage rigidities using firm surveys dates back
to the studies of Kaufman (1984) and Blanchflower and Oswald (1988) for the United
Kingdom, Holzer (1990), Blinder and Choi (1990), Bewley (1995, 1998, 1999) and
Campbell and Kamlani (1997) for the United States, and Agell and Lundborg (1995,
2003) for Sweden. In general, these studies found that firms do not cut wages because
they do not want to affect the motivation, effort and morale of workers. Consequently,
this leads to downward nominal wage rigidity. Similar results associated with efficiency
wage theories in explaining wage rigidities were found recently by Agell and
Bennmarker (2002, 2007) for Sweden, Franz and Pfeifer (2003, 2006) for Germany,
Zoega and Karlsson (2006) for Iceland, Copaciu et al. (2010) for Rumania, Kawaguchi
and Ohtake (2008) for Japan, and Amirault et al.(2009) for Canada. In addition, Franz
and Pfeifer (2003) and Agell and Bennmarker (2002, 2007) found that the existence of
collective agreements is another important factor in preventing wage cuts.
The Eurosystem Wage Dynamics Network (WDN), a research network composed of
economists from the European Central Bank and the central banks of the European
Union, conducted an ad hoc survey on price and wage setting behaviour among nearly
17,000 firms in 17 countries of the European Union between the end of 2007 and the first
half of 2008. The results of the WDN survey indicate the existence of significant
downward rigidity in base wages in the European Union, with important cross-country
differences. For example, downward nominal rigidity prevails in the Netherlands, Greece,
Germany, Austria and Portugal, whereas downward real rigidity is more prevalent in
Belgium, Finland, Luxembourg, Spain and Sweden. According to the survey, the most
important reasons for preventing wage cuts are the impact on work morale and effort,
preventing the most productive workers from leaving the firm, and labour regulations or
In particular, we applied our wage setting survey to 1,305 Colombian firms in the
country’s thirteen main cities, taking into account nine economic sectors and three firm
sizes. This survey has the advantage of using a representative sample of firms, which
3 For details on the WDN firm survey, see European Central Bank, Wage Dynamics in Europe: Final
Report of the Wage Dynamics Network (WDN), December 2009.
allows us to generalize the results to the population under study. As Campbell and
Kamlani (1997), we designed the survey to obtain answers for different occupational
groups, in our case, managers, professionals, technicians and assistants, and unskilled
workers, since the reasons for wage rigidity may differ across types of workers.
Regarding the response rate, it is important to mention that we obtained responses from
The survey asked firms how likely it is they will perform certain actions during a period
of economic slowdown. Then, with the survey results and using ordered logit models, we
empirically examine the firms’ responses, taking into account the firm-specific
information collected for the survey. The survey also asked firms why they do not reduce
wages in difficult times and provided respondents with a series of reasons based on the
more relevant theories, so as to test which of them explain wage rigidities in the
Colombian case. We also used ordered logit models to examine the firms’ responses in
The results of this study point to the presence of nominal and real downward wage
rigidities in Colombia.4 According to the survey, the most important reasons why
Colombian firms do not cut wages during difficult times are to prevent loss of the most
productive and experienced workers, not to affect worker’s effort and productivity, and
not to affect worker’s motivation. These reasons are related to the efficiency wage theory,
4 This finding confirms previous micro-economic evidence of wage rigidities in Colombia; see Iregui et al.
particularly to the adverse selection model, the shirking model, the gift-exchange model
and the fair wage-effort hypothesis. Interestingly, these results are similar to those found
in the literature for developed countries.
Survey evidence also suggests that firms can resort to other alternatives to adjust costs in
difficult times, besides changes in base wages, such as reducing non-statutory benefits
and variable pay, laying off employees, changing the type of employment contract and
hiring new workers at lower wages. The use of these strategies varies across economic
sectors and occupational groups.
This paper is divided into five sections, in addition to the introduction. In the second
section, we describe the survey design and sample selection. The third section analyses
the presence of downward nominal and real wage rigidities in Colombia and empirically
test firms’ responses to the related questions. Section four studies the reasons for
preventing wage cuts and empirically tests different theories on wage rigidities. In the
fifth section, we discuss alternatives other than changes in base wages that firms could
use to adjust labour costs during a period of economic slowdown. The final section
presents the main conclusions.
In this paper, the analysis is based on a unique survey of 1,305 Colombian firms. It was
designed to explore wage setting mechanisms, the nature and sources of wage rigidities,
and the link between wages and prices (see Iregui et al. 2009b). The survey also collects
data on several characteristics of the firms in question, such as the economic sector where
they operate, the kind of labour contracts they use, the existence of collective agreements
and different types of remuneration, among other features, which helped us to
characterize the firms in the empirical analysis.
The survey has the advantage of using a representative sample of firms. This allowed us
to generalize the results to the population under study: namely 39,004 small, medium and
large scale enterprises5, which are legally constituted and belong to all economic sectors,
except the public sector.6 The firms are located in 13 major cities7, which account for
70% of the formal employment in Colombia.
The sample selection was done by stratified random sampling, considering nine strata and
obtaining a final sample of 1,305 firms. The strata correspond to the following economic
sectors: agriculture, forestry and fishing; trade; construction; electricity, gas, water and
mining; manufacturing; financial services; transport, storage and communications;
education and health; and other services. In addition, firm size was considered as a
domain to guarantee that all sizes were represented in the final sample. With regard to the
response rate, it is important to mention that responses were obtained from 1,305 firms.
The firms that did not answer the questionnaire, for whatever reason, were replaced by
5 Firms with less than 10 employees were excluded.
6 The public sector was excluded, because the wages of public employees are set mainly by government
7 The cities are Bogotá, Bucaramanga, Barranquilla, Cali, Cartagena, Medellín, Manizales, Pereira and their
metropolitan areas. Barrancabermeja, Buga, Tuluá, Girardot and Rionegro were also included.
companies with similar characteristics. To do so, we used a sample surplus to maintain its
representativeness within the population.
In the design of the questionnaire, certain questions were adapted from the literature. In
particular, we considered the studies by Blinder and Choi (1990), Campbell and Kamlani
(1997), Bewley (1999), Agell and Lundborg (1995, 2003), and Franz and Pfeiffer (2006).
Preliminary versions of the questionnaire were discussed with senior specialists in survey
design and human resources managers; this enriched the survey.8
The selected firms were contacted first by telephone; those showing interest in answering
the survey were sent a letter explaining the academic purpose of the study and
emphasising the confidentiality of the information provided. Once the company agreed to
participate in our survey, a face-to-face interview was scheduled to apply the
questionnaire. The survey was directed to human resources personnel involved with wage
policies, who should be able to answer the questions for different occupational groups
(managers, professionals, technicians and assistants, and unskilled workers). The survey
was carried out during the first semester of 2009, when the Colombian economy was
showing signs of a slowdown in economic activity, low inflation and increasing
Finally, it is important to mention that all the results presented hereafter are generalized
for the population. The Coefficients of variation (cve) were calculated for each answer;
8 A Spanish version of the questionnaire is available in Iregui et al. (2009b), Appendix 4.
the coefficients obtained did not exceed 5%, which is an indicator of the reliability of the
Downward Nominal and Real Wage Rigidities
To assess whether wages are downward rigid, we asked firms about the likelihood of
performing certain actions during a period of economic slowdown, using a scale from 1
to 4, where 1 is not at all and 4 is very likely. To allow for comparisons, we calculated the
mean score of the answers. Following Blinder (1991), a mean score greater than or equal
to 3.0 is considered excellent and a score of less than 1.5 is very poor; a mean score
greater than or equal to 2.5 is considered to be reasonably strong.
In particular, to identify downward nominal wage rigidity (DNWR), the options of either
reducing or freezing base wages were considered. For downward real wage rigidity
(DRWR), the alternative of increasing basic pay at a rate lower than inflation was
included.9 Table 1 shows the percentage of responses not at all / not likely and likely /
very likely for each occupational position, as well as the mean scores obtained for the
9 According to Colombian law, the purchasing power of the minimum wage must be maintained. Then, the
previous alternatives can be considered only for base wages higher than the legal minimum wage.
How likely is your firm to carry out the following actions?
Occupational group Do not increase
Mean score* 2.33
Not at all / not likely 54.0
Likely / very likely 46.0
Mean score* 2.33
Not at all / not likely 53.8
Likely / very likely 46.2
and unskilled workers
Mean score* 2.04
Not at all / not likely 67.7
Likely / very likely 32.3
*Average score based on the following scale: 1 = not at all, 2 = not likely, 3 = likely, 4 = very likely.
Source: Authors calculations.
Pay raises below
the inflation rate
The results suggest the presence of DNWR, considering that, in all cases, more than 85%
of the firms indicated the option of reducing base pay was not at all / not likely and the
mean score was 1.5, indicating a very low likelihood of occurrence. In addition, more
than half the firms replied that the alternative of not increasing base wages was not at all
/ not likely. The option of pay raises below the inflation rate had a mean score of around
2.0 for all occupational groups and it is not at all / not likely for about 60% of the firms in
the case of managers and professionals and 70% of the firms for technicians, assistants
and unskilled workers, all of which provides evidence of DRWR. It is worth mentioning
that the results show no important differences by firm size. However, across sectors, the
results do show some variation.10 For instance, in financial services, the alternatives of
reducing base pay and not increasing base wages have a percentage of response for not
at all / not likely that is considerably higher than in the other sectors. In the construction
sector, the alternative of pay raises below the inflation rate has the highest response rates
for not at all / not likely compared to all occupations (73% on average).
The answers concerning wage rigidities are consistent with the results obtained when the
firms were asked about the last annual effective wage increase. Figure 2 shows the
histograms of the distribution of the average nominal wage change for each occupational
position between 2008 and 2009, when the country was showing signs of a slowdown in
economic activity. As illustrated, none of the companies cut wages and there is a spike
around the observed rate of inflation, 7.67%. In the case of unskilled workers, wage
changes were concentrated around this value for about 60% of the firms; however, for
managers, this proportion declines to about 40%. Furthermore, wage freezes are less
frequent among less-skilled workers, since they might be protected by collective
Next, to test the relevance of the firm’s characteristics for the responses, we estimated
ordered logit models for each action and occupational group. The dependent variable
increases with the likelihood of carrying out such actions. It takes values from 1 to 4,
where 1 = not at all, 2 = not likely, 3 = likely and 4 = very likely. The threshold
parameters estimated in all the models are statistically different from one another;
10 These results may be obtained from the authors upon request.
therefore, we maintained the four categories for the dependent variables in all the
Histograms of the distribution of the last nominal wage increase, 2009/2008
Source: Author’s calculations
The explanatory variables allow for differences in economic sectors and the location of
the firms (region); we considered trade and cities other than Bogotá (the nation’s capital)
as the reference categories in the regressions. Firm size also is included and is measured
by the number of employees (log (No. employees)). In addition, the share of managers
11 A Wald test was used to test the difference among the threshold parameters. The results of the tests, as
well as the marginal effects for all models, may be obtained from the authors upon request.
0.0 2.0 4.06.0 8.0 10.012.0 14.016.0
Percentage of firms
0.02.0 4.0 6.0 8.0 10.012.014.016.0
Percentage of firms
0.0 2.0 4.0 6.08.0 10.012.0 14.0 16.0
Percentage of firms
Technicians and Assistants
0.02.0 4.0 6.08.010.0 12.0 14.016.0
Percentage of firms
14 Download full-text
and professionals (skilled workers), the percentage of workers earning the minimum
wage (minimum wage earners), and the share of employees with a permanent
employment contract (permanent workers) were included to take into account the
characteristics and composition of the labour force. Moreover, a dummy variable that
takes the value of 1, if the firm has any form of collective agreement (collective
agreements), and a measure of union density (union members (%)) were considered to
evaluate the importance of collective wage agreements. Furthermore, we included
dummy variables to account for the presence of flexible benefits and variable pay.12
Finally, labour costs as a share of total costs were also included to approximate labour
Table 2 shows the ordered logit estimates for the alternatives do not increase base wages
and reduce base wages. According to the results for all occupational groups, the
probability that firms do not increase base wages in an economic slowdown increases
with the share of labour costs as a portion of total costs, as expected. Moreover, this
strategy in firms operating in the construction, manufacturing and financial services is
less likely than for firms in the trade sector (the reference category), where the high share
of temporary workers could affect the bargaining power of employees. Regarding the
composition of the labour force, in the case of managers and professionals the probability
that firms do not increase base wages decreases as the share of skilled workers increases.
This could be explained by the difficulty in recruiting employees of this type, as our
12 Flexible benefits correspond to a formal plan whereby employees can choose among different employer-
paid benefits or take cash. Variable pay corresponds to a form of compensation that links employee
payment to some measure of job performance.