Currency · United States · CFDs, forwards, futures, swaps · Yearly · United States Free

Effect of Corruption on FX Markets

Academic paper

Corruption, Carry Trades, and the Cross Section of Currency Returns

AuthorsKlaus Grobys; Jari‐Pekka Heinonen

Institute

FIUniversity of Vaasa
FIUniversity of Jyväskylä
?University of Jyväskyla
?University of Vaasa - Department of Accounting and Finance

Paper linkhttps://papers.ssrn.com/sol3/papers.cfm?abstract_id=2961743

Strategy in a nutshell

This strategy invests in 17 liquid currency pairs against the US dollar. Each year, currencies are ranked by the previous year’s Corruption Perception Index (CPI). The portfolio goes long on the five least-corrupt currencies and short on the five most-corrupt, creating a zero-cost portfolio rebalanced annually to capture potential currency performance differences linked to corruption levels.

Economic rationale

While traditional FX models focus on dollar and carry risk, corruption remains an underexplored factor. Research shows high corruption can negatively impact GDP and currency value, introducing additional risk for investors and influencing market behavior. This strategy seeks to exploit these corruption-related effects on currency returns.

Backtest performance

Annualised return6.04%

Volatility8.14%

Beta-0.013

Sharpe ratio0.74

Win rate46%

Full Python code

1from AlgorithmImports import *
2#endregion
3class EffectCorruptionFX(QCAlgorithm):
4def Initialize(self):
5self.SetStartDate(2010, 1, 1)
6self.SetCash(100000)
7self.symbols = {'CAD' : 'USDCAD',
8                'GBP' : 'USDGBP', 
9                'MXN' : 'USDMXN',
10                'EUR' : 'USDEUR',
11                'NOK' : 'USDNOK', 
12                'CHF' : 'USDCHF',
13                'SEK' : 'USDSEK',
14                'AUD' : 'USDAUD',
15                'NZD' : 'NZDUSD',
16                'JPY' : 'USDJPY', 
17                'HKD' : 'USDHKD',
18                'SGD' : 'USDSGD', 
19                'ZAR' : 'USDZAR',
20                'INR' : 'USDINR'
21                }
22for symbol in self.symbols:
23    data = self.AddForex(self.symbols[symbol], Resolution.Daily, Market.FXCM)
24
25self.cpi = self.AddData(CPIData, 'CPI', Resolution.Daily).Symbol
26self.quantile = 3
27first_key = [x for x in self.symbols.keys()][0]
28self.Schedule.On(self.DateRules.MonthStart(self.symbols[first_key]), self.TimeRules.AfterMarketOpen(self.symbols[first_key]), self.Rebalance)
29def Rebalance(self):
30if self.Time.month != 1: return
31cpi = {}
32for symbol in self.symbols:
33    if self.Securities[self.cpi].GetLastData():
34        cpi[symbol] = self.Securities['CPI'].GetLastData()[symbol]
35    
36if len(cpi) < self.quantile: 
37    self.Liquidate()
38    return
39sorted_by_cpi = sorted(cpi.items(), key = lambda x: x[1], reverse = True)
40quantile = int(len(sorted_by_cpi) / self.quantile)
41long = [x[0] for x in sorted_by_cpi[:quantile]]
42short = [x[0] for x in sorted_by_cpi[-quantile:]]
43
44self.Liquidate()
45
46long_count = len(long)
47short_count = len(short)
48
49for symbol in long:
50    if symbol == 'NZD':
51        self.SetHoldings(self.symbols[symbol], 1/long_count)
52    else:
53        self.SetHoldings(self.symbols[symbol], -1/long_count)
54for symbol in short:
55    if symbol == 'NZD':
56        self.SetHoldings(self.symbols[symbol], -1/short_count)
57    else:
58        self.SetHoldings(self.symbols[symbol], 1/short_count)
59# CPI data
60# NOTE: IMPORTANT: Data order must be ascending (date-wise)
61from dateutil.relativedelta import relativedelta
62class CPIData(PythonData):
63def GetSource(self, config, date, isLiveMode):
64return SubscriptionDataSource("data.quantpedia.com/backtesting_data/economic/CPI.csv", SubscriptionTransportMedium.RemoteFile, FileFormat.Csv)
65def Reader(self, config, line, date, isLiveMode):
66data = CPIData()
67data.Symbol = config.Symbol
68
69if not line[0].isdigit(): return None
70
71# Example File Format:
72# year;CAD;GBP;MXN;EUR;NOK;CHF;SEK;AUD;NZD;JPY;HKD;SGD;ZAR;INR
73# 2013;84;74;34;79;85;86;88;85;90;74;77;87;43;36
74#
75# YEARLY DATA
76split = line.split(';')
77
78data.Time = datetime.strptime(split[0], "%Y") + relativedelta(months=12)  # NOTE: Preventing of look ahaead bias. Add 12 months so this year we see last year's data.
79data.Value = split[1]
80data['CAD'] = int(split[1])
81data['GBP'] = int(split[2])
82data['MXN'] = int(split[3])
83data['EUR'] = int(split[4])
84data['NOK'] = int(split[5])
85data['CHF'] = int(split[6])
86data['SEK'] = int(split[7])
87data['AUD'] = int(split[8])
88data['NZD'] = int(split[9])
89data['JPY'] = int(split[10])
90data['HKD'] = int(split[11])
91data['SGD'] = int(split[12])
92data['ZAR'] = int(split[13])
93data['INR'] = int(split[14])
94return data
95# Custom fee model.
96class CustomFeeModel(FeeModel):
97def GetOrderFee(self, parameters):
98fee = parameters.Security.Price * parameters.Order.AbsoluteQuantity * 0.00005
99return OrderFee(CashAmount(fee, "USD"))