## Friday, 25 March 2011

### More on PPPs: theory this time

The basic problem with PPPs is the writing of the contract for whatever project is being undertaken. Depending on what can be contracted on and what can't, PPPs may or may not be the way to go. To, hopefully, make this a little clearer a look at a paper by Oliver Hart is useful.

In Hart (2003) Hart put forward a simple incomplete contracts model of PPPs. Consider a situation where, for example, a government wants a new prison built and run. Lets assume there are two periods, the “build” period followed by the “operate” period. There are (unverifiable) social benefits from the prison which we will call B. The (unverifiable) costs of the prison are denoted C. Both B and C are affected by investments that the builder can make. There are two forms of investment available to the builder, i and e. An increase in i increases B and decreases C and so is beneficial all round, while an increase in e decreases both B and C. This means the builder gains from e and “society” doesn’t. Therefore we will call i productive investment and e unproductive investment. The total costs of investment for the builder are i+e. i and e are unverifiable and thus can not be contracted on.

Now consider types of contracts for building and operating the prison. First, separate contracts for building and operating the prison. Call this unbundling. Second, a PPP contract, where the builder both builds and runs the prison. This is bundling. Under unbundling, the builder sets i=e=0. That is the builder builds the cheapest prison possible while staying within the contract. Under bundling the builder sets the marginal decrease in his costs, C, due to an increase in both e and i each equal to 1.
The trade-off between unbundling and bundling is simple. Under unbundling, the builder internalises neither the social benefit B nor the operating cost C. By setting i=e=0, he does too little of the productive investment, i, but the right amount of the unproductive investment, e. In contrast, under bundling or PPP, the builder again does not internalise B, but does internalise C. As a result, he does more of the productive investment, although still too little, but also more of the unproductive investment.

The model yields a simple conclusion. Conventional provision (‘unbundling’) is good if the quality of the building can be well specified, whereas the quality of the service cannot be. Under these conditions, inderinvestment in i under conventional provision is not a serious issue, whereas overinvestment in e under PPP may be. In contrast, PPP is good if the quality of the service can be well specified in the initial contract (or, more generally, there are good performance measures which can be used to reward or penalise the service provider), whereas the quality of the building cannot be. Under these conditions, underinvestment in i under conventional provision may be a serious issue, while overinvestment in e under PPP is not.
The upsot of all of this is that the choice between a conventional unbundled contract and a PPP bundled contract turns on whether it is easier to write contracts on the operating phase of the prisons life than on the building phase. So the usefulness of PPPs depends on what can or can't be contracted on.
• Hart, Oliver (2003). ‘Incomplete Contracts and Public Ownership: Remarks, and an Application to Public-Private Partnerships’, The Economic Journal, 113 (March), C69-C76.