Mexico: Selected Issues

International Monetary Fund

doi:10.5089/9781451956061.002.A003

Jump to Content Jump to Main Navigation

Mexico: Selected Issues

Author: International Monetary Fund

Publication Date:: 13 Dec 2007

eISBN:: 9781451956061

Language:: English

Keywords:: ISCR; CR; México; price differential; dummy post NAFTA; country pair; NAFTA member; price threshold; Exchange rates; North American Free Trade Agreement; Arbitrage; Trade liberalization

Download PDF (390.9 KB)

This Selected Issues paper analyzes external shocks and business cycle fluctuations in Mexico. The paper examines the relative importance of U.S. demand shocks—and other foreign disturbances—in explaining Mexican output fluctuations. It identifies the dynamic response of Mexico’s output to those shocks. The paper investigates which U.S. variables are most relevant to explaining business cycles in Mexico. It analyses potential spillovers and channels of transmission underlying the linkages between the United States and Mexican economies, and focuses on one aspect of the development of the Mexican private mortgage market, the market for mortgage-backed securities.

Abstract

III. Mexico’s Integration into NAFTA Markets: A View from Sectoral Real Exchange Rates and Transaction Costs¹

A. Introduction

1. Is Mexico reaping the full benefits of its integration within NAFTA markets? The analysis of relative price differentials across countries and sectors offers a way to evaluate the degree of market integration. The study of NAFTA members is of particular interest, allowing an assessment of whether regional trade liberalization has resulted in faster price convergence and smaller price differentials across countries, and, greater market integration.

2. The Law of One Price, or “LOOP,” states that identical goods should sell for the same price across countries when prices are expressed in a common currency. Evidence has shown, however, that prices of goods fail to fully equalize between countries, indicating that markets are not perfectly integrated—due to some kind of transaction costs that limit price arbitrage. Obstacles to integration include transport costs and (explicit or implicit) trade barriers. Our analysis sheds light on the following questions:

Is the degree of Mexico–U.S. integration similar to that of Canada—U.S. integration? We find that transaction costs are larger for the Mexico–U.S. country pair than for the Canada–U.S. pair.
Have markets become more integrated, with reduced transaction costs, after the introduction of NAFTA? Our results show that NAFTA significantly reduced price differentials between the U.S. and Mexico, as trade liberalization evidently reduced transaction costs, though this was not uniform across sectors.
What are some of the determinants of transaction costs? In addition to trade liberalization, sharing a common border and having lower exchange rate volatility are found to reduce transaction costs. However, it appears that industry or good-specific characteristics must account for a large part of transaction costs.

3. The empirical methodology analyzes dynamics in relative price adjustment and innovates by taking the perspective of an emerging market—Mexico.² Due to transaction costs, it may not be profitable to arbitrage away relative price differences across countries. When the marginal costs of arbitrage exceed the marginal benefit, there is a zone of no-trade and consequently prices in two locations fail to equalize. Outside this zone, arbitrage is profitable and the sectoral real exchange rate (SRER) can become mean-reverting. The existence of such “threshold band” requires the use of a nonlinear model—more specifically, a threshold autoregressive model (Tong, 1990; and Hansen, 1996, 1997). The estimated price thresholds are a measure of transaction costs, and the absence of relative price convergence is interpreted as a sign of weak market integration.³ We also attempt to identify determinants of transaction costs across country pairs, sectors, and over time.

4. The remainder of the paper is organized as follows. Section B reviews theoretical considerations on nonlinear dynamics in sectoral real exchange rates and presents the corresponding econometric methodology. The results are discussed in Section C. The determinants of transaction costs are studied in Section D. The last section summarizes and concludes.

B. Nonlinear Dynamics in Sectoral Real Exchange Rates

Theoretical underpinnings

5. According to the law of one price (LOOP), there should be no price differentials across countries for similar goods when prices are expressed in a common currency. At the aggregate level, the LOOP translates into purchasing power parity (PPP). The LOOP is based on the assumption that there are no transaction costs or impediments to trade that would prevent perfect arbitrage and allow sectoral price differentials—it relies on a frictionless goods arbitrage.

6. Deviations $x_{j t}^{i}$ from the LOOP for a sector j in country i at time t are defined as:

\begin{matrix} x_{j t}^{i} = s_{t}^{i} + p_{j t}^{i} - p_{j t} & (1) \end{matrix}

where $s_{t}^{i}$ is the logarithm of the nominal exchange rate between country i’s currency and the reference country, $p_{j t}^{i}$ is the logarithm of the price of good j in country i at time t and p_jt is the logarithm of the price of good j in the reference country at time t.

7. Ample empirical evidence (Isard, 1977; Richardson, 1978 and Giovannini, 1988) suggests that relative prices do not converge, or only in a very long-term horizon, and that price differentials are persistent. These studies also found that relative price differentials are significant and highly correlated with exchange rate movements.

8. The existence of transaction costs, in the form of transport costs or (explicit or implicit) trade barriers, is an explanation for lack of price convergence.⁴ Several theoretical studies account for the importance of transaction costs in modeling deviations from the LOOP (for example, Dumas, 1992; 1995 Sercu and others, ; O’Connell, 1998). These studies suggest that frictions to trade imply the presence of significant nonlinearities in sectoral real exchange rate dynamics. That is, transaction costs generate a band in which the marginal costs of arbitrage exceed the marginal benefit. Within this band, there is a zone of no-trade and consequently prices in two locations fail to equalize. Outside this band, arbitrage is profitable and the sectoral real exchange rate can become mean-reverting.

9. Empirical studies have investigated the presence of nonlinearities in deviations from the LOOP using a TAR model and focusing on developed markets. Obstfeld and Taylor (1997) find evidence of significant transaction costs in a sample of 32 international locations, using disaggregated data on clothing, food and fuel. Sarno and others (2004) provide support for nonlinear mean reversion, with considerable cross-country and sectoral variations. They use annual price data interpolated into quarterly for nine sectors and quarterly data on five exchange rates vis-à-vis the US dollar. Juvenal and Taylor (2007) study the presence of nonlinearities in deviations from the LOOP for 19 sectors in 10 European countries and find significant evidence of threshold adjustment with transaction costs varying considerable across sectors and countries.

Estimation Methodology and SETAR Model

10. To analyze patterns in relative price convergence, we model deviations from the LOOP using a self-exciting threshold autoregressive (SETAR) model for each sectoral exchange rate. More precisely, we investigate the presence of nonlinearities in deviations from the LOOP using a threshold-type model with two regimes. To analyze the characteristics of the threshold dynamics, we proceed in 3 steps:

First, we explore the validity of the nonlinear threshold model with respect to a null hypothesis of a random walk (unit root process). In other words, we first test for the existence of some degree of price convergence (whether partial or complete convergence), as opposed to no price convergence at all.
Second, for all cases in which some degree of price convergence is found, we assess whether price convergence is indeed partial—up to a threshold that we estimate—or complete. That is, we test whether a nonlinear model fits the data better than a linear one.
Finally, when we find evidence of nonlinear price convergence both in the pre- NAFTA and post-NAFTA periods, we test if the size of the threshold band is equal in both periods.⁵

11. The model tests for the existence of a threshold band, within which deviations from the LOOP are smaller than transaction costs and consequently are not arbitraged. In this case, the SRER would follow a unit root process—the LOOP would not hold. In the outer regime, deviations from the LOOP would be higher than transaction costs, making arbitrage profitable. In this case the process would become mean reverting. We test whether the autoregressive process followed by the real exchange rate switches across regimes.⁶

Under the assumption that adjustments from deviations from the LOOP are symmetric outside the threshold band, a simple three-regime TAR model may be written as:

\begin{matrix} Δ q_{j t}^{i} = [(ρ - 1) (q_{j t - 1}^{i} - k_{j}^{i})] 1 (q_{j t - d}^{i} > k_{j}^{i}) + [(ρ - 1) (q_{j t - 1}^{i} + k_{j}^{i})] 1 (q_{j t - d}^{i} < - k_{j}^{i}) + ɛ_{j}^{i} & (2) \end{matrix}

where $q_{j t}^{i}$ to be the demeaned component of the relative price difference $x_{j t}^{i}$ , given by $x_{j t}^{i} = c_{j}^{i} + q_{j t}^{i}$ ( $q_{j t}^{i}$ is estimated as an OLS residual), $k_{j}^{i}$ is the threshold parameter for sector j in country i, $q_{j t - d}^{i}$ is the threshold variable for sector j and country i. The parameter d accounts for the delay with which economic agents react to real exchange rate deviations.

The key variables, i.e., the threshold and autoregressive parameters, are estimated simultaneously using least squares via a grid search over $k_{j}^{i}$ .

C. Estimation Results

12. We use disaggregated monthly data on consumer price indices (CPI) for 18 sectors from January 1980 to December 2006, for Mexico, the United States and Canada.⁷

Testing for nonlinear price convergence

Table 1 reports the results of the estimation of the SETAR model.

13. The first step consists of testing the null hypothesis of a unit root (Figure 1). Our interpretation of such a case is that transaction costs are so large that arbitrage is not profitable and the threshold band is wide enough to contain the whole time series of the SRER. A number of observations follow from this first set of results:

Results provide a first indication that NAFTA led to greater integration between the United States and Mexico, with price equalization between locations taken as a sign of higher degree of market integration. Half of the SRERs in the pre-NAFTA period followed a unit root process and only four of them in the post-NAFTA period.
By contrast, results suggest that the Canadian and American markets have been more closely integrated, with a further improvement with NAFTA.

14. The second step—conducted only for cases in which SRER does not follow a unit root process—tests whether the nonlinear threshold model is superior when tested against a linear process in which no threshold exists. Table 1 reports the p-values indicating the significance level at which the linearity hypothesis can be rejected. In all cases the SETAR model clearly outperforms the linear one, confirming the existence of thresholds and therefore providing an estimate of transaction costs.

Table 1.

SETAR Estimation Results

Notes: This table shows the results from the estimation of the SETAR (1, 2, d) model in equation (5). k is the value of the threshold and p is the outer root of the TAR process. The estimation of k, p and d is done simultaneously via a grid search over k and d as described in section III. p-value Ha, p-value Hb and p-value He represent, respectively, the marginal significance levels of the null hypothesis of unit root in the outer regime, null hypothesis of linearity and null hypothesis of equality of thresholds pre- and post- NAFTA.

Estimated Transaction Costs⁸

15. Table 1 also reports the estimated thresholds for each SRER—that we interpret as a measure of overall transaction costs and reflecting market integration:

Across sectors, the results generally confirm that highly homogenous sectors—for example, fish products and fruits—show low threshold bands. This is a standard result in the literature for other country pairs (see Juvenal and Taylor, 2007).
For the United States–Mexico SRERs, evidence of a strong NAFTA effect is found. The range of transaction costs across sectors is smaller, from 7–32 percent in the pre-NAFTA period to 2–20 percent in the post-NAFTA period, also with a number of cases in which transaction costs go from “infinite” (unit root process) to measurable. Transaction costs bands are reduced in a number of sectors, suggesting greater market integration.
Overall, average transaction costs among NAFTA members are 34 percent higher between the U.S. and Mexico than between the U.S. and Canada. This result confirms previous evidence that the United States and Canada are the most integrated among NAFTA members.⁹

16. In comparison to the work of Juvenal and Taylor (2007), threshold bands among NAFTA members are on average slightly lower than between the United States and European countries. Interestingly, when considering the United Kingdom as a reference country, their estimated country average transaction costs range from 7 percent to 17 percent. The latter benchmark is probably most relevant to our study for comparison purposes given the process of liberalization among European countries and the distance factor.

Half-Lives

17. A usual measure of the speed of mean reversion is the half-life, the time it takes for the effect of 50 percent of a shock to die out. Table 2 reports the estimated half-lives (in terms of months) of price deviations from the LOOP, for the Mexico–U.S. SRER.¹⁰

Table 2.

Estimation of Half-Lives for Mexico-U.S. Sectoral Real Exchange Rates

(in months)

Notes: The columns show the half-life of the TAR model as a whole for a given shock estimated conditional on average initial history. The half-lives for a 10%, 20%, 30%, 40% and 50% shocks are computed by stochastic simulation using the generalized impulse response functions procedure developed by Koop and others (1996).

Table 2.

Estimation of Half-Lives for Mexico-U.S. Sectoral Real Exchange Rates

(in months)


	Shock (%)					Shock (%)
Sector	10	20	30	40	50	10	20	30	40	50
Bread
Meat	36	26	20	17	15	29	25	23	22	21
Fish						19	18	18	18	18
Dairy	20	15	11	9	8	7	5	5	5	5
Fruits						6	5	5	5	5
Vegetables	4	4	4	4	4	5	5	5	5	5
Non alcoholic beverages						7	7	6	6	6
Alcoholic beverages	13	12	12	11	11
Tobacco	18	12	8	7	6	8	7	7	7	7
Women’s clothing	10	10	10	9	9	5	5	5	5	5
Men’s clothing						10	8	8	7	7
Footwear	18	17	16	16	16	6	6	6	6	6
Fuel
Furniture						14	10	8	8	8
Medications						10	8	8	8	7
Vehicles	6	5	5	4	3	6	4	4	4	4
Gasoline
Photo	55	49	44	40	37	24	14	10	9	8
Average	20	17	14	13	12	11	9	8	8	8

Table 2.

Estimation of Half-Lives for Mexico-U.S. Sectoral Real Exchange Rates

(in months)


	Shock (%)					Shock (%)
Sector	10	20	30	40	50	10	20	30	40	50
Bread
Meat	36	26	20	17	15	29	25	23	22	21
Fish						19	18	18	18	18
Dairy	20	15	11	9	8	7	5	5	5	5
Fruits						6	5	5	5	5
Vegetables	4	4	4	4	4	5	5	5	5	5
Non alcoholic beverages						7	7	6	6	6
Alcoholic beverages	13	12	12	11	11
Tobacco	18	12	8	7	6	8	7	7	7	7
Women’s clothing	10	10	10	9	9	5	5	5	5	5
Men’s clothing						10	8	8	7	7
Footwear	18	17	16	16	16	6	6	6	6	6
Fuel
Furniture						14	10	8	8	8
Medications						10	8	8	8	7
Vehicles	6	5	5	4	3	6	4	4	4	4
Gasoline
Photo	55	49	44	40	37	24	14	10	9	8
Average	20	17	14	13	12	11	9	8	8	8

18. On average, relative price adjustment is significantly faster in the post-NAFTA period. For example, for a 10 percent shock, the average half-life pre-NAFTA was 20 months, while the average was reduced to 11 months in the post-NAFTA period. Our results also bring additional observations:

In the post-NAFTA period, there is less variation across different shock sizesthan in the pre-NAFTA period—suggesting that relative prices adjust more quickly, independently of the size of the price shock. In the post-NAFTA period, almost 60 percent of the SRERs adjust (by half) to a 10 percent shock within 6 to 9 months. In contrast, most (70 percent) SRERs take more than a year to adjust in the pre-NAFTA period, and 55 percent more than 18 months.
Half-lives vary substantially across sectors. Relative prices adjust relatively fast for homogenous goods, such as food products. The relative price of the more high-end products takes longer to adjust, for example furniture, and photographic equipment.

D. Determinants of Thresholds in Real Exchange Rates

19. In the spirit of the gravity models used to explain trade patterns, we investigate in this section the main determinants of the estimated transaction costs in sectoral real exchange rates. In their simplest form, gravity equations relate trade to distance between trading partners, as a proxy for transaction costs (see for example the initial work of Linneman (1969)). The models were enriched to account for other determinants of trade. Imbs and others (2007) studied the determinants of relative price dynamics for European countries, using a gravity-type model. They find that distance and exchange rate volatility are strongly correlated with thresholds and the half-life of exchange rate deviations.

The model is defined as:

\begin{matrix} k_{j}^{i} = λ_{j}^{i} + Σ_{k = 1}^{k} Φ_{j}^{i} (k) z_{j}^{i} (k) + ɛ_{j}^{i} & (9) \end{matrix}

where $z_{j}^{i}$ is a vector of explanatory variables. We assess in equation (9) whether transaction costs, measured by the threshold $k_{j}^{i}$ , are explained by selected explanatory variables.

Box 1.

Evidence on Price Convergence for Specific Goods

Our research also provides preliminary evidence on price convergence at the level of specific goods. Looking at such disaggregated price data has advantages, including greater comparability of goods across countries, and permitting assessment of the role of good-specific characteristics in price convergence.

Here we look especially at a number of goods for which some analysts have suggested that price competition may be relatively weak or even absent—corn, flour, rice, sugar, electricity, telephone services, and cement. We also report results for a comparator sample consisting of nine other specific goods. The main findings are as follows:

Again we find a significant reduction in relative price thresholds, from an average of 19 percent during the pre-NAFTA period, to 10 percent after the introduction of NAFTA.
Even after the introduction of NAFTA, relative price thresholds remained relatively high in the sample of interest, in most cases substantially above those in the comparator sample.

Table 3.

SETAR Estimation Results for Disaggregated Mexico - U.S. SRERs

Notes: See Table 1.“—” signifies no price convergence at all (unit root).

Table 3.

SETAR Estimation Results for Disaggregated Mexico - U.S. SRERs

	Pre-NAFTA		Post-NAFTA
	Threshold	Mean-reversion	Threshold	Mean-reversion
	k	ρ	k	ρ
Sample of interest
corn	0.25	0.91	0.21	0.95
flour	—	—	0.27	0.84
rice	—	—	0.20	0.82
sugar	0.29	0.83	0.06	0.89
electricity	0.06	0.93	0.15	0.87
landline telephone services	n.a.	n.a.	—	—
cement	0.30	0.98	0.17	0.95
concrete	0.33	0.97	0.12	0.92
Comparator sample
chicken parts	0.28	0.86	0.04	0.89
whole chicken	0.24	0.87	0.02	0.88
ground beef	—	—	—	—
ham	—	—	—	—
canned seafood	0.10	0.93	0.15	0.78
eggs	0.15	0.80	0.02	0.83
apples	—	—	0.05	0.86
oranges	0.13	0.90	0.07	0.85
tomato	0.08	0.75	0.09	0.70
potato	0.05	0.92	0.10	0.91
Full sample average	0.19		0.10

Notes: See Table 1.“—” signifies no price convergence at all (unit root).

Table 3.

SETAR Estimation Results for Disaggregated Mexico - U.S. SRERs

	Pre-NAFTA		Post-NAFTA
	Threshold	Mean-reversion	Threshold	Mean-reversion
	k	ρ	k	ρ
Sample of interest
corn	0.25	0.91	0.21	0.95
flour	—	—	0.27	0.84
rice	—	—	0.20	0.82
sugar	0.29	0.83	0.06	0.89
electricity	0.06	0.93	0.15	0.87
landline telephone services	n.a.	n.a.	—	—
cement	0.30	0.98	0.17	0.95
concrete	0.33	0.97	0.12	0.92
Comparator sample
chicken parts	0.28	0.86	0.04	0.89
whole chicken	0.24	0.87	0.02	0.88
ground beef	—	—	—	—
ham	—	—	—	—
canned seafood	0.10	0.93	0.15	0.78
eggs	0.15	0.80	0.02	0.83
apples	—	—	0.05	0.86
oranges	0.13	0.90	0.07	0.85
tomato	0.08	0.75	0.09	0.70
potato	0.05	0.92	0.10	0.91
Full sample average	0.19		0.10

Notes: See Table 1.“—” signifies no price convergence at all (unit root).

Box 2.

Real Exchange Rate Thresholds at the Aggregate CPI Level

Estimation of convergence thresholds can also be conducted at the level of the national CPI index, although the interpretation of the estimated thresholds is much less clear. Still, the results based on aggregate indices may be of some interest, and it turns out that they are broadly similar to the pattern of the sectoral findings. The results are reported in Table 4.

For all three country pairs, we find evidence of non-linear (partial) convergence of aggregate price levels for both the pre- and post-NAFTA periods. The size of the thresholds is significantly smaller in the post-NAFTA period, and is smaller for the U.S.—Canada country pair than for the U.S.—Mexico pair. After NAFTA, the estimated thresholds for the U.S.–Canada and the U.S.–Mexico pairs are reduced, respectively, from 13 percent to 10 percent, and from 18 percent to 14 percent, respectively.

This finding of thresholds at the aggregate level suggests that limitations to price convergence at the sectoral level can also be an issue of macroeconomic significance. This point may be worthy of analysis in empirical studies of (national-level) real exchange rates.

Table 4.

SETAR Estimation Results for Aggregate Price Indices

Notes: This table shows the results from the estimation of the SETAR (1, 2, d) model in equation (5). k is the value of the threshold and p is the outer root of the TAR process. The estimation of k, p and d is done simultaneously via a grid search over k and d as described in section III. p-value Ha, p-value Hb and p-value Hc represent, respectively, the marginal significance levels of the null hypothesis of unit root in the outer regime, null hypothesis of linearity and null hypothesis of equality of thresholds pre- and post- NAFTA.

20. The explanatory variables are intended to capture determinants of transaction costs. Given the small number of country pairs and their relative closeness, distance appears to be a poor proxy for transaction costs.¹¹ Instead, we include a dummy variable that takes value 1 when countries share a common border. The second variable is the volatility of the nominal exchange rate. Measured as the standard deviation of monthly exchange rate observations, the volatility variable is a proxy for uncertainty of the macroeconomic environment. Third, we include a measure of “tradability,” defined as the sum of imports and exports to the total output in a sector for a given country. Fourth, we use the number of establishments in each sector as a proxy for competition, or concentration, in each sector. Lastly, a dummy for the post-NAFTA period is used.

21. We examine the determinants of thresholds for the entire sample, including all three country pairs (we include here the Mexico—Canada pair), including the pre- and post-NAFTA periods. The 94 thresholds computed in the previous section constitute the observations.

Table 5.

Threshold Regressions

Notes: ^*** indicates a 1 percent degree of confidence, ^** indicates a 5 percent degree of confidence, ^* indicates a 10 percent degree of confidence.

Table 5.

Threshold Regressions

Determinants of Thresholds
	(1)	(2)
Distance	-0.041 ^*	-0.040 ^*
Dummy post NAFTA	-0.109 ^**	-0.116 ^**
Exchange Rate Volatility	4.020 ^***	3.991 ^***
Firms	-0.001	-
Tradability	-0.042	-
R²	0.33	0.33
N	94	94

Notes: ^*** indicates a 1 percent degree of confidence, ^** indicates a 5 percent degree of confidence, ^* indicates a 10 percent degree of confidence.

Table 5.

Threshold Regressions

Determinants of Thresholds
	(1)	(2)
Distance	-0.041 ^*	-0.040 ^*
Dummy post NAFTA	-0.109 ^**	-0.116 ^**
Exchange Rate Volatility	4.020 ^***	3.991 ^***
Firms	-0.001	-
Tradability	-0.042	-
R²	0.33	0.33
N	94	94

Notes: ^*** indicates a 1 percent degree of confidence, ^** indicates a 5 percent degree of confidence, ^* indicates a 10 percent degree of confidence.

22. Three variables appear significant: the post-NAFTA dummy, the border effect, and exchange rate volatility. For the latter two variables, the results are in line with findings in the rest of the literature. For example, Imbs and others (2007) find that distance, exchange rate volatility, tradability, and industry competition explain the level of thresholds. The dummy post-NAFTA is also strongly significant and negative, confirming our previous results that the introduction of NAFTA reduced transaction costs. However, our attempt to use sectoral variables to explain transaction costs is not successful: neither the number of firms in a sector nor the degree of “tradability” in a sector are found to be statistically significant (column (1) of Table 4). The poor quality of the data and the approximation in proxying intra-industry trade and sectoral competition are a probable explanation for the lack of significance. In column (2), the two variables are excluded, with little change in the results.

23. Using other types of models, the determinants of relative price differences between the United States and Canada have been extensively studied in the literature. The results are broadly consistent with our findings. Engle and Rogers (1996) the Canadian and US markets are not perfectly integrated and that distance and border are major determinants of price differentials between cities. Engle, Rogers and Yi Wang (2005) investigate the LOOP between US and Canadian cities using actual prices (instead of price indices). They find that absolute price differences between US and Canadian prices are higher than 7 percent. In addition, their results show that distance and border play a significant role in explaining price differentials between cities.

E. Summary of Results and Conclusion

24. Using a SETAR model, we find significant differences in transaction costs in different sectors and countries. Looking at the Mexico—U.S. and Canada—U.S. country pairs, the estimated price thresholds range from 2 percent to 32 percent.

Across sectors, the results generally confirm that highly homogenous sectors—fish and fruits—show low threshold bands.
Across country pairs, interpreting the size of the price threshold as a measure of market integration, we find significant differences between the three NAFTA members, with Mexico being relatively less integrated. Overall, average transaction costs among NAFTA members are 34 percent higher between the U.S. and Mexico than between the U.S. and Canada.

We also document the impact of NAFTA on the integration of the three countries and find that NAFTA substantially reduced transaction costs between Mexico and the U.S. while its impact was less marked between Canada and the U.S.

To analyze the adjustment of relative prices to shocks, we also compute the half-lives of the Mexico-U.S. sectoral real exchange rates, a measure of the time it takes for the effect of 50 percent of a price shock to die out. On average, the average half-life was substantially reduced after the introduction of NAFTA, going down from 20 months in the pre-NAFTA period to 11 months post-NAFTA.

25. The border effect and exchange rate volatility are found to be significant determinants of transaction costs. The dummy post-NAFTA is also strongly significant and negative, confirming that the introduction of NAFTA reduced transaction costs.

26. The analysis therefore supports the arguments that (i) emerging markets—in this case, Mexico—still face higher transaction costs than their developed counterparts; and (ii) trade liberalization may help in lowering relative price differentials between countries. We suspect that lack of competition may be a major determinant of high price thresholds. With limited data, we provide only tentative evidence on this issue.

27. From the point of view of Mexico, the findings imply that domestic goods’ prices today respond more fully, and more quickly, than in the past to either (i) a change in the domestic price in the U.S., or (ii) a change in the nominal exchange rate. While greater transmission of sectoral relative price shocks may have consequences for the conduct of monetary policy, it should be distinguished from the question of “exchange rate pass-through” to the overall consumer price index. Such overall pass-through is determined also by other factors, including monetary policy and the business cycle. However, in the last several years, the fluctuation of Mexico’s peso against the U.S. dollar has been fairly modest—with maximum and minimum monthly averages differing by only about 10 percent. Exchange rate movements within such a range are smaller than the transaction cost bands that we find for many sectors, implying that pass-through of such exchange rate changes to domestic prices will often be limited or even non-existent. For Mexico, therefore, it may now be that movements in U.S. prices—rather than nominal exchange rate fluctuations—are the more relevant source of variation in domestic prices of certain traded goods.

28. The main conclusion of the paper is that Mexico has made progress, but still has considerable room for improvement, in reducing barriers to goods market integration and achieving full benefits of globalization. It would be important to further analyze the reasons why transactions costs between Mexico-U.S. continue to exceed those for Canada -U.S. for many types of goods, and to determine whether these costs can be reduced through policy actions—for example, by developing logistics, transportation, and internal distribution mechanisms, or by enhancing the state of competition among domestic firms and reducing remaining barriers to external trade.

References

Chen, N., 2004, “The Behaviour of Relative Prices in the European Union: A Sectoral Analysis,” European Economic Review, Vol. 48, pp. 1257– 1286.
- Search Google Scholar
- Export Citation
Dumas, B., 1992, “Dynamic Equilibrium and the Real Exchange Rate in a Spatially Separated World,” Review of Financial Studies, Vol. 2, No 5, pp. 153– 180.
- Search Google Scholar
- Export Citation
Giovannini, A., 1998, “Exchange Rates and Traded Goods Prices”. Journal of International Economics, Vol 24, No 1–2, pp. 45– 68.
- Search Google Scholar
- Export Citation
González, M., and F. Rivadeneyra, 2004, “La Ley de un Solo Precio en México: Un Análisis Empírico”, Gaceta de Economía, ITAM, year 10—Vol. 19, pp. 91– 115.
- Search Google Scholar
- Export Citation
González, M., and F. Reivadeneyra, 1998, “Exchange Rates and Traded Goods Prices”. Journal of International Economics, Vol. 24, No 1–2, pp. 45– 68.
- Search Google Scholar
- Export Citation
Hansen, B.E., 1997, “Inference in TAR Models,” Studies in Nonlinear Dynamics and Econometrics, Vol. 2, pp. 1– 14.
- Search Google Scholar
- Export Citation
Heckscher, E.F., 1916, “Växelkurens grundval vid pappersmynfot,” Economisk Tidskrift Vol. 18, pp. 309– 312.
- Search Google Scholar
- Export Citation
Imbs, J., H. Mumtaz, M. Ravn and H. Rey, 2005, “Nonlinearities and Real Exchange Rate Dynamics,” Journal of the European Economic Association, Vol. 1, No 2-3, pp. 639- 649.
- Search Google Scholar
- Export Citation
Isard, P., 1977, “How Far Can We Push the Law of One Price,” American Economic Review, Vol. 67, No. 5. (Dec., 1977), pp. 942– 948.
- Search Google Scholar
- Export Citation
Juvenal, L. and M.P. Taylor, 2007, “Threshold Adjustment of Deviations from the Law of One Price,” University of Warwick, mimeo.
- Search Google Scholar
- Export Citation
Kapetanios, G. and Y. Shin, 2006, “Unit Root Tests in Three-Regime SETAR Models,” Econometrics Journal, Vol. 9, No. 2, pp. 252– 278.
- Search Google Scholar
- Export Citation
Mejía, P., 1999, “Classical Business Cycles in Latin America: Turning Points, Asymmetries and International Synchronization”, Estudios EconÓmicos, Vol. 14, pp. 265– 297.
- Search Google Scholar
- Export Citation
Mejía, P., 2000, “Asymmetries and Common Cycles in Latin America: Evidence from Markov Switching Models,” Economía Mexicana Vol. 9, pp. 189– 225.
- Search Google Scholar
- Export Citation
Obstfeld, M. and K. Rogoff, 2000, The Six Major Puzzles in International Economics. Is there a common cause? In: Bernanke, B. and Rogoff, K. (Eds.), National Bureau of Economic Research Macroeconomics Annual 2000. NBER and MIT Press, Cambridge, MA.
- Search Google Scholar
- Export Citation
Obstfeld, M. and A.M. Taylor, 1997, “Nonlinear Aspects of Goods-Market Arbitrage and Adjustment: Heckscher.s Commodity Points Revisited,” Journal of Japanese and International Economics Vol. 11, No. 4, 411– 479.
- Search Google Scholar
- Export Citation
O’Connell, P.G.J., 1998, “The Overvaluation of the Purchasing Power Parity,” Journal of International Economics Vol. 44, 71– 95.
- Search Google Scholar
- Export Citation
O’Connell, P.G.J. and Wei, S.-J., 2002, “The Bigger they are, the Harder they Fall: Retail Price Differences Across US cities,” Journal of International Economics Vol. 56, pp. 21– 53.
- Search Google Scholar
- Export Citation
Peel, D.A. and M.P. Taylor, 2002, Covered Interest Rate Arbitrage in the Interwar Period and the Keynes-Einzig Conjecture. Journal of Money, Credit and Banking Vol. 34, No 1, pp. 51– 75.
- Search Google Scholar
- Export Citation
Richardson, J.D., 1978, “Some Empirical Evidence on Commodity Arbitrage and the Law of One Price,” Journal of International Economics Vol 8, pp. 341– 351.
- Search Google Scholar
- Export Citation
Sarno, L., M.P. Taylor and I. Chowdhury, 2004, Nonlinear dynamics in deviations from the law of one price: a broad-based empirical study. Journal of International Money and Finance, Elsevier, Vol. 23, No 1, pp. 1– 25.
- Search Google Scholar
- Export Citation
Sercu, P.; Uppal, R. and Van Hulle, C., 1995, “The Exchange Rate in the Presence of Transaction Costs: Implications for Tests of Purchasing Power Parity,” The Journal of Finance, Vol. 50, No 4, pp. 1309– 1319.
- Search Google Scholar
- Export Citation
Tong, H., 1990, “Nonlinear Time Series. A Dynamic System Approach,” Clarendon Press, Oxford, UK.
- Search Google Scholar
- Export Citation
Torres, A., 2000, “Estabilidad en Variables Nominales y el Ciclo Económico: El Caso de México,” Documento de Investigación 2000-03, Banco de México.
- Search Google Scholar
- Export Citation

Prepared by Rodolphe Blavy and Luciana Juvenal. We would like to thank the staff of the Banco de Mexico for their helpful comments, Steven Phillips for his contributions at various stages of this paper, and Roberto Benelli, Roberto Garcia-Saltos, David Robinson, Lucio Sarno, and seminar participants at the IMF and at the LACEA conference. We also thank Modupeh Williams for excellent editorial assistance.

There is now an established literature on nonlinear behavior of sectoral real exchange rates for developed markets (see Obstfeld and Taylor, 1997; Sarno, Taylor and Chowdhury, 2004; Imbs and others, 2003; and Juvenal and Taylor, 2007),

An important contribution of our paper is methodological: to use recently developed testing techniques to confirm whether the autoregressive process outside the threshold band is different from the random walk observed inside the band.

Heckscher (1916) first pointed out at the possibility of nonlinearities in relative prices in the presence of trade frictions. In the case of Mexico, Gonzalez and Rivadeneyra (2004) investigate the LOOP between Mexican cities and provide empirical evidence that transactions costs (including tariff and non-tariff barriers) explain departures from the LOOP.

As noted in Hansen (1997), the conventional tests have asymptotic nonstandard distributions, approximated using a bootstrap procedure.

A forthcoming working paper discusses the econometric methodology in greater detail.

The data sources for the CPI indices are the Bank of Mexico, the U.S. Bureau of Labor Statistics and Statistics Canada. Monthly nominal exchange rates are period averages taken from the International Financial Statistics (IFS) of the International Monetary Fund (IMF).

To gauge the sensitivity of empirical results to underlying assumptions and variable definitions, we conduct three robustness checks. First, we consider the possibility of long-run trends in the measured price differentials arising from aggregation issues in price indices or from the presence of nontradable components (Harrod-Balassa-Samuelson effect). We also test the sensitivity of the results to (i) allowing for a different mean over the 1994–12 to 1995–12 period (corresponding to the Tequila Crisis), and (ii) restricting the estimation period to 1996–2006. Overall, our baseline findings are robust to these checks.

One possible alternative explanation for finding that thresholds are lower between the U.S. and Canada than between Mexico and the U.S. may be that goods are more homogenous between the first two countries. More generally, the comparability of the sectors may vary across country pairs. First, wealth effects may be at play. The relatively large income differences between Mexico and the U.S. and Canada affects the specific goods sampled in each CPI category. This may complicate the analysis, with the composition between luxury, middle, and ordinary products varying across countries. Second, statistical differences exist in the compilation of price level data, notably in adjustments for quality changes. A solution to this problem is to look at more disaggregated price indices and SRERs. Preliminary work on this is reported in Box 1.

Previous studies computed the half-life of the SETAR model in the outer regime, which depends on the parameter ρ, as in a linear model (ln (0.5)/ln(ρ)). This has the limitation that it does not consider the regime switching that takes place within and outside the band and provides misleading results. We compute the half-life taking into account the regime-switching nature of the SETAR model. This is important in the context of our model because the half-life takes different values depending on whether the SRER is within or outside the threshold band. The half-life is infinite with the threshold band and depend on ρ (more exactly, equal to ln(0.5)/ln(ρ)) outside the band. We compute the half-lives for a 10 percent, 20 percent, 30 percent, 40 percent and 50 percent shocks by stochastic simulation using the generalized impulse response functions procedure developed by Koop and others (1996).

Note that the three NAFTA countries studied here are all relatively large in terms of land area, so that for example the distance between two cities within a given country could well exceed the distance between two cities in different countries. This situation contrasts with the literature on price convergence within Europe.

Other Publishers

Save
Cite
Email this content

Share Link

Copy this link, or click below to email it to a friend
Email this content
or copy the link directly:

https://www.elibrary.imf.org/view/journals/002/2007/378/article-A003-en.xml
The link was not copied. Your current browser may not support copying via this button.

Link copied successfully

Mexico: Selected Issues

Author: International Monetary Fund

Volume/Issue:: Volume 2007: Issue 378

Publisher:: International Monetary Fund

ISBN:: 9781451956061

ISSN:: 1934-7685

Pages:: 79

DOI:: https://doi.org/10.5089/9781451956061.002

Keywords
III. Mexico’s Integration into NAFTA Markets: A View from Sectoral Real Exchange Rates and Transaction Costs1
References

View raw image

Extent of Price Convergence between Mexico—U.S. and Canada—U.S. Pre-NAFTA

View raw image

Estimated transaction costs (in percent /1): Mexico - U.S.: Pre-NAFTA vs. Post-NAFTA

View raw image

Canada - U.S. vs. Mexico - U.S. (post-NAFTA)

Table 5.

Threshold Regressions

Determinants of Thresholds
	(1)	(2)
Distance	-0.041 ^*	-0.040 ^*
Dummy post NAFTA	-0.109 ^**	-0.116 ^**
Exchange Rate Volatility	4.020 ^***	3.991 ^***
Firms	-0.001	-
Tradability	-0.042	-
R²	0.33	0.33
N	94	94

Notes: ^*** indicates a 1 percent degree of confidence, ^** indicates a 5 percent degree of confidence, ^* indicates a 10 percent degree of confidence.

Topics

Countries

Series

About

Resources

Abstract

III. Mexico’s Integration into NAFTA Markets: A View from Sectoral Real Exchange Rates and Transaction Costs1

A. Introduction

B. Nonlinear Dynamics in Sectoral Real Exchange Rates

Theoretical underpinnings

Estimation Methodology and SETAR Model

C. Estimation Results

Testing for nonlinear price convergence

Estimated Transaction Costs8

Half-Lives

D. Determinants of Thresholds in Real Exchange Rates

E. Summary of Results and Conclusion

References

Same Series

Other IMF Content

Other Publishers

Inter-American Development Bank

The World Bank

Share Link

Table of Contents

Headings

Figures

Metrics

III. Mexico’s Integration into NAFTA Markets: A View from Sectoral Real Exchange Rates and Transaction Costs¹

Estimated Transaction Costs⁸