Impact of Exchange Rate Misalignment on Capital Inflows

Impact of Exchange Rate Misalignment on Capital Inflows EXCHANGE RATE MISALIGNMENT AND CAPITAL INFLOWS: AN ENDOGENOUS THRESHOLD ANALYSIS FOR MALAYSIA ABSTRACT This study presents an attempt to investigate the impact of exchange rate misalignment on capital inflows in Malaysia. Specifically, a precise threshold value is estimated to examine when exchange rate misalignment suppresses capital inflows. To pursue these objectives, this study relies on the endogenous threshold analysis as of Hansen (1996, 2000). Results suggest that misalignment in terms of currency overvaluation, has a negative and significant effect when overvaluation is more than 15 percent. This estimate is consistent and robust despite the changes in the choice of explanatory variables. INTRODUCTION Foreign direct investment (FDI) has served as an important engine of growth via skills and technology transfer, creation of employment opportunities and expanding the capital stock in Malaysia. Since the 1997 Asian financial crisis, Malaysia is no longer the top 10 host for FDI. In fact, the rate of growth of FDI has dramatically decrease compared to that of the early 1990s. This is partly due to reverse investment (Mat Zin, 1999) and declining dependence on FDI to finance growth. However, this may also indicates the declining competitiveness of Malaysia in attracting FDI which warrants empirical research since it would be vital to investigate which factors that contributed to the deterioration of competitiveness. Since early 1980s, real exchange rate misalignment has become a standard concept in international macroeconomic theory and policy (Razin Collins, 1997). Hence, this study focuses on exchange rate misalignment as an indicator of capital inflow competitiveness in the case of Malaysia. Malaysia provides an interesting case as it is one of the largest recipients of FDI amongst its ASEAN counterparts. Another advantage of undertaking a single country study is the ability to delineate the assumption that countries are similar in terms of social, cultural, economic and political background (Sun et al., 2002). Therefore, only relevant economic determinants are accounted for to suit the Malaysian environment. The objective of this paper is to investigate the empirical relationship between capital inflows and exchange rate misalignment. Whilst existing literature focuses on the role of exchange rate, this study takes a step further to examine the impact of exchange rate misalignment on capital inflows. Specifically, we estimate a threshold value at which misalignment begins to significantly affect capital inflows. To the best of our knowledge, no published study has attempted to estimate a threshold value for exchange rate misalignment in Malaysia. Hence, this study intends to fill this gap. Based on the endogenous autoregressive threshold (TAR) model developed by Hansen (2000), we split the sample into high and low misalignment regimes. Results suggest that exchange rate misalignment due to overvaluation is detrimental to the influx of capital inflows. The next section provides a brief overview of FDI in Malaysia followed by a brief explication of the theoretical model and review of liter ature. The fourth section spells out the method pertaining to the objective. The penultimate section provides results and discussion and the final section concludes. CAPITAL INFLOWS IN MALAYSIA: RECENT TRENDS AND INCENTIVES The essence of export oriented-growth nexus somewhat depends on the inflow of foreign capital into the country. In the past, foreign direct investment has been the one of the major conduit for technology transfer, job creation and export-led growth to this country. To pursue this line of interest, the Malaysian government has designed various policies spanning the gamut of industrial specific incentives, taxation, and intellectual property protection to infrastructure support. The company tax rate for example has been reduced from 33 percent in 1987 to 27 percent in 2007 and 26 percent in 2008. Other tax incentives such as the investment tax allowance, tax relief for companies with pioneer status or high technology industries has continued until today with more industries be given the relevant status to reap the benefits of the incentives. Most recently, the government has liberalized bumiputera equity requirements for 27 sectors to further boost competitiveness. With reference to previous information, there was a surge in foreign direct investment (FDI) into Malaysia in the late 1980s and this trend continued until the onset of the 1997 Asian financial crisis. Another acute slump in the influx of FDI occured in 2001 when the economy was in a slight recession but picked up again in 2002 thereafter. With the recent burgeoning world recession following the American sub-mortgage crisis, it is expected that FDI will contract again (IMF, 2009). To capture a more vivid impact of misalignment on capital inflows, this study employs quarterly data from Bank Negara Malaysia (BNM – the central bank of Malaysia) instead of the UNCTAD data which are annual. Foreign capital inflows or investment inflows comprises three items: (i) equity investment, (ii) loans and (iii) real estate. Investment consists of equity investment in Malaysia by non-residents, loans obtained from non-residents and purchase of real estate in Malaysia by non-residents but excludes retained earnings (Source: Bank Negara Malaysia, Glossary, Monthly Bulletin Statistics January, 2009, p. 186-187). This study resorts to a specific measure of FDI, that is, foreign investment inflows. Data starts from 1991:Q1-2008:Q3, partly dictated by availability. THEORY AND REVIEW OF LITERATURE In this study, we rely on the portfolio balance approach to model the determinants of foreign capital inflows. This model has been successfully tested by Goh (2005) for Malaysia. Branson (1968) postulates that the proportion of foreign assets (Kf) in a given stock of wealth is a function of the domestic and foreign interest rates (i and i*), the measure of exchange rate expectation or risk (e) and the stock of wealth (w) expressed as: (1)Darby et al. (1999), augment this concept of exchange rate risk (e) into exchange rate volatility and exchange rate misalignment. Since this study focuses on the role of exchange rate misalignment, we substitute e with misalignment. Expressing the above equation at level yields, (2)Focusing on Z, the literature suggests a number of variables that determines capital flows. The enigmatic relationship between FDI and exchange rate nexus has been widely examined and most of the discussions root back to the work of Kohlhagen (1977), Cushman (1985), Froot and Stein (1991), Goldberg (1993) and Darby et al. (1999). The effect of exchange rate is less straightforward (Benassy-Quere et al., 2001). The mechanisms that exchange rate affects capital inflows can also be viewed via the wealth effect channel and the relative production cost channel (Xing, 2006). A devaluation of the currency of the host country makes local cost of production lower in terms of foreign currency, hence leading to higher returns from export-oriented industries. As for the wealth effect, a devaluation makes local asset cheaper which motivates investors to acquire more. Kohlhagen (1977) static model postulates that following depreciation in host countries, MNEs will increase their production capacity. In a two period dynamic model, Cushman (1985) suggests that adjusted expected real depreciation lowers the production cost which leads to increase in FDI flows. Similarly, Goldberg (1993) illustrates how sectoral profitability, location effects, and portfolio and wealth effects are important factors that determine investment an d their links with exchange rates. In her theoretical model, the direction of investment effects triggered by exchange rate movements is ambiguous, therefore, warrants empirical research. On contrary, in an imperfect information framework, Froot and Stein (1991) show that appreciation induces wealth effect of foreign investors, thus encouraging foreign investors to acquire more local assets. Empirically, there is quite a consensus that a depreciation of the exchange rate in the host country leads to a reduction of the FDI (Klein and Rosengren, 1994; Dewenter, 1995). There is however, a dearth of studies that empirically examine the relationship between FDI and exchange rate misalignment. Empirical attempts include Benassy-Quere et al. (2001) who advocate the benefits of depreciation may be offset by excessive volatility of the exchange rate. Blonigen (1997) illustrates how currency depreciation induces foreign firm to acquire firm-specific assets when markets are segmented. Hasnat (1999) study the impact of misalignment on FDI for five developed nations on annual data ranging from 1976-1995. All of these studies use misalignment as a control variable or a counterpart for exchange rate variability and is measured by a deviation from the purchasing power parity (PPP) values. Furthermore, most of these studies are based on the experiences of industrialized economies using panel data analysis framework. In short, a prolonged misalignment may affect long term business decisions as it affects costs. If the exchange rate is overvalued relative to the e stimated equilibrium level, investors may acquire more domestic assets for future capital gains in host country currency terms (Barrell and Pain, 1996). On the other hand, persistent overvaluation may reduce cost competitiveness of production in the host country, especially for export oriented products. Other traditional determinants of FDI can be demarcated into at least two categories – micro and macro determinants. The list of micro-determinants spans from market size, growth, labour costs, host government policies, tariffs to trade barriers. The macro-determinants include market size (Chakrabarti, 2001; Farrell et al., 2004; Kravis and Lipsey; 1992), openness (Edwards, 1990; Gastanaga et al. 1998; Hausmann and Fernandez-Arias, 2000; Aseidu, 2002), rate of inflation (Bajo-Rubia and Sosvilla-Rivero, 1994; Urata and Kawai, 2000), government budget, taxes (Gastanaga et al., 1998; Wei, 2000) and infrastructure (Wheeler and Mody, 1992; Urata and Kawai, 2000). Financial deepening is also another catalyst for FDI (Borensztein et al., 1998). Liquid liability, private credit and M3 serve as proxies. Increase in money supply fuels inflation which increases the cost of production in the host country rendering a negative relationship. However, increments in money supply supported by g rowth or higher productivity indicate increase in future purchasing power which can benefit market-seeking FDI. Finally, the degree of misalignment is computed based on the difference between the actual and the hypothetical equilibrium exchange rate. Accordingly, the estimation of the hypothetical equilibrium exchange rate relies on the theory advocated by Edwards (1994). This theory postulates that the real exchange rate is a function of several fundamental variables which includes the Balassa-Samuelson effect, trade openness, net foreign assets and government spending. Details are provided in Sidek and Yusoff (2009). METHODOLOGY AND DATA The question of when does misalignment begin to significantly affect capital inflows necessitate the existence of a non-linear relationship between these two variables. Thus, if such non-linear relationship exists, then it is possible to estimate an inflexion point, or a threshold value, at which the sign of misalignment may change or become significant. In the non-linear time series modelling, the threshold autoregressive model (TAR) is more popular since it offers a relatively simple specification, estimation and interpretation compared to other non-linear models. The origins of TAR models roots back to Tong (1980) where the main idea is to approximate a general non-linear autoregressive structure by a threshold autoregession with a small number of regimes. Hansen (1996, 2000) derives the asymptotic distribution of the ordinary least squares (OLS) estimates of the endogeneous threshold parameters which is used in this study. This section explains how equation (2) is estimated to incorporate threshold effect. According to Hansen (2000), threshold estimation is the act of splitting the sample into two regimes when the threshold value is unknown. One necessary precondition is that the threshold variable must be a continuous variable. In this study, the threshold estimation is carried out by splitting the sample into high misalignment and low misalignment regime. Since misalignment is a continuous variable, TAR model would be appropriate to engender the threshold value. Formally, the two-regime threshold regression model takes the form: where is the threshold variable which is used to split the sample into two regimes, is the threshold value which is unknown and must be estimated, denotes the dependent variable (capital inflow), represents a vector of explanatory variables and is the error term assumed to be white noise and i.i.d. Note that if the threshold value is greater than the threshold variable, equation (3) is estimated and vice versa. This allows the regression parameters to change with respect to . In order to write equations (3) and (4) in a single equation, a dummy variable is used which is defined as where {.} is the indicator function, with d=1 when and d = 0, if otherwise; and set , such that (3) and where and . Equation (5) allows all the regression parameters , and to be estimated and switch between the two regimes. The least square (LS) technique is used to estimate through minimization of the sum of squared errors function. To implement this, the model is expressed in matrix notation, hence, equat ion (5) is expressed as: (6) Define, (7) as the sum of squared error function. By definition the least squares estimators which is also the MLE when with i.i.d. , jointly minimize equation (7). This minimization process requires to be restricted to a bounded set . The concentrated sum of squared errors function is written as: (8) where is the value that minimizes . As takes values that is less than n, is uniquely described as: with (9) Focusing on the objective of this section, the first step is to examine whether there exist a threshold effect in the model. This requires the examination between the linear model vis-Ã  -vis the two-regime model, equation (5). The null hypothesis of no threshold effect is tested against an alternative hypothesis where threshold effect is present. Since TAR models have a non-standard distribution, Hansen (1997, 2000) develops a standard heteroscedasticity-consistent Langrange Multiplier (LM) bootstrap method to calculate the asymptotic critical value and the p-value. The second step is to examine whether the derived threshold value is statistically significant. This is done by differencing the confidence interval region based on the likelihood ratio statistic . Based on Hansen (2000), let C represent the desired asymptotic confidence interval (in this study at 95%) and be the C-level critical value and set . Assuming homoscedasticity, as , therefore, is the asymptotic C-level confidence region for . If the homoscedasticity condition is not fulfilled, then a scale likelihood ratio statistics of the residual sum of squared errors is defined as: (10)and the adjusted confidence region becomes such that is robust whether or not the heteroscedasticity condition holds. Simulation is set at 1000 replications as suggested by Hansen (2000). Also, is not normally distributed hence, the valid asymptotic confidence intervals of the estimated threshold values in the no-rejection areas defined as , where is a given asymptotic level; and the no- rejection region of the confidence interval is . If , than the null hypothesis of cannot be rejected. In addition, to examine the possibility of a second threshold value, the same exercise is repeated. Specifically, the empirical model to be tested which is based on equation (2) is defined as follows: (11) where K is capital inflows, Mis, R and M3 denote exchange rate misalignment, interest differentials and financial deepening, and Z represents the other control variables. Table 1 summarizes the description of data, measurement and sources used in this study. Table 1: Determinants of Capital Inflows (1991Q1-2008Q3) Variable Description Measurement Source I Foreign investment Total foreign investment inflow as a percentage of GDP BNM M3 Money supply M2 as a percentage of GDP IFS D Government deficit The difference between revenue and expenditure as a percentage of GDP BNM R Interest differential The difference between Malaysia and US 3-month T-Bill rates IFS T Taxation Government corporate tax revenue as a percentage of GDP BNM LL Liquid Liability Log International liquidity: banking institution liability, line. 7b.d IFS INFRA Infrastructure Log of spending on infrastructure as a percentage of GDP BNM IFS: International Financial Statistics, IMF, UNCTAD: United Nations Conference on Trade and Development, BNM: Bank Negara Malaysia Monthly Statistical BulletinDOS: Department of Statistics, Malaysia (various issues). RESULTS AND DISCUSSION Prior to time series analysis, we test for unit roots in order to avoid spurious regression. Three versions of unit root testing, namely the ADF, PP and KPSS tests are employed to examine whether the variables are stationary on level or otherwise. Table 3 indicates that the order of integration are mixed for a majority of variables. However, this study proceeds to examine the threshold effect by including lagged variables for I(1) variables in the OLS estimation. Moreover, equation (2) derived from the theory requires estimations at level. Table 2: Unit root test ADF PP KPSS Order of Integration Level 1st Diff Level 1st Diff Level 1st Diff I -3.7029* -7.9812* -3.5286* 14.00208 0.9008* 0.2305 I(0)/I(1) M3 -1.2741 -10.0951* -1.3334 -10.4699* 1.0229* 0.3588*** I(1) D -1.6297 -19.7087* -8.8219* -27.3774* 0.3649* 0.0894 I(0)/I(1) R -4.5405* -3.8179** -2.6509 -7.0649* 0.0711 0.0471 I(0)/I(1) INFRA -2.2527 -4.5270* -3.5053* -27.7776* 0.2234* 0.0813 I(0)/I(1) LL -3.0805 -6.5500* -2.4386 -6.7355* 0.1073 0.0607 I(0)/I(1) MIS -3.8075** -9.7442* -3.8076** -9.8483* 0.0662 0.0577 I(0) Note: *, ** and *** denote significance at 1%, 5% and 10% significant level. p-values are in parentheses. For ADF and PP test the null is no unit root (H0: Variable is stationary) whilst the null for the KPSS is the existence of unit root (H0: Variable is not stationary). The baseline regression constitutes the exchange rate misalignment, interest differential and a measure of financial development, M3. We present four additional models with different variables added to the baseline regression, namely liquid liability, government budget deficit, and infrastructure for sensitivity analysis. Hansen (2000) theoretical construct allows for two threshold effects, hence, the first step is to investigate the possible existence of such an effect. Prior to that, a threshold variable needs to be selected. Since the aim of this section is to examine at what percentage exchange rate misalignment actually hurts capital inflows, the appropriate threshold variable is the exchange rate misalignment. Upon choosing the appropriate threshold variable, the next step is to observe any evidence of a threshold effect and whether there exist one or more threshold by employing the heteroscedasticity-consistent Lagrange-multiplier (LM) test for a threshold based on Hansen (1996). To test under the null hypothesis of no threshold effect, p-values are calculated using a bootstrap analog which generates the dependent variable from the distribution , where is the OLS residuals from the estimated threshold model. With 1000 bootstrap replications, the p-values for the baseline threshold models (Table 3) using misalignment strongly suggest the existence of threshold effect at 0.000. Subsequently, this suggests that there is a sample split based on the effect of exchange rate misalignment. Table 3: Threshold Effects for the baseline model Model 1 First Sample Split F-Stats 51.4045 Bootstrap P-Value 0.000 Threshold Estimates -15.0260% 95% Confidence Interval -15.446% , -9.8360% Second Sample Split F-Stats 16.2171 Bootstrap P-Value 0.2890 Note: H0: No threshold effect. The threshold is based on the minimized sum of squared residuals. This illustrates the graph of the normalized likelihood ratio sequence as a function of the threshold in exchange rate misalignment. The estimated is the value which minimizes these graphs which range at =15.02-15.44%. The dotted lines on the graphs present the 95% critical values. For example, in model 1, the asymptotic 95% confidence interval set where crosses the dotted lines. The results suggest that there is ample evidence for a two-regime specification. Also, it is worth noting that 41 of the 71 observations fall into the 95% confidence interval, hence, requires an examination of the possible existence of a second sample split. Results in Table 3, show that second sample split renders insignificant bootstrap p-value thus, indicating no further regime split. Table 4 presents the results for baseline regression. For comparison purposes, this study provides the linear OLS model without the threshold effect and a two-regime model which accommodates the threshold effect. Basically, the variables confer the correct signs in line with the prediction of the theory. Misalignment has a negative and significant effect on capital inflows in regime 2. Interest differential is expected to confer a negative effect. Results indicate that interest differentials only affects capital inflows negatively in the regime 1 but is insignificant in the regime 2. Similarly, M3 has significant effect in both regime but is positive in the regime 1 but the sign switches in regime 2. Hence, splitting the sample gives a more indepth view of the effects of these basic variables on investment inflows. To reiterate, sample splitting allows the examination of whether the significant effect is present in both regimes or otherwise. The results show that below the threshold value of 15%, exchange rate misalignment may be negative but are not statistically significant. However, above the 15% threshold level, misalignment exerts both negative and significant impact on capital inflows. A 1% increase in misalignment (overvaluation) suppresses capital inflows by approximately 1.19%. The negative effect of exchange rate misalignment on capital inflows is consistent with the findings of Hasnat (1999). Barrel and Pain (1996) argue that an apparent currency misalignment persistent over some length of time may affect investment inflows decisions. A reasonable explanation is that the relative production costs may be higher as a result of such misalignment. If the ringgit is thought to be overvalued relative to its estimated equilibrium level, then foreign production may be discouraged by the prospect of future capital loss in home currency terms. Another issue which emerges after the 1997 financial crisis is that capital inflows must be managed since reversals are likely to cause severe damage to the economy. Reinhart and Reinhart (1998) calls for greater exchange rate flexibility which is meant to introduce two-way risks, therefore, discouraging speculative capital inflows. It is, however, only possible in the context of de facto peg or a tightly managed float. Furthermore, the effectiveness of this policy depends on how much policymakers are willing to allow the exchange rate to fluctuate. A large band denotes greater flexibility but risks having large nominal appreciation which connotes possible overvaluation of the currency. The result of this study suggests that overvaluation is detrimental to capital inflows if this band exceeds 15%. Hence, policymakers should keep exchange rate fluctuations well below this 15% threshold. Table 4: Baseline regression results on the effect of misalignment on capital inflows (1991:Q1-2008:Q3). Dependent variable is capital inflows. Model 1 Linear Model Threshold Model OLS without threshold Regime 1 Â £ 15.0259% Regime 2 > 15.0259% Misalignment -0.4267** (0.2115) -0.3186 (0.2573) -1.1955** (0.5712) Interest Differential -0.0250*** (0.0131) -0.0438* (0.01533) -0.0261 (0.0193) M3 0.2964* (0.0391) 0.2644* (0.0516) -0.5560* (0.1240) Constant 3.0468* (0.2779) 2.5394* (0.2593) 6.7313* (0.6099) No. of Observations 71 42 29 R2 0.3664 0.6484 0.4218 Notes: *, ** and *** denote 1%, 5% and 10% significance respectively. Standard errors in parentheses. Interest rate differential are consistently negative and significant in all specifications and in both regimes in majority of the threshold model. This stresses the role of interest rates in attracting capital inflows into Malaysia. Although the impact may be small, it is significant and the authorities should ensure that interest rates are kept at certain levels to maintain competitiveness of Malaysia as destination for capital investment. In this paper, the estimated impact of a 1% change in interest differential is expected to subdue foreign investment by 0.04 percentage point in the first regime and 0.03 percentage point in the second regime. The proxy for financial deepening, M3 is statistically significant in all models and in both regimes. Again, this signifies the importance of financial development in attracting capital investment into Malaysia. Interestingly, M3 is positive during the periods of low misalignment regime (regime 1) but becomes negative at higher misalignment regime (regime 2). During low misalignment, a 1% increase in M3 is expected to draw in 0.3 percentage point more investment inflow into Malaysia. This shows that in the lower regime, financial depth acts as an impetus to capital inflows. However, the situation reverse with 0.6 percentage point lower investment inflows is expected with a 1% increase in misalignment in the second threshold regime. Montiel (1999) explicitly explains this phenomenon where capital inflows increase reserves which then prompt an increase in the monetary base, M2 and M3. Such increases fuels further increments in domestic demand leading to real appreciation. Thus, any overvaluation of the currency may eventually have negative ramifications on capital inflows. Sensitivity analysis To check for the sensitivity of the estimated threshold value, Table 6 -7 and Figure 3 represents four other models which use different variables in addition to the baseline regression. The addition of taxes yields insignificant results without drastically changing the threshold value. Other additional variables such as government budget deficit and liquid liability are only significant in one of the two regimes . With the inclusion of additional variables, the estimated magnitude of each regressors differ slightly but maintains the same sign and significance level. For example a 1% increase in misalignment (overvaluation) suppresses capital inflows by 1.11-1.55 percentage point. The estimated impact of a 1% change in interest differential is expected to deter foreign investment by 0.04-0.05 percentage point in the first regime and 0.02-0.06 percentage point in the second regime. Similarly, during low misalignment, a 1% increase in M3 is expected to draw in 0.2-0.3 percentage point m ore investment inflow into Malaysia. An estimated 0.49-0.67 percentage point lower investment inflows is expected with a 1% increase in M3 in the second threshold regime. In view of the results, it seems evident that the exchange rate policy has important effect in attracting foreign capital inflows into Malaysia. Specifically, misalignment in terms of overvaluation should be kept lower than 15 percent to ensure that capital inflows remained unhurt. Table 5: Sensitivity Analysis: Threshold Effects Model 2 Model 3 Model 4 Model 5 First Sample Split F-Stats 71.1442 45.9364 53.3722 53.3722 Bootstrap P-Value 0.000 0.000 0.000 0.000 Threshold Estimates -15.4461% -15.0260% -15.0260% -15.0260% 95% Confidence Interval -15.446%, -15.025% -15.446%, -9.836% -15.446%, -0.0984% -15.446%, -0.0984% Second Sample Split F-Stats 16.4917 19.7585 22.9710 22.9710 Bootstrap P-Value 0.5310 0.3800 0.2420 0.2420 Note: H0: No threshold effect. The threshold is based on the minimized sum of squared residuals Table 6: Sensitivity Analysis for threshold estimates (1991:Q1-2008:Q3). Model 2 Linear Model Threshold Model OLS without threshold Regime 1 Â £ 15.4461% Regime 2 > 15.4461% Misalignment -0.4278*** (0.2216) -0.3497 (0.4143) -1.5593* (0.3135) Interest Differential -0.0250*** (0.0134) -0.0462* (0.0153) -0.0599* (0.0131) M3 0.2966* (0.0414) 0.2732* (0.0488) -0.5609* (0.0744) Liquid Liability -0.0029 (0.1709) -0.0634 (0.1932) 1.1843* (0.2615) Constant 2.9780* (0.2713) 2.5259* (0.2593) 6.1799* (0.3135) No. of Observations 71 41 30 R2 0.3842 0.6503 0.5986 Model 3 Linear Model Threshold Model OLS without threshold Regime 1 Â £ 15.0260% Regime 2 > 15.0260% Misalignment -0.4472** (0.2038) -0.3800 (0.2460) -1.1171*** (0.6229) Interest Differential -0.0254* (0.0126) -0.0505* (0.0140) -0.0237 (0.0221) M3 0.2844* (7.4922) 0.2521* (0.0472) -0.5391* (0.1477) Deficit -0.7655* (0.3059) -0.7380* (0.3099) -0.1841 (0.7174) Constant 3.0308* (0.2674) 2.5835* (0.2445) 6.6452* (0.7337) No. of Observations 71 42 29 R2 0.4285 0.6829 0.4230 Model 4 Linear Model Threshold Model OL S without threshold Regime 1 Â £ 15.0260% Regime 2 > 15.0260% Misalignment -0.2852 (0.2181) -0.2582 (0.2720) 1.2490** (0.5612) Interest Differential -0.0275** (0.0128) -0.0419* (0.0165) -0.0311 (0.0204) M3 0.3208* (0.0401) 0.2796* (0.0583) -0.5489* (0.1245) Tax 2.1899** (1.0761) 0.1283 (0.1457) 0.1260 (0.1720) Constant 3.0274* (0.4383) 2.2463* (0.4806) 6.5027* (0.7227) No. of Observations 71 42 29 R2 0.3665 0.6516 0.4300 Model 5 Linear Model Threshold Model OLS without threshold Regime 1 Â £ 15.0260% Regime 2 > 15.0260% Misalignment -0.3780*** (0.1977) -0.4495*** (0.2602) -1.3190** (0.6059) Interest Differential -0.0203 (0.0123) -0.0433* (0.0152) -0.0308 (0.0212) M3 0.2941* (0.0365) 0.2388* (0.0479) -0.6093* (0.1406) Infrastructure 3.0729* (3.3373) 0.0474** (0.0228) -0.0382 (0.0392) Constant 3.0709* (0.2569) 2.5698* (0.2346) 7.0433* (0.7173) No. of Observations 71 42 29 R2 0.4091 0.6815 0.4384 Notes: *, ** and *** denote 1%, 5% and 10% significance respectively. Standard errors in parentheses. CONCLUSION The objective of this chapter is to examine the impact of exchange rate misalignment on capital inflows. Results provide evidences of the negative impact of misalignment on capital inflows. To reiterate, overvaluation of the ringgit signals that Malaysia is less competitive vis-Ã  -vis other countries. In addition, this paper also estimates a specific threshold value; that is the degree of misalignment after which it begins to hurt capital inflows. By employing a recent technique by Hansen (1996, 2000), this study splits the sample into high misalignment and low misalignment regimes. This study shows that misalignments hurt capital inflows in the high misalignment regime or when misalignment is greater than 15 percent. This study also confirms the work of Goh (2005) who suggests that the portfolio balance model can capture the determinants of capital inflows in Malaysia. In particular, the results suggest that interest differential is an important determinant albeit, small, hence, p olicies should be direc Impact of Exchange Rate Misalignment on Capital Inflows Impact of Exchange Rate Misalignment on Capital Inflows EXCHANGE RATE MISALIGNMENT AND CAPITAL INFLOWS: AN ENDOGENOUS THRESHOLD ANALYSIS FOR MALAYSIA ABSTRACT This study presents an attempt to investigate the impact of exchange rate misalignment on capital inflows in Malaysia. Specifically, a precise threshold value is estimated to examine when exchange rate misalignment suppresses capital inflows. To pursue these objectives, this study relies on the endogenous threshold analysis as of Hansen (1996, 2000). Results suggest that misalignment in terms of currency overvaluation, has a negative and significant effect when overvaluation is more than 15 percent. This estimate is consistent and robust despite the changes in the choice of explanatory variables. INTRODUCTION Foreign direct investment (FDI) has served as an important engine of growth via skills and technology transfer, creation of employment opportunities and expanding the capital stock in Malaysia. Since the 1997 Asian financial crisis, Malaysia is no longer the top 10 host for FDI. In fact, the rate of growth of FDI has dramatically decrease compared to that of the early 1990s. This is partly due to reverse investment (Mat Zin, 1999) and declining dependence on FDI to finance growth. However, this may also indicates the declining competitiveness of Malaysia in attracting FDI which warrants empirical research since it would be vital to investigate which factors that contributed to the deterioration of competitiveness. Since early 1980s, real exchange rate misalignment has become a standard concept in international macroeconomic theory and policy (Razin Collins, 1997). Hence, this study focuses on exchange rate misalignment as an indicator of capital inflow competitiveness in the case of Malaysia. Malaysia provides an interesting case as it is one of the largest recipients of FDI amongst its ASEAN counterparts. Another advantage of undertaking a single country study is the ability to delineate the assumption that countries are similar in terms of social, cultural, economic and political background (Sun et al., 2002). Therefore, only relevant economic determinants are accounted for to suit the Malaysian environment. The objective of this paper is to investigate the empirical relationship between capital inflows and exchange rate misalignment. Whilst existing literature focuses on the role of exchange rate, this study takes a step further to examine the impact of exchange rate misalignment on capital inflows. Specifically, we estimate a threshold value at which misalignment begins to significantly affect capital inflows. To the best of our knowledge, no published study has attempted to estimate a threshold value for exchange rate misalignment in Malaysia. Hence, this study intends to fill this gap. Based on the endogenous autoregressive threshold (TAR) model developed by Hansen (2000), we split the sample into high and low misalignment regimes. Results suggest that exchange rate misalignment due to overvaluation is detrimental to the influx of capital inflows. The next section provides a brief overview of FDI in Malaysia followed by a brief explication of the theoretical model and review of liter ature. The fourth section spells out the method pertaining to the objective. The penultimate section provides results and discussion and the final section concludes. CAPITAL INFLOWS IN MALAYSIA: RECENT TRENDS AND INCENTIVES The essence of export oriented-growth nexus somewhat depends on the inflow of foreign capital into the country. In the past, foreign direct investment has been the one of the major conduit for technology transfer, job creation and export-led growth to this country. To pursue this line of interest, the Malaysian government has designed various policies spanning the gamut of industrial specific incentives, taxation, and intellectual property protection to infrastructure support. The company tax rate for example has been reduced from 33 percent in 1987 to 27 percent in 2007 and 26 percent in 2008. Other tax incentives such as the investment tax allowance, tax relief for companies with pioneer status or high technology industries has continued until today with more industries be given the relevant status to reap the benefits of the incentives. Most recently, the government has liberalized bumiputera equity requirements for 27 sectors to further boost competitiveness. With reference to previous information, there was a surge in foreign direct investment (FDI) into Malaysia in the late 1980s and this trend continued until the onset of the 1997 Asian financial crisis. Another acute slump in the influx of FDI occured in 2001 when the economy was in a slight recession but picked up again in 2002 thereafter. With the recent burgeoning world recession following the American sub-mortgage crisis, it is expected that FDI will contract again (IMF, 2009). To capture a more vivid impact of misalignment on capital inflows, this study employs quarterly data from Bank Negara Malaysia (BNM – the central bank of Malaysia) instead of the UNCTAD data which are annual. Foreign capital inflows or investment inflows comprises three items: (i) equity investment, (ii) loans and (iii) real estate. Investment consists of equity investment in Malaysia by non-residents, loans obtained from non-residents and purchase of real estate in Malaysia by non-residents but excludes retained earnings (Source: Bank Negara Malaysia, Glossary, Monthly Bulletin Statistics January, 2009, p. 186-187). This study resorts to a specific measure of FDI, that is, foreign investment inflows. Data starts from 1991:Q1-2008:Q3, partly dictated by availability. THEORY AND REVIEW OF LITERATURE In this study, we rely on the portfolio balance approach to model the determinants of foreign capital inflows. This model has been successfully tested by Goh (2005) for Malaysia. Branson (1968) postulates that the proportion of foreign assets (Kf) in a given stock of wealth is a function of the domestic and foreign interest rates (i and i*), the measure of exchange rate expectation or risk (e) and the stock of wealth (w) expressed as: (1)Darby et al. (1999), augment this concept of exchange rate risk (e) into exchange rate volatility and exchange rate misalignment. Since this study focuses on the role of exchange rate misalignment, we substitute e with misalignment. Expressing the above equation at level yields, (2)Focusing on Z, the literature suggests a number of variables that determines capital flows. The enigmatic relationship between FDI and exchange rate nexus has been widely examined and most of the discussions root back to the work of Kohlhagen (1977), Cushman (1985), Froot and Stein (1991), Goldberg (1993) and Darby et al. (1999). The effect of exchange rate is less straightforward (Benassy-Quere et al., 2001). The mechanisms that exchange rate affects capital inflows can also be viewed via the wealth effect channel and the relative production cost channel (Xing, 2006). A devaluation of the currency of the host country makes local cost of production lower in terms of foreign currency, hence leading to higher returns from export-oriented industries. As for the wealth effect, a devaluation makes local asset cheaper which motivates investors to acquire more. Kohlhagen (1977) static model postulates that following depreciation in host countries, MNEs will increase their production capacity. In a two period dynamic model, Cushman (1985) suggests that adjusted expected real depreciation lowers the production cost which leads to increase in FDI flows. Similarly, Goldberg (1993) illustrates how sectoral profitability, location effects, and portfolio and wealth effects are important factors that determine investment an d their links with exchange rates. In her theoretical model, the direction of investment effects triggered by exchange rate movements is ambiguous, therefore, warrants empirical research. On contrary, in an imperfect information framework, Froot and Stein (1991) show that appreciation induces wealth effect of foreign investors, thus encouraging foreign investors to acquire more local assets. Empirically, there is quite a consensus that a depreciation of the exchange rate in the host country leads to a reduction of the FDI (Klein and Rosengren, 1994; Dewenter, 1995). There is however, a dearth of studies that empirically examine the relationship between FDI and exchange rate misalignment. Empirical attempts include Benassy-Quere et al. (2001) who advocate the benefits of depreciation may be offset by excessive volatility of the exchange rate. Blonigen (1997) illustrates how currency depreciation induces foreign firm to acquire firm-specific assets when markets are segmented. Hasnat (1999) study the impact of misalignment on FDI for five developed nations on annual data ranging from 1976-1995. All of these studies use misalignment as a control variable or a counterpart for exchange rate variability and is measured by a deviation from the purchasing power parity (PPP) values. Furthermore, most of these studies are based on the experiences of industrialized economies using panel data analysis framework. In short, a prolonged misalignment may affect long term business decisions as it affects costs. If the exchange rate is overvalued relative to the e stimated equilibrium level, investors may acquire more domestic assets for future capital gains in host country currency terms (Barrell and Pain, 1996). On the other hand, persistent overvaluation may reduce cost competitiveness of production in the host country, especially for export oriented products. Other traditional determinants of FDI can be demarcated into at least two categories – micro and macro determinants. The list of micro-determinants spans from market size, growth, labour costs, host government policies, tariffs to trade barriers. The macro-determinants include market size (Chakrabarti, 2001; Farrell et al., 2004; Kravis and Lipsey; 1992), openness (Edwards, 1990; Gastanaga et al. 1998; Hausmann and Fernandez-Arias, 2000; Aseidu, 2002), rate of inflation (Bajo-Rubia and Sosvilla-Rivero, 1994; Urata and Kawai, 2000), government budget, taxes (Gastanaga et al., 1998; Wei, 2000) and infrastructure (Wheeler and Mody, 1992; Urata and Kawai, 2000). Financial deepening is also another catalyst for FDI (Borensztein et al., 1998). Liquid liability, private credit and M3 serve as proxies. Increase in money supply fuels inflation which increases the cost of production in the host country rendering a negative relationship. However, increments in money supply supported by g rowth or higher productivity indicate increase in future purchasing power which can benefit market-seeking FDI. Finally, the degree of misalignment is computed based on the difference between the actual and the hypothetical equilibrium exchange rate. Accordingly, the estimation of the hypothetical equilibrium exchange rate relies on the theory advocated by Edwards (1994). This theory postulates that the real exchange rate is a function of several fundamental variables which includes the Balassa-Samuelson effect, trade openness, net foreign assets and government spending. Details are provided in Sidek and Yusoff (2009). METHODOLOGY AND DATA The question of when does misalignment begin to significantly affect capital inflows necessitate the existence of a non-linear relationship between these two variables. Thus, if such non-linear relationship exists, then it is possible to estimate an inflexion point, or a threshold value, at which the sign of misalignment may change or become significant. In the non-linear time series modelling, the threshold autoregressive model (TAR) is more popular since it offers a relatively simple specification, estimation and interpretation compared to other non-linear models. The origins of TAR models roots back to Tong (1980) where the main idea is to approximate a general non-linear autoregressive structure by a threshold autoregession with a small number of regimes. Hansen (1996, 2000) derives the asymptotic distribution of the ordinary least squares (OLS) estimates of the endogeneous threshold parameters which is used in this study. This section explains how equation (2) is estimated to incorporate threshold effect. According to Hansen (2000), threshold estimation is the act of splitting the sample into two regimes when the threshold value is unknown. One necessary precondition is that the threshold variable must be a continuous variable. In this study, the threshold estimation is carried out by splitting the sample into high misalignment and low misalignment regime. Since misalignment is a continuous variable, TAR model would be appropriate to engender the threshold value. Formally, the two-regime threshold regression model takes the form: where is the threshold variable which is used to split the sample into two regimes, is the threshold value which is unknown and must be estimated, denotes the dependent variable (capital inflow), represents a vector of explanatory variables and is the error term assumed to be white noise and i.i.d. Note that if the threshold value is greater than the threshold variable, equation (3) is estimated and vice versa. This allows the regression parameters to change with respect to . In order to write equations (3) and (4) in a single equation, a dummy variable is used which is defined as where {.} is the indicator function, with d=1 when and d = 0, if otherwise; and set , such that (3) and where and . Equation (5) allows all the regression parameters , and to be estimated and switch between the two regimes. The least square (LS) technique is used to estimate through minimization of the sum of squared errors function. To implement this, the model is expressed in matrix notation, hence, equat ion (5) is expressed as: (6) Define, (7) as the sum of squared error function. By definition the least squares estimators which is also the MLE when with i.i.d. , jointly minimize equation (7). This minimization process requires to be restricted to a bounded set . The concentrated sum of squared errors function is written as: (8) where is the value that minimizes . As takes values that is less than n, is uniquely described as: with (9) Focusing on the objective of this section, the first step is to examine whether there exist a threshold effect in the model. This requires the examination between the linear model vis-Ã  -vis the two-regime model, equation (5). The null hypothesis of no threshold effect is tested against an alternative hypothesis where threshold effect is present. Since TAR models have a non-standard distribution, Hansen (1997, 2000) develops a standard heteroscedasticity-consistent Langrange Multiplier (LM) bootstrap method to calculate the asymptotic critical value and the p-value. The second step is to examine whether the derived threshold value is statistically significant. This is done by differencing the confidence interval region based on the likelihood ratio statistic . Based on Hansen (2000), let C represent the desired asymptotic confidence interval (in this study at 95%) and be the C-level critical value and set . Assuming homoscedasticity, as , therefore, is the asymptotic C-level confidence region for . If the homoscedasticity condition is not fulfilled, then a scale likelihood ratio statistics of the residual sum of squared errors is defined as: (10)and the adjusted confidence region becomes such that is robust whether or not the heteroscedasticity condition holds. Simulation is set at 1000 replications as suggested by Hansen (2000). Also, is not normally distributed hence, the valid asymptotic confidence intervals of the estimated threshold values in the no-rejection areas defined as , where is a given asymptotic level; and the no- rejection region of the confidence interval is . If , than the null hypothesis of cannot be rejected. In addition, to examine the possibility of a second threshold value, the same exercise is repeated. Specifically, the empirical model to be tested which is based on equation (2) is defined as follows: (11) where K is capital inflows, Mis, R and M3 denote exchange rate misalignment, interest differentials and financial deepening, and Z represents the other control variables. Table 1 summarizes the description of data, measurement and sources used in this study. Table 1: Determinants of Capital Inflows (1991Q1-2008Q3) Variable Description Measurement Source I Foreign investment Total foreign investment inflow as a percentage of GDP BNM M3 Money supply M2 as a percentage of GDP IFS D Government deficit The difference between revenue and expenditure as a percentage of GDP BNM R Interest differential The difference between Malaysia and US 3-month T-Bill rates IFS T Taxation Government corporate tax revenue as a percentage of GDP BNM LL Liquid Liability Log International liquidity: banking institution liability, line. 7b.d IFS INFRA Infrastructure Log of spending on infrastructure as a percentage of GDP BNM IFS: International Financial Statistics, IMF, UNCTAD: United Nations Conference on Trade and Development, BNM: Bank Negara Malaysia Monthly Statistical BulletinDOS: Department of Statistics, Malaysia (various issues). RESULTS AND DISCUSSION Prior to time series analysis, we test for unit roots in order to avoid spurious regression. Three versions of unit root testing, namely the ADF, PP and KPSS tests are employed to examine whether the variables are stationary on level or otherwise. Table 3 indicates that the order of integration are mixed for a majority of variables. However, this study proceeds to examine the threshold effect by including lagged variables for I(1) variables in the OLS estimation. Moreover, equation (2) derived from the theory requires estimations at level. Table 2: Unit root test ADF PP KPSS Order of Integration Level 1st Diff Level 1st Diff Level 1st Diff I -3.7029* -7.9812* -3.5286* 14.00208 0.9008* 0.2305 I(0)/I(1) M3 -1.2741 -10.0951* -1.3334 -10.4699* 1.0229* 0.3588*** I(1) D -1.6297 -19.7087* -8.8219* -27.3774* 0.3649* 0.0894 I(0)/I(1) R -4.5405* -3.8179** -2.6509 -7.0649* 0.0711 0.0471 I(0)/I(1) INFRA -2.2527 -4.5270* -3.5053* -27.7776* 0.2234* 0.0813 I(0)/I(1) LL -3.0805 -6.5500* -2.4386 -6.7355* 0.1073 0.0607 I(0)/I(1) MIS -3.8075** -9.7442* -3.8076** -9.8483* 0.0662 0.0577 I(0) Note: *, ** and *** denote significance at 1%, 5% and 10% significant level. p-values are in parentheses. For ADF and PP test the null is no unit root (H0: Variable is stationary) whilst the null for the KPSS is the existence of unit root (H0: Variable is not stationary). The baseline regression constitutes the exchange rate misalignment, interest differential and a measure of financial development, M3. We present four additional models with different variables added to the baseline regression, namely liquid liability, government budget deficit, and infrastructure for sensitivity analysis. Hansen (2000) theoretical construct allows for two threshold effects, hence, the first step is to investigate the possible existence of such an effect. Prior to that, a threshold variable needs to be selected. Since the aim of this section is to examine at what percentage exchange rate misalignment actually hurts capital inflows, the appropriate threshold variable is the exchange rate misalignment. Upon choosing the appropriate threshold variable, the next step is to observe any evidence of a threshold effect and whether there exist one or more threshold by employing the heteroscedasticity-consistent Lagrange-multiplier (LM) test for a threshold based on Hansen (1996). To test under the null hypothesis of no threshold effect, p-values are calculated using a bootstrap analog which generates the dependent variable from the distribution , where is the OLS residuals from the estimated threshold model. With 1000 bootstrap replications, the p-values for the baseline threshold models (Table 3) using misalignment strongly suggest the existence of threshold effect at 0.000. Subsequently, this suggests that there is a sample split based on the effect of exchange rate misalignment. Table 3: Threshold Effects for the baseline model Model 1 First Sample Split F-Stats 51.4045 Bootstrap P-Value 0.000 Threshold Estimates -15.0260% 95% Confidence Interval -15.446% , -9.8360% Second Sample Split F-Stats 16.2171 Bootstrap P-Value 0.2890 Note: H0: No threshold effect. The threshold is based on the minimized sum of squared residuals. This illustrates the graph of the normalized likelihood ratio sequence as a function of the threshold in exchange rate misalignment. The estimated is the value which minimizes these graphs which range at =15.02-15.44%. The dotted lines on the graphs present the 95% critical values. For example, in model 1, the asymptotic 95% confidence interval set where crosses the dotted lines. The results suggest that there is ample evidence for a two-regime specification. Also, it is worth noting that 41 of the 71 observations fall into the 95% confidence interval, hence, requires an examination of the possible existence of a second sample split. Results in Table 3, show that second sample split renders insignificant bootstrap p-value thus, indicating no further regime split. Table 4 presents the results for baseline regression. For comparison purposes, this study provides the linear OLS model without the threshold effect and a two-regime model which accommodates the threshold effect. Basically, the variables confer the correct signs in line with the prediction of the theory. Misalignment has a negative and significant effect on capital inflows in regime 2. Interest differential is expected to confer a negative effect. Results indicate that interest differentials only affects capital inflows negatively in the regime 1 but is insignificant in the regime 2. Similarly, M3 has significant effect in both regime but is positive in the regime 1 but the sign switches in regime 2. Hence, splitting the sample gives a more indepth view of the effects of these basic variables on investment inflows. To reiterate, sample splitting allows the examination of whether the significant effect is present in both regimes or otherwise. The results show that below the threshold value of 15%, exchange rate misalignment may be negative but are not statistically significant. However, above the 15% threshold level, misalignment exerts both negative and significant impact on capital inflows. A 1% increase in misalignment (overvaluation) suppresses capital inflows by approximately 1.19%. The negative effect of exchange rate misalignment on capital inflows is consistent with the findings of Hasnat (1999). Barrel and Pain (1996) argue that an apparent currency misalignment persistent over some length of time may affect investment inflows decisions. A reasonable explanation is that the relative production costs may be higher as a result of such misalignment. If the ringgit is thought to be overvalued relative to its estimated equilibrium level, then foreign production may be discouraged by the prospect of future capital loss in home currency terms. Another issue which emerges after the 1997 financial crisis is that capital inflows must be managed since reversals are likely to cause severe damage to the economy. Reinhart and Reinhart (1998) calls for greater exchange rate flexibility which is meant to introduce two-way risks, therefore, discouraging speculative capital inflows. It is, however, only possible in the context of de facto peg or a tightly managed float. Furthermore, the effectiveness of this policy depends on how much policymakers are willing to allow the exchange rate to fluctuate. A large band denotes greater flexibility but risks having large nominal appreciation which connotes possible overvaluation of the currency. The result of this study suggests that overvaluation is detrimental to capital inflows if this band exceeds 15%. Hence, policymakers should keep exchange rate fluctuations well below this 15% threshold. Table 4: Baseline regression results on the effect of misalignment on capital inflows (1991:Q1-2008:Q3). Dependent variable is capital inflows. Model 1 Linear Model Threshold Model OLS without threshold Regime 1 Â £ 15.0259% Regime 2 > 15.0259% Misalignment -0.4267** (0.2115) -0.3186 (0.2573) -1.1955** (0.5712) Interest Differential -0.0250*** (0.0131) -0.0438* (0.01533) -0.0261 (0.0193) M3 0.2964* (0.0391) 0.2644* (0.0516) -0.5560* (0.1240) Constant 3.0468* (0.2779) 2.5394* (0.2593) 6.7313* (0.6099) No. of Observations 71 42 29 R2 0.3664 0.6484 0.4218 Notes: *, ** and *** denote 1%, 5% and 10% significance respectively. Standard errors in parentheses. Interest rate differential are consistently negative and significant in all specifications and in both regimes in majority of the threshold model. This stresses the role of interest rates in attracting capital inflows into Malaysia. Although the impact may be small, it is significant and the authorities should ensure that interest rates are kept at certain levels to maintain competitiveness of Malaysia as destination for capital investment. In this paper, the estimated impact of a 1% change in interest differential is expected to subdue foreign investment by 0.04 percentage point in the first regime and 0.03 percentage point in the second regime. The proxy for financial deepening, M3 is statistically significant in all models and in both regimes. Again, this signifies the importance of financial development in attracting capital investment into Malaysia. Interestingly, M3 is positive during the periods of low misalignment regime (regime 1) but becomes negative at higher misalignment regime (regime 2). During low misalignment, a 1% increase in M3 is expected to draw in 0.3 percentage point more investment inflow into Malaysia. This shows that in the lower regime, financial depth acts as an impetus to capital inflows. However, the situation reverse with 0.6 percentage point lower investment inflows is expected with a 1% increase in misalignment in the second threshold regime. Montiel (1999) explicitly explains this phenomenon where capital inflows increase reserves which then prompt an increase in the monetary base, M2 and M3. Such increases fuels further increments in domestic demand leading to real appreciation. Thus, any overvaluation of the currency may eventually have negative ramifications on capital inflows. Sensitivity analysis To check for the sensitivity of the estimated threshold value, Table 6 -7 and Figure 3 represents four other models which use different variables in addition to the baseline regression. The addition of taxes yields insignificant results without drastically changing the threshold value. Other additional variables such as government budget deficit and liquid liability are only significant in one of the two regimes . With the inclusion of additional variables, the estimated magnitude of each regressors differ slightly but maintains the same sign and significance level. For example a 1% increase in misalignment (overvaluation) suppresses capital inflows by 1.11-1.55 percentage point. The estimated impact of a 1% change in interest differential is expected to deter foreign investment by 0.04-0.05 percentage point in the first regime and 0.02-0.06 percentage point in the second regime. Similarly, during low misalignment, a 1% increase in M3 is expected to draw in 0.2-0.3 percentage point m ore investment inflow into Malaysia. An estimated 0.49-0.67 percentage point lower investment inflows is expected with a 1% increase in M3 in the second threshold regime. In view of the results, it seems evident that the exchange rate policy has important effect in attracting foreign capital inflows into Malaysia. Specifically, misalignment in terms of overvaluation should be kept lower than 15 percent to ensure that capital inflows remained unhurt. Table 5: Sensitivity Analysis: Threshold Effects Model 2 Model 3 Model 4 Model 5 First Sample Split F-Stats 71.1442 45.9364 53.3722 53.3722 Bootstrap P-Value 0.000 0.000 0.000 0.000 Threshold Estimates -15.4461% -15.0260% -15.0260% -15.0260% 95% Confidence Interval -15.446%, -15.025% -15.446%, -9.836% -15.446%, -0.0984% -15.446%, -0.0984% Second Sample Split F-Stats 16.4917 19.7585 22.9710 22.9710 Bootstrap P-Value 0.5310 0.3800 0.2420 0.2420 Note: H0: No threshold effect. The threshold is based on the minimized sum of squared residuals Table 6: Sensitivity Analysis for threshold estimates (1991:Q1-2008:Q3). Model 2 Linear Model Threshold Model OLS without threshold Regime 1 Â £ 15.4461% Regime 2 > 15.4461% Misalignment -0.4278*** (0.2216) -0.3497 (0.4143) -1.5593* (0.3135) Interest Differential -0.0250*** (0.0134) -0.0462* (0.0153) -0.0599* (0.0131) M3 0.2966* (0.0414) 0.2732* (0.0488) -0.5609* (0.0744) Liquid Liability -0.0029 (0.1709) -0.0634 (0.1932) 1.1843* (0.2615) Constant 2.9780* (0.2713) 2.5259* (0.2593) 6.1799* (0.3135) No. of Observations 71 41 30 R2 0.3842 0.6503 0.5986 Model 3 Linear Model Threshold Model OLS without threshold Regime 1 Â £ 15.0260% Regime 2 > 15.0260% Misalignment -0.4472** (0.2038) -0.3800 (0.2460) -1.1171*** (0.6229) Interest Differential -0.0254* (0.0126) -0.0505* (0.0140) -0.0237 (0.0221) M3 0.2844* (7.4922) 0.2521* (0.0472) -0.5391* (0.1477) Deficit -0.7655* (0.3059) -0.7380* (0.3099) -0.1841 (0.7174) Constant 3.0308* (0.2674) 2.5835* (0.2445) 6.6452* (0.7337) No. of Observations 71 42 29 R2 0.4285 0.6829 0.4230 Model 4 Linear Model Threshold Model OL S without threshold Regime 1 Â £ 15.0260% Regime 2 > 15.0260% Misalignment -0.2852 (0.2181) -0.2582 (0.2720) 1.2490** (0.5612) Interest Differential -0.0275** (0.0128) -0.0419* (0.0165) -0.0311 (0.0204) M3 0.3208* (0.0401) 0.2796* (0.0583) -0.5489* (0.1245) Tax 2.1899** (1.0761) 0.1283 (0.1457) 0.1260 (0.1720) Constant 3.0274* (0.4383) 2.2463* (0.4806) 6.5027* (0.7227) No. of Observations 71 42 29 R2 0.3665 0.6516 0.4300 Model 5 Linear Model Threshold Model OLS without threshold Regime 1 Â £ 15.0260% Regime 2 > 15.0260% Misalignment -0.3780*** (0.1977) -0.4495*** (0.2602) -1.3190** (0.6059) Interest Differential -0.0203 (0.0123) -0.0433* (0.0152) -0.0308 (0.0212) M3 0.2941* (0.0365) 0.2388* (0.0479) -0.6093* (0.1406) Infrastructure 3.0729* (3.3373) 0.0474** (0.0228) -0.0382 (0.0392) Constant 3.0709* (0.2569) 2.5698* (0.2346) 7.0433* (0.7173) No. of Observations 71 42 29 R2 0.4091 0.6815 0.4384 Notes: *, ** and *** denote 1%, 5% and 10% significance respectively. Standard errors in parentheses. CONCLUSION The objective of this chapter is to examine the impact of exchange rate misalignment on capital inflows. Results provide evidences of the negative impact of misalignment on capital inflows. To reiterate, overvaluation of the ringgit signals that Malaysia is less competitive vis-Ã  -vis other countries. In addition, this paper also estimates a specific threshold value; that is the degree of misalignment after which it begins to hurt capital inflows. By employing a recent technique by Hansen (1996, 2000), this study splits the sample into high misalignment and low misalignment regimes. This study shows that misalignments hurt capital inflows in the high misalignment regime or when misalignment is greater than 15 percent. This study also confirms the work of Goh (2005) who suggests that the portfolio balance model can capture the determinants of capital inflows in Malaysia. In particular, the results suggest that interest differential is an important determinant albeit, small, hence, p olicies should be direc

Gender is a Construct Essay

Each person is born either male or female, these are biological facts. However, no matter how clean cut these biological facts may be, they have social implications. Biologically speaking, there are minimal differences in the ability of male and female persons, none that would indicate a less able sex. Yet the underlying social assumptions associated to sex, translate to gender roles that clearly define a perceived difference because of sex. Gender is a social expectation, constructed through time, insisted and demanded through generations. It is a perception of roles and abilities created by society to define men and women as separate groups (Lerner, p. 238, Wolfe, p. 27-34). Sandra Lee Bartky (p. 61-86) further explains that people are born male and female and not masculine and feminine. Femininity is a social ideology imposed upon women, an attribute which is achieved through forceful and repeated learning (Lee Bartky, p. 61-8). By defining gender as a construct we acknowledge that gender is not an attribute that is biologically defined. Gender does not come innately in a person. Instead, gender is defined and perpetuated by social assumptions and expectations. Gender ideologies determine what is expected of each person dependent on sex, while gender roles determine how each person is to act in fulfilling their expectations and how each person is to relate to each other. Gender is an idea that is socially constructed based on expectations of social roles. The roles the men and women are expected to have in a society, and the perceived ideals of masculine and feminine are formed through social expectations and not biological characteristics. Broadly gender ideologies relate to socially constructed roles that define division of labor, distribution of power, individual rights and responsibilities and differentiation as one relates to society. Works Cited Lee Bartky, Sandra. â€Å"Foucault, Femininity and the Modernization of Patriarchal Power. † Feminism and Foucult: Paths of Resistance. Northern University Press, 1988. Lerner, Gerda. The Creation of Patriarchy: Women & History. USA: Oxford University Press, 1987. Wolfe, Alan. â€Å"The Gender Question. † The New Republic 6 June: 27-34.

Six Sigma analysis - Free Essay Example

Sample details Pages: 23 Words: 6769 Downloads: 5 Date added: 2017/09/16 Category Analytics Essay Type Narrative essay Tags: Manufacturing Essay Did you like this example? This semester we chose to develop a Six Sigma analysis on the manufacturing process of computers at Dell, Inc. Our goal was to take the manufacturing process currently in place for the production of laptops and desktop PCs and maximize quality, efficiency, and the longevity of the computers. Historically, Dell has been known as an industry leader in supply chain management. They have been credited with developing supply chain processes that have come to be recognized as some of the most innovative not only in their industry but throughout all business sectors. All of these accolades made Dell an unlikely choice since there didn’t appear to be much room for improvement, at least from a supply chain standpoint. However, over the past few years Dell’s once firm lead on the personal computer market share has begun to deteriorate and they have since lost their hold of the leading market share to top competitor Hewlett-Packard. They are currently in second place in market share but just over the past fiscal year revenues have fallen 33% from the second quarter of 2009 compared to the second quarter of 2010. Some of this drop-off may be attributed to the economic recession; but regardless of external factors a 33% loss is not something to be ignored especially at a time when these types of losses could potentially become a growing trend. Our research indicated that over the past few years the amount of complaints Dell has received regarding faulty manufacturing and shortened life spans for their computers has been continuously growing; so we decided to focus our analysis on determining how to improve on Dell’s quality without diminishing their industry-renowned built-to-order process which is based upon speed and efficiency. Dell’s recent losses are a result of decreased quality and these have subsequently created a lack of trust in Dell’s brand. We set out to not only determine specifically what hardware or software issues these errors can be attributed to; but also in the process, re-strengthen Dell’s brand identity by increasing quality for their products. When we were choosing a company to study and run analysis on, Dell was not necessarily any of our group members’ first choice, primarily because of how successful their supply chain methods had been in previous years. We initially assumed there would be little we could do to improve the process. We began developing a decision by choosing three companies to pick from; Dell, Inc. , Nike, and Herr’s Potato Chip Company. We made a decision after entering several different characteristics into the Decision Lens software and evaluating how strongly we felt about each. Our analysis was based on five criteria which we determined to be the most important for the success of this project. The first criterion was the availability of data. For this project, it was critical to have access to information with as much detail as possible. Such data includes process descriptions, mission statements, business plans, financial earnings, sales, marketing strategies and customer feedback. Without such data, it would have been difficult to evaluate and identify a process that would benefit from a Six Sigma project. All participants in this project recognized the importance of this criterion as evidenced by a 0. 41 weight rating, the highest weight given to any of the criteria. The second criterion was the scope of potential improvement. If the company is already excelling in their processes and dominating the market, it would be difficult to find any room for improvement. One of the companies that we initially considered was Coca-Cola. We subsequently dropped the company from the list because as we could not find many areas we could improve upon. The next criterion was our familiarity with the product. We felt it was important to have at least some knowledge of the company and process before we began the proje ct. Prior experiences with the company or product could be used to assist in our process improvement. Also, we felt a certain level of awareness could provide us with a better understanding of the company from a customer perspective. Our fourth criterion was complexity of the processes involved. In our analysis, the more complex processes often result in higher chances for imperfection or failure. We also felt that processes that require a trained specialist to enhance would not benefit from our analysis because of our lack of understanding the methods. The fifth criterion was personal interest. This was to ensure we were all engaged and interested in working on the decided project. Our criteria were given weights of . 41 for availability of information, . 35 for scope of improvement, . 1 for familiarity of the product, . 08 for complexity of the process, and . 06 for personal interest. Under these criteria out alternatives returned values of . 47, . 4 and . 125 for Dell, Herr ’s, and Nike respectively with an inconsistency of . 017. Prior to conducting the analysis, we felt Herr’s would be the best company for the project due to our familiarity with the company’s products, its nearby headquarters and the availability of a tour of the manufacturing process. However, Dell made a much larger amount of information more easily accessible to the general public which we determined would be more beneficial for us during this project. Essentially, our goal for this project was to first identify those aspects of the Dell manufacturing process that were not operating properly with regard to efficiency and quality and then develop ways to improve them by decreasing the amounts of money and time necessary to complete them while not further decreasing quality. At one time, Dell had control of the market share with its successful â€Å"direct-to-customer† sale and complex supply models. Rather than manufacture the components it uses to build computers, Dell uses an intricate supply model that consists of almost zero stock inventories. The company has built strong, trust-based relationships with its suppliers. Each supplier is carefully chosen based on predetermined criteria which range from quality to warehouse location. However, Dell has recently lost ground in the computer market. This is due primarily to increased competition and rising computer component defects. These issues have occurred in both Dell’s hardware and software, most recently with defective batteries and motherboards. For our Six Sigma project, we selected the design defect issue because of the large number of complaints as well as the high rate of defect reoccurrence. These issues caused frustration among Dell’s customers and support centers. Also to date, Dell has failed to come up with a long-term solution that has effectively reduced the number of defective products. The name we chose for this process is â€Å"Design Qu ality Control. † This is because Dell, as mentioned previously, doesn’t manufacture computer components but rather orders them from its suppliers. Therefore, the design of the product and the assembly of the components are the major areas that Dell fully controls. The design is the first step of Dell’s production process. Dell engineers design and develop different styles and accessory options from which the customers can choose. Consequently, the design should be adequate and have undergone sufficient quality control procedures. A good design doesn’t necessarily result in a product free of defects but it helps to significantly reduce their occurrence. Over the past several years, the trust and reliability that Dell has built with its customers has eroded. During this time, competitors such as HP and Apple have made significant gains. Unfortunately, replacements sent to customers also often contain the same or new defects. Fixing the design defects adds additional costs to the users who need to ship the defective computers back to Dell as well as to the company itself that will have to replace the component or the product. Customer service and technology support teams are also spending considerable time troubleshooting flawed components and dealing with dissatisfied customers. It is important that Dell respond to the defect issue because, in a recessionary economy, customers are paying more attention to the Quality/Price ratio. Currently, Dell is running the risk of becoming known primarily as a company with faulty products. Such a reputation can damage sales, especially in a period when purchases of computers and other big ticket items are down overall. For this project we used the same concept of improvement used by Motorola and we targeted a 100-fold improvement. The starting point of the project is when Dell’s engineers begin gathering requirements for the new computer model or option. The process ends when the su ppliers of all parts or software are selected and an execution plan is created. We did not set out to change the assembly process but through our results we feel it should be addressed under a separate six sigma process. Our first constraint was the completion time for this project. We felt it needs to be started quickly so the company can start effectively competing again. The second constraint was that our project should not increase the design time. This is important as technology advances quickly and Dell needs to keep up with the developments at the same pace of its competitors or faster. The Design Quality Control project will have an important impact on Dell and its customers. By improving the design and engineering of Dell’s computers, there will ultimately be a lower work load for Dell’s repair and customer service departments. This will lead to reduced operating costs for these departments. The project will also have a positive impact on the cost of pro duction. With a successful implementation, Dell will be able to reduce the cost of product maintenance in addition to cutting down on the repair or replacement of defective units. It will also be possible for the company to decrease the number of employees in their call centers. This will permit Dell to focus on production and innovation at lower costs while increasing revenues. Statistics show that three of every five computers sold in the United States are defective or will have defects in less than one year of operation. This is a defect rate of 60% (https://answers. google. com/answers/threadview? id=304307 ). According to an independent study, the average cost of repairing a defective computer is $200. In the third quarter of 2008, Dell had 13. 6% of the global PC market (https://retailindustry. about. com/od/topusretail- companies/p/dellincprofile. htm) and about 176. 8 million customers (13. 6% of 1. 3 billion PCs https://news. zdnet. com/2100-9584_22-140272. html ). Based on the statistics above, the cost of fixing all of Dell’s defective units would be approximately $21. 2 billion (176. 8M*60%*200=$21. billion). The project would permit Dell to avoid the cost of replacing or repairing components and reduce the cost related to a higher call volume in the customer service departments. In addition, it would reduce the cost of warranties and reassembly of products. The design process was split into six parts: concept or idea creation, research and market study, feedback and development, testing and evaluation, product finalization, and finally action plan creation. Dell is currently using PTC Windchill software to design almost all of its product line from the concept to servicing. However, the software can only be as good as the data input and it cannot eliminate the need for testing and product evaluation. For this project, the Critical to Quality (CTQ) parameters required that a detailed, upgraded design plan be first completed and tested prior to following through with the project. Using these results, we determined to move forward with the project. Furthermore, all stakeholders, including shareholders, all employees, customers, and suppliers needed to be informed and consulted on the details of the plan prior to and during its completion. Also, employees that work directly in the customer service and repair divisions of the company needed to be reassured that these improvements are necessary for the continued growth of the company. In order for the project to be successful, they needed to know that no employee’s job is necessarily in jeopardy as a result of the project. Finally, to ensure continued quality after the completion of the project, the objectives needed to be synthesized with the overall strategy of the company. It is important to have a fairly specific cost estimate as well as a timetable for completion prior to beginning the project. We estimated that a full overview and re-engineered desi gn could be completed, tested, and entered into mass production within one year. However, since this is not a new device or even a completely new design, we felt that we should aim for project completion in approximately eight months. Timing is critical because the longer people continue to purchase potentially defective devices, the more the brand suffers. The goal was to deliver the improved model to the public as soon as possible but without rushing it through quality testing. Finance measurements also play a key role in the success of the project. The cost of this project cannot exceed the amount that the problem is currently costing the company. Given the potentially enormous cost of these defects, this project should be considered as more of a reinvestment in the company’s dedication to quality as opposed to another company cost. In design matters, Dell takes advantage of Small and Medium Business (SMB) feedback, historical purchasing data, and analysis of technology and industry trends to define the appropriate specifications for the majority of its notebooks and computers. The key to producing cost-effective notebooks and computers, while still incorporating some of the same primary design tenets as high-end models, lies in understanding how specific components can affect the cost and complexity of the system. For example, each memory slot, hard disk drive, PCI slot, rear or front I/O feature requires the incorporation of the connectors as well as the associated electrical components, motherboard space, reserve power capacity, cooling capacity, and associated mechanical structures. Similarly, each hard drive bay requires many of those same components in addition to a backplane board space for hot-pluggable configurations. Furthermore, each motherboard requires supporting electrical components and motherboard space. Notebooks and computers are designed to help meet the needs of small and medium sized businesses in a cost-effective way th rough base configurations that incorporate the minimum feature set. The basic problem of the design begins when incorporating the components of the computer. The lifetime of the components largely depends on the way the notebook or computer has been designed. For example, space should be provided for the motherboards to cool down. If the cooling capacity is not sufficient there is a chance that the components on the motherboard might fail. In setting up our process maps for testing during the initial run for our current state we did not expect to find optimal results. It was after this first run where we determined what our â€Å"current state† is and what our â€Å"desired state† would be. More specifically, this phase revealed to us exactly how much improvement is necessary for Dell to achieve their desired state. Once the initial run of testing was completed we were also able to more clearly determine the major fault-points. By indicating which aspects of the product and assembly are creating the most frequent trouble we were then able to establish the best strategy to catch and avoid those issues during future production. This would also give us a much more accurate estimation of total cost for the implementation of this process. As of right now we can only make rough estimations of what this strategy would cost Dell but after getting a better idea of how severe the problem is we could make a better projection. Upon being implemented, we expect the reliability of Dell’s hardware components to significantly improve ultimately leading to the overall advancement of Dell’s computers. We fully expect this process to be pivotal in the reshaping of Dell’s product quality and as a result of that improvement their image will also improve. In the long term, implementing continuous and consistent components reliability testing will force Dell to improve their product. More products will be tested and more products will s how up as containing errors. A larger variety of errors will be tested for and more errors than ever before will be detected. The more problems Dell looks for and identifies the more prevention they will be able to do for future production and this will ultimately lead to a decrease in problem errors. Overall, this process was designed to make Dell money through decreasing product repair costs and improving sales through brand enhancement. By improving Dell’s quality, we improve Dell’s image and this will lead to a return to a number one spot in industry market share trough increased sales. For our gage RR study, our first test was collecting information about the components’ lifetime (by reading the barcode printed by the manufacturing company on the back of each component and comparing it to the lifetime cycle of the computer as determined by the designers). For this exercise, the test consisted of two inspectors testing three different computers models and testing four components (Motherboard, Central Processing Unit (CPU), Memory, Hard Drive) for each model four times. TB1: Standard Order for Collecting Data for the Gage RR Study. Row| Unit| Component| Inspector| Measurement| Row| Unit| Component| Inspector| Measurement| 123456789101112131415161718192021222324252627282930313233343 536373839404142434445464748| 111111111111111111111111111111112222222222222222| MotherboardMotherboardMotherboardMotherboardCPUCPUCPUCPUMemo yMemoryMemoryMemoryHard driveHard driveHard driveHard driveMotherboardMotherboardMotherboardMotherboardCPUCPUCPUCP UMemoryMemoryMemoryMemoryHard driveHard driveHard driveHard driveMotherboardMotherboardMotherboardMotherboardCPUCPUCPUCP UMemoryMemoryMemoryMemoryHard driveHard driveHard driveHard drive| JohnJohnJohnJohnJohn John John John John John John John John John John John GeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorge GeorgeGeorgeGeorgeGeorgeGeorgeGeorgeJohnJohnJohnJohnJohn John John John John J ohn John John John John John John | To be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collected| 495051525354555657585960616263646566676869707172737475767778 98081828384858687888990919293949596| 222222222222222233333333333333333333333333333333| MotherboardMotherboardMotherboardMotherboardCPUCPUCPUCPUMemo ryMemoryMemoryMemoryHard drive Hard driveHard driveHard driveMotherboardMotherboardMotherboardMotherboardCPUCPUCPUCP UMemoryMemoryMemoryMemoryHard driveHard driveHard driveHard driveMotherboardMotherboardMotherboardMotherboardCPUCPUCPUCP UMemoryMemoryMemoryMemoryHard driveHard driveHard driveHard drive| GeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorge GeorgeGeorgeGeorgeGeorgeGeorgeGeorgeJohnJohnJohnJohnJohn John John John John John John John John John John John GeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorgeGeorge GeorgeGeorgeGeorgeGeorgeGeorgeGeorge| To be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collect edTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collected| TB2: Random Order for Collecting Data for the Gage RR Study. Row| Stand. Order| Unit| Component| Inspector| Measurement| Row| Stand. Order| Unit| Component| Inspector| Measurement| 1| 27| 1| Memory| George| To be collected| 49| 50| 2| Motherboard| George| To be collected| 2| 46| 2| Hard drive| John | To be collected| 50| 65| 3| Motherboard| John| To be collected| 3| 52| 2| Motherboard| George| To be collected| 51| 74| 3| Memory| John | To be collected| 4| 19| 1| Motherboard| George| To be collected| 52| 58| 2| Memory| George| To be collected| 5| 94| 3| Hard drive| George| To be ollected| 53| 39| 2| CPU| John | To be collected| 6| 3| 1| Motherboard| John| To be collected| 54| 5| 1| CPU| John | To be collected| 7| 6 4| 2| Hard drive| George| To be collected| 55| 24| 1| CPU| George| To be collected| 8| 90| 3| Memory| George| To be collected| 56| 4| 1| Motherboard| John| To be collected| 9| 53| 2| CPU| George| To be collected| 57| 63| 2| Hard drive| George| To be collected| 10| 20| 1| Motherboard| George| To be collected| 58| 47| 2| Hard drive| John | To be collected| 11| 28| 1| Memory| George| To be collected| 59| 44| 2| Memory| John | To be collected| 12| 33| 2| Motherboard| John| To be collected| 60| 76| 3| Memory| John | To be collected| 13| 60| 2| Memory| George| To be collected| 61| 34| 2| Motherboard| John| To be collected| 14| 85| 3| CPU| George| To be collected| 62| 92| 3| Memory| George| To be collected| 15| 10| 1| Memory| John | To be collected| 63| 41| 2| Memory| John | To be collected| 16| 87| 3| CPU| George| To be collected| 64| 55| 2| CPU| George| To be collected| 17| 25| 1| Memory| George| To be collected| 65| 37| 2| CPU| John | To be collected| 18| 38| 2| CPU| John | To be collec ted| 66| 84| 3| Motherboard| George| To be collected| 19| 61| 2| Hard drive| George| To be collected| 67| 57| 2| Memory| George| To be collected| 20| 49| 2| Motherboard| George| To be collected| 68| 80| 3| Hard drive| John | To be collected| 21| 30| 1| Hard drive| George| To be collected| 69| 43| 2| Memory| John | To be collected| 22| 81| 3| Motherboard| George| To be collected| 70| 2| 1| Motherboard| John| To be collected| 23| 83| 3| Motherboard| George| To be collected| 71| 40| 2| CPU| John | To be collected| 24| 51| 2| Motherboard| George| To be collected| 72| 48| 2| Hard drive| John | To be collected| 25| 8| 1| CPU| John | To be collected| 73| 31| 1| Hard drive| George| To be collected| 26| 29| 1| Hard drive| George| To be collected| 74| 13| 1| Hard drive| John | To be collected| 27| 69| 3| CPU| John | To be collected| 75| 35| 2| Motherboard| John| To be collected| 28| 54| 2| CPU| George| To be collected| 76| 22| 1| CPU| George| To be collected| 29| 59| 2| Memory| George| To be collected| 77| 88| 3| CPU| George| To be collected| 30| 17| 1| Motherboard| George| To be collected| 78| 93| 3| Hard drive| George| To be collected| 31| 16| 1| Hard drive| John | To be collected| 79| 70| 3| CPU| John | To be collected| 32| 72| 3| CPU| John | To be collected| 80| 42| 2| Memory| John | To be collected| 33| 79| 3| Hard drive| John | To be collected| 81| 14| 1| Hard drive| John | To be collected| 34| 86| 3| CPU| George| To be collected| 82| 77| 3| Hard drive| John | To be collected| 35| 91| 3| Memory| George| To be collected| 83| 67| 3| Motherboard| John| To be collected| 36| 12| 1| Memory| John | To be collected| 84| 15| 1| Hard drive| John | To be collected| 37| 23| 1| CPU| George| To be collected| 85| 32| 1| Hard drive| George| To be collected| 38| 9| 1| Memory| John | To be collected| 86| 73| 3| Memory| John | To be collected| 39| 68| 3| Motherboard| John| To be collected| 87| 18| 1| Motherboard| George| To be collected| 40| 66| 3| Motherboard| John| To be collec ted| 88| 36| 2| Motherboard| John| To be collected| 41| 7| 1| CPU| John | To be collected| 89| 78| 3| Hard drive| John | To be collected| 42| 6| 1| CPU| John | To be collected| 90| 89| 3| Memory| George| To be collected| 43| 21| 1| CPU| George| To be collected| 91| 95| 3| Hard drive| George| To be collected| 44| 26| 1| Memory| George| To be collected| 92| 62| 2| Hard drive| George| To be collected| 45| 1| 1| Motherboard| John| To be collected| 93| 71| 3| CPU| John | To be collected| 46| 11| 1| Memory| John | To be collected| 94| 96| 3| Hard drive| George| To be collected| 47| 75| 3| Memory| John | To be collected| 95| 56| 2| CPU| George| To be collected| 48| 45| 2| Hard drive| John | To be collected| 96| 82| 3| Motherboard| George| To be collected| For the second test we collected information about the components’ performance (we used CPUInfo software which read four performance criteria: Measured CPU Speed, Rated CPU Speed, Caches and Memory). For this exercise, the test consisted of two inspectors testing one computer three times during different phases of the computer lifetime (Phase 1= after assembly, Phase 2 = after 3 years of use and Phase 3 = end of 5 year of use). In each phase, the component performance is measured four times to obtain measurement for four performance criteria. TB1: Standard Order for Collecting Data for the Gage RR Study. Row| Phase| performance criteria| Inspector| Measurement| Row| Phase| performance criteria| Inspector| Measurement| 123456789101112131415161718192021222324252627282930313233343 36373839404142434445464748| 111111111111111111111111111111112222222222222222| Measured CPU SpeedMeasured CPU SpeedMeasured CPU SpeedMeasured CPU SpeedRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDCachesCachesCachesCachesMemoryMemoryMemoryMemoryMeasure d CPU SpeedMeasured CPU SpeedMeasured CPU SpeedMeasured CPU SpeedRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDCachesCachesCachesCachesMemoryMemoryMemoryMem oryMeasure d CPU SpeedMeasured CPU SpeedMeasured CPU SpeedMeasured CPU SpeedRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDCachesCachesCachesCachesMemoryMemoryMemoryMemory| TinaTinaTinaTinaTina Tina Tina Tina Tina Tina Tina Tina Tina Tina Tina Tina MariaMariaMariaMariaMariaMariaMariaMariaMariaMariaMariaMaria MariaMariaMariaMariaTinaTinaTinaTinaTina Tina Tina Tina Tina Tina Tina Tina Tina Tina Tina Tina | To be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collected| 495051525354555657585960616263646566676869707172737475767778 798081828384858687888990919293949596| 222222222222222233333333333333333333333333333333| Measured CPU SpeedMeasured CPU SpeedMeasured CPU SpeedMeasured CPU SpeedRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDCachesCachesCachesCachesMemoryMemoryMemoryMemoryMeasure d CPU SpeedMeasured CPU SpeedMeasured CPU SpeedMeasured CPU SpeedRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDCachesCachesCachesCachesMemoryMemoryMemoryMemoryMeasure d CPU SpeedMeasured CPU SpeedMeasured CPU SpeedMeasured CPU SpeedRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDRATED CPU SPEEDCachesCachesCachesCachesMemoryMemoryMemoryMemory| MariaMariaMariaMariaMariaMariaMariaMariaMariaMariaMariaMaria MariaMariaMariaMariaTinaTinaTinaTinaTina Tina Tina Tina Tina Tina Tina Tina Tina Tina Tina Tina MariaMariaMariaMariaMariaMariaMariaMariaMariaMariaMariaMaria MariaMariaMariaMaria| To be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collectedTo be collected| For the Random test, since we could not randomly select Phases for the same computer, we assumed we would use three identical computers with each one in a different Phase (Unit1= Phase 1, Unit2= Phase2, unit3=Phase3) TB2: Random Order for Collecting Data for the Gage RR Study. Row| Stand. Order| Unit| performance criteria| Inspector| Measurement| Row| Stand. Order| Unit| performance criteria| Inspector| Measurement| 1| 88| 3| RATED CPU SPEED| Maria| To be collected| 49| 15| 1| Memory| Tina | To be collected| 2| 2| 1| Measured CPU Speed| Tina| To be ollected| 50| 36| 2| Measured CPU Speed| Tina| To be collected| 3| 74| 3| Caches| Tina | To be collected| 51| 92| 3| Caches| Maria| To be collected| 4| 44| 2| Caches| Tina | To be collected| 52| 43| 2| Caches| Tina | To be collected| 5| 39| 2| RATED CPU SPEED| Tina | To be collected| 53| 45| 2| Memory| Tina | To be collected| 6| 27| 1| Caches| Maria| To be collected| 54| 4| 1| Measured CPU Speed| Tina| To be collected| 7| 91| 3| Caches| Maria| To be collected| 55| 19| 1| Measured CPU Speed| Maria| To be collected| 8| 12| 1| Caches| Tina | To be collected| 56| 81| 3| Measured CPU Speed| Maria| To be collected| 9| 71| 3| RATED CPU SPEED| Tina | To be collected| 57| 64| 2| Memory| Maria| To be collected| 10| 31| 1| Memory| Maria| To be collected| 58| 56| 2| RATED CPU SPEED| Maria| To be collected| 11| 80| 3| Memory| Tina | To be collected| 59| 42| 2| Caches| Tina | To be collected| 12| 24| 1| RATED CPU SPEED| Maria| To be collected| 60| 65| 3| Measured CPU Speed| Tina| To be collected| 13| 26| 1| Caches| Maria| To be collected| 61| 90| 3| Caches| Maria| To be collected| 14| 53| 2| RATED CPU SPEED| Maria| To be collected| 62| 58| 2| Caches| Maria| To be collected| 15| 83| 3| Measured CPU Speed| Maria| To be collected| 63| 54| 2| RATED CPU SPEED| Maria| To be collected| 16| 61| 2| Memory| Maria| To be collected| 64| 23| 1| RATED CPU SPEED| Maria| To be collected| 17| 75| 3| Caches| Tina | To be collected| 65| 57| 2| Caches| Maria| To be collected| 18| 22| 1| RATED CPU SPEED| Maria| To be collected| 66| 49| 2| Measured CPU Speed| Maria| To be collected| 19| 67| 3| Measured CPU Speed| Tina| To be collect ed| 67| 32| 1| Memory| Maria| To be collected| 20| 13| 1| Memory| Tina | To be collected| 68| 79| 3| Memory| Tina | To be collected| 21| 6| 1| RATED CPU SPEED| Tina | To be collected| 69| 46| 2| Memory| Tina | To be collected| 22| 33| 2| Measured CPU Speed| Tina| To be collected| 70| 41| 2| Caches| Tina | To be collected| 23| 10| 1| Caches| Tina | To be collected| 71| 38| 2| RATED CPU SPEED| Tina | To be collected| 24| 77| 3| Memory| Tina | To be collected| 72| 84| 3| Measured CPU Speed| Maria| To be collected| 25| 17| 1| Measured CPU Speed| Maria| To be collected| 73| 76| 3| Caches| Tina | To be collected| 26| 55| 2| RATED CPU SPEED| Maria| To be ollected| 74| 35| 2| Measured CPU Speed| Tina| To be collected| 27| 16| 1| Memory| Tina | To be collected| 75| 89| 3| Caches| Maria| To be collected| 28| 78| 3| Memory| Tina | To be collected| 76| 82| 3| Measured CPU Speed| Maria| To be collected| 29| 40| 2| RATED CPU SPEED| Tina | To be collected| 77| 51| 2| Measured CPU Speed| Maria| To be collected| 30| 68| 3| Measured CPU Speed| Tina| To be collected| 78| 62| 2| Memory| Maria| To be collected| 31| 73| 3| Caches| Tina | To be collected| 79| 34| 2| Measured CPU Speed| Tina| To be collected| 32| 37| 2| RATED CPU SPEED| Tina | To be collected| 80| 21| 1| RATED CPU SPEED| Maria| To be collected| 33| 69| 3| RATED CPU SPEED| Tina | To be collected| 81| 28| 1| Caches| Maria| To be collected| 34| 52| 2| Measured CPU Speed| Maria| To be collected| 82| 5| 1| RATED CPU SPEED| Tina | To be collected| 35| 95| 3| Memory| Maria| To be collected| 83| 18| 1| Measured CPU Speed| Maria| To be collected| 36| 30| 1| Memory| Maria| To be collected| 84| 85| 3| RATED CPU SPEED| Maria| To be collected| 37| 59| 2| Caches| Maria| To be collected| 85| 29| 1| Memory| Maria| To be collected| 38| 47| 2| Memory| Tina | To be collected| 86| 66| 3| Measured CPU Speed| Tina| To be collected| 39| 20| 1| Measured CPU Speed| Maria| To be collected| 87| 3| 1| Measured CPU Speed| Tina| To be collected| 40| 25| 1| Caches| Maria| To be collected| 88| 86| 3| RATED CPU SPEED| Maria| To be collected| 41| 60| 2| Caches| Maria| To be collected| 89| 48| 2| Memory| Tina | To be collected| 42| 1| 1| Measured CPU Speed| Tina| To be collected| 90| 94| 3| Memory| Maria| To be collected| 43| 63| 2| Memory| Maria| To be collected| 91| 70| 3| RATED CPU SPEED| Tina | To be collected| 44| 11| 1| Caches| Tina | To be collected| 92| 87| 3| RATED CPU SPEED| Maria| To be collected| 45| 14| 1| Memory| Tina | To be collected| 93| 50| 2| Measured CPU Speed| Maria| To be collected| 46| 7| 1| RATED CPU SPEED| Tina | To be collected| 94| 8| 1| RATED CPU SPEED| Tina | To be collected| 47| 72| 3| RATED CPU SPEED| Tina | To be collected| 95| 96| 3| Memory| Maria| To be collected| 48| 9| 1| Caches| Tina | To be collected| 96| 93| 3| Memory| Maria| To be collected| CTQ| Criteria| Test| Decision| Fast Processing Time | The process time should be short to allow Dell to compete. Dell needs to meet the product releas e date. | Compare the effective release date for each product with the original publicly communicated release date. | If the effective release date is the same or prior to the predetermined release date, then the process is satisfactory. If not the process  need improvement. | Product Functionality| -The product should turn on when the power switch is turned on and stay on until it is turned off. The product should respond accurately   and within 15 seconds to a user’s commands. | -Randomly select computers after assembly, connect to power and manually turn them on and off 10 times, each time for different legths (5 min, 30 min, 1 h, 12h and 24h)   Randomly select computers after assembly connect to power and manually turn them on while measuring the response time after each basic command. | -If the computer didn’t start or unexpectedly shutdown at any time during this test, then the design needs improvements. If the computer turned on when the switch was turned on for each test and turned off when requested then the design is satisfactory. If the response time was greater than 15 seconds then the design need improvements. If the response time is less or equal than 15 second then the design is satisfactory. | Component Reliability| All components should at least last for the product’s lifetime. (Lifetime for an average computer is 5 years) -Performance of all components should stay the same through the product’s lifetime (assuming that we start with high performance, high performance is 4 to 5 million arithmetic and logical operations in a second). | -Compare the lifetime of each component with the computers presumed lifetime. -Measure the performance of the same product during multiple phases of its lifetime. We will use the following performance criteria: *Measured CPU Speed*Rated CPU Speed*Caches*Memory (Dell may have better measurement tools but for this project we are going to use CPUInfo software as the evaluation too l). | -If at least one component of a computer is found with a lifetime cycle shorter than the lifetime cycle of the computer, then design needs improvement. If all of the components lifetime cycle exceed the product lifetime cycle, then the design is satisfactory. -If one performance criterion for a computer is different from Dell’s original setup of the performance target during the design phase at any time during the product lifetime cycle, then the design needs improvement. If all performance criteria stays identical to Dell’s original criteria for the entire lifetime cycle, then the design is satisfactory. Product Reproducibility| -Product should be reproducible at any time and at any quantity with the same customer’s requirements and characteristics | Compare products side by side  | If the products have the same exact components, speed, performance, quality, then the design is satisfactory. If any difference was identified then the design needs improvem ents. | Standardized Production Steps | Each product should go though the same exact production and quality control steps: design, assembly,  evaluation and shipment. | Follow the production steps for the same and for different products. | If the any step was missing or was added then the design needs improvement. If the same steps were followed, then the design is satisfactory. | For our two-way ANOVA test our goal was to set up an analysis that would allow us to determine the degree of dependency between the number of replaced parts (defects) and four different factors: type of Dell computers used (notebook, laptops or desktops), type of users (student/school, home/office or IT professionals/developers), type of design (custom design, partially custom design or Dells standard design) and time of use (one year, three years or five years). For this study we planned to randomly select male and female owners of Dell computers and separate them into 81 groups. Each group would con tain 10 individuals that satisfy the criteria determined by the combination of factors and levels as shown in the table. During the study, we would ask each individual about one type of computer even if the individual owns multiple types of Dell computers. The question would be, â€Å"Was any part of your computer replaced since you purchased your computer from Dell Inc.? † This experiment has nine levels on each side of the table:   (custom design, partially custom design and standard design)*(one year, three years and five years) and (student/school, home/office and IT professionals/developers)*(notebook, laptops and desktops) In this example, we are interested in testing the following Null Hypotheses:   H1: The number of parts replaced does not depend on the type of use   H2: The number of parts replaced does not depend on the type of the omputer used   H3: The number of parts replaced does not depend on the type of the Design   H4: The number of parts replace d does not depend on the number of years of use Our anticipation was to find that the custom design computers are subject to higher rate of defects than the partially custom design computers. Our hypothesis was that the error rate gets higher as more time passes and as the uses become more sophisticated. We did not anticipate a significant difference for the defect rate among the different types of Dell computers. In our analysis we set out to find if there is in fact a relationship among the usage, age, user, and design and the number of times parts need to be replaced. The rationale for this analysis was so if we do find a relationship between any of these factors we could then make recommendations to Dell on how to prevent these types of malfunctions for future users. Ideally, we would like to be able to find some form of relationship because it would enable us to create preemptive measures and allow Dell to alter design specifications for these products. If a relationship exi sts this could be extremely beneficial to Dell because they would then be able to request information from their customers about usage types, and the length of time they planned to use the computer, etc and then Dell could create the computer based on these specifications. Two-Way ANOVA Table Factors| Student/School| Home/Office| IT Professionals/Developers| | Notebook| Laptops| Desktops| Notebook| Laptops| Desktops| Notebook| Laptops| Desktops| Custom Design| One year| 4| 5| 5| 4| 7| 5| 5| 7| 7| | Three years| 6| 7| 6| 5| 6| 5| 5| 7| 6| Five years| 7| 7| 5| 6| 7| 4| 7| 8| 7| Partially Custom Design| One year| 5| 5| 4| 4| 5| 3| 7| 6| 3| | Three years| 4| 3| 4| 4| 6| 4| 5| 6| 6| | Five years| 4| 5| 5| 5| 5| 5| 5| 7| 6| Dells Standard Design| One year| 4| 4| 2| 0| 2| 4| 1| 2| 3| | Three years| 3| 4| 3| 1| 4| 3| 1| 4| 2| | Five years| 4| 4| 1| 4| 4| 3| 2| 6| 2| *The data in this table are fakes. The next aspect we covered was building a simulation of the company’s current manufa cturing supply chain as well as a comparative â€Å"improved† supply chain model, and finally an â€Å"ideal† model to show how the company could create a seemingly â€Å"perfect† model assuming all other variables and resources would not interfere. For all three simulations we used a warm-up period of 2,400 minutes with results collected after 4,800 minutes. The working time was 16 hours per day with a five day week. The total run time for each simulation was 7,200 minutes and each operation kept the same operating time for the three maps. However, the percentage of â€Å"Yes† decisions improved from Current through Ideal. It’s also important to mention for the purposes of these maps that they were slightly modified since we assumed that each order received was for one single unit although realistically this process is actually being conducted hundreds and possibly even thousands of times each day concurrently with one another. In the curren t state of the process map we were able to produce sixty nine computers per week, in the improved state of the process map the simulation ended up producing one hundred and thirteen computers per week and in the Ideal state of the process map the simulation ended up producing one hundred and nineteen computers. From the results of the simulations, we noticed an improvement in the number of units produced after the implementation of multiple quality controls steps and improvement of the feedback between customers, customer service, suppliers and production unit. Given the results, we believed that the improved process is worth implementing from a quality and efficiency perspective. However, we don’t have a good estimate for the cost. Our goal in improving the current process was to cut down on errors and customer complaints by adding additional product and quality check points while also attempting to maintain the same or similar total times. The increase from sixty-nine co mputers produced to one hundred thirteen between the current to improved states was a large but expected increase but the one hundred thirteen to one hundred nineteen was not as large of an increase as we had expected indicting to us that our improved state, although not necessarily the ideal process, was incredibly close to what we would consider an idyllic model. Listed on the following three pages in three separate process maps are the results we got upon running Dell’s current, improved and ideal. Ultimately, through the use of process maps and the results of statistical analysis based on the numbers that were available for us for this project we feel that we have not only been able to successfully improve on Dell’s current supply chain model but also advise them on a number of important methods and strategies to implement and control these new models and ensure that they will thrive within their current system. Overall we feel Dell needs to make improvements on their computer designs in relation to the where specific parts will be placed for specific users. As stated previously, the life of a computer can be greatly influenced by how the user utilizes it. Certain components of the same model may need to be upgraded or replaced in a much shorter span of time because of how the computer is most commonly used. Dell already has the resources to determine this information while building the computer so that they can prevent these issues before they ever arise and extend the life of their machines. This means researching new, innovative designs to determine if simply rearranging the placement of certain parts alone would correct the problem or if it would require a set of upgraded parts. Building a strong, long-term relationship with their suppliers would also benefit Dell. Suppliers could help to make very valuable recommendations to Dell as well as aid in the reengineering of Dell’s designs. In combination with research is also the statistical testing. Determining whether there is correlation between errors and usage or errors and specific parts all begins with statistical testing and should be continued with as many different facets of the manufacturing process as necessary. A large part of implementing a six sigma strategy is establishing goals and accountability throughout all levels of an organization. This is important to not only track progress but also to reinforce that this is a company-wide agenda not just centered on manufacturing, design, and upper management. All employees should feel a sense of responsibility and personal drive to take action in making recommendations for improvements. Also, if employees feel that top management is truly on board and involved in these improvements it will send a strong message throughout the organization that this strategy is for the long-term and not just a passing phase. Finally, in closing, it is also important that Dell not forget and disregard what they do well. Their core competency is, and has been, for over a decade, their quick and efficient supply chain model. They successfully revolutionized the omputer industry with their built-to-order service and the incredible speed and efficiency at which they could produce personalized computers. So they certainly should not abandon the strategies that have brought them success. They should however increase the quality checks within the current model (which is what we did in our process maps). Speed and efficiency mean nothing if the product isn’t built properly and needs to be returned, costing the company money and aggravating customers. Among other initiatives, Dell’s mission statement outlines the goal to produce computers of the highest quality so this should be held to the highest standard in the manufacturing process. Don’t waste time! Our writers will create an original "Six Sigma analysis" essay for you Create order