There is no fact of the matter as to what NGDP is, but that’s probably OK
So, I’m feeling fairly sanguine about my conceptual issues with NGDP. So long as the capital intensity of the economy does not change unduly over a short space of time, the fact that we treat capital goods differently to intermediate goods doesn’t bother me too much for NGDP-targeting purposes. I’d say that NDP is certainly the correct measure of welfare, and if you try and think about how you’d give GDP some kind of micro-foundation in utility theory, I think you’d see why that is pretty quickly. I remain very open to views as to what the best nominal expenditure figure for central banking purposes is, but I think NGDP is probably fine even if you get some issues with accounting decisions as to what to count as ‘capital’ in borderline cases. And I definitely think that we don’t want to be getting into the question of how (or whether) we should be depreciating capital when it is idle during a recession, and what impact that has on Nominal NDP (for example, straight line depreciation would make recessions look worse relative to, say, the machine hours method).
Googling last night on the topic of calculating GDP in practice, I stumbled across a paper hosted on the BEA website, entitled ‘Taking the Pulse of the Economy: Measuring GDP‘ (Landefield, Seskin & Fraumeni) originally published in the Journal of Economic Perspectives. It’s actually quite interesting. Let me begin by quoting the paper:
In the United States, the GDP and the national accounts estimates are fundamentally based on detailed economic census data and other information that is available only once every ﬁve years. The challenge lies in developing a framework and methods that take these economic census data and combine them using a mosaic of monthly, quarterly, and annual economic indicators to produce quarterly and annual GDP estimates. For example, one problem is that the other economic indicators that are used to extrapolate GDP in between the ﬁve-year economic census data—such as retail sales, housing starts, and manufacturers shipments of capital goods—are often collected for purposes other than estimating GDP and may embody deﬁnitions that differ from those used in the national accounts.
Yikes! There’s a really good summary on p200-201 of all the various metrics that are used to calculate the various components of GDP. It’s safe to say that the list initially provided me with little comfort. However, it does appear that revisions at the time of the census in the past have been low
For the last ﬁve benchmark revisions of GDP, which correspond to the census years 1982, 1987, 1992, 1997, and 2002, the nominal level of GDP was revised an average of 1.1 percent, and the growth rate between benchmark years was revised an average of 0.26 percentage point. The corresponding mean absolute revisions to the nominal level of GDP and the growth rate were similar in magnitude because most of the revisions were upward.
This does give me some comfort, subject to the key question that the benchmarking exercise is accurate. And I think it probably is – the assumptions you’d need to make for ‘fitting’ the data in order to avoid a large one-time revision would likely, over time, get fishier and fishier. Eventually some government statistician would work out they could make a name for themselves by whistleblowing on the whole thing.
So, whilst I have one more farewell post in the works featuring GDP accounting, I will be looking at ‘G’ rather than ‘I’. What I would say is that it has been surprisingly difficult for me to get a grip on what GDP ‘means’. Market Monetarists are fond of saying that everyone else starts their analysis with real GDP, whereas they (‘we’, maybe? I think I count as one) start with goods/services and the money that is exchanged for them. This is what allows them (us) to diagnose a recession as monetary disequilibrium. But I guess this exercise has led me one step further back even from that – I see the building blocks as money and transactions, some of which count towards NGDP and some which don’t, and that it is surprisingly tricky to understand how we build up those transactions into more meaningful economic indicators.