The failure of Australian legislation on indirect discrimination to detect the systemic racism which prevents Aboriginal people from fully participating in the workforce

Government figures put the current indigenous unemployment rate at around 23%, 3 times the unemployment rate for other Australians. This thesis aims to assess whether Australian indirect discrimination legislation can provide a remedy for one of the causes of indigenous unemployment - the systemic discrimination which can result from the mere operation of established procedures of recruitment and hiring. The impact of those practices on indigenous people is examined in the context of an analysis of anti-discrimination legislation and cases from all Australian jurisdictions from the time of the passing of the Racial Discrimination Act by the Commonwealth in 1975 to the present. The thesis finds a number of reasons why the legislation fails to provide equality of opportunity for indigenous people seeking to enter the workforce. In nearly all jurisdictions it is obscurely drafted, used mainly by educated middle class white women, and provides remedies which tend to be compensatory damages rather than change to recruitment policy. White dominance of the legal process has produced legislative and judicial definitions of "race" and "Aboriginality" which focus on biology rather than cultural difference. In the commissions and tribunals complaints of racial discrimination are often rejected on the grounds of being "vexatious" or "frivolous", not reaching the required standard of proof, or not showing a causal connection between race and the conduct complained of. In all jurisdictions the cornerstone of liability is whether a particular employment term, condition or practice is reasonable. The thesis evaluates the approaches taken by appellate courts, including the High Court, and concludes that there is a trend towards an interpretation of reasonableness which favours employer arguments such as economic rationalism, the maintenance of good industrial relations, managerial prerogative to hire and fire, and the protection of majority rights. The thesis recommends that separate, clearly drafted legislation should be passed to address indigenous disadvantage and that indigenous people should be involved in all stages of the process.

Modeling multivariate autoregressive conditional heteroskedasticity with the double smooth transition conditional correlation GARCH Model

In this paper, we propose a multivariate GARCH model with a time-varying conditional correlation structure. The new double smooth transition conditional correlation (DSTCC) GARCH model extends the smooth transition conditional correlation (STCC) GARCH model of Silvennoinen and Teräsvirta (2005) by including another variable according to which the correlations change smoothly between states of constant correlations. A Lagrange multiplier test is derived to test the constancy of correlations against the DSTCC-GARCH model, and another one to test for another transition in the STCC-GARCH framework. In addition, other specification tests, with the aim of aiding the model building procedure, are considered. Analytical expressions for the test statistics and the required derivatives are provided. Applying the model to the stock and bond futures data, we discover that the correlation pattern between them has dramatically changed around the turn of the century. The model is also applied to a selection of world stock indices, and we find evidence for an increasing degree of integration in the capital markets.

Conditional coskewness and asset pricing

We explore the empirical usefulness of conditional coskewness to explain the cross-section of equity returns. We find that coskewness is an important determinant of the returns to equity, and that the pricing relationship varies through time. In particular we find that when the conditional market skewness is positive investors are willing to sacrifice 7.87% annually per unit of gamma (a standardized measure of coskewness risk) while they only demand a premium of 1.80% when the market is negatively skewed. A similar picture emerges from the coskewness factor of Harvey and Siddique (Harvey, C., Siddique, A., 2000a. Conditional skewness in asset pricing models tests. Journal of Finance 65, 1263–1295.) (a portfolio that is long stocks with small coskewness with the market and short high coskewness stocks) which earns 5.00% annually when the market is positively skewed but only 2.81% when the market is negatively skewed. The conditional two-moment CAPM and a conditional Fama and French (Fama, E., French, K., 1992. The cross-section of expected returns. Journal of Finance 47,427465.) three-factor model are rejected, but a model which includes coskewness is not rejected by the data. The model also passes a structural break test which many existing asset pricing models fail.

Tax morale and conditional cooperation

Why so many people pay their taxes, even though fines and audit probability are low, is a central question in the tax compliance literature. Positing a homo oeconomicus having a refined motivation structure sheds light on this puzzle. This paper provides empirical evidence for the relevance of conditional cooperation, using survey data from 30 West and East European countries. We find a high correlation between perceived tax evasion and tax morale. The results remain robust after exploiting endogeneity and conducting several robustness tests. We also observe a strong positive correlation between institutional quality and tax mmorale. Keywords: Tax morale; Tax compliance; Tax evasion; Pro-social behavior; Institutions

Magnocellular and parvocellular pathway mediated luminance contrast discrimination in amblyopia

To evaluate whether luminance contrast discrimination losses in amblyopia on putative magnocellular (MC) and parvocellular (PC) pathway tasks reflect deficits at retinogeniculate or cortical sites. Fifteen amblyopes including six anisometropes, seven strabismics, two mixed and 12 age-matched controls were investigated. Contrast discrimination was measured using established psychophysical procedures that differentiate MC and PC processing. Data were described with a model of the contrast response of primate retinal ganglion cells. All amblyopes and controls displayed the same contrast signatures on the MC and PC tasks, with three strabismics having reduced sensitivity. Amblyopic PC contrast gain was similar to electrophysiological estimates from visually normal, non-human primates. Sensitivity losses evident in a subset of the amblyopes reflect cortical summation deficits, with no change in retinogeniculate contrast responses. The data do not support the proposal that amblyopic contrast sensitivity losses on MC and PC tasks reflect retinogeniculate deficits, but rather are due to anomalous post-retinogeniculate cortical processing of retinal signals.

12 fingers or one, it’s how you play? Genetic discrimination in the Australian workforce

For quite some time, debate has raged about what the human race can and should do with its knowledge of genetics. We are now nearly 60 years removed from the work of Watson and Crick who determined the structure of deoxyribonucleic acid (DNA), yet our opinions as how best to employ scientific knowledge of the human genome, remain as diverse and polarised as ever. Human judgment is often shaped and coloured by popular media and culture, so it should come as no surprise that box office movies such as Gattaca (1997) continue to play a role in informing public opinion on genetics. In order to perform well at the box office, movies such as Gattaca take great liberty in sensationalising (and even distorting) the implications that may result from genetic screening and testing. If the public’s opinion on human genetics is strongly derived from the box office and popular media, then it is no wonder that the discourse on human genetics is couched in the polar parlances of future utopias or future dystopias. When legislating in an area like genetic discrimination in the workforce, we must be mindful of not overplaying the causal link between genetic predisposition towards a disability and an employee’s ability to perform the inherent requirements of their job. Genetic information is ultimately about people, it is not about genes. Genetic discrimination is ultimately about actions, it is not about the intrinsic value of genetic information.

Discrimination or Differential Involvement? A Review of the Research Exploring the Impact of Indigenous Status on Sentencing

Existing court data suggest that adult Indigenous offenders are more likely than non-Indigenous defendants to be sentenced to prison but once imprisoned generally receive shorter terms. Using findings from international and Australian multivariate statistical analyses, this paper reviews the three key hypotheses advanced as plausible explanations for these differences: 1) differential involvement, 2) negative discrimination, 3) positive discrimination. Overall, prior research shows strong support for the differential involvement thesis, some support for positive discrimination and little foundation for negative discrimination in the sentencing of Indigenous defendants. Where discrimination is found, we argue that this may be explained by the lack of a more complete set of control variables in researchers’ multivariate models.

Exponentiated gradient algorithms for conditional random fields and max-margin Markov Networks

Log-linear and maximum-margin models are two commonly-used methods in supervised machine learning, and are frequently used in structured prediction problems. Efficient learning of parameters in these models is therefore an important problem, and becomes a key factor when learning from very large data sets. This paper describes exponentiated gradient (EG) algorithms for training such models, where EG updates are applied to the convex dual of either the log-linear or max-margin objective function; the dual in both the log-linear and max-margin cases corresponds to minimizing a convex function with simplex constraints. We study both batch and online variants of the algorithm, and provide rates of convergence for both cases. In the max-margin case, O(1/ε) EG updates are required to reach a given accuracy ε in the dual; in contrast, for log-linear models only O(log(1/ε)) updates are required. For both the max-margin and log-linear cases, our bounds suggest that the online EG algorithm requires a factor of n less computation to reach a desired accuracy than the batch EG algorithm, where n is the number of training examples. Our experiments confirm that the online algorithms are much faster than the batch algorithms in practice. We describe how the EG updates factor in a convenient way for structured prediction problems, allowing the algorithms to be efficiently applied to problems such as sequence learning or natural language parsing. We perform extensive evaluation of the algorithms, comparing them to L-BFGS and stochastic gradient descent for log-linear models, and to SVM-Struct for max-margin models. The algorithms are applied to a multi-class problem as well as to a more complex large-scale parsing task. In all these settings, the EG algorithms presented here outperform the other methods.

Sparseness vs estimating conditional probabilities : some asymptotic results

One of the nice properties of kernel classifiers such as SVMs is that they often produce sparse solutions. However, the decision functions of these classifiers cannot always be used to estimate the conditional probability of the class label. We investigate the relationship between these two properties and show that these are intimately related: sparseness does not occur when the conditional probabilities can be unambiguously estimated. We consider a family of convex loss functions and derive sharp asymptotic results for the fraction of data that becomes support vectors. This enables us to characterize the exact trade-off between sparseness and the ability to estimate conditional probabilities for these loss functions.