I’m still nagging away at the implications of the fallacy that spectrum exists. Sorry.
I’ve been struck recently that many if not most definitions of property rights seem to turn on a relationship to an asset. For example, Gary Libecap in Contracting for Property Rights defines them as "the social institutions that define or delimit the range of privileges granted to individuals to specific assets" (1990:1); or Yoram Barzel in The Economic Analysis of Property Rights: "Property rights of individuals over assets consist of the rights, or the powers, to consume, obtain income from, and alienate these assets" (1997:2). Such definitions set out to define rights which assure the owner of an asset that they can derive value from that asset.
However, one can have rights to create value that do not require the existence of an underlying asset – unless, of course, one takes the position that the existence of a property right necessarily implies an asset. 
Therefore, let me distinguish between any property right, which is an asset in itself, and a property right to exploit an asset, which entails two assets: the right itself, and the underlying asset. All assets can lead to property rights – perhaps tautologically, in that something might not be counted as an asset if it does not have rights associated with it – but not all property rights require assets.
It always helps to make things concrete. One property right without an underlying asset is a New York taxi cab medallion: it's a right to operate, but there isn't an underlying asset. The right is tied to a particular place (New York), but that place isn't the asset.
Another common asset-less right is a franchise, that is, an agreement to sell a company's products exclusively in a particular area or to operate a business that carries that company's name.
Perhaps my favorite is a trademark, that is, a word, symbol, or phrase, used to identify a particular manufacturer or seller's products and distinguish them from the products of another. One might use the word “Wired” to brand a magazine, but the word isn’t the asset; when I last counted about a year ago, there were about 27 distinct trademarks using the word "wired" in the US.
Notice that permission for an agent to behave in a particular way is the essence of all these rights – and of rights that require assets, too. Therefore, I’d contend that behavior is the key to property rights, and assets are optional.
There are of course many property rights to assets, from owning a pencil to the right to extract oil in a particular region. Note that the underlying assets don't have to be tangible: an algorithm over which one has a patent is a perfectly viable intangible asset (perhaps made so exactly by the property right).
This distinction between property rights that do and do not require underlying assets matters: if one assumes an underlying asset where there is none, one is liable to over-assign rights.
For example, if trademark regulation assumed that the word being used was the asset, then it might give the owner of the trademark the right to all possible (commercial) uses of the word. There would be only one “Wired” trademark in the US, let’s say owned by Condé Nast; the companies who wanted to use the word to sell cologne, art supplies, energy drinks, stationery, electronic door chimes or automobile wheels would be out of luck. This would be a loss because an entrepreneur could apply the letters w-i-r-e-d to some new product that couldn’t be confused with a magazine without seeking (and probably failing to get) Condé Nast’s permission.
Similar reasoning applies to radio regulation. The existence of radio licenses doesn’t mean that there is an underlying asset, “spectrum”. 
If one regards a radio channel as an asset, then (Anglo-American) regulators have shown a proclivity to grant an expansive array of rights. Following the norm of technology and service neutrality, they have defined operating rights so broadly that pretty much preclude all operations that radiate energy in that channel, regardless of its harm to the licensee, in order to allow the licensee to operate in any conceivable way.  Such a broad definition forecloses new entry by potentially useful but non-interfering services
A broad definition also forecloses future arrangements of radio operating rights that are not tied to channel-based world view. Bands and channels, as regulatory constructs, are in large part a consequence of the two-stage super-heterodyne radio design that first filters a broad range of frequencies at the "RF stage", and then after down-conversion picks out a narrow range at the "IF stage". This is an old-fashioned approach that is increasingly becoming obsolete  - but it is enshrined in regulation.
Barzel, Yoram, Economic Analysis of Property Rights, Cambridge University Press 1989, second edition 1997
Libecap, Gary D., Contracting for Property Rights, Cambridge University Press 1990
 A view of property rights that does not require the existence of underlying assets is not identical to the "bundle of rights" approach taught in law school property classes; there it's taken as a given that there's an underlying asset - paradigmatically, real estate - and the bundle explains how it can be simultaneously “owned" by multiple parties.
 The emergence of the spectrum concept suggests that this is, indeed, the conclusion that has been drawn. Perhaps the reasoning that a property right must entail an asset is one of the reasons why “spectrum” has become such an entrenched concept.
 I’m ignoring allowed inter-channel interference; for a discussion of that case, see my report on the meeting held at Silicon Flatirons, “Defining Inter-Channel Operating Rules”
 See e.g. Soni & Newman 2009, "Direct conversion receiver designs enable multi-standard/multi-band operation", RF Designline