'Large' Irreducible Pure Forecast Errors for Commodity Outputs
Table 33a: Key results for AccStrucSMD
1. Why did the model erroneously give poor prospects to CutStone ?
CutStone had a USAGE error of 24% versus the smaller trend forecast error of 3%; hence it was situated well above the 45 degree line. The key results for this commodity are shown in Table 45. The actual outcome for CutStone output (x0dom) was a 12.2% rise over the 1998‐2005 period. This followed 13.9% growth from 1992‐1998. The extrapolated trend was therefore a further 16% output expansion versus the USAGE forecast of a 14.3% contraction. Table 46 shows the main users, cost structure and other information of interest of the 1998 database used in the forecast. The following observations can be made:
Producers purchased 60% of domestic output (Section 3 of Table 46).
Households purchased 37% of domestic output (Section 3 of Table 46).
Intermediate demand was driven by building and construction industries (Section 4a of Table 46).
There was 33% import penetration (Section 5 of Table 46).
Labour was the main factor input cost (Section 6b of Table 46).
From 1992 to 1998 growth in domestic demand for CutStone was driven by households. The total rise in household demand for the commodity was 50.6% (see Table 45). Household demand for domestically produced CutStone (x3cs) increased 44.8% (this is not shown the results table). Intermediate input demand for the domestically produced commodity (x1csi) on the whole was relatively flat during this period. USAGE calculated modest rises in the taste and preference indicators for households and producers (a3com and cont_ac), which were projected forward. In the case of households, USAGE predicted total demand (undifferentiated by source) to grow by 25.1% over the seven year forecast horizon. However, household demand for domestically produced CutStone was expected to increase by a slower 14.5%. Where producer demand is concerned, Table 47 shows that USAGE underestimated the growth of the four largest intermediate purchasers of CutStone. (The USAGE error was far greater than the trend error for the two largest of these.) Given that production demand was the larger share of output, the model forecast an overall 14.3% reduction in x0dom for CutStone.
BAS1 Above/(Below)
Proportion Actual Forecast 45o Line
Other new construction 43 OthrConstruc 42% 10% ‐1% 16
New office, industrial and commercial buildings construction 42 IndComBuild 28% 2% ‐17% 15
New residential 1 unit structures, nonfarm 33 Nresident1 11% 40% 13% (21)
New residential garden and high‐rise apartments construction 36 GardHighrise 5% 39% 13% (2)
x0dom 1998‐2005
SIC Name USAGE
Commodity
Table 47: Key results for the main purchasing industries of CutStone
The actual result was 12.2% output growth, which was largely driven by a strong increase in tastes and preferences for CutStone by producers and households. Recall that the combined change in household tastes (a3com) was projected forward to be 6.0%. This essentially means that at any given set of prices and per capita income, consumption per household of CutStone would be 6.0% higher in 2005 than in 1998.64 In reality, household tastes towards CutStone soared by 48.9%; driving an 88.5% increase in consumption demand for the domestically produced commodity. A similar story emerges for the contribution to output of CutStone‐using technical and taste change (ac). This was projected to be 3.2% higher in the forecast, when in fact the increase was 22.5%. These were clearly the key drivers behind the error.
It is also worth noting the significant difference in technological change parameters. On the supply side, primary factors comprised 50% of total input costs (Section 6a of Table 46). In the forecast, all primary factor augmenting technical change (a1prim) indicated a 27.7% improvement in primary factor efficiency. This meant that the CutStone industry was projected to require 27.7% less primary factors to produce the same level of output whilst holding all other inputs constant. The contribution of all primary factor augmenting technical change to total input costs (cont_a1prim) was estimated to be an overall cost reduction of 14.9%. In reality, this efficiency measure deteriorated by 33.4%, and its contribution to total input costs rose 15.5%.
64 More precisely, the consumption per household of CutStone in 2005 would be 6*(1 – share of CutStone in
2. Macro perspective
Although stone remains an important building material, new construction materials and methods developed during the twentieth century have limited its use almost entirely to a finishing element of mostly decorative value.65 In addition, any forward looking comments made around the 1990s were reasonably cautious:
“The long‐term industry outlook was generally lackluster for the early 2000s. Limited opportunities for further productivity gains, coupled with greater foreign competition, were expected to hurt many industry sectors. Although traditional domestic markets, such as construction, experienced expansive growth in the booming economy of the late 1990s, superior synthetic substitutes continued to make gains. Due to the strength of the construction industry in the late 1990s, the cut stone industry did experience steady growth between 1997 and 2000, when the value of shipments increased from $1.24 billion to $1.63 billion.
Because of stone's weight‐to‐value ratio, moreover, opportunities for U.S. export growth are slim with the exception of niche specialty stones. U.S. producers exported about 2 percent of production in the late 1990s. A bright spot on the horizon for the industry is the expected continued surge in historical restoration projects that require considerable amounts of stone to replace damaged pieces from the original construction.”66
3. Conclusion
The modeller may have viewed this overall cautious outlook as being consistent with the downbeat USAGE forecast for the commodity. Moreover, the building and construction boom that occurred mostly during the second half of the forecast period played a key role in the forecast error. Excessive borrowing across many sectors was fuelled by exceptionally low interest rates post the events of “September 11”; lax lending standards; piecemeal regulation; and financial product innovation. The extent and longevity of this boom did not seem to have been expected by industry experts. However, a track record of overly accommodative monetary policy from the mid‐1990s and steady industry growth in 1997 and 1998 may have provided some clues that the general outlook was overly guarded. On balance, it is difficult to say, conclusively, that the modeller could have produced a better forecast for CutStone. Perhaps, if negative growth was seen to be too pessimistic, a zero growth forecast – at most – might have been worked into the model.
65 http://www.answers.com/topic/cut‐stone‐and‐stone‐products, visited 11 September 2009. 66 http://www.answers.com/topic/cut‐stone‐and‐stone‐products, visited 11 September 2009.