woensdag 19 februari 2014

Modelspecificatie: RESET test

Ramsey's specificatietest (RESET test) onderzoekt of er quadratische of kubieke termen moeten toegevoegd worden aan het model. Het komt neer op een F-test voor de gecombineerde significantie van de coëfficiënten van de toegevoegde voorspelde waarden uit een lineair model.

maandag 17 februari 2014

The Wage Curve

In 1995 publiceerden David Blanchflower en Andrew Oswald The Wage Curve. Daarin vonden ze een quasi universele negatieve relatie tussen de werkloosheid en de lonen. Dit klinkt weinig verbazingwekkend: indien de werkloosheid hoog is, is er een overaanbod aan werknemers, die elkaar onderbieden.

Toch is dit moeilijk te begrijpen vanuit een klassiek supply & demand denkkader. Nemen we een normale dalende vraagcurve en een stijgende aanbodscurve. Het totale aanbod is inelastisch, en bepaalt de werkloosheid (het verschil met het evenwicht). Dan zijn er twee mogelijkheden:

  1. De vraag daalt, waardoor het evenwicht zich verplaatst langs de aanbodscurve, en de werkloosheid toeneemt.
  2. Men bevindt zich niet in het evenwicht, maar ergens op een punt met een lager loon en boven de aanbodscurve. 
Dit zijn twee eigenaardige posities. In de eerste moeten we een verklaring vinden voor de daling van de vraag. Depreciatie van het kapitaal (uitgestelde investeringen) of technologische achterstand zijn de eerste kandidaten. Naderhand verwachten we echter dat het totale aanbod endogeen is, en zich eveneens zal bijstellen. In het tweede geval is er een imperfecte marktwerking. In het schema bevinden we ons op een punt op of boven de aanbodscurve en onder het equilibrium, waar bij perfecte concurrentie hogere lonen mogelijk zijn. Dit is niet het geval, bijvoorbeeld door beperkte arbeidsmobiliteit, waardoor de werkgever als een monopsonist opereert en louter rekening dient te houden met de (lokale) aanbodscurve. In dit geval kan de werkloosheid als exogeen opgevat worden: neemt de werkloosheid af, dan is er effectief druk om hogere lonen uit te betalen.

Het zou interessant zijn een decompositie te doen van de wage curve volgens het aandeel vraagverschuiving en het aandeel marktimperfectie.

MathJax

To include LaTeX formatting in a blog, one should just load the MathJax script if needed. The example below also loads AMSmath and AMSsymbols, so simple formatting goes easily. Put this on top of the page:

<script src="http://cdn.mathjax.org/mathjax/latest/MathJax.js" type="text/javascript">    
    MathJax.Hub.Config({
        HTML: ["input/TeX","output/HTML-CSS"],
        TeX: { extensions: ["AMSmath.js","AMSsymbols.js"], 
               equationNumbers: { autoNumber: "AMS" } },
        extensions: ["tex2jax.js"],
        jax: ["input/TeX","output/HTML-CSS"],
        tex2jax: { inlineMath: [ ['$','$'], ["\\(","\\)"] ],
                   displayMath: [ ['$$','$$'], ["\\[","\\]"] ],
                   processEscapes: true },
        "HTML-CSS": { availableFonts: ["TeX"],
                      linebreaks: { automatic: true } }
    });
</script>

Andrew Oswald

http://www.andrewoswald.com/

Britse economist met interesse in arbeidseconomie, happiness, etc.

Publiceerde samen met Blanchflower uitgebreid rond de Wage Curve, het negatieve verband tussen werkloosheid en lonen.

donderdag 23 januari 2014

The meaning and measurement of wage drift

Meaning

Wage drift is the difference in the growth rate of the actual and (collectively) negotiated pay, as it appears in collective agreements made in multi-employer (branche level) or single-employer (company level) wage negotiations. Collective agreements imply the involvement of worker's

Wage drift is a relevant concept in the context of free wage negotiations and the right for collective bodies to take part in those, representing the members as well as non-members in many cases. When wage drift is close to zero, it indicates that collective bargaining has a large influence on wage setting. This influence may be direct, through high coverage (e.g. high unionization or the extension of agreements) or indirect, through the adoption of the same wage growth in individual negotiations or non-negotiated wage arrangements.

When wage drift is positive (i.e. the actual pay increase is higher than the negotiated pay increase), there are elements beyond negotiated pay that push wages higher. When wage drift is negative, there should be a lack of coverage, as the negotiated pay increase is strictly applicable as a minimum to the workers covered and in stable employment.

Measurement

The wage that is monitored is the wage which has been negotiated. This means that when premiums are negotiated, they should be included in both the measure of the actual wage growth and the contractual wage growth. In many cases, however, only data on the base salary will be available.

We therefore have four measures of wage drift:

  • Multi-employer with premiums
  • Multi-employer without premiums
  • Single-employer with premiums
  • Single-employer without premiums

We suspect that the wage drift will be lower and variability in the wage drift will be weaker when measured including single-employers, premiums, or both. The reason for this is that decentralization and negotiations on premiums allows fine tuning and therefor actual pay will more closely correspond to negotiated pay. This will in particular be the case when economic growth is strong.

Two major problems appear when attempting to measure wage drift: there is no good indicator of actual wage growth, and indicators of negotiated pay increases are biased because of the changing composition of the aggregates they represent. The first point is particularly problematic. Comparing labour cost figures as they appear in national accounts with wage surveys reveals the issues with consistency of the former and representativity of the latter. Accountancy tricks and taxation influence labour cost, while very few surveys have enough information to allow branche level analysis. In some countries, notably in Scandinavia and Belgium, public administration is well equipped and provides data that is complete and consistent. It would be interesting to compare national accounts with such a source to evaluate the difference we may expect in other countries as well.

The second point relies on the researcher. If one obtains branche level data for consecutive years, it is possible to correct the negotiated pay increase by estimating the effect of known confounding variables. Some important variables in this respect are age (because of seniority pay increases), and schooling (because of college wage premiums) or occupational composition (technological change). The effect of these variables should be distracted from the observed negotiated pay increase. This is standard practice in wage drift analyses in the Netherlands. An alternative approach is to observe the wage growth for the 10th percentile of the workers that remain in stable employment. This group, however, differs strongly from the workforce at any moment in time, even at the lower percentile, or may especially so, because of the reasons just outlined.

Explanations

As noted before, coverage is an important explanatory variable in establishing the link between actual pay and negotiated and contractual wages.

woensdag 22 januari 2014

Minimum wages and marginal profit

In 1946, Stigler did not believe minimum wages could be any good.

Wages represent marginal profit, so imposing a minimum wages equals increasing marginal productivity, or discharging unproductive employees.

It should be noted that a discharged employee can always go to a sector with a lower minimum wage, until there is only the national minimum wage or migration left, in which case unemployment because likelier.

But my thoughts are the following: is it not a much simpler explanation to consider the market as imperfect. In a supply and demand framework, one would be to the left of the equilibrium. Employers might hire more employees in order to optimize profit, but for some reason they don't do it. They all prefer to higher a few workers less, as if it was a collusive decision to keep a pool of workers unemployed, reasoning unemployment will increase their value (schooling, reduce aspirations, increase motivation, etc.). Certainly, as for many companies, the pool of possible employees is actually well known. They are aware of the fact that the emptier the pool, the higher the likelihood that the pre-estimated productivity will be wrong. Therefor, they rather select one worker when there are 20 more available in a town, then when it is the last one.

As a result, increasing the minimum wage literally changes nothing. Well, it does increase the labour supply and puts some limit on labour demand, but it may well be the case that from the point out of equilibrium one was in, the minimum wage is payable and not even touching marginal productivity, which moves with the labour demand curve. Actually, this should be the case. We know what there is to prove: minimum wages do not alter employment. The only question remaining is why the equilibrium is not regarded as an optimal position.

Stigler refers to this in the paragraph on Employer wage determination. Arguably, multi-employer collective agreements speak in favour of this option. However, he believes this wage cannot be above marginal productivity. Again, this might be wrong: there can be wage compression within firms or industries, by paying high-wage earners below marginal productivity, which is quite possible for the very same reasons as stated above. Don't these employees start their own business instead to obtain a higher income? It seems that risk aversion acts as a barrier to such movements.

donderdag 9 januari 2014

Industry classifications

International Standard Industry Classification (ISIC)

This is the classification used by the United Nations.


Statistical Classification of Economic Activities in the European Community (NACE)

This is the classification used in the European Union.