Description

09/11/2014
Download Document
Título: Assessing the impact of financial liberalization on stock market volatility in selected developing countries
Autor(es): Ilene Grabel
Fonte: Journal of Development Studies. 31.6 (Aug. 1995): p903.
Tipo de documento: Article
Copyright: COPYRIGHT 1995 Frank Cass & Company Ltd.
http://www.tandf.co.uk/journals/titles/00220388.asp
Resumo:
This article argues for the importance of measuring stock market volatility following financial liberalisation in
developing c

All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.

Related Documents

Share

Transcript

09/11/2014 Download Documentfile:///C:/Users/127/Downloads/Download%20Document.html 1/11
Título: Assessing the impact of financial liberalization on stock market volatility in selected developing countriesAutor(es): Ilene GrabelFonte:
Journal of Development Studies.
31.6 (Aug. 1995): p903.Tipo de documento: ArticleCopyright: COPYRIGHT 1995 Frank Cass & Company Ltd.http://www.tandf.co.uk/journals/titles/00220388.aspResumo: This article argues for the importance of measuring stock market volatility following financial liberalisation indeveloping countries. Three alternative indices for measuring volatility are developed; these are used to examinethe view that financial liberalisation induces increased asset price volatility. Based on the limited data available,this view is corroborated in the majority of countries investigated.Texto completo: I. INTRODUCTIONMuch has been written about the disappointing experiences of many less developed countries (LDCs) thatimplemented aggressive financial liberalisation (FL) programmes in the mid- to late 1970s and early 1980s.These contradicted the predictions of economists McKinnon and Shaw,(1) the early architects of the neoclassicaltheory of FL, who argued that FL in LDCs would induce a virtuous cycle of increased savings, investment andgrowth [McKinnon, 1973; 1991; Shaw, 1973]. In the event, many countries which followed the McKinnon-Shaw prescription suffered financial instability and crises, reduced savings and disappointing growth.(2) Nevertheless, FL has been taken to be uniformly successful in one important respect, namely, in encouraging theformation of equities markets where they did not previously exist, and encouraging their deepening where they predated the reforms. Indeed, recently there has been a proliferation of scholarly and popular articles which extolthe virtues of these rapidly expanding markets for LDC economic performance and for investors worldwide[Drake, 1985; Errunza and Losq, 1987; Je Cho, 1986; Papaioannou and Duke, 1993; Rowley, 1986]The expansion of equities markets in many LDCs has been truly impressive. In 1980 LDC stock markets listedsome 5,531 domestic companies and had a market capitalisation of $86,125 US million and an annual tradingvolume of $23,672 million. By the end of 1992, 36 LDCs had stock markets listing a combined total of 13,217individual domestic companies with a combined market capitalisation of $774,093 million and an annual tradingvolume of $594,685 million.(3)Most commentators have argued that this explosion of equities markets has been an unmitigated benefit for thesecountries. However, Singh [1993], Calamanti [1983], and Samuels and Yacout [1981] argue from a Keynesian perspective that the expansion of LDC stock markets threatens to induce speculation and financial crises and amisallocation of savings and investment, to the detriment of real sector growth and stability.(4)This view of increased volatility and its detrimental consequences runs counter to the assumptions of the FLhypothesis. The latter predicts a decrease in volatility following FL, and in any event discounts itsmacroeconomic effects. From a Keynesian perspective, whether, and to what degree, expanded stock markets area boon or impediment to real sector growth and stability in LDCs depends in part on the level of volatility theyexhibit: markets with frequent and severe price swings might be much more apt to induce short-term speculativeinvestment practices and might also be more likely to induce broader macroeconomic instability. Yet, empiricalresearch to date has failed to explore the techniques appropriate for measuring price volatility in LDC stock markets, and hence, its actual levels.An empirical assessment of the effects of stock market volatility on LDC economies requires in the first instancea means to measure actual levels of volatility in these markets. This objective motivates this article. It presentsthree approaches to measuring stock market volatility in LDCs and presents the results of empirical applicationsof these measures to a sample comprising several Latin American and Asian LDCs that implemented FL programmes in the 1970s and 1980s. The empirical claims of the article are necessarily quite modest given theconstraints imposed by data availability. The measurement techniques proposed, however, will be useful tofuture empirical research as new data become available. Given these limitations, attention is necessarilyrestricted to those LDCs with established equities markets prior to FL and further to the subset of countries for which stock market data are available. Using the measures presented herein, the argument explored is that FLinduces increased asset price volatility.
09/11/2014 Download Documentfile:///C:/Users/127/Downloads/Download%20Document.html 2/11
The article is organised in the following manner. Section II explores the macroeconomic significance of volatilityfor LDCs, sketching briefly the manner in which high volatility might induce important macroeconomic effects.Section III explores various theoretical and empirical measurement issues, and presents three distinct measuresof volatility. In section IV a case is made for a limited use of volatility indices (VIs). Section V describes theactual scope of FL programmes, and presents the results of the volatility measures. These are on the wholefavourable to the volatility-inducing argument. The paper concludes with a reflection on the implications for future empirical research.II. THE MACROECONOMIC SIGNIFICANCE OF VOLATILITY FOR LDCSThe Keynesian view advanced here contradicts a key assumption of the neo-classical FL hypothesis. In this view,the financial deepening associated with FL should decrease overall volatility by increasing the numbers of bothinvestors and tradeable shares, and by encouraging the increased production and dissemination of reliableinformation via the increased profit opportunities which attend financial deepening. Furthermore, even if FLleads to an increase in volatility, the capital asset pricing model [Merton, 1980] suggests that increases involatility would not impair macroeconomic performance provided that stock returns incorporated appropriaterisk premia (that is, that markets are efficient) [Chou, Engle and Kane, 1992].From a Keynesian perspective, volatility may be expected to result from the quickened pace of financialtransactions which FL allows. Moreover, volatility may be self-exacerbating: volatility forces investors toshorten their time horizons for both offensive (profit-seeking) and defensive (loss-minimising) reasons, with the paradoxical effect of inducing increased volatility. This implies that increases in market volatility may lead toreductions in real-sector investment activities. [Keynes, 1964: Ch. 12; Singh. 1993].Increasing volatility might also have deleterious effects on the macroeconomy via increasing financial fragility.This could dampen overall economic activity, and in the event of an exogenous shock (such as an unexpecteddramatic increase in the interest rate), could lead to forced asset sales and a cumulative debt-deflation. In sum,the success of FL in introducing mechanisms of rapid asset price adjustment may introduce increased volatilityinto the economy and may, as a consequence, undermine macroeconomic stability and economic growth(5).These potentially injurious effects of FL have not been explicitly explored in the expansive empirical literaturethat tests the McKinnon-Shaw FL hypothesis.(5)III. METHODOLOGY OF VOLATILITY INDEXESTwo classes of VIs will be developed. What is termed the neo-classical VI (NC-VI) is based on the theoreticalview that assets yield some 'normal' return over time based on their underlying fundamental value.(7) Themagnitude of the deviation from the asset's fundamentals-based return constitutes asset volatility. What is termedthe Keynesian VI, on the other hand, presumes that asset returns have no fundamentals-based centre of gravity.(8) Volatility in the Keynesian case is simply given by the magnitude of asset return fluctuations.Derivation of the NC Volatility Index (NC-VI)The NC-VI is based on a return-to-normality model of asset returns [Fortune, 1989]. In this model, the realisedreturn on an asset in any period t ([R.sub.t]) is equal to the sum of its expected return for that period ([E.sub.t])and the innovation in returns ([I.sub.t]). The return-to-normality model is attractive from the neoclassical perspective because it side-steps the thorny theoretical issue of determining which variables ought to be includedin the set of 'fundamentals' which, in this view, determine asset returns. In an efficient market the expected returnin any period t is a function of the information available through the previous period; that is, [E.sub.t][where]t-1.The innovation in returns is a consequence of new information that becomes available in period t. The innovationterm is assumed to have standard statistical properties (namely, it is randomly distributed with a mean of zeroand a constant variance). The critical assumption in the return-to-normality model is that the departure of ex-postreturns from expected returns will eventually be corrected by arbitrage activities.The return on an asset is given then by:[R.sub.t] = [E.sub.t][where]t-1 + [I.sub.t] (1)where E([I.sub.t]) = 0 and E[[I.sub.t].sup.2] = [Sigma]2. In period t, the expected return on an asset is the sum of the normal or historical return, R', and some departure from R' proportional to the previous period's deviationfrom R'.(9) That is:
09/11/2014 Download Documentfile:///C:/Users/127/Downloads/Download%20Document.html 3/11
[E.sub.t]t-1] = R[prime] + [Theta] ([R.sub.t-1] - R[prime]) (2)where [Theta] ranges from zero to one, inclusive. Simplifying and substituting (2) into (1):[R.sub.t] = [(1 - [Theta])R[prime]] + [Theta][R.sub.t-1] + [I.sub.t] (3)Defining a equal to (1 - [Theta])R[prime] we can rewrite (3):[R.sub.t] = a + [[Theta][R.sub.t-1] + [I.sub.t] (4)We can estimate a linear equation in which the independent variables explain the expected value of [R.sub.t](and R[prime]), and the residual measures the innovation in [R.sub.t]. Having extracted the innovation in returnsfrom the time series data on returns, we can then construct an index of volatility [Fortune, 1989: 27].Before presenting the standard index of volatility in any single period (called here the 'NC-VI, type 2'), we willdepart from the standard volatility literature and develop a straightforward index of volatility that is also basedon the return-to-normality model (called here the 'NC-VI, type 1'). NC Volatility Index, Type 1The type 1 NC-VI allows a stark comparison of market volatility under two different regulatory regimes. Thetype 1 index will allow us to determine whether the variance of the innovation in stock prices has increasedsignificantly with FL. This comparison of market volatility in the pre- and post-FL periods will be undertakenusing the Goldfeld-Quandt test (for homoscedasticity) [Kmenta, 1986: 292-4]. This test involves orderingchronologically the observations of returns and omitting one-third of the observations in the middle of the entiresample (eliminating the transition period), leaving two equal-sized groups of observations.(10) Through OLS wecan extract the innovations in stock prices from the pre- and post-FL periods. Then, we create a ratio of the sumof squared innovations of prices divided by the share price index (SPI) for each year from the post-FL perioddivided by the same ratio for the pre-FL period. The type 1 NC-VI is written:[Mathematical Expression Omitted]Rejection of the null hypothesis (of equal variances) is consistent with the expectation of increased marketvolatility following FL. NC Volatility Index, Type 2While the type 1 VI allows comparison of market volatility in two periods (namely, the pre- and the post-FL periods), the type 2 NC-VI provides a measure of market volatility in any single period t. Using this VI, volatilityin period t is defined as the ratio of the squared innovation in period t [Mathematical Expression Omitted] to thevariance of the innovation over the entire sample period, [[Sigma].sup.*2]. The NC-VI, type 2 is defined:[Mathematical Expression Omitted]Given that the mean of this index is unity,(11) a period with an index value greater than one may be consideredone of high (relative) volatility; conversely, a period with a value of less than one would be considered one of low volatility [Fortune, 1989: 15].The Keynesian Volatility IndexThe Keynesian VI differs from the NC-VI in its refusal to posit a return to normality or a fundamentals-drivenrate of return. The NC-VIs (both type 1 and 2) impose a return-to-normality model and use autoregression toimpose constant mean, homoskedasticity, and non-autocorrelation.(12) Hence one must interpret NC-VIs withsome caution. In the Keynesian VI, volatility is measured by the rapidity and magnitude of changes in market prices alone. An asset is relatively volatile during one period if its price moves by larger amounts during this period than during others. This volatility is of interest because rapid and large swings in asset prices threaten thesecurity of collateral and the value of capital bases, regardless of the causes of volatility.The Keynesian VI measures market volatility over some interval as the coefficient of variation.(13) TheKeynesian VI for some interval is thus defined:VI = [Sigma]/mean price (7)
09/11/2014 Download Documentfile:///C:/Users/127/Downloads/Download%20Document.html 4/11
IV. TOWARD A LIMITED USE OF VOLATILITY INDEXESThe use of VIs in the corporate and international financial literature remains controversial [Bryant, 1987: Ch.6].(14) There are inherent difficulties in using the NC return-to-normality model for the purpose of detectingincreasing price volatility. Critically, fitting the values of the coefficient using standard techniques (for example,autoregression or moving average) entails correcting in advance for autocorrelation and heteroskedasticity. But if the argument presented here is correct, then we would expect to find autocorrelation and heteroscedasticity andits apparent elimination via statistical techniques prior to the construction of the types 1 and 2 NC-VIs wouldseverely bias the results against a finding of increasing volatility. But the converse, is also true: if, to take theexample of a type 1 NC-VI, the test statistic does prove significant, then there is good support for the increasingvolatility argument. Nevertheless, given all the attending conceptual problems, even such a finding should betreated with caution. This suggests the need to Utilize also a measure of volatility that overcomes the inherent bias of the neoclassical model. The Keynesian VI accomplishes this.The modesty of the claims regarding the actual VIs presented here should again be acknowledged. Present datalimitations restrict greatly the range of countries for which VIs can be calculated, and, even for those countriesfor which data are available, gaps of a few years in the data collected from 1984 to 1993, limit even further thecalculation of VIs.HypothesesWe would expect to find increases in volatility (in the post- as compared to the pre-FL period) using the type 1 NC-VI. Using the type 2 NC-VI, we would expect to observe a preponderance of VIs (for any single period t) below 1.0 prior to FL and above 1.0 following FL. This assessment will be made by comparing the respectivemeans of the pre- and post-FL (type 2 NC-) VIs. In order to minimise concerns about measuring volatility duringthe transition period to FL, in calculating the means of the pre-and post-FL VIs we will omit the four years of observations that span the transition.(15)Turning to the Keynesian VI, we will compare volatility in the pre- and post-FL periods by inspecting the meansof the VIs for each of these periods (as with the type 2 NC-VI). Ideally, the Keynesian VI should corroborate thefindings of the types 1 and 2 NC-VIs. Observations spanning the four years around the implementation of FL areagain omitted in order to eliminate temporary distortions due to uncertainty about and adjustments to the regimeshift.A final point regarding interpretation of the findings is in order. A finding of increased volatility (using any of the VIs) may be subject to the claim that exogenous shocks accounted for part of the increased volatility.V. PRESENTATION OF VOLATILITY INDEXESGovernments in Latin America undertook abrupt FL in the mid-1970s as a key component of full-scale economicliberalisation programs. Financial markets were liberalised in Uruguay in 1973, in Colombia and Venezuela in1973-75, in Chile in 1974-75, in Brazil in 1976, in Argentina in 1976-77, and in Mexico in 1977. It is commonlyheld that Asian countries' FL experiments of the early 1980s were more successful than those of Latin America.(16) Asian governments were apparently able to benefit from the mistakes of the abrupt deregulation undertakenin Latin America. In general, financial markets were liberalised more gradually in Asian countries. FL began in1978 in Malaysia, followed by the Philippines in 1980, South Korea in 1981-83, and in Indonesia in 1983-84.(17)In the countries investigated below, comprehensive FL programs having the following attributes wereimplemented: interest rate and loan ceilings were removed on some loans and deposit accounts, governmentalcredit allocation programmes were dismantled or greatly reduced, financial markets and institutions werediversified and deepened so that pre-existing institutions could participate in a wider array of activities,regulations were removed with the intent of encouraging the flowering of new institutions such as brokeragefirms, markets (such as futures and options), and instruments (such as commercial paper), fixed or multipleexchange rates were removed, and measures aimed at encouraging that conditions of competitiveness and freeentry prevailed in the financial system [Fry, 1988: 62-3]. It should be noted however that FL is a complex andevolving process which differs across countries, particularly in terms of sequencing and timing of the individualreforms. This of course complicates empirical tests of FL, inducing the problem of dating the regime shift. Thedates used here are from the International Monetary Fund (IMF), and reflect its judgement as to when (in eachcountry) policy changes surpassed the threshold at which we can meaningfully speak of FL.(18)

Search

Similar documents

Tags

Related Search

Assessing the impact of landfill leacheate onThe Impact of Differential Fees on EnrollmentThe impact of teachers qualification on studeThe Impact of New Technology on Higher EducatThe Impact of Social Networking on HealthThe Impact of Community Media on Social and PThe Impact of Intellectual Capital on the ProImpact of Macroeconomic Variables on Stock PrThe Impact of Niyogi Report on Hindu NationalThe Impact of Computers on Developmental Traj

We Need Your Support

Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks