Economy ISSN: 2313-8181 Vol. 1, No. 1, 20-31, 2014 www.asianonlinejournals.com/index.php/Economy 20 Political Instability and the Effectiveness of Economic Policies: The Case of Thailand from 1993-2013 Jonathan E. Leightner 1 --- Tomoo Inoue 2 1 Hull College of Business Administration Georgia Regents University, Georgia, USA 2 Faculty of Economics Seikei University, Kichijoji-kitamachi, Musashino-shi Tokyo, Japan Abstract This work is licensed under a Creative Commons Attribution 3.0 License Asian Online Journal Publishing Group 1. Introduction Thailand’s political/economic history from 1993 to 2013 is filled with changes in Thailand’s economic policies, political situation, and external environment. This time period involves the opening of an international banking center, a speculative attack, the collapse of an exchange rate that had been successfully fixed for thirteen years, the acceptance of a 17.2 billion dollar loan from the IMF along with the conditions that came with that loan, the political rise of Thaksin Shinawatra, the biggest land-slide election victory in Thai history, another election in which the opposition did not win enough seats to even initiate a censor debate, government condoned killing of drug lords and Muslims, the use of political power to enrich the Shinawatra clan, massive street protests in Bangkok, the calling of a snap election that the opposition boycotted and that resulted in a constitutional crisis, Thailand’s king telling Thailand’s judges to fix the situation or wear chamber pots over their heads to cover their shame, an annulling of the boycotted election, a coup, new elections, protesters taking over Bangkok’s international airport, a new government, different protestors taking over the central business/shopping area of Bangkok, the army dispersing the protesters while several malls burn, the election of Thaksin’s sister, Yingluck, as Thailand’s first female prime minister, and a failed rice manipulation scheme by the government (Leightner, 2007b). Immediately after this time period there was another coup. It is impossible to create a macroeconomic model that would capture the effects of all of these structural changes, but all of these changes could potentially modify the effectiveness of government policies by directly affecting the economy or by affecting internal and/or external business expectations. Even if such a model could be developed, there would be insufficient data to estimate it because the shortest temporal length for gross domestic product (GDP) data for Thailand is quarterly and there are too few observations between major structural changes. Fortunately, Leightner (2015) has developed a regression procedure that produces unbiased estimates that capture the influence of omitted variables without having to identify, quantify, find proxies for, or model the influence of these important, but omitted variables. Between 1993 and 2013, the Thai economy suffered from massive changes in its economy, political situation, and external environment. Given data constraints and the rapidity of the changing situation, it is impossible to create an adequate macroeconomic model for Thailand during this time period. Thus we use a statistical method designed to solve the omitted variables problem with regression analysis. This method produces a separate slope estimate for every observation which makes it possible to see how omitted variables are affecting the estimated relationships over time. We use this method to estimate dGDP/dG, dGDP/dMB, dGDP/dX, dGDP/de, and dGDP/dReserves using quarterly Thai data from 1993 to 2013 where GDP is gross domestic product, G is government consumption, MB is the monetary base, X is exports, e is the baht/US$ exchange rate, and Reserves are foreign reserves. We find that the pro-equality policies of the Thaksin regime were helpful and that export driven growth is no longer a viable option for Thailand. Although we approve of Thaksin’s economic policies, we disapprove of other aspects of his regime. Keywords: Fiscal policy, Monetary policy, Trade policy, Thailand, Omitted variables. http://creativecommons.org/licenses/by/3.0/ Economy, 2014, 1(1): 20-31 21 This paper uses Leightner (2015) method to estimate dGDP/dG, dGDP/dMB, dGDP/dX, dGDP/de, and dGDP/dReserves using quarterly Thai data from 1993 to 2013 where GDP is gross domestic product, G is government consumption, MB is the monetary base, X is exports, e is the baht/US$ exchange rate, and Reserves are foreign reserves. Leightner’s method produces a separate slope estimate for every observation which makes it possible to see how omitted variables affected the estimated relationships. We find that export driven growth is no longer a viable policy for Thailand and that the economic policies of the Thaksin regime were helpful to the Thai economy. However, it is important to realize that we do not condone the corruption, the extra-judicial killings of drug lords and Muslims, and the undercutting of the institutions required by democracy that are associated with the Thaksin regime. Leightner (2015) shows (using data from the 17 countries using the euro, the USA, the UK, Japan, Russia, Brazil, and China) that the world needs a pro-equality approach in order to solve the current world-wide economic malaise that is due to a surplus of savings and under-consumption. Thaksin’s economic policies were pro- equality. It is a shame that those effective policies came bundled with a corrupt, power-hungry, and immoral regime. Section II will provide an intuitive explanation of the statistical technique used in this paper. Section III presents the empirical results, and Section IV concludes. 2. An Intuitive Explanation of the Analytical Technique Used Omitting important variables from the analysis is one of regression analysis’ most serious problems. For example, if just equation (1) is estimated, while not considering the fact that equation (1)’s slope is a function of other variables (equation 2), then a constant slope is found, in contrast to the true slope which varies. Y = α0 + β1 X (1) β1 = α1 + α2 q (2) Y = α0 + α1X + α2 Xq (3) We substitute equation 2 into equation 1 to produce equation 3, which provides a convenient way to model the omitted variables problem. We initially assume that all variation from the fitted line (error) is due to omitted variables. Leightner (2015) shows that, in this case, the “error” for the ith observation from estimating equation (1) without considering equation (2) is α2Xi(qi – E[q]) where E[q] is the mean value of q. The standard approach to the omitted variable’s problem is to use instrumental variables. However, this standard approach requires instruments that are almost perfectly correlated to the omitted variables while not being independently related to the dependent variable. Finding and justifying such variables is usually impossible. Once instruments are selected, their relationship to the omitted variables and the relationship between the omitted variables and the dependent variable must be correctly modeled. All of these conditions are impossible to meet for a subject as complex as the effectiveness of macroeconomic policies. For recent papers that express concern over omitted variable bias see Abbott and Allen (2011), Angrist and Alan (2001), Black and Lisa (2001), Botosan and Marlene (2002), Cellini (2008), DiPrete and Markus (2004), Harris and Keith (2007), Mustard (2003), Pace and James (2010), Paterson and Kevin (2002), Scheffler et al. (2007) , Sessions et al. (2006), Streams and Edward (2004), and Swamy et al. (2003). Fortunately, Branson and Knox (2000) explain that the observations at the top of any given data set would be associated with the most favorable values for all omitted variables (in other words, the values of the omitted variables that would lead to the largest values for the dependent variable, ceteris paribus). Building on this intuition, Leightner (2002) developed a new analytical technique named “Reiterative Truncated Projected Least Squares” (RTPLS) that solves the omitted variable problem of regression analysis without using instrumental variables and their unreasonable assumptions. Leightner (2008) and Leightner and Tomoo (2012b) created the second and third generation of the technique respectively: RTPLS2 and RTPLS3. Leightner and Tomoo (2012b) also produce an argument that RTPLS3 is unbiased. Leightner (2015) uses the fourth generation, RTPLS4. His simulations show that the average ratio of Ordinary Least Squares (OLS) error to RTPLS error is 3.8 when the effect of omitted variables on the true slope is 10 times the size of random error. Furthermore, the OLS/RTPLS4 error ratio is more than 28 when the effect of omitted variables on the true slope is 100 times the size of random error. As random error approaches zero and as the sample size increases, the OLS/RTPLS error ratio approaches infinity. Published studies that used RTPLS, RTPLS2, RTPLS3, or RTPLS4 in applications include Leightner and Tomoo (2012a; 2012b; 2009; 2008a; 2008b; 2007) and Leightner (2015; 2013; 2011a; 2011b; 2010a; 2010b; 2008; 2007a; 2005a; 2005b; 2002). Figure 1 will be used to explain RTPLS. To construct Figure 1, we generated two series of random numbers, X and q, which ranged from 0 to 100. We then calculated a series for the dependent variable, Y, using equation (4): Y = 100 + 10 X + 0.6 q X (4) ∂Y/∂X for equation (4) equals 10 + 0.6 q. This slope will take numerical values that range from ten (when q equals zero) to seventy (when q equals one hundred). Since seventy is seven times bigger than ten, q makes a seven hundred percent difference to the true slope in this example. In Figure 1, we identified each point with that observation’s value for q. Notice that the upper edge of the data corresponds to relatively large qs – 91, 96, 98, 98, 98, 96, 95, and 94. The lower edge of the data corresponds to relatively small qs – 17, 0, 3, 4, and 11. This makes sense since as q increases so does Y, for any given X. For example, when X approximately equals 85, reading the values of q from top to bottom of Figure 1 produces 86, 80, 75, 68, 65, 54, 49, 43, 35, 27, and 21. Thus the relative vertical position of each observation is directly related to the values of q. If, instead of adding 0.6qX in equation 4, we had subtracted 0.6qX, then the smallest qs would be on the top and the largest qs on the bottom of Figure 1. Either way, the vertical position of observations captures the influence of q. Also realize that the omitted variable, q, represents the combined effects of all forces that are not included in the analysis. For example, if there are 500 forces that are omitted where 300 of them are positively related to Y and 200 are negatively related to Y, then the observations on the frontier will correspond to when the 300 variables are at their largest levels and the 200 are at their lowest levels. Economy, 2014, 1(1): 20-31 22 Where q is biggest, along the upper edge of Figure 1, the slope is also biggest. Likewise, where q is smallest, along the bottom edge of Figure 1, the slope is also smallest. The relative vertical position of the observations that correspond to any given X is related to the true slope. Imagine that a researcher does not know what q is and that he, thus, omits it from his analysis. In this case, OLS produces the following estimated equation: Y = -59.85 + 45.27X with an R-Squared of 0.6524 and a standard error of the slope of 3.338. Although, this OLS regression may look successful, it is not. Remember that the true equation is Y = 100 + 10 X + 0.6 q X. Since q ranges from 0 to 100, the true slope (true derivative) ranges from 10 to 70 and OLS produced a constant slope of 45.27. OLS did the best it could, given its assumption of a constant slope; OLS produced a slope estimate of approximately 10 + 0.6 E(q) = 10 + 0.6(53) = 42. OLS is always biased when it assumes a constant slope when, in reality, the slope varies. The most important implication from Figure 1 is that the relative vertical position of different observations contains information about the combined influence of all omitted variables. RTPLS4 uses this relative vertical position to solve the omitted variables problem. Furthermore, using RTPLS4 does not require that the researcher find appropriate proxies for the omitted variables, measure the omitted variables, or model how the omitted variables affect the dependent variable. Indeed, a researcher can use the relative vertical position of observations to capture the combined effects of all omitted variables even when he cannot name the important omitted variables. Fig-1. The inuttion behind BD-RTPLS4 RTPLS4 uses Data Envelopment Analysis (DEA) to draw a frontier through the top data points in Figure 1. All the data is then projected vertically upwards to this frontier. Since the observations on the frontier are the observations for which the omitted variables are most favorable (the values for q are at their highest levels along the Economy, 2014, 1(1): 20-31 23 top edge of Figure 1), this projection makes every observation correspond to the most favorable omitted variables values. In other words, this projection equalizes the influence of omitted variables to their most favorable level. However, to the right of the top most observation (the 94 in the upper right hand corner in Figure 1), DEA draws a horizontal line extending to infinity. The projection of all observations vertically upward may result in some observations being projected to this horizontal section. For example, the 81 which is closest to the upper right hand corner of Figure 1 would be projected to a horizontal section of the frontier. Because this horizontal section has nothing to do with the true relationship between X and Y, it should not be included when OLS is used to find a slope estimate for the projected data. The OLS slope estimate found for the projected data (after truncating off any horizontal part of the frontier) is a slope estimate for when q is at its most favorable level. We call this slope estimate the “truncated projected least squares” slope estimate (TPLS) for the first iteration. A new column is then added to the data and this TPLS slope estimate is put into that data column for the observations that did not need to be projected upwards – i.e. the observations through which DEA drew the initial frontier. The observations that did not need to be projected upwards are then cut from the original data file and pasted into a second data file and the procedure repeated using the original data file (sans all previous iterations’ frontier observations). We reiterate this process, peeling the data down, layer by layer, from the top of the data to the bottom of the data. The first iteration of this process produces a TPLS slope estimate for when omitted variables cause the dependent variable, Y, to be at its highest numerical level. The second iteration produces a TPLS slope estimate for when omitted variables cause the dependent variable to be at its second highest level, etc. When the regression at the end of an additional iteration would use fewer than ten observations, this process is terminated (the remaining observations will be located at the bottom of the data). Once the data set has been peeled from the top to the bottom, we return to the original data set and peel it up from the bottom to the top. When peeling up from the bottom, we project the data downward to the lower boundary of the data, we truncate off any lower left horizontal region, we run OLS regressions through the truncated projected data, we cut the frontier observations from the original data set and paste them (along with their TPLS estimates) into the second data set, and we reiterate this process until there are fewer than 10 observations remaining (the remaining observations will be at the top of the original data). By peeling the data both directions – from bottom to top and from top to bottom – all observations will have at least one TPLS estimate associated with it and some (in the middle of the data) will have two. All the “peeling up” and “peeling down” TPLS estimates (with the corresponding original data) are put into the second data file. We then use this second data file to run a final regression where these TPLS estimates minus Y/X are the dependent variable and 1/X is the sole independent variable. The following derivation provides the rational for this final regression. Y = α0 + α1 X + α2 Xq (Equation 3 repeated) (5) ∂Y/∂X = α1 + α2q (Derivative of equation 5) (6) Y/X = α0/X + α1 + α2q (Dividing equation 5 by X) (7) α1 + α2q = Y/X - α0/X (Rearranging equation 7) (8) ∂Y/∂X = Y/X - α0/X (From equations 6 and 8) (9) ∂Y/∂X - Y/X = - α0/X (rearranging equation 9) (10) The α0 estimated in this final regression and the data for Y/X and X are plugged into equation 9 to produce a separate RTPLS4 slope estimate for each observation. Alternatively, Generalized Least Squares (GLS) could be used to estimate α0 from equation (1) and the resulting α0 along with data on Y/X and X could be plugged into equation (9); Inoue et al. (2014) name this alternative approach “Variable Slope Generalized Least Squares” (VSGLS). Theoretically VSGLS produces the best linear unbiased estimate (BLUE) for α0 because equation (1) has heteroscedastic error (Aitken, 1935); however simulations show that VSGLS produces between twice and three times the error of RTPLS4 when sample sizes of 250 observations are used and all error is due to omitted variables. Thus, under these conditions, RTPLS4 is “better than BLUE.” However, when random error is added, RTPLS4 and VSGLS perform equally well (Inoue et al., 2014) . Figure 1 can be used to better explain the role of the final regression in the RTPLS4 process. If all the upper frontier observations had the exact same omitted variable values (perhaps 97), then the resulting TPLS estimate would be exactly equivalent to the true slope for the frontier observations. However, Figure 1 shows that the observations on the upper frontier were associated with omitted variable values of 91, 96, 98, 98, 98, 96, 95, and 94. The resulting TPLS slope estimate would perfectly fit a q value of approximately 96 (the mean of 91, 96, 98, 98, 98, 96, 95, and 94). When a TPLS estimate for a q of 96 is associated with qs of 91, 96, 98, 98, 98, 96, 95, and 94, some random variation (both positive and negative variation) remains. By stacking the results from all iterations when peeling down and up, and then conducting this final regression, this random variation is eliminated. Realize that Y is co-determined by X and q. Thus the combination of X and Y should contain information about q. This final regression exploits this insight in order to better capture the influence of q. RTPLS4 generates reduced form estimates that include all the ways that X and Y are correlated. Thus, even when many variables interact via a system of equations, a researcher using RTPLS4 does not have to discover and justify that system of equations. In contrast, traditional regression analysis theoretically must include all relevant variables in the estimation and the resulting slope estimate for ∂Y/∂X is for the effects of just X – holding all other variables constant. RTPLS4’s reduced form estimates are not substitutes for traditional regression analysis’ partial derivative estimates. Instead traditional regression analysis and RTPLS4 are compliments that capture different types of relationships. One disadvantage of RTPLS4 is that it cannot determine the mechanism by which the independent variable affects the dependent variable. On the other hand, RTPLS4 has the significant advantage of not having to model and find data for all the forces that can affect Y in order to estimate dY/dX. Both RTPLS4 and standard regression techniques estimate “correlations.” Neither one of them prove “causation.” We created confidence intervals for each RTPLS4 estimate by grouping the estimate with the 2 estimates before and after it and then using equation (11), which is based on the Central Limit Theorem. Economy, 2014, 1(1): 20-31 24 99% confidence interval = mean + (s/√n)tn-1, α/2 (11) In equation (11), “s” is the standard deviation, “n” is the number of observations which is 5, and tn-1, α/2 is 4.032 as taken off the standard t table. By always considering a given estimate and the 2 estimates before and after it, we created a moving confidence interval (much like a moving average) for a given set of RTPLS estimates. This 99% confidence interval can be interpreted as meaning that there is only a one percent chance that the true average for a given RTPLS estimate with the two RTPLS estimates before it and the two RTPLS estimates after it will lie outside of the given range. Given this interpretation, it is possible for a given RTPLS estimate to lie outside of its confidence interval if the other four estimates around it are all noticeably above or below the given RTPLS estimate. 3. The Data and the Empirical Results The data is given in Table 1. The data for G, MB, and X are in millions of current baht. The exchange rate data is given in Thai baht per US dollar, and the foreign reserves are given in millions of current US dollars. The empirical results are given in Table 2 where the columns labeled “upper” and “lower” give the upper and lower bounds for a 99% confidence interval for the estimates in the column immediately to the left of these labels. Figures 2 through 6 depict the empirical results. Table-1.The Data (GDP, G, MB, and X in millions of Baht; e in Baht per US $; reserves in millions of US dollars held at the end of the quarter) GDP G MB X e reserves 1993 Q1 755554 71336 247009 278949 25.4893 22239.4 Q2 755573 75095 251122 278051 25.2309 23979.8 Q3 811118 89321 253105 314353 25.2344 25225.3 Q4 842977 80230 271418 330152 25.3676 25438.8 1994 Q1 886103 84356 292341 329351 25.4176 26672.6 Q2 870964 82994 290243 332490 25.2050 28340.5 Q3 896836 102342 298691 356983 24.9855 29950.2 Q4 975438 84695 313430 391962 25.0403 30279.0 1995 Q1 1033855 102381 346179 425287 24.9711 30119.5 Q2 1026365 98676 356481 421084 24.6360 34958.3 Q3 1032857 115885 358564 438776 24.9432 35866.1 Q4 1093135 97461 377820 466527 25.1481 37026.7 1996 Q1 1116552 111052 404999 461125 25.2593 38982.5 Q2 1146094 111159 401172 441700 25.3116 39830.0 Q3 1154274 125651 402961 442540 25.3478 39537.0 Q4 1194121 121654 430313 464545 25.5124 38724.5 1997 Q1 1158084 117538 459817 471972 25.8872 38065.6 Q2 1165717 113471 469964 467382 25.4479 32353.0 Q3 1182021 130782 445963 589362 32.6166 29612.2 Q4 1226788 114914 453171 743399 40.2875 26967.7 1998 Q1 1210828 116037 461473 776676 46.6776 27680.0 Q2 1117120 111876 441534 645169 40.3994 26571.7 Q3 1112059 150137 443404 671389 41.1398 27290.8 Q4 1186440 133641 464258 630719 36.9918 29535.9 1999 Q1 1159803 119749 476324 628982 37.0998 29936.1 Q2 1108838 129285 431217 625559 37.2054 31433.9 Q3 1152229 144485 429386 692836 38.4051 32360.2 Q4 1216209 139522 516166 755931 38.8182 34780.6 2000 Q1 1231245 131830 499570 746880 37.7030 32283.9 Q2 1189978 130009 457647 735051 38.8022 32142.0 Q3 1212115 156021 468615 874782 41.0781 32249.8 Q4 1289393 139947 506770 930571 43.5539 32661.3 2001 Q1 1284700 136881 519127 837246 43.3145 32294.7 Q2 1257209 143445 505812 844237 45.6110 31611.5 Q3 1270065 164660 506510 857128 45.1024 32635.4 Q4 1321528 136131 534152 842139 44.3687 33048.4 2002 Q1 1355115 152036 560397 820135 43.6763 33614.5 Q2 1325184 145700 552827 833325 42.7687 36790.6 Q3 1343999 166374 563519 898844 42.0522 37652.0 Q4 1426345 139781 587629 946700 43.3824 38923.7 2003 Q1 1471707 146465 627072 946261 42.9705 37631.7 Q2 1424519 156390 607517 913278 42.2336 39327.1 Q3 1457881 179425 611538 979831 41.2460 40264.3 Q4 1563262 153722 651981 1047196 39.7335 42147.7 2004 Q1 1583692 161420 681915 1059950 39.1582 43036.4 Q2 1568023 183434 675812 1108574 40.2158 43306.1 Q3 1606091 199173 691135 1182636 41.2266 44767.5 Q4 1731670 176568 763277 1236708 40.2704 49831.7 2005 Q1 1716030 193301 776977 1147003 38.5887 48681.1 Q2 1691863 204321 763749 1244214 40.0950 48357.3 Q3 1780615 239439 756188 1430853 41.3042 49795.2 Q4 1904385 206588 794965 1396009 41.0472 52065.9 2006 Q1 1948891 220051 837441 1379062 39.3269 55265.8 Q2 1900243 230628 795231 1381043 38.1247 58057.4 Q3 1945831 262614 789179 1522271 37.6822 61592.7 Q4 2049974 212694 826019 1495178 36.5307 66984.8 2007 Q1 2096403 245357 845252 1470577 33.9916 70863.0 Q2 2047536 254724 823516 1473854 32.6376 72999.5 Q3 2107739 295134 831429 1575307 31.4757 80686.7 Q4 2273519 244055 880730 1739843 31.0903 87455.1 Economy, 2014, 1(1): 20-31 25 GDP G MB X e reserves 2008 Q1 2283347 259812 922963 1662721 31.0426 109970.3 Q2 2283267 268331 906861 1739789 32.2996 105676.2 Q3 2305387 314762 898575 1934054 33.8802 102421.6 Q4 2208465 277937 967818 1604962 34.8593 111008.0 2009 Q1 2199600 279823 971511 1449590 35.3349 116216.3 Q2 2196020 288743 975599 1392586 34.7276 120811.0 Q3 2246467 344407 969872 1616066 33.9632 131755.7 Q4 2399464 300955 1032435 1721810 33.3039 138417.6 2010 Q1 2560083 316260 1082431 1767053 32.8966 144094.1 Q2 2471448 315689 1086842 1764641 32.3844 146759.2 Q3 2490045 359799 1076872 1820548 31.6276 163235.3 Q4 2583245 318279 1141793 1851057 29.9859 172128.9 2011 Q1 2744960 334477 1244944 2058719 30.5442 181584.0 Q2 2652994 337697 1206049 2013730 30.2842 184894.3 Q3 2688075 397216 1217604 2220098 30.1460 180112.7 Q4 2454105 328140 1311562 1817403 31.0048 175123.8 2012 Q1 2798211 355744 1317609 2038340 30.9973 179247.8 Q2 2811487 368186 1337258 2098657 31.2885 174689.1 Q3 2801625 442386 1334503 2224797 31.3531 183627.2 Q4 2964026 378014 1409036 2167418 30.6748 181608.0 2013 Q1 2999658 371474 1435458 2148647 29.7952 177802.9 Q2 2955431 404666 1408188 2061985 29.8939 170841.1 Q3r 2924215 481986 1397179 2286810 31.4804 172286.3 Q4 3018145 384517 1492954 2255011 31.7307 167232.5 Sources: G, X, and GDP: National Economic and Social Development Board; Reserves and MB: The Bank of Thailand; e: http://fx.com Table-2.The Empirical Results: dGDP/ dG dMB dX de* dres 1993 Q1 4.82 upper lower 1.76 upper Lower 0.80 upper lower 23.26 upper lower 0.61 Upper Lower Q2 4.58 1.73 0.80 23.50 0.56 Q3 4.47 5.79 4.16 1.93 2.02 1.69 0.89 1.09 0.72 23.94 24.53 23.11 2.74 6.06 -0.75 Q4 5.38 6.00 4.23 1.92 2.02 1.74 0.94 1.12 0.77 24.06 24.66 23.45 3.97 6.47 0.41 1994 Q1 5.62 5.98 4.32 1.93 1.95 1.89 1.08 1.11 0.87 24.35 24.91 23.75 5.40 6.09 2.64 Q2 5.53 6.70 4.47 1.89 2.07 1.83 1.02 1.15 0.92 24.44 25.49 23.77 4.55 7.66 3.06 Q3 4.74 6.87 4.58 1.92 2.12 1.84 1.02 1.20 0.97 24.86 26.12 23.90 5.17 9.96 3.04 Q4 6.66 7.04 4.66 2.08 2.12 1.85 1.13 1.23 0.98 25.43 26.61 24.17 7.71 10.50 3.59 1995 Q1 6.08 7.04 4.59 2.06 2.11 1.90 1.18 1.23 1.03 25.97 26.61 24.80 9.69 10.40 5.12 Q2 6.23 7.26 5.26 1.98 2.10 1.95 1.17 1.21 1.12 26.26 26.53 25.44 8.13 10.07 7.18 Q3 5.36 7.14 5.26 1.98 2.07 1.94 1.14 1.27 1.12 25.99 26.44 25.89 8.11 10.31 7.70 Q4 6.99 7.28 5.33 2.04 2.07 1.94 1.20 1.39 1.08 26.26 26.57 25.97 9.48 10.58 7.61 1996 Q1 6.35 7.26 5.22 1.96 2.10 1.95 1.27 1.47 1.10 26.33 26.68 25.97 9.61 11.00 8.11 Q2 6.61 7.09 5.82 2.05 2.10 1.96 1.39 1.50 1.18 26.51 26.73 26.19 10.14 11.68 8.85 Q3 5.91 6.74 5.91 2.07 2.15 1.82 1.41 1.47 1.26 26.54 26.83 26.00 10.43 11.82 9.29 Q4 6.43 6.86 5.91 2.03 2.17 1.74 1.43 1.45 1.32 26.68 26.86 26.04 11.67 13.16 9.35 1997 Q1 6.35 6.78 5.71 1.82 2.12 1.73 1.33 1.53 1.12 26.01 29.40 21.22 10.93 15.09 9.30 Q2 6.65 7.19 5.77 1.80 2.08 1.75 1.36 1.56 0.90 26.52 30.37 16.46 13.09 18.25 9.16 Q3 5.89 7.33 5.81 1.93 2.03 1.76 1.10 1.47 0.76 20.79 29.52 12.49 14.86 19.36 10.16 Q4 7.09 7.33 5.79 2.00 2.03 1.75 0.93 1.36 0.71 17.06 26.70 11.50 17.97 18.65 12.15 1998 Q1 6.89 7.73 4.61 1.93 2.04 1.74 0.87 1.09 0.78 14.65 20.71 13.33 16.94 18.54 12.44 Q2 6.31 7.72 4.58 1.80 2.02 1.73 0.91 1.04 0.81 16.47 18.71 14.36 14.12 18.55 12.50 Q3 4.66 7.32 4.64 1.78 1.94 1.72 0.86 1.06 0.81 16.14 19.23 14.28 13.56 16.90 12.54 Q4 5.80 6.78 4.59 1.86 1.87 1.74 1.04 1.06 0.83 18.36 19.03 15.75 15.05 15.67 11.66 1999 Q1 6.25 6.43 4.46 1.76 1.94 1.72 1.00 1.06 0.83 18.16 19.01 16.19 13.95 15.45 11.31 Q2 5.39 6.36 4.98 1.82 1.95 1.69 0.92 1.05 0.85 17.83 18.47 17.33 11.67 15.47 11.32 Q3 5.13 6.55 4.95 1.93 1.94 1.69 0.90 1.00 0.87 17.50 18.40 17.36 12.68 15.54 11.29 Q4 5.77 6.41 4.99 1.73 1.97 1.72 0.91 0.94 0.88 17.65 18.25 17.25 13.63 15.54 11.29 2000 Q1 6.22 6.45 4.84 1.82 1.99 1.73 0.94 0.98 0.78 18.25 18.43 16.60 15.15 15.51 12.47 Q2 5.99 6.62 5.13 1.90 1.97 1.73 0.90 0.97 0.76 17.52 18.62 15.83 13.94 16.81 12.82 Q3 5.13 6.81 5.18 1.90 1.94 1.82 0.78 0.97 0.76 16.66 18.44 15.41 14.58 17.53 13.36 Q4 6.27 6.72 5.14 1.91 1.93 1.84 0.81 0.93 0.76 16.06 17.69 14.94 16.76 17.82 13.53 2001 Q1 6.38 6.72 4.84 1.85 1.92 1.83 0.90 0.92 0.77 16.13 16.84 14.95 16.80 17.59 14.66 Q2 5.89 7.00 5.18 1.85 1.91 1.83 0.86 0.95 0.80 15.20 16.41 15.08 16.30 17.58 15.85 Q3 5.21 6.98 5.17 1.87 1.88 1.84 0.86 1.01 0.81 15.43 16.56 15.04 16.18 18.41 15.61 Q4 6.68 6.94 5.17 1.87 1.89 1.81 0.94 1.02 0.82 15.91 16.79 14.96 17.53 18.46 15.18 2002 Q1 6.21 6.93 5.06 1.84 1.89 1.80 1.00 1.02 0.85 16.32 17.13 15.30 18.24 18.48 15.04 Q2 6.27 7.39 5.41 1.81 1.89 1.80 0.95 1.01 0.89 16.53 17.11 15.86 15.85 18.75 15.33 Q3 5.60 7.67 5.36 1.81 1.88 1.79 0.90 1.02 0.89 16.90 17.24 16.22 15.99 19.84 14.98 Q4 7.26 7.70 5.44 1.88 1.88 1.79 0.94 1.01 0.90 16.76 17.35 16.46 17.58 19.55 14.92 2003 Q1 7.24 7.72 5.24 1.83 1.89 1.79 0.99 1.01 0.90 17.13 17.79 16.52 19.39 19.58 15.66 Q2 6.48 7.97 5.75 1.81 1.92 1.80 0.98 1.01 0.93 17.21 18.98 16.17 17.35 19.98 16.65 Q3 5.83 7.97 5.75 1.86 1.91 1.80 0.95 1.01 0.95 17.78 19.76 16.43 17.78 20.42 17.00 Q4 7.49 7.78 5.56 1.90 1.91 1.80 0.98 1.01 0.93 18.99 19.88 16.97 19.48 20.29 17.01 2004 Q1 7.26 7.79 5.36 1.85 1.90 1.82 0.99 1.01 0.90 19.37 19.65 17.72 19.56 20.21 17.86 Q2 6.30 8.04 5.77 1.84 1.90 1.82 0.93 1.02 0.90 18.78 19.74 18.35 19.07 19.93 18.98 Q3 6.00 7.76 5.75 1.86 1.88 1.80 0.91 1.05 0.89 18.51 20.47 18.16 19.30 20.18 18.94 Q4 7.48 7.49 5.62 1.85 1.88 1.78 0.97 1.03 0.88 19.57 20.49 18.17 19.86 20.20 18.95 2005 Q1 6.75 7.56 5.32 1.79 1.93 1.75 1.03 1.04 0.84 20.34 20.50 18.38 20.01 20.87 19.00 Q2 6.27 7.84 5.54 1.79 2.01 1.73 0.93 1.05 0.86 19.46 20.44 19.05 19.64 22.31 18.76 Q3 5.72 7.56 5.62 1.93 2.04 1.74 0.87 1.08 0.86 19.32 21.25 18.87 20.86 22.79 19.08 Q4 7.23 7.49 5.57 1.99 2.06 1.80 0.98 1.06 0.86 20.04 21.92 18.68 22.32 22.79 19.05 http://fx.com/ Economy, 2014, 1(1): 20-31 26 dG dMB dX de* dres 2006 Q1 6.99 7.53 5.36 1.94 2.06 1.90 1.03 1.06 0.86 21.14 22.63 19.01 21.84 22.82 18.99 Q2 6.45 8.00 5.68 1.98 2.11 1.92 0.99 1.05 0.93 21.56 23.56 19.68 19.95 22.80 18.47 Q3 5.84 7.88 5.66 2.06 2.15 1.92 0.93 1.09 0.92 22.05 25.41 19.95 19.54 21.72 18.26 Q4 7.70 7.77 5.55 2.09 2.14 1.99 1.02 1.09 0.92 23.32 26.87 20.46 19.53 20.48 17.92 2007 Q1 6.87 7.81 5.22 2.10 2.15 2.05 1.06 1.09 0.93 25.33 28.32 21.37 19.11 20.46 16.74 Q2 6.42 8.21 5.54 2.10 2.22 2.04 1.03 1.06 0.98 26.08 29.58 22.82 17.88 19.96 16.43 Q3 5.75 7.94 5.60 2.15 2.22 2.06 1.00 1.08 0.98 27.43 29.93 24.72 16.93 20.15 14.03 Q4 7.63 7.97 5.62 2.22 2.22 2.08 1.00 1.05 0.98 28.84 29.71 25.94 17.51 19.03 13.34 2008 Q1 7.20 8.01 5.42 2.13 2.23 2.11 1.05 1.08 0.92 28.94 29.50 26.37 14.02 18.09 13.23 Q2 6.97 7.87 5.84 2.16 2.31 1.96 1.01 1.09 0.92 27.82 29.98 25.05 14.58 17.55 12.28 Q3 6.02 7.38 5.84 2.21 2.28 1.87 0.92 1.17 0.90 26.65 29.43 24.05 15.26 15.66 12.18 Q4 6.46 6.99 5.82 1.95 2.26 1.81 1.04 1.24 0.88 25.34 27.94 24.11 13.21 15.73 11.33 2009 Q1 6.39 6.81 5.34 1.93 2.19 1.81 1.15 1.25 0.90 24.95 26.85 24.57 12.54 15.28 10.51 Q2 6.18 7.01 5.37 1.92 2.02 1.90 1.20 1.21 1.00 25.37 27.67 24.16 12.04 13.32 11.15 Q3 5.33 7.18 5.34 1.98 2.08 1.89 1.06 1.22 1.04 26.24 29.35 23.94 11.42 12.90 11.33 Q4 6.60 7.22 5.35 2.01 2.08 1.91 1.08 1.20 1.03 27.67 30.03 24.84 11.97 12.67 11.26 2010 Q1 6.79 7.21 5.20 2.07 2.07 1.95 1.15 1.15 1.04 28.99 30.51 26.10 12.62 12.84 10.56 Q2 6.52 7.19 5.82 1.98 2.07 1.95 1.10 1.15 1.06 28.91 32.01 26.90 11.78 12.91 10.20 Q3 5.78 7.35 5.81 2.01 2.07 1.92 1.08 1.15 1.05 29.71 33.11 27.70 10.71 12.70 10.03 Q4 6.82 7.30 5.80 1.98 2.02 1.92 1.11 1.12 1.05 31.96 33.61 28.44 10.70 11.79 10.03 2011 Q1 6.98 7.34 5.43 1.95 2.02 1.91 1.08 1.14 0.97 32.44 33.61 29.88 11.03 11.12 10.31 Q2 6.64 7.29 5.67 1.93 2.12 1.65 1.05 1.13 0.97 32.11 33.42 30.21 10.34 11.32 9.74 Q3 5.73 7.24 5.67 1.94 2.09 1.64 0.97 1.14 0.97 32.49 33.53 30.24 10.80 11.73 9.63 Q4 6.22 7.01 5.72 1.63 2.06 1.64 1.06 1.14 0.97 30.08 33.40 30.23 9.78 12.20 9.50 2012 Q1 6.71 6.99 5.24 1.88 2.03 1.64 1.11 1.14 0.96 32.31 33.36 30.21 11.47 12.30 9.74 Q2 6.52 7.21 5.43 1.86 2.00 1.64 1.09 1.15 1.01 32.09 34.13 29.94 11.85 12.82 9.79 Q3 5.40 7.47 5.47 1.86 1.88 1.85 1.02 1.18 1.02 31.96 35.11 30.91 11.22 12.85 10.94 Q4 6.75 7.36 5.41 1.88 1.88 1.86 1.12 1.21 1.01 33.73 35.69 31.23 12.23 13.31 11.07 2013 Q1 6.97 7.39 4.85 1.87 1.88 1.86 1.15 1.21 1.00 34.96 35.61 31.52 12.70 13.46 11.25 Q2 6.29 7.54 5.26 1.87 1.90 1.81 1.18 1.20 1.04 34.55 35.38 32.14 12.96 13.65 12.02 Q3r 5.21 1.86 1.05 32.61 12.67 Q4 6.78 1.81 1.10 32.95 13.61 *for the dGDP/de results, GDP was measured in units of 5 billion baht. This was done so that Figure 5 would be easier to understand. As shown in Table 2, dGDP/dMB was 2.03 in the fourth quarter of 1996, immediately prior to George Soros’ speculative attack against the Thai baht in the first half of 1997. This value means that for every one million increase in the monetary base for Thailand, GDP increased by 2.03 million. This multiplier had fallen to 1.78 by the third quarter of 1998 for a 12.3 percent decline. The bottom line in Figure 2 shows that the monetary base remained unusually constant between the third quarters of 1997 and 1999. This was due to the IMF conditions imposed on Thailand when Thailand accepted an IMF bailout in August 1997. By 2000, the monetary base began to steadily increase. The upper lines in Figure 2 show that the dGDP/dMB multiplier was unusually stable during most of the Thaksin administration (2001-2006), rose noticeably in the last year of Thaksin’s administration (between the third quarters of 2005 and 2006), continued to rise after the coup that displaced Thaksin and during the pro-Thaksin Samak/Somchai regime. Between the second quarter of 2005 and the third quarter of 2008, dGDP/dMB rose by 23.5 percent. Under the Democrats (first quarter of 2009 – second quarter of 2011), dGDP/dMB fell and became more unstable. The economic policies of the Thaksin regime were more pro-equality than that of the Democrats. A shift away from pro-equality policies will decrease the marginal propensity to consume (MPC) which should decrease all government multipliers. Soon after Thaksin’s sister, Yingluck, was elected in July 2011, dGDP/dMB fell for one quarter, but then stabilized. Figure-2. dGDP/dMB Economy, 2014, 1(1): 20-31 27 Leightner (2015) found that, for the USA and Japan (between1980 and 2012), increases in the money supply were associated with declines in the money multiplier and that decreases in the money supply were associated with increases in the money multiplier. However, the Thai results do not produce a similar pattern. Indeed a comparison of the upper and bottom lines in Figure 2 implies that the variations found in dGDP/dMB were not due to noticeable changes in the growth of the monetary base; instead these variations must be due to other factors like monetary policy interacting with other government policies or due to expectations. Figure 3 and Tables 1 and 2 show that both government consumption and the government consumption multiplier, dGDP/dG, exhibited an annual pattern where government consumption increased in the third quarter and dGDP/dG fell in the third quarter. This implies that cuts in government consumption (in the fourth quarter) have a stronger negative effect on GDP that increases in government consumption (in the third quarter) have a positive effect. Leightner (2015) found a similar pattern for Austria, Brazil, Cyprus, Estonia, Greece, Luxembourg, Portugal, and Slovakia in recent years. For example, in Thailand when government consumption increased by 74.2 billion baht in the third quarter of 2012 (442.4 – 368.2), dGDP/dG was 5.40, implying that GDP rose by 400.68 billion baht (74.2 x 5.40). However, when government consumption was cut in the fourth quarter of 2012 by 64.4 billion baht (378.0 – 442.4), dGDP/dG was 6.75 implying a decline in GDP of 434.7 billion baht (6.75 x 64.4). Since 6.75 is twenty five percent bigger than 5.40, the fourth quarter decline in government consumption caused GDP to fall by 25 percent more than an equal increase in government consumption would cause GDP to rise. Government consumption would be a much more effective tool if government consumption was smoothed throughout out the year instead of spiking in the third quarter and then being cut in the fourth quarter to reduce the deficit. Figure-3. dGDP/dG If we eliminate this annual pattern by taking the annual average value for dGDP/dG, we find that annual dGDP/dG steadily grew from 4.81 in 1993 to 6.49 in 1997 for a 35 percent increase. However, after the baht collapsed on July 2, 1997 and Thailand took out the IMF loan in August 1997, annual dGDP/dG fell to 5.91 in 1998 and continued to decline to 5.63 in 1999 for a 13.3 percent decline. During the first half year of the Thaksin regime (quarters 1 and 2 of 2001) there was a pending constitutional court case against Thaksin for corruption, and this pending court case constrained what the Thaksin administration could do. The annual dGDP/dG for 2001 was 6.04. Between 2002 and 2008 (the remaining Thaksin years, the year of the coup, and the pro-Thaksin Samak/Somchai regime), annual dGDP/dG ranged between 6.33 and 6.77. However, annual dGDP/dG fell to 6.13 in 2009, the first year under the next Democrat regime. The Democrats in 2010 started to implement more pro-equality programs than they previously had and annual dGDP/dG rebounded to 6.48. Again Keynesian theory would predict a higher multiplier under pro-equality regimes (ceteris paribus) because increasing equality increases the marginal propensity to consume. Under Yingluck, Thaksin’s sister, annual dGDP/dG slightly declined from 6.39 to 6.31. Table 2 and Figure 4 show the dGDP/dX results. Notice that dGDP/dX for Thailand is always much lower than dGDP/dG. Furthermore, dGDP/dX for Thailand was often less than one meaning that a one million baht increase in exports increased GDP by less than one million baht. From the perspective of a simple Keynesian model, the Thai results are surprising. Since GDP = C + I + G + X – M (where C = private consumption, I = Investment, G = Government Consumption, X = exports, and M = imports), one would expect dGDP/dX to equal dGDP/dG (because both G and X are added into GDP in the same way). There are several reasons why dGDP/dX would be less than dGDP/dG in a more complex Keynesian model. Government consumption is often done for the benefit of the country in contrast to exports which are done for the Economy, 2014, 1(1): 20-31 28 benefit of private citizens and companies. Thus government consumption is more likely to be pro-equality while exporting is often correlated with increased inequality. Greater equality causes the marginal propensity to consume to raise causing government policy multipliers to increase. Many Asian countries, especially prior to 2008, embraced an export driven growth strategy which explicitly entailed greater inequality. For example China (prior to 2008) embraced an export driven growth model that employed a suppressed exchange rate, which Leightner (2015) shows hurts consumers and helps producers, and the suppression of wages which further hurts consumers. Please understand that we are NOT anti-trade. There are real gains from trade. However, many countries have tried to increase those gains by employing market distortions, like suppressed wage rates and exchange rates, and these distortions are no longer able to drive sustainable growth in today’s world. The remainder of section III will explore the issues dealing with trade in more depth. Figure-4. dGDP/dX Figure-5. dGDP/de Figure 5 shows the dGDP/de results where “e” is the baht/US dollar exchange rate and GDP is in units of five billion baht. The dGDP/de value of 23.26 in the first quarter of 1993 means that for every one baht increase in the baht per US dollar exchange rate, Thailand’s GDP increased by 116.3 billion baht (5 billion x 23.26). Notice that an increase in “e” is a depreciation of the baht and a decrease in “e” is an appreciation of the baht. Thailand’s dGDP/de steadily increased from 116.3 billion baht in the first quarter of 1993 to 132.6 billion baht in the second quarter of 1997. The top line in Figure 5 shows that when the Thai baht was floated on July 2, 1997, the exchange rate (baht/dollar) immediately rose to 32.62 baht/dollar and kept rising until it hit 46.68 baht/dollar in the first quarter of 1998 (if we were to examine daily values of the baht, instead of quarterly values, the baht hit a high of 55.5 baht/dollar). This depreciation of the baht (dollars/baht falling or baht/dollar rising) is correlated with dGDP/de Economy, 2014, 1(1): 20-31 29 falling from 132.6 billion baht in the second quarter of 1997 to 73.25 billion baht in the first quarter of 1998 for a 44.8 percent decline. This means that the immediate boost to GDP in 1997-1998 from the depreciation of the baht was 44.8 percent less than what the Thais expected based on data immediately prior to the fall. Figure 5 also shows that dGDP/de steadily rose during the Thaksin years (2001-2006) and accelerated its increase in the year of the coup (2007) and during the pro-Thaksin Samak/Somchai regime. However, when the Democrats took over in 2008, dGDP/de initially fell but rebounded in 2009. During the Yingluck years (2011-2013), dGDP/de was the highest it had been since 1993, but was also slightly more volatile. Notice that when the baht/dollar exchange rate increases, dGDP/de falls, and that when the baht/dollar exchange rate decreases, dGDP/de rises. This means that the positive effects from the baht/dollar exchange rate rising are weaker than the negative effects of the baht/dollar exchange rate falling. For example when the exchange rate went from 31.04 baht/dollar in the first quarter of 2008 to 35.33 baht/dollar in the first quarter of 2009, dGDP/de went from 144.7 billion baht to 124.75 billion baht for a 20 percent decline. In contrast, when the exchange rate went from 35.33 baht/dollar in the first quarter of 2009 to 29.99 baht/dollar in the fourth quarter of 2010, dGDP/de rose from 124.75 to 159.8 billion baht for a 28 percent increase. Therefore appreciations of the baht (baht/dollar falling) hurt the Thai economy more than equal depreciations of the baht help. This asymmetric affect could be because negative news has a stronger effect on expectations than positive news has. Many countries that embrace export driven growth models suppress their exchange rates below market clearing levels by printing more of their currencies and then exchanging their currencies for US dollars, US treasury bills, or other foreign currencies. When a country accumulates US dollars (for example) in order to drive down their exchange rate, they must forever hold those dollars, because if they ever use those dollars, then the value of the dollar would fall causing the value of their own currencies to rise, ceteris paribus. Other countries wanting to accumulate US dollars, which they never use, has been a wonderful deal for the US – the US gets imports in exchange for printed dollar bills that never get used. It is like the world giving the US loan after loan and never asking to be repaid. [However, Leightner (2015) discusses how China (who now holds almost 4 trillion dollars of foreign reserves as of April 2014) now has significant leverage against the USA because China can threaten to dump those reserves, of which more than ½ are US dollars.] This “accumulating US dollars to suppress the exchange rate and increase exports” deal “might” be worth it for the accumulating countries if dGDP/d(foreign reserves) exceeds the cost of the foreign reserves. We emphasize the word “might” because a complete analysis would need to analyze the opportunity costs involved. However, it is safe to say that if the resulting increase in GDP does not exceed the cost of the reserves, then accumulating the foreign reserves were not worth it as a growth strategy (however, it might be worth it as a strategy to reduce the risk of speculative attacks). Table 2 and Figure 6 depict the dGDP/d(reserves) results for Thailand where GDP is measured in millions of baht and foreign reserves are measured in millions of US dollars at the end of the quarter. Since dGDP/d(reserves), as shown in Figure 6, never exceeds the baht/dollar exchange rate, as shown in Figure 5, Thailand’s accumulation of foreign reserves were not worth it as a growth strategy. In other words, for every dollar of reserves that Thailand accumulated, Thailand’s GDP increased by much less than a dollar. Thailand gave the US a dollar worth of imports for printed paper (that it is illogical for Thailand to ever cash in under an export driven growth model) but Thailand got less than a dollar’s increase in GDP. Figure-6. dGDP/d (Foreign Reserves) Figure 6 shows Thailand’s official holdings of foreign reserves (in units of 10 billion dollars); however, these official holdings of foreign reserves do not depict the true drop in foreign reserves that occurred in 1997. At the beginning of 1997 Thailand had 38.7 billion dollars of foreign reserves. When George Soros launched a speculative Economy, 2014, 1(1): 20-31 30 attack against the Thai baht in the first half of 1997, the Thai government defended its fixed exchange rate by buying up surplus baht on the market. However, the Thai government did not want the world to know how much of its foreign reserves it was using (because that knowledge would have caused more foreigners to sell their baht) thus, the Thai government hid its defense of the Thai baht by using forward contracts, the effects of which do not immediately show up in the official foreign reserves accounts. Actually, in the first half of 1997, Thailand spent approximately 35 billion dollars of foreign reserves (if the forward contracts are included) defending the baht (recall that in January 1997, Thailand only had 38.7 billion dollars of foreign reserves). Figure 6 shows that during the Thaksin regime (2001-2006) Thailand’s holdings of foreign reserves steadily increased at a relatively slow rate resulting in Thailand’s foreign reserves of 32.3 billion dollars in the first quarter of 2001 increasing to only 58.1 billion dollars by the second quarter of 2006 for an 80 percent increase in six years. In contrast, in the six years between the coup that displaced Thaksin (3 rd quarter of 2006) and when his sister, Yingluck became prime minister (3 rd quarter of 2011), Thai holdings of foreign reserves increased from 58.1 billion dollars to 184.9 billion dollars for a 218 percent increase. Furthermore, during the Thaksin six years, dGDP/d(reserves) rose from 16.8 to 19.95 for an 18.8 percent increase. In contrast, during the six years after the Thaksin regime, dGDP/d(reserves) fell from 19.95 to 10.34 for a 48.2 percent decline. Figure 6 also shows that when Yingluck, Thaksin’s sister, took over the Thai government she stopped accumulating foreign reserves and dGDP/d(reserves) began to rebound. As Leightner (2015) shows, accumulating foreign reserves to suppress an exchange rate in order to increase exports also increases the degree of inequality which reduces all government multipliers. Thus, Thaksin and Yingluck not accumulating massive amounts of foreign reserves during their regimes was consistent with their pro-equality agendas. According to the Keynesian model, government multipliers are higher when the marginal propensity to consume (MPC) is higher and the MPC increases under government policies that increase equality. 4. Conclusion For a country to grow, there must be production increasing investment. Investment requires two things – savings to fund the investment and the reasonable belief that what investment produces will sell (Leightner, 2015). Many of our growth models emphasize the role of savings to fund investment while ignoring the role of consumption in providing a reason to invest. Leightner (2015) shows the world is currently suffering from a surplus of saving, which is currently either (1) sitting idle, (2) seeking a return from deception or rent, or (3) funding speculative bubbles and these three things are happening because there is insufficient consumption to justify investing that saving in the expansion of production. Thailand suffered from a speculative bubble between 1993 and 1996, this bubble gave George Soros the evidence he needed to successfully conduct a speculative attack against the Thai baht. The collapse of the Thai baht was devastating to the Thai economy. The political rise of Thaksin Shinawatra was a direct consequence of the collapse of the Thai baht and of the conditionality imposed by the subsequent IMF bailout loan. The Thaksin regime used a pro-equality strategy which is what is needed in a world suffering from a surplus of savings. The empirical results of this paper show that Thaksin’s pro-equality economic policies were appropriate and good for the country. Unfortunately, Thaksin was corrupt, power hungry, condoned the killing of drug dealers and Muslims, and undercut the institutions needed for democracy. The results of this paper are consistent with what Leightner (2015) found for the seventeen countries using the euro, the USA, the UK, Japan, Brazil, and Russia. Leightner (2015) shows that pro-equality policies are the solution to the current world-wide economic malaise and that export driven growth will no longer work in our current world. We appreciate the research help given by Yumiko Deevey and of Kunihiro Fujio. References Abbott, J.K. and K.H. Allen, 2011. An embarrassment of riches: Confronting omitted variable bias and multiscale capitalization in hedonic price models. The Review of Economics and Statistics, 93(4): 1331-1342. Available from http://www.mitpressjournals.org/doi/abs/10.1162/REST_a_00134. Aitken, A.C., 1935. On least squares and linear combinations of observations. Proceedings of the Royal Society of Edinburgh: 55: 42–48. Angrist, J.D. and K.B. Alan, 2001. Instrumental variables and the search for identification: From supply and demand to natural experiments. Journal of Economic Perspectives, 15(4): 69-85. Black, S.E. and L.M. Lisa, 2001. How to compete: The impact of workplace practices and information technology on productivity. The Review of Economics and Statistics, 83(3): 434-445. Botosan, C.A. and P.A. Marlene, 2002. A re-examination of disclosure level and the expected cost of equity capital. Journal of Accounting Research, 40(1): 21-40. Branson, J. and L.C.A. Knox, 2000. Taxation and economic growth in New Zealand. In taxation and the limits of government. Ed. G.W. Scully and P.J. Caragata. Boston: Kluwer Academic. pp: 37-88. Cellini, S.R., 2008. Causal inference and omitted variable bias in financial aid research: Assessing solutions. The Review of Higher Education, 31(1): 329-354. DiPrete, T., A. and G. Markus, 2004. Assessing bias in the estimation of causal Effects: Rosenbaum bounds on matching estimators and instrumental estimation with imperfect instruments (Berlin, Wissenschaftszentrum Berlin fur Sozialforschung). Harris, A.l.L. and R. Keith, 2007. Schooling behaviors or prior skills? A cautionary tale of omitted variable bias within oppositional culture theory. Sociology of Education, 80(2): 139-157. Inoue, T., L.D.M. Pierre and L.E. Jonathan, 2014. Several related solutions to the omitted variables problem, (Unpublished Manuscript). Leightner, J.E., 2002. The changing effectiveness of key policy tools in Thailand, Institute of Southeast Asian studies for East Asian development network, EADN Working Paper No. 19(2002) x0219-6417. Leightner, J.E., 2007b. Thailand’s financial crisis: Its causes, Consequences, and implications. Journal of Economic Issues, 41(1): 61-76. Leightner, J.E., 2008. Omitted variables and how the Chinese Yuan affects other Asian currencies, International Journal of Contemporary Mathematical Sciences, 3(14): 645-666. Leightner, J.E., 2015. The limits of fiscal, monetary, and trade policies: International comparisons and solutions. Singapore: World Scientific. Leightner, J., E., 2010a. China’s fiscal stimulus package for the current international crisis, What does 1996-2006 tell us? Frontiers of Economics in China, 5(1): 1-24. Leightner, J., E. and I. Tomoo, 2008a. Capturing climate’s effect on pollution abatement with an improved solution to the omitted variables problem. European Journal of Operational Research, 191(2): 539-556. Leightner, J.E., 2005a. Fight deflation with deflation, not with monetary policy. The Japanese Economy: Translations and Studies, 33(2): 67- 93. http://www.mitpressjournals.org/doi/abs/10.1162/REST_a_00134 Economy, 2014, 1(1): 20-31 31 Leightner, J.E., 2005b. The productivity of government spending in Asia: 1983-2000. Journal of Productivity Analysis, 23(1): 33-46. Leightner, J.E., 2007a. Omitted variables, confidence intervals, and the productivity of exchange rates. Pacific Economic Review, 12(1): 15-45. Leightner, J.E., 2010b. How China’s holdings of foreign reserves affect the value of the US dollar in Europe and Asia. China & World Economy, 18(3): 24-39. Leightner, J.E., 2011a. Chinese overtrading, In steven rosefielde, Masaaki Kuboniwa, and Satoshi Mizobata Eds., Two Asias: The Emerging Postcrisis Divide, (Singapore: World Scientific Publishers). Leightner, J.E., 2011b. Fiscal stimulus for the USA in the current financial crisis: What does 1930-2008 tell us? Applied Economic Letters, 18(6): 539-549. Leightner, J.E., 2013. The changing effectiveness of monetary policy. Economies, 1(3): 49-64. Available from http://www.mdpi.com/2227- 7099/1/3/49. Leightner, J.E. and I. Tomoo, 2007. Tackling the omitted variables problem without the strong assumptions of proxies. European Journal of Operational Research, 178(3): 819-840. Leightner, J.E. and I. Tomoo, 2008b. The effect of the Chinese Yuan on other Asian currencies during the 1997-1998 Asian crisis. International Journal of Economic Issues, 1(1): 11-24. Leightner, J.E. and I. Tomoo, 2009. Negative fiscal multipliers exceed positive multipliers during Japanese deflation. Applied Economic Letters, 16(15): 1523-1527. Leightner, J.E. and I. Tomoo, 2012a. Is China replacing the USA as an engine for global growth? International Economics & Finance Journal, 7(1): 55-77. Leightner, J.E. and I. Tomoo, 2012b. Solving the omitted variables problem of regression analysis using the relative vertical position of observations. Advances in Decision Sciences. Available from http://www.hindawi.com/journals/ads/2012/728980/. Mustard, D.B., 2003. Reexamining criminal behavior: The importance of omitted variable bias. The Review of Economics and Statistics, 85(1): 205-211. Pace, K.R. and L.P. James, 2010. Omitted variable biases of OLS and spatial lag models. Progress in spatial analysis: Advances in Spatial Science Part 1: 17-28. Available from http://www.springerlink.com/content/p518x94h58261u59/abstract/?target=print. DOI 10.1007/978-3-642-03326-1_2. Paterson, R.W. and B.J. Kevin, 2002. Out of sight, out of mind? Using GIS to incorporate visibility in hedonic property value models. Land Economics, 78(3): 417-425. Scheffler, R.M., B.T. Timothy and R.K. Jennifer, 2007. The role of social capital in reducing non-specific psychological distress: The importance of controlling for omitted variable bias. Social Science & Medicine, 65(4): 842-854. Sessions, D.N., Lonnie and K. Stevans, 2006. Investigating omitted variable bias in regression parameter estimation: A genetic algorithm approach. Computational Statistics & Data Analysis, 50(10): 2835-2854. Streams, S.C. and N.C. Edward, 2004. Time to include time to death? The future of health care expenditure predictions. Health Economics, 13(4): 315-327. Swamy, P.A.V.B., I.-L. Chang, M. Jatinder, S. and T.S. George, 2003. Correcting for omitted-variable and measurement-error bias in autoregressive model estimation with panel data. Computational Economics, 22(2): 225-253. Views and opinions expressed in this article are the views and opinions of the authors, Economy shall not be responsible or answerable for any loss, damage or liability etc. caused in relation to/arising out of the use of the content. http://www.mdpi.com/2227-7099/1/3/49 http://www.mdpi.com/2227-7099/1/3/49 http://www.hindawi.com/journals/ads/2012/728980/ http://www.springerlink.com/content/p518x94h58261u59/abstract/?target=print