Michaela Neumayr & Astrid Pennerstorfer, WU Vienna University of Economics and Business.
What amount of donations can you expect from people when fundraising for your organization? Who are the people who do not donate any money at all, and who are those who donate substantial amounts? Basically, charitable donations are considered to strongly vary across income classes, and fundraisers know that the amounts of individuals’ donations increase with their financial resources. But what proportion of income can you expect to be donated by people of different income groups? Which income groups are more ‘generous’, those with higher or lower incomes?
Such questions have been discussed in scholarly research since the 1990s and answers to these questions are surprisingly diverse and even contradictory. In the debate concerning the “charitable giving profile” all kinds of shapes have been proposed. Some argue that it follows a U-shaped curve, with individuals at both ends of the income distribution donating the highest proportions of their income. For the US, for instance, those below an annual income of 10,000 USD donate about 4.6% of their income, while those with an income higher than 150,000 USD give 2.2%, and those in the middle 1.4%. Other studies do not find that lower-income groups are more generous, but rather describe the charitable-giving profile as a flat curve with an upward slope for higher-income groups. Yet other studies doubt the upwards swing on the right-hand side and instead find higher relative donation levels for lower income groups. We could go on with an account of yet other assumed shapes of previous studies, but we stop at this point to tell you about our study.
Our study The Relation Between Income and Donations as a Proportion of Income Revisited: Literature Review and Empirical Application consists of two parts. First, we systematically review 26 empirical studies that investigate the relation between income and proportion of income given, trying to make sense of the variety of existing results. In this review, one important explanation for diverging findings is the great number of varying methodological approaches in the previous studies. Second, we use the findings of this systematic review for an empirical application on Austrian income tax data (n = 20,000). In this second part, we are able to demonstrate how the various methods that have been used in previous studies lead to diverging findings. Based on that, we are able to tell which of the previous findings can be considered certain.
Returning to the first part of our study, we find three major reasons why results of previous studies vary so much:
– The first refers to the sample analysed. When describing the charitable giving profile it is crucial whether specific parts of the population are included in the sample (e.g. non-donors) or represented by a group that is large enough to draw unambiguous conclusions (e.g. high-income groups). Additionally, we find that both representative survey data and administrative tax data have their advantages and flaws for use in analysis of the charitable giving profile.
– The second refers to the specification of variables. Previous studies have used different income and donation measures and this variation can also influence the result.
– The third refers to the method of analysis. The method how the charitable giving shape was determined varied greatly in previous studies ranging from mere visual inspection of donation/income ratios for different income classes, over bivariate to multivariate analyses.
In the second part, we take a sample of Austrian income tax data set to show how results regarding the charitable giving profile change when you vary the sample, define the variables differently or vary the statistical method. We find a visual inspection of the relation between income and the proportion of income donated to be especially unsatisfactory because results vary greatly depending on income class definitions and because visual inspection leaves a lot of room for interpretation. Results of various regression analyses point towards a U-shaped charitable giving profile. However, this is not what we find in the results when applying a more precise estimation technique (semi-parametric estimation), which displayed a downward-sloping relation for most income levels. We find this final method especially helpful for this question, as it allows to exactly describing the relationship between the income and proportion donated without any restrictions.
What can fundraisers take away from this for their practice? While our study has shown that results of previous analysis vary with the method in use, three conclusions regarding the relation between income and the proportion donated are safe to say. First, people in the lowest income group donate the largest proportion of income. While the share itself varies depending on the country context, it is always higher compared to other income groups within the respective country. Second, the profile of the charitable giving curve seems to be fairly flat for middle and high income groups. This means, middle and high income groups can be expected to give the same share of their income. Third, regarding the very high income groups the relation is more difficult to describe Important to note, this very high income group refers to the top 5-3% of the income profile only. Within this group, fundraisers still have to find out for themselves what amounts can be achieved, while for all other groups it is known statistically what proportions and thus amounts can be expected.
Click here to access the full NVSQ research article: Neumayr, M. & Pennerstorfe, A. (2020). The Relation Between Income and Donations as a Proportion of Income Revisited: Literature Review and Empirical Application, Nonprofit & Voluntary Sector Quarterly, https://doi.org/10.1177/0899764020977667.
NVSQ Blog 