A. FDI and Exchange Rates: A Review of the Literature

There are relatively few papers in the literature on FDI determinants that consider how changes in exchange rates might affect export-platform FDI.⁷ This is the channel implicitly assumed in our model. The mechanism we suggest is the following: Imagine a firm whose net-worth, or expected future stream of income flows, is affected by the exchange rate. Hence, the rate of return on capital in such a firm also changes with exchange rate movements, which in turn affects the level of investment in the firm, whether foreign or domestic. A typical example is the export industry. As domestic exchange rates appreciate, exports become less competitive compared to their foreign equivalents, reducing the expected future stream of income of the firm that produces them. We would therefore expect investment in this industry to go down.

The empirical literature on FDI has made considerable progress in explaining the patterns observed in the data, nevertheless without firm theoretical basis. The majority of these studies examine the determinants of FDI destination by firms, both at regional (within a country) and country level using reduced form regressions. Among these determinants, the most common are the market-size of the recipient region or country, regulations (such as taxation laws and trade barriers), institutions (such as legal protection of assets, corruption, well-functioning markets and infrastructure) and exchange rates. Our focus will generally be on exchange rates.⁸

One of the first papers to study this relationship is Froot and Stein (1991) who present a model with imperfect capital markets, in the sense that the internal cost of capital is lower than borrowing from external sources. Thus, an appreciation of home currency leads to an increase in firm wealth and provides the firm with greater internal funds for investment relative to rival firms in the foreign country that have experienced the devaluation of their currency. Thus, following a depreciation in its currency, a country will receive more FDI. The authors provide empirical evidence for their claim using inward FDI data for the US.

Blonigen (1997) suggests an alternative mechanism for the same result. He notes that FDI by a firm involves acquiring assets that are transferable within the firm and across markets, without a currency transaction (such as firm specific technology, managerial skills, etc.). Thus, a depreciation of the home currency will make these assets cheaper for foreign firms while leaving their nominal return unchanged (since after acquisition these assets can be transferred abroad at no cost and yield revenues in home currency). In other words, a depreciation of a country’s currency will induce a “fire sale” of such transferable assets to foreign firms operating in global markets, versus domestic firms that may not have such access. Blonigen uses industry-level data on Japanese mergers and acquisition FDI into the US to test this hypothesis. He finds strong support for the existence of increased inward US FDI by Japanese firms in response to real dollar depreciations relative to the yen. From a more macroeconomic perspective, Eichengreen and Tong (2005) cite anecdotal evidence that the depreciation of East Asian exchange rates encouraged fire-sale FDI.

B. Evidence in Favor of Export-Platform FDI

There is strong evidence in favor of export-platform FDI in East Asian and Latin American economies, and particularly in China.⁹ According to Whalley and Xin (2006), Foreign Invested Enterprises (FIEs) account for over 50 percent of exports from and 60 percent of imports into China; although they employ only 3 percent of the workforce.¹⁰ According to Hanson, Mataloni and Slaughter (2001), when compared to FDI into OECD and Latin America, the exports-to-sales ratio of FIEs in East Asia is the highest. Their data, which covers a period up to 1998, shows that in East Asia 50 percent of sales of FIEs are exported to third countries. This figure was 40 percent for Mexico and 30 percent for the rest of Latin America.

It is likely that the competitiveness of the export sector could be a major explanatory factor for the co-movements we observe between FDI flows and exchange rates. In summary, the literature has derived some important firm-level models of how exchange rate levels or volatility can affect FDI flows, but there is little work to assess the policy relevance of any of these theories.

C. Consumption

The second issue we would like to consider is the effect of increasing consumption on capital flows. An increase in consumption will cause interest rates to rise, since for the same level of output households now invest less. Higher interest rates imply lower asset prices, which tend to attract FDI. This effect is unambiguous. Consider now a simple thought experiment in which nominal exchange rates are revalued. Thus the foreign currency price of domestic assets increases. However, a revalued currency also increases consumption by households because their purchasing power has increased. Increased consumption, as explained above, will tend to increase FDI through lower asset prices. Therefore, the overall effect of a revaluation policy on FDI flows is ambiguous.

It is interesting to see whether such policies are also advocated in reality and if so, what they have achieved. There is an ongoing debate in the literature about the appropriate policy tools to “correct” global imbalances. Proposed policies include those aimed at boosting consumption, including revaluations of the exchange rate. This is consistent with our story, although we maintain that the effect of a nominal revaluation on the direction of net capital flows is ambiguous.¹¹

We mentioned already the intention of the Chinese government to shift spending from investment to social spending and to oversee “an increased role for private consumption” (Kuijs, 2005). The IMF’s view, as stated in its Asia Pacific Regional Economic Outlook (2006), is that raising private consumption in emerging Asia plays a crucial role both for achieving sustainable growth in the region and for reducing the global imbalances, as observed through a comparison of current account figures across countries.¹²

A few stylized facts about consumption patterns in East Asia are important to note here. By most accounts, consumption as a share of GDP in East Asia is low compared to other regions (see Tables 1 and 2 in the appendix). Also, these ratios have been stagnant (or decreasing in the case of China) during the past decade despite rising incomes in the region. Finally, demographics play an important role in this trend, so that even without any countervailing government policies, the consumption share is set to rise as populations in the region age and begin to dissave.

The dominant view is that there was a surge in precautionary savings following the Asian crisis, which has held back consumption even though incomes have risen dramatically. The absence of social safety nets, under-developed financial markets that enable households to diversify risks, and a large proportion of the population saving heavily as they approach retirement age, have contributed to this precautionary measure.

A recent study by the IMF examines this common perception.¹³ Pre-crisis consumption levels, given economic fundamentals, are used to form out-of-sample forecasts for consumption levels after the crisis. The study finds that private consumption in East Asia recovered after the crisis, and in many cases exceeded the out-of-sample forecast. This offers evidence that precautionary saving declined as populations grew more confident about subsequent growth. The one case where the private consumption-to-GDP ratio is consistently below the out-of-sample forecast is China. The report concludes that precautionary savings in China may have risen permanently in the wake of the Asian crisis.

Overall, there seems to be a consensus, at least in policy circles, on the fact that consumption in East Asia is low by international standards. There also appears to be agreement on how to remedy the consequences of this fact—by shifting government expenditure from investment to consumption to speed up the transition to a sustainable equilibrium.

This paper offers a more rigorous evaluation of such policy recommendations. The policy advice has generally recommended an increase in government consumption expenditure, such as social safety nets, health care, and education. Here we will consider direct government transfers such as tax-cuts, which are generally simpler to implement in practice, and which should have similar welfare implications.

A. Production

There are 8 factors of production and 3 types of financial assets:

1-5. Capital types	9. Foreign currency
6. Urban labor	10. Rural labor
7. Domestic currency	11. Land
8. Bank deposits

View Table

1-5. Capital types	9. Foreign currency
6. Urban labor	10. Rural labor
7. Domestic currency	11. Land
8. Bank deposits

The five types of capital correspond to five aggregate non-agricultural productive sectors. We could have any number of capital types without affecting the structure of the model, our choice of five is essentially arbitrary.¹⁴ Each of these factors and financial assets is replicated in each period and, accordingly, has a price in each period. Period 1 domestic currency is the numeraire.

B. Current Production

An input-output matrix, A_t, is used to determine intermediate and final production in period t. Corresponding to each sector in the input-output matrix, sector specific value added is produced using capital and urban labor for the non-agricultural sectors, and land and rural labor in agriculture. The different factors are allocated across the economy so that agriculture uses land and rural labor, and all other sectors use one of the five capital types plus urban labor. Accordingly, capital is perfectly mobile across a given sub-sector, but is immobile across other sub-sectors. Labor, on the other hand, may migrate from the rural to the urban sector.

The specific formulation of the firm’s problem is as follows. Let y_kij, y_Lij be the inputs of capital and urban labor to the jth non-agricultural sector in period i. Let Y_Gt be the outstanding stock of government infrastructure in period i. The production of value added in sector j in period i is then given by:

where we suppose that public infrastructure may act as a productivity increment to private production.

Sector j pays income taxes on inputs of capital and labor, given by t_kij, t_Lij, respectively, in period i. Agriculture is taxed on its use of labor. Hence the effective price for labor and capital paid by sector j is:

Thus if ${\tilde{\mathit{P}}}_{\mathit{\text{Kij}}}$, ${\tilde{\mathit{P}}}_{\mathit{\text{Lij}}}$ are the prices of capital and labor in period i, then the prices charged by enterprises, p_i, are given by

Where va(P,Y_Gi) is the vector of cost-minimizing value-added per unit of output.

C. Domestic Investment

We suppose that domestic investors produce each type of sectoral capital via a sector-specific investment technology that uses inputs of capital and labor to produce new capital. Domestic investment is carried out by the private sector and is entirely financed by domestic borrowing in the home currency. Let us define the following notation.

t_ki= Profit (capital) tax rate (percent)

C_Hi= The cost of producing the quantity H of capital in period i.

r_t= The interest rate in period i.

P_Ki= The return to capital in period i.

P_Mi= The price of money in period i.

δ = The rate of depreciation of capital.

Suppose, then, that the rental price of capital in period 1, or whatever the current period might be, is P₁. The cost of borrowing must equal the present value of the return on new capital. Hence:

where r_j is the interest rate in period j, given by:

and P_Bj is the price of a bond in period j. Accordingly, the investor takes out a loan from the banking system to cover his costs. This loan then becomes an asset of the banking system.

D. Bankruptcy

We make one further assumption about the behavior of the individual firm. The firm, like all other agents in the model, optimizes with a 2 period time horizon for which it knows all prices.

After the second period it assumes that future interest rates and returns to capital will remain the same as in period 2. Hence P_Ki =P_K2, r_i=r₂; i>2. If at some point the present value of investment, as given in equation (3), falls below the corresponding value of debt service, then the sector is unable to pay its debt obligations which were incurred to finance this investment. Accordingly, the bank which holds these assets now holds corresponding bad debts. This situation might occur if, after the investment was incurred, the interest rate rose or the rate of return to capital fell, due to some unanticipated event. We assume that a bankrupt firm cannot invest.

E. Foreign Direct Investment

We model the foreign investor in an, admittedly, ad hoc way based upon the empirical estimates described in the next section. The foreign investor, considering direct capital investment in the home country, looks at 3 endogenous variables. These are the past (most recent) rate of growth in real GDP in the country, the current relative nominal interest rates in the home country and the rest of the world (the US interest rate), and the expected change in the home country’s exchange rate (against the US $). Thus we would expect that the change in FDI would be given by:

Here e_i denotes the exchange rate in period i while r_Fi denotes the foreign (US $) interest rate. GDP_dom is the home country’s GDP.

Thus if domestic interest rates are higher than foreign rates, then domestic asset prices will be lower than foreign asset prices, leading to an increase in FDI. Similarly, if a depreciation of the exchange rate is expected, then a decline in domestic asset prices is also expected. Hence we expect that α should be positive. The second term on the RHS is the domestic growth rate. The next section will show that, for the set of countries on which our analysis is based, this is also positively related to changes in FDI. Accordingly, β should also be positive.

FDI is allocated across sectors according to the existing share of the nominal value of sectoral capital in total nominal capital. Thus let W_Kij denote the total stock of capital of type j at the beginning of period i. Let KRATIO_ij represent the share of the nominal value of capital of type j in period i in the total nominal value of the existing capital stock, given by:

Then nominal FDI going to sector j in period i is given by:

Thus FDI “follows” prior capital formation in the sense that it is allocated according to existing shares of nominal capital. Clearly this does not reflect optimizing behavior on the part of foreign investors, who are following the investment decisions made by optimizing domestic investors, who have allocated past investments based on sectoral rates of return. Nonetheless, we believe that it is possible to justify this modeling approach by claiming that observed behavior of domestic investors may represent the best available information to the foreign investor.

F. Banking

The banking sector in our model is quite simple. We will suppose that there is at least one bank for each non-agricultural sector of the economy. Each bank lends primarily to the sector with which it is associated. The banks are, however, not fully specialized in the sector to which they correspond. We will make the simplifying assumption that each bank holds 50 percent of the outstanding debt of its particular sector. It then holds the remainder of its assets uniformly distributed across each of the remaining sectors in the economy. Suppose, for example, that there are 5 sectors. In this case bank 3, for example, holds 50 percent of the debt of sector 3, and 12.5 percent of the remainder of its assets are in each of sectors 1,2,4, and 5. We make this assumption of diversification of assets in order to avoid a possible situation in which the insolvency of a particular sector leads to the automatic insolvency of its related bank.

We will suppose that banks follow a strategy of lending that looks at the risks associated with their borrowers. That is, as their borrowers become more insolvent, the banks ration credit to those borrowers.¹⁵ We will choose a simple functional form that connects credit rationing to borrower insolvency. Suppose that C_Hij is the demand for borrowing by sector j in period i. Suppose also that bank k has DEF_ik percent of its total assets in default in period i. Let δ_k ≥ = 0 be a parameter specific to bank k, and let be βjk the share of borrowing by sector j taken by bank k. Sector j then receives loans L_ji where:

Thus if there are no bank assets in default, then no credit rationing takes place. If assets are in default, then the credit demanded by sector j for investment is reduced by each bank proportionally to the share of that bank’s defaulted assets in total assets. The parameter δ_k is a measure of the risk aversion of banks. Higher values of δ_k indicate a more rapid contraction of credit in response to bad loans. We impose a solvency requirement on the banking system. Namely, if α percent of a bank’s assets are in default, caused by a corresponding insolvency in its borrowers, then the bank is declared insolvent. At this point a fraction of the bank’s deposits are seized by the government.¹⁶ In particular, depositors in the bank find part of their deposits frozen. We use a simple rule to determine the fraction of a bank’s deposits that are seized. If DEF_ik is the share of bank k’s assets that are in default in period i, as before, then regulators seize ω_kDEF_ik of the bank’s deposits, where ω_k is a bank specific parameter. This seizure of deposits correspondingly reduces the bank’s ability to lend. Thus the bank’s supply of loans, and hence its assets, are determined by the demand for loans from the productive sectors of the economy, as well as the risk imputed to potential borrowers

G. Consumption

There are two types of consumers, representing rural and urban labor. We suppose that the two consumer classes have differing Cobb-Douglas demands. The consumers also differ in their initial allocations of factors and financial assets. The consumers maximize intertemporal utility functions, which have as arguments the levels of consumption and leisure in each of the two periods, subject to intertemporal budget constraints. The consumer saves by holding money, domestic bank deposits, and foreign currency. He requires money for transactions purposes, but his demand for money is sensitive to the interest rate. In addition, the consumer’s demand for bank deposits is sensitive to his perception of the solvency of the banking system. In particular, as banks increasingly incur bad loans, the consumer’s interest elasticity of money declines, causing him to reduce his bank deposits.¹⁷

In order to avoid unreadable subscripts, let 1 refer to period i and 2 refer to period i+1. The consumer’s maximization problem is then:

such that:

where:

P_i = price vector of consumption goods in period i.

x_i = vector of consumption in period i.

C_i = value of aggregate consumption in period i (including purchases of financial assets).

N_i = aggregate income in period i.

t_i = vector of sales tax rates in period i.

P_Lui = price of urban labor in period i.

α₂ = elasticity of rural/urban migration.

P_ki = price of capital in period i.

K₀ = initial holding of capital.

P_Ai = price of land in period i.

A₀ = initial holding of land.

δ = rate of depreciation of capital.

P_Mi = price of money in period i.

x_Mi = holdings of money in period i.

P_Bi = discount price of a certificate of deposit in period i.

π = domestic rate of inflation in period i.

r_i, r_Fi = the domestic and foreign interest rates in period i.

e_i = exchange rate in terms of units of domestic currency per unit of foreign currency in period i.

TR_i = transfer payments from the government in period i.

a, b, α, β = estimated constants.

d_i = constants estimated from model simulations.

DEF = The value of non-performing assets in the banking system.

ASSET = Total assets of the banking system.

c = a functional form that depends negatively upon the ratio of non-performing assets to total assets in the banking system.

The left hand side of equation (5a) represents the value of consumption of goods and leisure, as well as of financial assets. The next two equations contain the value of the consumer’s holdings of capital and labor, as well as the principal and interest that he receives from the domestic and foreign financial assets that he held at the end of the previous period. The equation C_i = N_i then imposes a budget constraint in each period. Equation (5d) is a standard money demand equation in which the demand for cash balances depends upon the domestic rate of inflation and the value of intended consumption. There is, however, one modification. The inflation elasticity, c, is inversely related to the share of non-performing bank assets in total assets. As non-performing assets rise, c declines.

Equation (5b) says that the proportion of savings made up of domestic and foreign interest bearing assets depends upon relative domestic and foreign interest rates, deflated by the change in the exchange rate. Finally, equation (5c) is a migration equation that says that the change in the consumer’s relative holdings of urban and rural labor depends on the relative wage rates.

In period 2 we impose a savings rate based on adaptive expectations, as in equation (5e). The constants (d_i) are estimated by a simple regression analysis, based on the previous periods. The constants are updated after each two period segment by running a regression on the previous t - 2 periods. Thus savings rates are endogenously determined by intertemporal maximization in period t, but are determined by adaptive expectations in period t + 1.¹⁸

When period t + 2 begins, the consumer’s holdings of financial assets may be different than those incorporated in the above problem, since defaults may have occurred. The consumer then optimizes again for periods t + 2, t + 3, based on his new, unexpected holdings of financial assets at the beginning of period t + 2.

H. The Government

The government collects personal income, corporate profit, and value added taxes, as well as import duties. It pays for the production of public goods, as well as for subsidies. In addition, the government must cover both domestic and foreign interest obligations on public debt. The deficit of the central government in period 1, D1, is then given by:

where S₁ represents subsidies given in period 1, G₁ is spending on goods and services, while the next two terms reflect domestic and foreign interest obligations of the government, based on its initial stocks of debt. T₁ represents tax revenues.

The resulting deficit is financed by a combination of monetary expansion, as well as domestic and foreign borrowing. If ∆y_BG1 represents the face value of domestic bonds sold by the government in period 1, and C_F1 represents the dollar value of its foreign borrowing, then its budget deficit in period 2 is given by:

where r₂(∆y_BG1 + B₀) represents the interest obligations on the initial domestic debt plus borrowing from period 1, and e₂r_F2(C_F1 + B₀) is the interest payment on the initial stock of foreign debt plus period 1 foreign borrowing.

The government finances its budget deficit by a combination of monetization, domestic borrowing, and foreign borrowing. We assume that foreign borrowing in period i, C_Fi, is exogenously determined by the lender.

I. The Foreign Sector

The foreign sector is represented by a simple export equation in which aggregate demand for exports is determined by domestic and foreign price indices, as well as world income. The specific form of the export equation is:

where the left hand side of the equation represents the change in the dollar value of exports in period i, π_i is inflation in the domestic price index, Δe_i is the percentage change in the exchange rate, and π_Fi is the foreign rate of inflation. Also, Δy_wi represents the percentage change in world income, denominated in dollars. Finally, σ₁ and σ₂ are corresponding elasticities.

The combination of the export equation and domestic supply responses then determines aggregate exports. Demand for imports is endogenous and is derived from the domestic consumers’ maximization problems. Gross capital inflows are determined by FDI, from the previous section. The supply of foreign reserves y_FGi, available to the government in period i is given by:

Here x_Fi represents the demand for foreign assets by citizens of the home country, so x_F(i-1)- x_Fi represents private capital flows.C_Fi represents exogenous foreign borrowing by the home government.

Finally changes in the money supply in period i,∆M_si, are now given by:

where ∆y_Mi is determined by the government’s financing of its budget deficit, and ∆OMO_i represents money created via open market operations. The remainder of the right hand side represents the domestic currency value of the balance of payments.

A. Methodology

In calibrating the IO matrices, we only used Chinese data as a very stylized example. To estimate the FDI flow elasticities however, we use a panel data set comprising of 11 countries in developing Asia.²⁰ Although we are interested in finding the elasticities of FDI flows within a specific country, we believe using a panel was appropriate for a few reasons: First, from a practical point of view, unlike the case for IO matrices, there is not enough annual data on FDI available for any single country to obtain estimates with reasonable confidence intervals. Second, the model does not aim to explain the economic movements of a particular country. Furthermore, by using a fixed effect panel data model for our estimation, we have restricted our attention to variations of FDI with respect to macro variables of interest. We have specified our estimation equation as follows:

The dependent variable, ∆FDI, is measured as the change in the GDP share of FDI from the previous year. FDI (%GDP) has been increasing rapidly over the last 2 decades, hence the data is highly non-stationary. The rate and the consistency of growth of FDI flows are not matched by the developments in any other macro economic indicator over the same period. Therefore, we first need to transform these series into a stationary measure. To do this, we use first differenced FDI as our dependent variable. As Figure 1 in the Appendix shows, the FDI series are clearly non-stationary both in East Asia and in OECD. This non-stationarity is almost entirely removed by first differencing the FDI series (Figure 2, Appendix).

The independent variables, GDP growth (% annual), real interest rates (% annual), and year-on-year average nominal devaluation (%) are all stationary and therefore are used in the regressions in their original form. We also experimented with several other variables on the right hand side, including inflation, and lending rates, as well as lags and leads of growth and devaluation. However, only in the parsimonious specification above did we obtain beta coefficients that were statistically significant and robust to different estimation methods.

Given that FDI series are integrated of order 1, regressing FDI on lagged FDI values and independent variables will result in “super consistency” of estimates, though we know that in small samples, the coefficient is biased downwards and the estimated t- and F- values are incorrect. Therefore, we rule out this specification. We also rule out regressing FDI on a time dummy and other independent variables, since this specification will give rise to a spurious regression. That is,. the standard errors are very low simply because the FDI series is increasing over time (time variance dominates the variance in every other stationary variable on the right hand side).

We used three main variations of the panel data model to estimate equation (8): First, we used a simple fixed effect regression with robust standard errors. In our second specification, we use a fixed effect model with first-order autoregressive disturbances.²¹ Finally, we use a feasible GLS model, which estimates the panel in the presence of AR(1) autocorrelation within panels and cross-sectional correlation and heteroskedasticity across panels. Specifically, we have assumed that, within panels there is AR(1) autocorrelation and that the coefficient of the AR(1) process differs across panels.

B. Data and Samples

We use data from three different country groups: Developing Asia, G20²² and OECD²³ for the period 1985-2005. Although we only use the estimates for the developing Asia in our simulations, the estimation for other country groups was carried out as a robustness check. We collected our data from the World Bank’s World Development Indicators database (WDI). Some regressions also used data on FDI from UN Foreign Direct Investment database. The results are almost identical.

C. Results

The results are presented in Table 1. P-values are in parentheses below each point estimate. We are only presenting the elasticities with respect to devaluation (efd), real interest rates (efr) and growth (efg), as presented in the final form of our regression. Other variables including inflation, lending rates and different lags of independent variables were not associated with FDI flows in a statistically significant way and hence their elasticities are not presented here.

Table 1.

Elasticities of Changes in FDI Inflows with Respect to Real Interest Rates, Growth and Changes in Exchange Rate

(1) Simple FE(2) AR(1) (3) FGLS

Table 1.

Elasticities of Changes in FDI Inflows with Respect to Real Interest Rates, Growth and Changes in Exchange Rate

	East Asia			OECD			G20
	(1)	(2)	(3)	(1)	(2)	(3)	(1)	(2)	(3)
efr	0.162** -0.02	0.186** -0.02	0.145** -0.02	0.041 -0.61	0.038 -0.68	0.03 -0.6	0.012 -0.41	0.013 0.577	0.008 -0.47
efg	0.198*** -0.01	0.229*** 0	0.179*** 0	0.213** -0.05	0.235** -0.05	0.117 -0.15	0.083** -0.04	0.102** 0.033	0.062** -0.04
efd	0.032* -0.06	0.039** -0.05	0.027* -0.06	-0.783 -0.6	-0.815 -0.62	-0.87 -0.5	0 -0.56	0 0.623	0 -0.65

(1) Simple FE(2) AR(1) (3) FGLS

View Table

Table 1.

Elasticities of Changes in FDI Inflows with Respect to Real Interest Rates, Growth and Changes in Exchange Rate

	East Asia			OECD			G20
	(1)	(2)	(3)	(1)	(2)	(3)	(1)	(2)	(3)
efr	0.162** -0.02	0.186** -0.02	0.145** -0.02	0.041 -0.61	0.038 -0.68	0.03 -0.6	0.012 -0.41	0.013 0.577	0.008 -0.47
efg	0.198*** -0.01	0.229*** 0	0.179*** 0	0.213** -0.05	0.235** -0.05	0.117 -0.15	0.083** -0.04	0.102** 0.033	0.062** -0.04
efd	0.032* -0.06	0.039** -0.05	0.027* -0.06	-0.783 -0.6	-0.815 -0.62	-0.87 -0.5	0 -0.56	0 0.623	0 -0.65

(1) Simple FE(2) AR(1) (3) FGLS

There are a few interesting points to notice about our results: First, the point estimates using different estimation methods are very close for all elasticities in all three regions. Except for one case (the elasticity of changes in FDI flows with respect to growth in G20), we either reject or accept the statistical significance of each independent variable in all three estimations methods. In developing Asia, we find a statistically significant elasticity of FDI with respect to all three variables of interest, namely changes in the official (nominal) exchange rate, real interest rates and GDP growth. These are the estimates used in our simulations. The signs of these elasticities are consistent with our explanation for the mechanism through which these variables affect FDI flows: A one percent higher real interest rate differential increases the change in FDI flows (%GDP) from the previous year by 0.2 percent. We expect this elasticity to be positive, as capital flows to a country with lower asset prices. A one percent nominal devaluation (increase in average nominal exchange rate with respect to the previous year) increases the flows by 0.03 percent. We interpret this channel as the effect of the devaluation on the competitiveness of export sector. Finally, growth matters for FDI flows: a one percent higher GDP growth increases our dependent variable by 0.2 percent contemporaneously.

In other country groups, namely emerging market countries of G20 and OECD countries, the only significant relationship is between changes in FDI (%GDP) with contemporaneous growth rate. Our interpretation of this result is that higher FDI each year is directly counted as higher investment in national accounting data. However, to the extent that there is serial correlation in growth rates, one hypothesis could be that higher growth this year would mean higher expected growth next year, which induced higher investment this year. It is also interesting to note that the elasticity of FDI (%GDP) growth with respect to the GDP growth rate are of the same order of magnitude for the three country sets: efg is estimated to be between 0.10 and 0.25 depending on the particular specification and the region.

Finally, we should emphasize that the relationships we establish do not provide a robust explanation of cyclical changes in FDI flows. However, for the purposes of calibrating our model they provide us with a reasonable range of values. Our model converges for all the values in the confidence intervals of our point estimates.

A. A Base-Case Simulation

In order to use our model for counter-factual simulations, we first generate an equilibrium, using benchmark policy parameters, over an eight-year time horizon.²⁵ We take initial allocations to have their values for China for 1996, and all government policy parameters also have Chinese values. We take tax rates to have their estimated effective values. Government current and capital expenditures are given their historical values, and are assumed to be maintained at the same real level for all periods of the simulation. We also suppose that the Central Bank maintains a fixed exchange rate, in U.S. dollars, with the rate being fixed at the level of the first year. A sector is unable to repay its debt, as in Section III.D, when the present value of the future stream of earnings from the investment becomes less than the corresponding debt obligations. Finally, we will suppose that the bank solvency requirement, α as in Section III.B, is 8 percent. Thus if a bank’s non-performing assets are greater than 8 percent of its total assets, then a portion of the bank’s deposits are seized and depositors are unable to retrieve that share of their assets.

As in Section III.B, if the bank’s borrowers default on their loans, then the bank loses ω_kDEF_ki of its deposits, resulting in a wealth shock to depositors. In order to be specific, we will let ω_k = 1 for all banks. We will suppose that banks do not ration credit when their borrowers begin to default. Table 2 reports the simulation outcomes for the main macroeconomic variables for this base case.

Table 2.

Base Case

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

Table 2.

Base Case

Period		1		2		3		4		5		6		7		8
Nominal GDP 1/		100.0		127.5		142.6		183.4		190.4		244.9		245.6		315.7
Real GDP 1/		100.0		107.2		110.9		119.0		123.1		132.0		136.6		146.2
Price Level		100.0		118.9		128.6		154.1		154.7		185.6		179.8		216.0
Nominal rate		0.9		2.7		1.7		7.4		2.3		10.5		2.8		12.6
Real interest rate				-13.6		-6.0		-10.4		1.9		-7.9		6.1		-6.3
Budget deficit 2/		7.7		6.9		9.3		8.3		10.6		9.6		9.8		9.6
Trade balance 2/		10.4		7.5		4.7		2.1		1.2		-1.1		-1.1		-3.2
Net capital stock at end of period 8				Percent of bank assets in default at end of period:								US$ FDI in Period 8 (Base period = 100)
				Period			4		6		8	105.4
Sector 1	100.0			Bank 1			0.0		0.0		0.0	Utility levels
Sector 2	100.0			Bank 2			0.0		0.0		0.0	Consumer
Sector 3	100.0			Bank 3			0.0		0.0		0.0	1	2		3
Sector 4	100.0			Bank 4			0.0		0.0		0.0	100.0	100.0		100.0
Sector 5	100.0			Bank 5			0.0		0.0		0.0

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

View Table

Table 2.

Base Case

Period		1		2		3		4		5		6		7		8
Nominal GDP 1/		100.0		127.5		142.6		183.4		190.4		244.9		245.6		315.7
Real GDP 1/		100.0		107.2		110.9		119.0		123.1		132.0		136.6		146.2
Price Level		100.0		118.9		128.6		154.1		154.7		185.6		179.8		216.0
Nominal rate		0.9		2.7		1.7		7.4		2.3		10.5		2.8		12.6
Real interest rate				-13.6		-6.0		-10.4		1.9		-7.9		6.1		-6.3
Budget deficit 2/		7.7		6.9		9.3		8.3		10.6		9.6		9.8		9.6
Trade balance 2/		10.4		7.5		4.7		2.1		1.2		-1.1		-1.1		-3.2
Net capital stock at end of period 8				Percent of bank assets in default at end of period:								US$ FDI in Period 8 (Base period = 100)
				Period			4		6		8	105.4
Sector 1	100.0			Bank 1			0.0		0.0		0.0	Utility levels
Sector 2	100.0			Bank 2			0.0		0.0		0.0	Consumer
Sector 3	100.0			Bank 3			0.0		0.0		0.0	1	2		3
Sector 4	100.0			Bank 4			0.0		0.0		0.0	100.0	100.0		100.0
Sector 5	100.0			Bank 5			0.0		0.0		0.0

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

We notice that our model generates moderate rates of growth in real GDP, with the average rate over the 8 years being 5.6 percent. The budget deficit rises slightly, and then stabilizes, while interest rates rise somewhat, in response to the higher budget deficit. The trade surplus disappears, and is replaced by a deficit as the fixed exchange rate becomes overvalued.²⁶ No borrower defaults on his loan and no bank fails. FDI gradually rises over time, as the average domestic interest rate is 5.1 percent, higher than the assumed US interest rate of 2 percent. Hence domestic assets are cheaper than foreign assets, given the fixed exchange rate. At the same time there is positive growth, also leading to an increase in FDI.

Before analyzing the policy experiments, two points are worth emphasizing here as they clarify some of the seemingly counter intuitive results. The first is related to changes in domestic vs foreign investment: while the foreign investor cares about nominal interest rates, only real interest rates matter for the domestic investor’s decision. This is because the foreign investor borrows abroad, at the assumed constant nominal rate of 2 percent, to finance his purchase of an asset in the domestic currency. His rate of return, therefore, is determined by the difference between the nominal increase in the price of the asset converted into dollars, less the 2 percent paid on borrowed funds. Thus, the foreign investor cares about nominal domestic rates as well as exchange rates. Whereas the domestic investor borrows by issuing domestic currency denominated bonds, on which she pays domestic nominal rates, to finance her purchases of assets, on which she earns the real rate plus inflation in prices. Hence only the real interest rate matters for her investment decision.

The second observation is that domestic investment can rise following a revaluation, despite rising real interest rates. The explanation here is that the income effect of a revaluation allows the agent to increase both his consumption expenditure (on cheaper imported goods) and investment. This will not necessarily be the case for all choices of labor-leisure elasticities.

B. Increasing Consumption by a Revaluation

Suppose now that the authorities decide to carry out a revaluation of the exchange rate. They might do so as a way to increase domestic consumption and welfare, or to reduce FDI. As an illustrative example, we carry out an annual revaluation of 9.5 percent.²⁷ All exogenous parameters, other than those having to do with the open market operation, remain the same as in the previous exercise. Table 3 gives the results of this exercise.

Table 3.

9.5 Percent Revaluation

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

Table 3.

9.5 Percent Revaluation

Period		1		2		3		4		5		6		7		8
Nominal GDP 1/		100.0		116.2		131.7		165.0		163.8		203.7		194.1		237.0
Real GDP 1/		100.0		107.0		111.0		119.0		122.9		131.3		136.5		146.1
Price Level		100.0		107.0		118.7		139.3		133.3		155.1		142.3		162.9
Nominal rate		-3.1		1.4		0.9		6.1		1.4		10.9		2.2		15.7
Real interest rate				-5.2		-9.0		-9.6		6.0		-4.7		11.4		1.1
Budget deficit 2/		7.7		7.7		9.9		8.8		11.8		10.7		13.8		12.7
Trade balance 2/		10.3		7.2		4.1		1.3		-1.2		-6.1		-5.5		-10.3
Net capital stock at 3/ end of period 8				Percent of bank assets in default at end of period:								US$ FDI in Period 8 (Base period = 100)
				Period			4		6		8	95.4
Sector 1	101.3			Bank 1			0.0		0.0		0.0	Utility levels
Sector 2	103.3			Bank 2			0.0		0.0		0.0	Consumer
Sector 3	100.8			Bank 3			0.0		0.0		0.0	1	2		3
Sector 4	101.1			Bank 4			0.0		0.0		0.0	102.4	111.2		75.2
Sector 5	103.1			Bank 5			0.0		0.0		0.0

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

View Table

Table 3.

9.5 Percent Revaluation

Period		1		2		3		4		5		6		7		8
Nominal GDP 1/		100.0		116.2		131.7		165.0		163.8		203.7		194.1		237.0
Real GDP 1/		100.0		107.0		111.0		119.0		122.9		131.3		136.5		146.1
Price Level		100.0		107.0		118.7		139.3		133.3		155.1		142.3		162.9
Nominal rate		-3.1		1.4		0.9		6.1		1.4		10.9		2.2		15.7
Real interest rate				-5.2		-9.0		-9.6		6.0		-4.7		11.4		1.1
Budget deficit 2/		7.7		7.7		9.9		8.8		11.8		10.7		13.8		12.7
Trade balance 2/		10.3		7.2		4.1		1.3		-1.2		-6.1		-5.5		-10.3
Net capital stock at 3/ end of period 8				Percent of bank assets in default at end of period:								US$ FDI in Period 8 (Base period = 100)
				Period			4		6		8	95.4
Sector 1	101.3			Bank 1			0.0		0.0		0.0	Utility levels
Sector 2	103.3			Bank 2			0.0		0.0		0.0	Consumer
Sector 3	100.8			Bank 3			0.0		0.0		0.0	1	2		3
Sector 4	101.1			Bank 4			0.0		0.0		0.0	102.4	111.2		75.2
Sector 5	103.1			Bank 5			0.0		0.0		0.0

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

We notice a number of changes, as compared to the benchmark case of Table 2. The revaluation has caused a considerable deterioration in the trade balance, as might be expected. In addition, FDI is now lower in the final period than in period 1. Although nominal interest rates are rising, revaluation of the domestic currency is high enough to reduce FDI flows. FDI is also significantly lower than the final period outcome in Table 2, as asset prices continue to rise for the foreign investor, both through a nominal revaluation and a lower nominal interest rate compared to the base case. Rising demand for imports has caused consumption of domestic output to decline. The real value of savings rises, as does domestic investment which more than substitutes for the FDI decline. As a result, there has been a moderate increase in end of period sectoral capital stocks. Note that this is consistent with the intuition we provided above: real interest rates, which have risen in this case relative to the base case, are the determining factor for domestic investors. There have been welfare gains for both domestic consumers.

How far can the revaluation be pushed while still remaining economically viable? Suppose as a final experiment we increase the rate of revaluation to 10 percent annually. Table 4 gives the results.

Table 4.

10.0 Percent Revaluation

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

Table 4.

10.0 Percent Revaluation

Period		1		2		3		4		5		6		7		8
Nominal GDP 1/		100.0		115.8		130.8		163.9		162.1		201.4		191.0		232.1
Real GDP 1/		99.8		107.1		110.5		118.2		122.5		131.3		136.3		145.6
Price Level		100.2		108.1		118.4		138.7		132.3		153.4		140.1		159.4
Nominal rate		-3.3		1.4		0.8		5.8		1.2		10.7		2.0		15.9
Real interest rate				-6.0		-8.0		-9.7		6.1		-4.5		11.7		1.9
Budget deficit 2/		7.7		7.3		9.9		8.9		11.8		10.7		13.9		12.8
Trade balance 2/		10.3		7.2		4.1		-1.5		-1.1		-6.5		-5.9		-11.1
Net capital stock at 3/ end of period 8				Percent of bank assets in default at end of period:								US$ FDI in Period 8 (Base period = 100)
				Period			4		6		8	94.6
Sector 1	101.7			Bank 1			0.0		0.0		16.0	Utility levels
Sector 2	103.3			Bank 2			0.0		0.0		63.0	Consumer
Sector 3	100.8			Bank 3			0.0		0.0		15.1	1	2		3
Sector 4	101.2			Bank 4			0.0		0.0		28.7	102.4	111.7		73.2
Sector 5	103.4			Bank 5			0.0		0.0		15.6

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

View Table

Table 4.

10.0 Percent Revaluation

Period		1		2		3		4		5		6		7		8
Nominal GDP 1/		100.0		115.8		130.8		163.9		162.1		201.4		191.0		232.1
Real GDP 1/		99.8		107.1		110.5		118.2		122.5		131.3		136.3		145.6
Price Level		100.2		108.1		118.4		138.7		132.3		153.4		140.1		159.4
Nominal rate		-3.3		1.4		0.8		5.8		1.2		10.7		2.0		15.9
Real interest rate				-6.0		-8.0		-9.7		6.1		-4.5		11.7		1.9
Budget deficit 2/		7.7		7.3		9.9		8.9		11.8		10.7		13.9		12.8
Trade balance 2/		10.3		7.2		4.1		-1.5		-1.1		-6.5		-5.9		-11.1
Net capital stock at 3/ end of period 8				Percent of bank assets in default at end of period:								US$ FDI in Period 8 (Base period = 100)
				Period			4		6		8	94.6
Sector 1	101.7			Bank 1			0.0		0.0		16.0	Utility levels
Sector 2	103.3			Bank 2			0.0		0.0		63.0	Consumer
Sector 3	100.8			Bank 3			0.0		0.0		15.1	1	2		3
Sector 4	101.2			Bank 4			0.0		0.0		28.7	102.4	111.7		73.2
Sector 5	103.4			Bank 5			0.0		0.0		15.6

^1/All indices are normalized for this base simulation.

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

Real GDP continues to decline, as compared to Table 4, and the trade balance deteriorates further. FDI also declines compared to the base case, as well as to Table 4, due to the increased price of domestic assets. Consumption of domestic output continues to decline, while savings increase, leading to yet higher domestic investment that replaces FDI. As a result, the rate of return on domestic investment declines, and real interest rates continue to rise. By the final period sector 2 defaults on its debts. Sector 2 (manufacturing) is large relative to the economy so its default spreads across all banks, each of which is now above the 8 percent failure rate. Accordingly, the entire banking system has now collapsed. We also notice that each consumer has realized an intertemporal utility increase, given the decrease in the price of imports, despite the fact that real GDP has slowed. Essentially cheap imports have substituted for domestic output, and there has been an increase in consumption of leisure.

Of course this new outcome is not sustainable, so we need to ask if it is possible to achieve similar welfare gains by policies other than exchange rate revaluations.²⁸ There are many such potential policies, and it is not possible to analytically solve for particular tax rates or spending levels that will give a particular welfare outcome. However, as a simple example of what could be accomplished by increasing consumption, we will impose a decrease in the personal income tax. The assumed effective tax rate on personal income was 5 percent in the previous examples. We will now suppose that it is reduced to 3 percent. The outcome is given in Table 5.

Table 5.

Personal Income Tax Reduction

^1/All indices are normalized for this base simulation

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

Table 5.

Personal Income Tax Reduction

Period		1		2		3		4		5		6		7		8
Nominal GDP 1/		101.7		129.4		146.1		187.9		196.7		253.0		255.8		328.7
Real GDP 1/		100.3		107.6		110.9		118.8		123.0		131.7		136.1		145.7
Price Level		101.3		120.3		131.8		158.1		159.9		192.1		187.9		225.6
Nominal rate		1.2		2.9		2.0		7.7		2.6		11.0		3.0		13.1
Real interest rate				-13.4		-6.9		-10.2		1.4		-7.6		5.3		-5.8
Budget deficit 2/		8.8		8.0		10.4		9.5		11.9		10.9		13.2		12.0
Trade balance 2/		9.9		7.3		4.1		1.6		0.6		-1.6		-1.8		-3.8
Net capital stock at 3/ end of period 8				Percent of bank assets in default at end of period:								US$ FDI in Period 8 (Base period = 100)
				Period			4		6		8	105.6
Sector 1	98.7			Bank 1			0.0		0.0		0.0	Utility levels
Sector 2	98.4			Bank 2			0.0		0.0		0.0	Consumer
Sector 3	98.6			Bank 3			0.0		0.0		0.0	1	2		3
Sector 4	98.4			Bank 4			0.0		0.0		0.0	102.9	102.1		95.2
Sector 5	98.9			Bank 5			0.0		0.0		0.0

^1/All indices are normalized for this base simulation

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

View Table

Table 5.

Personal Income Tax Reduction

Period		1		2		3		4		5		6		7		8
Nominal GDP 1/		101.7		129.4		146.1		187.9		196.7		253.0		255.8		328.7
Real GDP 1/		100.3		107.6		110.9		118.8		123.0		131.7		136.1		145.7
Price Level		101.3		120.3		131.8		158.1		159.9		192.1		187.9		225.6
Nominal rate		1.2		2.9		2.0		7.7		2.6		11.0		3.0		13.1
Real interest rate				-13.4		-6.9		-10.2		1.4		-7.6		5.3		-5.8
Budget deficit 2/		8.8		8.0		10.4		9.5		11.9		10.9		13.2		12.0
Trade balance 2/		9.9		7.3		4.1		1.6		0.6		-1.6		-1.8		-3.8
Net capital stock at 3/ end of period 8				Percent of bank assets in default at end of period:								US$ FDI in Period 8 (Base period = 100)
				Period			4		6		8	105.6
Sector 1	98.7			Bank 1			0.0		0.0		0.0	Utility levels
Sector 2	98.4			Bank 2			0.0		0.0		0.0	Consumer
Sector 3	98.6			Bank 3			0.0		0.0		0.0	1	2		3
Sector 4	98.4			Bank 4			0.0		0.0		0.0	102.9	102.1		95.2
Sector 5	98.9			Bank 5			0.0		0.0		0.0

^1/All indices are normalized for this base simulation

^2/As a percentage of GDP.

^3/Capital stocks are normalized to levels of the benchmark case of Table 2.

As expected, the budget deficit rises compared to the case of Table 2. Following a tax cut GDP rises initially relative to the base case, as it is expected. However, due to rising budget deficit, the final period GDP is lower than that in the base case (see Figure 1). This is also intuitive: As the government issues debt to monetize the deficit interest rates rise, slowing down capital formation.

Read GDP Series Following a Tax Cut

(% change relative the base case)

Citation: IMF Working Papers 2007, 278; 10.5089/9781451868418.001.A001

• Download Figure
• Download figure as PowerPoint slide

View Full Size

Read GDP Series Following a Tax Cut

(% change relative the base case)

Citation: IMF Working Papers 2007, 278; 10.5089/9781451868418.001.A001

• Download Figure
• Download figure as PowerPoint slide

Read GDP Series Following a Tax Cut

(% change relative the base case)

Citation: IMF Working Papers 2007, 278; 10.5089/9781451868418.001.A001

• Download Figure
• Download figure as PowerPoint slide

The increase in nominal interest rates has caused FDI to increase, as compared to the benchmark case, although the decline in domestic investment is greater than the increase in FDI due to rising real interest rates. Lower levels of borrowing by domestic investors have caused no loan defaults, and the banking system is solvent. Consumer 1 (urban), is at approximately the same utility level as in Table 3, although Consumer 2 is at a lower level. The trade balance improves significantly, and is only slightly lower than in the base case of Table 2. The small deficit is because of lower exports: a higher price level combined with fixed exchange rates means higher export prices for the foreigners. Thus the tax cut has achieved the objective of increasing consumption, as reflected in higher utility for both consumers. Consumer 1 is better off with the tax cut than with the revaluation, while the opposite is true for Consumer 2.

We thus have an indication that it is possible to have general intertemporal welfare increases, without financial failures, by the use of fiscal policy in the form of tax changes. Exchange rate changes, in the form of revaluations, may lead to larger welfare gains by certain consumer categories, but can become financially unsustainable.