5 Measured New Zealand growth rates over time
One point that should be noted when considering New Zealand’s average growth rate over a particular sub-period is that the growth rate can be quite sensitive to the endpoints (or time period) chosen. Table 9 shows New Zealand’s growth rate based on OECD data for a number of different length sub-periods. To illustrate the sensitivity of the average growth rate for a period, consider the average growth rate for the period 1988 to 1994 (a 6 year window with window endpoint 1994). The average growth rate for this sub-period is 0.63 percent. Compare this to the growth rate for the period 1987 to 1993. The average growth rate for this period was -0.10 percent per annum. These growth rates differ significantly yet the start and end of the six year period under consideration differ by only one year.
|Window endpoint||Window length|
Growth Rates calculated using average annual change method.
Example: the growth rates with window endpoint 1999 and window length 8 is calculated using real GDP data for the years 1991 through to 1999.
Bold figures show the highest average growth rate for each window length. For example if one focuses on growth rates for sub periods that are 7 years long, the highest growth rate for any period of this length was 2.5% and this relates to the period 1992 to 1999.
Alternatively, this same point can be illustrated when the endpoint is fixed and the length of the sub-period differs by a single year. For example the average growth rate for the period 1991 to 1997 was 2.30 percent. Extending this period back just one year results in a growth rate for the period 1990 to 1997 of 1.32 percent. This is nearly a whole percentage point lower. These differences are a result of the variability of the annual growth rates. Due to this variability it is often desirable to measure trend growth, which loosely put implies measuring growth rates between two years that are similarly placed during the growth cycle, for example, peak to peak. The objective of this paper is, however, to document New Zealand’s historical growth performance over time and not to determine New Zealand’s trend (or potential) growth rate.
Table 10 gives the ranking of the New Zealand growth rate for each cell in Table 9 within the 26 OECD countries included in the OECD dataset used for this paper. For each possible sub-period shown in the table, the average growth rates of the other 25 OECD countries have been calculated and New Zealand’s ranking within these growth rates computed. As Table 10 shows New Zealand’s growth rate for most sub-periods has been towards the bottom of the OECD (lowest possible ranking is 26). Periods where performance has been in the top half are rare and not sustained for long periods of time.
|Window endpoint||Window length|
Possible rankings range from 1 (highest growth rate for the period in the OECD) to 26 (lowest growth rate for the period in the OECD). Bold cells show New Zealand’s highest ranking in each column.
* indicates that the growth rate ranking is sufficiently high to be categorised as being in the top half of the OECD.
Bearing in mind the sensitivity of NZ growth rates to the sample period, the results shown in Tables 9 and 10 may highlight signs of improved performance by the NZ economy over the last decade. In Table 9, the decade with the highest growth rate out of any decade long period in the table was the most recent decade ending in 2000. However, while this period also resulted in New Zealand’s highest growth ranking out of all decade long periods, the rate of growth achieved was still insufficient to register New Zealand in the top half of OECD growth rates.
It should be noted that exactly the same growth rates are obtained when using OECD data from publications such as National Accounts of OECD Countries (OECD, 2002) regardless of whether real GDP per capita is converted into a common currency using exchange rates or PPPs. This is because the OECD converts all the observations in a country’s real GDP per capita series (expressed in the country’s national currency) using the exchange rate or PPP rate for a single year. A transformation that involves either multiplying or dividing all observations in a series by some constant has no impact on the growth rate of the transformed series.
Tables equivalent to Tables 9 and 10 based on the PWT, Maddison and Haugh’s calibrated real GDP data sources are provided in Appendix D. Nuxoll (1994) raises a concern that data construction techniques used in constructing series such as those contained in the PWTs may have inadvertently introduced a spurious correlation between growth rates and income. Nuxoll argues that (based on what he calls the Gerschenkron proposition) any income index using fixed prices to measure growth rates would tend to understate the growth rates for less developed countries and overstate the growth rates for more developed countries relative to the national income accounts.
The PWT draw heavily on the work of the International Comparison Project (ICP). The ICP estimates real expenditure in a large number of countries based on what are termed “international prices”. International prices are constructed using the Geary-Khamis formula for international prices. This results in the international price of a good depending little on the prices in low-income countries, countries with small populations or low or relatively small demand for the good.
The ICP only produces expenditure estimates of real GDP for a few years and consequently to construct the annual series that appear in the PWT, Summers and Heston extrapolate estimates for real consumption, investment, government spending and net foreign balance for a large number of years. These estimates are based on international prices. “The estimates for real consumption, investment, government spending, and net foreign balance were combined with the growth rates for the same series in existing World Bank national-accounts data. This amounts to assuming that these series measured in terms of international prices grow at the same rate as these series measured in domestic prices. The result is a series of estimates for each year, all measured in terms of international dollars.” (Nuxoll, 1994)
Consequently, real total GDP measures from the PWT and national accounts estimates differ because of the price weights used. The PWT use international prices whereas national accounts uses domestic prices. If this results in the share in GDP of consumption, investment, government or net foreign balance differing between the PWT and the national accounts, the growth rates obtained from the different sources will differ.
Nuxoll (1994) notes that international prices are a synthetic set of average prices across countries, so they are not drawn directly from one country. He also states that prices in Hungary are the closest to the international prices used in the ICP and PWTs. Nuxoll’s research ultimately finds that “Current versions of the Penn World Table do not systematically distort the data, because of the very high level of aggregation. Nonetheless, the growth rates in Penn World Tables do differ from national accounts.” (Nuxoll, 1994). Nuxoll goes on to argue that the use of real GDP series measured in domestic prices is more reliable than using series expressed in international prices, because domestic prices characterise the trade-offs faced by people in the country. An awareness of the sorts of problems associated with the use of different price weights is, however, still desirable for empirical work.
- For example, the real GDP series for New Zealand expressed in 1995 prices and exchange rates (US dollars) is obtained by converting each value in a real GDP per capita series expressed in 1995 prices and valued in New Zealand by dividing by the 1995 exchange rate with the US dollar. Likewise, the real GDP series for New Zealand expressed in 1995 prices and PPPs (US dollars) is obtained by converting each value in a real GDP per capita series expressed in 1995 prices and valued in New Zealand by dividing by the 1995 PPP with the US dollar. Note, OECD publications express the exchange rate for the New Zealand and US dollars in terms of the number of New Zealand dollars a US dollar will buy. In New Zealand exchange rates tend to be expressed terms of the number of units of a foreign currency one New Zealand dollar will buy.
- For more details see Geary(1958) and Khamis (1967).