I've spent a lot of time with Higgs's thinking about GDP. Much of his argument centers around an argument of authority invoking Simon Kuznets. The problem is that Kuznets never really figured out how to deal with government in GDP accounting becuase it was developed before WWII and the rise of the "welfare state" proper. In Kuznet's conception GDP was a kind of welfare accounting and that goods and services were categorized in terms whether they were welfare inducing now (C) or welfare inducing later (I). However, some goods defied this simple categorization, in particular munitions The issue was how do you characterize something like munitions? If it is not welfare enhancing today or tomorrow, the argument runs, then they must be intermediate goods and thus not part of GDP*. But of course you now run into the problem that you are arbitrarily assigning goods as intermediate and final goods based on whether you think they are welfare enhancing or not. The solution to the problem is to adopt the approach the BEA eventually did, which is simply trace the production process of goods to their "end purchaser" and call those final goods. However, this means that GDP is not a measure of welfare, it is an accounting identity.
Now, to the extent that one is tempted to use NIPA as a proxy for welfare the Higgsian view should be criticized because it ignores the flip side of national product: national income. If the government pays someone to fix a broken window or dig a trench, that person's welfare has been increased regardless of whether he has produced anything strictly welfare enhancing for someone else (I'll come back to the net effect in a second).
The second prong in the Higgsian critique is that government purchases do not take place at market prices.
Here I'll let Robert Fellner explain it:
To see why, we first have to understand why spending at all can be considered a measure of wealth. Upon reflecting on this matter, we realize it has to do with prices (specifically market prices) and the information that they convey. Whereas one finds his lot improved by the purchase of 5 wheelbarrows at the market price of 10 dollars a wheelbarrow, we can conclude the individual, and thus the economy as a whole, is wealthier to the tune of the utility that 50 dollars in spending has granted him, specifically the subjective value of the additional 5 wheelbarrows. Moreover the spending of 50 dollars represents the creation of these 5 wheelbarrows and of course the additional utility they grant, hence why they were purchased. This is the key. Wealth is not measured simply by the dollar amount of spending. Wealth comes from the goods and services provided in exchange for money. This is why spending matters and is accurate as an indicator of wealth. If any part of this process is diluted, the quality of spending as an accurate indicator of wealth diminishes greatly.
So why does consumer spending work well in this regard, where government spending fails? The answer lies in the prices. In a free market, all participants are subject to the profit and loss test. Namely, if one consistently spends more than he earns, he eventually becomes bankrupt and removed from the market altogether. In order to prevent this "death by free-market" one must learn to generate a profit; which is done by allocating resources efficiently. As all market participants engage with one another in this task, prices emerge for all the various goods and services within the economy that reflect their valuation to the economy as a whole (the price of course being derived both from the subjective valuation of the good as measured against the scarcity of the good, put more simply: supply versus demand).
This ignores three basic things. First, it ignores the work of Alfred Chandler (And J.K. Galbraith, btw) who has in great detail explained how large firms in certain industries vertically and horizontally integrate specifically to avoid having their production process (and prices) dictated by "the market" but rather prices are determined by negotiations between divisions of the firm. In similar fashion, the government does not simply dictate prices but rather negotiates them with firms (when they have some degree of monopsony power at least). Is negotiation really a less efficient means of price discovery than anonymous markets? Are anonymous markets really the dominant free market structure anyway? I'm less inclined to assume no to both of the questions than an Austrian is. Anyway, I'm not denying gold plating but I am suggesting that the price the government pays is not a priori less arbitrary than the price early adopters pay for an Iphone (to illustrate this from a different angle).
That may have been a clumsy segue but there is a more fundamental problem with the idea that "market prices" capture welfare better than government prices. Even assuming all C and I are allocated by perfectly competitive markets competitive prices only capture the exact utility produced by the good for the last (marginal) buyer. Everyone else in the market is accruing either a consumer surplus or a producer surplus. To capture that involves a whole series of arbitrary assumptions that are impossible to make and best avoided taking NIPA for what it is.
Again, though, if we look at income it may be easier to make welfare statements. If the government buys a loaf of bread from the baker it seems easier to make welfare statements (though I still advise against it even if I am as prone to do it as everyone else). The baker's welfare is unambiguously higher if the government (or anyone) pays $100 for his loaf of bread instead of $1. Unlike the troubles you run into with trying to make wealth/utility statements about who purchases what for who and why we can make a relatively uncontroversial statement that more income is better.
Okay, so now on to net effects. There are definitely certain conditions under which the broken window falacy is wrong. The increase in GDP from fixing a broken window from one of the many curiosities of NIPA accounting that, again, underscore why you shouldn't act like GDP literally means total wealth, utility or welfare. If I grow my own wheat and bake my own bread it doesn't count as GDP but if I buy it from the baker it does despite the fact that the same "utility" or wealth is created when I make my own bread (ignoring issues of scale efficencies***). Likewise, fixing a broken window shows up as GDP even though I as an individual would be better off if the market transaction never happened and I spend my broken window money on components of GDP I actually want to spend them on.
However, if the government borrows money to fix my window there are conditions (that aren't that ridiculous to meet and where the WWII example is instructive) where there can a net increase in GDP as income as welfare. The relevant variable here is the Debt to GDP ratio (not the absolute level of debt, so take your Ricardain equivalence somewhere else) which allows the broken widow theory to work because there is no full redistribution through taxes. The government can borrow today, give me the money to fix my window and then sit on the debt for as long as it needs to and let the burden of that debt shrink (under certain growth/interest rate conditions) away as nominal GDP growth increases, paying only the cost to service the debt as it shrinks. This spending, by the way, also increases nominal GDP grow at the point in time that I pay for my new window assuming no crowding out it will directly off set itself and there will be no increase in Debt/GDP. You can also tell a story where this could happen through the private sector provided as favorable borrowing terms as (or the ability to set interest rates like) the government. However, one should be reminded that the likes of Paul Krugman only advocate the building of FEMA death camps when the market has failed to produce full employment. In which case the government is also making savers better off because it is giving individuals something to hold that pays at least some kind of risk adjusted return in olight of a lack of investment. Individual svaes can be made better off even though on aggregate the whole point is to finance today with something that will be insignificant tomorrow.
Now, we can debate whether the relevant variables that determine the shrinking of the debt hold or that the inflationary bias is too much, or the size of the multiplier but the basic principle stands.
*For the uninitiated: GDP is the purchase of newly produced final goods and services. This is distinct from intermediate goods which are inputs into the production of final goods and services. The reason for this distinction is becuase final goods and services already contain all the costs of the inputs (and factors of production**) that went into making them. We use final goods and services as a measure of output becuase final goods and services contain the value of everything that goes into making them and so we get a full measure of the "output" of the economy.
**A "factor of production" is something that does not get "used up" in the production of a good. This is distinct from an "input" which does get used up. For instance when a baker bakes a loaf of bread the wheat he uses is an input but his labor (and the oven, his capital) do not get used up are factors of production. The price of a loaf of bread (a final good if I buy it) contains in it the cost of the wheat, the cost of the bakers labor, the interest he has to pay on the oven he bought with a loan and the profit the baker gets from owning the bakery (all payments to the factors of production).
.*** Presumably though scale inefficiencies are irreverent from a utility point of view since I must be on net gaining at least some utility from going through the trouble to bake my own bread (assuming I have the income to do it) instead of just buying it.