Output Volatility: GDP v. GNP

While macroeconomists spend a lot of effort trying to understand the volatility of output, we spend less time worrying about whether output should be measured by real GDP or real GNP. As we discussed here, the difference between GNP and GDP (known as net income from abroad) is quite modest for the U.S. If we look at quarterly data from 1970 to today, average real GDP growth has been 0.769% per quarter (annualized, this amounts to less than 3.1% per year), while average real GNP growth has been 0.767%. In other words, not much difference since net income from abroad was only 0.7% at the beginning of this period with the ratio being half of that for the end of the period. The standard deviation of real GNP growth was 0.852% over this period as compared to 0.847% for real GDP growth. In fact, the only significant differences between the two series seems to be the last quarter of 2001 when real GNP growth exceeded real GDP growth by 0.6% and the first quarter of 2002 when real GDP growth exceeded real GNP growth by 0.5%.

As I was preparing this post on Ireland’s economic growth, I recalled a paper from Colm McCarthy entitled Volatility in Irish Quarterly Macroeconomic Data, which notes:

The quarter-to-quarter volatility in real macroeconomic aggregates, including gross output (GDP) and gross income (GNP), is very pronounced for the Irish data, more so than for other OECD countries.

The principal source of volatility in Irish real GDP lies in the recorded figures for industrial output.

Within the industrial output category, which is the aggregate of manufacturing and construction, the excess volatility can be traced to a small number of manufacturing sub-sectors.

These sub-sectors are known to be dominated by exporting multinationals, whose shares in recorded output greatly exceed their shares in employment or in the generation of domestic demand. Recent patterns of growth in these manufacturing sub-sectors show very sharp rises in output, exports and Gross Value Added (GVA), unaccompanied by commensurate movements in employment or payroll.

As I read his introduction, I was wondering if much of the GDP volatility might be due to variations in the degree of transfer pricing manipulation. McCarthy’s conclusion certainly raises the issue of transfer pricing:

The Irish economy has a high GDP/GNP ratio as well as large short-run variations in value-added components other than wages. This paper finds that the Irish quarterly macro data are being seriously distorted by the unusual structure of Irish Manufacturing. This relates to the, no doubt perfectly legitimate, activities of some multinational sectors. These sectors have output valuations dominated by profits rather than wages, and fluctuations in output composition or in transfer pricing or both are clearly at the source of the quarter-to-quarter volatility in overall activity as measured. Care is warranted in interpreting the seasonally adjusted data anyway, since the seasonal factors are based on a period of unprecedented change in the Irish economy. The ESA-95 national accounting rules do not cope easily with a large multinational sector and transfer pricing, and we suggest that the CSO might consider some method of smoothing the Net Factor Payments numbers, and perhaps some other components, on a quarterly basis.

Figure 1 of his paper shows how unusually high Ireland’s GDP/GNP ratio is, while figure 2 points out that this ratio has increased over the last couple of generations. Figure 2, however, only points out that Ireland’s real GNP growth has been less than its real GDP growth. McCarthy also notes that Ireland’s real GNP growth is almost as volatile as its real GDP growth.