Abstract
This paper assesses the impact of a disruption to capital inflows by examining past episodes of capital inflows in New Zealand and other countries. It also reviews the IMF’s Global Economy Model (GEM), which is used to provide some estimates of the equilibrium relationship between New Zealand’s real effective exchange rate and real commodity prices. The analysis also suggests that permanent changes in non-energy commodity prices can have a significant impact on New Zealand’s equilibrium exchange rate.
III. The Declining Importance of Tradable Goods Manufacturing: How Much Can Growth Theory Explain?15
A. Introduction
50. New Zealand, like most industrial countries, has been experiencing a decline in the relative importance of tradable goods production. Based on a 3-year moving average, goods production as a share of New Zealand GDP declined by 3¼ percentage points between 1995 and 2004, from roughly 28¼ percent to 25 percent.16 Although this decline is toward the high side when compared to some other industrial countries (bottom panel in Figure III.1), the decline in the non-commodity based manufactured goods is quite similar to many other countries. The difference in New Zealand is largely accounted for by a slightly larger decline in the share of commodities, which was exceeded only by Norway, Sweden, and Iceland.17
Share of Tradable Goods Production in GDP
Citation: IMF Staff Country Reports 2008, 164; 10.5089/9781451973877.002.A003
Source: National Accounts data.Share of Tradable Goods Production in GDP
Citation: IMF Staff Country Reports 2008, 164; 10.5089/9781451973877.002.A003
Source: National Accounts data.Share of Tradable Goods Production in GDP
Citation: IMF Staff Country Reports 2008, 164; 10.5089/9781451973877.002.A003
Source: National Accounts data.51. As laid out in Baumol (1967), the declining relative importance of goods production can arise from more rapid technological progress in the goods sector.18 Notionally, there is more scope for adopting labor-saving technological advancements in goods production. Despite faster productivity growth in the goods sector, labor mobility between the goods and services sectors enforces nominal wage equality. The required lower real producer wage in the service sector (because of lower productivity) is achieved with a rising relative price of services and, therefore, an increase in the share of services in output. Globalization is the open economy extension of Baumol, further driving down the relative price of tradable manufactured goods. In this note, simulations with the IMF’s Global Economy Model (GEM) are used to estimate how much of the historical decline in goods production in New Zealand can be accounted for by differing rates of productivity growth between the goods and services sectors in New Zealand and its major trading partners.
52. The simulation analysis suggest that faster tradable sector productivity growth in New Zealand and its major trading partners accounts for a large portion of the relative decline in tradable goods production. Simulating GEM over a ten-year period incorporating the productivity gap between the tradable and nontradable sectors seen between 1995 and 2004 results in a decline in tradables production in New Zealand of close to 3 percent, just under the 3.3 percent seen historically. Roughly 2/3 of the decline is accounted for by manufactured goods with the remainder accounted for by commodities.
B. GEM and the Simulation Experiment
53. GEM is a multi-region, multiple-good model of the world economy that is derived completely from optimizing foundations.19 The version of the model used here has been configured with three types of goods: a nontradable good; a tradable non-commodity good (manufactures); and a tradable commodity good (commodities). Because the model is derived from optimizing foundations, changes in the relative importance of each good will be driven by technology and tastes. The model has been calibrated to represent five regions: New Zealand; Australia; the United States; Emerging Asia; and the rest of the world.
54. Over the 1995 to 2004 period, productivity growth has been faster in the tradables sector than in nontradables, and fastest in Emerging Asia. The annual growth in labor productivity in the tradables and nontradables sectors for select countries and regions are in graphed in Figure III.2 and summarized in Table III.1.20 In all countries/regions, tradable sector productivity has exceeded that in the non-traded sector by roughly 1½ to 4 percentage points. The smallest gap being in the United States and the largest in emerging Asia, with New Zealand and Australia having gaps close to that in the United States.
Annual Labor Productivity Growth
Citation: IMF Staff Country Reports 2008, 164; 10.5089/9781451973877.002.A003
Source: Fund staff calculations.Annual Labor Productivity Growth
Citation: IMF Staff Country Reports 2008, 164; 10.5089/9781451973877.002.A003
Source: Fund staff calculations.Annual Labor Productivity Growth
Citation: IMF Staff Country Reports 2008, 164; 10.5089/9781451973877.002.A003
Source: Fund staff calculations.Average Annual Labor Productivity Growth 1995 to 2004
The Euro Area, the United Kingdom, and Japan are used to proxy the rest of the world.
Average Annual Labor Productivity Growth 1995 to 2004
| New Zealand | Australia | Emerging Asia | United States | Rest of World1 | |
|---|---|---|---|---|---|
| Tradables | 2.6 | 2.9 | 6.5 | 3.5 | 3.2 |
| Nontradables | 0.7 | 1.5 | 2.6 | 2.0 | 0.5 |
| Productivity Gap | 1.90 | 1.4 | 3.9 | 1.5 | 2.7 |
The Euro Area, the United Kingdom, and Japan are used to proxy the rest of the world.
Average Annual Labor Productivity Growth 1995 to 2004
| New Zealand | Australia | Emerging Asia | United States | Rest of World1 | |
|---|---|---|---|---|---|
| Tradables | 2.6 | 2.9 | 6.5 | 3.5 | 3.2 |
| Nontradables | 0.7 | 1.5 | 2.6 | 2.0 | 0.5 |
| Productivity Gap | 1.90 | 1.4 | 3.9 | 1.5 | 2.7 |
The Euro Area, the United Kingdom, and Japan are used to proxy the rest of the world.
55. The simulation experiment is set up to replicate the trend productivity gap between the tradable and nontradable sectors in New Zealand and its major trading partners over the 1995 to 2004 period. It is assumed that the productivity growth gap exists equally in the commodities and manufactures tradable sectors. Ideally the shock would be implemented by matching the actual historical gaps between non-tradables and each of the two tradable goods in the model. However, the available data does not split the tradable goods sector into these two components.21
C. Simulation Results
56. The declines in the shares of tradable goods resulting from the tradable sector productivity gap simulation broadly match the data. The simulation results for New Zealand, Australia and the United States are presented in Table III.2. In terms of the total decline in the share of tradable goods in GDP, the simulation results match the data surprisingly well for both New Zealand and Australia. In the cases of New Zealand and Australia, the simulated declines are just slightly below what occurred, while for the United States, the simulated decline is only about 75 percent of the decline in the data. This could reflect the fact that the region configuration used here does not adequately capture the trading relationships of most importance for the United States, for example that with Canada.
Change Over Ten Years in Share of GDP (In Percent)
Due to data limitations, the calculations for the United States are not based 3-year moving average but rather the decline between 1998 and 2006.
Change Over Ten Years in Share of GDP (In Percent)
| New Zealand | Australia | United States1 | ||||
|---|---|---|---|---|---|---|
| Simulated | Actual | Simulated | Actual | Simulated | Actual | |
| Manufactures | −1.9 | −1.9 | −1.3 | −1.7 | −2.0 | −2.5 |
| Commodities | −1.0 | −1.4 | −0.9 | −1.0 | −0.1 | −0.4 |
| Total | −2.9 | −3.3 | −2.2 | −2.7 | −2.1 | −2.9 |
Due to data limitations, the calculations for the United States are not based 3-year moving average but rather the decline between 1998 and 2006.
Change Over Ten Years in Share of GDP (In Percent)
| New Zealand | Australia | United States1 | ||||
|---|---|---|---|---|---|---|
| Simulated | Actual | Simulated | Actual | Simulated | Actual | |
| Manufactures | −1.9 | −1.9 | −1.3 | −1.7 | −2.0 | −2.5 |
| Commodities | −1.0 | −1.4 | −0.9 | −1.0 | −0.1 | −0.4 |
| Total | −2.9 | −3.3 | −2.2 | −2.7 | −2.1 | −2.9 |
Due to data limitations, the calculations for the United States are not based 3-year moving average but rather the decline between 1998 and 2006.
57. Looking at the 2004 to 2007 period, unbalanced growth can explain only about half of the estimated decline in the share of tradable goods production in New Zealand GDP. Using real volumes and computing nominal shares with slightly imperfect deflators suggests that tradable goods production relative to GDP in New Zealand declined by 1.9 percentage points over the 2004 to 2007 period. Commodities production declined by roughly 0.8 percentage points, with manufactures declining by 1.1 percentage points. Extending the simulation for an additional three years and assuming that the relative gaps in productivity growth were identical to those over the 1995 to 2004 period results in a simulated decline in tradable goods production as a share of GDP of 0.9 percentage points. Roughly 1/3 of this occurs in commodities with the remaining 2/3 coming in manufactures. Because the time period is quite short and the data is estimated, this difference should be interpreted cautiously. However, this result does suggest that the experience in the last three years is beyond what unbalanced growth can explain. Time will tell if this is simply short-term volatility, potentially reflecting the impact of competitive pressures on margins, or a more permanent decline resulting from the temporarily elevated exchange rate.
D. Conclusions
58. GEM simulations suggest that unbalanced growth can explain much of the relative decline in the share of tradable goods in GDP in New Zealand, Australia and the United States over the 1995 to 2004 period. Simulations incorporating the historical gap between productivity growth in the tradables and non-tradable sectors lead to declines in the relative share of tradable goods broadly matching the declines contained in the data. However, there are some differences that arise in the decomposition of the declines into their commodities and manufactures components. These differences should not be surprising as they likely reflect the fact that other macroeconomic shocks, not considered in this analysis, have undoubtedly had important effects over the sample period examined.
References
Baumol, W., 1967, “Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis.” American Economic Review, June, pp. 415 -426.
Hunt, B., and A. Rebucci, 2005, “The U.S. Dollar and Trade Deficit: What Accounts for the Late 1990s?” International Finance Vol. 8, No. 3, pp. 399 -434.
Hunt, B., 2007, “U.K. Inflation and Relative Prices Over the Last Decade: How Important was Globalization?” IMF Working Paper, WP/07/208.
Laxton D., and P. Pesenti, 2003, “Monetary Policy Rules for Small, Open, Emerging Economies.” Journal of Monetary Economics, 50, pp. 1109 -1146.
Prepared by Ben Hunt, (Ext. 3-6361).
Given the year-to-year volatility in GDP shares, 3-year moving averages present a more reliable picture of the trend change. For Canada and the United States, data limitation prevented using a moving average. Further, for the United States the change is between 1998 and 2006. For Canada, Norway, the U.S., and the U.K., oil production is removed from commodities because the large increase in oil prices distorts the picture.
Commodities were defined to be agriculture, fishing, forestry, mining and the production of food and beverages. It was not possible to include food and beverage production in commodities for South Africa, the United Kingdom, and the United Sates.
Other theories advance to explain the relative decline in the importance of tradable goods production are specialization (which leads to outsourcing and thus reclassification of activities previously performed in-house by manufactures) and changing consumer preferences as income rises.
For a detailed descriptions of GEM’s structure and dynamic adjustment properties see Laxton and Pesenti (2003) and Hunt and Rebucci (2005).
The computations are derived using the database maintained in the IMF’s Research Department to support the assessment of real effective exchange rates.
The shock is implemented assuming that people must learn about the persistence in productivity growth. When agents have perfect foresight under long-lived shocks that have significant implications for wealth, rational expectations models, like GEM, can produce adjustment dynamics unlike that seen in actual data. To address this and generate closer-to-real-world adjustment dynamics, the shock is implemented assuming that each period, agents must generate forecasts of the persistence in productivity growth. Here the learning is calibrated so that agents initially learn slowly about the persistence. However, as the duration of the shock increases, agents start to learn more quickly. See Hunt (2007) for a description of the uncertainty framework and an illustration of the speed of learning.
