3.3 Summary of the structural equations
The second stage of the model development is to specify the functional form of the structural equations for each variable. The functional form for each equation must be consistent with the block exogeneity restrictions and reflect information from prior theoretical research or from empirical single equation and VAR research.
Tables 2 and 3 summarise the structural equations and lag structure. The rows show the dependent variable in each equation. The columns show which of the variables appear as explanatory variables in each equation. The shaded cells indicate these. Table 2 shows the contemporaneous relationships. Table 3 shows the lagged variables that appear in the equations for each dependent variable.
The unshaded cells in Table 2 represent the contemporaneous restrictions in the matrix B0. There are 119 unshaded cells or contemporaneous restrictions. The number required to just identify the model is 78.
3.4 The International Economy Block
The set of international variables included in the first New Zealand VAR model developed by Wells and Evans (1985) was limited to the exchange rate and international trade prices. This choice was based on the reasonable argument that under the pervasive regulatory environment that applied during their sample period to private international capital transactions, domestic financial markets and exchange rates, the main channels through which international activity was likely to impact on the domestic economy were the export and import price channels. Furthermore, they observed at that point in time econometric results from structural models had not revealed any significant effect of interest rate differentials on private international capital movements (Wells and Evans, page 424).
International linkages are likely to have become more complex and more pervasive as a result of deregulation of New Zealand trade and financial markets since 1984. Financial deregulation is likely to have created direct financial linkages, increasing the sensitivity of domestic financial markets to international financial conditions. Export product and market diversification suggest that New Zealand tradeable goods production is likely to be more responsive to international demand conditions. Evidence consistent with this argument can be found in Conway (1998). Using a structural VAR model that included US GDP and US interest rates, Conway found the proportion of the variance of New Zealand real GDP attributable to US GDP and US interest rates increased when the sample period was truncated to exclude the period prior to 1985:1.
Theory and subsequent empirical results provide strong justification for expanding the set of international variables to allow for more direct linkages with New Zealand domestic variables in addition to the international trade price links captured by Wells and Evans. Our model therefore includes foreign real output, foreign interest rates and foreign real asset returns. The international block is a Wold recursive system in the contemporaneous variables. The contemporaneous causal ordering runs from foreign real output to foreign interest rates and real asset returns. Foreign real output (
) depends upon lags of foreign nominal interest rates (
) and foreign real asset returns (
).
Foreign nominal interest rates respond to contemporaneous movements in foreign real output and to lags in foreign real output and foreign interest rates. This aspect of the model is therefore similar in structure to the corresponding component of Dungey and Pagan’s (2000) model for Australia. As they point out, if the typical interest rate decision rule for central banks is approximated by a Taylor rule, then the equation for the foreign interest rate is likely to be mis-specified since it implies that central banks only vary interest rates in response to an output gap. Although the full model does include international prices, they are only representative of the prices of goods exported and imported by New Zealand and are unlikely to be sufficiently representative of the prices of concern to the world’s major Central banks.
Foreign real asset returns () respond contemporaneously to foreign real output and foreign interest rates, and to lags in these variables and its own lags. The inclusion of reflects growing integration of New Zealand financial markets with international markets since financial deregulation during the mid 1980s implied by Conway’s (1998) results. Several single equation investigations (including Dyer, 1994; Grimes, 1994; Rae, 1995; and Eckhold, 1998) justify attempting to capture more directly the significance of international financial market conditions on New Zealand interest rates and financial asset returns.
3.5 The International Trading Prices Block
Several classes of models have been used to evaluate external price effects in New Zealand (see Wells and Easton, 1986). These have included for example the classical homogeneous product model (applied by Bailey, Hall and Phillips, 1980), and the Scandinavian model cast in terms of traded and non-traded goods (applied by Ursprung, 1984). Some models have attempted to capture these international price effects by using the portmanteau terms of trade variable (Kim, 1994; Hansen and Hutchison, 1997; and Conway, 1998) or international oil prices to represent common global shocks (Selover and Round, 1995) or to act as a proxy for foreign demand conditions (Claus, 2002).
Others have argued that the Australian three-goods model of the dependent economy (developed by Salter, 1959 and Swan, 1963), which emphasises the distinction between imports, exports and non-traded goods, and has exogenous terms of trade, is a more appropriate basis for modelling the impact of international traded goods prices on the New Zealand economy. New Zealand models structured in this way include Wells and Evans (1985), Buckle and Pope (1985) and Szeto (2002). The basis for adopting this three-good approach is that New Zealand’s industrial structure falls between the typical characteristics of developed and developing economies. Consequently, a substantial proportion of New Zealand’s imported goods are inputs to the production process and a significant proportion of its exported goods are primary based products with a low proportion of exportable production absorbed by home consumption.
Within the context of this type of commodity structure, Findlay and Rodriguez (1977), Bruno and Sachs (1985) and Buckle and Pope (1985) demonstrated that rising import prices, including oil prices, have a stagflation effect on the domestic economy. In the contemporary environment of a floating exchange rate and inflation targeting by the New Zealand Central bank, the impact on domestic inflation would be smaller but the short-run effect on output could be accentuated. Because export commodities make up only a small proportion of final consumption, rising export prices will increase real incomes and have a small impact on domestic inflation.
The effects of import and export price shocks on domestic real output and inflation therefore need not necessarily be mirror images of each other. This argument is supported by the results from Wells and Evans (1985). They found an increase in export prices raised private sector output and employment and had virtually no effect on output prices. An increase in import prices reduced private sector output and employment but also raised output prices, the classical stagflation effect. Their results were estimated using data from 1961:1 to 1981:1.
Changes to the New Zealand policy and trading environment and changes in industrial structure since the mid 1980s are unlikely to have diminished the merits of the ‘Australian model’ and the relevance of Wells and Evans’ results. New Zealand’s industrial structure during the 1980s and 1990s still warrants the separate identification of import and export price shocks. Furthermore, recent international research has demonstrated the utility of differentiating between commodities in the context of understanding business cycles in small open economies (see for example Kose, 2002).
Consistent with the small open economy assumption, we model the foreign currency price of New Zealand exports (
) and the foreign currency price of New Zealand imports (
) as responding to contemporaneous and lagged foreign real output and to their own lags.
