اثر سرطان پستان و روده بزرگ بر نتایج بازار کار؛ عوارض متوسط و گرادیانهای آموزشی
|کد مقاله||سال انتشار||تعداد صفحات مقاله انگلیسی||ترجمه فارسی|
|16257||2013||15 صفحه PDF||سفارش دهید|
نسخه انگلیسی مقاله همین الان قابل دانلود است.
هزینه ترجمه مقاله بر اساس تعداد کلمات مقاله انگلیسی محاسبه می شود.
این مقاله تقریباً شامل 14009 کلمه می باشد.
هزینه ترجمه مقاله توسط مترجمان با تجربه، طبق جدول زیر محاسبه می شود:
- تولید محتوا با مقالات ISI برای سایت یا وبلاگ شما
- تولید محتوا با مقالات ISI برای کتاب شما
- تولید محتوا با مقالات ISI برای نشریه یا رسانه شما
پیشنهاد می کنیم کیفیت محتوای سایت خود را با استفاده از منابع علمی، افزایش دهید.
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Journal of Health Economics, Volume 32, Issue 6, December 2013, Pages 1028–1042
We estimate causal effects of breast and colorectal cancer on labour market outcomes 1–3 years after the diagnosis. Based on Danish administrative data we estimate average treatment effects on the treated by propensity score weighting methods using persons with no cancer diagnosis as control group. We conduct robustness checks using matching, difference-in-differences methods and an alternative control group of later cancer patients. The different methods give approximately the same results. Cancer increases the risks of leaving the labour force and receiving disability pension, and the effects are larger for the less educated. Effects on income are small and mostly insignificant. We investigate some of the mechanisms which may be important in explaining the educational gradient in effects of cancer on labour market attachment.
Improvements in cancer screening, detection and treatment have led to an increase in the number of people surviving cancer and an increase in cancer survivorship as a chronic health condition; see e.g. Cutler (2008). Some types of cancer often strike people at working age, and the effects of cancer on labour market outcomes are important for the individual and society. First, if an individual cancer survivor leaves the labour force because of the cancer, it may be an indication of serious negative (long-term) effects on health and well-being. Similarly, reductions in working hours or earnings may be an indication of negative effects on health and well-being. Also, negative effects on employment and earnings for cancer survivors represent an economic loss to society in the same way as the loss of lifetime earnings due to cancer mortality (Bradley et al., 2008). Previous studies have found socioeconomic gradients in cancer survival (e.g. Dalton et al., 2007 and Carlsen et al., 2008a) which may involve differences in the timing of diagnosis, in treatment, in recurrences/second cancers and in other factors. For cancer survivors, these factors may also be important for potential social inequality in effects of cancer on labour market outcomes, but other factors may be important as well, for instance differences in job options, which may depend on education, labour market experience and type of job before the cancer diagnosis. This paper focuses on labour market outcomes (labour market attachment, earnings and total income) of cancer survivors up to three years after the diagnosis. We estimate causal effects of cancer using different identification strategies and investigate if there are educational gradients in these effects. We also analyse some mechanisms which might explain the educational gradients. Different forms of cancer may have very different consequences for survivors in terms of labour market outcomes. In this paper we focus on two forms of cancer with high survival rates which strike relatively many people of working age: Breast cancer and colorectal cancer.1 In a Grossman (1972) type model a negative health shock, e.g. due to cancer, may reduce labour supply by increasing time spent on being ill, raising tastes for leisure, decreasing productivity and increasing time needed for investment in health maintenance. However, if health insurance is tied to employment as it is in the US, a negative health shock may increase labour force participation and work effort for those who are still able to work, since the costs of losing their job and thereby their health insurance may increase; see Currie and Madrian (1999) and Bradley et al. (2012). Since we use Danish data and since the whole population in Denmark is covered by public health insurance, we expect negative effects of cancer on labour market participation and earnings. The contributions of the present paper are the following. We use a large longitudinal administrative Danish dataset. We select for the years 2000–2004 all cases of breast and colorectal cancers of 30–60-year-olds who did not have cancer earlier and who survived at least three years after the diagnosis. We estimate separate effects for each type of cancer comparing with a large control group of people with no cancer diagnosis. There are no problems of attrition (except for death and emigration). Our sample is much larger than other population-based prospective and longitudinal studies estimating impacts of cancer on labour market attachment and earnings; see e.g. the survey by Steiner et al. (2004) and additional references in Moran et al. (2011). We are able to condition on a large number of baseline variables including demographics, education, labour market outcomes and health conditions. We observe outcomes for each person in each year for several years before and after the base year (year of diagnosis). We consider many outcomes including labour market attachment, earnings and total income before and after taxes. The large number of cancer cases in the dataset allows us to estimate separate effects for each type of cancer with great precision. We estimate the ‘average treatment effect on the treated’ using propensity score weighting techniques to estimate the counterfactual outcomes of cancer patients in case they did not have cancer, conditioning on baseline variables. We conduct robustness checks using matching estimators and difference-in-differences estimations. In other robustness checks we use an alternative control group of later cancer patients, which to our knowledge is a new approach in the estimation of effects of cancer on labour market outcomes. We investigate if there is social inequality (educational gradients) in the effects of cancer on labour market outcomes and explore some potential mechanisms behind such inequality.2 We also present separate estimates for survivors with and without recurrences and/or second cancers. We find that for patients with breast or colorectal cancers, the average risk of leaving the labour force increases by 5–8 percentage points 1–3 years after the diagnosis. There is a significant social gradient. For patients with no education beyond compulsory school the effect is 8–11 percentage points whereas it is 1–5 percentage points for patients with a further or higher education. A large share of the patients leaving the labour force receive disability pension which implies leaving the labour force permanently due to health conditions. The negative effects on total income are small. Estimates also indicate only very small negative effects on hourly wages and earnings for those who work (although these estimates may be biased due to induced sample selection, since those who remain employed, in spite of having cancer, may have a greater commitment to their careers, suggesting that the true effects could be more negative). The educational gradients in the risk of leaving the labour force are not explained by differences in the risks of metastasised cancer at diagnosis or second cancers/recurrences across education groups, since these differences are very small. Effects of cancer on the risk of leaving the labour force are much larger for those who were blue-collar workers at baseline than for those who were white-collar workers, and stratification by job type at baseline reduces educational gradients in effects of cancer on labour market status. The contents of the remainder of this paper are as follows. Section 2 contains a short survey of related studies. In Section 3 we discuss the empirical methods. Section 4 describes the data. In Section 5 we present results, and Section 6 contains conclusion and discussion.
نتیجه گیری انگلیسی
We estimate effects of breast and colorectal cancer on labour market outcomes and income for three-year survivors in Denmark using different estimation strategies: propensity score weighting (based on the selection on observables assumption) and dif-in-dif (allowing for unobserved individual fixed effects), and we use two different control groups: individuals with no cancer diagnosis, and an alternative control group of later cancer patients. Results are robust to these alternative estimation methods and control groups. We find significant effects of breast and colorectal cancer on labour market participation and disability pension take-up, and a significant educational gradient: Three years after the diagnosis the risk of being out of the labour force increases by about 10 percentage points for people with no education beyond compulsory school, by about 7 percentage points for people with a vocational education, and by about 4–5 percentage points for those with a further or higher education. The corresponding figures for the risk of disability pension three years after the diagnosis are 9, 5–6 and 3–4 percentage points. For both types of cancer the negative effects on total income are small. Estimates indicate only very small negative effects on hourly wages and earnings for those who work, but these estimates may be biased due to induced sample selection, since those who remain employed, in spite of having cancer, may have greater commitment to their careers, suggesting that the true effects could be more negative. Several mechanisms may be important in explaining the educational gradient in effects of cancer on labour market attachment. Firstly, as we have shown, individuals with a lower education have considerably higher baseline risks of leaving the labour force and of receiving disability pension. This may indicate that this group is more vulnerable in general, and therefore it may not be surprising that effects of cancer are larger in absolute terms. Thus, even though we find substantial and significant educational gradients in absolute terms, the differences between education groups are smaller in relative terms, i.e. when considering, e.g. the increased risk of leaving the labour force due to cancer divided by the baseline risk in case of no cancer. Secondly, economic incentives may be important for individuals who are able to work. Thus, people with low education and low income may have stronger incentives to not working and to apply for disability pension because of a higher replacement ratio of public benefits, but on the other hand, people with a higher education and higher income will typically have access to more supplementary compensation from labour market pension schemes or private pension or insurance schemes. Our findings indicate that differences in economic incentives are not very important. Thus, we find only very small negative income effects of cancer and we do not find an educational gradient in income effects. However, there may be significant heterogeneity in coverage, and we do not observe the potential non-participation income of individuals who continue to work, so it might be that highly educated individuals with a high income and poor supplementary coverage have a higher propensity to continue working instead of applying for disability pension. Thirdly, individuals with a higher education might be in better health after cancer treatment because they are more aware of likely symptoms so that the cancer was discovered at an earlier stage, or because they receive different treatment, or because they are more efficient in terms of following advice from doctors during treatment and after. However, we found that the risks of metastasised cancer at diagnosis or second cancers/recurrences are not smaller for patients with a higher education. Fourthly, a larger share of people with a lower education may have to accept physically demanding jobs which may be especially difficult to attend to after a cancer disease. Our results based on stratification by type of job at baseline (indicators of blue- and white-collar workers) are consistent with this hypothesis. Thus, some of the educational gradients in effects of cancer on employment are reduced considerably according to point estimates. However, standard errors in this analysis are large since very few with a further education are blue-collar workers and very few white-collar workers are low-educated. This is not surprising, especially since the indicators of job type are primarily based on the level of qualifications needed in the job rather than how physically demanding the job is. Nevertheless, if one of the mechanisms behind the educational gradients in effects of cancer on employment, which we find, is that a larger share of low-educated patients had physically demanding jobs at baseline, retraining policies targeted on this group which increase their chances of finding other types of jobs might be effective. Interventions that improve early detection and treatment and reduce symptoms of cancer patients in general may reduce educational gradients in employment effects of cancer, since improved ability to work may be more important for the low-educated who may to a greater extent have to accept physically demanding jobs. It is important to note that effects of cancer (and other serious health conditions) on labour market outcomes and income may to some extent depend on the institutional setting, especially the welfare system, labour market policies and the prevalence and coverage of private health insurance and pension schemes. Therefore, our findings may not generalise to other industrialised countries, especially countries with very different institutional settings such as the US. This may in particular be true for effects on income and earnings and for very short-term employment effects. Our estimates of employment effects three years after diagnosis are largely consistent with results in Moran et al. (2011) and Short et al. (2008) for the US.