The causes of crime, and the best policies to reduce the incidence of crime, are matters of considerable debate. From an economic perspective, the perceived probability of being convicted if arrested, and the expected length of a resulting prison sentence, are two factors that might be expected to have an impact on the incidence of crime. If income has diminishing marginal utility, then having a low income may also contribute to the probability that an individual will offend. However, the microeconomic models that generate these propositions assume that the individual is rational. An alternative view is that the relevant individuals are not capable of properly assessing risky future events, and that increased conviction rates and longer prison sentences would have little impact on their behaviour, other than to remove them from general society for a period of time. Unfortunately, datasets that might shed light on these issues are rare.

This file contains data on men whose first adult arrest in California occurred after 1972. Their arrest records from the California Department of Justice have been merged with official earnings records from the California Employment Development Department on the basis of social security number, to provide a dataset that contains information on both income-related variables and variables related to arrest, conviction and imprisonment.

The binary variable arr86 takes a value of 1 if the individual was arrested in 1986 and 0 otherwise. This will be taken as a measure of whether or not the individual engaged in criminal activity in 1986. The variable pcnv is the proportion of previous arrests that resulted in a conviction. This will be taken as a measure of the individual's judgement of the probability of being convicted. The variable avgsen is the average length of prison sentence served by the individual for their previous convictions. This may be used as a measure of the expected prison sentence if convicted. The dataset also contains other variables that may be relevant, including variables measuring the income and employment of the individual.

Use this dataset, and techniques you have learned in ECON232, to investigate the factors that are related to the probability that an individual commits a crime. Of particular interest are the following questions:

1.Are longer prison sentences likely to reduce the incidence of crime?

2.Are policies that increase the probability of arrest (e.g. more police patrols) likely to reduce the incidence of crime?

3.Are higher employment rates likely to reduce the incidence of crime?

4.Are improved income support schemes (e.g. higher social security payments) likely to reduce the incidence of crime?

For each of the above questions, you should provide information on both the statistical significance of the relevant factor, and the economic significance (i.e. if the relevant factor was changed by a particular amount, by how much do you estimate that the probability of an individual committing a crime would change?). You must submit two files for assessment - a written report that has been saved in the PDF format, and a Gretl session file. The requirements for these files are explained in detail below.

Your report should be written in the style of a university essay (i.e. it should have a title, an introduction, some paragraphs of content, and a conclusion). It should provide a clear statement of your test results and estimates, a clear description of the econometric techniques and results that you used to generate your results, and a clear, convincing justification of the techniques that you used. Your main objective is to convince the marker that your results are credible, so for every estimator or statistic that you use in your answers, you should clearly state the properties that you believe the estimator or statistic has in your application, and explain why you think these properties hold in your particular case (for example, if you believe that the classical assumptions are reasonable for your model, then the OLS estimator would be unbiased and efficient, and this would be a good reason to base your answers on the OLS method, provided that you can convince the marker that the classical assumptions hold). If you think that there are any weaknesses in your results or approach, then you should state them clearly.

Your report should consist of fewer than 1000 words (possibly much fewer). It should not include appendices. Instead, any tables, figures, etc that you think are relevant should be included in the text of your report at the point at which they are discussed. You should proofread your work and ensure that the spelling and grammar are correct. You should use a font type and size that are easy to read (e.g. Times New Roman 12). Any equations should be typeset using your software package's equation editor (or equivalent). Tables, figures, etc should have titles and appropriate labels. Your report should be saved as a PDF file1. Recent versions of Microsoft Word are able to save files in the PDF format, as are OpenOffice, LibreOffice and many other document preparation software packages. If you are familiar with LaTeX, then this is likely to be a better way to prepare your report. You should check that you are able to create a PDF file with your software before you start typing your report. If you need help creating PDF files, ask your tutor. Marks will be deducted for poor presentation and, in extreme cases where the marker is unable easily to understand parts of your assignment, part (or all) or your assignment may attract no marks.

In addition to submitting your written report, you must also submit a Gretl session file containing your computational work. All models that you have estimated, tests that you have conducted, plots that you have created, etc, should be saved as icons in Gretl and labelled in such a way that the marker may easily find everything that was generated using Gretl and is mentioned in your written report. The Gretl session file that you submit will not be separately marked but may be used by the marker to understand your written report better. Students who do not submit a Gretl session file, or submit a file which does not include all of the computational work reported, may perform extremely poorly since the marker may find it difficult to come to an informed judgement about the quality of model used, the econometric techniques used, or the rationale for the techniques and model.

If you have performed any calculations in a spreadsheet, then you should also submit it in a Microsoft Excel format. The version of the spreadsheet that you submit must be cleaned up and clearly annotated so that the marker can understand what you have done. If you perform calculations outside Gretl, but do not submit them in a spreadsheet, then it is likely that you will receive a low mark since the marker may not be able to understand what you have done.

注：本栏目重在收集一些海外留学文书的题目，以便加加留学编辑深入了解海外教育方式与发展形势，从而拓展个人陈述、推荐信等文书的写作思路。