دستمزدها، اشتغال، گردش کار و دسترسی به بازارهای کار محلی

In this paper, we extend a dynamic efficiency wage model to the case of multiple local labour markets that interact through migration. Firms are concerned about turnover costs. The quitting behaviour of workers is a function of local labour market conditions, non-wage income and the costs and benefits of migration to other local labour markets. A synthetic micro sample of 20,302 observations from the 1986, 1991 and 1996 New Zealand Censuses of Population and Dwellings provides evidence supporting the theory. Across subgroups, the wages of workers with relatively inelastic local labour supply and/or lower geographical mobility are relatively more responsive to changes in the local employment rate. The evidence is consistent with the notion that local employers engage in monopsonistic competition with respect to the employment of such workers.

During the 1990s some empirical phenomena in the labour market were documented that appeared to contradict the standard competitive model. An often-cited example is Card and Krueger's (1994) case study of the impact of a minimum wage increase, where, contrary to competitive theory, they found that an increase in the minimum wage increased employment in fast food outlets in New Jersey, United States. Another interesting case is the research of Blanchflower and Oswald (1990), who, using American and British data, found evidence for an inverse relationship between the level of pay of individuals and the prevailing local unemployment rate, which they labelled the ‘wage curve’. Subsequently, they and many others have reported additional evidence for this relationship.1 A common thread running through such findings could be the presence of an upward-sloping supply curve facing the individual employer in a local labour market, rather than the perfectly elastic one of the competitive model. It is therefore no surprise that monopsony has gained a rather more prominent position in the theory of labour demand (see, e.g., Boal and Ransom, 1997 and Manning, 2003). Nevertheless, outright monopsony seems a rather extreme market form, given that most firms face some competition in recruitment and barriers to entry into specific markets have been reduced rather than increased in recent decades. Thus, rather than assuming a single buyer of labour, Bhaskar and To (1999) formulated a theory of monopsonistic competition in which there is free entry, but with the establishment of new firms constrained by lump-sum start-up costs. Each employer then has some market power in the labour market, even though the firm employs only a small fraction of the work force. This wage setting power results from horizontal job differentiation, which is a form of worker heterogeneity that leads to workers preferring certain jobs on the basis of non-wage characteristics. Monopsonistic competition may be responsible for the responses to unemployment or institutional shocks in local labour markets being quantitatively small. One of the main reasons for horizontal job differentiation is the geography of local labour markets. Varying costs of job search and commuting create heterogeneity, even among otherwise identical workers, simply due to differing residential locations. If commuting costs are rather important, Bhaskar and To (1999, p. 195) find that employment will increase following a modest increase in the minimum wage. The present paper develops this theme further by focussing on frictions in the labour market at a greater spatial scale than commuting, namely wage outcomes across well-defined local labour markets, linked through migration. Again heterogeneity of workers is introduced, but reservation wages now vary across workers because the lump-sum costs of migration between localities will depend on the location of their current job. The geography of labour markets then influences the relationship between local wages and employment. In addition, using a simple search model, Sato (2000) shows that as long as there are productivity differentials across local labour markets, those with higher productivity have higher equilibrium wages and lower unemployment rates. Even with low mobility costs, spatial real wage differentials can then persist in equilibrium because higher productivity regions have larger populations that result in utility-offsetting congestion costs (commuting costs and land rent). In this paper we will carry out an empirical analysis of local wages and employment by means of data on a set of 30 urban labour markets in New Zealand. As with the models described above, the geography of labour markets and fixed costs of production are central features of our model which lead to monopsonistic tendencies in these local labour markets. We formulate a model that extends the approach of Campbell and Orszag (1998), who, in turn, combined elements of a dynamic efficiency wage model due to Phelps (1994) with endogenous quitting behaviour as in Salop (1979). We generalise quitting behaviour by including non-wage income and the possibility of migration to other local labour markets. Long-run inter-regional equilibrium is ensured by the equalisation of expected utility across local labour markets, as in Harris and Todaro (1970). Exogenous shocks that originate within the local labour market (such as a permanent change in the level of amenities) generate equilibria that lead to a positive long-run correlation between wages and unemployment, while shocks generated outside the local labour market lead to a negative correlation. Given an assumption of small local labour markets, the latter shocks are likely to dominate and generate Blanchflower and Oswald's wage curve, albeit with a varying elasticity in our model. The paper extends previous research on the wage curve phenomenon in a number of ways. Firstly, the specification of our regression equations follows directly from the theoretical model, which enables us to recover parameters of the quit-rate function without having access to labour turnover data. Secondly, we introduce a spatial dimension, namely the accessibility of local labour markets, into estimation of the unemployment elasticity of pay. Thirdly, we control for spatial variation in real non-wage income, such as the variation in the purchasing power of national social security benefits. A fourth innovation in the present paper is that we link calculated wage curve elasticities for groups of workers to levels of geographical mobility observed for such workers in other research. For example, the extent of local monopsony captured by the wage curve appears to be greater for less geographically mobile groups, such as the lower skilled (e.g., Mauro and Spilimbergo, 1999). Finally, we estimate the model proposed in this paper with data on New Zealand. This has several advantages. The first is that New Zealand has a small population of just over four million, with local labour markets dispersed over an area that is between that of Great Britain and Japan. While there is a mixture of large (such as Auckland and Wellington) and small urban labour markets, many of the local labour markets are ‘thin’ by international standards. The use of New Zealand therefore allows us to exemplify a case in which the geography of local labour markets may generate non-competitive influences on wage setting. Moreover, the distance between the country's 30 urban labour markets is such that inter-urban commuting would be virtually absent, so that there is little confounding of migration and commuting responses. A further interesting feature of the New Zealand case is that this country underwent a decade of deregulation and restructuring since the mid 1980s. The impact of this has been well documented in the international literature (e.g., Evans et al., 1996). In the present context, of particular importance is legislation that was introduced in 1991 aimed at enhancing labour market flexibility (e.g., Morrison, 2003). By considering how the unemployment elasticity of pay changed over the decade with 1991 as midpoint, we can assess whether the reform may have had an impact on the unemployment elasticity of pay. A ‘synthetic’ micro-level sample was generated from three New Zealand population censuses (1986, 1991 and 1996).2 Each observation consists of a group of wage and salary earners at the local labour market level, defined according to their employment status, ethnicity, gender, age, education and occupation. For each group, an estimate of median annual income and mean weekly hours worked was obtained. This resulted in 20,302 useful observations. In addition, several local labour market characteristics were obtained for each urban area. Some of these were obtained from census data, some from other sources. The next section sets out the theoretical model. Section 3 discusses the nature of the available data and provides some descriptive statistics. The empirical support for the model is discussed in Section 4. Section 5 sums up.

In this paper we examined the role that local labour market conditions play in wage determination. We extended a labour turnover cost model proposed by Campbell and Orszag (1998). While their model is essentially non-spatial, in this paper we explicitly incorporated the interaction between local labour markets. The theory was put to the test by means of a ‘synthetic’ micro-sample of 20,302 observations in 30 New Zealand local labour markets, derived from 1986, 1991 and 1996 New Zealand census data. The employment rate was found to enter the wage equation with a positive coefficient, consistent with the notion of a short-run wage curve. In addition, more remote labour market areas coincide with lower earnings, ceteris paribus. Moreover, local quits respond to available wages in other local labour markets. A key conclusion of the study is that differences in the extent to which local labour market conditions affect the wages of subgroups of workers are important. Some evidence of a wage curve was found for men, but not for women, even after controlling for endogeneity of the employment rate. One explanation for this, consistent with the turnover cost model, is that the labour force participation of female workers is more elastic than that of males. A similar argument may also explain why the employment elasticity of pay is relatively low for workers aged under 26. Furthermore, the earnings of non-Europeans, part-time workers and those with less education are more vulnerable to the state of the local business cycle. On the whole, our model and empirical evidence are consistent with the notion that local employers engage in monopsonistic competition with respect to the employment of less mobile workers. In general, the elasticities found for hourly earnings did not differ greatly from those found for annual income, suggesting that hours worked varied little with local labour market conditions. Blanchflower and Oswald (1994) described their empirical findings on wages and unemployment as featuring ‘two of the variables that most interest policymakers’ (p. 1). The results of this study suggest that the wages of certain groups of workers are more responsive to changes in local employment prospects than those of others. Any attempt to reduce this vulnerability will then require increasing the geographical mobility of these workers. Future research needs to address the cause of the observed variation in the slope of the wage curve by estimating the quit-rate function directly using labour turnover data.