In this essay, I’m going to talk some of the requirements that have to be fulfilled if someone wants to create a trust. (As that trust has been deliberately created, it will of necessity be an express trust.) These requirements can be simply summarised.
First of all, you can’t unintentionally create a trust so the most basic requirement is that you manifest an intention to create a trust. (This is known as the ‘certainty of intention’ requirement.)
Secondly, you can’t create a trust over property that you don’t currently own so you have to own the trust property before you can create a trust over it.
Thirdly, you have to identify the property which is to be the subject matter of the trust – otherwise the courts won’t know what item of property is to be held on trust. (This is known as the ‘certainty of subject matter’ requirement.)
Fourthly, the trustees of the trust and, more importantly, the courts have to know enough about who you intend to be benefited by the trust and how for them to give effect to your intentions. (This is known as the ‘certainty of objects’ requirement.)
Fifthly, in the case where you intend to create a trust not by declaration (i.e. by taking an item of property that belongs to you and declaring that you hold it on trust for someone else) but by transferring property belonging to you to someone else to be held on trust for a third party, title to the putative trust property must be transferred to the intended trustee or trustees before the trust will come into existence.
Sixthly, if you want to create a trust over a piece of land, you must express your intention to create the trust in writing, otherwise the trust will be unenforceable: Law of Property Act 1925, section 53(1)(a),(b).
This essay focusses on the second, third and fourth requirements.
Suppose that I know my aunt is going to leave me a very valuable silver coffee pot after she dies, in her will. I say to you, ‘I declare that when my aunt dies and I get the coffee pot, I will hold it on trust for you’. This declaration is of absolutely zero effect – if my aunt dies and I get the coffee pot, I’ll be free to do what I like with it. More tricky is the following situation. Suppose that my aunt has died but her executors have yet to give effect to her will and so I have yet to receive the coffee pot which she left me in her will. At that point I say to you, ‘I declare that I will hold the coffee pot on trust for you’. Whether or not my declaration will have had the effect of imposing a trust over the coffee pot for your benefit depends crucially on whether I had an interest in the coffee pot at the time I made that declaration. The decision of the Privy Council in Commissioner of Stamp Duties v Livingston (1965)indicates that at the time I make my declaration, I do not have any kind of interest in the coffee pot – I merely have a right to sue the executors to get them to perform their duty to give effect to the will. So, again, my declaration will be of zero effect and when I do eventually get the coffee pot, I will be free to do what I like with it.
Certainty of subject matter
Suppose I own 1000 bottles of wine and I’ve stored them in a large wine cellar. One day, I declare that I hold 500 of them on trust for you without making it clear which 500 bottles I’m talking about. My declaration will be of zero effect because the courts simply won’t know which of my bottles of wine are to be held on trust for you. It’s necessary for them to know this because – what if 20 of the bottles of wine are destroyed in an earthquake which hits the wine cellar? The courts need to know whether the bottles of wine which have been destroyed were held on trust for you or not – and they will not if I haven’t already made it clear exactly which of my 1000 bottles of wine are held on trust for you. Notice that the same point holds true if the 1000 bottles are identical in every respect – if 20 of them were destroyed in a storm and we had to determine whose bottles of wine had been destroyed, it would be of no comfort to know that the 20 destroyed bottles were identical in every respect to the surviving 980.
In Hunter v Moss (1994) the requirement that there be certainty of subject matter before a trust can come into existence was weakened somewhat. In that case A owned 950 shares in Company X and declared that he held 50 of them on trust for B without specifying which of the 950 shares were to be held on trust for B. The trust was upheld as valid. The decision can be criticised.
What if A, after making his declaration of trust, had sold 500 of his shares in Company X and used the proceeds of the sale to acquire shares in Company Y. We will need to know whether in doing this A sold the shares in Company X that he held on trust for B and invested the proceeds in Company Y. (If A loses his investment in Company Y he might well want to claim that he did sell B’s shares and invest the proceeds in Company Y; on the other hand, if A makes a lot of money in investing in Company Y, B might well want to claim that his shares contributed to that investment.) But because A didn’t make it clear which of his 950 shares were held on trust for B, we can’t tell whether B’s shares were sold and the proceeds invested in Company Y. In order to avoid this difficulty, the Court of Appeal should have held in Hunter v Moss that the declaration of trust was invalid for lack of certainty of subject matter.
One of the reasons the Court of Appeal gave for finding that the trust was valid in Hunter v Moss was that if A had left B 50 of his 950 shares in Company X in his will, the bequest would have been perfectly valid – why should it matter that A chose not to give B the shares in his will but instead chose to allow B to enjoy them in his lifetime by declaring that he held them on trust for B? But the crucial difference is that if A had left B 50 of his 950 shares in Company X in his will, the executors of A’s estate would have taken 50 of the 950 shares and handed them over to B, thus separating B’s shares from the other shares in Company X. In Hunter v Moss nothing was done to separate out the shares in Company X that were to be held on trust for B and the shares which A remained absolute legal owner of.
Similarly, if A had 1000 identical bottles in his wine cellar and left 50 of them to B in his will without identifying which 50 should go to B, there would be no problem with the bequest – the executors of A’s estate would simply go into the wine cellar, take 50 bottles out and give them to B, thus separating B’s bottles from the other wine bottles. Where there might be a problem is if A had 1000 bottles in his wine cellar – all of different types and quality – and he left 50 bottles from his wine cellar to B in his will. The bequest here would probably be invalid because the executors would not be able to tell which wine bottles they were supposed to give to B. A would not have told them enough about his intentions for them to give effect to them.
Certainty of objects
This takes on to our next point which is that for a valid trust to be created, the person creating the trust (who is known as the settlor) must give the courts enough information about who he intends to be benefited under the trust, and how, for the courts to give effect to that intention. The crucial question in applying this requirement is – do we know enough about the settlor’s intentions to give effect to them? If we do, then we have no problem. If we don’t, then the courts can’t uphold the trust which was sought to be created. To see this approach in action, let’s look at some bequests in a will made by A which raise ‘certainty of objects’ issues:
(1) ‘£100 to be given to each of my friends.’
Now the concept ‘friend’ is pretty fuzzy, but if we know of at least one person who was A’s friend then we know enough to give effect to A’s intention in making this bequest. Say we know for sure that however fuzzy the concept ‘friend’ is, B was definitely one of A’s friends. So we know that A, in making this bequest, intended that B should get £100. So we give effect to that intention and give B £100. Okay – now there may be other people who we’re not really sure about, as we don’t know whether they were A’s friends or not. These people won’t get anything under this bequest – we don’t know enough about them to know whether or not A intended them to get £100. But this won’t affect the validity of the gift to B. This was the result reached in Re Barlow (1979), where the testatrix specified that her ‘family and friends’ should be allowed to buy paintings from her estate at a special discount. The gift was held to be valid: those who could clearly establish that they were ‘family’ or ‘friends’ and who wanted to buy a painting should get a discount. The fact that there would be others who could not clearly establish that they were ‘family’ or ‘friends’ and who could not therefore get a discount was neither here nor there. Note that you certainly should not read Re Barlow as saying that any gift for ‘friends’ will be valid. As the next example shows, it all depends on the terms of the gift.
(2) ‘£10,000 to be distributed equally among my friends.’
(Notice that this isn’t an example of a discretionary trust – the executors have no discretion as to how to distribute the £10,000 among A’s friends; they are told to distribute the amounts in equal shares.)
Now here we don’t know enough to give effect to A’s intentions in making this bequest. To know how much each friend should get, we would have to know how many friends A had – and the fuzziness of the concept ‘friend’ makes it impossible for us to put a figure on how many friends A had. So our basic problem here is one of uncertainty of subject matter – we don’t know enough here to know how much each person should get under the bequest. The bequest will therefore be invalid.
(3) ‘£10,000 to be distributed equally among my ten best friends.’
Here the problem we had in (2) is solved for us. We know how much each of the ten best friends are to get – they are to get £1,000 each. So this bequest is identical to a bequest which goes ‘£1,000 to be distributed to each of my ten best friends’, which looks pretty similar to the bequest in (1). Now – if we know of one person who was definitely one of A’s ten best friends, then we know enough to give effect to A’s intentions in setting up the bequest. Say B was definitely one of A’s ten best friends. In that case, we have no problem – we simply give B £1,000. Now, we may not know who were the other nine people A was referring to when he made his bequest, but that shouldn’t affect the validity of the gift to B.
(4) ‘£10,000 to be distributed as my executors see fit among the inhabitants of my hometown.’
Applying our yardstick – do we know enough to give effect to A’s intentions? – A clearly didn’t intend here either that the executors should give out the £10,000 in equal shares to all the inhabitants of his hometown or that the executors should consider the merits of each and every inhabitant of his hometown in deciding how to distribute the £10,000. So we don’t need to know – in order to give effect to A’s intentions – how many people live in A’s hometown and we don’t need to have – in order to give effect to A’s intentions – a complete list of all the inhabitants of A’s hometown. So the fact that we don’t know how many people live in A’s hometown and the fact that we can’t draw up (because it would take too long) a complete list of all the inhabitants of A’s hometown should not invalidate this discretionary trust. And indeed the House of Lords took the same view in McPhail v Doulton (1971)
Instead, the House of Lords said that a discretionary trust for a given class of people will be invalid if you can’t tell of any given individual in the world whether they fall within or without that class. So – according to McPhail v Doulton – the discretionary trust here will fail if there exist some people in the world of whom we can’t say whether or not they are inhabitants of A’s hometown. For example, suppose B owns a holiday home in A’s hometown – is B to be counted as an inhabitant if he only spends a month a year there? If you are in doubt – you don’t know whether B should be counted as being inside or outside the class to be considered – then, according to McPhail v Doulton, the discretionary trust will be invalid.
This seems too strict. If we just ignored B and decided to give out the money to those who are clearly inhabitants of A’s hometown, that would be consistent with A’s intentions. If so, then giving effect to A’s intentions doesn’t require us to know of any given person in the world whether they do or do not fall within the class ‘inhabitants of my hometown’. We simply need there to be a good number of people who clearly fall within that class so that giving the money out among those people would be consistent with A’s intentions. (The question of what counts as a ‘good’ number of people must be judged against A’s intentions. Suppose I set up a discretionary trust: ‘£100,000 to my trustees to be distributed as they see fit among those I taught while I was at Pembroke College, Cambridge.’ The records of who I taught have been irretrievably lost but about ten people can prove that I taught them while I was at Pembroke. And suppose that we know I must have taught about 1,000 people in my time at Pembroke. Giving my money out among the ten people who can prove I taught them while I was at Pembroke wouldn’t really be consistent with my intentions when I set up the discretionary trust – I intended that my trustees would distribute the money across a much wider range of people. But if 500 people could prove that I taught them while I was at Pembroke, that might be constitute a good number of people.)
At the moment, the law occupies a halfway house between the strict position taken in McPhail v Doulton – for a discretionary trust to be valid, you’ve got to be able to tell of anyone in the world whether or not they fall within the class to be considered – and the position I’d like it to take, under which a discretionary trust will be valid if a sufficiently large number of people clearly fall within the class to be considered.
If the reason why you can’t tell of any given individual in the world whether or not they fall within the class to be considered is because the class is defined in vague or uncertain terms – in other words, is conceptually uncertain – then the strict position taken by the House of Lords in McPhail v Doulton will prevail and the discretionary trust will be automatically held to be invalid. So a discretionary trust for ‘my friends’ will be invalid because the definition of the class is such that there will be some people of whom we will not be able to say whether or not they count as being ‘my friends’ (they are not clearly my friends, but they are not clearly not my friends either). And this is so even if there is a good number of people who clearly are my friends. In such a case, as I have said, holding that the discretionary trust is invalid seems overly strict – the trustees of the trust would be able perfectly well to give effect to my intentions by just considering those who are clearly ‘my friends’.
But if the reason why you can’t tell of any given individual whether or not they fall within a particular class is that while the class is clearly defined, there are evidential difficulties that prevent us saying of some people whether or not they fall within the class, the Court of Appeal has held – in Re Baden’s Will Trusts (No 2) (1972)– that the strict approach adopted in McPhail v Doulton should not be followed.
In that case, the Court of Appeal considered whether a discretionary trust for ‘relatives of employees’ of the testator’s company was valid or not. Stamp LJ defined relatives as meaning ‘next of kin’ and found that it was perfectly possible to tell of any given individual whether or not they were a ‘relative’ of an employee of the testator’s company. So he held that the trust was valid according to the test for validity laid down in McPhail v Doulton.
In contrast, the majority – Megaw and Sachs LJJ – defined ‘relatives’ much more widely. They held that A will be a ‘relative’ of an employee B if A and B are descended from a common ancestor. On this definition, the class to be considered in Re Baden’s Will Trusts (No 2) was conceptually certain. The definition of the class didn’t create any problem with telling whether a given individual fell within the class or not. With perfect family trees, one could easily tell whether someone was a ‘relative’ of an employee. But the problem was that we don’t have perfect family trees. So the class ‘relatives of employees’ was conceptually certain but evidentially uncertain. Evidential difficulties meant that it was impossible to tell of large numbers of individuals whether or not they were ‘relatives of employees’. Despite this, both Megaw and Sachs LJJ held that the discretionary trust should be declared valid: evidential uncertainty should not, they held, defeat the trust. In so ruling, they departed from the strict test laid down in McPhail v Doulton. However, they argued – contrary to the position taken by Stamp LJ – that the strict test in McPhail v Doulton was not intended to apply in cases of evidential uncertainty, only in cases of conceptual uncertainty where vagueness or uncertainty in the definition of the class made it difficult to tell of any given individual whether they fell within or without the class. However, Megaw and Sachs LJJ disagreed over when evidential uncertainty would render a discretionary trust invalid. Sachs LJ took the very liberal position that so long as the evidential uncertainty didn’t prevent us being able to tell of at least one person that they fell within the class to be considered, then the trust should be declared valid. Megaw LJ was more cautious, holding that evidential uncertainty would not render a discretionary trust invalid so long as we could still tell of a sufficiently large number of people that they fell within the class to be considered.
I prefer Megaw LJ’s position; indeed, as I have already said, I think that the same test should be applied to determine whether a discretionary trust that is afflicted by conceptual uncertainty is valid. Stamp LJ’s approach – holding that a discretionary trust will only be valid if the class to be considered is both conceptually and evidentially certain – is too strict: we don’t need to tell of everyone in the world whether or not they fall within the class to give effect to the settlor’s intentions. And Sachs LJ’s position is too liberal: it opens the door to our holding that a discretionary trust is valid even when evidential difficulties mean that there are only two or three people who clearly fall within the class to be considered when the settlor might have intended the trustees of the trust to select from a much bigger class.
Having said all that, it doesn’t seem that there is anything in Re Baden’s Will Trusts that will save the legacy we are considering here. The class ‘inhabitants of my hometown’ seems to be conceptually, rather than evidentially, uncertain. The definition of the class is what makes it difficult for us to tell whether or not a given individual falls within the class to be considered, not any lack of knowledge of the facts on the ground. (One useful way of testing whether we are dealing with a situation of conceptual uncertainty, as opposed to evidential uncertainty, is to ask yourself: ‘If we had perfect information, would we be able say of any given individual whether or not they fall within the class to be considered?’ If the answer is ‘no’ then the class is conceptually uncertain.)
(5) ‘£10,000 to be distributed as my executors see fit among the people in this photograph, taken of Cambridge Market Square on October 16 2012 (names and addresses of everyone in the photograph attached to the photograph).’
Again we are dealing here with a discretionary trust, but this time there is no problem with certainty of objects as the information we have allows us to tell of any given individual whether or not they fall within the class to be considered. But there is still a big problem with giving effect to A’s intentions here. The class is capricious in the sense that the executors will have no idea what they are supposed to considered in deciding how to give out the £10,000 allocated for distribution among the members of this class. They have no idea what criteria A wanted them to have in mind when choosing that this one should get some money and this one shouldn’t get anything. Even splitting the money equally among all the members of the class would be capricious as the executors have no assurance that equal distribution would have been consistent with A’s intentions. So the fact that the executors have no idea how to go about giving effect to A’s intentions in giving out the £10,000 will render this trust invalid, for capriciousness. Of course, the problem of capriciousness could be solved if A had separately supplied the executors with a letter of wishes as to how they should go about exercising that discretion. This may account for why so-called ‘intermediate’ or ‘hybrid’ trusts (under which money is held on trust to be distributed to ‘anyone in the world’ except for the settlor and his family) have been upheld by the courts even though they look capricious (see Re Manisty (1974) and Re Hay (1981)). The courts know very well that the trustees of these trusts know exactly what they are supposed to do under these trusts; and that these trusts have been very deliberately set up as a tax planning device to allow the settlor to determine through his trustees how the money that he has settled on them on trust is used while making it clear to the tax authorities that the settlor and his family have absolutely no taxable interest in that money.
(6) ‘£1 million to be distributed among everyone in the world sharing the same surname as me.’
Again we are dealing with a discretionary trust here, and there is again no problem with certainty: we can tell of any given individual whether or not they fall within the class to be considered by simply looking at their surname. This trust looks at first sight like it might be invalid for capriciousness. But appearances deceive: this trust is not actually capricious. This is because there is one distribution that will be consistent with A’s intentions in making this provision, and that is equal distribution among everyone with the same surname as him. An unequal distribution – where some in the class get something and others get nothing – would be capricious as the executors would have no assurance that such a distribution would be consistent with A’s intentions. However, the executors could be confident that an equal distribution would be consistent with A’s intentions, as he must have felt some affinity for everyone who shared the same surname as him in making this provision. However, an equal distribution would be impossible here as it is impossible to determine how many people share the same surname as A. Given this, the discretionary trust here will be declared to be invalid for administrative unworkability.
This is, in my view, the problem that Lord Wilberforce intended to address when he suggested in McPhail v Doulton that a discretionary trust would be invalid on the ground that it was ‘administratively unworkable’ if the class to be considered was ‘so hopelessly wide as not to form “anything like a class”’. Textbooks and later decisions (in particular, R v District Auditor, ex parte West Yorkshire MCC (1986)) have focussed on the word ‘wide’ as suggesting that a discretionary trust would be invalid for administrative unworkability where the trust was for a really huge class of people. But a discretionary trust will be perfectly workable for even really huge classes of people if the trustees know well enough what they should be looking for in selecting among those classes.
No – the following analysis seems more sensible. In McPhail v Doulton, the House of Lords moved away from the ‘complete list’ test for determining whether a discretionary trust was valid, according to which a discretionary trust would be invalid if you could not draw up a complete list of all the members of the class to be considered under the trust. They did this because they did not think the trustees of a discretionary trust needed a complete list of members of the class to be considered to give effect to the intentions of the settlor setting up the trust. (As I have observed above, the test for validity that the House of Lords adopted instead – the ‘is or is not’ test: can you tell of any given individual whether they are or are not within the class to be considered? – can be criticised on exactly the same ground.) In so ruling, they opened up the possibility of discretionary trusts for very large classes of people being declared valid for the very first time. However, the House of Lords must have been aware of the possibility that their ruling could give rise to a problem where: (1) a settlor S creates a discretionary trust for a very large, but certain, class of people; and (2) the only way of giving effect to S’s intentions in creating that trust is to distribute the trust funds equally among the members of the class to be benefited under that trust. In such a case, the trustees of the trust fund would be trapped. The discretionary trust would be valid and therefore have to be given effect to. But in order to give effect to the trust, the trustees would need to compile a complete list of the members of the class to be benefited under the trust, when the size of the class makes compiling such a list impossible. It was in order to give the trustees of such a trust an escape route, that Lord Wilberforce suggested that a discretionary trust might be invalid for ‘administrative unworkability’.
So, for me, ‘administrative unworkability’ does not target the problem (such as it is) of giving effect to a discretionary trust for a very large class of people. It is intended instead to deal with the problem created by settlements such as that in our situation (6) – where the only way of giving effect to the settlor’s intention to benefit a very large class of people is to distribute the trust monies equally among all the members of that class.