China’s Service Trade and Dwindling Current Account Surplus

GuanTao(管涛)andWangXiaotong(王霄彤).................................................................................................

Abstract: Recent years have seen sharp increases in China’s trade deficit in services, especially in tourism, arousing concerns over a potential capital flight from China. Such concerns appeared to be justified by China’s balance of payments data, whose statistical scope was officially modified retroactively since 2014. To overcome the impact of changing statistical approach, this paper uses banks’ cross-border customer payment and receipt data to re-estimate China’s service trade, and finds that China’s service imports did not increase abnormally from 2014 to 2016 as some researchers reckoned. With the re-estimated service trade data, we find that the relationship between China’s GDP per capita and travel imports as a share of service imports since 2001 tallies with those of emerging Asian industrial countries before the Asian Financial Crisis in 1997, indicating that China’s travel import growth in recent years did not deviate from the long term trend and international experience. Furthermore, this paper finds that China’s travel imports did not significantly influence the stock and housing markets in the US, Hong Kong and the EU as major travel destinations, thus refuting the argument that massive capital has fled China for overseas markets.

Keywords: service trade, tourism trade, capital flight, external balance JEL Classification Codes: F1, F3, F4

DOI: 1 0.19602/j .chinaeconomist.2020.05.02

1. Introduction

Since peaking in 2007, China’s current account surplus as a share of GDP has steadily shrunk by 8.6 percentage points to reach 1.3% in 2017 and further dipped to -0.4% in the first half of 2018, turning into a current account deficit rarely seen over the past couple of decades.

Judging by international balance of payments data, China’s trade deficit in services is chiefly to blame for its dwindling current account surplus as a share of GDP. Since 1998, China has been running a persistent service trade deficit, which amounted to 292.2 billion US dollars in 2018 (see Figure 1), up 276.9 billion US dollars from 2009, or 1.42 times the reduction in its current account surplus by 194.1 billion US dollars from 2009 to 2018. Swelling tourism trade deficit is a main contributor to China’s bulging service trade deficit. Since 2009, China has been running a tourism trade deficit, totaling 237 billion US dollars in 2018, up 232.9 billion US dollars from 2009 (see Figure 1), or 84% of the increase in service trade deficit from 2009 to 2018.

The IMF (2017) hailed China’s falling ratio of current account surplus to GDP as an achievement of proactive economic rebalancing, and forecasted the ratio to drop below 1% to reach 0.2% by 2022. Yet China’s economic rebalancing has also raised doubts. Since most service trade is settled with commercial

papers, the authenticity of transactions is hard to verify, foreign exchange flows often elude supervision, and growing service trade deficit becomes an implicit conduit for illegal capital flight (Yu, 2017). With an overestimated service trade deficit, China’s current account surplus might be higher than what official data suggest (Setser, 2016).

Economists at the US Federal Reserve have conducted a quantitative analysis of China’s service trade deficit. Wong (2017) contends that anomalies in China’s tourism trade deficit from 2014 to 2016 contain a capital flight for purchasing overseas financial assets but disguised as tourism transactions. Adjusting for the capital flight in travel imports, China’s current account surplus as a share of GDP would be about 1% higher than China’s official figure in 2016.

This paper attempts to employ more accurate data to exclude the impact of change in statistical approach and other factors to re-estimate China’s actual service trade deficit and uncover the reality behind anomalies in the service trade deficit.

2. Fed Reserve’s Research Approach May Be Flawed

Most studies on capital flight have employed either of the following two methods: first, the World Bank’s residue method, in which “outward capital flight occurs when sources of funds exceed uses of funds based on international balance of payments data” (Claessens, 1993; Kant, 1998; Kar and Dev, 2013); the other method is to compare goods trade data reported by a home country and its trading partners to estimate the amount of illegal capital flight that evades supervision (Bhagwati, 1981; Cardoso and Dornbush, 1989; Song, 1999). Due to scant service trade data, it is uncommon for capital flight to be estimated with service or tourism trade data, but capital flight through tourism transactions does exist in the real world. For instance, the amount of capital flight can be estimated with tourism transactions data, and applied to the US balance of payments data of 1861-1900 (Simon, 1960). Given the rising share and importance of trade in services and tourism, this paper will estimate the size of China’s capital flight through trade in services as a response to the Fed Reserve’s study.

2.1 Potential Problems with Wong’s Approach

By comparing China’s international balance of payments data and relevant data of its trading partners, Wong (2017) finds an abnormal increase in China’s travel imports over the 2014-2016 period, which cannot be explained by statistical factor or economic fundamentals. This increase is negatively correlated with China’s economic growth and positively correlated with expectations for renminbi depreciation. The implication is that growth in China’s travel imports is unlikely to have stemmed from overseas service consumption by Chinese nationals and should be attributable to their purchase of overseas financial assets instead. That is to say, there is a capital flight from China disguised as travel imports. If such a capital flight is attributed to capital account rather than current account, China’s actual current account surplus as a share of GDP in 2015 and 2016 should be 1% higher than what data suggest.

Yet in 2016, China changed the statistical approach for tourism trade by incorporating bank card payment records and surveyed data about the ratio of cash spending during international travels. The new statistical approach also captures investment transactions disguised as travel spending, such as overseas property and insurance investments, as much as data were available. Data of 2014 and 2015 have been retroactively adjusted (International Balance of Payments Analysis Group, SAFE, 2017). Change in the statistical approach alone, therefore, would inevitably lead to an increase in travel imports in 2014, which cannot be overlooked even if a capital flight had inflated China’s travel imports as argued by Wong.

2.2 Potential Problems with Wong’s Research Model

Wong’s (2017) study is based on two models. One is a trade mirror model that estimates China’s import data from the export data of countries from which China imports tourism services. The assumption is that China’s travel imports from a foreign country mirror the country’s travel exports. If there is a misstatement in a country’s travel export data, which stems from a capital flight disguised as travel spending, there must be a discrepancy between China’s travel imports and the trading partner’s travel exports, which should be the size of capital flight. Wong estimated China’s capital flight during 2014-2016 with this method.

Yet the reality is that most countries do not distinguish two-way trade in their trade data. Under the tourism account, they only provide import and export data with the rest of the world without detailing trade with individual countries. In using the trade mirror model, Wong employs the travel export data of China’s major trading partners to the rest of the world rather than specifically to China. Given the rapid growth in China’s outbound travels amid the booming economy, China’s travel imports would be apparently underestimated if the average level of overseas travels from the rest of the world is to be followed. The so-called mirror data would not tally with China’s official data and would lead one to conclude that massive capital has fled China under the disguise of travel spending.

Another model is the gravity model in trade theory, which mainly applies to goods trade but is instead used by Wong to estimate the size of tourism trade. Based on an analysis of the gravity equation with two-way tourism trade data created by the World Bank for 199 countries tracing back to 1985, Wong has estimated the extent to which tourism trade is influenced by factors like GDP, distance and language. With China’s available two-way trade data updated till 2011, Wong has estimated China’s travel data of 2014-2016 based on coefficients from the gravity equation, and the results are smaller than China’s official data.

However, the gravity model is not without problems. First, the coefficients from the model are average values for various years and cannot reflect rapid growth in China’s economy and tourism trade in more recent years. Second, the gravity model is normally employed for analysis of goods trade rather than service trade, and the determinants of service trade, such as distance and language, may not be the same with those of goods trade. The model’s applicability to tourism trade is, therefore, to be verified. According to SAFE data, China’s top 10 trading partners in 2017 were China’s Hong Kong,

the US, Japan, the UK, Australia, Germany, South Korea, Canada, Singapore and China’s Taiwan. Except for a small surplus with Singapore, China ran service trade deficits with all the other nine trading partners, including the US, China’s Hong Kong, Australia, Canada, Japan, the UK and Germany by the descending order of service trade deficits (International Balance of Payments Analysis Group, SAFE). Obviously, the magnitude of China’s service trade deficits has little to do with distance. Third, with the gravity model, Wong employs coefficients based on 1980-2000 data to estimate 2000-2016 data, but such a simple counterfactual test lacks scientific basis since coefficients from simple regression are only applicable to sample years while coefficients for other years may change.

3. Re-estimated China’s Service Trade Data with Banks’ Customer Payments Data

The most critical problem with Wong’s paper is that it employs non-comparable scopes of data with a sudden change in the statistical scope at a certain time point. As mentioned before, SAFE started to “formulate travel income and spending data based on data collected from travel payment channels, including credit cards and debit cards, remittances and cash,” and retroactively adjusted data of 2014 and 2015 (International Balance of Payments Analysis Group, SAFE, 2017). Therefore, there is an abrupt change in the scope of international balance of payments data in 2014 compared with 2013. In case of an external shock, its impact cannot be separated from the changing statistical scope. Given the change in statistical scope, this paper will employ banks’ cross-border customer receipts and payments data to estimate tourism trade.

Unlike the international balance of payments data, there is no major change in the statistical scope of banks’ cross- border customer payments data, which are free from the interference of changing statistical scope. Moreover, banks’ cross-border customer payments data are more consistent with actual tourism trade data. While tourism trade occurs through credit card, debit card, remittances and cash, banks’ customer payments data include all such data except for cash.

Of course, banks’ customer payments data are not exactly the same with international balance of payments data. Customer payments data only include service trade income and spending of the nonbanking sector (including the non-banking financial sector, corporate sector and household sector) processed through banks, excluding the service trade income and spending of the banking sector itself. Another major difference lies in the calculation of freight cost and insurance premiums, which are accounted as part of goods trade in banks’ customer income and spending data. Yet in the international balance of payments accounting, freight cost and insurance premiums are accounted as service trade. By common practice, freight cost is about 4% of import value, and insurance premiums are roughly 1%. To make banks’ cross-border customer receipt and payments data more comparable with international balance of payments data, this paper accounts 5% of goods trade in the customer payments data under service trade. In this manner, this paper calculates service imports in banks’ cross-border customer payments after adjusting for freight costs and insurance premiums. Since tourism trade makes up 50% to 90% of overall service trade, banks’ customer service trade data may serve as a proxy variable of China’s travel imports.

As can be learned from Figure 2, China saw a significant increase in its service trade deficit as a share of GDP in 2014, which corresponds to what Wong describes as anomalies in tourism trade. However, banks’ cross-border payments data suggest a steady increase in China’s service trade deficit (except for a decrease in China’s capital inflow and outflow over recent couple of years). Before 2014, banks’ customer payments data significantly exceeded the international balance of payments data, but after the adjustment of statistical scope in 2014, the two started to converge. The implication is that banks’ customer payments data adjusted for freight cost and insurance premiums are more temporally consistent and better data for depicting tourism trade (the underestimation of service trade spending