Information, belief elicitation and threshold effects in the 5X1000 tax scheme: a framed field experiment

Abstract

In this paper, we study by means of a framed field experiment on a representative sample of the population the effect on people’s charitable giving of three, substantial and procedural, elements: information provision, belief elicitation and threshold on distribution. We frame this investigation within the 5X1000 tax scheme, a mechanism through which Italian taxpayers may choose to give a small proportion (0.5%) of their income tax to a voluntary organization to fund its activities. We find two main results: (i) providing information or eliciting beliefs about previous donations increases the likelihood of a donation, while thresholds have no effect; (ii) information about previous funding increases donations to organizations that received fewer donations in the past, while belief elicitation also increases donations to organizations that received most donations in the past, since individuals are more likely to donate to the organizations they rank first.

Appendices

Appendix A: Instructions

1.1 General instructions

Sardegna Solidale and the University of Cagliari are grateful for your decision to participate to this research. We ask you to fill the following questionnaire in all its parts. Try to answer autonomously to all questions and, in case of necessity, ask for the support of our researcher. Answers will be evaluated by us in anonymous form and elaborated in aggregate. Researchers will not be able in any case to retrieve the respondent’s identity.

1.2 The choice questionnaire task

By filling this questionnaire, you will take part in a lottery organized by the Department of Economics and Business of the University of Cagliari. The drawn will take place on the 31st of October 2013, one all the questionnaires will be collected. The winner’s identity will be kept anonymous to the researchers and she/he will be contacted by the Sardegna Solidale volunteers by means of the numeric code you received.

The first and unique prize is equal to €1000. However, this money will not go to the winner. She/he can decide whether to give it or not to one charitable organization among those from the list below. If the prize is not given the winner will not receive anything anyway.

Now we ask you to imagine that you have already won the prize. You now have € 1000 that you could give or not to give to one among the not-for-profit organization listed below.

If after the drawn you will result as the lottery winner, the decision you are about to make will be implemented for real. That means that if you decided to give to some organization the prize, such an organization will receive the money for real, otherwise, if you decided not to give, the prize will not be distributed.

1.3 (In the no-info treatment)

Tick the box corresponding to your choice:

Tick your preferred option	Organizations
$\square$	NO DONATION
$\square$	EMERGENCY
$\square$	UNICEF – ITALIA
$\square$	L.A.V. _ LEGA ANTIVIVISEZIONE
$\square$	WWF _ WORLD WIDE FOUNDATION ITALIA
$\square$	GREENPEACE
$\square$	AMNESTY INTERNATIONAL - SEZIONE ITALIANA
$\square$	CARITAS ITALIANA
$\square$	BANCO ALIMENTARE ONLUS
$\square$	ADMO _ ASSOCIAZIONE DONATORI MIDOLLO OSSEO

1.4 (In the info treatment)

Tick the box corresponding to your choice:

Tick your preferred option	Organizations	Funding Received in 2011 through 5X1000 (€)
$\square$	NO DONATION	–
$\square$	EMERGENCY	11,023,415.00
$\square$	UNICEF—ITALIA	5,460,307.00
$\square$	L.A.V._LEGA ANTIVIVISEZIONE	1,176,578.00
$\square$	WWF_WORLD WIDE FOUNDATION ITALIA	1,021,070.00
$\square$	GREENPEACE	758,835.00
$\square$	AMNESTY INTERNATIONAL—SEZIONE ITALIANA	753,674.00
$\square$	CARITAS ITALIANA	193,890.00
$\square$	BANCO ALIMENTARE ONLUS	170,351.00
$\square$	ADMO_ASSOCIAZIONE DONATORI MIDOLLO OSSEO	68,828.00

1.5 (In the no-info+beliefs elicitation treatment)

Before making your choice we ask you to order (by assigning a specific rank) each organization in terms of how much funding you think they received the previous year through the 5X1000 mechanism (denote with 1 the organization that raised more money and with 9 the one that raised less and with all the other numbers 2-8 those in the intermediate positions)

Tick the box corresponding to your choice:

Tick your preferred option	Organizations	Ranking Indicate the position of each organization in term of funding received latest year (1 = first... 9 = last)
$\square$	NO DONATION	–
$\square$	EMERGENCY
$\square$	UNICEF—ITALIA
$\square$	L.A.V. _ LEGA ANTIVIVISEZIONE
$\square$	WWF _ WORLD WIDE FOUNDATION ITALIA
$\square$	GREENPEACE
$\square$	AMNESTY INTERNATIONAL - SEZIONE ITALIANA
$\square$	CARITAS ITALIANA
$\square$	BANCO ALIMENTARE ONLUS
$\square$	ADMO _ ASSOCIAZIONE DONATORI MIDOLLO OSSEO

1.6 (In the no-info+beliefs elicitation + threshold treatment)

Before making your choice we ask you to order (by assigning a specific rank) each organization in terms of how much funding you think they received the previous year through the 5X1000 mechanism (denote with 1 the organization that raised more money and with 9 the one that raised less and with all the other numbers 2–8 those in the intermediate positions)

Note that if the amount of the aggregate donations is greater than a given threshold, only a fraction of the € 1000 will be actually distributed to the organization.

Tick the box corresponding to your choice:

Tick your preferred option	Organizations	Ranking Indicate the position of each organization in term of funding received latest year (1 = first... 9 = last)
$\square$	NO DONATION	–
$\square$	EMERGENCY
$\square$	UNICEF–ITALIA
$\square$	L.A.V. _ LEGA ANTIVIVISEZIONE
$\square$	WWF _ WORLD WIDE FOUNDATION ITALIA
$\square$	GREENPEACE
$\square$	AMNESTY INTERNATIONAL - SEZIONE ITALIANA
$\square$	CARITAS ITALIANA
$\square$	BANCO ALIMENTARE ONLUS
$\square$	ADMO _ ASSOCIAZIONE DONATORI MIDOLLO OSSEO

Appendix B: Clustered analysis

1.1 B.1 Information effect (NoInfot vs InfoT)—Sect. 3.2

Information provision leads donors to act systematically more in favor (with respect to the no information treatment) of the less endowed charities of cluster C.

The main result is confirmed also in this alternative clustered setting (see Table 8; Fig. 3). The categorical distribution of recipients under InfoT is contrasted against the NoInfoT distribution. Donations density moves from cluster A (−5 percentage points) and B (−11 percentage points) in favor of cluster C’s charities (+16 percentage points). This qualitative polarization of the donation behavior results to be highly statistically significant according to Pearson $X^{2}$ test ($p\hbox { value}<0.001 {\vert }\hbox { tag}: \hbox {b}$).

Table 8 Donations received by each of the nine organizations in the previous fiscal year through the 5X1000 mechanism

Fig. 3

Distribution of donations (by treatment)—clustered

1.2 B.2 The beliefs effect (NoInfoT vs BeliefT)–Sect. 3.3

Also in this case, the analysis at cluster level helps to better highlight the underling dynamic (Table 8; Fig. 3). Visual inspection of Fig. 3. suggests how the U-shaping effect is mainly driven by an increase of donations in favor of better endowed charities belonging to cluster A (+10 percentage points) and a correspondent decrease of donations to the less endowed ones of cluster B (−8 percentage points) and C (−2 percentage points). As for the previous test, also in this further clustered configuration the Pearson $X^{2}$ test rejects $(p\hbox { value}=0.008\, {\vert }\, \hbox {tag}: \hbox {d})$ the null hypothesis of independence between experimental conditions and donation allocations to the clustered recipients.

1.3 B.3 Threshold effect (BeliefT vs Belief&ThresholdT)—Sect. 3.4

The same result holds true when the clustered categorization is considered. Visual inspection of Fig. 3 (see also Table 8) confirms how the two clustered categorical distributions display analogous patterns of donations by treatments (cluster $\hbox {A}\approx 35\%$; cluster $\hbox {B}\approx 16.5\%$; cluster $\hbox {C}\approx 48.5\%$). The Pearson $X^{2}$ test fails to reject $(p\hbox { value}>0.7 {\vert } \hbox {tag: f})$ the null hypothesis of independence between experimental conditions and donation allocations to the clustered recipients (Tables 9, 10).

Table 9 Bonferroni–Holm stepwise multiple-hypotheses-testing procedure (Holm 1979)

Table 10 Distribution of donations (by information, belief elicitation and threshold)—clustered