Copyright 2009 Pearson Prentice Hall. All rights reserved.JPMorgan Chases Foreign Exchange Forecasting AccuracyA case study in evaluating an individual institutionsforeign exchange forecasting accuracyVeselina (Vesi) Dinova Had been asked by her director at Teknekron (US) to review the exchange rate forecasting accuracy of her companys primary financial services provider, JPMorgan Chase (JPMC). Vesi had focused on the three primary currencies the companys operations revolved around , the dollar, the euro, and the yenTeknekron had relied upon JPMC for most of its currency advisory services for years, and the forecasts provided by JPMC were regularly used in sales and sourcing decisions including pricingThe rise of the euro against the dollar in recent years had, however, raised the interest in the accuracy of these forecasts. US dollar/euro spot exchange rateVesi plotted the forecasts provided by JPMC and the actual spot exchange rate for the 2002-2005 period, in 90-day incrementsAs illustrated by Exhibit 1, the results were not encouraging. Although JPMC had hit the actual spot rate dead-on in both May and November 2002, the size of the forecasting errors, and direction of movement, seemed to increase over timeJP Morgans FX Forecasting2Copyright 2009 Pearson Prentice Hall. All rights reserved.Exhibit 1 Monthly Average Exchange Rates: US$/3Copyright 2009 Pearson Prentice Hall. All rights reserved.Growing Tension Over AccuracyWhat was most worrisome to Vesi was that in a good part of 2004, JPMC was getting the direction wrong. In February 2004 they had forecast the spot rate to move from the current rate of $1.27/ to $1.32/, but in fact the dollar had appreciated dramatically in the following three month period to close at $1.19/. Although Teknekron used a weighted moving average of the actual spot and forecasted rate in its foreign currency pricing (in this case, in euros), the directional error had caused the firm to average in a much weaker dollar than what had happenedThe buyer had been irritatedTeknekrons ExposureWhereas Teknekron did most of its sales in North America and Europe (hence the euro-based pricing), its sourcing was confined to the United States and JapanTeknekrons Japanese suppliers provided 3% discounts for Japanese yen-denominated invoicing, which Teknekron had traditionally been happy to take (the discount) and provide (yen payments)But this required Vesis company to manage and control its cost of goods sold including the yen-denominated costsJP Morgans FX Forecasting4Copyright 2009 Pearson Prentice Hall. All rights reserved.Japanese yen/US dollar spot rateVesi now turned to the predicted accuracy of JPMC on the yenExhibit 2 provides an overview of that analysisOnce again, although the dollar was consistently falling against the yen, the forecasting accuracy at least by eyeballing the graphic, was not encouraging. The most recent quarter had closed at 108/$, although the forecast had been to follow trend to 96/$.JP Morgans FX Forecasting5Copyright 2009 Pearson Prentice Hall. All rights reserved.Exhibit 2 Monthly Average Exchange Rates: Japanese Yen per U.S. Dollar6Copyright 2009 Pearson Prentice Hall. All rights reserved.1.How would you actually go about calculating the statistical accuracy of these forecasts? Would Vesi have been better off using the current spot rate as the forecast of the future spot rate, 90 days out?2.Forecasting the future is obviously a daunting challenge. All things considered, how well do you think JPMC is doing?3.If you were Vesi, what would you conclude about the relative accuracy of JPMCs spot rate forecasts? JP Morgans FX Forecasting: Case Questions7Copyright 2009 Pearson Prentice Hall. All rights reserved.1.How would you actually go about calculating the statistical accuracy of these forecasts? Would Vesi have been better off using the current spot rate as the forecast of the future spot rate, 90 days out?As shown in the following table, Vesi would actually have been much better off using the JPMorgan forecasts than simply using the current spot rate. The average error for the period shown was 1.2% for JPMorgan on the Us dollar/Euro, while the spot rate would have been 2.7%. Similarly, the average error for the period shown for the Japanese yen/US dollar spot rate was 0.2% for JPMorgan, while the spot rate would have been 1.7%.JP Morgans FX Forecasting: Case Questions8Copyright 2009 Pearson Prentice Hall. All rights reserved.JP Morgans FX ForecastingThe exchange rate forecasting error is calculated as a simple percent difference the differencebetween the forecasted rate and the actual ending rate as a percent of the actual rate.9Copyright 2009 Pearson Prentice Hall. All rights reserved.2.Forecasting the future is obviously a daunting challenge. All things considered, how well do you think JPMC is doing?JPMorgan seems to be doing okay for the period shown. As noted under question 1, the forecasting error is significantly less for both exchange rates than if a simple random walk (current spot rate forecast) had been used. The period under consideration is one of a definitive positive/negative trend for both exchange rate pairs. As is typically the case in forecasting, the biggest errors typically arise at the turning points when the price or rate changes direction. 3.If you were Vesi, what would you conclude about the relative accuracy of JPMCs spot rate forecasts? In both cases (exchange rate pairs), JPMC is consistently underestimating the actual exchange rate change. That is what the negative mean on the forecast error indicates. Given that both exchange rates are following strong trends over the sample period, and given the inherent conservatism likely to dominate the thinking of forecasting services, the sign of the average error is understandable. And the size of the error is reasonably small and relatively accurate. JP Morgans FX Forecasting: Case Questions10Copyright 2009 Pearson Prentice Hall. All rights reserved.