Financial Development and Trade Liberalization

October 29, 2020 at 06:25 pm

RESEARCH DIVISION

Working Paper Series

David Kohn,

Fernando Leibovici

and

Michal Szkup

Working Paper 2020-038A

https://doi.org/10.20955/wp.2020.038

October 2020

FEDERAL RESERVE BANK OF ST. LOUIS

Research Division

P.O. Box 442

St. Louis, MO 63166

The views expressed are those of the individual authors and do not necessarily reflect official positions of the Federal Reserve Bank of St. Louis, the Federal Reserve System, or the Board of Governors.

Federal Reserve Bank of St. Louis Working Papers are preliminary materials circulated to stimulate discussion and critical comment. References in publications to Federal Reserve Bank of St. Louis Working Papers (other than an acknowledgment that the writer has had access to unpublished material) should be cleared with the author or authors.

Financial Development and Trade Liberalization∗

David Kohn

Pontiﬁcia Universidad Cat´olica de Chile

Fernando Leibovici

Federal Reserve Bank of St. Louis

Michal Szkup

University of British Columbia

October 2020

ABSTRACT

We study the role of ﬁnancial development on the aggregate eﬀects and welfare implications of reducing international trade barriers on production inputs such as physical capital and intermediates. We document that ﬁnancially underdeveloped economies feature a slower response of real GDP, consumption, and investment following trade liberalization episodes that improve access to imported production inputs. We set up a quantitative general equilibrium model with heterogeneous ﬁrms subject to ﬁnancial constraints and estimate it to match salient features from Colombian plant-level data. We ﬁnd that the adjustment to a decline of import tariﬀs on physical capital and intermediate inputs is signiﬁcantly slower in ﬁnancially underdeveloped economies in line with the empirical evidence. Moreover, we ﬁnd that ﬁnancial development increases the welfare gains from trade liberalization; low-income agents beneﬁt from higher wages while exporters beneﬁt from a depreciated real exchange rate and lower capital costs.

Keywords: ﬁnancial development, trade liberalization, welfare. JEL: F1, F4.

∗We thank George Alessandria for helpfully discussing an earlier version of this paper. We thank Marcela Eslava, Franz Hamann, and Daniel Osorio for their advise on various aspects of the Colombian data. We also thank seminar participants at the Society of Economic Dynamics, University of British Columbia, National Bank of Poland, Insper, Universidad Alberto Hurtado, Universidad de Chile, NYU Alumni Conference, Cordoba Economics Meeting, Sociedad de Economia Chilena, European Econometric Society Meeting, Uni- versidad Di Tella, Central Bank of Chile, LACEA-LAMES Meeting, Universidad Diego Portales, RIDGE International Macro Workshop, and Econometric Society Virtual World Congress. We thank Alvaro Cordero, Gonzalo Garcia, Vicente Jimenez, and Matthew Famiglietti for excellent research assistance. David Kohn acknowledges ﬁnancial support from CONICYT, FONDECYT Iniciaci´on 11170457. The views expressed in this paper are those of the individual authors and do not necessarily reﬂect oﬃcial positions of the Federal Reserve Bank of St. Louis, the Federal Reserve System, or the Board of Governors. Emails: davidkohn@uc.cl, ﬂeibovici@gmail.com, m.szkup@ubc.ca.

1 INTRODUCTION

A key channel through which openness to international trade promotes growth and economic development is by making production inputs cheaper. Cheaper access to imports of physical capital and intermediate inputs allow ﬁrms to increase their production scale and reduce their prices, leading to increased real GDP and consumption growth (Wacziarg and Welch 2008, Estevadeordal and Taylor 2013, Amiti and Konings 2007). Despite these potential beneﬁts, trade liberalization is often resisted as a policy to promote economic development, particularly in less developed economies. In this paper, we investigate whether these economies have less to gain from trade liberalization.

A salient feature of developing economies that may limit the gains from lowering international trade barriers are frictions in ﬁnancial markets (Manova 2013, Caggese and Cunat 2013, Kohn et al. 2016, 2020, Brooks and Dovis 2019). Credit market frictions can limit the degree to which ﬁrms are able to adjust their production in response to lower prices of imported physical capital and intermediates, thereby limiting the degree to which ﬁrms in these economies beneﬁt from trade liberalization. As a result, the impact of trade liberalization on the economy might be severely mitigated.

In this paper we document that, indeed, ﬁnancially underdeveloped economies grow slower than their ﬁnancially developed counterparts after reducing international trade barriers on capital and intermediate inputs. Our empirical approach extends the analysis conducted by Estevadeordal and Taylor (2013) to investigate the role of ﬁnancial development on the dynamics of key aggregate outcomes following trade liberalization. We ﬁnd that ﬁnancially underdeveloped economies feature a persistently slower response of real GDP, consumption, and investment following trade liberalization.

We then investigate the extent to which cross-country diﬀerences in ﬁnancial development can account for the diﬀerential response that we document across countries following trade liberalization. To do so, we study a quantitative general equilibrium model of international trade with ﬁrms heterogeneous in productivity and subject to ﬁnancial frictions. We estimate the model to match salient features of Colombian plant-level data and use it to quantify the implications of reducing international trade barriers on imports of physical capital and intermediate inputs. Consistent with the data, we ﬁnd that diﬀerences in ﬁnancial development signiﬁcantly reduce the short- and medium-run impact of trade liberalization on real GDP, consumption, and investment. Finally, we ﬁnd that ﬁnancial development not only decreases welfare gains from trade liberalization but makes welfare gains more unequal across agents with exporters and rich agents beneﬁting more than poor and less-productive,non-exporters.

We make three contributions to the existing literature that studies the eﬀects of trade liberalization. First, we use cross-country data to document that trade liberalization has stronger positive eﬀects on GDP, capital, and consumption in economies that are more ﬁnancially developed. Our second contribution is to set up a model to quantify the impact of ﬁnancial development on the aggregate gains from a unilateral trade liberalization. Finally, we use our model to investigate the distributional eﬀects of trade liberalizations and how these vary with ﬁnancial development. In doing so, we contribute to a recent literature that studies the unequal distribution of gains from trade (Autor et al., 2014, 2016, Carroll and Hur, 2020a,b, among others).

Our model consists of a small open economy populated with entrepreneurs that are heterogeneous in productivity who can trade internationally and are subject to ﬁnancing con- straints. Entrepreneurs produce diﬀerentiated varieties that they can sell both domestically and abroad, subject to trade costs as in Melitz (2003). Production requires physical capital, labor, and intermediate inputs (i.e. materials), and entrepreneurs face a collateral constraint that limits the amount they can borrow to accumulate capital. Capital and materials are produced using both domestic and imported varieties, with the latter subject to tariﬀs.

We then calibrate the model to match key features of Colombian plant-level data and use it to quantify the role of credit frictions in shaping the economy's response to trade liberalization. To do so, we investigate the aggregate dynamics as well as the welfare and distributional eﬀects in response to a unilateral reduction of import tariﬀs on physical capital and intermediate inputs. We contrast our ﬁndings with those corresponding to an economy with higher ﬁnancial development. Consistent with the data, we ﬁnd that diﬀerences in ﬁnancial development signiﬁcantly reduce the short- and medium-run impact of trade liber- alization.

Lower import tariﬀs on capital and intermediates allow ﬁrms to access these goods at a lower cost, leading them to increase their production scale and thus aggregate output, consumption, and investment. Access to ﬁnance plays a key role in allowing ﬁrms to do so, since cheaper inputs increase their optimal scale, increasing the optimal amount of physical capital that needs to be accumulated. Therefore, in economies with underdeveloped ﬁnancial markets, the adjustment of ﬁrms' production scale in response to trade liberalization is substantially slower.

Cross-country diﬀerences in ﬁnancial development have signiﬁcant implications for the welfare and distributional eﬀects of trade liberalization. First, we ﬁnd that the welfare gains from trade liberalization are larger in ﬁnancially developed economies since ﬁrms in these economies are able to reap the beneﬁts from trade liberalization quicker than those in ﬁnancially underdeveloped economies. Second, we ﬁnd that ﬁnancial frictions exacerbate

the unequal distribution of the gains from trade liberalization: These are larger among productive constrained ﬁrms, as cheaper capital and intermediate inputs allow them to relax their borrowing constraints, facilitating achieving their optimal scale of operation. Moreover, welfare gains are larger across wealthy entrepreneurs and exporters. In addition, we (I) decompose the welfare eﬀects of trade liberalization to identify the key channels that account for our results, and (II) document the impact of trade liberalization on consumption, income, and wealth inequality.

Consistent with previous studies from the literature, our ﬁndings suggest that the impact of reducing international trade barriers on imports of capital and intermediate inputs diﬀers substantially from lowering barriers on imports of ﬁnal consumption goods. Yet, we ﬁnd that the overall impact of reducing import tariﬀs on both consumption and production inputs is dominated by the latter.

Our ﬁndings thus provide a rationale for the higher resistance to trade liberalization in less developed economies: There might simply be less to gain from trade openness in these economies, particularly in the short- and medium-run, given the impact of frictions in ﬁnancial markets in slowing down the adjustment to the post-liberalization environment.

RELATED LITERATURE

Our paper contributes to several strands of the literature that investigate the impact of trade liberalization. On the one hand, our work is related to a broad empirical literature on the aggregate impact of trade liberalization. For instance, Sachs et al. (1995) and Wacziarg and Welch (2008) consider a large number of trade liberalization episodes and show that, on average, they are associated with an increase in GDP and investment.1 However, they document that there is substantial heterogeneity across countries. Similar patterns have been uncovered by Greenaway et al. (1997). Our ﬁndings suggest that some of this heterogeneity can be explained by diﬀerences in ﬁnancial development as well as by diﬀerences across the types of goods being liberalized. Irwin (2019) provides a detailed survey of recent empirical work on the eﬀects of lower import barriers on economic growth.

On the other hand, our work is related to recent quantitative studies that evaluate the gains from trade liberalization using dynamic models with heterogeneous ﬁrms in the spirit of Melitz (2003). Recent papers in this literature include Costantini and Melitz (2008), Alessandria and Choi (2014), Alessandria et al. (2018), Fieler et al. (2018), Carroll and Hur (2020a,b), Burstein and Melitz (2011), among others. These and other related studies point to the importance of ﬁrm-level dynamics in accounting for the aggregate eﬀects of lowering

1See also Pavcnik (2002), Goldberg et al. (2009), and Topalova and Khandelwal (2011).

international trade barriers. Alessandria et al. (2020) provide a detailed review of recent developments in this literature.

Recent empirical studies point to the importance of reducing international trade barriers on imports of capital and intermediate inputs for the gains from trade liberalization. For instance, Estevadeordal and Taylor (2013) point to the key role played by improved access to both of these types of goods in allowing countries to grow following trade liberaliza- tion. Similarly, Amiti and Konings (2007) documents the diﬀerential impact on productivity of improving access to imported intermediate inputs rather than ﬁnal consumption goods. Goldberg et al. (2010), Gopinath and Neiman (2014), and Halpern et al. (2015), among others, also point to the importance of access to imported intermediates as a source of growth. We contribute to this literature by investigating the role of diﬀerences in ﬁnancial development on the impact of reducing international trade barriers on imported capital and intermediate inputs.

As in Ravikumar et al. (2017) and Anderson et al. (2015), we ﬁnd that capital accumulation and access to intermediates play a key role on the response to trade liberalization. Motivated by their ﬁndings, our work focuses on the role of international trade barriers on capital and intermediate inputs and its interaction with ﬁnancial development. Thus, we contribute to the literature that investigates the impact of ﬁnancial development on the gains from trade liberalization (Caggese and Cunat, 2013 and Brooks and Dovis, 2019). We show that the role of ﬁnancial frictions on the gains from trade liberalization depends crucially on the type of goods being liberalized and, in particular, that the dynamics of the economy following improved access to capital and intermediate inputs is shaped by the level of ﬁnancial development.

More broadly, we contribute to a large literature that studies the aggregate consequences of ﬁnancial frictions. Buera et al. (2011), Midrigan and Xu (2014), Moll (2014) show that ﬁnancial frictions induce capital misallocation, leading to potentially signiﬁcant aggregate distortions. Kohn et al. (2016, 2020), Leibovici (2016), Manova (2013), and Chaney (2013), among others, study the impact of ﬁnancial frictions on trade ﬂows. We complement these studies by investigating how ﬁnancial frictions shape the response of the economy to a trade liberalization.

In the next section we conduct the empirical analysis. We set up the model in Section 3. In Section 4 we discuss the mechanism through which ﬁnancial frictions aﬀects the response to trade liberalization in our model. In Section 5 we present the quantitative results. Section 6 concludes.

Empirical evidence

In this section, we document salient features of the dynamics of key aggregates after reducing barriers to international trade. We begin by contrasting countries that reduced average import tariﬀs relative to countries that did not. Then, we investigate whether these eﬀects diﬀer by ﬁnancial development. To isolate the role of ﬁnancial development in allowing countries to adjust productive capacity to trade liberalization, we focus on tariﬀ reductions on imports of production inputs such as capital or intermediate goods.

Our data and approach build on the work of Estevadeordal and Taylor (2013), which we extend to investigate the implications for a broader set of aggregate outcomes as well as the role of ﬁnancial development. Following their work, we focus on trade liberalization episodes that took place during the 1990s (the period that Estevadeordal and Taylor (2013) refer to as the "Great Liberalization" experiment).

2.1 Data

Our empirical analysis employs three types of data. First, we use data on import tariﬀs to identify trade liberalization episodes. Second, we use data on the level of ﬁnancial development to partition countries undergoing trade liberalization based on the intensity of the ﬁnancial frictions. Finally, we use data on aggregate variables whose dynamics we contrast across countries based on whether they reduced trade barriers or by their level of ﬁnancial development. We now describe the data we use in more detail.

Import tariﬀs We utilize two data series to measure the degree of trade openness across countries: average import tariﬀs across all goods and average tariﬀs on imports of capital and intermediate inputs. We follow Estevadeordal and Taylor (2013) and measure tariﬀs at two points in time: before or early in the trade liberalization process (1985-1993), and late or after the trade liberalization took place (1999-2004). All averages are computed without weights. Both series are obtained from Estevadeordal and Taylor (2013).

Average import tariﬀs across all goods are collected from the Economic Freedom in the World 2005 database and available for 75 countries. Average import tariﬀs on capital and intermediate inputs are most favored nation (MFN) applied tariﬀs and compiled for 49 countries. The latter tariﬀs were are obtained from UNCTAD's TRAINS (Trade Analysis and Information System) for the post-liberalization period, were compiled from a variety of sources by Estevadeordal and Taylor (2013) for the pre-liberalization tariﬀs.

Aggregate outcomes We use data from the Penn World Tables 9.1 (Feenstra et al. 2015) to document the dynamics of the following variables: GDP, consumption, investment, investment of machinery and non-transport equipment, exports, and imports. All variables are per capita and expressed in constant prices at national currency.

Financial development We measure the degree of ﬁnancial development across countries using the World Bank's Global Financial Development database (Cihak et al. 2012). We restrict attention to the amount of domestic credit provided to the private sector as a share of GDP (GFDD.DI.14), which is the most popular measure used in the literature on consequences of ﬁnancial development (see, for example, King and Levine, 1993 or Manova, 2013).

2.2 Aggregate dynamics after trade liberalization

We begin by investigating the extent to which countries that underwent trade liberalization experienced aggregate dynamics that diﬀered systematically from their counterparts whose trade barriers did not change.

Trade liberalization episodes

We focus on average import tariﬀs and identify countries that liberalized international trade based on the change of these tariﬀs between the pre-liberalization period (1985-1993) and the post-liberalization period (1999-2004). As in Estevadeordal and Taylor (2013), we identify liberalizers as those countries whose import tariﬀ change was below the median over this period.

Table 1: Trade liberalization episodes based on all goods

	Avg. tariﬀ change	Avg. initial tariﬀ	Avg. ﬁnal tariﬀs	# of countries

Nonliberalizers	−0.91%	10.08%	9.17%	43
Liberalizers	−18.58%	30.87%	12.29%	42

Table 1 presents the average initial and ﬁnal tariﬀs as well as the average tariﬀ change across liberalizers and nonliberalizers. We ﬁnd that nonliberalizers experienced a negligible decline of average tariﬀs over this period, a reduction of 0.91 percentage points on average. Moreover, these countries already had low tariﬀs in the pre-liberalization period, 10.08% on average. In contrast, we ﬁnd that liberalizers reduced their tariﬀs by 18.58 percentage points on average. These countries had very high tariﬀs in the pre-liberalization period,

30.87% on average, but they almost closed the gap in tariﬀ levels with nonliberalizers in the post-liberalization period.2

Aggregate dynamics

We now investigate dynamics of key aggregate outcomes following a reduction in international trade barriers, contrasting countries that liberalized trade with those that did not. As described above, we focus on the dynamics of GDP, consumption, investment, investment in machinery and equipment, exports, and imports.

To do so, for each outcome variable in each country, we ﬁrst express it as a log deviation relative to its value in 1975. Then, for each year and variable, we compute the average log deviations relative to 1975 across liberalizers, and we then do the same for nonliberalizers. Liberalizers and nonliberalizers might have experienced diﬀerent dynamics not only due to the impact of trade liberalization but also because they were already featuring diﬀerent trend levels of growth prior to trade liberalization. We address this issue by expressing the variables as log deviations relative to a linear trend whose slope captures the average growth rate on the decade prior to the pre-liberalization period: 1975-1985.3

Figure 1 plots the average dynamics of each of the aggregate variables of interest across countries that liberalized international trade and those that did not, computed as described above.We ﬁnd that all variables exhibit similar dynamics across liberalizers and nonliberal- izers over the period 1975 to 1990; one exception is exports, which is already 0.20 log points higher across liberalizers by 1990. Their paths, however, diverge widely from 1990 onwards. For instance, real GDP in countries that reduced import tariﬀs is 0.20 log points higher in year 2000 relative to 1975 than it would have been expected based on the 1975-1985 trend. In contrast, real GDP in nonliberalizer countries did not diﬀer in year 2000 relative to 1975 from the value one would have expected based on the 1975-1985 trend.

While we observe very similar patterns for consumption and investment, there is a large diﬀerence in magnitudes across these variables. The diﬀerence in investment between liberalizers and nonliberalizers is approximately equal to 0.40 log points by the end of the period, whereas the gap in consumption across these country groups only reaches 0.10 log points by the early 2000s. Moreover, this gap is even wider when restricting attention to investment in machinery and equipment, allowing us to abstract from residential investment and investment in structures which is less aﬀected by changes in import tariﬀs. Liberalizers feature investment in machinery and equipment that is approximately 0.60 log points

In Section 5 below, we investigate the response of the economy to a change in import tariﬀs of similar magnitude to the one reported here, under diﬀerent varying degrees of ﬁnancial development.
We normalize the trend to ensure that the deviations from the trend are equal to 0 in 1975.

log GDP, chg. from 1975	.15	log C, chg. from 1975			log I, chg. from 1975
.3									.6
.2					.1					.4
.1					.05					.2
0					0					0
-.1	1980	1990	2000	2010	-.05	1980	1990	2000	2010	-.2	1980	1990	2000	2010
1970	1970	1970
		Year					Year					Year

log I (M&E), chg. from 1975		log X, chg. from 1975		log M, chg. from 1975
.6					.8					.8
.4					.6					.6
.2					.4					.4
0										.2
				.2					0

-.2					0					.2
										-
1970	1980	1990	2000	2010	1970	1980	1990	2000	2010	1970	1980	1990	2000	2010
		Year					Year					Year
			Trade liberalization			No trade liberalization

Figure 1: Aggregate dynamics after trade liberalization

higher than their nonliberalizer counterparts by the end of the sample. Finally, we observe that liberalizers also feature a larger increase of both imports and exports following trade liberalization.

The above ﬁndings are consistent with those of Wacziarg and Welch (2008) and Es- tevadeordal and Taylor (2013) and we interpret this evidence as suggesting that capital accumulation plays a key role on the impact of trade liberalization.

2.3 Trade liberalization of production inputs and ﬁnancial development

The above ﬁndings and together with the results of earlier studies suggest that capital accumulation is likely to play a key role in accounting for the aggregate dynamics observed following trade liberalization. One implication of this ﬁnding is that insofar ﬁnancial un- derdevelopment prevents ﬁrms from undertaking their desired investment decisions, it might slow down the response of capital and other aggregate variables after trade liberalization decreasing potential gains from reduction of tariﬀs. We now investigate the extent to which this is the case by restricting attention to tariﬀ reductions on imports of capital and inter-

mediate inputs, which are more likely to increase the economy's desired productive capacity than tariﬀ reductions on other types of goods.

Trade liberalization episodes

As in the previous subsection, we classify countries as undergoing trade liberalization if the average change in their import tariﬀs for capital and intermediate inputs is below the median across countries. We then partition trade liberalizers as ﬁnancially underdeveloped if their average domestic credit to GDP ratio over the whole period 1975-2004 is below the median across countries.

Table 2 presents the average tariﬀ changes and credit-to-GDP ratios across nonliberal- izers, liberalizers that are ﬁnancially underdeveloped, and liberalizers that are ﬁnancially developed. As in the analysis above, we observe that nonliberalizers feature much smaller tariﬀ changes than liberalizers. Yet, we ﬁnd that ﬁnancially underdeveloped liberalizers on average reduced their import tariﬀs relatively more than countries with high credit-to-GDP ratios: 22.65 vs. 15.19 percentage point reductions in import tariﬀs, respectively.

Table 2: Trade liberalization episodes based on capital and intermediate inputs

	Avg. tariﬀ change	Avg. credit/GDP	# of countries

Nonliberalizers	−2.17%	0.62	25
Liberalizers, low credit/GDP	−22.65%	0.24	12
Liberalizers, high credit/GDP	−15.19%	0.66	11

Table 2 also shows that the partition of liberalizers by ﬁnancial development features signiﬁcant diﬀerences in their average credit to GDP ratios over the 1975 to 2004 period. While ﬁnancially underdeveloped liberalizers have an average credit to GDP ratio equal to 0.24, this value is 2.75 times larger on average across countries classiﬁed as ﬁnancially developed.4

Aggregate dynamics

We now investigate the extent to which ﬁnancially underdeveloped economies indeed experienced a slower adjustment to trade liberalization than ﬁnancially developed economies. As described above, we restrict attention to the average change of each outcome variable relative to 1975 after removing a linear trend for the growth over the period 1975-1985.

4 In Section 5 below, we investigate the response of a ﬁnancially underdeveloped economy to a change in import tariﬀs of production inputs. Our baseline economy features a Credit-to-GDP ratio equal to 0.24.

	GDP elasticity			C elasticity			I elasticity
1					1					3

.5					.5					2
										1
0					0
										0
-.5					-.5					-1
1970	1980	1990	2000	2010	1970	1980	1990	2000	2010	1970	1980	1990	2000	2010
		Year					Year					Year

	I M&E elasticity			X elasticity			M elasticity
6					3					4

4					2
										2
2					1
0					0					0

-2					-1					-2
1970	1980	1990	2000	2010	1970	1980	1990	2000	2010	1970	1980	1990	2000	2010
		Year					Year					Year
	Trade liberalization, low credit			Trade liberalization, high credit

Figure 2: Trade liberalization of production inputs and ﬁnancial development

Given that economies with high vs. low credit-to-GDP ratios experienced trade liberalizations of diﬀerent magnitudes, as shown in Table 2, we compute the elasticity of each outcome variable to changes in tariﬀs. In the rest of this section we therefore abstract from countries that did not reduce tariﬀs. Figure 2 plots the average dynamics of each of the aggregate variables of interest across economies with low vs. high average credit-to-GDP ratios that reduced import tariﬀs on capital and intermediate inputs.

We ﬁnd that all variables exhibit similar dynamics over the period 1975-1990 across both types of economies, but the path of most variables diverges signiﬁcantly from around 1990 onwards; one exception is exports, whose dynamics over the whole sample is very similar across both types of economies. For instance, while the elasticity of GDP to changes in tariﬀs by the end of the sample is approximately equal to one in ﬁnancially developed economies, this elasticity is approximately half as large in ﬁnancially underdeveloped economies. More- over, the elasticity of all variables except exports is at least twice as large in ﬁnancially developed economies.

These ﬁndings suggest that diﬀerences in ﬁnancial development across countries might have a signiﬁcant impact on the aggregate dynamics following trade liberalization. We now investigate the extent to which this is the case using a quantitative general equilibrium model

of international trade with frictions in ﬁnancial markets.

3 Model

We consider a small open economy populated by a unit measure of entrepreneurs, representative producers of composite consumption goods, representative producers of composite investment goods, and the rest of the world. Entrepreneurs produce diﬀerentiated varieties by operating a ﬁrm and choose whether to sell their output internationally. Composite consumption and investment goods are produced by combining domestic and imported varieties. Finally, the rest of the world demands the varieties produced by entrepreneurs and is the source of imported goods.

3.1 Economic environment

3.1.1 Entrepreneurs

Entrepreneurs are inﬁnitely-lived and risk-averse. Preferences over streams of composite consumption goods are represented by the expected lifetime discounted sum of a constant

	∞		c1−γ
relative risk-aversion period utility function, E0	Pt=0	βt	t	, where γ denotes the coeﬃcient
1−γ

of relative risk aversion and β is the entrepreneurs' discount factor. Finally, E0 denotes the expectation operator taken over the realizations of productivity shocks, described below, conditional on the information set in period zero.

Technology Entrepreneurs produce diﬀerentiated varieties by operating a production technology

yt = zt ktαnt1−α1−αm mtαm ,

(1)

where zt denotes the entrepreneurs' idiosyncratic level of productivity, kt is the capital stock, nt is the amount of labor hired, and mt denotes the amount of intermediate inputs used (e.g., materials). We assume that entrepreneurs use composite investment goods as intermediate inputs. We refer to α(1 − αm) ∈ [0, 1] as the capital share and to αm ∈ [0, 1] as the share of intermediate inputs in production. Labor is hired at a wage rate wt denominated in units of ﬁnal consumption goods. Idiosyncratic productivity zt follows a time invariant AR(1) process, ln zt = (1 − ρ)µ + ρ ln zt−1 + εt, where εt is distributed Normal with mean zero and standard deviation σε.

Every period, entrepreneurs are endowed with a unit of labor that they supply inelastically to a competitive labor market. Entrepreneurs can also accumulate capital internally

by transforming composite investment goods purchased in period t into physical capital in period t + 1. Capital depreciates at rate δ after being used for production, leading to a law of motion for capital that is given by

kT+1 = (1 − δ)kT + xT,

(2)

where xT denotes gross investment.

MARKET STRUCTURE Entrepreneurs are monopolistically competitive and choose the quantities and prices at which to sell domestically and abroad subject to their respective demand schedules. In the domestic market, the demand schedules solve the problem of the producers of composite consumption and investment goods, while the demand schedule in the international market is taken as given from the rest of the world.

INTERNATIONAL TRADE Entrepreneurs can choose to export, but exporting entails additional variable and ﬁxed costs. In particular, ﬁrms pay a ﬁxed cost F , in units of labor, every period that they export. Furthermore, exporters are subject to two ad-valorem trade costs: (i) a tariﬀ τF charged by the rest of the world on the domestic economy's exports, and (ii) an iceberg trade cost τ ≥ 1 which requires ﬁrms to ship τ units for every unit that arrives at destination. τ captures variable costs other than tariﬀs, such as shipping costs, foreign marketing costs, or costs due to damages incurred during transit of goods.

FINANCIAL MARKETS Individuals have access to international ﬁnancial markets, where they can borrow or save by trading a one-periodrisk-free bond at real interest rate r. The interest rate is taken as given from the rest of the world. However, entrepreneurs face a borrowing constraint that limits the amount they can borrow. They can only borrow up to a fraction of the value of the capital stock at the time that the loan is due for repayment.

In particular, let dT+1 denote the amount borrowed by entrepreneur i in period t, which is due for repayment in period t + 1. Then, in addition to the natural borrowing limit, dT+1 has to satisfy

dT+1 ≤ θPK,TkT+1,

(3)

where θ ∈ [0, 1], and PK,T is the price of capital in period t so that PK,TkT+1 captures the current price of the total capital stock owned by the entrepreneur.

We denote the net worth of entrepreneurs in period t as aT, which is given by aT+1 = PK,TkT+1 − dT+1/(1 + r). Given this deﬁnition, the borrowing constraint can be alternatively

expressed as:	1 + r
Pk,tkt+1 ≤	(4)
	at+1
1 + r − θ

Equation (4) shows that the borrowing constraint faced by entrepreneurs limits the amount of capital that they can operate with. In particular, the current value of next period's capital stock has to be lower than a multiple of the entrepreneur's net worth in period t + 1.5 Note also that the tightness of the borrowing constraint is increasing in the price of capital.

Timing The timing of the entrepreneurs' decisions is as follows. At the beginning of the period, entrepreneurs hire labor and purchase intermediate inputs to produce their diﬀerentiated variety to be sold domestically and possibly also abroad. If they decide to export, they also pay ﬁxed export costs. Entrepreneurs are paid their labor and interest income, and then use these resources to repay debt due from the previous period as well as to consume and save up for next period.

At the end of the period, agents observe the following period's productivity shock. Then, they issue debt and choose next period's level of physical capital given the amount of net worth they chose to carry over.6

Entrepreneurs' problem Given the setup described above, the entrepreneurs' problem consists of choosing choose sequences of consumption (ct), labor (nt), intermediates (mt), investment (xt), export status (et), and prices and quantities (yh,t, ph,t, yf,t, pf,t) at which to sell the varieties in each of the markets (subscript h denotes the domestic market, while subscript f denotes the foreign market), in order to maximize their lifetime expected utility. In addition to the borrowing constraint described above and the market-speciﬁc demand schedules described below, entrepreneurs' choices are subject to a sequence of period-by- period budget constraints, given by

ct + Pk,txt + dt = wt + [ph,tyh,t + et (ξtpf,tyf,t − wtF ) − wtnt − Pk,tmt] +	dt+1	+ Tt, (5)

	1 + rt

where ξ is the real exchange rate. The entrepreneurs' choices are also subject to a sequence of period-by-period laws of motion for capital, kt+1 = [(1 − δ)kt + xt], and production technologies yh,t + τ yf,t = zt ktαn1t −α1−αM mαt M . Tt is a lump sum transfer that reimburses

5As discussed below, the entrepreneurs' net worth in period t + 1 is equal to their savings in period t. Since kT+1 = (1 − δ)kT + xT it follows that constraint (4) is actually a constraint on entrepreneur's investment in period t. This explains why the capital is priced at the price PK,T rather than PK,T+1.

6Following Buera and Moll (2015), this timing assumption allows us to eliminate uninsured idiosyncratic investment risk, thus simplifying the quantitative solution of the model by combining capital k and debt d, into a single state variable, net worth, a, where aT+1 = PK,TkT+1 − dT+1/(1 + r).

entrepreneurs with the revenue collected from the tariﬀs levied on imports.

3.1.2 Composite consumption good

There is a representative producer of composite consumption goods that operates a constant elasticity of substitution technology to aggregate domestic varieties produced by entrepreneurs with imported varieties produced by the rest of the world. This composite consumption good is used by entrepreneurs for consumption. Each period, the problem of the producer of composite consumption goods is then given by

Z 1

max Yc −	ph(i)yh,c(i)di − (1 + τc)ξpm,cym,c
yh,c(i),ym,c	0							(6)

		Z0					σ
	s.t. Yc =	yh,c(i)	σ di + ωcym,cσ			,
			1	σ−1	σ−1		σ−1

where τc > 0 denotes a domestic tariﬀ on imported varieties used to produce the composite consumption good and ωc is the relative weight of imported goods in the production of the consumption bundle. Notice that we normalize the price of the composite consumption good to 1.

3.1.3 Composite investment good

Similarly, there is a representative producer of composite investment goods that operates a constant elasticity of substitution technology to aggregate domestic varieties produced by entrepreneurs with imported varieties produced by the rest of the world. As described above, this composite investment good is used both to increase the stock of physical capital and as an intermediate input in the production of varieties by entrepreneurs. Each period, the problem of the investment good producer is then given by

max PkYk −	ph(i)yh,k(i)di − (1 + τk)ξpm,kym,k
yh,k (i),ym,k	0					(7)

	Z0				σ
s.t. Yk =	yh,k(i)	σ di + ωkym,kσ		,
		1	σ−1	σ−1	σ−1

where τk > 0 denotes a domestic tariﬀ on imported varieties used to produce the composite investment good and ωk is the relative weight of imported goods in the production of the investment good. Finally, Pk denotes price of the composite investment good relative to the price of the composite consumption good.

3.1.4 Import tariﬀs: Revenues and transfer

As described above, imports of varieties used to produce composite consumption goods are subject to a tariﬀ given by τc, while imports of varieties used to produce composite investment goods are subject to a tariﬀ given by τk. We assume that the total revenue collected by these tariﬀs is reimbursed to entrepreneurs as a lump-sum transfer T . Then, we have that:

T =τcξpm,cym,c + τkξpm,kym,k

(8)

3.1.5 Rest of the world

The rest of the world demand varieties from domestic entrepreneurs and supplies varieties to consumption and investment good producers. The foreign demand for domestic varieties is assumed to be given by a standard downward-sloping demand function yf,t = [(1 + τf )pf,t]−σ Yf,t, where Yf,t is an aggregate demand shifter for the rest of the world and pf,t is denominated in units of the foreign ﬁnal good. The supply of varieties used to produce composite consumption goods is assumed to be perfectly elastic at a price pm,c. Similarly, the supply of varieties used to produce composite investment goods is assumed to be perfectly elastic at the price pm,k.

Finally, the rest of the world trades bonds with domestic entrepreneurs in ﬁnancial mar- kets. As described above, these bonds trade at real interest rate r.

3.2 Recursive formulation

We now present the recursive formulation of the entrepreneurs' problem in a stationary equilibrium. Let v(k, d, z) denote the value function of an entrepreneur with capital k, debt d, and productivity z who decides how much to consume in the current period and how much to save for the future (i.e., how much net worth a′ carry to the next period). Deﬁne g(a, z) as the value function of an entrepreneur with net worth a and productivity z, who decides

how to allocate his savings between capital and debt. Recall that we deﬁne a′ = pkk′ − d′

1+r

to be the net worth that the agents accumulate for the future.7 Then, we have that:

	c1−γ	′	′
v (k, d, z) = max		+ βEz′ [g (a	, z	)]
1 − γ
c,a′

subject to	(9)

+ a′ + d = w + (1 − δ)Pkk + π(k, z)

a′ ≥ 0,

7WE ASSUME A ≥ 0.

where π(k, z) denotes the proﬁts of an entrepreneur with capital k and productivity z:

π(k, z) =

max

phyh + eξpf yf − wn − Pkm − ewF

{pH,yH,pF ,yF ,n,m,e∈0,1}

subject to

(10)

yh + eτ yf = z kαn1−α(1−αM) mαM

yh = p−h σ(Yc + PkσYk), yf = (pf (1 + τf ))−σ Yf .

The value function g(a′, z′) is given by

g(a′, z′) = max v(k′, d′, z′)

	k′,d′
subject to:	Pkk′ = a′ +	d′
1 + r

	d′ ≤ θPkk′

As in Buera and Moll (2015) and others, the above value functions can be combined such that the problem features only two state variables: productivity z and net worth a.8

3.3 Equilibrium

Let S ≡ A × Z denote the state space of entrepreneurs and let s ∈ S denote an element of the state space. Let φ denote a measure over S. Then, for a given value of the interest rate r, a recursive stationary competitive equilibrium of this economy consists of aggregate prices {w, ξ, Pk}, policy functions {d′, k′, e, c, m, n, yh, yf , ph, pf , Yc, Yk, ym,c, ym,k}, value functions v and g, and a measure φ : S → [0, 1] such that:

Policy and value functions solve the entrepreneurs' problem
Policy functions solve the problem of producers of composite consumption goods
Policy functions solve the problem of producers of composite investment goods
Labor market clears:

[n(s) + e(s)F ] φ(s)ds = 1

5. Market for consumption good clears:

c(s)φ(s)ds = Yc

8See the Appendix for the details.

6. Market for investment good clears:

[x(s) + m(s)] φ(s)ds = YK

7. Measure φ is stationary

This deﬁnition of the stationary equilibrium is standard: It requires that all agents behave optimally taking prices w, ξ, PK as given, that prices are such that markets clear, and that the measure of entrepreneurs is stationary.

Mechanism: The role of ﬁnancial development

Throughout the rest of the paper we investigate the impact of lowering tariﬀs on imports of investment and intermediate inputs and the role played by ﬁnancial development on the magnitude of these eﬀects. In this section we describe the mechanism through which this policy change aﬀects allocations in our model; in the following sections we examine these eﬀects quantitatively.

4.1 Cheaper access to imported production inputs

A unilateral reduction of τK makes imports of intermediate inputs and investment goods cheaper. This aﬀects the domestic economy through two channels. First, it reduces the cost of producing the composite investment good. As a result, both materials and capital become cheaper, decreasing the production costs. Second, it leads to a reallocation of demand by producers of the composite investment good from domestic to imported varieties. Thus, a reduction of τK has both positive and negative direct eﬀects on the domestic economy.

The change of τK also induces general equilibrium eﬀects. As domestic production costs decline, domestic producers reduce their prices and increase their competitiveness relative to imports competition. This eﬀect oﬀsets at least partially the reallocation of demand from domestic to imported varieties by producers of composite investment goods. But, more importantly, it also leads producers of composite consumption goods to reallocate their demand from imported to domestic varieties, as the latter become cheaper when tariﬀs on production inputs are reduced. Thus, total domestic sales are likely to increase. The higher demand for domestic varieties leads to an increase in the demand for labor and, thus, wages increase. The domestic economy thus becomes richer when tariﬀs on imported production inputs are reduced.

These gains, however, are not spread evenly across individuals: Productive and wealthy entrepreneurs capture most of the gains from the higher demand for domestic varieties. Yet, our model features signiﬁcant channels through which these gains are redistributed. First, equilibrium wage changes allow our economy to automatically redistribute some of these gains to entrepreneurs with lower productivity or net worth. Real wages increase in response to the higher demand for domestic varieties and this eﬀect disproportionately beneﬁts lower productivity or net worth entrepreneurs for whom wages constitute a higher share of their total income. Second, these agents are negatively aﬀected by the lower tariﬀs revenues, which aﬀect more the low net worth agents for whom these constitute a higher share of their income.

Third, our model features redistribution from the domestic economy, which is better oﬀ after the policy change, to the rest of the world, whose environment remains unchanged. The decline in the price of domestic varieties leads to a lower price of the consumption composite, leading to a real depreciation. This leads to higher exports and further increases domestic output.

Thus, we conclude that a reduction of tariﬀs on production inputs such as investment goods and intermediate inputs leads to an increase of consumption and exports. As a result, it leads to an economic boom with increased production by all entrepreneurs. While the gains are unevenly spread across entrepreneurs, signiﬁcant forces exist to redistribute the gains across individuals.

4.2 The role of ﬁnancial development

Financial development can limit the degree to which the domestic economy is able to beneﬁt from the forces described above. In an economy with less developed ﬁnancial markets (a lower θ), the pre-liberalization stationary equilibrium is likely to feature a higher share of constrained entrepreneurs. Thus, as tariﬀs and production costs are reduced, a higher fraction of entrepreneurs is unable to expand production to the extent desired, thereby limiting the degree to which ﬁrms in this environment can beneﬁt from trade liberalization.

Over time, however, entrepreneurs are able to accumulate funds internally, relaxing the borrowing constraints and increasing their scale of production closer to their desired scale. Given that the lower tariﬀs increase the proﬁtability of sales regardless of the scale of pro- duction, this allows ﬁrms to disproportionately relax their constraints, partially oﬀsetting the initial dampening eﬀects of lower ﬁnancial development. Overall, however, ﬁnancial un- derdevelopment limits the positive eﬀects of reducing τK through its impact on the share of ﬁnancially constrained entrepreneurs.

The above discussion treats the share of constrained exporters as unchanged. However, a lower τK also leads to a reduction in the share of ﬁnancially constrained exporters. This is implied by the resulting decline in the price PK of the composite investment good eﬀectively relaxes the borrowing constraint, allowing ﬁrms to purchase a higher amount of physical capital per unit borrowed in ﬁnancial markets. Thus, these eﬀect further amplify the positive impact of this policy change.

The role of ﬁnancial development on the impact of this policy change is, then, ambiguous ex ante. On the one hand, lower ﬁnancial development tends to dampen the positive direct and indirect eﬀects of a decrease in τK . On the other hand, it relaxes borrowing constraints, which tends to strengthen the eﬀects of lowering τK . Determining which eﬀect dominates as well as the aggregate and distributional eﬀects requires a careful quantitative investigation, which we perform in the following section.

5 Quantitative analysis

In this section, we investigate quantitatively how ﬁnancial development aﬀects the aggregate, distributional, and welfare eﬀects of a unilateral trade liberalization that reduces tariﬀs on imports of intermediate and capital goods.9

To quantify the role of ﬁnancial development we consider the unilateral trade liberalization that Colombia underwent in the late 1980s. We ﬁrst calibrate the model to match key features of Colombian plant-level data, an economy characterized by a low level of ﬁnancial development. We use the model to examine the aggregate eﬀects of a decrease in import tariﬀs designed to resemble the one observed in Colombia between 1988 and 1992. We contrast our baseline economy with a counter-factual economy featuring a high-level of ﬁnancial development (but otherwise calibrated to match key features of Colombian plant-level data). In particular, we compare the aggregate and distributional eﬀects, both in the long-run and along the transition, between our baseline economy and those implied by the counter-factual economy with developed ﬁnancial markets.

Colombia's unilateral trade liberalization is ideally suited for our analysis. In the early 1990s the Colombian government undertook an ambitious trade reform program in an environment with underdeveloped ﬁnancial markets. By the end of 1990, the government had eliminated the majority of non-tariﬀ import barriers, while reducing average import tariﬀs from 32 to 12 percent between 1990 and 1992. Yet, despite the large scale trade liberal-

9We focus on unilateral trade liberalizations as such were common in the 1980s and 1990s when many developing countries opened up to trade following the Washington consensus. In the Online Appendix, we present the results of a multilateral trade liberalization (i.e., trade liberalization featuring a simultaneous decrease of both import and export tariﬀs).

ization, there was substantial skepticism about the economy's potential to accrue the large potential gains from trade. In particular, at the time of trade liberalization the World Bank pointed out to the lack of credit and under-developed capital markets as a major obstacle for fully realizing the potential beneﬁts of trade reforms and was concerned that Colombian trade liberalization "is seriously constrained by the existing ﬁnancial sector" (page 5, World Bank 1993). Thus, Colombia's trade liberalization is a perfect case for analyzing the eﬀect of ﬁnancial development on the gains from a unilateral trade liberalization.

5.1 Calibration

To calibrate the model, we partition the parameter space into two groups. The parameters in the ﬁrst group are pre-assigned either to values observed in the data or to values commonly used in the literature. The parameters in the second group are estimated jointly to match key moments of plant-level and aggregate data from Colombia.

Pre-assigned parameters The ﬁrst group of parameters is presented in Table 3 and consists of γ, σ, δ, r, ρ, α, αM, τC, τK , τF , τ , pM,C, pM,K , PF , and YF . We set the risk aversion parameter γ to 2, the elasticity of substitution σ to 4, and the depreciation rate δ to 0.1; these are standard values used in the literature. In addition, we set the interest rate r to 0.06. Following previous studies (Buera et al. 2011, Midrigan and Xu 2014, Kohn et al. 2020), we also assume that idiosyncratic productivity shocks have high persistence and set

to 0.90.

We set the share of intermediates αM in the production of tradable goods equal to 0.50, consistent with plant-level data from Colombian manufactures, described above. The capital share of output is equal 0.60. Next, we set export and import tariﬀs -both to consumption and capital goods- equal to 32 percent, the value of average import tariﬀs observed in Colombia in 1988, just prior to trade liberalization. Import tariﬀs in the rest of the world are assumed to be equal to domestic import tariﬀs, equal to 0.32, and we normalize the iceberg trade cost to 1.10

Finally, we normalize the price index in the rest of the world to 1, the foreign aggregate demand to 10, and the prices of imported capital and consumption goods to 1.

10 In our model, given that we calibrate β to match the net-exports-to-GDP ratio, we cannot identify separately τ , ωC, and ωK . Any change in τ aﬀects imports-to-GDP ratio and the share of consumption imports in total imports, the two target moments that we use to calibrate ωK and ωC, respectively. Adjusting ωK and ωC to match these two moments following a change in τ undoes any impact that a change in τ has on our economy.

Table 3: Pre-assigned parameters

Parameter Value Description

2 Risk aversion

4 Elasticity of substitution

0.1 Capital depreciation rate

0.06 Interest rate

0.9 Persistence of prod. shocks

0.6 Share of capital

αm	0.5	Share of intermediate inputs
τc	0.32	Consumption imports tariﬀs
τk	0.32	Capital imports tariﬀs
τf	0.32	Exports tariﬀs

1 Iceberg trade costs

pm,c, pm,k	1	Price of ym,c, ym,k
Pf	1	Foreign price index
Yf	10	Aggregate demand rest of the world

Calibrated parameters The set of calibrated parameters consists of the ﬁxed export cost F , the standard deviation of the productivity shocks σε, the relative weights of imported goods in the production of investment and consumption goods, ωk and ωc, the degree of ﬁnancial development θ, and the discount factor β.

To estimate these parameters we target salient features of both plant-level data from Colombian manufactures and aggregate data. In particular, we study the Annual Manufacturing Survey, which is collected by the Departamento Administrativo Nacional de Estadis- tica (DANE) and surveys all manufacturing plants with at least 10 workers.11 Following Fieler et al. (2018), we use data from 1982 to 1988 to calibrate the model for the period prior to tariﬀ reduction implemented in subsequent years. We supplement this dataset with data from the World Bank.

We choose {F, σε, ωk, ωc, θ, β} to match the following moments: (i) the share of ﬁrms that export, (ii) the size of exporters relative to non-exporters (as given by the ratio between the average domestic sales of exporters and the average domestic sales of non-exporters), (iii) the share of imports in consumption, (iv) the aggregate share of imports in GDP, (v) the average amount of domestic credit extended to the private sector as a percentage of GDP between 1988 and 1992 (credit-to-GDP ratio) as reported by the World Bank, and (vi) the net exports-to-GDP ratio. In particular, we follow the Simulated Method of Moments, choosing these parameters to minimize the squared distance between the above moments of the model and their data counterparts.

11This data has been used before by Roberts and Tybout (1997), Ruhl and Willis (2017), and Fieler et al. (2018), among others.

Table 4: Calibrated parameters

Parameter	Value	Target moment	Data	Model
F	0.466	Share of exporters	0.11	0.11
σε	0.157	Exporters' domestic sales premium	5.68	5.68
ωc	0.209	C imports share	0.266	0.266
ωk	0.287	Imports / GDP	0.124	0.124
θ	0.205	Credit / GDP	0.244	0.244
β	0.835	Net exports / GDP	-0.029	-0.029

We report the calibrated parameters and the target moments in Table 4. As observed in the table, our model can match the targets moments very closely.

Financially developed economy We contrast our baseline economy described above with a counterfactual economy with developed ﬁnancial markets. Given our desire to contrast the welfare implications between these economies, we keep the discount rate unchanged across them; thus, we set β = 0.835, as in our baseline economy.12 We then estimate F , σ, ωc, ωk, and θ, to match moments (i)-(v) described above, except that we now target the average credit-to-GDP ratio across OECD countries between 1988 and 1992; a set of economically- and ﬁnancially developed economies. Thus, examining the implications of trade liberalization in the counter-factual economy allows us to quantify the eﬀect that trade liberalization would have had in Colombia had it had the level of ﬁnancial development of the most ﬁnancially developed economies at the time. Table 5 reports the calibrated parameters and targets for this counterfactual ﬁnancially-developed economy.

Table 5: Calibrated parameters - Financially developed

Parameter	Value	Target moment	Data	Model

F	0.735	Share of exporters	0.11	0.11
σε	0.148	Exporters' domestic sales premium	5.68	5.68
ωc	0.257	C imports share	0.266	0.266
ωk	0.335	Imports / GDP	0.124	0.124
θ	0.796	Credit / GDP	1.25	1.25
β	0.835	Same as baseline	-	-

5.2 Trade liberalization on intermediate and investment goods

Using our calibrated baseline and counterfactual economies, we now investigate the extent to which ﬁnancial development aﬀects the aggregate dynamics following a trade liberaliza-

12Otherwise, diﬀerences in the welfare implications of trade liberalization could be simply accounted by diﬀerences in discount rates.

tion and its welfare implications. To do so, we consider the stationary equilibrium of our calibrated model and examine the impact of a reduction in import tariﬀs on intermediate and capital goods of the magnitude observed in Colombia between 1988 and 1992. In par- ticular, we consider a one-time, unexpected, and permanent reduction of import tariﬀs from 32 percent to 12 percent.

The timing of the trade liberalization is as follows. In period 1, the economy is in a stationary equilibrium, and we refer to it as the pre-liberalization period. In particular, at the end of period 1, agents choose assets, capital and debt for the following period expecting to remain in the pre-liberalization stationary equilibrium. At the beginning of period 2, trade liberalization occurs and agents learn the full path of tariﬀs from then on.13 Thereafter, the economy begins its transition to its new steady state.

We divide our analysis into two parts. First, we examine how lowering import tariﬀs impacts aggregate variables and how these eﬀects vary with ﬁnancial development, θ. In particular, in Section 5.2.1 we examine the dynamics of key aggregate statistics following the tariﬀ reduction contrasting our baseline and ﬁnancially developed economies. In Section 5.2.2 we then investigate the welfare eﬀects of trade liberalization, its distributional consequences, and we analyze how these depend on the level of ﬁnancial development.

5.2.1 Aggregate eﬀects of trade liberalization

Long-run eﬀects. We begin by analyzing the long-run aggregate eﬀects of decreasing imports tariﬀs to capital and intermediate goods on key aggregate variables: Output, capital, consumption, exports and imports. To do so, we contrast the steady-state values of these variables before and after trade liberalization. Table 6 reports the impact that a reduction of τK has on these variables in the baseline (θ = 0.21) and in the ﬁnancially developed (θ = 0.8) economies.

Consider ﬁrst the baseline economy. Table 6 shows that the long-run eﬀects of decreasing τK are positive and quantitatively large: Real GDP increases by 1.4%, capital by 6.6%, and consumption by 3.4%. Real exports increase by 26% in response to trade liberalization, while real imports increase by 36%.14 Finally, the net exports-to-GDP ratio improves from -0.029 to -0.019 despite the large increase of imports.

To understand these results note that the lower import tariﬀs on intermediates and capital goods reduce the cost of production inputs, thus acting akin to a positive supply shock: Capital and intermediate goods become cheaper (the price of capital falls 1.7%),

13 The results do not change signiﬁcantly if agents learn about the trade liberalization one period in advance. See the results for "Expected trade liberalization" in the Online Appendix for more details.

14 We compute real variables using a Laspeyres price index, keeping prices constant at their initial steady state level.

Table 6: Steady state eﬀects (τK ↓)

	θ = 0.21	θ = 0.80
Real GDP	1.41%	1.65%
Capital	6.61%	6.58%
Consumption	3.38%	3.43%
Real exports	26.4%	19.9%
Real imports	35.8%	42.3%
Price of capital	-1.71%	-1.79%
Wage	6.23%	6.47%
Real exchange rate	5.49%	4.22%

leading to an increase in capital and output. As the domestic ﬁnal good becomes cheaper, the real exchange rate depreciates (by 5.5% in the baseline) inducing an increase of exports. The higher demand for capital and intermediates lead to an increase in the demand for labor which, along with the falling price of the ﬁnal consumption goods, increase the real wage (by 6.2% in the baseline). Thus, consumption increases as a result of higher proﬁts and wages.

We ﬁnd that diﬀerences in ﬁnancial development have a modest eﬀect on the long-run impact of trade liberalization. In particular, GDP and imports increase somewhat more in the ﬁnancially developed economy (1.7% vs 1.4%, and 42.3% vs 35.8%, respectively), while capital and consumption increase almost identically in the two economies (6.58% vs. 6.61% and 3.43% vs. 3.28%, respectively). Somewhat surprisingly, exports increase relatively less in the ﬁnancially developed economy (19.9% vs. 26.4%). This is accounted by the lower exchange rate depreciation experienced by the ﬁnancially developed economy following a reduction in τK .

The above results show that diﬀerences in ﬁnancial development have little impact on the aggregate eﬀects of trade liberalization in the long run. However, we show below that these results mask large diﬀerences in aggregate outcomes along the transition as well as substantial diﬀerences in welfare gains.

Transitional dynamics. We next examine the eﬀects of lowering import tariﬀs on capital and intermediate goods along the transition to the new steady state. Figure 3 plots the response of real GDP, capital, real exports and consumption following a reduction of τK in the baseline economy (blue solid line) and its counter-factual ﬁnancially developed counterpart (red dashed line) while Figure 4 plots the respective dynamics of prices.

Consider ﬁrst the baseline economy. The top left panel in Figure 3 shows that GDP declines on impact. This is driven by a decline in domestic sales, due to an increase in

% Real GDP

0.02
0.01
0
	Baseline
	Financially Developed
-0.01
10	20	30	40	50	60

% Real Export Sales

0.3
0.2
0.1
0
10	20	30	40	50	60

% Capital

0.08
0.06
0.04
0.02
0
10	20	30	40	50	60

% Consumption

0.04
0.03
0.02
0.01
0	Baseline

	Financially Developed
-0.01
10	20	30	40	50	60

Figure 3: Transition dynamics following a reduction in τK

% Real Wage

0.06
0.04
0.02
0
10	20	30	40	50	60

% Real Exchange Rate

0.06
0.04
0.02
		Baseline
		Financially Developed
0
10	20	30	40	50	60

% Price of Capital

0
-0.005
-0.01
-0.015
-0.02
10	20	30	40	50	60

Figure 4: Prices dynamics following a reduction in τK

foreign competition that leads to a drop in capital good producers' demand for the varieties of the least productive ﬁrms, which do not beneﬁt as much from a decrease in the cost of production driven by a fall τK . The large drop in the cost of capital and intermediate goods (bottom panel in Figure 4) results in an investment boom and increases the stock of capital between periods 2 and 3 as productive ﬁrms ﬁnd it optimal to expand their scale. Aggregate consumption declines on impact given the relatively large fraction of constrained entrepreneurs who increase their capital stock by initially sacriﬁcing consumption (bottom right panel in Figure 3). Finally, we ﬁnd that exports respond strongly on impact due to a large increase in the real exchange rate (top right panel in Figure 4) which makes exporting more proﬁtable and encourages entrepreneurs to reallocate their sales towards the foreign market.15

After the initial period, the baseline economy converges slowly towards its ﬁnal steady state driven by a slow accumulation of capital due to presence to ﬁnancial frictions. As capital increases so does the demand for labor, leading to higher wages. Thus, wages steadily increase throughout the transition (top left panel in Figure 4).

In contrast to our baseline economy, we ﬁnd that in the ﬁnancially developed economy trade liberalization does not have a negative eﬀect on impact: Both GDP and consumption remain essentially unchanged. Moreover, we ﬁnd that capital increases relatively more than in the baseline economy, since the looser ﬁnancial constraint allows entrepreneurs to increase investment without sacriﬁcing consumption. Thus, entrepreneurs are able to expand their sales at a higher rate than entrepreneurs in the baseline economy. As a consequence, the growth rates of GDP, capital, and consumption in the ﬁnancially developed economy initially surpass those featured by the baseline economy, leading the ﬁnancially developed economy to converge faster to its new steady state.

These diﬀerences in the aggregate variables across the two economies are mirrored by diﬀerences in the price dynamics across them. Figure 4 shows that wages increase by more in the ﬁnancially developed economy driven by stronger demand for labor associated with a larger expansion of capital and output. On the other hand, the real exchange rate increases by less in the more ﬁnancially-developed economy as entrepreneurs increase by more their demand for ﬁnal domestic goods. Finally, we see that the price of capital decreases by almost identical amounts in both economies in response to a decline in import tariﬀs.

15As discussed in Kohn et al. (2020), collateral constrains restrict ﬁrm-level total sales, but do not restrict entrepreneurs' allocation of sales across markets. Thus, following a depreciation, entrepreneurs increase their foreign sales (even if they cannot increase their total sales due to binding ﬁnancial constraints) by reallocating some of their sales from the domestic to foreign market. See also Almunia et al. (2018) for empirical evidence on this reallocation channel.

5.2.2 Welfare and distributional eﬀects

In this section, we analyze how ﬁnancial development aﬀects the welfare gains from trade liberalization, both in the aggregate and across the distribution of agents.

To understand how lower tariﬀs on intermediate and capital goods aﬀect welfare, note that it has both positive and negative eﬀects on entrepreneurs. Lower tariﬀs make investment goods cheaper, which decreases the cost of production and makes entrepreneurs more competitive. A decrease in the price of investment goods in turn has positive general equilibrium eﬀects: it leads to a real depreciation that boosts the proﬁts of exporting ﬁrms. Moreover, some of these gains to exporters are distributed to other entrepreneurs through an increase in wages.

However, lowering tariﬀs also leads to some negative eﬀects on welfare. On the one hand, it decreases the demand of investment good producers for domestic varieties, with particular impact on non-exporters. On the other hand, lower tariﬀs also generate lower tariﬀ revenues, hurting the poorest agents since these revenues are redistributed across all agents. Nevertheless, the overall impact of lowering τK is to lead to an economic boom, at least after the ﬁrst few periods.

These positive and negative eﬀects of trade liberalization are aﬀected by the extent of ﬁnancial development. In particular, ﬁnancial underdevelopment limits the extent to which constrained entrepreneurs can beneﬁt from a decrease in the cost of capital goods, since they are limited in their potential to expand production. However, the decrease in the cost of capital also relaxes ﬁrms' borrowing constraints. We quantify the net impact of these channels on welfare across the two economies in the rest of this section.

Aggregate welfare. We ﬁrst investigate how ﬁnancial development aﬀects aggregate welfare gains. To do so, we compute a "consumption-equivalent" welfare measure, as in Mendoza et al., 2009. Speciﬁcally, we compute the aggregate welfare gains, G, as:

		RS vT (s)φ0(s)ds	1

G =		1−γ	− 1	(11)

	RS v0(s)φ0(s)ds

where G consists of the permanent change of consumption across all agents required to make agents indiﬀerent between remaining in the economy without trade liberalization and moving to one with lower tariﬀs as described above. Moreover, φ0(s) is the stationary measure in the initial steady state, v0(s) is the value function of an entrepreneur in state s prior to trade liberalization, and vT (s) is the value function of an entrepreneur in state s in period 2 right

after trade liberalization is announced and realized.

Table 7 reports our ﬁndings.16 The ﬁrst row reports the welfare gains of trade liberalization as deﬁned in Equation 11. The second row reports the welfare gains excluding the transition; that is, restricting attention to the comparison between the initial and ﬁnal steady states. Thus, comparing the ﬁrst and second rows allows us to evaluate the welfare cost of the transition.

TABLE 7: WELFARE (τK ↓)

	θ = 0.21	θ = 0.80

Overall	0.13%	0.98%
Excluding transition	2.58%	2.76%

We ﬁnd that the welfare gains from trade liberalization are lower in the economy with ﬁnancial frictions: only 0.13% compared to almost 1% in the ﬁnancially developed economy. Notice that this diﬀerence is smaller once we exclude the transition: the second row shows that if both economies adjusted immediately to the ﬁnal steady state, then welfare would increase by 2.58% in the baseline and 2.76% in the ﬁnancially developed economy. Thus, we conclude that the diﬀerence in welfare gains between the two economies is largely accounted by the faster transition to the ﬁnal steady state in the ﬁnancially developed economy. That is, ﬁnancial frictions slow down the adjustment to the environment with lower tariﬀs, reducing the welfare gains from this policy change.

WELFARE DISTRIBUTION. We next investigate the welfare gains from a decrease in tariﬀs on capital and intermediate goods along the entrepreneurs' asset and productivity distribution.

To do so, we compute an individual-speciﬁc"consumption-equivalent" welfare measure that we use to contrast welfare changes across diﬀerent types of agents. In particular, for every entrepreneur in a given state s = (a, z) who lives in the steady-state of the economy without trade liberalization we compute the permanent percentage change of consumption g(s) that would make him indiﬀerent from moving to an economy featuring the decline τK examined above.

Analogous to Mendoza et al. (2009), g(s) can be computed as the solution to:

g(s) =	v0	(s)	1	− 1,	(12)
		vT	(s)		1−γ

where v0(s) and vT (s) are deﬁned as above. To compute the average welfare gains from trade

16Speciﬁcally, we report G × 100.

liberalization across a given group of entrepreneurs with s ∈ S, we deﬁne GS as

GS ≡ g(s)φ0(s)ds,	(13)
S

where φ0(s) is the stationary measure at the initial steady state.

We are interested in understanding (1) how the welfare gains from trade liberalization vary across the joint distribution of assets and productivity, and (2) whether there are speciﬁc regions across this distribution where entrepreneurs are worse oﬀ following the decrease in τK . To answer these questions we partition the agents into asset and productivity quintiles. We then compute the welfare gains and the share of entrepreneurs with negative welfare gains (i.e., share of losers) in each bin (i, j), where i denotes the i-th quintile of the asset distribution (i ∈ {1, ..., 5}) and j denotes the j-th quintile of the productivity distribution (j ∈ {1, ..., 5}). Figure 5 presents our results.

			1
			0.9
1			0.8
			0.7
0.5			0.6

			0.5
			0.4
0
			0.3
1
2			0.2

3			5
	4	0.1
4
	3

5		2	0
assets	1	productivity

Figure 5: Welfare (τK ↓) distribution

The top panels of Figure 5 illustrate the welfare gains from trade liberalization across the diﬀerent net worth and productivity quintiles. The top left panel presents the results

for the baseline, while the top right panel presents the results for the ﬁnancially developed economy. Not surprisingly, we observe that the welfare gains are higher on average in the more ﬁnancially developed economy; the gains are uniformly higher across all net worth and productivity bins. However, we ﬁnd that there are signiﬁcant diﬀerences in the distribution of these gains across agents. In the baseline economy the gains are mostly concentrated among the richest and more productive entrepreneurs while in the ﬁnancially developed economy the welfare gains are distributed more uniformly.

The bottom panels of Figure 5 illustrate the share of losers for each asset and productivity bin. The bottom left panel presents the results for the baseline economy, while the bottom right one presents those for the ﬁnancially developed economy. We ﬁnd that there are signiﬁcant diﬀerences in the extent to which trade liberalization leads to losses across these economies. On the one hand, a substantial share of agents (14.2%) is made worse oﬀ by trade liberalization in the baseline model; mostly agents with low productivity and low assets.17 In contrast, all agents are better oﬀ in the ﬁnancially developed economy.

TABLE 8:	WELFARE (τK ↓)


	θ = 0.21	θ = 0.80

Winners	0.26%	1.02%
Losers	-0.02%	- %
Exporters	1.03%	1.48%
Non-exporters	0.12%	0.96%
Entrepreneurs	0.63%	1.24%
Workers	0.07%	0.94%
Wealthy	0.75%	1.19%
Poor	0.16%	1.00%

To better understand the heterogeneous welfare gains from lowering tariﬀs on capital and intermediate goods, in Table 8 we now compute the welfare gains for exporters versus non- exporters, "wealthy" versus "poor" agents, and "entrepreneurs" versus "workers".18 We ﬁnd that exporters in both economies gain more than non-exporters. However, this diﬀerence is much larger in the baseline economy due to a smaller wage increase and a larger real depreciation. In the ﬁnancially developed economy a sharp wage increase redistributes some

17These entrepreneurs lose on average 0.02%, compared to an average gain of 0.26% enjoyed by the rest of the agents; see Table 8, below.

18We deﬁne wealthy agents as those in the top 10% of the wealth distribution, while the rest is deﬁned as poor. We deﬁne individuals to be workers if they supply more labor than they hire (N< 1), while the rest is deﬁned to be entrepreneurs (N ≥ 1).

of the welfare gains from exporters to non-exporters, while a larger increase in the demand for ﬁnal goods leads to a lower real exchange rate depreciation further limiting the gains of exporters. These diﬀerential eﬀects from wages can also be seen when comparing workers and entrepreneurs, as there are lower diﬀerences in welfare gains between entrepreneurs and workers in the more ﬁnancially developed economy.

Table 8 also shows that assets play a minor role in determining the welfare gains in the ﬁnancially developed economy, while these play an important role in the baseline economy. In the latter economy, wealth is crucial to beneﬁt from trade liberalization, as it allows individuals to adjust their production scale immediately; in the former economy wealth is not as necessary as agents can borrow from deeper ﬁnancial markets.

Decomposition of welfare gains. We now investigate the channels that account for the welfare results reported above. To do so, we decompose the aggregate and group-speciﬁc welfare gains into four channels through which changes in τK aﬀect welfare.

The ﬁrst channel consists of the eﬀect of changes in tariﬀ revenue on welfare; the "tariﬀs income" channel. Following trade liberalization, tariﬀ revenue declines, which reduces the amount of tariﬀs revenues that gets redistributed to the agents. The second channel consists of the impact of τK on the price of investment goods and the demand for domestic varieties by investment good producers; the "investment" channel.

The above two channels capture the "direct eﬀects" of trade liberalization. The remaining two channels consist of general equilibrium eﬀects. The third channel consists of the impact of trade liberalization on the real exchange rate; the "exchange rate" channel. Cheaper materials lead to cheaper domestic varieties and consumption good and, thus, to a depreciated real exchange rate.19 Finally, wages increase following trade liberalization given the increased demand for labor; the "wage" channel.

In order to quantify the strength of these channels, we perform a series of partial equilibrium computations. In particular, we follow Carroll and Hur (2020b) and introduce measure- zero agents: Agents that optimize in response to prices and aggregate quantities that are diﬀerent from the ones observed in equilibrium, but do not aﬀect equilibrium prices and aggregate quantities. Thus, to compute the "tariﬀs income" channel, we assume that along the transition and in the ﬁnal steady state our zero-measure agents face the initial-steady state prices and aggregate quantities but receive reduced tariﬀs revenues as observed along the equilibrium path following a trade liberalization. Similarly, to compute "investment channel" we assume that measure-zero agents only face the equilibrium changes in PK and

19Note that part of this eﬀect is driven by a change in the demand for domestic varieties by consumption good producers.

	Table 9:	Welfare decomposition (τK ↓)


θ = 0.21	Tariﬀs income	Investment	Exchange rate	Wage	Total
	(τK )		(PK , YK )	(ξ, YC)	(w)

All agents	-1.6%		0.7%	0.3%	0.9%	0.22%
Exporters	-0.7%		1.0%	1.2%	-0.4%	1.03%
Non-exporters	-1.7%		0.6%	0.2%	1.0%	0.12%
Wealthy	-0.7%		0.8%	0.9%	-0.2%	0.75%
Poor	-1.7%		0.7%	0.2%	1.0%	0.16%
Entrepreneurs	-0.9%		0.9%	0.7%	-0.1%	0.63%
Workers	-1.8%		0.6%	0.1%	1.2%	0.07%

θ = 0.8	Tariﬀs income	Investment	Exchange rate	Wage	Total
	(τK )		(PK , YK )	(ξ, YC)	(w)

All agents	-1.6%		0.9%	0.5%	1.2%	1.02%
Exporters	-0.8%		1.2%	1.5%	-0.4%	1.48%
Non-exporters	-1.7%		0.8%	0.4%	1.5%	0.96%
Wealthy	-0.8%		1.0%	1.1%	-0.1%	1.19%
Poor	-1.6%		0.9%	0.4%	1.4%	1.00%
Entrepreneurs	-1.0%		1.2%	1.1%	0.0%	1.24%
Workers	-1.7%		0.8%	0.3%	1.6%	0.94%

YK implied by a decrease in τK (with other quantities ﬁxed at their initial steady state level); for the "exchange rate" channel, we assume these agents only face the equilibrium changes of ξ and YC; and, ﬁnally, for the "wage" channel, the measure-zero agents only experience a change in wages as observed along the equilibrium path.

Table 9 reports the welfare decomposition for the baseline and ﬁnancially developed economies. We ﬁnd that the decline in tariﬀs income has a very large negative eﬀect on welfare and is the main reason why welfare gains are modest in both economies. The average loss due to lower tariﬀ income is -1.6% in both the baseline and ﬁnancially developed economies. This negative eﬀect is particularly strong among workers, poor agents, and non-exporters since tariﬀs revenue represent a larger proportion of income for these agents.20

20 In Colombia, following the trade liberalization tariﬀs revenues represented 1.3% of GDP in 1990 and

In contrast, the investment and the exchange rate channels contribute positively to welfare and are the main channels through which agents beneﬁt in the aggregate in both economies. Entrepreneurs, wealthy agents, and exporters beneﬁt relatively more since they can adjust their production by more in response to the higher domestic and foreign demand. These channels are even stronger in the ﬁnancially developed economy as agents are able to expand their production more on impact.

Finally, we ﬁnd that higher wages help redistributing part of the welfare gains from en- trepreneurs, wealthy agents, and exporters (who beneﬁt more from investment and exchange rate channels) towards workers, poor agents, and non-exporters. Since the strength of this channel is directly linked to the extent that agents can expand their production following trade liberalization, this channel is stronger in the ﬁnancially developed economy.

We have thus shown how diﬀerent equilibrium channels aﬀect welfare following a trade liberalization, and how these eﬀects are distributed across agents. In the following section, we explore in detail how inequality changes following trade liberalization.

Trade liberalization and inequality. We now investigate the implications of trade liberalization and ﬁnancial development for consumption, income and wealth inequality. Table 10 presents statistics on inequality changes for these variables, contrasting the baseline and ﬁnancially developed economies.21 We focus on two measures of inequality: the Gini coeﬃcient and the P 75/P 25 ratio, which consists of the ratio of the 75th percentile to the 25th percentile of the distribution for each variable of interest.

First, we ﬁnd that there is a pecking order of inequality across the various variables under analysis. In both economies and according to both inequality measures, consumption inequality is the lowest, it is slightly higher for income, and almost twice as high for wealth. Furthermore, we ﬁnd that in both economies trade liberalization increases all measures of inequality. Interestingly, this is the case even for the ﬁnancially developed economy, where trade liberalization increases welfare for all agents. Finally, we don't ﬁnd a clear rank across the two economies in the impact of trade liberalization on inequality: Whether or not inequality increases more in one economy or the other appears to critically depend on the particular inequality measure under analysis. In the Online Appendix we show that the increase of inequality following trade liberalization is primarily related to the reduction of

dropped to 1.0% of GDP on average for the period 1992-1998 (OECD data). In the Online Appendix, we investigate the eﬀects of trade liberalization in a counterfactual economy in which tariﬀs revenues do not decline after trade liberalization, but in which prices and quantities are allowed to adjust and clear the markets. Consistent with this exercise, we observe a much larger increase in welfare for all agents since we abstract from the negative "tariﬀs income" eﬀect depicted here.

21 Income is the sum of wages, proﬁts from domestic and foreign markets, tariﬀs revenues and asset income. Wealth is assets.

Table 10: Trade liberalization and inequality (τK ↓)

	θ = 0.21			θ = 0.80
	Pre-liberalization % Change		Pre-liberalization	% Change


		Consumption
Gini	0.34	1.37%		1.35%
	0.32
P75/P25	2.08	0.15%		1.96	0.25%
		Income
Gini	0.37	1.39%		1.40%
	0.34
P75/P25	2.11	0.49%		1.94	0.01%
		Wealth
Gini	0.61	0.83%		0.85%
	0.68
P75/P25	4.16	6.50%		4.85	7.46%

the redistributed tariﬀ revenue which raises inequality since it consists of the reduction of an income source common to all agents.

Table 11: Trade Liberalization and Income Inequality (τK ↓)

Baseline vs. Data

Pre-liberalization % Change

		Model
Gini	0.37	1.4%
P75/P25	2.11	0.5%
		Data
Gini	0.54	3.7%
P75/P25	3.31	3.6%

Notes: Data from Colombia Poverty Report, World Bank, November 1992, Table 3. Pre- and post- trade

liberalization correspond to years 1988 and 1995, respectively.

Table 11 contrasts income inequality in the baseline model with data. To do so, we use data from the World Bank's Poverty Report for Colombia and focus on years 1988 and 1995 as our empirical counterpart for the pre- and post- trade liberalization results. First, notice that inequality is much higher in the data according to both measures of inequality. This is to be expected since all agents in our model are both workers and entrepreneurs and thus, for instance, our model doesn't feature very poor unemployed or hand-to-mouth agents, nor our model can account for the high degree of wealth concentration of the richest households.

Second, notice that our model can explain 38% of the observed increase in the Gini coeﬃcient from 1988 to 1995 (1.4% vs. 3.7%) and 14% of the observed increase in the P 75/P 25 ratio.

5.3 Trade liberalization on all goods (τK and τC)

In the quantitative analysis above we focus on the eﬀects of a unilateral reduction of tariﬀs charged on imports of investment and intermediate goods. The reasons we focus on import tariﬀs charged on these goods are threefold: (i) a reduction of tariﬀs on capital and intermediate goods makes capital goods cheaper, leading ﬁrms to increase their capital stocks, thus making ﬁnancial frictions more likely to distort allocations by limiting capital accumulation; (ii) most imports in emerging countries consist of these types of goods (3/4 of all Colombian imports, as reported in the calibration section); and, ﬁnally, (iii) there is an easier case to be made by policy-makers for reducing these tariﬀs, since they have a direct impact on domestic ﬁrms' productivity.

Yet, tariﬀs are usually reduced on all types of goods, not just on capital and intermedi- ates. Thus, in this section we investigate the sensitivity of our ﬁndings to considering trade liberalization of consumption goods in addition to capital and intermediates.22 To do so, we use our calibrated baseline and counterfactual economies to investigate the extent to which ﬁnancial development aﬀects the aggregate dynamics following trade liberalization and its welfare implications. As above, we consider the stationary equilibrium of our calibrated model and examine the impact of a reduction of import tariﬀs on consumption, capital, and intermediates, of the magnitude observed in Colombia between 1988 and 1992. In particu- lar, we consider a one-time, unexpected, and permanent reduction of import tariﬀs from 32 percent to 12 percent. The timing of trade liberalization is analogous to the analysis above.23

5.3.1 Aggregate eﬀects of trade liberalization

Long-run eﬀects. We begin by analyzing the long-run aggregate eﬀects of decreasing imports tariﬀs on key aggregate variables: Output, capital, consumption, exports and imports. To do so, we contrast the steady-state values of these variables before and after trade liber- alization. Table 12 reports the impact that a reduction of τK and τC has on these variables in the baseline (θ = 0.21) and in the ﬁnancially developed (θ = 0.8) economies.

Consider ﬁrst the baseline economy. Table 12 shows that the long-run eﬀects of decreasing τK and τC are signiﬁcatively lower than in the case in which only τK is reduced: Real GDP decreases by 0.45% (vs. an increase of 1.4% when only τK is reduced), capital increases by

See the Online Appendix for the eﬀects of reducing import tariﬀs on only consumption goods.
We present only the main results. See the Online Appendix for additional details.

Table 12: Steady state eﬀects (τK ↓, τC ↓)

	θ = 0.21	θ = 0.80
Real GDP	-0.45%	-0.10%
Capital	4.80%	5.18%
Consumption	2.21%	2.66%
Real exports	32.5%	24.1%
Real imports	46.9%	55.9%
Price of capital	-0.92%	-0.93%
Wage	5.93%	6.66%
Real exchange rate	7.38%	5.90%

4.8 (vs. 6.6%), and consumption by 2.2% (vs. 3.4%). In response to trade liberalization, real exports and imports increase by more in this case, 32.5% vs 26% and 47% vs 36%. Despite the large increase of imports, the net exports-to-GDP ratio improves from -0.029 to -0.017.

While a reduction of import tariﬀs on intermediates and capital goods reduce the cost of production inputs and has a positive overall eﬀect in the economy, a reduction of τC negatively impacts all variables except exports. To understand why this is the case note that as imported consumption varieties become cheaper, domestic consumption good producers switch from domestic to foreign varieties. Thus, a reduction of τC leads to a decrease in domestic sales. This decreases entrepreneurs' proﬁts leading to a fall in capital and consumption. On the other hand, these negative eﬀects are partially oﬀset by the lower price of ﬁnal consumption which results in a real depreciation. This increases exports, partially oﬀsetting the negative impact on domestic sales. In addition, the lower price of ﬁnal goods the increases real wage, yet not enough to oﬀset the negative impact due to increased foreign competition.24

As in the previous section, diﬀerences in ﬁnancial development have modest eﬀects on the long-run impact of trade liberalization. In particular, GDP, capital, consumption, and imports increase somewhat more in the ﬁnancially developed economy (or decrease by less, in the case of GDP). Exports increase relatively less in the ﬁnancially developed economy. This is accounted by the lower exchange rate depreciation experienced by the ﬁnancially developed economy following a tariﬀ reduction.

Transitional dynamics. We next examine the eﬀects along the transition to the new steady state. Figure 6 plots the response of real GDP, capital, real exports and consumption

24 Note that a similar negative eﬀect operates when only τK changes. However, in that case the resulting decline in the price of investment goods allows domestic entrepreneurs to compete more eﬀectively in the foreign and domestic markets. This eﬀect dominates the initial negative impact of an increase in foreign competition following a reduction of τK .

Table 13: Welfare following decrease in tariﬀs

	θ = 0.21	θ = 0.80
τK	0.13%	0.98%
τC	-0.84%	-0.59%
τK & τC	-0.42%	0.50%

following a reduction of τK and τC in the baseline economy (blue solid line) and its counter- factual ﬁnancially developed counterpart (red dashed line) while Figure 7 plots the respective dynamics of prices.

Notice that these ﬁgures look similar to Figure 3 and Figure 4, except that all lines are below their values when only tariﬀs on capital goods are reduced. Thus, when we additionally reduce tariﬀs on consumption goods, all aggregate variables are negatively aﬀected (except exports), somewhat counteracting the positive eﬀect of a reduction in tariﬀs on capital goods. As before, GDP, capital and consumption increase slowly following the trade liberalization, but ﬁnancial development allows for a much faster increase of these variables.

5.3.2 Welfare and distributional eﬀects

Finally, we brieﬂy discuss how ﬁnancial development aﬀects the welfare gains from trade liberalization, both in the aggregate and across the distribution of agents.

Table 13 reports the welfare eﬀects implied by a simultaneous decrease in tariﬀs on all types of varieties (i.e., a decrease in τK and τC). A fall in tariﬀs on capital goods leads to an increase in welfare, while a drop in tariﬀs on consumption goods has a negative impact on welfare. Aggregate eﬀects on welfare are negative in the baseline economy, equal to -0.42%, while they are positive in the ﬁnancially developed economy, equal to 0.50%, when both tariﬀs are reduced at the same time.

Welfare distribution. Finally, we consider the distribution of welfare eﬀects across agents. To better understand the heterogeneous welfare gains from lowering tariﬀs, in Table 14 we now compute the welfare gains for exporters versus non-exporters, "wealthy" versus "poor" agents, and "entrepreneurs" versus "workers". As above, the average eﬀects are negative in the baseline economy, while they are positive in the ﬁnancially-developed economy. We ﬁnd that exporters in both economies gain more than non-exporters, entrepreneurs gain more than workers and the top 10% of the agents wealth distribution gain much more than the bottom 90% in the wealth distribution. These diﬀerential eﬀects are more unequal in the baseline economy than in the ﬁnancially developed one.

Thus, these results are broadly consistent with the results presented in the previous

% Real GDP

0.01
		Baseline
		Financially Developed
0
-0.01
-0.02
-0.03
10	20	30	40	50	60

% Real Export Sales

0.4
0.3
0.2
0.1
0
10	20	30	40	50	60

% Capital

0.06
0.04
0.02
0
10	20	30	40	50	60

% Consumption

0.03
0.02
0.01
0
-0.01
10	20	30	40	50	60

Figure 6: Transition dynamics following a reduction in τK and τC

% Real Wage

0.08
0.06
0.04
0.02
0
10	20	30	40	50	60

% Real Exchange Rate

0.08
0.06
0.04
0.02
		Baseline
		Financially Developed
0
10	20	30	40	50	60

% Price of Capital

0
-0.005
-0.01
-0.015
10	20	30	40	50	60

Figure 7: Prices dynamics following a reduction in τK and τC

section for the case in which only tariﬀs to capital and intermediate goods are reduced.

TABLE 14: WELFARE (τK ↓,τC ↓)

	θ = 0.21	θ = 0.80

All agents	-0.29%	0.57%
Winners	0.68%	0.57%
Losers	-0.46%	- %
Exporters	0.86%	1.47%
Non-exporters-0.43%	0.46%
Entrepreneurs	0.28%	1.00%
Workers	-0.50%	0.43%
Wealthy	0.73%	1.06%
Poor	-0.30%	0.56%

6 CONCLUSION

In this paper, we study the role of ﬁnancial development on the aggregate eﬀects and welfare implications of reducing international trade barriers on capital and intermediate inputs. We formulate a quantitative general equilibrium model with heterogeneous ﬁrms subject to ﬁnancial constraints, estimate it to match salient features from Colombian plant-level data, and ﬁnd that the adjustment to a trade liberalization is signiﬁcantly slower in ﬁnancially un- derdeveloped economies. Moreover, we ﬁnd that ﬁnancial development increases the welfare gains from trade liberalization: low-income agents beneﬁt from higher wages while exporters beneﬁt from a depreciated real exchange rate and lower capital costs.

Our paper contributes to the literature along three dimensions. First, we document that ﬁnancially underdeveloped economies grow relatively slower than their ﬁnancially developed counterparts after reducing import tariﬀs on capital and intermediate goods. Our second contribution is to quantify the extent to which cross-country diﬀerences in ﬁnan- cial development account for the observed diﬀerences in aggregate dynamics following trade liberalization. Consistent with the data, we ﬁnd that diﬀerences in ﬁnancial development signiﬁcantly reduce the short- and medium-run impact of trade liberalization on real GDP, consumption, and investment. Our third contribution is to investigate the implications of these ﬁndings for both aggregate welfare as well as for the distribution of welfare gains across various population sub-groups. Consistent with our aggregate quantitative ﬁndings, we ﬁnd

that aggregate welfare increases relatively more in the ﬁnancially developed economy. In addition, we show that the distribution of welfare gains is considerably more unequal when the economy is less ﬁnancially developed.

More broadly, our ﬁndings provide a rationale for the higher resistance to trade liberalization in less developed economies: There might be less to gain from trade openness in these economies, particularly in the short- and medium-run. Therefore, our results imply that trade liberalization might need to be pursued together with reforms aimed towards improving ﬁrms' access to ﬁnance. In addition, our ﬁndings suggest that ﬁnancial development might not only provide the signiﬁcant welfare gains that have already been documented in the literature (Buera et al. 2011, Midrigan and Xu 2014), but it might also help ensuring more equally distributed welfare gains from trade liberalization, thus increasing the support for such policies across diﬀerent types of agents.

REFERENCES

ALESSANDRIA, G. AND H. CHOI, "Establishment Heterogeneity, Exporter Dynamics, and the Eﬀects of Trade Liberalization," Journal of International Economics 94 (2014), 207- 223.

ALESSANDRIA, G., H. CHOI AND K. RUHL, "Trade adjustment dynamics and the welfare gains from trade," Technical Report, Working Paper, 2018.

ALESSANDRIA, G. A., C. ARKOLAKIS AND K. J. RUHL, "Firm Dynamics and Trade,"

Technical Report, National Bureau of Economic Research, 2020.

ALMUNIA, M., P. ANTRAS` , D. LOPEZ RODRIGUEZ AND E. MORALES, "Venting out:

Exports during a domestic slump," (2018).

AMITI, M. AND J. KONINGS, "Trade liberalization, intermediate inputs, and productivity: Evidence from Indonesia," American Economic Review 97 (2007), 1611-1638.

ANDERSON, J. E., M. LARCH AND Y. V. YOTOV, "Growth and trade with frictions: A structural estimation framework," Technical Report, National Bureau of Economic Research, 2015.

AUTOR, D. H., D. DORN AND G. H. HANSON, "The China shock: Learning from labor- market adjustment to large changes in trade," Annual Review of Economics 8 (2016), 205-240.

AUTOR, D. H., D. DORN, G. H. HANSON AND J. SONG, "Trade adjustment: Worker-level evidence," The Quarterly Journal of Economics 129 (2014), 1799-1860.

BROOKS, W. AND A. DOVIS, "Credit market frictions and trade liberalizations," Journal of Monetary Economics (2019).

BUERA, F. J., J. P. KABOSKI AND Y. SHIN, "Finance and Development: A Tale of Two Sectors," American Economic Review 101 (2011), 1964-2002.

BUERA, F. J. AND B. MOLL, "Aggregate Implications of a Credit Crunch," American Economic Journal: Macro 7 (2015).

BURSTEIN, A. AND M. J. MELITZ, "Trade liberalization and ﬁrm dynamics," Technical Report, National Bureau of Economic Research, 2011.

CAGGESE, A. AND V. CUNAT, "Financing Constraints, Firm Dynamics, Export Decisions, and Aggregate Productivity," Review of Economic Dynamics 16 (January 2013), 177-193.

CARROLL, D. AND S. HUR, "On the distributional eﬀects of international tariﬀs," in Work-

ing Paper (2020a).

CARROLL, D. R. AND S. HUR, "On the heterogeneous welfare gains and losses from trade," Journal of Monetary Economics 109 (2020b), 1-16.

CHANEY, T., "Liquidity Constrained Exporters," NBER Working Papers 19170, 2013.

CIHAK, M., A. DEMIRGUC¨¸-KUNT,E. FEYEN AND R. LEVINE, "Benchmarking ﬁnancial

systems around the world," (2012).

COSTANTINI, J. AND M. MELITZ, "The dynamics of ﬁrm-level adjustment to trade liber- alization," The organization of ﬁrms in a global economy 4 (2008), 107-141.

ESTEVADEORDAL, A. AND A. M. TAYLOR, "Is the Washington consensus dead? Growth, openness, and the great liberalization, 1970s-2000s," Review of Economics and Statistics 95 (2013), 1669-1690.

FEENSTRA, R. C., R. INKLAAR AND M. P. TIMMER, "The next generation of the Penn World Table," American economic review 105 (2015), 3150-82.

FIELER, A. C., M. ESLAVA AND D. Y. XU, "Trade, quality upgrading, and input linkages: Theory and evidence from colombia," American Economic Review 108 (2018), 109-46.

GOLDBERG, P., A. KHANDELWAL, N. PAVCNIK AND P. TOPALOVA, "Trade liberalization

and new imported inputs," American economic review 99 (2009), 494-500.

GOLDBERG, P. K., A. K. KHANDELWAL, N. PAVCNIK AND P. TOPALOVA, "Imported

intermediate inputs and domestic product growth: Evidence from India," The Quarterly journal of economics 125 (2010), 1727-1767.

GOPINATH, G. AND B. NEIMAN, "Trade adjustment and productivity in large crises," American Economic Review 104 (2014), 793-831.

GREENAWAY, D., W. MORGAN AND P. WRIGHT, "Trade liberalization and growth in developing countries: Some new evidence," World Development 25 (1997), 1885-1892.

HALPERN, L., M. KOREN AND A. SZEIDL, "Imported inputs and productivity," American Economic Review 105 (2015), 3660-3703.

IRWIN, D. A., "Does Trade Reform Promote Economic Growth? A Review of Recent Evidence," Technical Report, National Bureau of Economic Research, 2019.

KING, R. G. AND R. LEVINE, "Finance and growth: Schumpeter might be right," The quarterly journal of economics 108 (1993), 717-737.

KOHN, D., F. LEIBOVICI AND M. SZKUP, "Financial Frictions and New Exporter Dynam- ics," International Economic Review 57 (2016), 453-486.

---, "Financial frictions and export dynamics in large devaluations," Journal of International Economics 122 (2020), 103257.

LEIBOVICI, F., "Financial Development and International Trade," Working Paper, 2016. MANOVA, K., "Credit Constraints, Heterogeneous Firms, and International Trade," Review

of Economic Studies 80 (April 2013), 711-744.

MELITZ, M. J., "The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity," Econometrica 71 (November 2003), 1695-1725.

MENDOZA, E. G., V. QUADRINI AND J.-V.R´ıOS-RULL, "On the welfare implications of ﬁnancial globalization without ﬁnancial development," in NBER International Seminar on Macroeconomics 2007 (University of Chicago Press, 2009), 283-312.

MIDRIGAN, V. AND D. Y. XU, "Finance and Misallocation: Evidence from Plant-Level Data," American Economic Review 104 (November 2014), 422-58.

MOLL, B., "Productivity Losses from Financial Frictions: Can Self-Financing Undo Capital Misallocation?," American Economic Review 104 (2014), 3186-3221.

PAVCNIK, N., "Trade Liberalization, Exit, and Productivity Improvements: Evidence from Chilean Plants," The Review of Economic Studies 69 (2002), 245-276.

RAVIKUMAR, B., A. M. SANTACREU AND M. SPOSI, "Capital Accumulation and Dynamic Gains from Trade," (2017).

ROBERTS, M. J. AND J. R. TYBOUT, "The Decision to Export in Colombia: An Empirical Model of Entry with Sunk Costs," American Economic Review 87 (1997), 545-564.

RUHL, K. J. AND J. L. WILLIS, "New exporter dynamics," International Economic Review 58 (2017), 703-726.

SACHS, J. D., A. WARNER, A. ˚ASLUND AND S. FISCHER, "Economic reform and the

process of global integration," Brookings papers on economic activity 1995 (1995), 1-118. TOPALOVA, P. AND A. KHANDELWAL, "Trade liberalization and ﬁrm productivity: The

case of India," Review of economics and statistics 93 (2011), 995-1009.

WACZIARG, R. AND K. H. WELCH, "Trade liberalization and growth: New evidence," The World Bank Economic Review 22 (2008), 187-231.

WORLD BANK, "Colombia - Export Development Project (English)," Technical Report, Washington, DC: World Bank, 1993.

Attachments

Original document
Permalink

Disclaimer

Federal Reserve Bank of St. Louis published this content on 29 October 2020 and is solely responsible for the information contained therein. Distributed by Public, unedited and unaltered, on 29 October 2020 22:24:05 UTC

Labour and Tories clash on UK economy as both hint at tax cuts	01:30pm	AN
Zimbabwe approves licensing of Musk's Starlink internet service	01:28pm	RE
Fire at entertainment venue kills at least 24 people in western India	12:59pm	RE
Pope kicks off 'World Children's Day' at Rome's Olympic Stadium	12:54pm	RE
World Court's order on Rafah does not rule out entire offensive, Israel says	12:40pm	RE
Elon Musk plans xAI supercomputer, The Information reports	12:31pm	RE
EU should follow U.S. on tariffs over Chinese goods, Italy minister says	12:21pm	RE
LGBTQ+ Poles call for action to boost rights at Pride march	12:19pm	RE
Heatwave challenges India's massive election	12:14pm	RE
Deadly Russian strikes hit Kharkiv hardware store	12:14pm	RE
Macron says he would debate with far-right's Le Pen ahead of EU vote	12:10pm	RE
Yellen sees White House moving quickly to name FDIC nominee	11:51am	RE
Vadilal Industries Limited Reports Earnings Results for the Fourth Quarter and Full Year Ended March 31, 2024	11:48am	CI
As vote nears, South Africa's Ramaphosa promises to 'do better'	11:47am	RE
US 'deeply concerned' over China military drills in Taiwan Strait, State Dept says	11:28am	RE

Stock Market News in real time

Financial Development and Trade Liberalization

All News: More news

Headlines

Most Read News