bangladesh infln.doc

7/28/2019 Bangladesh infln.doc

1/3

Bangladesh has been experiencing a rapid growth in the general price level in recent years. The rateof inflation has crept up steadily since July 2009, rising from an average of 2.3 per cent during2008/09 to a peak of 12 per cent in September 2011. The inflation rate declined to 9.9 per cent inApril 2012. The rapid rate of inflation has become a major economic and social problem. Unless this istackled forcefully and with some urgency it could become a substantial political debacle for theGovernment when it seeks re-election in the next 18 months. It is also important that right policychoices are made in the effort to control inflation based on sound analysis.

The Policy Debate: There is much policy debate, often influenced by populist perceptions, about whatfactors cause inflation. One popular debate concerns the role of nominal exchange rate in managinginflation. There are quite a few policy makers, researchers and businesses who believe that thedepreciation of the exchange rate is the primary culprit underlying rapid inflation in Bangladesh. Thisgroup believes that the government should basically pursue a fixed nominal exchange rate policy. Theunderlying logic is the standard cost-push argument for inflation. Exchange rate raises the taka priceof imported inputs that pushes up the cost of production and that in turn fuels inflation.The other debate is the role of international commodity prices. The rising global food and fuel

prices are seen as the main culprit underlying inflation in Bangladesh. In this debate inflation istemporary and the government has little control over inflation except to make efforts to insulatedomestic food and fuel prices from rising through price controls and subsidies. The influence of thispopulist argument on policy making is large and illustrated by the rapidly growing subsidy bill of thegovernment that has now run into almost 4.0 per cent of gross domestic product (GDP), equivalent to

over 35 per cent of total tax revenue.A third major policy debate is the perceived trade-off between growth and inflation. The argumenthere is that developing countries like Bangladesh have no choice but to tolerate significant inflation asa price for economic growth and development. This is a more substantive argument based onquantitative research done by the Bangladesh Bank that argues that there is a trade-off betweeninflation and growth well up to 6.0-8.0 per cent rate of inflation. Efforts to reduce inflation below thisthreshold level will have an adverse effect on growth.The policy debate suggests that there is considerable confusion and disagreement on the nature ofinflation in Bangladesh and factors that explain inflation. Given these issues, it is important that policyanalysis for inflation control must be based on a careful review of the data and proper quantitativeanalysis.Quantitative Analysis: Time series data and related analysis can be very helpful in understandingmacroeconomic developments, making projections for the future and developing policy responses totackle unhappy macroeconomic outcomes. The analysis of inflation is a good example of how proper

time series analysis can help the government to control inflation and stabilise the macroeconomy. Wewill like to stress the importance of "proper time series analysis" because in its absence we can easilyreach erroneous conclusions that, if applied to policy making, can do substantial damage to themacro-economy.Stationarity and Causality Tests for Proper Time Series Analysis: A key requirement of this "proper"analysis is to first start with a good theory about causality. Many things tend to move together overtime. Without a sound analysis of how developments are correlated and what is the cause and what is

the effect there is either a risk of spurious correlation or mis-specification of the relationship.Economic theory helps avoid the problem of spurious correlation, but it sometimes does not helpidentify the causality. This problem of establishing causality in time series data has received a great


2/3

deal of attention in quantitative economic research and considerable progress has been made inrecent years to help identify proper causality, thereby facilitating better policy making and economicforecasting.Working independently in different time periods, two researchers, CliveGranger of the University ofNottingham in England and Christopher Sims of the Princeton University in USA, pioneered thequantitative methods for establishing causality. Both received the Nobel Prize inEconomics; Grangerin 2003 (unfortunately he died in 2009) and Sims in 2011. The statistical

technique developed to establish causality is known as the Granger-Sims test.Good practice quantitative research using time series data first needs to ensure that the data are"stationary" (to enable meaningful predictions) and that causality is established using the Granger-Sims test before deciding which variable is the cause (also called the independent or exogenousvariable) and which variable is the effect (also known as the dependent variable or endogenousvariable).Inflation and GDP Growth Models for Bangladesh: Turning to the Bangladesh situation, we needtestable model to quantitatively determine the factors that explain inflation. Drawing from economicliterature and the policy debate in Bangladesh, there are basically three models of inflation. The

simplest model, which appeals most to populists and is well understood in Bangladesh, is the cost-push model of inflation where the rate of inflation is determined by cost factors such as internationalfood and fuel prices, other sources of imported inflation reflected by world inflation, and the nominalexchange rate changes. The second model is the well-known monetarist hypothesis popularised byNobel Laureate Milton Friedman where inflation is determined by the excess of monetary growth over

the rate of growth of GDP. In this model inflation is purely a monetary phenomenon. The third andmore widely accepted model is one that combines both cost-push and monetary factors. Which factordominates at any point in time is determined on the basis of proper time series analysis.For this policy paper, we develop a generalised inflation model that combines both cost-push andmonetarist variables. The model is specified as follows:INF= f (GM2, GGDP, DNER, GIFP),Where:INF= rate of inflation;GM2= growth of broad money;GGDP= rate of growth of GDP;DNER= depreciation of the nominal exchange rate;GIFP= rate of growth of international food prices.The model basically says that inflation is determined by the rate of growth ofmoney supply (broadlydefined), the rate of growth of GDP, the nominal exchange rate, and the rate of growth of

international food prices. Other cost factors such as international inflation could also be introduced andwill be considered as we go along.Since there is a policy debate surrounding the relationship between inflation and rate of growth ofGDP, we also estimate a model for GDP growth that allows for feedback from inflation to GDP growthand money supply growth to GDP growth. A simple GDP growth model is specified below.GGDP= f (GM2, INF, INV/GDP, GLab, T/GDP),Where:GGDP= rate of growth of GDP;INF= rate of inflation;GM2= rate of growth of broad money;INV/GDP= investment rate;GLab= growth rate of labor forceT/GDP= trade to GDP ratioThe GDP growth model says that the growth of GDP depends upon the rate of investment, the growth

of labour force, trade to GDP ratio (to allow for openness effect), the rate of inflation and the rate ofgrowth of money supply.Stationarity Test Results: The data we use is from 1981-2011, which gives a fairly large number of

observations to estimate stable long-term relationships between the variables. Before we estimatethese two equations, we need to test for "stationarity" and "causality". The standard Dickey-Fullerstationarity tests show that the data for GDP growth, Inflation, M2 growth, International food priceinflation, Depreciation of the nominal exchange rate, and Labour force growth are all zero orderstationary. The investment rate is not zero order stationary. However, the investment level (INV) iszero order stationary. Similarly, the trade to GDP ratio is not zero order stationary but the change inthe ratio (D T/GDP) is stationary.


3/3

Granger Causality Test Results: Mere evidence of correlation does not indicate causality. Economictheory can help. Thus, there is little debate about the causality of the following relationships: worldfood price inflation causes domestic inflation; labour force growth causes GDP growth; investment ratecauses GDP growth; and trade-openness causes GDP growth. However causality relationship betweeninflation and money supply growth; between GDP growth and inflation; between GDP and moneysupply growth; and between inflation and nominal exchange rate changes are debatable. So the datawere checked for causality using the Granger causality test. The results are:

1) Inflation and GDP growth do not Granger cause each other.2) M2 (broad money) growth Granger causes Inflation but not the other way round.3) Inflation Granger causes Nominal exchange rate depreciation but not the other way round4) GDP and M2 growth Granger cause each otherThe result that inflation causes exchange rate depreciation rather than the other way round indicatesthat it cannot be used as a determinant of inflation. The joint causality between GDP growth and M2growth suggests that M2 cannot be used on the right hand side of GDP growth equation. Instead aninstrumental variable (growth of private credit GPC) is used after ensuring its stationarity, Grangercausality test and relevance as an instrument.

Estimation Results for Inflation: The estimated regression result for inflation equation is shown below:INF= 8.28 (0.002) + 0.21(0.010) GM2+ 0.015 (0.66) GIFP- 0.99 (0.026) GGDPR-squared=0.3183; Adj. R-squared= 0.2425; p values in bracketsThe results show that the growth of broad money (GM2) has a strong and positive effect on domesticinflation (INF), while GDP growth has a strong and negative effect on inflation. These results are

consistent with the findings of many other researches. International food price inflation has the rightsign but its coefficient is not significant, suggesting that while it may play a significant role in theshort-term it is not a significant factor for long-term inflation. Introduction of international inflationvariable did not show up as significant.Estimation Results for GDP Growth: The estimated regression result for the GDP growth equation isshown below:GGDP= 4.39 (0.00) + 0.0031 (0.00) INV + 0.172 (0.355) GPC -0.14 (0.056) INF+0.069 (0.218) DT/GDP -0.098 (0.626) GLFR-squared= 0.615; Adj R-squared= 0.61; p values in bracketThe results show that investment (INV) is a strong and positive determinant of GDP growth. Inflation(INF) has a strong and negative relationship with GDP growth. Growth of private credit (GPC) has thecorrect sign but its coefficient is insignificant. Trade liberalisation (DT/GDP) has a positive effect onGDP growth, although its coefficient has a relatively low significance. The labour force growth (GLF)variable comes up insignificant and with the wrong sign. This could reflect a measurement problem.

Dr Sadiq Ahmed is Vice-Chairman, Policy Research Institute (PRI), Bangladesh. This write-up on'Explaining Inflation in Bangladesh' is a policy note that was presented at a recent policy dialogue inDhaka, organised by PRI, Bangladesh. [email protected]
mailto:[email protected]:[email protected]

bangladesh infln.doc

Documents