Thursday, 1 September 2016

Does increased flexibility in labour institutions makes labour markets more resilient to recessions? The new CGR working papers by Prof Pedro Martins present positive evidence

Recessions hit hard and fast while wages and labour conditions are sticky and difficult to adapt, which tends to lead to high unemployment levels during business cycle downturns. We can see an example of this in the unemployment levels of some European economies.  According to Eurostat, Spain had an unemployment rate of the 22.7% in April 2015, a rate that skyrocketed from 8.1% in January 2008. Similarly Greece’s unemployment rate went from 8% in 2008 to 25.6% in 2017. Portugal has also experienced a bumpy ride, albeit a more moderate one. The unemployment rate started at similar levels to Spain and Greece in 2008, peaked around 17.5% in January 2013 and start decreasing afterwards down to 13% in April 2015.

The three countries reformed their labour markets to increase employment and economic growth. These reforms took place in a complex regulatory space involving workers and businesses and affected multiple dimensions: type of contracts, bargaining structure, overtime, work taxes, etc. The unemployment figures above indicate that reforms do not always succeed in their aims of reducing unemployment. However a quick glance to the Portuguese unemployment evolution suggest that they must have done something right. The last two CGR working papers provide a valuable insight into these drivers of these results.

Pedro Martins, -Professor of Applied Economics at the School of Business and Management, Queen Mary, University of London,- explores two labour reforms conducted by the Portuguese government in 2012. In the first place, he assesses the effects of reforming overtime premiums and secondly the effects of increasing the duration of fixed term contracts.

In the first paper -Can overtime premium flexibility promote employment? Evidence from a law reform -  Prof Martins evaluates a 2012 reform in labour law on overtime premium wages. This reform was one of the conditions of the Memorandum of Understanding signed by the Portuguese Government to access the support funds from the European Union and the International Monetary Fund. The reform came into force in August 2012 after parliamentary approval preceded by a tripartite agreement between the government, employer and union confederations; it established that the first hour of overtime would be subject to a premium of 25% (down from a 50% premium), and following overtime hours would be subjected to a premium of 37.5% (down from 75%). Weekends and bank holidays overtime would reduce their premium from 100% to 50%.  These reforms might have a significant effect in the Portuguese economy given that -as Prof Martins details- paid overtime corresponded to 1.2% of the total wage bill of the economy in 2011 and affected over one third of the workers in firms that pay overtime.

Prof Martins explains how such a reform could affect the trade-off between overtime and employment. If a firm have to choose between facing a temporary production increase through more hours or more workers, they would tend to prefer adding more hours to existing workers giving the quasi-fixed costs of employment. Therefore, the Portuguese overtime reform would lead to an increase use of overtime. However, the effects on employment are less clear, as Prof Martins illustrates with Figure 1.

Figure 1: Theoretical effects of an overtime premium cut

Figure 1 illustrates a case in which the increase in overtime availability also increases the employment level.  In the case of a firm not using overtime -point C- their preferences regarding workers and hours are expected to remain unchanged. However, it might be possible that a firm already paying overtime moves to increase both their overtime and employment demand after the overtime premium is cut -moving from point a to point b. This might be because paying for less overtime makes both the employment of existing workers and new hires more attractive to firms already using overtime.

These theoretical assumptions are coherent with the changes experienced in overtime premiums after the reform is introduced. This is illustrated with Figure 2 -Figure 3 in the paper- where we can observe the change in overtime premium between October 2011 and October 2012 for workers that had the same overtime hours in the two time periods.

Figure 2: Distribution of 2012-11 worker-level change in overtime premium (same overtime hours)
Prof Martins points out that the distribution peaks on -50% and 0, highlighting that firms can be divided mostly between those that made an extensive use of the new legislation and reduced their workers overtime premium by half and firms that made no use of the new legislation. The paper then exploits this distribution researching two main questions: what are the determinants that drive a firm to reduce or maintain their overtime premiums? And what are the effects of reducing it?

Answering the first question, Prof Martins finds that -in line with the theoretical assumptions- firms more prone to overtime reductions tend to be labour intensive, have lower hourly base pay but higher overtime base pay and are members of employers’ associations. To answer the second question, Prof Martins compares the outcomes of interest across firms that reduce their overtime premiums and firms that not. In order to minimise self-selection biases, he uses a difference-in-differences matching’ approach. 

As described in the CGR post about the effects of the Olympic Games in happiness, Difference-in-Difference techniques seeks to exploit quasi-natural experiments in order to determine the treatment effect of a policy intervention. The problem that social scientists face in analysing policy intervention is the inexistence of the perfect control groups that exist in experiments in which interventions are assigned randomly. Difference-in-Differences techniques overcome this difficulty by identifying a group that -although may vary in some characteristics- has a parallel pre-intervention trend in the observed variables of interests.

The matched Difference-in-Differences strategy adopted by Prof Martins departs from the same approach but matches treated and non-treated firms along a series of relevant covariates. In this case firm age, firm exports, employment, sales or firm types among other factors observed in the pre-treatment and post-treatment period. This ensures that every treated firm is compared with the most similar non-treated firm. Therefore, the observed differences between firms that reduced overtime premiums and those that not reduced overtime premiums capture the causal effect of the legislation.
Following this approach, Prof Martins finds that firms that reduced their overtime bill were able to increase, in relative terms, the volume of overtime hours, but also of employment, total hours and sales. These findings are consistent with the theoretical prediction previously outlined: the reduction in overtime costs increases the demand not only of overtime but also of workers in general.

In the second paper - Should the maximum duration of fixed-term contracts increase in recessions? Evidence from a law reform - Prof Martins analyses a similar marginal change in employment conditions. In this case, the reform involves an increase of Fixed-Term Contracts (FTCs) maximum duration from three to four and a half years.

FTCs differ considerably from permanents contracts in terms of worker protection, which makes them especially attractive for employers in contexts with highly restricted individual dismissals, as is the case of Portugal. Prof Martins highlights that, with a 22% share of FTCs in total private-sector employment, Portugal has the third largest percentage of workers under FTCs in the European Union. In the context of an ongoing recession, the Portuguese government proposed a measure of “extraordinary renewal” that would allow employers to extend FTCs up to four and a half years, a measure that came into force in January 2012.

Critically for the evaluation conducted, the new law established that, amongst FTCs approaching their maximum duration, only those that reached the duration threshold from January 11th 2012 would be allowed to extend their duration. So, FTCs signed during up to early January 2009 would not be eligible (their maximum duration was still set at three years) while those signed in the later months of 2009 would be eligible for extension.

Such discontinuity designs in eligibility are highly appealing for evaluation purposes as they allow one to build a control group that effectively only differentiates from the treated group in the random characteristic that determines their eligibility. In this case it is highly reasonable to assume that the characteristics of the FTCs signed in, say, January 2009 and those signed in, say, February2009 only differ in their eligibility for FTCs extension, implying that observed differences between the two groups can appropriately capture the causal effect of FTCs reform.

Figure 3 -Figure 1 in the paper- presents graphic evidence of how this discontinuity design affected the conversion share of FTCs to permanent contracts and the number of hirings per month. 

Figure 3: Number of workers and converstion rates per hiring month

Month 0 corresponds to February 2009; the first month when FTCs contracts can extend their duration. The red line represents the number of hirings in said months -that took place back in 2009- and the blue dots the conversion rate of FTCs into permanent contracts. As can be observed, there is a significant drop in conversion rates after the FTCs became eligible for extension, suggesting that Portuguese firms have taken advantage of the measure and extended their FTCs. The question is what are the effects then of FTCs extension in terms of employment and workers’ mobility?

Prof Martins addresses the effects of FTCs extension over three main potential transitions: conversion to permanent, employment status, and mobility to different firms. The results points to negative effects on conversion rates, as suggested by the graphical evidence. While employment does not seem to have increased significantly, inter-firm mobility fell. Basically workers that lose their jobs because they reach the limits of FTCs tend to find other jobs quickly, however the FTC extension allows firms to keep workers that would have otherwise have been dismissed and then moved elsewhere - which may have significant effects on the productivity levels of the labour market.  These results are latter confirmed by a difference-in-differences approach that also controls for observed worker and firm covariates and unobserved effects at firm level. As Prof Martins concludes, increased flexibility of FTCs can be useful during recessions, leading to longer contracts and less inter-firm mobility.

From a more general perspective, the reasons why it tends to be difficult to increase labour flexibility in economic downturns is because they have to overcome the interests of existing employers and union confederations in keeping the status quo or contain changes, an interest and veto power that increases the more stringent is the regulation. It is this clash between recipes for reforms and the political realities of passing these reforms that often ends in reforms at the margin, creating segmented pockets of flexibility in otherwise rigid markets.

As Prof Martins points in his papers, there is abundant literature critical of these reforms “at the margin” arguing that increases labour market segmentation while has negligible effects on reducing unemployment. However, as the last CGR working paper shows, the parameters in which labour market institutions can be reformed are highly heterogeneous and their consequences depend on which dimensions are reformed. The results of both articles, evaluating the effects of reform at the margin of overtime payment and FTCs duration, present strong evidence for greater flexibility on selected labour market institutions during recessions.  It is a case that Prof Martin puts forward in his FTCs’ working paper, concluding that his findings:

highlight the potential of greater flexibility in the legal parameters of FTCs - and possibly other labour market regulations and institutions (not only unemployment benefits but possibly also tax wedges, activation practices, minimum wages, etc.) - over the business cycle as a tool to minimise the resulting employment fluctuations. Such rules in labour market policy making could successfully complement those in place in macroeconomics (Taylor 1993), especially given the challenges that many countries currently face in their monetary and fiscal policies.

If you are interested, you can read the complete working papers in the following links:

Or if you have a broad interest in the work of Prof Martins on labour markets you can visit the post summarizing his inauguration lecture:

No comments:

Post a Comment