From “One Country, Two Systems” to Monetary Integration?

1. From “One Country, Two Systems” to Monetary Integration? A Talk at the Hong Kong Institute for Monetary Research (presenting parts of the findings of my research fellowship in October-December 2001) Tsang Shu-ki Department of Economics Hong Kong Baptist University www.hkbu.edu.hk/~sktsang 31 January 2002

2. Introduction • One country, multiple monetary systems • Hong Kong and Mainland China: one country, two currencies • Hong Kong , Macau and Mainland China: one country, three monetary systems • And with Taiwan? • Historical precedents of multiple currencies: civil war times

3. The facts • The dual system in Hong Kong-Mainland China does have some novel features and asymmetries. • Hong Kong is an international financial centre, with the 4th highest per capita GDP in the world. Mainland China is a reforming socialist market economy with capital controls and a good deal of institutional rigidities remaining.

4. The facts • The Hong Kong special administrative region (SAR) is to decide its own monetary policies in accordance with Articles 110-113 of the Basic Law. • “The Hong Kong dollar and the Renminbi will circulate as legal tender in Hong Kong and the mainland respectively. The HK dollar will be treated as a foreign currency in the mainland. Likewise, the Renminbi will be treated as a foreign currency in Hong Kong.” (Chen, 1996)

5. The facts • “…. one country, two currencies, two monetary systems and two monetary authorities which are mutually independent" (Yam, 1996). • Before 1978, the existence of the Hong Kong dollar as a convertible currency served China well. As much as ¼ to 1/3 of Chinese foreign exchange earnings came from/through Hong Kong. • After the launching of the economic reforms, Hong Kong has become China’s most important trading port and “external investor” as well as a stepping-stone for foreigners.

6. Currency substitution versus transaction convenience • In the earlier years of reforms, hoarding of HK dollars by mainlanders might have been driven by a fear of the devaluation of the Renminbi, and therefore can be regarded as a form of “currency substitution” . • The situation now is more that of “transaction convenience” with the credibility of the Renminbi on the increase. The higher degree of economic integration means that it would reduce transaction costs for Chinese parties to accept and to store Hong Kong dollars.

7. Currency substitution versus transaction convenience • The other side of the story must also be told: Renminbi is also increasingly accepted for transaction purposes in Hong Kong. • Unlike China, of course, Hong Kong allows the circulation of foreign currencies although the Hong Kong dollar is the only legal tender. • Table 1 gives an estimate of HK dollars circulating in S. China. Y is the currency to GDP ratio, and X is the series of years from 1966 to 1987. The assumption is that the C/GDP ratio would converge to the long run average of about 4% in mature economies.

9. Theories updated • As Barandiaran and Tsang (1997) argued, supporting the status quo amounts to addressing critically the arguments for monetary unification, the alternative to the coexistence of the two currencies. • To be politically correct, unification would mean the demise of the Hong Kong dollar. • The situation is unlike Europe's ongoing economic and monetary integration, in which politics has played a dominant role (Wyplosz, 2001).

10. Costs and benefits of monetary unification • The benefits of unification are related mainly to (a) the transaction costs of currencies and (b) the risk posed by exchange rate variations. Unification would reduce the transaction costs and the risk of exchange rate variations only between the Hong Kong dollar and the Renminbi but not between the Renminbi and other currencies. • According to Barandiaran and Tsang (1997), the net benefit could well be negative for Hong Kong.

11. Forms of monetary coexistence • Barandiaran and Tsang (1997) distinguished three forms of coexistence : (1) spontaneous competition, (2) legal competition, and (3) monopoly. • They characterised the status quo as (1) spontaneous competition although it was rather limited to southern China.

12. (1) Spontaneous competition • It turns out that spontaneous competition has continued after 1997 but currency substitution is largely replaced by transaction convenience. • Further integration plus developments in Hong Kong after the East Asian crisis: “Let’s go north and consume and work!” • Equalisation of factor and output prices?

13. “Failure” of (2) legal competition and (3) monopoly • Failure of options (2) (legal competition) and (3) (monopoly) is not surprising. • With rising confidence about the Chinese economy and concern about “political correctness”, option (2), giving the HK dollar the status of legal tender or legally allowing HK dollars to circulate in parallel with the Renminbi in the Mainland, is really a non-starter. • Monopoly is the official position. But given the difficulties of strict implementation and the informal benefits of “transaction convenience”, in S. China at least, why bother to crack down on spontaneous competition, a game in which the Renminbi is not losing?

14. China-HK as an OCA? • The theory of optimum currency areas (OCAs) Pioneered by Mundell (1961) and McKinnon (1963). • The push towards euroisation in Europe gave rise to a new generation of theories and applied works (see Talvas, 1993 and Lafrance and St-Amant, 1999).

15. The empirics of OCAs • Methods to test where regions form an OCA range from statistical analyses to cointegration and VAR procedures. Given the constraints of short Chinese data series, I can only apply two methods (Tsang, 2002): • Variance analysis (VA): looking at the standard deviations of different variables across the regions concerned. Growth and inflation, e.g. should not show undue variations within an OCA. • Principal components analysis (PCA): examining the results of a “common shock” to the different variables across the regions concerned.

16. The empirics of OCAs • The principal components analysis (PCA) approach we have used aims at decomposing the common shocks to an economic variable in different regions into positive (symmetric) and negative (asymmetric) shocks. This is complementary to the von Hagen and Neumann's (1994) individual shock approach, and an extension of Caporale (1993) by Ma and Tsang (1997).

17. The empirics of OCAs • To generate estimates of the shocks or innovations to a set of economies, one can use the following vector auto-regression (VAR): yt =  + .yt-1 + t (1) where yt is a vector of values for an important economic variable, say GDP or RPI, over different economies or regions,  represents a vector of constants,  is coefficient matrices, and t is vector of disturbance terms.

18. The empirics of OCAs • The estimated t 's are subjected to principal components analysis after they are normalised so that their expected value is equal to 0 and their variance equals 1 (et). The normalised eigenvectors, f1, ... fM, of the correlation matrix of the et's are called the loading factors. The squares of the loading factors show the weight applied to each component in expressing each series as a function of the components.

19. The empirics of OCAs • Hence the square of fij, which is the i-th element of fj, represents the percentage of the variance of region i's variable explained by j-th principal component included in the model. These squares show the percentages of the fluctuation of the economic variable that can be explained by "common shocks", i.e. the "principal components", or shocks that have economy-wide effects.

20. The empirics of OCAs • As the eigenvectors, or the loading factors, are normalised so that Σj1fij2=1 for any given region i, the calculated symmetric and asymmetric shocks for each region are also normalised. The decomposition into symmetric shocks for region i is given as follows:

21. The empirics of OCAs

22. The empirics of OCAs • If the symmetric reaction outweighs the asymmetric reaction for a particular region, it would constitute a piece of evidence that the region could derive net benefit by being a member of the wider community, say, an OCA. • In contrast, if the negative (asymmetric) component outweighs the positive (symmetric) component, the region would react to a “common shock” in the opposite (or very different) direction compared with the other regions in the hypothetical OCA. It would have suffered instability if it had joined the OCA.

23. The empirics of OCAs • Some controversies surround this empirical interpretation. Diversification may be good for an OCA: if there are deep adjustment mechanisms (e.g. fiscal smoothing or long-term capital markets)--the two Mundell’s (1961 versus 1973a, b) a là McKinnon (2001). • However, there is also the issue of the “endogenous OCA” versus the “exogenous OCA”. • I adopt Mundell’s (1961) criteria here.

24. The findings for China-HK • Five sets of variables are used in the empirical tests which span the period of 1978 to 1999 or 2000 (remember they are all regional data): • (1)GDP (annual); • (2)Fixed asset investments (annual); • (3)Export trade (annual); • (4)Monetary variables: deposits and loans (annual); • (5)Prices: consumer price indexes (CPI’s) (monthly) and stock prices (daily).

25. The findings for China-HK • China now has 27 provinces and four municipalities under central control (Beijing, Tianjin Shanghai and Chongqing), making up a total of 31 “regions”. But data for some provinces or municipalities are too short or sporadic to be of any use. I have reported in each of the tables of the attached paper about the regions used.

26. The findings for China-HK • The preliminary results, as presented in the 14 tables of Tsang (2002), point to a picture which is very sceptical of a positive answer. The empirical findings based on disaggregated historical data show some signs of real and nominal convergence (GDP and inflation) only between Hong Kong and some parts of Eastern China, driven by trade (see Tables 2, 6 and 13 of Tsang (2002)).

27. The findings for China-HK • This seems to be consistent with the actual trajectory of economic integration, which is not detected in Liang (1999) and Ma and Tsang (1997). • However, all other results, even those based on higher frequency consumer and stock price data, fail to provide evidence that Hong Kong and the Mainland as a whole as yet constitute an OCA.

28. Forms of Monetary Integration • I. The status quo • II.The pegging of the HK dollar (or Macau’s pataca and the New Taiwan dollar) to the Renminbi • III. A national currency, “chino”, could be launched and the Renminbi, HK dollar, pataca and NT dollar might be pegged to it at agreed rates • IV. Reminbi-sation or chino-isation.

29. Forms of Monetary Integration • Options I, II, and III would meet no legal problems for Hong Kong (Tsang, 1999) and Macau. Option IV would require the amendment of their respective Basic Law. • For Taiwan, the constraints are political as well as economic. Option III is less unpalatable than option IV.

30. Option II for Hong Kong? • Should the HK dollar be re-pegged to the Renminbi when the latter becomes convertible? • The economic integration between Hong Kong and Mainland has been phenomenal. On the surface, they are one of the largest trading partners and external investors to each other. And the largest listed company in HK is China Mobile, accounting for over 10% of total stock market capitalisation.

31. Option II for Hong Kong? • However, the integration should not be exaggerated. Trade-wise, for example, intermediate trade has dominated, while the US remains a crucial market for end products after the outward processing portions are netted out, as shown in the following Table 2. • From the Mainland’s perspective, Hong Kong is also not its largest trading partner, as shown in Table 3.

33. Table 3Destinations and Sources of China’s Exports and Imports in 2000 (%) Exports Imports Total trade Japan 16.7 18.4 17.5 US 20.9 9.9 15.7 HK 17.9 4.2 11.4 S. Korea 4.5 10.3 7.3 Germany 3.7 4.6 4.2

34. Option II for Hong Kong? • There is also the problem of trade invoicing. Most of Hong Kong and Mainland China’s external trade transactions are invoiced in the US$ or the Yen. • The problem can be alleviated if both the Mainland and Hong Kong’s external trade is invoiced in similar proportions to the US$ and Yen and other currencies and both sides practice the same exchange rate regime. • This does not seem to be case.

35. Option II for Hong Kong? • So far, intra-trade and invoicing do not justify any further strengthening of monetary integration. However, a recent proposal is to have a Hong Kong-Mainland China Free Trade Agreement. That proposal, if implemented, would have the end result of promoting intra-trade, and hence enhancing the net benefit of monetary integration. • Over the long term, what needs to be monitor is the patterns and proportions of external and internal trade, even after the Renminbi has become full convertible.

36. Monetary union? • Options III and IV represent steps towards a monetary union. The highest common denominator of the literature suggests that a monetary union requires, amongst other things such as political considerations: • (1) real convergence; • (2) nominal convergence; • (3) risk-sharing mechanisms against asymmetric shocks; and • (4) factor mobility.

37. Real versus nominal convergence • A monetary union requires economic convergence (Mundell, 1961). However, there is a controversy on what “convergence” means. The key contention is that between the so-called “traditional view” (De Grauwe, 1996) which stressesreal convergence (GDP growth, investment trends, trade, factor mobility etc.) against the new view (Mélitz, 1988) which emphasizesnominal convergencefor the monetary union (inflation, interest rates, exchange rates etc.).

38. Real versus nominal convergence • A compromised view is presented in the 2000 Annual Report of the Eesti Pank (Bank of Estonia, 2001). Both are necessary but there may be conflicts. One example is that of real growth versus inflation. • Economists in Estonia, one of the candidate countries in the enlargement process of the EU, are talking about real annual growth rates of about 3%, and inflation rate of about 2%, above the EU averages. One of the main reasons is the trade-oriented real growth pattern of those candidate countries and the so called “Balassa-Samuelson effect”. (UNECE, 2001).

39. Real versus nominal convergence:the Case of the Mainland and HK • Regarding real convergence, the prospects of look “good” (for the Mainland, not for HK) in terms of GDP. Other requirements will be discussed later. • As to nominal convergence, the growth of both Hong Kong and the Mainland has been externally-oriented. Both suffered from strong inflationary bouts in the past two decades. That brings into consideration: • 1. The Balassa-Samuelson (B-S) effect; and • 2. The Dutch disease.

40. Nominal convergence for monetary union? • Both arise from sharp increases in the demand for tradables, which raises their prices. Price and income effects spill over to nontradables, where growth in productivity is slower. Hence export-led economies tend to suffer from an inflation bias (Sachs and Larrain, 1993). • The difference between the B-S effect and the Dutch disease is that in the latter, the demand is for “new” tradables, which crowds out the traditional tradables. The inflationary pressures tend to be stronger under the Dutch disease.

41. Nominal Convergence in Mainland China-HK? • Imai (1999) investigates the case of inflation in HK and comes out with the conclusion that it was due to Dutch disease rather than the B-S effect. In other words, HK’s tradables (services to an opening China) were “new”. • They may no long be so in the future. Hence the inflationary bias in HK may disappear. • Another consideration is the “equalisation of factor prices” across an increasingly permeable border.

42. Nominal Convergence in Mainland China-HK? • Guillaumont Jeanneney and Hua (2001), study Chinese provincial inflation differentials in 1992-99 and find that the B-S effect cannot be rejected. Although it is not a study of China as a whole vis-à-vis the rest of the world. It does not seem too far-fetched to speculate that the Balassa-Samuelson effect is also working in that context. • With WTO accession and the further opening up of China, the B-S effect may persist, which would mean that price convergence between HK and the Mainland may come in the future.

43. Risk-sharing in a monetary union • In a monetary union, the giving up of monetary and exchange rate policies need to be balanced by risk-sharing mechanisms among members, unless goods and factor prices are very flexible, unlikely even given nominal convergence. These include: • (1)Fiscal smoothing mechanism such as either independent fiscal policies among members (Beetsma and Bovenberg, 1997) or fiscal federalism by a central authority using taxation policies, fiscal transfers and insurance schemes, e.g. progressive federal income taxes (Hallett, Hutchison and Jenson, 1999). 

44. Risk-sharing in a monetary union • (2)Market-based mechanisms through the functioning of financial and credit markets, which share risk among different member regions through cross equity holdings and borrowing and lending, (Athanasoulis and van Wincoop, 1998). • Smoothing through fiscal transfers and credit markets is found to be higher in Canada than that in the US (Antia, Djoudad and St-Amant, 1999), while it is achieved mainly through capital and credit markets in the US (Asdrubali and Kim, 1999).

45. Risk-sharing in a monetary union • Independent fiscal policies and fiscal federalism are actually two contrasting views. Beetsma and Bovenberg (1997) and Fatás (1998) stress that federalism may weaken fiscal discipline consistent with a monetary union, while Bayoumi and Masson (1997) argue for the superiority of a nation-wide fiscal stabiliser.

46. Risk-sharing in Mainland-HK? • Independent fiscal policies are the norm under the Basic Law, which rules out tax-transfers mechanisms under fiscal federalism. However, the fiscal policies of Mainland China and Hong Kong are very different. In a monetary union, Hong Kong might “suffer” from fiscal expansionism in Mainland China, if the present trends continue. Then Hong Kong may have to practice a more pro-active fiscal policy to offset asymmetric shocks, which is however constrained by Article 107 of the Basic Law.

47. Risk-sharing in Mainland-HK? • Hong Kong would then have to rely on market-based risk-sharing, through the capital and the credit markets. Then it depends on the degree of integration between the financial and credit markets in the two economies. Presently, the integration is relatively one-sided: Chinese enterprises get listed in Hong Kong, absorbing Hong Kong capital; while Hong Kong banks lend to the Chinese side, more than the other way around. A better two-way flow is required for smoothing in a monetary union.

48. Labour Mobility in a Monetary Union • Factor mobility is a key to monetary union, as stressed by Mundell (1961) and the traditional view (De Grauwe, 1996) about real convergence. Labour mobility is often stressed. • Some would regard the lack of labour mobility as the Achilles heel of the EMU in the multi-cultural and multi-linguistic Europe.

49. Labour Mobility in a Monetary Union • No such cultural and language barriers exist between Mainland China and Hong Kong. What exist are political and economic barriers. The 150/day inbound quota to Hong Kong may have to be balanced with labour mobility northwards to attain a real convergence that is beneficial to Hong Kong and aspiring migrants and renders a monetary union an optimal choice.

50. Concluding remarks • The empirical results presented here suggest that Mainland China and Hong Kong are yet to constitute an OCA, despite some limited evidence of convergence with Eastern China. • Various possible forms of monetary integration are set out for the future in the eventuality of the full convertibility of the Renminbi. • Even for the re-pegging of the HK dollar to the Renminbi (option II), there does not seem to be any pressing need, unless intra-trade surges and other conditions prevail.