A sectoral analysis of wage responsiveness to employment conditions

Get Complete Project Material File(s) Now! »

The Phillips curve in the South Africa literature

This section presents an overview of historical developments regarding the Phillips curve in South Africa. As reported by du Plessis and Burger (2006), evidence of early contributions regarding this topic in the country stretches all the way back to the 1970s and includes works of Hume (1971), Strydom and Steenkamp (1976), Strebel (1976), etc. These studies have in common the fact that they all followed closely the framework of Phillips (1958) in investigating the trade-o§ between nominal wages and unemployment. As the authors highlight, the main drawback about the application of this framework for South Africa was the inaccuracy of the measure of unemployment, which promptly led authors to move away from unemployment to rather focus on output gap. Therefore, in doing so, Truu (1975) and Strydom and Steenkamp (1976) Önd a signiÖcant trade-o§ between output gap and ináation but only when the sample is restricted to the 1960s.
This consequently raised the question of the presence of the Phillips curve in South Africa. More importantly, if this presence is then justiÖed, the form that it takes was another relevant interrogation. Indeed one thing certain about the South African literature regarding the Phillips curve is mainly how to reconcile theory and empirical realizations. Du Plessis and Burger (2006) essentially emphasize that this ambiguity has divided researchers on the topic into two factions. A Örst group of authors (Strebel, 1976, Nell, 2000, and Burger and Marinkov, 2006) essentially focused on non linear speciÖcations in an attempt to at least replicate and identify in the business cycle periods when the trade-o§ between ináation and output gap associated with the Phillips curve might have held. On the other hand, a second group adopted a then unorthodox approach by abandoning the hope of including a demand e§ect in the ináation equation for South Africa, either explicitly (Pretorius and Small, 1994) or rather implicitly (Fedderke and Schaling, 2005).

The basic Model

This model assume indivisible labour with all the variations in hired input taking place in the form of variations in employment. There is a large representative household with a continuum of members represented by the unit square and indexed by a pair (i; j) 2 [0; 1] [0; 1]. The Örst dimension (indexed by i 2 [0; 1]) represents the type of labour in which a given household member is specialized. The second dimension on the other hand (indexed by j 2 [0; 1]) deÖnes his disutility from work. This disutility is given by t j ‘ if he is employed, zero otherwise. ‘ 0 deÖnes the elasticity of the marginal disutility of work and t > 0 is an exogenous preference shifter which we also refer to as a labour supply shock given the impact it has on labour supply. Following Merz (1995), Gali (2011) deÖne a utility that is logarithmic in consumption. Further, there is full risk sharing among household members.

Data

Labour market data in South Africa are notoriously not very reliable and subjected to extensive change in deÖnition. Our quarterly data covers the period 1970Q1-2013Q4. We use a large set of di§erent variables and di§erent deÖnitions of labour market conditions. The baseline speciÖcation includes, Consumer Price Index as a measure of price ináation and two alternative sources of wage data namely the remuneration in the private sector, and unit labour costs in the manufacturing. Wage ináation is measured as the centered four quarter di§erence of the log of nominal wage expressed in percentage terms. The same applies for price ináation. The cyclical unemployment, measured as di§erence from the mean, is really usable only from 2000Q1 to 2014Q1. To have a longer speciÖcation we need to substitute the unemployment measure with more reliable employment measures, in particular private sector employment and manufacturing employment. The private sector employment has gone through a series of revision and the data are not always comparable through time. Nevertheless we try statistically to reduce the e§ect of these distortions. Manufacturing employment is the most reliable measure, but it is only a proxy of the overall labour market conditions.
The employment variables are de-trended using the Hodrick-Prescott Ölter to analyze variable employment as its deviation from the steady state value, while the unemployment series is demeaned of the average value of 23 percent unemployment rate, that implicitly we assume is the natural rate of unemployment. Data sources include the South African Reserve Bank, Quantec and the Saint Louis Federal Reserve Bank database. Before commenting on the regression analysis, it is worth to have a quick look at the data to be used in estimating speciÖcation (2.16). The basic hypothesis common with the old Phillips curve speciÖcation is that there is a negative relationship between wage ináation and unemployment. In Ögure 1 we display this relationship between for the period 2000-2014 in two scatter plots of wage ináation and unemployment to check if such a relationship applies in the case of South Africa.

A sectoral analysis of wage responsiveness to employment conditions

Economic theory suggests that there is a positive link between wage and labour productivity. Essentially, when output per worker rises, this creates an incentive for Örms to increase their demand for workers which ultimately results in an increase in workers compensation. This theory is backed up by empirical evidence in countries including Israel (Lavi and Sussman, 2001), Australia (Kumar et al, 2009), the United States (Strauss and Wohar, 2004), as well as South Africa (Fallon, 1992, Fallon and da Silva 1994, Wakeford, 2004, and Klein, 2012). Most of these studies however conclude of a weak link between wage and labour productivity as far as the South African labour market is concerned. The absence of a strong relationship between these two variables has direct implications for ÖrmsíproÖtability, which in turns may have acute repercussions in terms of job creation, and Önally, in terms of unemployment.
This has often been highlighted in the literature as an explanation for the severe job shedding the economy witnessed in the aftermath of the 2008 Önancial crisis. Furthermore, it is important to note that the weak link between wage and labour productivity can be explained by the presence of noises of macroeconomic and/or institutional nature. Klein (2012) for instance argues that the presence of these factors may create a wedge between the two variables which may explain why gains in labour productivity are not fully translated to wage increases. The main factors highlighted in his study include price and wage rigidities, labour adjustment costs, and other structural factors (market regulations, entry restrictions, etc.).

Wage rigidities, bargaining power and monetary policy

Early studies of labour market dynamics, which were essentially based on a search and matching framework developed by Diamond, Mortensen and Pissarides (see Mortensen, 2011 for a review of the setup), failed to reconcile theoretical labour market behaviour with observed moments. The same applies to early New Keynesian works that did not account for frictions in the labour market in terms of wage rigidities and persistent unemployment. The myth of the divine coincidence (Blanchard and Gali, 2007) was thus introduced. In particular, this would mean for central bankers that stabilizing ináation would result into negligible undesirable áuctuations in unemployment (output gap). Subsequent New Keynesian models introduced nominal wage rigidities essentially based on the work of Erceg et al (2000), a framework we used in the previous chapter, to show the empirical failure of the divine coincidence. Indeed, there appears to be a trade-o§ between the stabilization of ináation and that of unemployment, which in itself means the existence of a sacriÖce ratio monetary authorities around the world face.

The social planner equilibrium

We assume a benevolent social planner who solves the problem facing technological constraints and labour market frictions present in the decentralized economy. He internalizes the e§ects of changes in the labour market tightness on hiring costs and the resource constraint. Since there is symmetry in preferences and technology, e¢ ciency requires that identical quantities of goods be produced and consumed, meaning Ct (i) = Ct for all i 2 [0; 1]. Also, labour market participation has no cost but instead, it has a social beneÖt since it decreases hiring costs. The social planner always chooses an allocation with full participation. This necessarily does not imply full employment since both a disutility, and increases in hiring costs come as a result of higher employment.

Contents :

• 1 Introduction
• 2 Estimation of a New Keynesian Wage Phillips Curve
  • 2.1 Introduction
  • 2.2 The Phillips curve in the South Africa literature
  • 2.3 The Model
  • 2.3.1 The basic Model
  • 2.3.2 Extension of the model
  • 2.4 Empirical results
  • 2.4.1 Data
  • 2.4.2 Estimation results
  • 2.5 A sectoral analysis of wage responsiveness to employment conditions
    • 2.5.1 A simpliÖed model
    • 2.5.2 Empirical Study
  • 2.6 Conclusion
• 3 Monetary Policy in an Economy with High Structural Unemployment
  • 3.1 Introduction
  • 3.2 Wage rigidities, bargaining power and monetary policy
  • 3.3 The Model
    • 3.3.1 Household
    • 3.3.2 The Örms
    • 3.3.3 The equilibrium
    • 3.3.4 Log Linearization
  • 3.4 Simulation
    • 3.4.1 Calibration
    • 3.4.2 Impulse responses
  • 3.5 Estimation and historical decomposition
    • 3.5.1 Estimation results
    • 3.5.2 Historical decomposition
  • 3.6 Estimation analysis
    • 3.6.1 Impulse response function for South Africa
    • 3.6.2 Welfare analysis
  • 3.7 Conclusion
  • 3.8 Appendix A
  • 3.9 Appendix B
• 4 Labour Market E§ects of Public Employment
  • 4.1 Introduction
  • 4.2 Public employment e§ects: theory and stylized facts
  • 4.3 Model
    • 4.3.1 General assumptions
    • 4.3.2 General Setting
    • 4.3.3 Households
    • 4.3.4 Workers
    • 4.3.5 Private sector Örms
    • 4.3.6 Government
  • 4.4 Calibration
  • 4.5 Impulse response functions
  • 4.5.1 E§ects of di§erent levels of bargaining power
  • 4.5.2 Di§erence of productivity between private and public sectors
  • 4.6 Analysis of the results and optimal policy response
  • 4.7 Conclusion
  • 4.8 Appendix A
  • 4.9 Appendix B
• 5 Summary and Concluding Remarks
  • 5.1 Summary
  • 5.2 Limitations and lines for further research
• 6 Bibliography

GET THE COMPLETE PROJECT
Structural Unemployment, Labour Market Dynamics and the Transmission of Monetary Policy in South Africa