The Treasury

Global Navigation

Personal tools

2.3  New Zealand evidence

To date there have been only a few studies that look at terms of trade trends and volatility and their impact on economic growth in a New Zealand context. One such study was Grimes (2006). He found that approximately half the variance in annual GDP growth between 1960 and 2004 can be explained by the level and volatility of the terms of trade. He found that the level of the terms of trade had a positive effect on economic growth while the volatility in the terms of trade (particularly import prices) had a negative impact on economic growth. This paper extends the methodology of Grimes (2006) by using slightly different measures of real export and import prices. This is discussed below in Section 2.3.1.

Another paper looking at the importance of the terms of trade for economic growth in New Zealand was Fox, Kohli and Warren (2003). Using a modified Diewert-Morrison decomposition, they separate out the contributions to New Zealand’s GDP growth from total factor productivity (TFP) growth, labour and capital utilisation, the terms of trade, and the trade balance for the period of 1983 to 2001. They find that although the terms of trade did not make the largest contribution to economic growth on average over this period, there were times when movements in the terms of trade made significant contributions to economic growth. This paper extends the work of Fox et al up to 2005 in order to include the recent increase in the terms of trade from 2003. It is found that in 2004 and 2005, the terms of trade accounted for close to two-fifths of real GDP growth in those years. A more detailed discussion of this method and the results are displayed in Appendix 2.

Buckle, Kim, Kirkham, McLellan and Sharma (2002) is another study which looks at the impact of terms of trade movements on the New Zealand economy. Their research was based on the development of a structural vector-autoregressive model of the New Zealand economy to help decompose the contributions to New Zealand business cycles. Part of their analysis looked at the impact of international variables such as export and import prices. Rather than use a “portmanteau” terms of trade variable, they separate out the contributions from export and import prices as their impact on GDP fluctuations varies over the sample period. They found that shocks to export prices tend to have relatively long cycles while import price shocks are much more volatile. They conclude that these international variables (along with foreign interest rates, foreign output and foreign equity returns) have had a significant influence on New Zealand’s business cycle.

2.3.1  An empirical test

As discussed above, Grimes (2006) tested the impact of both the level and volatility of the terms of trade on New Zealand’s GDP growth. His approach begins with an equation of the form shown by (1) below

(1)     .

where . is the annual growth rate in GDP between year t-1and t, . is the logarithm of the terms of trade in year t, and. is the 10 year moving standard deviation of the terms of trade.

As the terms of trade is a ratio of export and import prices, Grimes (2006) suggests that it is appropriate to test whether export and import prices individually affect economic growth. Therefore (1) above is extended into (2) where the level of the terms of trade, . , and the volatility of the terms of trade, . , are separated into their real export and import price components,

(2)     .

where . is real export prices, . is real import prices, . is the 10 year moving standard deviation of . , and . is the 10 year moving standard deviation of . .

Unlike in Grimes (2006) however, which calculates real export and import price indices by deflating by the New Zealand consumer price index, this paper deflates the export and import price indices by the IMF’s Manufacturing Unit Value (MUF) series (which is common in this form of analysis).[4] The period tested is also 11 years longer, i.e. 1950 to 2005.[5] The results of this analysis are displayed in Table 1 below. Like in Grimes, the unrestricted regression (2) is initially estimated and then tested to see whether it can be restricted to the form of (1). This is done through the specification of four different regressions numbered Regression 1 to Regression 4.

Table 1 – Results from the Estimation of Equations (1) and (2)
Independent Variables Regressions - Dependent Variable: .
1 2 3 4 smoothed^
. 0.725***
      (0.225)
. -0.375*** -0.301** -0.389*** -0.497***
(0.144) (0.153) (0.148) (0.144)
. 0.113** 0.106** 0.123*** 0.072*** 0.029***
(0.050) (0.043) (0.030) (0.020) (0.007)
. -0.040 -0.029
(0.051) (0.043)      
.
         
. -0.022 -0.019
(0.024) (0.019)      
. -0.057** -0.096*** -0.114*** -0.064***
(0.029) (0.024)   (0.017) (0.016)
. -0.000**
    (0.000)    
Constant -0.052 -0.038 -0.119*** -0.035** -0.003
(0.050) (0.038) (0.037) (0.027) (0.012)
. 0.211 0.349 0.279 0.354 0.711

***, ** and * denote rejection of the null hypothesis at the 1%, 5% and 10% significance levels respectively. Newey-West standard errors are presented in parentheses.

^ This equation uses a three year centred moving average of GDP as the dependent variable as well as lagged independent variables.

Regression 1 corresponds with (2) and it is found that real export prices and real import price volatility are significant at the 5% significance level while real import prices and real export price volatility are insignificant. Regression 2 includes a two year lag of the dependent variable (which is significant at the 1% level) after finding that the residuals from Regression 1 showed some evidence of serial correlation. This increases the explanatory power in comparison to Regression 1 markedly. Regression 3 drops all the insignificant variables from Regression 2, namely real import prices and export price volatility, and import price volatility is replaced by total terms of trade volatility. As found in Grimes (2006), the explanatory power falls when the total terms of trade volatility is added and so import price volatility is used as the chosen volatility measure.

Using a Wald test, Grimes (2006) finds that the coefficients on . and . are equal in absolute value but of opposite sign. He therefore is able to replace these two variables with . . However, in the regressions tested below, this Wald test is rejected and therefore it is only the level of real export prices rather than the level of the terms of trade that enters into Regression 4. The actual and predicted annual GDP growth is displayed in Figure 3.

Although this paper uses different measures of real export and import prices (and hence volatility) to Grimes (2006), the conclusions are similar. Where they do differ is that rather than the total level of the terms of trade having a positive effect on economic growth, as it does in Grimes, it is only the level of real export prices that are found to have a significant impact on economic growth. Like in Grimes, this paper finds that the volatility in import prices has a highly significant negative effect on economic growth while export price volatility is found to be insignificant.

These are interesting conclusions. It is surprising that the level of real import prices did not have an impact on economic growth. One would expect higher import prices to have an adverse effect on economic growth through lower household consumption and lower investment, although if the higher prices were enough to encourage people to purchase domestic substitutes, then this may actually be positive for economic growth (particularly in the short-term). This is because imports would fall even though consumption stays at the same level or increases, leading to higher GDP. However, it may not be so positive in the longer-term as households’ (and firms’) purchasing power would have fallen. This could see a drop in economic welfare. It is not surprising however, that the volatility in import prices has a significant negative impact on economic growth. The majority of New Zealand’s capital investment goods are sourced from overseas. If a firm saw that the prices of imported capital goods were very volatile, it may deter them from making new investment decisions. This would have negative impacts on economic growth.

Figure 3 – Actual and Predicted GDP Growth
.
Source: Statistics New Zealand, Author’s calculation

As in Grimes (2006), Regression 4 is then finally tested using a three year centred moving average of GDP growth. The results are displayed in the far right column of Table 1 and as in Grimes, the amount of variation increases considerably (71% in this case).[6] This shows that once some of the variability in annual GDP growth is removed, the specification of Regression 4 is able to explain a considerable amount of the remaining variability. That is, annual GDP growth can be explained by lags of itself plus the level of real export prices and real import price volatility.

The next section looks in more detail at how the level and volatility of New Zealand’s export and import prices have changed over time.

Notes

  • [4]Other studies that use the MUV index as a deflator include Cashin, Liang & McDermott (2000), Cashin, McDermott & Scott (1999), Bleaney & Greenaway (1993) and Cuddington & Liang (1998).
  • [5]For a description of the data used in this section see Appendix 1.
  • [6]Also tested is the effect of regime changes on these results such as the 1985 float of the New Zealand dollar. This did not have a major effect on the results (as in Grimes (2006)).
Page top