Thai Monetary Policy Under the Inflation Targeting Regime


 This paper explores the macroeconomic effects of inflation targeting in Thailand. Furthermore, this study uses a nonlinear new Keynesian model under the dynamic stochastic general equilibrium framework with price indexation to analyze the monetary policy under inflation targeting in Thailand. The model is estimated using a Bayesian statistic for the Thai economy. It shows that inflation is more stabilized and inflation persistence has fallen after adopting inflation targeting. The paper also indicates that the Bank of Thailand is more responsive to the deviation of inflation from its target using inflation targeting. The key monetary mechanism exists through changes in the real interest rate which affect aggregate demand. It is worth noting that the larger the inflation targeting rate is, the lower the steady state output from its steady state level given no trend inflation.


Introduction
A nominal anchor is widely agreeing to be a way to achieve the primary objectives of monetary policy which are to maintain low and stable inflation and ultimately to stabilize output around potential. Establishing a credible nominal anchor delivers price stability because it attempts to pin down inflation expectations. Since early 1990 inflation targeting has been adopted by an increasing number of countries, including Thailand in 2000. The rational for this rising adoption is that inflation targeting provides conditions for a credible monetary policy. It creates a condition for a credible medium term anchor for inflation expectations. More importantly, inflation targeting results in independent monetary policy and by implication flexibly responds to short run shocks and keeps output at potential. As Mishkin (1999) argued inflation targeting has a key advantage that it is readily understood by the public and enables monetary policy to respond to shocks to domestic economy.
The attempt of economists to investigate interactions between the real economic performance at the aggregate level and monetary policy gives rise to a framework for monetary policy analysis. Gali (2008)  This paper is structured as follows. Section 2 reviews the monetary policy framework in Thailand. Section 3 describes the effects of inflation targeting on Thai macroeconomic performance. Section 4 introduces the nonlinear new Keynesian model with price indexation and price dispersion. Section 5 discusses the rational of the Bayesian estimation in a DSGE model. Section 6 conducts an empirical analysis based on zero inflation at the steady state, describing transmission mechanisms and impulse responses of monetary policy rules and exogenous shocks. Section 7 mentions how changes in inflation targeting influence the Thai economy. Section 8 discusses policy implications. Section 9 provides conclusions.

Monetary policy framework in Thailand
Targeting the exchange rate was the Thai monetary policy strategy from November 1984 to June 1997. The fixed exchange rate regime in Thailand involved fixing the baht to the U.S. dollar to support long term economic growth. Nonetheless, in the early 1990s the Thai financial market was liberalized and opened to the foreign capital market, leading to a lending boom, loan losses and banks' balance sheets deteriorated. The Thai baht had been overvalued for several years before 1997 due to a large current account deficit. The floating exchange rate system was then adopted on June 1997 and Thailand was under the IMF program. The monetary policy strategy after the adoption of a floating exchange rate was a monetary target by which Bank of Thailand targeted domestic money supply to control the fluctuation of interest rates and to stabilize prices and to lead to sustainable growth eventually. Under monetary targeting, a strong relationship between the money aggregate and goal variability such as inflation is necessary in achieving price stabilization. However, that relationship became inconsistent over time.
Therefore, the Bank of Thailand, after the IMF exit, decided to switch its monetary policy to a flexible inflation targeting regime in May 2000. The Bank of Thailand then adopted the policy rate as an operational tool in maintaining the target rate of inflation. The primary goals are to ensure sustainable growth and to maintain price stability. The Bank of Thailand anticipated that the inflation targeting regime would rebuild confidence and its credibility. The core inflation targeting was the Thai monetary anchor after May 2000. The core inflation excludes fresh food and energy prices which are highly volatile in the short run. The core inflation targeting results in low volatility of the target measure and emphasizes inflationary pressures coming from the demand side. Movement in overall prices can also be described by changes in core inflation. Although the headline inflation could deviate from the core in the short run, they move closely in the long run.
The Bank of Thailand adopted the target range for quarterly average core inflation from 0 to 3.5 percent from May 2000 to August 2009. The reason was that the ceiling of 3.5 percent was close to headline inflation of Thailand's main trading partners and competitors. The inflation targeting during this period therefore focused on export competitiveness with a sufficiently wide target range for the monetary policy to deal with any temporary shocks. After August 2009 the Bank of Thailand set a narrower range of inflation targeting from 0.5 to 3.0 percent. To reduce the chance of deflation , the lower bound was adjusted upward. The objectives of the low and narrow range were to build up confidence of consumers and businesses and ultimately to support sustainable growth.
However, the new target was set for an annual average of headline inflation at 2.5 percent with a tolerance band of ±1.5 percent in 2015. The rationales behind this new target are firstly the core inflation compared to the past failed to reflect the headline inflation for a much longer time period. Secondly, the Bank of Thailand wanted to improve its communication to the public on monetary policy decisions. The headline inflation reflected changes in the cost of living because it accounts for changes in prices of all goods and services in the CPI basket. Cost of living was crucial for consumption and the saving decisions by households and for investment and price setting by firms. The public was able to recognize what composed the cost of living under headline inflation targeting. The point target then enhances monetary policy effectiveness in anchoring long term inflation expectations. At the same time, the range band is wide enough to provide a cushion for temporary shocks.
Recently, at the end of December 2019, the Bank of Thailand decided to set the inflation range target of 1.0 -3.0 percent for the medium-term horizon and for the year 2020. This range target adoption is to support the potential growth and to reflect the changing inflation dynamics from technological progress and the transition into an aging society.

Thai macroeconomic effects of inflation targeting
One critical issue is whether the adoption of an inflation targeting framework affects Thai macroeconomic performance. This section attempts to survey Thai facts by investigating changes in Thai inflation, growth and interest rates prior and after the adoption of the inflation targeting regime. Figure 1 illustrates Thai inflation during the inflation targeting regime. Corbo et al. (2001) argue that managing inflation to hit the target is not a proper way to measure the success of inflation targeting. The primary responsibility of the inflation targeting framework is to bring down the level of inflation to an appropriate level for longer term price stability. Schaechter et al., (2000) argue that a central bank is able to maintain its credibility with temporary deviations from the target.
Inflation targeting in Thailand started from 2000 and has proved largely successful in bringing down inflation. Table 1 shows the average level of inflation during monetary and inflation targeting. The table also reports average inflation using three different types of inflation targeting for comparison. Inflation went from 3.8 percent on average prior to adoption to approximately 2.1 percent after inflation targeting. Table 1 also compares inflation in different types of inflation targeting. Inflation went from 2.6 percent under core inflation targeting with a wide range to 2.5 percent under core inflation targeting within a narrow range. After the adoption of headline inflation targeting, the inflation level has been brought down significantly to 0.3 percent on average. Pétursson (2005) conclude from 21 inflation targeting countries that inflation targeting leads on average to a 2.5 percent to more than 3 percent fall in inflation.
One important question arises about how inflation targeting contributes to stabilizing inflation or reducing inflation fluctuations. It is obvious from Table 1 that for Thailand the volatility of inflation has been reduced after inflation targeting. The fluctuations went from 4.4 prior to adoption to 1.9 after inflation targeting. During the inflation targeting era, it is also clear that inflation volatilities have decreased. The fluctuation in headline inflation targeting is at 0.8. These results are consistent with findings in Mishkin (2003) and Neumann and von Hagen (2002) which argue that inflation targeting stabilizes inflation.
It is interesting to see how inflation targeting affects the inflation persistence. The univariate AR(2) model in line with Pétursson (2005) is used to find how inflation targeting changes inflation persistence in Thailand. The parameter is a constant term, is the inflation at time , −1 is the inflation at time − 1, −2 is the inflation at − 2 . 1 , 2 and are the coefficient. The variable is a dummy for after the adoption of inflation targeting = 1. Falling average inflation is described by the second order polynomial trend, ( ) where ( ) = 2 and is the error term at .
Given the structure of the model, the persistency of inflation before targeting is explained by 1 + 2 and by 1 + 2 + after targeting. 2 The inflation persistence has reduced if 2 It is necessary to have enough data to estimate a reliable persistence. Thai data is divided only into two periods before and after targeting. Quarterly data from 1981q1 to 2000q1is for before targeting and from 2000q2 to 2019q1 for after targeting. The coefficients 1 = 1.307 , 2 = −0.473 and = −0.123 are all statistically significant at the 1 percent level. For a group of inflation targeting countries, inflation targeting has reduced inflation persistence (See Pétursson (2005)).
is statistically negative. As noted by Kuttner and Posen (1999), because inflation targeting is a monetary policy anchor to combat an increase in inflation, inflation should exhibit less persistence given a temporary price shock. Therefore, the issue on inflation persistence involves the credibility of monetary policy. The credibility of monetary policy has increased with the reduction in inflation persistence. In Thailand, table 1 documents that inflation persistence has fallen from 0.83 to 0.71 after adopting inflation targeting. The result indicates that the credibility of Thai monetary policy has been improved after adopting inflation targeting. The inflation targeting is likely to have a positive effect to the behavior of expected inflation. The medium-term inflation expectation over the pre-inflation targeting period was higher than that of the inflation targeting period. Buddhari and Chensavasdijai (2003) points that inflation expectations appear to become more firmly anchored to lower inflation following the change in the regime in 2000. The inflation targeting framework has convinced businesses and consumers that the Bank of Thailand will successfully resist any persistent movements of inflation from the target band.
Regarding the effectiveness of inflation targeting on the real economy and business cycle fluctuation, one may doubt that this anchor is sufficiently flexible and could obstruct an economic expansion as in Firedman and Kuttner (1996). However, Mishkin (1999) points out that inflation targeting will eventually benefit growth and provide a favorable growth record for many inflation targeting countries after the adoption. Truman (2003) andBall andSheridan (2003) also note that the inflation targeting regime has no harm on growth. As reported in Table 1, average real GDP growth in Thailand during the monetary targeting was approximately 2.5 percent but after the adoption of inflation targeting it clearly increased to 3.9 percent. These results suggest that the growth record of Thailand under inflation targeting was favorable compared to monetary targeting. Comparing results between the core and headline inflation periods, growth declined slightly under the headline inflation targeting. Output growth volatility measured by the output growth standard deviation has generally declined over time. The fluctuations of Thai output growth were 5.6 before and reduced to 3.1 after adoption. Table 1 shows that the headline inflation targeting regime provides the lower output growth variability at 0.6 comparing to core inflation regimes at 3.7 on average. Given the Thai experience, the inflation targeting regime in Thailand has not generally damaged output growth and at the same time has been flexible enough for monetary policy to curb temporary shocks in the variation in output. in Thailand went from 6.8 percent before adoption to 2.2 percent after inflation targeting.
Comparing between the core and the narrow range core inflation targets, the narrower range has a lower nominal interest rate. The headline inflation targeting results in the lowest level of nominal interest rates at 1.5 percent. These results suggest that inflation targeting has improved public understanding and increased the credibility of Thai monetary policy. After adopting the inflation range target of 1.0-3.0 percent in December 2019, an outbreak of the Covid-19 started and spread around the world. The economic impact of the virus has severely been affected the Thai economy due to Thailand's openness to trade and as a tourism hub. The Bank of Thailand cut its policy rate several times from 1.25 percent in January to 0.5 percent in July 2020. The first seven months of 2020, the nominal interest rate on average falls to 0.87 percent and the average inflation during this period is -1.1 percent.

The New Keynesian Model
The phenomenon of economic fluctuations and money non-neutrality are believed by Keynesian proponents to happen because of imperfect competitiveness. Thus, the new Keynesian theory was developed based on the structure of the classical theory by combining the crucial assumptions of monopolistic competition and nominal rigidities. The model is composed of a single final good and a continuum of intermediate goods. Dixit and Stiglitz (1977) introduced monopolistic competition by which firms produce differentiated goods and have some market power to set the price of the goods they produce. Calvo (1983b) formulated a form of staggered pricing to capture the nominal rigidity. The non-linear new Keynesian model is built for Thai monetary policy analysis during the inflation targeting regime.

Household cost minimization problem
The economy is populated by a representative household. Firstly, the household would like to consume a final consumption good , at the lowest cost. Households choose an optimal combination of the intermediate goods that minimize the cost of achieving this level of the final good. Following Dixit and Stiglitz (1977), the consumption index is given by: The monopolistically competitive firms produce intermediate goods for consumption.
Assume the existence of a continuum of firms indexed by the subscript , where is distributed in the unit interval, ∈ [0,1]. Therefore, a continuum of intermediate goods is produced. Firm produces good ( ) and its price is ( ) . is the elasticity of substitution.
The household seeks to minimize its expenditure ∫ ( ) 1 0 ( ) subject to a basket of goods given by [2]. This results in the aggregate price index and a consumption demand equation of each differentiated good i:

Household utility maximization problem
Secondly, given the cost of achieving any given level of from [2], households optimally choose consumption good and labor to maximize their expected utility with respect to their period budget constraint. The preferences follow an external habit formation utility function as below 3 : is a coefficient of persistence in habits. ∈ (0,1) is the consumption and labor share and stands for the risk aversion coefficient while its inverse is the intertemporal elasticity of substitution. The proportions of time for leisure and work are and respectively. Thus, The budget constraint is given by: where is the stock of financial assets at the end of period , is the gross real interest rate paid on assets held at the beginning of period to pay out interest in period + 1, is the rental rate, is the real wage rate and, is investment and are lump-sum taxes. The law of motion of capital is governed over time by: Capital formation incorporates capital adjustment costs denoted by the function . 4 The parameter refers to the existence of costs in terms of investment changes between periods 3 Assume that ( , ) > 0, ( , ) > 0, ( , ) < 0 and ( , ) < 0.
with (1) = ′ (1) = 0 and ′′ (1) ≥ 0 implying that there is cost associated with changing the level of investment, that is cost is zero at steady state, and that this cost is increasing in the change in investment. as in Christiano, Eichenbaum and Evans (2005). With the adjustment cost, the optimal capital stock and investment decision are now separated. All variables are expressed in real terms relative to the price of output.
The household's utility maximization leads to the optimal intertemporal allocation equations as below: where , is the marginal utility of consumption at . , . It shows that the expected discounted return on a riskless bond is equal to 1. The marginal rate of substitution between the consumption equation and leisure is equal to real wage as in [9]. is the Tobin's Q representing the market value of the total installed capital over the replacement cost of that capital. is the Tobin's Q marginal ratio and is defined as = ⁄ .
[10] indicates that the value of current installed capital depends on its future expected value, taking into account the depreciation rate and the expected rate of return.
[10] could be rewritten as 1 = Ε [ +1 Λ , +1 ] where +1 = ( +1 + (1 − ) +1 )⁄ and is the gross return on captial. It indicates that the expected discounted return on capital over the period to + 1 is equal to 1. Therefore, the model equates the expected discounted return on a riskless bond with that of capital. The first order condition for investment is expressed in [11].

Firms
The production sectors are divided into two parts, a final good and intermediate good producers.

Final good sector
The final good is produced by a firm that aggregates intermediate goods into a single composite good using the following Dixit and Stiglitz aggregator: The firm takes as given the price of intermediate good ( ) and the price of the composite final good and then maximizes profits given a production function as in equation [12].
This results is the demand of intermediate good :

Intermediate goods sector
The intermediate goods are produced by a continuum of monopolistically competitive firms. Each intermediate good is produced by only one firm using labor and capital based on the following Cobb-Douglas production function: where is the productivity process and is end of period capital stock. Each firm acknowledges the demand curve it faces or ( ) = ( ). In this stage, firms take as given the prices of production function factors, the nominal wage and nominal capital rental rate and determine the amount of labor and capital to be hired to minimize costs. The firm's cost minimization problem leads to the following conditions: where represents a nominal marginal cost. [15] indicates that labor demand is the product of the nominal marginal cost and the marginal product of labor. Labor demand is equal to wage. The product of the nominal marginal cost and the marginal product of capital is capital demand. Capital demand is equal to rental rate, as in [16].
Using [15] and [16], the cost function and the instantaneous real profit function of firm can be written respectively as: where = ⁄ is a real marginal cost. Having described the firm profit, the model then adds the feature of price stickiness by considering the case of a staggered price setting established by Calvo (1983). The Calvo model assumes that some firms can adjust prices but other firms cannot. In any period, the probability that each firm will not adjust its price is and the probability that each firm will change its price is 1 − . Given that, the expected time a price remains unchanged is 1 1 − ⁄ then the key parameter measures the degree of price rigidity. All firms adjusting in period face the same problem, so all adjusting firms will set the same price. Let * be the optimal price chosen by all firms adjusting at time . The aggregate of all prices in the economy will be: we can rewrite the aggregate price level as: Consider a firm, , that has a chance 1 − to reset its price at . The firm's decision problem is to choose * ( ) to maximize its discounted real profits as below: Using the demand curve for good , the real profit is given by: where represents the fraction of firms that do not adjust price. All firms adjusting in period face the same problem, so all adjusting firms will set the same price. Let * be the optimal price chosen by all firms adjusting at time . The profit maximization results in the first order condition for optimal choice of * as: Since all firms face the same marginal cost, the index is dropped. The first order condition can be rearranged further resulting in optimal pricing behavior of intermediate goods: where ( − 1) ⁄ represents the mark up, describing the difference between the price and the marginal cost.
[23] indicates that the firm will mark up the price over its current marginal cost if the degree of price stickiness or = 0, which is the condition of the flexible price model. In the case of the sticky price model, nonetheless firm will mark up the price over the weighted average of flow of future marginal costs. When setting the price, the higher the value of , the further the distance in the future of marginal costs that the firm must take into account. Price dynamics with the introduction of markup shock is shown in the appendix.

The price indexation
The price indexation is introduced to explain the inflation persistence as in Christiano, Eichenbaum and Evans (2005) and Smets and Wouters (2007). For firms that do not reoptimize with probability , their prices are partially indexed to last period's aggregate inflation: where −1 = −1 −2 ⁄ and ∈ [0,1] is the magnitude of the indexation factor. The aggregate price index [18] is then determined by the optimal price setting and no reoptimizing price setting: Rearranging further yields the expression that relates optimal price, current inflation and past inflation: A firm setting its price in period and may not reset its price in future periods, its future demand is determined by the optimally chosen price takes into account future inflation.
Recursively solving [24] and putting the results in the demand function [13] yields the firm's future demand: The price behavior with no reoptimization is given by: The discounted real profit [20] can be rewritten as: The optimal price setting under the price indexation is that a firm chooses its price to maximize its discounted real profits [29] using [27] and [28] for the duration that it cannot reoptimize its price. The first order condition for the above profit maximization is: The first order condition can be rearranged further yielding the optimal pricing behavior: * = ( − 1) It is necessary to transform the first order condition for pricing to recursive definitions in Another way of indexing assumes that prices are indexed to a weighted average of last period inflation and trend inflation. Denote the two weights by and ̅ , the dynamic equations [33] and [34] can be replaced Π = /( −1 * (1−̅ ) ) in which * is the inflation target. Smets and Wouters (2007) assumed that = ̅ so that the effect of trend inflation is eliminated.

Output equilibrium
The  Yun (1996). Therefore the equilibrium relation between aggregate demand and aggregate factor inputs is given by: Notice from [37] that price stickiness causes an inefficiency through the price dispersion mechanism. Note that Δ = 1/( * ⁄ ) .

The effects of monetary policy
Taylor (1998)

The shocking process specification
Suppose the model economy experiences an unanticipated variation in total productivity, government spending and mark-up shocks. The effects of the aggregate supply and demand shocks can be evaluated through the technology shock and monetary policies.
The law of motion for the exogenous shocks are assumed to follow a first-order autoregression process AR(1) and they can be expressed in log form around the steady state. The exogenous forcing processes to technology, government spending and mark-up shocks are respectively shown below: The parameter measures how persistent each shock is. In other words, how important is that last period shock in determining how large the shock is in this period. The variables are innovations to each random shock and they are normally distributed and serially uncorrelated.

The Bayesian estimation
Before the late 1990s, researchers used optimization methods such as the maximum model likelihood and the priors on the DSGE parameters is the objective of the Bayesian influence process. The posterior is the density of parameters given existing data. Using the Bayesian rule, the posterior distribution can be computed as: where p( . ) stands for a probability density function and stands for the DSGE model. The gamma distribution is clearly for parameters restricted to be positive. The inverse gamma has support on an open interval that excludes zero and is unbounded. It is commonly the distribution for the standard deviation of the shock processes. The normal distribution reveals the extent of the uncertainty that surrounds the steady state value. The prior distributions of the parameters are listed in Table 2. Table 2 contains the priors for the model, following Smets and Wouters (2007). The figures are reported with the 90 percent highest posterior density interval. Table 3 shows the mean of the posterior distributions comparing with that of prior distribution. The model predicts that the degree of price rigidity is = 0.1792 in Thailand. Given that, the expected time a price remains unchanged is 1.22 quarters which comes from 1 (1 − ) ⁄ .

The dynamic effects of monetary policy rules and exogenous shocks
Given the estimated parameters from the Bayesian estimation, the model structure enables us to study alternative monetary and fiscal policies and to analyze the economic impact of productivity and mark up disturbances. This section explores the Thai economy's equilibrium responses as percentage deviations about the steady state to four different shocks.
The first shock is a 1 percent supply side shock to total factor productivity, , and the second is a 1 percent demand side shock to government expenditure , . Figures 3 and 4 illustrate the impulse response as a proportional deviation about the steady state to the technology and the government spending shock respectively. The third shock is a 1 percent shock to monetary policy rules, , and the last shock is a 1 percent mark-up shock, , .   However, if Thai output is lower than its potential level by 1 percent, the Bank of Thailand reduces its policy rate by just only 2 basis points. The larger value of indicates that the Bank of Thailand is more responsive to the deviation of inflation from its target.
The key monetary mechanism exists through changes in the real interest rate which affect consumption and investment. The increase in the real interest rate leads households to attempt to postpone their consumption and motivates firms to cut their investment. As a result, the aggregate demand falls. The rise in the nominal interest rate causes inflation and the output gap to fall immediately. During the inflation targeting regime in Thailand, it is seen that the 1 percent positive policy shock generates an increase in the real rate by 2 percent deviation from its steady state, and decrease in inflation by nearly percent from its steady state and a decrease in output by approximately by 0.08 percent deviation from its steady state as depicted in figure  As long as the central bank is able to affect the real interest rate through its control of the nominal interest rate, monetary policy can affect real output. Hours worked and real wage fall on the impact. An increase in the interest rate causes the discounted future profit of a firm to decline. Therefore, price of capital and investment falls from the impact.
In the case of cost push or markup shock, figure 6 shows the impulse response to a 1 percent increase in the markup shock. An increase in markup leads to a higher level of profits for companies operating in the economy. Firms raise prices because of a higher markup. A positive markup shock leads to a fall in output, consumption, investment and hours worked for 4 quarters after the impact. The real interest rate reduces on the impact and gradually goes back up. Consumption and investment fall along with output. The shock generates an increase in inflation at the same time reducing in output. The central bank responses by raising the interest rate but the size is relatively small.

The inflation targeting, price indexation and the steady state
Ascari (2004) shows that when trend inflation is considered, both steady state and dynamic properties of the model change dramatically. One way to add trend inflation to the model is indexation. Yun (1999) and Jeanne (1998)  For zero inflation at the steady state, * = 1, we obtain = ∆ = 1 and = ( 1 1− ).
Nonetheless if * > 1 we obtain that < ( ), reflecting that the steady state markup will be higher than it would be if there were no trend inflation. At the steady state, the real wage is equal to the real marginal cost.
The lower the marginal cost, the greater the difference between the wage and the marginal product of labor. Price dispersion also creates an inefficient resource allocation, leading to a loss of aggregate output. Therefore, the model predicts a negative relationship between the inflation targeting rate and aggregate output at the steady state. A higher inflation targeting rate should be associated with lower steady state output. The standard new 8 The hours work at the steady state is estimated by using the average hours in a week is 48 hours (8 hours a day and 6 days a week), divided by total time available 98 hours (14 hours a day and 7 day a week), multiplied by the fraction of population that works is 0.69 (the average labor force divided by the population 15 years and older) Keynesian model with no trend inflation creates a higher level of long run output relative to the one with trend inflation. The reasonable explanation is that in the model with no trend inflation, inflation in the long run is fully stabilized. Inflation enlarges the effect of price dispersion to aggregate output. Therefore, there is no price dispersion and output is at the highest possible level. and that of the model without trend inflation. The model with no tend inflation is the benchmark. Table 4 documents the effect of the inflation targeting rate and the past inflation indexation to the steady state for the Thai economy during headline inflation targeting 2015q1 to 2019q1. The experiment in this section is that the prices are indexed to last period's inflation. Given a price indexation parameter, , the higher is the inflation targeting rate, the greater the output is below its long run level of no trend inflation. For example, for inflation targeting at 2.5 percent and = 0.5, steady state output is lower than its steady state level of no inflation at the steady state by 1.29 percent. Long run investment is consistently lower than its long run level of no inflation at the steady state by 1.98 percent.
However, given the new inflation target range is 1-3 percent, if the Bank of Thailand chose its inflation targeting at 2 percent which is the mid-point of the new range, long run output would be higher. The steady state output is lower than its steady state level of no inflation at the steady state by only 0.74 percent. Long run investment is lower than its long run level of no inflation at the long run by 1.12 percent. Table 4 shows that if = 0.5, steady state output is lower than its steady state level at no inflation at the steady state by only 0.37 percent for inflation targeting at 1.5 percent.
Long run investment is lower than its long run level at no inflation at the steady state by 0.55 percent. Additionally, given inflation targeting, table 4 shows that the higher is the degree of price indexation, the smaller is the output below its steady state level. The reasonable explanation is that at the steady state past inflation equals the inflation targeting rate and that at a higher degree of price indexation, the lower is the price dispersion at the steady state. Table 5 shows the effect of the inflation targeting rate and the average inflation indexation on the steady state for Thai economy during headline inflation targeting 2015q1 to 2019q1. The experiment is that the prices are indexed to a weighted average of last period's inflation and trend inflation. If ̅ = 1, the indexation is identical to the case of the past inflation indexation. Given the inflation targeting rate and ̅ = 0, when price indexation parameter rises, the price dispersion rises and output is further below its steady state level. If = ̅ = 0.5, the indexation is identical to the case of no trend inflation, leading to output at the benchmark. The deviation of the degree of past inflation indexation from 0.5 will have quite a similar impact to output. Given the inflation targeting and the price indexation parameter , an increase in the indexation parameter ̅ results in a ushape in output deviation from its steady state level. In any case, from Table 5, an increase in the inflation targeting rate causes output to be further below its steady state level. Both the past inflation indexation and the average inflation indexation to the steady state yield the same result -that is, the larger the inflation targeting rate is, the lower the steady state output from its steady state level given no trend inflation.

Policy implications
The model suggests the routes through which the level of inflation targeting is costly for monetary policy in the long run. In the long run, price dispersion and markup distortion are minimized when the level of the inflation target is low. Ambler (2007) reviews that under the New Keynesian model with inflation targeting, price dispersion is an increasing function of trend inflation and causes long run output to be a decreasing function of inflation targeting rates. There is a negative trade off between the level of inflation target and output in the long run. The model according to this study predicts that in Thailand a lower level of the inflation targeting rate generates a higher output in the long run.
However, Ascari and Ropele (2007) show that in the short run the level of the inflation target alters the relation between inflation and output and in turn influences the dynamics of inflation. In other words, the level of the inflation target affects the slope of the Phillips curve. The output gap is decreasing in the inflation targeting rate. Therefore, a decline in the central bank's inflation targeting rate enhances the relationship between inflation and output gap. The Phillips curve is steeper and monetary policy is more effective.
In the second quarter of 2019, Thailand's economy grew at its slowest rate in nearly five years. According to Thailand's Office of the National Economics and Social Development Council, Thai growth slowed sharply. The first three quarters of 2019, the Thai economy grew only 2.5 percent which was clearly slower than growth in 2018 at 4.1 percent. Thai growth was not only lower than expected but also further below its potential growth. Thai potential growth is estimated at 3.7 percent. Under the condition that the economic activities and the employment rate below their natural levels, the economic growth is slower and the economy enters recession. Over time, wages and prices are falling in response to lower aggregate demand leading to low level of inflation. As we have seen, Thai inflation has continuously been low and below the inflation target for several years since adopting headline inflation targeting. The Thai average inflation for the first eleven months was slightly lower than 1 percent which was the lower bound set by the Bank of Thailand.
Failing to achieve a constant rate of inflation in line with inflation expectations, the Thai economy by implication cannot keep output at its potential level. Thai output has been lower than its potential output after adopting headline inflation targeting. If the Thai economy has persistently been operated below its normal level, this critically is the structural problem. Thailand needs to create skilled workers and Thai firms must exhibit innovation.
The Thai educational system should deliver the required human capital. Investment and particularly investment in R&D must increase.
The relatively low level of inflation causes the Thai baht to appreciate in the long run. The Federal Reserve Bank has been able to manage US inflation around its target at 2 percent while Thai inflation has only been around 1 percent. Under purchasing power parity, Thailand inflation has been lower than US inflation for the last six years, resulting in the current situation that the Thai baht has appreciated against the US dollar to the highest level in last six year in October 2019. The low level of economic activity also leads to low inflation and a low expected future inflation. The low expected future inflation causes the real interest rate to rise and in turn lowers investment and output. The world inflation is on a decreasing trend according to the IMF report. Inflation is decreasing because of structural changes such as the expansion of e-commerce, price competitiveness, the low price of oil and technological progress. Beside the low demand in Thailand, these factors in some degrees cause Thai inflation to be low as well.
Therefore, the level of inflation targeting set by the Bank of Thailand should reflect the Thai current economic situation. Lowering the inflation target could describe better Thai actual inflation and result in increased effectiveness of monetary policy. Adjusting the inflation targeting down to the target range of 1.0-3.0 percent should be the appropriate strategy.

Conclusion
The The model shows that the Bank of Thailand is more responsive to the deviation of inflation from its target. If Thai inflation is lower than its target by 1 percent, the Bank of In the end, there is a long run negative inflation-output trade off in the choice of a steady state inflation rate for the Thai economy. The study finds that the higher the inflation targeting rate, the lower the output is below its long run level of no trend inflation. In addition, the higher the degree of price indexation, the smaller the output below its steady state level. Indeed, inflation is costly through price dispersion. This price dispersion increases at higher levels of inflation targeting and leads to a loss of output. Therefore, the study recommends continued use of inflation targeting with a reduced target level. Recently, the inflation target range of 1.0-3.0 percent adopted by the Bank of Thailand in December 2019 should be the proper policy.

Household cost minimization problem
The Lagrangian function corresponding to this problem is: The first order conditions for the minimization are given by: This results in the aggregate price index for consumption which equals to : where is the Lagrangian multiplier. The consumption demand for each differentiated good m can be written as 9 : − 9 The price elasticity of demand for good i is  . If  , the individual goods become substitute goods and individual firms have less market power.

Household utility maximization problem
The Lagrangian function associated with the household maximization problem can be defined as: Notice that if = 0 , there are no adjustment costs in investment and = 1 or =

Final good firm profit maximization problem
The maximization problem of a representative firm in the final good sector is:

Intermediate good firm profit maximization problem
The Lagrangian function corresponding to this problem is: The corresponding first order conditions are: Notice that the Lagrange parameter associated with the technological restriction represents the shadow price of change in the ratio of the use of capital and labor services, implying that the Lagrange parameter measures nominal marginal costs = . Where is the markup shock.

Price dispersion
Define ̂ as the aggregate factor inputs and using = ∫ ( ) 1 0 , yields the aggregate factor inputs:     Table 4 The effect of the inflation targeting rate and the past inflation indexation on the steady state for the Thai economy during headline inflation targeting 2.5% ± 1.5% -0.3648 -0.2154 -0.1088 -0.0400 * is the inflation targeting at the mid range is price indexation parameter, indexing to past inflation The figures are the percent deviation of each parameter from its natural level of zero inflation at the steady state Table 5 The effect of the inflation targeting rate and the average inflation indexation on the steady state for the Thai economy during headline inflation targeting 2.5% ± 1.5% * = 2.5% * = 1.5% ̅ = 0 ̅ = 0.5 ̅ = 1 ̅ = 0 ̅ = 0.  All variables are in the deviation from its steady state form. XX is X deviations from its steady state. Y is output. C is consumption. I is investment. H is working hours. W is real wage. R is ex post real interest rate. Rn is nominal interest rate. π is inflation. Q is Tobin's Q. The vertical axis is percentage and the horizontal axis is quarters.

Figure 4 Government spending shock
Note: Same as figure 3